Testimony, Belief and Unbelief in the Problem of Faith and Reason

A thesis submitted to the University of Manchester for the degree of Doctor of Philosophy in the Faculty of Humanities

2017

Finlay P Malcolm

School of Social Sciences

2

CONTENTS Abstract ...... 6

Declaration and Copyright Statement ...... 7

Acknowledgements ...... 8

0. INTRODUCTION ...... 9

0.1 Situating the Project ...... 9 0.2 Chapter Overviews ...... 13 1. NEW DIRECTIONS IN FAITH AND REASON ...... 15

1.1 Overview ...... 15 1.2 The Problem of Faith and Reason ...... 16 1.2.1 Belief ...... 16 1.2.2 Evaluating Faith & Belief ...... 19 1.3 Epistemic Justification ...... 22 1.3.1 The Value of Epistemically Justified Beliefs ...... 22 1.3.2 Evidentialism ...... 23 1.3.3 Evidence ...... 27 1.3.4 Foundationalism ...... 33 1.3.5 Non-Inferential Justification ...... 37 1.4 Sources of Doxastic Justification and Normativity ...... 40 1.4.1 Prudential Justification ...... 41 1.4.2 Normative Reasons for Belief ...... 42 1.5 The PFR Redux ...... 46 1.5.1 An Epistemic Formulation of the PFR ...... 46 1.5.2 Response 1: Non-Doxastic Theories of Faith ...... 50 1.5.3 Response 2: The Divine Testimony Theory ...... 52 1.6 Conclusion ...... 56 2. THE NATURE OF FAITH ...... 57

2.1 Overview ...... 57 2.2 Analysing Faith ...... 58 2.3 Abrahamic Faith ...... 65 2.3.1 Propositional Faith and Relational Faith ...... 66 2.3.2 Trust ...... 69

3

2.3.3 Testimonial Trust ...... 78 2.3.4 Trust and Testimony in Abrahamic Faith ...... 85 2.3.5 Relational Faith as the Cause of Propositional Faith ...... 87 2.3.6 Propositional Faith ...... 90 2.4 Conclusion ...... 94 3. NON-DOXASTIC FAITH...... 95

3.1 Overview ...... 95 3.2 The Non-Doxastic Theory ...... 95 3.3 Four Arguments for the Non-Doxastic Theory ...... 101 3.3.1 Doubt...... 101 3.3.2 Language ...... 107 3.3.3 Volition ...... 112 3.3.4 Pragmatism ...... 115 3.4 The Fictionalist Objection ...... 120 3.4.1 The Objection...... 120 3.4.2 The Problem with Fictionalist Faith ...... 125 3.4.3 The Importance of Truth ...... 127 3.5 Conclusion ...... 133 4. DIVINE TESTIMONY ...... 135

4.1 Overview ...... 135 4.2 Faith: ‘The Ear of the Soul’...... 135 4.3 Divine Speech ...... 139 4.3.1 Authorial and Deputised Discourse ...... 139 4.3.2 Appropriated Discourse ...... 145 4.3.3 Appropriated Discourse in Theology ...... 148 4.4 Conclusion ...... 150 5. JUSTIFYING FAITH IN DIVINE TESTIMONY ...... 152

5.1 Overview ...... 152 5.2 Reductionism and Non-Reductionism in Testimony ...... 152 5.2.1 Non-Reductionism ...... 153 5.2.2 Reductionism ...... 156 5.3 Modified Reductionism in Divine Testimony ...... 161 5.3.1 The Absent Speaker ...... 161

4

5.3.2 Locke’s Modified Reductionism ...... 162 5.3.3 Justifying Faith Reductively ...... 166 5.4 Non-Reductionism in Divine Testimony ...... 170 5.4.1 Deputised Discourse: Wahlberg ...... 170 5.4.2 Appropriated Discourse: Plantinga ...... 174 5.4.3 Defeaters and Epistemic Authority ...... 185 5.5 Conclusion ...... 190 6. CONCLUSION ...... 191

6.1 Overview ...... 191 6.2 Faith in Testimony ...... 191 6.3 Justified Religious Faith ...... 193 6.4 Faith without Belief ...... 195 References ...... 198

Word Count (including footnotes): 79, 519

5

Abstract This thesis evaluates two current theories of propositional faith, and explores the prospects of each theory for responding to the perennial problem of faith and reason (PFR). According to the first theory – the non-doxastic theory of faith (NDT) – faith that p does not require belief that p. NDT can be used to respond to the PFR when we accept that beliefs held on faith do not meet the required standards for epistemic justification, but deny that all instances of faith require belief. Where someone’s faith does not involve belief, her faith may be adequately responsive to the epistemic reasons in her possession. According to the second theory – the divine testimony theory (DTT) – faith requires trusting the testimony of a divine being where this involves believing the testimony. One can look to defend DTT by drawing from the significant resources that have recently been developed in the philosophy of testimony to assess the standards of justification in contexts of divine testimony. DTT takes the PFR head on, but addresses it from a fresh perspective. In this thesis I defend three claims. First, NDT is untenable as a theory of faith and hence fails as a response to the PFR. Second, DTT is defensible as a theory of faith. Third, by using the resources from the epistemology of testimony one can use DTT to defend the justification of propositional faith held on divine testimony, and hence can adequately respond to the PFR. Chapter 1 presents a version of the PFR by setting out a fairly typical account of epistemic justification, and shows precisely how NDT and DTT can be used to respond to the PFR. Chapter 2 defends DTT as a theory of faith within the Abrahamic religions by giving an analysis of both trust and trust in testimony, and tying these accounts to Abrahamic faith. Chapter 3 presents the non-doxastic theory of faith as both an objection to DTT, and as a response to the PFR in its own right. NDT is rejected on the grounds that the arguments used in its favour ultimately fail, and that it cannot distinguish faith from kinds of pretence like fictionalism. Chapter 4 looks at several ways by which God might be thought to speak to people, and Chapter 5 develops an epistemological theory of faith in divine testimony. By using the resources from general theories of testimony, Chapter 5 considers and rejects several ways of defending DTT before augmenting the work of Alvin Plantinga to give a robust and defensible theory of the justification of beliefs held on divine testimony, and a promising response to the PFR. The Conclusion suggests ways of advancing the research from this thesis. 6

Declaration and Copyright Statement

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes.

Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made.

The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions.

Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=24420), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses.

7

Acknowledgements

I would like to thank my supervisors, Michael Scott, David Liggins and Chris Daly for guiding my work and providing incisive feedback throughout my time as a doctoral researcher. Thanks also to the philosophy department at the University of Manchester for many stimulating and helpful discussions, and for providing an enjoyable social environment within which to study. I also thank the University of Manchester for awarding me with a President’s Doctoral Scholarship, without which this research would not have been possible. I have also received excellent feedback which has aided my research from my attendance at numerous conferences, especially the Value of Faith conference in San Antonio in January 2016, for which I am grateful for the invitation from Trent Dougherty, and for discussion of my ideas with Daniel Howard- Snyder, Daniel McKaughan, and Jonathan Kvanvig.

Parts of Chapter 3 of this thesis have been published as ‘Faith, Belief and Fictionalism’ (with Michael Scott) in Pacific Philosophical Quarterly, and as ‘Can Fictionalists Have Faith?’ in Religious Studies. Thanks to the reviewers and editors of these journals for their feedback. The ideas and arguments from those papers has since been developed upon significantly to produce the work for Chapter 3.

There are many, many people who supported me through the process of applying to undertake my PhD, and who have been supportive of me throughout the last three years. The list is too long to name names but you all belong to my families in Guildford and, now, in Manchester too. I can’t express my thanks to you enough. I must also add my thanks to my parents who have always let me pursue my own path, and have given me so much support throughout the years.

No words, however, can sufficiently describe how appreciative I am to my wonderful wife, Lilia Malcolm. She gave up so much to support me, and has walked every day of my PhD life supporting and encouraging me through good times and hard times. This thesis, and who I am as a person, does not exist without her support, and it is to her that I dedicate this work.

8

0. INTRODUCTION

0.1 Situating the Project

Religious epistemology is primarily concerned with two connected questions: (1) How can we have religious knowledge – particularly knowledge of God? (2) Are our religious beliefs in general, and our beliefs about God in particular, rational?1 These questions are connected in the sense that religious knowledge is thought to require rational beliefs. I understand beliefs with a religious subject-matter to involve transcendent entities like God, angels, and demons; doctrines like the trinity; divine plans for us and for the world; places like heaven and hell; and supernatural events like the resurrection. Religious epistemology, broadly construed, concerns the epistemic status of our beliefs with regard to this sort of subject-matter. One of the main contributing factors to the notable increase in philosophical attention paid to religious epistemology in particular over the last few decades has been the concurrent rise in attention paid to epistemology in general. In epistemology, there has been a greater emphasis placed on the justification of our beliefs arising from individual sources of knowledge like perception, along with developments of externalist theories of knowledge like ‘reliabilism’ (Goldman 1979). The most well received developments in religious epistemology have taken theories from general epistemology, and applied them to questions concerning knowledge with a religious subject-matter. The locus classicus of this comes from the work of Alvin Plantinga (1983, 2000) who developed a way of arguing that religious beliefs are justified without requiring any evidence for their support. He does so by hypothesising that humans have a faculty of perception that allows us to form beliefs about God in a basic, immediate way.2 This theory marks one way of applying a general theory of epistemic justification into the context of religious belief. If a theory is plausible in

1 Other definitions are less precise than this, but seem to say roughly the same thing. For instance, Dougherty and Tweedt claim that ‘Religious epistemology is the study of how subjects’religious beliefs can have, or fail to have, some form of positive epistemic status (such as knowledge, justification, warrant, and rationality) and whether they even need such status appropriate to their kind’ (2015, p.547). 2 Plantinga calls this the ‘sensus divinitatis’. A similar, though slightly more limited theory is Alston (1991).

9 general epistemology, then that plausibility may carry through into the applied context too. Part of this thesis attempts to make a similar kind of move. Over the last 25 years or so, there has seen a major research programme developed in the epistemology of testimony. It is now widely held that testimony is a source of a basic kind of belief, and often is a source of knowledge. This claim runs parallel to a claim that Plantinga was working on, according to which perception is a source of a basic kind of belief. In the latter case, we have seen the development of a tradition of philosophical research that investigates just how perception can deliver knowledge of God to a perceiver. Has there been any attention paid to the former case of testimony? From my own research I have found very little, outside of a few works, belonging largely to what is sometimes called ‘analytic theology’. This might be surprising since religious testimony has surely been addressed in much of the discussion around miracles since the work of Hume. But the context to the testimony considered by Hume is a matter of human-to- human interaction. This is not the kind of testimony that’s isomorphic to the case of perception that was of interest to Plantinga. What we are looking for is not so much a trust in the testimony of people who have had religious experiences, but a trust in the testimony that comes from God himself. In effect, we want a theory of testimony as coming from the very speech of God; in other words, a theory of divine testimony. Now, the reason that such a theory has been largely confined to theology is because it looks like what we have here is a request for a theory of divine revelation, and this is a particularly theological topic. But philosophical approaches to this theme can move the field of religious epistemology forwards in a way that, from what I have found at least, has received almost no attention in the philosophy of religion. Moreover, it’s not at all obvious that God speaking is the same as God revealing, and so a theory that works from philosophical accounts of speech may bring different ideas into play compared with theological accounts of revelation. Looking back on the two questions that I said were central to religious epistemology, I want to show how this application of theories of testimony to religious epistemology is a matter of claiming that testimony is a way of having knowledge of God, and that beliefs based on this testimony can be rational in much the same way that beliefs based on everyday testimony can be rational. This application of testimony to religious epistemology is the main strand of research in this thesis. It’s important to note, though, that the two questions I attributed to religious epistemology are prompted 10 by a long tradition of scepticism, not so much directed towards religious belief, but to religious faith. This tradition, which is sometimes called the problem of faith and reason, goes back to early antiquity (Helm 1999). It construes faith as the antithesis to reason, finding faith to be a matter of believing things of a religious nature in a way that is epistemically deficient. My attempt to bring the epistemology of testimony to bear on religious epistemology is not merely opportunistic; it actually falls out of an intuitively appealing account of the nature of religious faith. This theory holds that faith requires trusting God’s own testimony. Hence, the approach I take in this thesis of applying theories of testimony to religion both seeks to address the central questions in religious epistemology, and does so from the perspective of the problem of faith and reason, by tying testimony to a particular account of the nature of faith. This account, I will argue, has been defended right through history, from Antiquity in the work of Clement of Alexandria, in the Middle Ages with St. Thomas Aquinas, through to early modern philosophy, particularly in the work of John Locke, and more recently from several authors including Linda Zagzebski and Anthony Kenny. In several later chapters I will draw together the work of these various authors to explain the nature of this account of faith, and to defend its epistemic legitimacy. A second strand of research in this thesis explores a more radical approach to the problem of faith and reason – one that is fairly recent, but is currently very popular. According to this view, faith does not require belief at all. Someone could be an agnostic, but in virtue of other things, like her behaviour and the positive evaluation she makes of the object of religion, she may still have faith. This theory is sometimes called ‘non-doxastic faith’ (Bishop 2016). The response that the theory gives to the problem of faith and reason is quite unique. It says, in effect, that one’s faith can be epistemically rational even if one lacks sufficient epistemic reason for belief because one’s faith can track whatever epistemic reasons one has, even where one’s confidence falls below the level required for belief. Ordinarily we might think that in this kind of case, someone suffers a loss of faith. Here, the contention is that faith is still maintained, and moreover, it can be rationally held. Non-doxastic faith stands opposed to my divine testimony theory in an important respect, but is consistent with it in another. It is opposed in the sense that on the divine testimony theory I defend, faith in divine testimony requires belief in what God says. However, non-doxastic faith is consistent with the divine testimony theory in that, even though non-doxastic faith maintains that faith does not require belief, it’s 11 still the case that faith may be a matter of cognitively, though non-doxastically, endorsing divine testimony. In that case, this cognitive endorsement could be a matter of accepting, rather than believing the testimony. The stance I take in this thesis on non-doxastic faith is largely negative. I argue in detail that this theory is false, and in so doing, I represent one of the few voices opposing the theory amongst philosophers of religion.3 Indeed, non-doxastic faith has received widespread support from the most influential philosophers working on religion, including Richard Swinburne, Robert Audi, William Alston and Jonathan Kvanvig, to name but a few. Whilst I disagree with the central claim of this theory, I do believe that there is some merit to it. In particular, religious faith has too often been thought of as simply identical with religious belief. The value of non-doxastic theories of faith is that they have shown the importance of many other attitudes that are central to religious life. These can include acceptance, hope, trust, assent, plans, commitments, the emotions and various other affective attitudes. There is clearly value in exploring these other features of religious life. The point I want to make is that these cannot replace belief within faith, even though they form an important part of one’s general religious response. So, we have seen how this thesis is situated with respect to the problem of faith and reason and in the questions generally asked by religious epistemologists. For clarity, we could say that I am looking to address two questions that are each tied to a particular theory of faith:

(1) How can the divine testimony theory of faith utilise recent theories from the epistemology of testimony to show how such faith is epistemically justified?

(2) Is non-doxastic faith a defensible theory, and if so, does it offer an adequate response to the problem of faith and reason?

Since the theories in (1) and (2) are partly denials of one another (in the sense that divine testimony requires belief, whilst non-doxastic theory does not require belief), I will use the non-doxastic theory to provide a major line of objection to the theory I develop for divine testimony. If I can successfully show that non-doxastic faith is untenable, then that would remove one line of objection from the divine testimony

3 My opposition to this view is published in Malcolm and Scott (forthcoming) and Malcolm (forthcoming). These publications are some of the only criticisms of the theory in the literature at present (an exception is Mugg 2016), and are discussed at length in Chapter 3.

12 theory. It would also show that (2) is not an option for those who would like to respond to the problem of faith and reason by way of appeal to the non-doxastic theory.

0.2 Chapter Overviews

The purpose of this thesis is to evaluate two different ways in which we might respond to the problem of faith and reason (PFR) by exploring two different accounts of faith. My process for doing so is to begin by fully articulating the problem, then by defending one account of faith and rejecting another, and finally, by determining how the remaining account can overcome the PFR. In Chapter 1 (‘New Directions in Faith and Reason’) I situate the PFR in terms of my own account of epistemic justification. I outline a fairly typical foundationalist epistemology according to which non-basic beliefs must be supported by other justified beliefs to be justified themselves, whereas basic beliefs, which just are sources of evidence, are justified according to various other standards. From this account I frame the PFR in terms of a failure to meet the standards of epistemic justification, and outline, in depth, the divine testimony theory and non-doxastic theories of faith, and how they may each respond to this alleged failure In Chapter 2 (‘The Nature of Faith’) I defend a theory of religious faith I call Abrahamic faith according to which faith requires both trust and a propositional attitude. I begin by addressing the meta-philosophical challenge that faith cannot actually be analysed, and provide a method for analysing faith which overcomes this challenge. I then give an analysis of trust, and explore trust in the particular instance of trust in testimony, and use this to give an account of trust in divine testimony. I connect the discussion on trust with an account of relational faith, which I distinguish from propositional faith, and then offer an account of how the two kinds of faith are related in Abrahamic faith. I propose that relational faith is often the cause of propositional faith. I end with an account of propositional faith, which, I maintain, requires belief. In Chapter 3 (‘Non-Doxastic Faith’) I present the non-doxastic theory of faith as both an objection to my account of propositional faith, and as a response to the PFR in its own right. I begin by setting out the theory itself, before examining the four leading arguments in favour of it, each of which I find to be unpersuasive. I then level an objection against non-doxastic faith, which is that non-doxastic faith grants faith to

13 pretence positions such as fictionalism. After explaining how this objection works, I consider the best responses available to it, but find these to give a mere Pyrric victory. I conclude that propositional faith does require belief. If this is so then non-doxastic faith cannot be used as a way to respond to the PFR. In Chapter 4 (‘Divine Testimony’) I pick up where I left off in Chapter 2 by considering the prospects of the divine testimony account of faith. I begin by giving a closer analysis of the claim that faith involves believing God’s testimony, and find several sources in the historical literature that have defended a similar claim. I then move on to look at what it is for God to testify. Three accounts of divine discourse are derived from the work of Nicholas Wolterstorff, each of which have some merit and provide a basis, in the final chapter, for analysing the epistemic status of the beliefs formed from them. In Chapter 5 (‘Justifying Faith in Divine Testimony’) I take up the issues posed by the PFR once again by looking at recent theories in the epistemology of testimony, and applying these to the specific case of divine testimony. I analyse both reductive and non-reductive accounts of divine testimony, and discuss and reject the only theories in the current literature that have dealt with the issue from the perspective of recent work in the epistemology of testimony. I then dedicate significant time to evaluating the work of Alvin Plantinga. I argue that he can be interpreted as giving an account of faith as believing divine testimony, and I augment his theory with support from recent work in human testimony. The result is, I believe, a workable theory of the epistemology of faith in divine testimony that will stand up to scrutiny, marks an improvement on Plantinga’s prior work, and offers a plausible response to the PFR. In the Conclusion, I summarise the central points from the thesis, and make recommendations for where this research could be taken further.

14

1. NEW DIRECTIONS IN FAITH AND REASON

1.1 Overview

This chapter gives an articulation of one of the best known problems in the philosophy of religion – the problem of faith and reason (PFR). The PFR has been addressed at virtually every stage in the history of philosophy back to the classical period.4 I will argue that the PFR is an epistemological problem according to which faith fails to meet the standards required to be justified with respect to knowledge. I will explain how the PFR works by outlining an intuitive epistemological framework in which to situate it.5 Following this I will present two novel approaches to overcoming the problem. It will be the place of the remainder of this thesis to evaluate these two approaches, and so this chapter involves setting the scene within which the further work will take place. In §1.2 I will begin by giving a brief sketch of the predominant features of belief. Since belief is scrutinised and evaluated both in this chapter and throughout the thesis, I will begin by saying what belief is. Then I present an initial statement of the PFR by putting forward a thesis of epistemic justification according to which beliefs are justified only if they are based on good reason. The PFR can then be stated as an argument if we accept those accounts of faith that claim that faith involves believing without good reason. The nature and structure of epistemic justification is then given greater attention in §1.3 where I outline a kind of evidentialism, and in so doing offer an account of evidence and the sorts of things that might qualify as evidence, before adopting foundationalism as the final component to justification. The aim of this section is not so much to give a robust defence of a particular account of epistemic justification as it is to outline one within which to situate the PFR. Modern versions of the PFR are often given in similar terms to the version I present in this chapter.6 In §1.4 I consider the implications of the account of epistemic justification for the normativity of belief. Three distinct norms are presented and I discuss one method for weighing these norms in order to identify obligations on what one should believe.

4 An excellent collection of work is Helm (1999). 5 It is not my aim to offer an overview of what many others have said about the PFR – there are other good sources that achieve this, for instance, Swinburne (2005), Plantinga and Wolterstorff (1983) and Helm (2001). 6 See Plantinga (1983) for discussion of evidentialist approaches to justification.

15

Taking these implications into account, in §1.5 I offer an extended and reformulated version of the PFR which argues that faith is not justified, and hence it cannot lead to knowledge. This section concludes with two potential responses to the PFR to be explored and evaluated further in subsequent chapters.

1.2 The Problem of Faith and Reason

1.2.1 Belief

We hold many beliefs about all kinds of different things. For instance, we may believe that we are seeing, touching or smelling a particular object; that certain past events have or have not occurred, and particular future events will or won’t occur; that our friends have similar emotions and desires to us; and that the earth orbits a nearby star in our galaxy. When we talk about beliefs, we usually say that they have propositional content. Propositions are abstract entities that capture the content of our beliefs.7 We believe that p. Typically in language, the propositional content scoping within an attitude is indicated by ‘attitude+that’, followed by propositional content. I believe (attitude) that ‘the milk is in the fridge’ (proposition). Beliefs are thus one kind of what are standardly called propositional attitudes. There are several other attitudes we can take towards a proposition. We might fear that p, hope that p, desire that p, or know that p. Propositional attitudes can be contrasted with mental states that lack propositional content such as itches, pains, and moods.8 Our believing, or doxastic attitudes towards something stand in a relation of truth to that thing. So, believing that the sky is blue involves taking the proposition, the sky is blue, to be true. This is one kind of doxastic stance we can take toward something. Another is to believe that something is not true, or to disbelieve that thing. The third doxastic attitude we can have is to suspend judgment regarding something, which is to be agnostic regarding its truth-value, and thus to neither believe, nor disbelieve that it is the case.

7 For detailed explanation see Schroeder (2006). 8 Ratcliffe (2008) provides an interesting discussion of several of these kinds of non- propositional states.

16

On a standard and widely-held view, doxastic attitudes are mental states that we have.9 My belief that there is milk in the fridge, my disbelief that it’s 10.30am, and my suspension of judgment over whether England will win the cricket match are all mental states towards things that I think are, respectively, true, false, and of which I am undecided. I can be right or wrong with what I believe and what I disbelieve (though not with suspension of judgment). There may be no milk in the fridge, and it might actually be 10.30am. Given this, we can say that belief and disbelief are truth- conditional mental states. Attitudes that are truth-conditional are usually called ‘cognitive’. Truth-conditional cognitive states like belief can be contrasted with non- cognitive states that are non-truth-conditional. Desires, wants, wishes and plans are all kinds of non-cognitive states whose propositional content represents how we would like the world to be, rather than how we take it truly to be.10 In addition to beliefs, there are other kinds of mental states that are truth- conditional, and hence cognitive. Some of these attitudes include acceptance, assumption and assent. So, if I accept that it’s 10.30am, I needn’t actually believe this to be true, but what I accept will be true or false. Acceptance and other non-doxastic cognitive attitudes involve acting as though a certain proposition is true, even though we don’t believe it to be true. These attitudes will be given greater attention in Chapter 3. A requirement for determining whether beliefs are true or false is in terms of correspondence with what they represent. In this way we can say that a belief is true providing the propositional content is true. When someone believes the proposition, there is milk in the fridge, the belief will be true providing there actually is milk in the fridge. It will be false if the belief fails to correspond with what it represents.11 A metaphor sometimes used for construing belief as aiming to truly represent the world around us is in terms of ‘direction of fit’ (Smith 1994). Beliefs and other cognitive states are said to have a mind-to-world direction of fit such that it is incumbent upon them to represent the world as it truly is.12 In contrast, desires and

9 Alternative theories include that belief is a behavioural disposition (see Schwitzgebel 2002 and 2013), and that there are no beliefs (Churchland 1981; Stich 1983). 10 This distinction between beliefs and desires can be attributed to Hume (1978, Bk III, part I, §I, pp. 457-58; Bk II, part III, §III, pp. 415-16). 11 In §1.3.3 I will discuss further some accounts of correspondence. 12 We sometimes say that beliefs are directed towards the world. This is called ‘intentionality’ (Searle 1983).

17 non-cognitive states have a world-to-mind direction of fit, in that they represent the world the way we would like it to be.13 One of the most commonly agreed upon features of belief is that they are adopted involuntarily. It doesn’t seem like we choose to believe what we do; rather, our beliefs are formed in us through our experiences and internal reflections. For example, if I believe that London is the capital of England then presumably I have adopted this belief from my time looking at maps of England, and attending school where I was taught this fact. I can’t now choose to believe instead that Manchester is the capital of England. Attempting to directly do so at will would not be belief, but would be some form of pretence or acceptance. I can, however, indirectly choose to believe something by having experiences that will form that belief in me. Suppose I currently believe that the time is a quarter to twelve. I can’t directly choose to believe, instead, that it’s a quarter past twelve. However, I can look at a clock and if it says that it’s a quarter past twelve, rather than a quarter to twelve, then my beliefs will normally revise themselves and I will come to believe that it’s a quarter past twelve. I will have come to the new revised belief through indirectly voluntary means. It follows from this discussion that belief is a kind of mental state we find ourselves with, not an action we choose to engage in. As John Heil states it, a belief is ‘the result of things done, but it is not in itself a thing done’ (1984, p.60). One important difference between doxastic cognitive attitudes, like belief, and non-doxastic cognitive attitudes, like acceptance, is their voluntariness. Whilst a non- doxastic cognitive attitude will meet virtually all of the characteristics outlined in this section, they will not be involuntary. When someone accepts p, that person chooses to do so. Moreover, the need for someone to accept p is because that person lacks the belief that p. The person does not believe it to be the case that p, and herein belief is importantly distinguished from the range of non-doxastic cognitive attitudes. Beliefs also play an important role in guiding our practical decision making. For instance, according to a popular theory of human motivation and action, we are motivated to act on our desires and intentions, which work in conjunction with our

13 See Bricke (2000, chapter 2) for a useful elaboration on these metaphors, and a connection to the work of Hume.

18 beliefs to direct our behaviour.14 So, if we desire to get to the train station and hold beliefs about how to get there, then we will use these beliefs to guide our actions when making our way toward the train station. This can have important implications if, say, you are the leader of a country trying to decide whether to declare war on the basis of those propositions you believe and those you don’t. If what we believe guides our actions, then in most cases, we will need our beliefs to be accurate.

1.2.2 Evaluating Faith & Belief

How should we go about appraising our beliefs? One option is to ask whether, if we hold doxastic attitude D (be it belief, disbelief or suspension of judgment): what right do we have to hold D? Another way of putting this, which is standard practice in epistemology, is to ask: what justifies our holding of D? According to one broad definition, ‘justification is the right standing of an action, person, or attitude with respect to some standard of evaluation’ (Watson 2016). For instance, the actions of an individual can be justified under the law, or an organisation’s decisions can be unjustified against a moral standard. One kind of justification, or ‘right standing’ we can have for our beliefs is their evaluation with respect to knowledge. When a belief has a right standing to knowledge, it is standard practice to call this epistemic justification. One way to evaluate D is, then, to ask whether D is epistemically justified.15 We will see in §1.3.2 why this gives D right standing with respect to knowledge, but I should note before continuing that our beliefs can be evaluated against other standards too. Some of these methods of evaluation, in particular prudential justification, will be discussed in §1.4, but the problem I am outlining in this section only concerns the evaluation of our beliefs in epistemic terms.

14 This theory is sometimes called the ‘Humean theory of motivation’, since its most well- known formulation can be traced to Hume (1978, see note.10). Smith (1994) offers a defence of the theory, and Nottelmann (2011) provides a helpful overview. 15 The terms epistemic justification, epistemic rationality, and epistemic warrant are used interchangeably (often this varies by writer’s preference. Plantinga (2000) favours use of the latter two, but he has technical definitions for all three). Sometimes changes in the status of one of these has implications for the status of another. For instance, we might say that if D for person S is epistemically justified, then D is rational for S. I don’t doubt that what I will say about epistemic justification will be thought to have implications for rationality and warrant. I will tend to focus on justification though, for the sake of economy and simplicity. However, as a general rule I take justification, warrant and rationality to be broadly equivalent.

19

We have now specified one means of evaluating our doxastic attitudes – in terms of epistemic justification. But what justifies a belief in this way? One intuitive answer is that I am justified in holding D providing I have a good reason for holding D. We seem to demand this of people all the time when they say claim to know something. For instance, if we are catching a flight together next week and you tell me you believe we need to be at the airport at 2pm, then I will want to know what reason you have for your belief. If you tell me that you just checked the tickets and it instructs us to be there at 2pm then I will be happy and take you to have a good reason – I will think you have justification for your belief, and this is the sort of reason required of you to have knowledge. So, you having a good reason for your belief is sufficient for you being justified in your belief. However, if you tell me you’re just guessing, then I will take you have a bad reason – I will think you lack justification for your belief. So, it seems like having a good reason for your belief is necessary for you being justified in your belief with respect to knowledge. We can give an initial formulation of this thesis as follows:

(EJ): A person S has epistemic justification for holding doxastic attitude D if and only if S has a good reason to hold D.

At present, EJ is not explicit enough to account for epistemic justification since we haven’t yet given an account of what it is to have a ‘good reason’. This will be addressed in §1.3.2. It does, for now though, provide us with a means for outlining the problem to be addressed in this thesis, so at this point I want to use what has been said so far about belief to give a statement of this problem. We can begin to see our problem by noting that the following statement is entailed by EJ:

(EJe) If a person S does not have a good reason to hold D, then S is not epistemically justified in holding D.

What implication does (EJe) have for faith – particularly faith towards religious claims? Well, according to some authors, religious faith involves holding beliefs without good reason. Consider the following claim made by philosopher David Hume:

Our most holy religion is founded on Faith, not on reason; and it is a sure method of exposing it to put it to such a trial as it is, by no means, fitted to endure. (Hume 1975, p.130/§100)

20

On one interpretation of Hume’s statement, by ‘Our most holy religion’, he means the beliefs that religious people hold that are taught by a particular religious institution, and in his context, that religion was Christianity (though his claims here will fit other religions too). For instance, religious people tend to believe propositions such as, ‘God exists’, ‘God is all loving’, ‘miracles are performed today’, and ‘Jesus was God incarnate’. Of these beliefs, Hume claims that they are ‘founded on Faith’, which we might read as believed on faith. So, according to Hume, religious people hold religious beliefs on faith. He contrasts those beliefs that are formed in this way from those that are based on ‘reason’. We don’t have a statement yet for what it is to hold beliefs on good reason, but let’s presume that Hume here means something similar to what I have been talking about, namely, that being based on good reason gives a belief a positive evaluation in epistemic terms. We can summarise Hume’s account of faith as follows:

(F) Faith involves holding doxastic attitude D without good reason. 16

Now we are in a position to see the problem for the religious believer. If someone with religious faith accepts both Hume’s account of faith (F) and (EJe), then it looks like her beliefs will lack epistemic justification. Taking these together, we get the following initial formulation of the PFR:

(EJe) If a person S does not have a good reason to hold doxastic attitude D, then S is not epistemically justified in holding D. (F) Religious faith involves holding doxastic attitude D without good reason. (C) Therefore, doxastic attitude D held on religious faith is not epistemically justified.

The statement given in the conclusion (C) applies to all doxastic attitudes held on religious faith, and hence every instance of faith that involves belief, whether the belief

16 Hume’s account is echoed in contemporary writings as well. For instance, Steven Pinker has recently defined faith as ‘believing something without good reasons to do so’ (2006). Brian Leiter (2012) says that ‘Religious beliefs, in virtue of being based on ‘faith,’ are insulated from ordinary standards of evidence and rational justification, the ones we employ in both common sense and in science’ (pp.33–34), and that ‘religious belief in the post- Enlightenment era involves culpable failures of epistemic warrant’ (p.82). Another example is Walter Kaufman: ‘Faith means intense, usually confident, belief that is not based on evidence sufficient to command assent from every reasonable person (1978, p.2). These views are also widespread in the popular writings of the ‘New Atheists’.

21 has religious content or not, will be epistemically unjustified. This initial formulation of the problem leaves a number of questions open that I will seek to answer in the following sections. First, what is it to have a good reason for D? Second, what is so good about having epistemic justification – what makes this property of belief worth obtaining? Third, why must we accept (F)? Are there alternative accounts, or must faith involve holding beliefs without good reason? In what follows of this chapter, I will answer these questions, and in so doing, formulate a further, and much clearer version of the PFR. Let’s take the first and second questions together, and thus begin with epistemic justification.

1.3 Epistemic Justification

1.3.1 The Value of Epistemically Justified Beliefs

In section §1.2.1, I proposed that beliefs play an important role in our practical decision making. Beliefs are hugely significant in the decision making that we undertake on a daily basis, and in some cases, our having true beliefs can be a serious matter. For example, doctors rely on their own beliefs, and the beliefs of their colleagues, when making critical decisions about patient treatments. Even in the day- to-day decisions all people make, we want our beliefs to be true since they will affect whether or not we’re on time for important meetings, who we decide to trust with what we value most, and even the kinds of foods we need to eat to keep healthy. Because our beliefs play such a critical role in our practical reasoning, they are valuable if they are accurate: D is valuable in our practical reasoning if D is true. Granted this claim, we are then able to draw the connection between truth and epistemic justification. According to EJ, if a belief is epistemically justified then it is valuable. We just said that a belief is valuable if it is true. It follows from this claim and EJ that what makes epistemically justified beliefs valuable is in their connection to truth. Let’s add further that epistemically justified beliefs are held with a ‘good reason’, from which this argument suggests that to hold D with good reason increases the likelihood that D is true. This way of understanding the role of epistemic justification is not new. It has been proposed by several others before, including in the work of Lawrence Bonjour:

The basic role of justification is that of a means to truth…We cannot…bring it about directly that our beliefs are true, but we can presumably bring it about

22

directly…that they are epistemically justified. And, if our standards of epistemic justification are appropriately chosen, bringing it about that our beliefs are epistemically justified will also tend to bring it about…that they are true. If epistemic justification were not conducive to truth in this way…then epistemic justification would be irrelevant to our main cognitive goal and of dubious worth. Epistemic justification is therefore in the final analysis only an instrumental value, not an intrinsic one…Any degree of epistemic justification, however small, must increase to a commensurate degree the chances that the belief in question is true…for otherwise it cannot qualify as epistemic justification at all. (Bonjour 1985, pp.7-8)17

Bonjour is saying here that the greater degree of epistemic justification a belief has, the more likely it is that the belief will be true.18 Supposing I am right that a belief has value in our practical decision making when that belief is true. Well, on the basis of what Bonjour has said, the more epistemic justification we have for our beliefs, the more of this kind of value the beliefs will have. An epistemically justified belief will be of greater value to us than one that is not justified this way because it is more likely to be true and we require true beliefs to make the best decisions when it comes to our practical reasoning. Several questions present themselves to us at the point. Why should it be the case that epistemically justified beliefs are more likely to be true? What implications does this discussion have for what we ought to believe? Can our beliefs have value even if they are not epistemically justified? The ensuing sections will give a thorough account of epistemic justification, and in so doing will offer answers to each of these questions.

1.3.2 Evidentialism

To begin, let’s remind ourselves that in accordance with EJ, S is epistemically justified in holding D providing S has a good reason to hold D. Now, this notion of a ‘good reason’ was derived from intuition concerning my example of two individuals catching a flight together. In one case, the person read the tickets and this was

17 This way of articulating epistemic justification is endorsed by Goldman (1989), and defended in Steglich-Petersen (2013). Some interesting alternatives – though none of which I wish to endorse – are summarised in Watson (2016). 18 For more on the relation between epistemic justification and truth see Littlejohn (2012).

23 considered a good reason for holding his belief. In the contrasting case he merely guessed and this was thought to be a bad reason to hold it. Now, in the former case, the individual has some evidence for his belief – namely, that he has looked at the ticket. In the latter case, he has no evidence. This example suggests that it is built into our concept of what makes a good reason for belief that a belief is based on evidence. Let’s broaden the scope of the examples a little to see just how common-sense this view is, and how it connects to knowledge. Suppose you are considering how to vote in an upcoming election and you want to know which party has the most pro- environmental policy. What should you do? Imagine a food company who are planning to include a particular preservative in their new product and want to know what health risks or benefits it carries. How should they find this out? Consider two friends who are discussing the quickest driving route on holiday and want to know for sure which way is quickest. What would be the best means of getting this information? Perhaps the most plausible answer to each question is this: look at the evidence. For all three cases, the evidence will help them get to the truth – it is the presence of evidence that justifies a belief, and leads to an increase in likelihood that the belief will be true. The evidence will assist each person in each example to form beliefs that are more likely to be true. We can use these examples as a basis for constructing an argument in favour of the view that epistemic justification requires evidence – a thesis typically referred to as ‘evidentialism’:19

Argument 1 (A1): (1) Doxastic attitude D is epistemically justified if and only if D is held for a good reason (EJ); (2) If D is based on evidence then D is held for a good reason (from the examples); (3) Therefore, if doxastic attitude D is epistemically justified then D is based on evidence.

A1 is one of the most straightforward ways to show the conceptual connection between epistemic justification and evidence. To make things clearer, I will need to say more

19 The following account is given by Conee and Feldman (2004, p.1): ‘evidentialism is a view about the conditions under which a person is epistemically justified in having some doxastic attitude toward a proposition. It holds that this sort of epistemic fact is determined entirely by the person's evidence.’

24 about the nature of evidence, but before I do, I want to present a few further arguments in favour of evidentialism. The foregoing examples were not only being used to show the connection between justification and evidence, but between increased evidence and increased likelihood of truth. In the previous section, we saw that Bonjour and others claimed that epistemic justification is a means of increasing the likelihood of our beliefs being true. But if increased evidence is what delivers on this demand, then it is evidence that makes a belief epistemically justified. We can see this argument as follows:

Argument 2 (A2): (4) If D is epistemically justified, then D is more likely to be true; (5) If D is more likely to be true then D is based on evidence; (6) Therefore, if D is epistemically justified then D is based on evidence.

There are two further arguments for evidentialism that rely on a conceptual connection between evidence and knowledge. The examples at the beginning of this section all specify a subject who is not only seeking to form a belief, but wants to gain knowledge. Now, a traditional statement of the necessary conditions for propositional knowledge is that knowledge requires a justified true belief.20 As the three examples suggest, for one to know p, one must obtain evidence in favour of p. If a person S’s evidence E is sufficient for p, then it would follow that, providing one believes p on the basis of E, and it is true that p, then S knows p.

Argument 3 (A3): (7) Epistemically justified true belief is necessary for knowledge; (8) Knowledge requires having evidence; (9) Therefore, epistemically justified true belief requires having evidence.21

20 The famous counterexample to a necessary and sufficient condition reading of this analysis is given by Gettier (1963). I assume, along with Kvanvig (2003), that these conditions are at least necessary, but that knowledge will require either a further, fourth condition, or a specific statement of the current conditions, before the analysis is both necessary and sufficient – one that will ‘un-Gettier’ the analysis. 21 A variant on this argument that focuses on the normative features of belief can be found in Dougherty (2011, p.2).

25

This argument provides a further reason for why epistemic justification has value – namely, because it is required for knowledge.22 Importantly though, it also demonstrates the connection between epistemic justification and evidence. According to (9), wherever one has an epistemically justified true belief, one must have evidence for that belief. But this does not result from the truth of the belief, but from the belief’s justification. I can form a true belief that the preservative I want to include in the new food brand will not harm our customers, and do so by simply guessing, but in this case I haven’t consulted any evidence. So, evidence can come apart from true belief. Once I look through the academic research on the impact of the preservative on health, then at that point I will have consulted the evidence. If I believe on this basis that there will be no harm done to our customers, and this belief is true, then I could be said to meet the conditions for EJ – I have a good reason for my belief. This establishes the connection between epistemic justification and evidence: it is the possession of evidence that grants epistemic justification to D – it is the evidence that fills out what it is to possess a ‘good reason’ for D. The final argument makes the same point as the previous, but by using reductio ad absurdum:

Argument 4 (A4):

(10) Epistemically justified true belief is required for knowledge; (11) If epistemically justified true belief does not require evidence, then one can have knowledge even when the evidence entirely opposes one’s belief; (12) Epistemically justified true belief does not require evidence (assumption for reductio); (13) Therefore, one can have knowledge even when the evidence entirely opposes one’s belief.

According to (13), I can consult all of the evidence that there is or will ever be available for p, and even when all this evidence points to ~p, and if I continue to believe p, I can still know p. So, I can commission dozens of clinical trials to show the effect of my preservative on health, all of which show that it causes health problems, and yet,

22 Although see Kvanvig (2003; 2010) for an important problem with this: the ‘swamping problem’.

26 believe that it does not cause health problems, and somehow know that it does not cause health problems. This would seem to be an absurd result. The conclusion that I wish to draw from A1-A4 is simply that epistemic justification requires evidence – that, for D to be epistemically justified, D must be based on evidence.23 So, epistemic justification is understood in terms of evidence, and the possession of evidence increases the likelihood of a belief being true, thus, epistemic justification increases the likelihood of a belief being true. We can now give a new formulation of EJ in which a ‘good reason’ is understood in terms of possession of evidence:

(EJ): A person S has epistemic justification for holding doxastic attitude D if and only if D fits the evidence S possesses for D.

This has quite a particular knock on effect for the PFR. In §1.2.2 I restated a claim made by Hume which enabled us to formulate this problem. Earlier in that very work Hume issues the following dictum: ‘A wise man, therefore, proportions his belief to the evidence’ (1975, p.110/§87). If proportioning one’s beliefs to the evidence is part of basing them on reason for Hume, then faith, which is not based on reason, would not proportion beliefs to the evidence. Furthermore, if epistemic justification is measured in terms of evidence then faith is not epistemically justified. As we just saw, if a belief is not epistemically justified, then there is no guarantee that it is any more likely to be true than a belief based on guesswork. For a belief to lack credibility in this respect of likelihood to being true, then perhaps we ought not believe it – it would be unwise to do so, given the importance of belief to practical reasoning. It is these issues that are at the heart of the PFR, and a revised version of the problem that takes these reflections into account will be given in §1.5. Although we have clarified that good reasons for a belief are evidential, we still haven’t said what evidence is. What do we mean when we talk about evidence?

1.3.3 Evidence

23 This is stated explicitly in Kim (2008, p.543): ‘the concept of evidence is inseparable from that of justification. When we talk of ‘evidence’ in an epistemological sense we are talking about justification: one thing is ‘evidence’ for another just in case the first tends to enhance the reasonableness or justification of the second’.

27

According to one definition of evidence offered by Benjamin McMyler (2011, p.133), we can ‘construe evidence broadly to encompass anything that genuinely counts in favor of the truth of a proposition’. On this account, then, evidence for p counts in favour of the truth of p. This definition seems to be compatible with examples in which we look for evidence to support a proposition. Returning to my earlier examples, a ticket issued by a particular airline which states that my flight departs at 2pm is evidence for the proposition that my flight departs at 2pm. If it’s evidence, then it also counts in favour of the truth of that proposition, presumably in this case because it’s been issued with some authority. A map shows that the route from Manchester to London is longer down the M1 than the M6 and M40, and this counts as evidence for the proposition that the route from Manchester to London is longer down the M1 than the M6 and M40. It counts in favour of the truth of that proposition, presumably because maps are generally reliable. Let’s adopt McMyler’s definition then as a formal statement of the nature of evidence:

(E): Evidence E is that which counts in favour of the truth of a proposition p.

On this definition, we can draw a contrast between evidence, on one hand, and misleading evidence on another. The former will count in favour of the truth of a proposition, whilst the latter may count in favour of a proposition, but not its truth. Hence, misleading evidence would not be considered evidence according to (E). Of course the mere existence of objective evidence for the truth of a proposition p will not suffice to justify one’s belief in p. Rather, one must be in possession of this evidence. That is precisely what EJ states – that when and only when one is in possession of evidence for p is D justified. The flight ticket is evidence for my flight time, but if I hold a belief about my flight time on guesswork, the belief won’t be based on the evidence. I must base the belief on the evidence for it to be justified. As such, evidentialism can be thought of as ‘a supervenience thesis according to which facts about whether or not a person is justified in believing a proposition supervene on facts describing the evidence that the person has’ (Conee and Feldman 2004, p.1). When person S has evidence E for p, and holds D towards p on the basis of E, then S is justified in holding D. So, the justification of D for S supervenes on the evidence S possesses for D and bases D on. If either D was not held on the basis of E (even if S was in possession of E), or S does not possess E and yet holds D, then S would not be

28 justified in holding D. In either case the supervenience base for D’s justification has come away from the evidence. Supposing I am in possession of some evidence for my belief. What sort of thing am I then in possession of? If we look at some examples of the sorts of things we ordinarily count as an evidential basis for a belief, we can get an answer to this question.

Case 1 (perceptual):24 Bob wants to know if it is warm outside to see if he should wear a jumper or not. He steps outside and it feels warm to him, and so he comes to believe that it is warm outside. He takes this belief as evidence upon which to base his decision not to wear a jumper. The evidence in this case is a particular kind a mental state – a perceptual belief – and specifically, a feeling of warmth. We often use sensory perception as a source for justifying our beliefs, and the same kind of example could be run for, say, if we want to know whether it is sunny or cloudy outside, or if it is raining. Perceptual beliefs are formed through any of our senses, and these beliefs provide us with evidence upon which we can justify other beliefs. This kind of evidence is a posteriori.

Case 2 (testimony): Mary wants to know if the bank is open after 5pm so she asks the cashier in the bank and he says that the bank is open after 5pm. She believes the cashier, and takes this belief as evidence that the bank is open after 5pm. The evidence Mary has for the proposition in question just is her belief acquired from testimony.25 We often form our beliefs on the testimony of others, and look to determine the truth- value of a proposition on the basis of these beliefs. In fact, according to some (Coady 1992), testimonially acquired belief is ubiquitous. This kind of evidence is also a posteriori.

Case 3 (memory): John wants to know if he had porridge for breakfast this morning. He introspects for a moment, recalls that he had porridge for breakfast this morning,

24 An epistemology which regards sensory beliefs as necessary and sufficient for knowledge would be strictly empirical (‘strictly’ in the sense that only perceptual beliefs can count as evidence). A recent defence of such a project is found in Bonjour (2003). 25 This view of testimony as a source of evidence presupposes a popular theory in the epistemology of testimony known as non-reductionism. A converse theory, in which our justification for testimonially acquired beliefs ultimately reduces to – is inferred from – other kinds of basic justification is called reductionism. A detailed discussion of these two theories will be covered in chapter 5. For a clearly articulated discussion of the two theories, see Lackey (2011).

29 and takes this as evidence that he had porridge for breakfast this morning. As with cases 1 and 2, we have a mental state – a memory belief – that plays the role of evidence in justifying the proposition. Again, we have another form of a posteriori belief.

Case 4 (a priori intuition): Sarah wants to know if Mark is taller than Jane. She knows that Mark is taller than Harry, and that Harry is taller than Jane. When she reflects on this, it immediately appears to her as though Mark is taller than Jane. Now Sarah wants to know if 2+1=3. She looks at the sum 2+1 and it immediately appears to her that 2+1=3. Finally, Sarah wonders whether something can be both black and white all over. From reflection, it immediately appears to her that something cannot be both black and white all over. In all three cases, Sarah forms a belief on what we might call rational or a priori intuition. These intuitions are evidence for the subsequent beliefs that she holds. This intuition is immediate, and gained a priori.

All four cases provide examples of a mental state – a belief – that a subject possesses at a certain time which we take as evidence for a certain proposition.26 This view, according to which it is our mental states that are our evidence is sometimes called ‘mentalism’.27 This position is clearly summarised by Mittag:

Evidence for or against p is, roughly, any information relevant to the truth or falsity of p…[O]nly facts that one has are relevant to determining what one is justified in believing, and in order for one to have something in the relevant sense, one has to be aware of, to know about, or to, in some sense, “mentally possess” it. The sort of evidence the evidentialist is interested in, therefore, is restricted to mental entities (or, roughly, to mental “information”). (Mittag 2004, Page number unavailable)

There are a number of points here that are highly relevant to what has already been said. First, evidence for p counts in favour of the truth of p – which is consistent with (E) – and evidence against p counts in favour of the falsity of p – which is, roughly, the converse of (E). Second, for D to be justified for S, S must be in possession of

26 These four kinds of belief do not represent all the forms of evidence we can have, but they are perhaps the most common, and will be sufficient for my purposes in this thesis. 27 Conee and Feldman (2001) argue for this theory in relation to epistemic internalism. It is central to their theory of evidentialism as well: ‘As to what constitutes evidence, it seems clear that this includes both beliefs and sensory states such as feeling very warm and having the visual experience of seeing blue’ (Conee and Feldman 1985, n.2, p.32). Mentalism is also defended in Swinburne (2011) and Turri (2009). For an overview see Sylvan (2016).

30 evidence for D. And partially following from this, third, to be in possession of evidence is a matter of holding certain beliefs and other mental and sensory states – it is these states that are our evidence and justify our beliefs in certain propositions. These observations are especially relevant for succinctly formulating a statement of epistemic justification. What we see is that a belief’s status as epistemically justified is dependent on the evidence base supporting it at a particular time – something observed earlier with respect to evidentialism as a supervenience thesis. If we no longer base a belief on evidence, or the evidence falls away from a belief in some way – perhaps because it has been forgotten – then a belief loses its justified status. One further point is that, according to the most recent formulation of EJ, our beliefs must fit the evidence in some way. One way of articulating what it means for a belief to fit the evidence is in terms of that belief matching the evidence. By way of analogy, we might think that a picture matches that which it pictures (Dougherty 2011, p.9). In a similar way, a belief that the day is warm outside matches our perceptual beliefs that it feels warm outside. Our belief that the sky is cloudy matches our visual experience of grey when we look up. And this interpretation of fittingness works with other kinds of evidence too. So, our belief that 20x37=740 matches our evidence (if the following is indeed evidence for us) that 10x37=370 and 370x2=740. This is, then, one way to interpret what it means for a belief to fit with the evidence. An alternative way, as suggested by Hume’s remarks, is to believe proportionally to the evidence.28 Here we get the notion of a belief, having as one of its properties, a credence – a precise degree to which a proposition is believed.29 In this sense, the credence we assign to a proposition should fit with the evidence that we have for it. In some instances, as with the proposition it is cloudy, which is believed because we have a clear visual perception of greyness above us, it would fit our evidence to assign a credence of 1 (or almost 1) to that proposition. Some propositions do not warrant such support. A scientist in the early stages of research might only have

28 The view is articulated by John Locke: ‘The mind, if it will proceed rationally, ought to examine all the grounds of probability, and see how they make more or less, for or against any probable proposition, before it assents to or dissents from it, and upon a due balancing of the whole, reject or receive it, with a more or less firm assent, proportionably to the preponderance of the greater grounds of probability on one side or the other.’ (1924, Bk IV, Ch 15, §5) 29 Since Ramsey (1926), it has been thought that credences can be measured by betting behaviour. See Huber (2009) and Buchak (2014b) for discussions.

31 suggestive evidence that points to the truth of a hypothesis, and it might then only fit the evidence she has to assign a credence of .1 or .2 to that hypothesis. Likewise, we might hold beliefs about indeterminate propositions, such as future events. The belief that tomorrow it will rain might be held with credence .8 because you live in a particular part of the world with plenty of rain. However, a weather forecast predicting sunshine all day could give you reason to alter your credence to .4. In effect, the forecasters are providing testimonial evidence that conflicts with your previous inductive evidence, and fitting your belief to these conflicting sources is a matter of balancing the relative likelihoods of the future event occurring on the basis of these sources. These two theories of fittingness – matching and proportionality – are not mutually incompatible. It seems possible to hold both theories together, and even if they aren’t, I don’t here want to offer reasons for deciding between them. They merely give a few plausible ways for conceiving what it is for a belief to fit one’s evidence. With all of this said, we are in a position to offer a further statement of epistemic justification:

(EJ): Doxastic attitude D toward proposition p is epistemically justified for person S at time t if and only if having D toward p fits the evidence S has for p at t.30

There is one clear, and extremely well-known problem facing this account of epistemic justification. Consider the following argument. According to EJ, only evidence justifies our belief toward a proposition. But, according to mentalism about evidence, our evidence consists in a set of beliefs and sensory states. It follows that, either these ‘evidential-belief’ states are justified by other sources of evidence, or that our evidential-belief states are not justified by other sources of evidence. However, this presents a dilemma since, on the first disjunct, even if our evidential-belief states can be justified by other sources of evidence, those justifying sources of evidence will also be evidential-belief states which in turn will require justification from other sources of evidence. This will create a regressive justification chain that goes on ad infinitum.

30 With a few minor tweaks, this is the same thesis defended by Conee and Feldman (1985, p.15), which is arguably the most widely favoured thesis of evidentialism about epistemic justification.

32

This is one version of what is often called ‘the regress problem’.31 Without an end to the regress, no beliefs will be justified. Now, the second disjunct is equally problematic since, if our evidential-belief states are not themselves justified, then how can we reasonably expect them to confer justification to those beliefs that depend on them? This would make EJ indefensible since we would have no good reason for thinking that any belief that fits the evidence for a proposition is any more likely to be true, simply by virtue of being based on evidence, since we have no good reason for thinking the evidence itself is reliable. Evidentialism can escape through the horns of this dilemma by endorsing a theory that I will briefly introduce next: foundationalism. As we will see, endorsing foundationalism as a way out of the regress problem comes at a cost to the evidentialist. Let’s first see how it can supplement our current theory of epistemic justification.

1.3.4 Foundationalism

To address the dilemma at hand, we need to begin by thinking about how our different beliefs are related in such a way that they justify other beliefs that we hold. Suppose, for example, that working away in my room one evening I hear a noise outside – a light pattering on the ground. I immediately form a perceptual belief that ‘there is a pattering noise outside’. Let’s call this perceptual belief P1. Now, I remember distinctly that ‘If I hear a pattering noise outside, it means that it is raining outside’.

Call this memory belief P2. From these two beliefs, I form a third belief, inferred from them that ‘it is raining outside’ – P3.32 Now, P1 and P2 are my evidence for P3. In that sense I infer the latter belief from the two former beliefs, and am justified in believing

P3 on that basis according to EJ. Continuing this example, let’s presume that I hold a third evidential-belief – this time one derived from the testimony of my spouse – whose propositional content is that ‘the washing is hanging up outside’ – P4. This belief, when taken together with my inferred belief, P3, justifies my believing a fifth

31 This argument is used as the main motivation for adopting foundationalism about justification. It traces back as far as Aristotle’s Posterior Analytics (1994, Bk I and II). For some discussion of the role of this argument and how foundationalism overcomes it, see Sosa (1980) and Audi (2003). 32 For a lengthy treatment of the role propositional content plays in reasoning, see Brandom (2000).

33 proposition, P5, that ‘it is raining on the washing that is hanging up outside’. And now let’s add to these five beliefs a sixth, inductive belief, P6, that ‘when it rains on the washing, the washing gets wet’, and finally a seventh, P7, inferred from P6 and P5, that ‘the washing outside will get wet’. The following diagram represents the structure of the justification for these beliefs:

P7

P5

P3

P1 P2 P4 P6

With evidential-beliefs P1, P2, P4 and P6 at the bottom of the justification structure, we have the architecture to the justification of my beliefs.33

According to EJ, P3, P5 and P7 are epistemically justified since they fit some evidence that is held at a particular time, and are based on that evidence. Worth noting though is that P5 and P7 are not only based on evidential-beliefs, but are also inferred from non-evidential beliefs as well. However, this presents no epistemic problem since those beliefs themselves are based on evidence, and so their status is epistemically justified. As such, epistemic justification is transmitted up the chain from evidential and non-evidential beliefs, providing the latter are justified by an evidential belief (in the way that P3 is), or the belief they are inferred from is justified by an evidential belief (in the way that P5 is), and so on. So, the structure of the justification for beliefs

P3, P5 and P7 is inferential or indirect, being transferred from other beliefs. And this

33 With foundationalism, the pyramidal shape is usually the converse, with basic beliefs fewer than inferential beliefs, but if we continued on the example, then presumably there would be less basic beliefs, and more inferential beliefs, giving the pyramid its traditional upside-down structure.

34 model example can be extended to show innumerable other beliefs, including those that are only formed on the basis of inferential beliefs. From this position we can clearly see the force of the dilemma presented at the end of the previous section. According to EJ, for beliefs P1, P2, P4 and P6 to be justified, they must be based on evidence. The existing structure of justification represented above does not base any of those beliefs on evidence, and so they are not justified. Even if they could be based on evidence, then that evidence itself will need to be justified. Suppose, for instance, we wanted to base P1 on a prior evidential-belief,

P0, whose propositional content could be something like, ‘I hear pattering’. If P0 were then taken as evidence for P1 then P0 would also need to fit some piece of evidence, and so the regress ensues. In order to address this issue, many philosophers endorse foundationalism about justification. A fairly simple theory, foundationalism claims that all inferentially justified beliefs rest on a foundation of non-inferentially justified beliefs. The standard terminology is to call non-inferential beliefs basic and inferential beliefs non-basic. The dispute over foundationalism usually centres on which beliefs lay at the foundation. According to some ‘strong’ approaches to the theory, only those beliefs that are indubitable can have the status of basic.34 It is much more common now, though, to endorse a moderate foundationalism according to which beliefs can be fallible and dubitable.35 As a definition, we can take modest foundationalism to be the claim that

…the basic beliefs possess knowledge adequate justification even though these beliefs may be fallible, corrigible, or dubitable. A corollary to modest foundationalism is the thesis that the basic beliefs can serve as premises for additional beliefs. The picture then the modest foundationalist offers us is that of knowledge (and justification) as resting on a foundation of propositions whose positive epistemic status is sufficient to infer other beliefs but whose positive status may be undermined by further information. (Poston 2016, page number unavailable)

34 Descartes (1996) is usually associated with this position, who restricted as basic belief in what he regards as indubitable propositions like ‘I am thinking’. 35 For a defence of this position, and a rejection of strong foundationalism, see Alston (1976a and 1976b), Williamson’s (2000) ‘anti-luminosity argument’, and Plantinga (1983).

35

If we endorse foundationalism as part of a theory of epistemic justification, we can see that a number of the sources of evidential-beliefs appealed to thus far will be appropriate as basic beliefs. Consider perceptual beliefs, for instance. They are sources of evidence for other, non-basic beliefs, and are generally a reliable basis for knowledge. Moreover, we don’t tend to infer them from any other beliefs. But nevertheless, our perceptual faculties are fallible, and so they can and sometimes do generate false beliefs. This is not problematic for modest foundationalism though since it does not require that our basic beliefs be infallible. Importantly, adopting foundationalism solves the regress problem according to which we face a dilemma that either our evidential-belief states are justified by other sources of evidence, or that our evidential-belief states are not justified by other sources of evidence. We needn’t take hold of either horn since evidential-belief states are basic, non-inferential foundational beliefs which are justified in some different way, and so do not need to be justified by other sources of evidence. This is widely held by modest foundationalists for perception, memory and a priori intuition. As noted earlier, a number of philosophers also endorse this view for testimony, but this will be given greater discussion in Chapter 5. By adopting foundationalism we see that EJ is actually an evidentialist thesis about epistemic justification for inferential, non- basic beliefs:

(EJ): Inferential (non-basic) doxastic attitude D toward proposition p is epistemically justified for person S at time t if and only if having D toward p fits the evidence S has for p at t.

With this revised definition of epistemic justification, we have a way of avoiding the regress problem, and a thesis that fits neatly into a broader foundationalist epistemology. I mentioned at the end of the previous section that endorsing foundationalism to avoid the regress problem comes at a cost for the evidentialist. She is left without an explanation for why basic, non-inferred beliefs are justified. Why should we think such beliefs can sit at the base of our epistemic architecture? EJ as it currently stands offers no answer to this question. A full account of epistemic justification requires an explanation of this issue, and ideally, the explanation will merge with the given evidentialist account. In the next section, I will propose several theoretical options for how the evidentialist can accomplish this.

36

1.3.5 Non-Inferential Justification

One way we can see clearly the problem facing the evidentialist is by revisiting the preliminary intuitions we had about the nature of justification. We found that justification appears to require believing for a good reason. EJ gives this reason for non-basic beliefs: they are justified when they fit the evidence. The possession of evidence is a good reason for belief. The question arises, then, as to what good reason someone might have for holding a basic belief. The foundationalist can’t simply respond that she is justified in holding her basic beliefs for no reason at all, or else she faces a charge of arbitrariness.36 A belief that is held arbitrarily would by definition not be based on any reason at all, let alone one we would consider good.37 The way around this problem is to give some reason as to why basic beliefs are justified, even though this justification will be defeasible. There are several theories that have attempted to do just this, by proposing some property P that can belong to a basic doxastic attitude D, whereby if and only if D has P, then D is non- inferentially justified. Although I won’t take a principled stance on which theory I consider to be most accurate, I will briefly outline three plausible candidates for what P might be, under the assumption that it is possible, if not likely, that there is some property P that can justify D.38 The first option is a theory that has come to be known as phenomenal conservatism (PC). According to PC, if it seems to S that p, then, in the absence of defeaters, S thereby has at least some justification for believing that p (see Huemer 2007, p.30). It seems to me that the day is warm, and unless I have some defeater for this seeming, then I have justification for believing the day is warm. The notion of defeat will be given greater attention in Chapter 5, but suffice it to say for now that a

36 For an objection to this effect, see Klein (1999, 2000 and 2004). The arbitrariness objection goes further by arguing that if the foundationalist does give a reason, she is committed to infinitism about justification since we enter back into the infinite regress of reasons for belief. For responses to this objection in particular (rather than the general issue of arbitrariness), see Bergmann (2004), Howard-Snyder (2005), and Howard-Snyder and Coffman (2006). 37 This sort of concern may have been what led epistemologists away from the notion of basic beliefs as self-justifying; as not requiring anything else to justify them. For instance, Chisholm (1977) proposed that we think of a ‘directly evident’ state, such a perception, as that which ‘constitutes its own evidence’ (p.25). 38 A range of plausible candidates is outlined and briefly assessed in Goldman (2008).

37 defeater condition in an account of justification permits that one’s justification can be undercut by rival considerations. I might have the flu, which would defeat my seeming that the day is warm. Despite the claim that seemings are defeasible in this way, advocates of PC will claim that there are many instances in which a proposition seems or appears true to us, and this seeming is the cause of our belief in the proposition. Given this, PC maintains that it would be overly sceptical to claim that seemings do not give justification if one has no defeaters for it. Thinking back to the basic beliefs discussed in §1.3.3, in each case a proposition was believed because it seemed true, and there was no defeater cited that counted against any of the seemings. There have been several arguments given in favour of PC, and a number raised against it.39 However, for my purposes I want to put these aside, and simply ask whether PC gives us a solution to the arbitrariness problem by positing seemings as the property that can justify basic beliefs. One potential concern is that PC appears to be committed to the view that seemings are basic beliefs from which other beliefs are inferred. If that were true, we would need some non-inferential justification for seemings themselves, whereas PC is a theory that attempts to use seemings as a justification for basic beliefs. However, PC needn’t be committed to such a view. The mere fact that it seems to S that p does not entail that S believes that p, even though S may hold the higher-order belief that she believes that it seems to her that p. The seeming itself is non-doxastic and merely offers a ground for justification and an explanation for why S comes to believe that p. So, seemings are one possible candidate for P. The second option I want to propose is Feldman’s (2003) theory of proper- responsiveness. According to this theory, a basic belief is spontaneously formed, and is ‘justified provided it is a proper response to experiences and is not defeated by other evidence the believer has’ (p.74). To support this thesis Feldman gives the example of walking into a room and seeing a table. He claims that believing that there is a table in the room ‘is a suitable thing to believe given that experience’, where to ‘believe something that does not fit that experience at all, such as that there is an elephant in the room, would not be a proper response to that experience’ (p.74).

39 A leading advocate of PC has been Huemer (2001, 2006, 2007, 2009), and other arguments for PC include McGrath (2013), Tucker (2010) and Skene (2013). Objections to PC have included Brogaard (2013), Conee (2013), DePaul (2009), DePoe (2011), Littlejohn (2011), Markie (2005, 2006, 2013) and Siegel (2013).

38

The notion of ‘proper-responsiveness to experience’ is helpfully clarified by a further example:

Compare a novice birdwatcher and an expert walking together in the woods, seeking out the rare pink-spotted flycatcher. A bird flies by and each person spontaneously forms the belief that there is a pink-spotted flycatcher there. The expert knows this to be true but the novice is jumping to a conclusion out of excitement. The expert has a wellfounded belief but the novice does not. In the same situation, both the novice and the expert may have wellfounded beliefs about the color, shape and size of the bird they see. This suggests that there is some relevant difference between such properties as being gray with pink spots and about 4 inches long and properties such as being a pink-spotted flycatcher. One might say that the former are “closer to experience” than the latter. Anyone with proper vision can discern the former in experience. This is not true of the latter. (Feldman 2003, p.75)

We can extract from this what Feldman means by proper-responsiveness. When we have certain experiences, the beliefs we immediately form from them are proper given what appears, for Feldman, to be two conditions. First, that the beliefs match up with the experience itself, and second, that they are consistent with other background beliefs and training we may have had.40 The novice arguably meets the first condition, but not the second, which explains why her response is not justified. The expert meets both, which explains why hers is. Given that proper-responsiveness concerns basic beliefs, it seems like a plausible candidate for P. The third option, which is very similar to the theory proposed by Feldman, is sometimes called process reliabilism (PR). Reliabilist theories of epistemic justification claim that a person S’s belief that p is justified if and only if S’s belief that p is formed by a reliable process (see Goldman 1979, 2008). According to its main advocate Alvin Goldman, ‘a cognitive mechanism or process is reliable if it not only produces true beliefs in actual situations, but would produce true beliefs…in relevant counterfactual situations’ (1976, p.771). By way of this definition, Goldman endorses the truth-conduciveness of epistemic justification that was emphasised by Bonjour in §1.3.1. A reliable process is that which increases the likelihood of obtaining true

40 The ‘training’ condition bears some similarity to virtue theories of epistemic justification, and Sosa’s (1991, 2007) theory in particular.

39 beliefs, and so a basic belief formed through a reliable process is more likely to be true than one which is not. What is the difference between PR and Feldman’s proper-responsiveness? It seems as though a properly responsive belief could simply be the result of a reliable belief-forming process. We can accept this claim, but still show that there is a difference between the two theories. For one thing, although Feldman claims that justification for basic beliefs consists in one making a proper response to experience, he doesn’t explain how this process actually works. The process reliabilist can step in at this stage and account for the relevant process in terms not being ‘from experience to belief but from receptor stimulation (to experience) to belief’ (Goldman 2008, p.77). By focusing on the role of the receptor stimulation, the process reliabilist can offer an account of what prior training and experiences have influenced the receptivity of a subject’s experiences to make them more conducive to forming true beliefs. Giving an account of this will help us to understand better why, for instance, the expert and not the novice is justified in his belief. So, PR offers us a further, more extended account of what P could be. I have given three options for non-inferential justification of basic beliefs, which I take to be plausible candidates for whatever property P justifies basic beliefs. A belief that instantiates this property will not be held arbitrarily. With this discussion in mind, the following gives a final account of epistemic justification.

(EJ): Non-inferential (basic) doxastic attitude D toward proposition p is epistemically justified for person S at time t if and only if D possesses property P, which non-inferentially justifies D.

Inferential (non-basic) doxastic attitude D toward proposition p is epistemically justified for person S at time t if and only if having D toward p fits the evidence S has for p at t.

I want to move now to briefly outline other kinds of justification for our beliefs. The notion of prudential justification has come to play an important role within the PFR, so it’s worth taking note of it at this stage, and analysing it alongside what has already been said of epistemic justification.

1.4 Sources of Doxastic Justification and Normativity

40

1.4.1 Prudential Justification

I have shown that epistemic justification is a matter of evaluating a belief in terms of its standing with respect to knowledge. In §1.3.2, I showed that knowledge is partly constituted by a true belief (A3), and that believing a proposition on the basis of evidence for that proposition is more likely to yield a true belief (A2). Epistemic means for assessing belief are then standards for determining a belief’s likelihood of being true for the purposes of gaining propositional knowledge. There are other ways of evaluating our beliefs, though, which are non-epistemic, since they are not concerned with truth-conducive factors for belief. Instead, they advise when it is prudent, or in line with certain practical aims we may have, to believe a proposition, even if that proposition is false or highly unlikely to be true given the evidence or other epistemic factors bearing on the proposition. For instance, suppose your friend has been accused of a crime and it would be beneficial for her if you believe she is innocent since this would give her some motivation to get through this difficult time. Further, imagine that the evidence is pretty damning and it looks like your friend is guilty. According to EJ, your belief will only be epistemically justified if you believe that your friend is guilty since that belief fits the evidence you presently have. Nevertheless, if you believe, instead, that your friend is innocent, this belief is still open to positive appraisal, but rather than epistemic appraisal, the appraisal is given in prudential terms. The support your friend receives in the knowledge you believe she’s innocent could be invaluable to her psychological well-being. In that case, it seems as though you are justified in retaining the belief that she is not guilty, but justified purely in prudential, rather than epistemic terms.41 Another case that is regularly cited in the literature – partly because it has some empirical basis – is believing you will overcome a disease in order to increase your likelihood of overcoming it.42 You may be handed clear evidence for the claim that your disease will get the better of you sooner rather than later, but if overcoming the disease can be aided by your believing that the disease won’t kill you, then it seems prudent to believe against the evidence in this case. Again, your belief will lack

41 This is quite a typical example from the ethics of friendship used to support pragmatic reasons for belief (see Keller 2004 and Kawall 2013). 42 In his defence of prudential reasons for belief Reisner (2008; 2009) cites this as one of his examples.

41 epistemic justification, but this may not be problematic. This is not a time for being concerned with having your beliefs match the evidence; rather, this is literally a matter of life and death, and it seems perfectly reasonable to grant positive, rational appraisal, to the belief that you will live, even if for purely pragmatic reasons. Evidently, then, a belief can be evaluated according to more than one standard. According to this notion of prudential justification, D toward p is justified iff it fits with our practical goals. Moreover, this thesis will have implications for the things that we ought to believe, i.e. for the norms that govern our doxastic states. As I want to show in the following section, there are norms that govern our beliefs, and these are independent of one another, but also require balancing together. They will have important ramifications for the way that the PFR is stated.

1.4.2 Normative Reasons for Belief

I was motivated to analyse epistemic justification because it looked like we require our beliefs to be true if they are going to guide our behaviour accurately. From this, the following principle looks highly plausible: D toward p is valuable for person S in terms of guiding S’s practical decision-making if D is true.43 It is one of the great benefits of knowledge that we may feel heightened confidence in those propositions that we use in our practical reasoning. Having justification for our beliefs increases the likelihood of their being true, and, conceptually speaking, turns true belief into knowledge.44 Even though beliefs are involuntary mental states and not actions, it does still seem open to us to place obligations on our beliefs in the same way that we would for actions.45 If we want knowledge then we ought to believe in accord with whatever

43 I opt for this sufficiency condition on the truth of belief rather than the stronger claim that ‘belief aims at truth’ because of the considerations brought to bear from prudential justification. My position is that belief may well have more than one aim, though truth will likely be the primary aim. For an overview of this discussion see Chan (2013). 44 For an important critique of this conceptual account, see Williamson (2000). See also, note 13 for a discussion of ‘Gettier’ cases. My claim here ought to be taken as a necessity condition: one knows p only if one is justified in believing p. It still isn’t clear what further conditions are required for knowledge. 45 The idea that beliefs are not suited to normative evaluation can be traced to Kant’s ought implies can principle: ‘Now of course the action must be possible under natural conditions if the ought is directed to it’ (1998 A548/B576). Recently, this has been discussed in the debate over doxastic voluntarism (Steup 2000). So, if our beliefs are involuntary but only voluntary

42 standard is appropriate for epistemic justification. I have defended a view according to which epistemic justification is a matter of believing those propositions that fit our current evidence. So, if person S wants knowledge then she ought to believe those propositions that currently fit with her evidence (inferentially), or that have some non- inferentially justifying property P. In terms of specifying an epistemic norm, this can be defined in relation to a desire to attain knowledge. Such a norm is proposed by Feldman (1988, p.243): ‘to say that you epistemically ought to believe p is to say that given that you want to achieve epistemic excellence, you ought to believe p’. Our beliefs can thus have a certain quality to them, and excellence as an epistemic quality, I suggest, is a property of a belief that is instantiated when a belief meets certain standards of justification, such that epistemic excellence in turn yields knowledge. It is, then, epistemically obligatory to believe a proposition that we have epistemic justification for, providing we want to attain knowledge. And with this claim, we see the connection between justification and obligation, which can be formulated as follows:

(EO): If person S wants to know p, then S is epistemically obligated to believe p if and only if S has epistemic justification for p.46

Note that the obligation is conditional: epistemic obligations are contingent on an agent’s desire to attain knowledge – a natural disposition according to some authors (Zagzebski 2012, pp.33-43). Should it be the case that an agent has no interest in knowing p, she isn’t obligated to believe p even if she is justified in believing p.47 On this account, then, being epistemically justified is not equivalent to being epistemically obligated – the former only entails the latter when the agent has the desire for knowledge.48 Also of note is that obligations come with reasons.49 In this case, being epistemically justified to believe p is a reason for being obligated to believe p. Finally, we must remember that this obligation is epistemic – nothing has been said here about actions are subject to obligation, then beliefs cannot be subject to obligation. I will sidestep this problem given that I take there to be sufficient solutions to it (see Heil 1984, and Feldman 1988). 46 I am thus setting up the question of doxastic duty in terms of hypothetical, rather than categorical imperatives. 47 For a similar account of rationality, see Foley (1993). 48 In this respect I deviate from Conee and Feldman who claim that ‘being epistemically obligatory is equivalent to being epistemically justified.’ (1985, p.19) 49 For an account see Littlejohn (2012, chapter 7).

43 non truth-conducive factors, or ethical obligations. This is purely an obligation provided you want to attain knowledge. As was argued in the previous section, it isn’t always of primary importance that our belief that p is propositional knowledge. Often we want our beliefs to help us attain non-epistemic goals. To outlive your disease is a goal that someone may be able to get closer to attaining through believing that she will outlive it, and if you have this goal, then you are prudentially obligated to believe the relevant proposition. The norm being described here is purely pragmatic – it obligates you to believe instrumentally, and in accord with certain goals you wish to achieve. We can state this norm as follows:

(PO): If person S wants to attain goal g, and believing p will partly or fully enable S to achieve g, then S is prudentially obligated to believe p.50

Now, supposing a person S has been diagnosed with cancer and her goal is to stay alive, and, moreover, she knows that believing she will stay alive will help her to achieve that goal. In that case she is prudentially obligated to believe she will stay alive according to PO. However, she also wants to know whether or not she will survive, and the evidence points to an unlikely outcome in this respect. She, then, equally appears to be epistemically obligated to believe that she won’t stay alive according to EO. In cases like this where there is normative conflict, S must weigh up the different pro-tanto oughts for belief that she has, which is an ought that holds as long as there are no countervailing oughts which outweigh it.51 She must determine which pro-tanto ought carries the greater weight and believe accordingly, at which point she will follow the best course of action all things considered. In religion there is an important and influential tradition, tracing its heritage to Blaise Pascal (1692, §343), which endorses a similar thesis to PO. According to this view, if our goals are to achieve eternal life, and believing that God exists can help you to attain this goal, then you are prudentially obligated to believe that God exists.52

50 There may be other means for S to achieve g other than believing p. If that were the case, S may have conflicting prudential obligations, and it will be a matter of balancing these to determine the most effective course of action. 51 See Reisner (2008) for a helpful discussion of how to balance pro-tanto obligations. A collection of relevant work on how to weigh our reasons is Lord and Maguire (2016). 52 A famous defence of this approach is given by William James (1896) who argued that religion offers itself as a ‘momentous good’ and its claims are worth believing in order to

44

One can have this obligation even if one is epistemically obligated to believe that God does not exist. Again, as with the prior case, determining what to believe is a contingent matter, conditional on the goals of the subject – be they epistemic or prudential – and the relevant weights attributable to each. I want to address one final norm of belief that is often discussed in relation to faith and reason – moral obligation. Arguably the origins of the debate over doxastic norms is W. K. Clifford’s essay The Ethics of Belief (1879). In that work Clifford introduces an example of a ship-owner who, having sufficient evidence to lead him to believe that his ship is not seaworthy, puts aside this evidence and believes that it is in fact seaworthy in order to collect money from passengers who wish to use the ship to sail abroad. In the example the ship goes down at sea causing the deaths of all of its passengers. In light of this example Clifford concludes that ‘It is wrong always, everywhere, and for anyone to believe anything on insufficient evidence’ (p.186). Although Clifford is sometimes interpreted as giving an epistemic norm here, I want to briefly argue that the kind of wrong Clifford has in mind here is actually moral. Note, first, that it is not the belief that appears to be immoral but the decision to act in accordance with the belief for which the ship-owner is to be blamed. However, the belief is in some way part of the decision to act. This is because beliefs play an important role in practical reasoning, so presumably if the ship-owner had have believed that his ship was not seaworthy, he would have been less likely to have sent it out. We might say that his believing it seaworthy increased the likelihood of his loading it up with passengers, though this belief was not decisive – we must also factor in his desires, ill-will, etc. Consider another case in this regard. A person who holds certain racist beliefs has an increased likelihood of performing an immoral racist act compared with if that same person held no racist beliefs. Since we are morally obligated to refrain from performing immoral acts, if believing a certain proposition makes us more likely to perform those acts, then we ought not believe that proposition. This gives us the following doxastic norm:

(MO): If believing p increases the likelihood that person S will perform some immoral act a, then S is morally obligated to disbelieve p.

attain this good. For a contemporary defence of Pascal and his approach see Jordan (2006) and Bishop (2007). A recent account that uses decision theory to defend the pragmatic view is Buchak (2012).

45

I have attempted to draw out from Clifford what he intended with his principle. Even though it was formulated as a kind of morality thesis about evidentialism, his examples concern the moral implications of acting according to false beliefs. But doing so is not always immoral. Supposing I believe, against the weather forecast, that it won’t rain today and so I choose to not take out an umbrella with me. Now, even if it rains, and so clearly I have believed and acted against the evidence, it doesn’t look like I have done anything immoral. I may have been epistemically obligated, but there was no moral obligation on me. The same isn’t true of the ship-owner. His false belief carries important moral implications and he ought to believe in a way that will lead him to act morally. Note how this weighs with EO and PO. The ship-owner does not want to know p, so he isn’t epistemically obligated to believe that the ship is not seaworthy. But, he does want to get some money together for himself from his paying customers, so he’s prudentially obligated to believe that it is seaworthy. So far he ought to believe that the ship is seaworthy. However, doing so will cause harm stemming from an immoral act, and as such, he ought not believe that his ship is seaworthy. In fact, the moral ought outweighs the other two in his case. As such, the ship owner has an all things considered obligation to believe that the ship is not seaworthy, stemming from the weight of MO. With these reflections at hand we can see that Clifford’s thesis that believing against the evidence is always immoral is false. Evidentialism is not a moral but an epistemic thesis. Nevertheless, there are moral norms that govern our beliefs, and these are to be weighed alongside other doxastic obligations.

1.5 The PFR Redux

1.5.1 An Epistemic Formulation of the PFR

We are now in a position to offer a full statement of the PFR, and to know what the implications are of this problem for propositional faith. The argument will divide in two important respects. First, the PFR is a claim about the epistemic status of faith – namely, that propositional faith involves unjustified belief, and hence cannot be knowledge. Second, and following from this claim, the PFR takes a view on the normative status of propositional faith, namely, that epistemically speaking, one ought not have propositional faith. Let’s begin with the first part of this argument.

46

Recall in §1.2 Hume’s account of faith which was that faith involves holding a belief without good reason.53 We can augment this claim somewhat to bring it in line with what has since been said about having a ‘good reason’ according to EJ. So, using a paraphrased version of EJ, combined with Hume’s account of faith – what we might call the evidentialist objection – we can infer that

(1) The beliefs that constitute a person S’s faith that p do not fit the evidence, nor are they non-inferentially justified (evidentialist objection);

This claim is not yet enough to show that propositional faith is evidentially or epistemically deficient, since it doesn’t state that all instances of propositional faith include belief. Only instances of faith that involve belief will be lacking in evidence according to (1). It is still open for someone to accept (1) and yet deny that their faith is constituted by a belief. The advocate of the PFR thus needs to assume the following claim:

(2) For any person S, S’s faith that p requires that S believes that p;

It seems relatively uncontroversial to assume (2), and it is certainly assumed by Hume and others (although questioning (2) will be crucial to the non-doxastic response to the PFR, to be discussed shortly). Now, from (1) and (2) you derive the following lemma:

(3) So, for any person S, S’s faith that p does not fit the evidence, nor is it non- inferentially justified (from (1) and (2)).

Returning to EJ, we know when a belief is justified and, conversely, when it is unjustified. An unjustified belief is one that fails to meet the requirements specified by the necessary and sufficient conditions. As such, EJ entails that

(4) Beliefs are epistemically unjustified iff they do not fit the evidence nor are they non-inferentially justified (EJ entailment); and it follows from (2), (3) and (4) that

53 It is crucial to acknowledge here that the doxastic attitude in question that is identified as part of faith is a belief, not suspension of judgment, and not disbelief. The attitude I am picking out by the term ‘belief’ is a positive state – a confidence in the truth of a proposition, or an involuntary taking of the proposition to be true.

47

(5) So, any person S’s faith that p is epistemically unjustified (from (2), (3) and (4)).

Returning to the arguments used to establish evidentialism, we can recall from discussion of arguments 3 and 4 that knowledge requires, not only a true belief, but one that is epistemically justified:

(6) Knowledge requires an epistemically justified true belief (A3, premise (7));

Since, according to (5), propositional faith is epistemically unjustified in virtue of the beliefs that constitute it, then, taken with our conceptual claims about knowledge (6), it follows that

(7) So, any person S’s faith that p is not knowledge (from (5) and (6)).

From conclusions (5) and (7) we gain a clear picture of the epistemic status of faith. According to (5) faith cannot be epistemically justified since the beliefs that constitute it do not fit the evidence nor are they non-inferentially justified. According to A2 in favour of evidentialism, epistemically justified beliefs are more likely to be true. There is then no guarantee that faith, and the beliefs that constitute it, have any more chance of being true than a belief held on guesswork. Furthermore, (7) tells us that propositional faith cannot be knowledge, and given the greater instrumental value of knowledge compared with mere unjustified belief, faith has a decreased instrumental value. These are problems that plague faith given the value we attribute to rational, evidentially-grounded belief. Finally, let’s look at these implications for what we ought to believe. To paraphrase EO slightly for ease, and assuming (2),

(8) If S wants to know p, then S is epistemically obligated to have faith that p iff S is epistemically justified in having faith that p (EO paraphrase, assuming (2)).

Affirming the antecedent, but taking into account conclusion (5), we see the following argument:

(9) S wants to know p. (10) Faith that p is epistemically unjustified (5); (11) So, it’s not the case that S is epistemically obligated to have faith that p (from (8) and (10)).

48

The claim made by (11) is what we might call the weak obligation argument, which is that we are simply under no epistemic obligation to have faith, regardless of the content of the proposition. There is a stronger argument, though, which relies on the claim that, not only does faith lack evidential support, but that it goes against the evidence. To begin with, consider the following entailment from EO in which a person wants to know whether or not p, and hence wants to know whether or not to adopt an attitude of faith (which entails belief) toward p:

(12) If S wants to know whether she epistemically ought to have faith that p or not, then, iff S is epistemically justified in believing p is S epistemically obligated to have faith that p, and iff S is epistemically justified in believing that not-p is S epistemically obligated to not have faith that p (EO entailment);

Now, let’s affirm the antecedent again, and bring in a strong epistemic objection to belief in p

(13) S wants to know whether she epistemically ought to have faith that p or not; (14) S is epistemically justified in believing not-p (‘strong epistemic objection’); from which it follows:

(15) So, S is epistemically obligated to not have faith that p (from (12) and (14)).

Conclusion (15) is what we might now call the strong obligation argument. It has been argued that when the content of the proposition is theistic or generally religious in nature, that we not only lack evidence in favour of p, but that the evidence in fact favours not-p.54 Not only do we have two conclusions which tell us that the epistemic status of faith cannot be knowledge ((5) and (7)), but we also have two conclusion that either do not obligate us to have faith (11), or obligate us not to have faith (15). As was argued in §1.4.2, these obligations need to be weighed against other obligations we may have. So, the prudential benefits that come with having faith might outweigh the epistemic obligations, and hence having faith may be permissible, even if, say, (10) and (14) are defensible. Such a point is argued for in the pragmatist tradition of Pascal and James discussed earlier. The epistemic concerns cannot be alleviated though

54 A common argument for this claim comes from the problem of evil and other objections to theistic belief. See Le Poidevin (1996) and Mackie (1983) for discussion.

49 without rejecting one of the premises. According to the conclusions reached thus far, whatever propositions are believed that constitute faith cannot be justified, and hence cannot be known. It follows that propositional faith is epistemically defective, and hence a poor basis for decision-making and a target for easy objection. How might the defender of propositional faith respond to these problems? Here, and for the remainder of this thesis, I will outline and evaluate two responses.

1.5.2 Response 1: Non-Doxastic Theories of Faith

The first strategy for responding to the PFR does so by rejecting (2). According to this response, faith that p does not require belief that p, and so not every instance of propositional faith will involve a propositional belief, which, as (1) implies, is epistemically unjustified. The advocate of this approach can then concede (1), but since faith needn’t involve belief, then (3) and, importantly, (5) will not always follow. According to Daniel Howard-Snyder, an articulate supporter of this view, the evidentialist objectors

…identify faith with believing something on insufficient reason or evidence…[I]f they’re right, faith requires belief—but it does not. For although…one can have faith that p when one merely accepts p, or merely assents to p, or (belief-less-ly) assumes p, one cannot believe p in that condition. (Howard-Snyder 2013a, p.369)

Howard-Snyder’s account of propositional faith requires a cognitive attitude, but that attitude needn’t be doxastic. It can be filled in by some non-believing state. Given this, the evidentialist objection (1) will have no force against propositional faith that is not constituted by belief. This same point is made in another non-doxastic account, this time given by Daniel McKaughan:

Because faith in God does not require justified belief that God exists, even if one were to accept both the evidentialist requirement…and the claim that the arguments or evidence for God’s existence do not adequately support belief…the conclusion that faith in God is thereby unjustified or in some sense irrational, does not follow. (McKaughan 2013, p.115)

McKaughan’s theory is similar to Howard-Snyder’s, but note that he is not only concerned with propositional faith, but also faith in God. This kind of faith is often 50 accounted for in terms of a something like trust or behavioural commitment. On another similar theory, Jonathan Kvanvig has defended a view of faith according to which one’s commitments and plans, arising from one’s affective attitudes and experiences, are sufficient for religious faith. Moreover, he uses this account in the same way as McKaughan and Howard-Snyder to reject (2) of the PFR when responding to the objections of the ‘New Atheists’:

Written in the tradition of concern over the clash of faith and reason, all such approaches think of faith as something belief-like or at least requiring quite specific cognitive commitments, so that the central question about religious faith is whether the beliefs involved can be held rationally. And then the question of whether religious faith makes any kind of sense turns into the question of whether adequate grounds can be found for religious belief. But if my description of a life of faith is where we start the discussion, there will be no specific cognitive contents that can be identified as the ones that are both epistemically problematic and essential to affective faith as such. On the picture of faith developed here, people of this kind of (religious) faith might include various kinds of skeptics and agnostics regarding the existence of God. (Kvanvig 2013, p.127)

The three accounts I have presented share a number of features in common. First, they all defend an account of faith in which faith does not require belief. Second, they use this account to overcome the PFR, and in terms of the argument as I have stated it, they do so by rejecting (2). Third, they seek to justify faith, or in some way make faith seem rational, without doing so in epistemic terms. Another advocate of this approach, Robert Audi, states that we need to take seriously ‘the possibility that faith, as a central element in religious commitment, can be rational even if theistic beliefs with the same content should turn out not to be’ (Audi 1991, p.213). Unsurprisingly, the kind of rationality or justification that tends to be utilised to defend faith is prudential; they often appeal to the sorts of arguments raised in §1.4.1.55 I will refer to the non-doxastic theory of faith according to which faith does not require belief as NDT (the non-doxastic theory). In Chapter 2 I will discuss propositional faith further, but in Chapter 3 I will give an extended outline of NDT, and evaluate the arguments used in its favour. If advocates of this position are going

55 See Buchak (2012) for a recent defence.

51 to use their approach to successfully overcome the PFR then that success rests on their theory being properly established through convincing argument. Before moving on I need to make two observations about NDT as an approach to resolving the PFR. First, the doxastic attitude picked out in (1) and (2) is a positive belief. However, non-doxastic theory requires some doxastic attitude, just not a believing one. It is widely held by its defenders that propositional faith can be constituted by suspension of judgment, as well as belief, and furthermore, that it is incompatible with disbelief. When suspension of judgment is the doxastic state identified in premise (1), most advocates of NDT in religion would want to reject (1) since they see that suspension of judgment is the appropriate doxastic state to adopt given the evidence we possess for religious propositions. As such, faith would not be epistemically unjustified since it would meet the standards set by EJ; hence (5) can be rejected. Despite this, a faith that does not involve positive belief cannot be knowledge since knowledge requires belief, so even if (5) can be rejected, (7) still holds. However, and this is the second point, if advocates of NDT want to endorse (1) when it is filled in by a positive belief, then they customarily also accept (7). Even when faith involves belief, on this view, it is epistemically unjustified and so does not lead to knowledge. This is problematic since, as mentioned in the previous section, it will have decreased instrumental value compared with beliefs that are justified. The only way around this problem is by rejecting (1). So, even if it turned out that NDT is a defensible position, its advocates will still face the problems that follow from (7) when faith involves belief. The approach I will consider next attempts to overcome to PFR by rejecting (1).

1.5.3 Response 2: The Divine Testimony Theory

The non-doxastic approach to responding to the PFR is a very recent project. Far more established though is the use of natural theology to attempt to demonstrate that belief in God is epistemically justified. This tradition goes back several millennia. Modern approaches to natural theology either attempt to develop arguments for theistic belief from empirical observation or from a priori premises.56 In either case, providing we are happy to grant evidential status to such observations and premises, then if a belief

56 For overviews of these different arguments see Swinburne (2004) and Dougherty and Walls (forthcoming).

52 in the proposition that God exists is based on these arguments, and the arguments appear to be sound, then that belief will meet the standards set by EJ. Consider, for instance, the cosmological argument, which reasons partly from empirical observations about the causation of physical objects, including our own universe, and partly from a priori premises concerning causation, to the conclusion that the universe must have an uncaused cause outside of itself (Craig 1979). Suppose we can take these observations as basic, either grounded in direct perception or testimony of another’s perception, and that belief in God can be inferred from these basic beliefs. Then we have evidence for the belief that God exists. And this kind of process can be followed for other arguments as well, though of course one’s belief that God exists will only be justified by these arguments if it is in fact based on them. One weakness of natural theology is its limited scope. Whilst some arguments may be used to justify propositions such as ‘God exists’, ‘there is an uncaused cause’, ‘the cause of the universe must be inconceivably powerful’, ‘the creator of the world loves beauty’, ‘the creator of humans values morality’, ‘God must be simple’ etc., these propositions are fairly narrow in scope. How could natural theology be used to justify the numerous other claims that are distinctive of each individual religious tradition? Take Christianity for example, whose central claims include ‘Jesus is God’, ‘salvation is for all’, ‘God made everyone with purpose’, ‘prayers are heard and answered’, ‘God has performed many miracles’, ‘there are angels’, ‘there is an eternal afterlife’, etc. What is the epistemic status of a person’s belief in these propositions and many more like them? Can anyone have evidence for these enough to satisfy EJ? The arguments typical of natural theology don’t address belief in these sorts of propositions.57 The source of these propositions is clearly the religious texts that form a foundation upon which to derive the claims central to the tradition. In some cases, as in the Abrahamic traditions of Christianity, Judaism and Islam, these texts are believed to be divinely inspired, and believers take this to warrant belief in the propositions found in the texts. The ubiquity of belief in propositions that have their source in sacred texts is an important observation since it broadens the epistemic demand to justify one’s beliefs beyond the central proposition that God exists, for which one can appeal to natural theology. Deference to natural theology no longer seems to be an

57 There is agreement on this in Kenny (1992).

53 option for many of the propositions believed on no other basis than sacred texts. As such, the demands set by EJ are much more widespread than is usually thought. Some authors regard this problem as central to the PFR:

The basic impetus for the problem of faith and reason comes from the fact that the revelation or set of revelations on which most religions are based is usually described and interpreted in sacred pronouncements, either in an oral tradition or canonical writings, backed by some kind of divine authority. (Swindal 2001, page number unavailable)

This is a quite different way of looking at the problem. Often the PFR is stated as a concern over the lack of evidence for the proposition ‘God exists’. From this alternative perspective, though, the question of the justification of one’s religious beliefs, which will be many and varied, is a question about the justification of believing propositions on the basis of alleged divine revelation. Those beliefs are justified providing believing in divine revelation is itself justified. One tradition in Christian philosophical theology draws such a close link between the nature of faith and revelation that it in fact defines faith simply as ‘belief that a proposition is true because God has revealed it’ (Kenny 1992, p.50).58 John Locke is one notable advocate of this view, who defines faith as

…the assent to any proposition, not thus made out by the deductions of reason, but upon the credit of the proposer, as coming from God, in some extraordinary way of communication. This way of discovering truths to men, we call revelation. (Locke 1924, Bk IV, Ch XVIII, §2)59

This definition of faith is perhaps too strong. For instance, it seems possible to have faith that God exists where this attitude was not formed by believing it on God’s own testimony. Nevertheless, a more moderate stance on Locke’s claim, that believing propositions on God’s testimony is a sufficient source for obtaining propositional faith with religious content doesn’t look problematic at all. It raises the interesting question that I proposed above as to whether such faith is epistemically justified.

58 A similar claim is made about this tradition by G.E.M Anscombe: ‘At one time there was the following way of speaking: faith was distinguished as human and divine. Human faith was believing a mere human being; divine faith was believing God.’ (1979, p.142) 59 A current advocate of this theory is Linda Zagzebski: ‘religious faith is believing God’ (2012, p.190).

54

I will refer to the theory that faith can be a matter of believing divine testimony as DTT (the divine testimony theory). This kind of religious faith is commonly found in the three Abrahamic religions, and DTT can be used as a means for addressing the PFR. In Chapter 2 I will give an analysis and defence of DTT as a theory of faith, and in Chapters 4 and 5 I will explore how DTT may overcome the PFR. There are several philosophical theories that have been developed in the last few decades that provide significant resources for showing how DTT may address the PFR which have as yet been left unutilised. What I have in mind is the burgeoning literature from the epistemology of testimony. Epistemologists working in this field have set out positions on the epistemic status of beliefs formed on testimony. According to those who hold a reductionist view of testimony, one’s justification for a belief reduces to other sources of basic justification, such as induction and perception (Fricker 1995). On the opposing theory of non-reductionism, testimony is itself a basic source of justification (Lackey 2003). Suppose we view religious texts and other putative sources of divine speech as sources of divine testimony. We can then use reductionism and non- reductionism in testimony to determine under what conditions beliefs based on these putative sources of divine testimony will be justified. Despite the availability of these theories to philosophers of religion, only limited attention has been paid to them as a means of justifying religious faith.60 I want to explore this route of justification further to examine its plausibility. This investigation will be approached as a means to rejecting (1), and if (1) can be effectively refuted such that beliefs held on this alleged divine testimony are justified, then conclusions (5) and (7) will not follow. As such, religious faith will not be epistemically defective and beliefs held on faith can be knowledge. This can be approached by two routes. First, by endorsing reductionism about testimony, and thereby determining whether one can have evidence that God has testified to them. Second, by endorsing non-reductionism, and thereby determining whether beliefs formed on divine testimony can be non-inferentially justified in some way.61 I will explore both options at some length in Chapter 5.

60 Two exceptions are Lamont (2004; 2009) and Wahlberg (2014). 61 This latter project can be seen as a new approach to the ‘Reformed Epistemology’ discussed in the Introduction.

55

1.6 Conclusion

The problem of faith and reason, I have argued, is an epistemic objection to propositional faith. It is formulated when we defend the view that beliefs must be epistemically justified if they are to count as knowledge, but maintain the beliefs constituting faith cannot be epistemically justified. I have endorsed a theory of epistemic justification that is broadly evidentialist and foundationalist. With this account of epistemic justification providing us with a precise formulation of the PFR, it will now be the task of this thesis to outline and evaluate responses to this problem. In order to analyse the plausibility of the responses to the PFR outlined in the foregoing, we need to discuss the nature of faith itself. As I will show in the next chapter, it is possible to defend DTT, and in Chapter 3, I will show that there are a number of serious problems with NDT. In the next chapter, then, I will explain how one can defend the divine testimony theory of faith.

56

2. THE NATURE OF FAITH

2.1 Overview

At the end of Chapter 1 I introduced the two theories of faith that I will now focus on as responses to the problem of faith and reason (PFR): Non-Doxastic Theory (NDT) and Divine Testimony Theory (DTT). In this chapter I will begin to evaluate DTT by exploring how the notion of divine testimony is tied to a kind of faith found in the three Abrahamic religions. I will refer to this as Abrahamic faith. My aim is to show that Abrahamic faith has two constitutive parts: first, a relational faith involving trust in God, and second, faith as a propositional attitude. In analysing Abrahamic faith, I will offer an account of each of these kinds of faith as well. As we saw in Chapter 1, DTT maintains that religious beliefs are often acquired by trusting God’s testimony. In order to explain how this works I will develop an account of trust, and use it to show what it is to trust someone’s testimony. I will then apply this to divine testimony, and show how trust in God’s testimony is often part of Abrahamic faith, and how trusting God’s testimony is related to propositional faith. This will enable me to provide a full account of Abrahamic faith In §2.2 I discuss an immediate obstacle to the possibility of analysing faith. This comes from the meta-philosophical claim that faith is a ‘family resemblance’ concept and so does not admit any systematic analysis. I will endorse this view to an extent, and concede that there is not one unified kind of faith, but ultimately find that providing we distinguish the kind of faith we want to analyse from other kinds, we can specify an analysable concept. §2.3 states that Abrahamic faith is the kind of faith I will analyse, and introduces propositional and relational faith. I then set out in detail how Abrahamic faith is constituted by relational faith by explaining what trust is. §2.3.3 draws from the general account of trust to give a theory of trust in testimony. This plays a pivotal role in the later chapters where I consider trust in God’s testimony. In §2.3.4 I discuss what role trust plays in Abrahamic faith, before arguing in §2.3.5, that trust in God’s testimony is related to propositional faith by the fact that it is often a cause of propositional faith. The chapter concludes in §2.3.6 with an account of propositional faith, which I argue is a complex attitude requiring belief and a positive evaluation of its object.

57

2.2 Analysing Faith

What is faith? In order to answer this question, I want to begin with a problem that faces any attempt to answer it, and use this problem to motivate my own method for analysing faith. The problem comes from the claim that there is no unified concept of faith. Some authors (Sessions 1994, p.253) have used this claim to say that no analysis can be given for faith. According to these authors, although there are resemblances between the different concepts for faith, there is no means of construing faith as a concept. This conclusion is arrived at with support from Ludwig Wittgenstein’s (1953) remarks concerning ‘family resemblance’. In this section I will explain how this argument works, and agree with its conclusion that faith is not a unitary concept, but defend the claim that faith can be analysed providing we identify the specific kind of faith to be analysed. My view is that in order to say what faith is our method must first be to say what kind of faith it is that we want to account for. Let’s begin by motivating the claim that there is no unified concept of faith. There are two sources for this motivation. First, there are numerous theories of faith that appear to be both plausible, and yet to be describing different concepts, some of which are incompatible. Second, there are numerous felicitous uses of ‘faith’ that appear to refer to many different concepts. Beginning with the first claim, recent accounts of faith take it to be, inter alia:62 belief in faith propositions,63 belief without adequate justification,64 belief plus various affective states that are formed via a sense of the divine,65 belief formed through divine revelation,66 a doxastic venture,67 a non-

62 This list is far from exhaustive. For further summaries see Bishop (2016) and Buchak (2017a). 63 This is perhaps the most common folk understanding of faith. Swinburne (2005, p.138-141) attributes the view to St. Thomas Aquinas. 64 I addressed this notion, which I attributed to Hume, Pinker, Leiter, and the New Atheists, in the previous chapter. 65 See Plantinga (2000, chapters 8 and 9) for a contemporary account. This will be discussed further in §5.4.2 of this thesis. 66 This model will be one of the two the main focal points of discussion throughout this thesis. For one notable endorsement of this view see Locke (1924, Bk IV, Ch XVIII, §2). 67 Versions of this view belong in the ‘pragmatist’ tradition of faith, notably in the work of Pascal (1660/1962) and James (1896). A contemporary account is Bishop (2007).

58 doxastic venture,68 action without belief,69 trust,70 hope without belief,71 and an allegiance to an ideal.72 Some of these accounts may be compatible. For instance, faith could be belief in faith propositions held without adequate justification, as I attributed to some authors in Chapter 1. Others are incompatible. For example, the theory that faith is a doxastic venture contradicts the theory that faith is a non-doxastic venture. Other theories may be compatible but are very different, which is the case with belief in faith propositions and action without belief. If each of these theories is in contention as a plausible account of faith but several are incompatible with each other, this raises the prospect that at least some of the theories are describing a different concept of faith. The second means by which we can motivate the claim that faith is not a unitary concept is to elicit intuitions about the nature of faith from appropriate uses of ‘faith’. If it’s appropriate to use ‘faith’ in some contexts and not others, then this might give some clue as to what we take faith to be. However, when we review different uses of ‘faith’, we are presented with a diversity of felicitous uses. Consider the following sentences:73

(1) John has faith that God created the world. (2) Mary places her faith in God’s loving plan for her life. (3) Claire has a faith. (4) Peter is a man of faith. (5) Daniel believed it on faith. (6) Kathy acted on good faith. (7) Steven kept faith with his wife.

It’s possible that each locution corresponds to a different kind of faith. Robert Audi (2008, pp.92-93) has argued for this position, saying that in instances like (1), which he calls ‘propositional faith’, there is ‘a certain positive disposition toward the

68 The next chapter will be dedicated to critiquing this account. For examples, see Howard- Snyder (2013a), Alston (1996) and Schellenberg (2005; 2014). 69 See McKaughan (2013), Pojman (1986) and Swinburne’s ‘pragmatic faith’ (2001; 2005, pp.147-151) 70 Swinburne (2005, pp.142-147) attributes this to Martin Luther. See also Schellenberg’s (2005, pp.127-166) account of ‘operational faith’. 71 See Muyskens (1979). Pojman (1986) could also be interpreted in this sense. 72 See Kvanvig (2013) and Dewey (1934). 73 This list is partly adapted from Audi (2008, pp.92-96).

59 proposition that this is so’. He distinguishes this from cases like (2), calling them ‘attitudinal faith’, which ‘implies certain attitudes, such as reverence and trust’. The next five he defines as follows: (3) ‘creedal faith, i.e., a religious faith, the kind one belongs to by virtue of commitment to its central tenets’; (4) ‘global faith, the kind whose possession makes one a person of faith and can qualify one as religious provided that the content of the faith is appropriate’; (5) ‘doxastic faith, illustrated by believing something ‘on faith’’74; (6) ‘acceptant faith, referred to when someone is said to accept another person, or a claimed proposition or proposed action, ‘in good faith’ or, sometimes, ‘on faith’’; (7) ‘allegiant faith, which is roughly fidelity, as exemplified by ‘keeping faith’ with someone’. Audi’s point is that since each of the uses of ‘faith’ in (1)-(7) look felicitous, and each sentence implies a different kind of faith, then this is evidence that there are a number of different kinds of faith. As with the different theories of faith I considered above, many of these kinds of faith look incompatible. For instance, there doesn’t seem to be any relevant similarity between the fidelity found in keeping faith with someone (7), and believing a proposition on faith (5).75 The diversity does not end there, though, since there is also faith in the context of different religions to consider. For example, according to some Buddhist scholars, faith is ‘a state of conviction or resoluteness that keeps one firmly rooted in practice…with resolute conviction in a state of clearness, tranquillity and freedom’ (Park 1983, p.16).76 As with the accounts discussed at the beginning of this section, this theory may be compatible with the ‘allegiance to an ideal’ account. One model of faith taken from Hinduism shares deep similarities to the ‘belief formed through divine revelation’ account that I want to discuss further in this chapter. According to this view developed by the theologian K. Satchidananda Murty (1959), ‘our awareness of God is his self-revelation’ (p.283), and faith involves ‘reliance on a self-authenticated source of evidence’ (p.301), namely, that very self-revelation.

74 Doxastic faith is likely what is referred to in the ‘belief without adequate justification’ model of faith (see note 3 above). The view is often associated with Kierkegaard (1941, p.204). 75 Some authors defend a more parsimonious interpretation of the uses of ‘faith’, but still conclude that there are several different concepts of faith. For example, Howard-Snyder (2016) argues that the seven kinds endorsed by Audi can be reduced to three – ‘propositional faith’, ‘relational faith’ (equivalent to Audi’s ‘attitudinal faith’) and ‘global faith’. 76 See Ladd Sessions (1994, chapter 4) for a more detailed overview.

60

Although there are some similarities between some of the different religious conceptions of faith, there are other concepts that may be highly plausible, but that are incompatible with other plausible concepts, as I claimed above. This is true also of the different ways of using ‘faith’, and the concepts implied by that use. Should we conclude from the evident plurality of different accounts of faith and uses of ‘faith’ that it’s not possible to give a unified analysis of faith? Those who adopt this view are often influenced by Wittgenstein’s theory of ‘family resemblance’.77 This theory claims that our concepts do not have necessary conditions that jointly determine their nature. To make this point, Wittgenstein invites us to consider the concept games:

I mean board-games, card-games, ball-games, Olympic games, and so on. What is common to them all? —Don’t say: “There must be something in common, or they would not be called ‘games’”—but look and see whether there is anything in common to all.—For if you look at them you will not see something that is common to all, but similarities, relationships, and a whole series of them at that. (§66)

To make his point that games lack any features common to all types he illustrates with a few possible conditions that there may be to define all of them:

Are they all “amusing”? Compare chess with noughts and crosses. Or is there always winning and losing, or competition between players? Think of patience. In ball games there is winning and losing; but when a child throws his ball at the wall and catches it again, this feature has disappeared. Look at the parts played by skill and luck; and at the difference between skill in chess and skill in tennis. Think now of games like ring-a-ring-a-roses; here is the element of amusement, but how many other characteristic features have disappeared! (§66)

At this point, having determined that games are not united by any common feature, Wittgenstein gives his family resemblance theory:

I can think of no better expression to characterize these similarities than ‘family resemblance’; for the various resemblances between members of a

77 The theory is discussed in his Philosophical Investigations (1953, §§65-71).

61

family: build, features, colour of eyes, gait, temperament, etc. etc. overlap and crisscross in the same way.—And I shall say: ‘games’ form a family. (§67)

And given the diversity in faith we have outlined thus far, it might be tempting to conclude something similar, namely, that faith is a family of concepts that bear a family resemblance to one another, but do not share necessary and sufficient conditions. And this is precisely the position taken by William Ladd Sessions in his wide-ranging study, The Concept of Faith:

[T]he term “faith” is not univocal in all its applications; it bears no single core meaning or invariant type of referent–not throughout history, not world-wide today, not even within “single” traditions of faith such as Christianity and Buddhism. Faith is not one single thing but many different things that more or less resemble one another, in various ways. To return to an analogy we used earlier, faith is like a family of variously related but vastly differing individuals, not like an assembly of variously garbed genetic clones. (Sessions 1994, p.253)

This conclusion is clearly taken directly from Wittgenstein’s theory, such that a challenge to one is a challenge to the other. Let’s briefly assess the family resemblance theory, before determining how to proceed with the analysis. The challenge that Wittgenstein is posing is that we cannot identify individually necessary and jointly sufficient conditions for our concepts because there is nothing in common that they all share. However, for some concepts at least, this claim is too strong. This can be illustrated with the well-known philosophical examples of bachelor and vixen. To be a bachelor, one must be an unmarried man: being an unmarried man is both necessary and sufficient for being a bachelor. To be a vixen one must be a female fox: being a female fox is both necessary and sufficient for being a vixen.78 In both of these cases, we identify members of a family by their satisfaction of certain conditions. This is different from identifying them by their resemblance to one another. Someone could resemble a bachelor by, say, being young, living alone, being promiscuous, and not wearing a wedding ring. However, if this person was either female or married then this person would not be a bachelor since this person does not satisfy the necessary conditions for being a bachelor. For these

78 See Suits (1967) for an analysis of games.

62 concepts at least, Wittgenstein’s argument fails since it does not acknowledge the difference between being a member of a family and bearing a family resemblance. For certain concepts, something must satisfy certain necessary and sufficient conditions in order to have the concept applied to it, and bearing a resemblance to those who do satisfy these conditions is insufficient for it to be a member of the concept. We can highlight this distinction between family and family resemblance further by considering examples of human families. In order to be a child of my mother and father’s biological family I must meet certain necessary conditions, namely, that I have a genetic heritage stemming directly from both my mother and father. If I satisfy these conditions then this is sufficient for me to be a child of their biological family. But this is insufficient for bearing resemblance to either my mother or father. I could bear no resemblance to either of them but still be a child of their biological family. Conversely, I could bear striking resemblance to a man or woman without being a child of their biological family.79 It looks like Wittgenstein’s claim that concepts are united by no common features that are necessary and sufficient for that concept is wrong for the concepts of bachelor, vixen, and being a biological child of two parents. This is due to a failure to distinguish between being a family member, which has necessary conditions, and resembling a family, which does not. In a similar way, we might wonder whether some of the examples considered earlier are genuine instances of faith, or whether they merely resemble faith. For instance, perhaps we could say that having propositional faith is genuine, whereas the concept of global faith is some imposter that merely resembles faith. To argue for this claim would require us to show why some plausible theories of faith and uses of ‘faith’ are not genuine. Furthermore, for those concepts that we think are genuine, we might try to find a way to combine them so that we have a unified concept. Once we have eliminated the concepts we think are imposters, and combined those that we think are genuine, we could attempt to give necessary conditions that are common across all the cases that remain. To try to eliminate some theories of faith and combine the remainder into one concept does not present the most obvious way to proceed. This is because some of the examples given above look like very plausible cases of faith, and yet are

79 For an extended discussion of this argument with different examples see McGinn (2011, chapter 2).

63 incompatible with other highly plausible instances of faith. Take locutions (3) and (7) for example. Someone can subscribe to a particular religion and hence have the faith referred to in (3). There appears to be nothing evidently problematic with this kind of faith. With regard to (7) though, a man can satisfy this kind of faith by demonstrating steadfast faithfulness to his spouse. Both are genuine examples of faith – neither merely resembles faith. Nevertheless, they are prima facie distinct concepts and so it doesn’t look like they can be defined by the same set of necessary conditions. So, neither kinds of faith should clearly be discounted, but neither do they share the same set of common features. It looks, then, like Wittgenstein’s family resemblance claims are vindicated for the concept of faith. Indeed, family resemblance is also justified by the fact that these concepts also share some similarities. For instance, one could show steadfast faithfulness to one’s religion, hence showing crossover between (3) and (7). If we concede Wittgenstein’s point about family resemblance for faith, is it possible to give an analysis of faith? To answer this question let’s consider an analogy with the concept friend. Say we wanted to give an analysis for friend, and one necessary condition for friend is that person S is a friend only if S is someone with whom we have a friendship. Whilst this looks necessary for friend in some contexts, it doesn’t look necessary in others. In the House of Commons of the UK parliament, for instance, a person is a friend in the sense of right honourable friend providing that the person is being addressed in that context and is a member of the addressee’s own political party. There needn’t be a friendship between the two individuals despite the fact that the concept friend applies to them. So, we could reject the friendship condition specified above in order to identify a set of conditions that will apply to all cases of friend. Alternatively, we could allow that friend varies by kind, and accept that some kinds of friends will lack necessary conditions that apply to others, even though they may also share some necessary conditions. As such, there will be a set of necessary conditions for friend in the context of the UK parliament, and a set of conditions for friend in other contexts. If we endorse this claim, then we are saying that at least some concepts vary by kind, and sometimes we must specify a particular context in order to accurately define the concept. Regarding faith, then, if we are going to say what faith is, then we need to first say what kind of faith we are referring to. The answer to this will take into account the context of the faith in question.

64

This methodology rejects a family resemblance view in one respect since it claims that faith is an analysable concept; that there are necessary and sufficient conditions for faith. However, in another respect it also accepts family resemblance since it agrees that there is no analysis that can be given for all concepts of faith since faith varies by kind. In order to analyse faith, we must specify the kind of faith to be analysed. There are two benefits to this approach. First, we can still say what faith is, and thus we can still appraise the epistemic status of faith. If no analysis of faith were possible, then presumably it would not be possible to say whether faith is justified or not. Second, since we can say what faith is, we can also say what faith is not, that is, we can still reject certain theories as failing to account for faith. In some cases, we might want to say that a theory that offers a description of faith does not provide a genuine instance of faith, but a mere resemblance of faith. In other cases, the theory may neither resemble faith, nor offer an accurate description of faith. What is of importance is to first state explicitly what kind of faith we are analysing, and then to offer arguments that persuasively set out the conditions for that faith. With this methodology in hand, I will begin in the next section by identifying what kind of faith I want to analyse.

2.3 Abrahamic Faith

The kind of faith I will be analysing is common to the three Abrahamic religions and it is therefore typical to see this kind of faith exhibited in the followers of these religions both throughout history and in the present day. We can also see a religious ideal of Abrahamic faith exemplified in the biblical character of Abraham. However, despite its religious pedigree, this kind of faith need not only be held towards religious entities like God, but can be exhibited in non-religious objects as well, in particular, towards other people. That is, the attitudes constitutive of what I refer to as Abrahamic faith are found in followers of the three Abrahamic religions, but are exhibited towards non-religious objects and propositions as well. In order to analyse Abrahamic Faith I will begin by introducing two other kinds of faith which are the two putative constituents of Abrahamic faith. These are propositional faith and relational faith. Both were briefly considered by Audi in the previous section in locutions (1) and (2) respectively (although Audi referred to (2) as ‘attitudinal faith’). After introducing these kinds of faith I will analyse the nature of

65 relational faith in detail by looking at the nature of trust. I will then discuss how they are each involved in Abrahamic religious faith, and how they relate to one another, before giving an analysis of propositional faith.

2.3.1 Propositional Faith and Relational Faith

Propositional faith is fairly widespread and can be held towards many different propositional contents. For instance, I can have faith that my team will win the game, or faith that my spouse will remain committed to me, or faith that God exists. It is typical for the expression ‘faith that’ to refer to propositional faith. There are a number of different theories concerning the nature of propositional faith, though they often bear similarity. Recall that Audi defined propositional faith as ‘a certain positive disposition toward the proposition that this is so’. Similarly, J. L. Schellenberg claims that propositional faith concerning religious propositions ‘involves some kind of positive and assenting attitude toward religious propositions’ (2005, p.108).80 Both of these descriptions of the attitude in question look like they could also apply to propositional belief. However, as will be discussed in Chapter 3, both Audi and Schellenberg (among others) maintain that belief is not the only cognitive attitude that can meet their descriptions of propositional faith. In §2.3.6 I will offer my own account of propositional faith, which will claim that it also requires a positive evaluative component. Propositional faith is widespread within the lives of people who are committed to the Abrahamic religions. Christians, for example, will typically have faith that Christ was resurrected, or that God is sovereign, or that God answers prayers, etc. It would be unusual if someone claiming to be a Christian, or a Muslim, or Jewish, did not also claim to have faith towards propositions with a relevant religious subject- matter. It seems to follow, then, that propositional faith is required by Abrahamic faith. There is another kind of faith that is generally thought to be part of Abrahamic faith, which we can call, following Howard-Snyder (2017), relational faith.81 Faith in

80 Two other leading accounts of propositional faith are Howard-Snyder (2013a) and Buchak (2012). 81 Other proposals include ‘operational faith’ (Schellenberg 2005) and ‘attitudinal faith’ (Audi 2008). In my view, the former is too impersonal, whereas relational faith is usually interpersonal, and the latter is not distinctive enough since propositional faith is also a kind of attitude.

66 the Abrahamic religions involves an interpersonal relationship between a human person and a divine being, and this is manifest in our concept of relational faith. Consider the character of Abraham himself. In the biblical book of Genesis, it records that God promises to miraculously give Abraham a child, and that child would found the nation of Israel. In order to receive this blessing, he is required to move to a foreign country and do what God asks. Despite being advanced in years – so much so that only a miracle could give him a child – he trusts what God has told him, putting aside his doubts and relying on God for his provision.82 Abraham goes through with God’s commands, moving to a foreign land, and continuing to trust in God’s promise for his life of having a son. Abraham’s behaviour exhibits trust, and is a standard example of relational faith in the religious domain. Someone need not be aware of the faith of Abraham to exhibit Abrahamic faith – it is simply that relating to God in a trusting way is characteristic of religious faith. Outside of religion, relational faith is manifest in interpersonal relationships between people, such as that between friends and spouses, and many features of the trust exhibited by Abraham will also be true of the faith in these kinds of relationships. It is typical to distinguish relational faith from propositional faith by use of the expression ‘faith in’ as opposed to ‘faith that’.83 This distinction, as made in theology, goes back at least as far as Martin Luther in the 16th century:

[T]here are two ways of believing. In the first place I may have faith concerning God. This is the case when I hold to be true what is said concerning God. Such faith is on the same level with the assent I give to statements concerning the Turk, the devil and hell. A faith of this kind should be called knowledge or information rather than faith. In the second place there is faith in. Such faith is mine when I not only hold to be true what is said concerning God, but when I put my trust in him in such a way as to enter into personal relations with him, believing firmly that I shall find him to be and to do as I have been taught. . . . The word in is well chosen and deserving of due attention. We do not say, I believe God the Father or concerning God the

82 This narrative is largely recorded in Genesis chapters 12-15. The specific passage that refers to a discussion between Abraham and God where he expresses his doubts but ‘believes’ God is Genesis 15: 1-6. 83 See Price (1965) for a helpful discussion of how belief-in contrasts with belief-that. There are a number of relevant similarities for faith.

67

Father, but in God the Father, in Jesus Christ, and in the Holy Spirit. (Luther 1907, p.1:203)

According to this distinction, propositional faith, or faith that p, is a matter of holding true certain propositions, whereas relational faith, or faith in, involves the trust typical of a personal relationship. What is central to relational faith, according to Luther, is trust. Since trust is important for developing relationships between persons, it looks plausible that trust is involved in relational faith. Someone can be related to God by having faith in God where this faith is a matter of trusting in God. It has been popular amongst other authors too to claim that relational faith involves trust.84 For instance, Richard Swinburne maintains that, according to one concept of faith (what he calls ‘The Lutheran View’), ‘faith involves both theoretical beliefs-that…and a trust in the Living God’ (2005, p.142). One argument for the claim that faith involves trust, which is used by J. L. Schellenberg, is by appeal to uses of ‘faith in’: ‘How could it be correct to say that I have faith in God or the Dharma or democracy or the educational system if I am not inclined to place my trust in the items mentioned?’ (Schellenberg 2005, p.109). I agree that having faith in these things will typically involve being inclined to place trust in them. However, I don’t think it follows that every use of ‘faith in’ implies trust. Suppose I have faith in the England cricket team to turn their poor form around. This attitude might be better characterised by a confident expectation or by hope in my team, rather than by trust. Nevertheless, talk of ‘faith in’ something often does imply trust in that thing, and so to say we have faith in someone can be to say we trust that person, and hence to say we have faith in God can be to say that we trust God. Regardless of whether the use of the expression ‘faith in’ always implies trust, there is no doubt that trust plays a critical role in relating people to each other in non- religious contexts, and to God in certain religious contexts.85 When it is said that someone’s faith is a matter of trust, we are using the concept of relational faith. I have now introduced two concepts of faith that are required by the faith of the Abrahamic religions. I want to move on to explore relational faith in detail before discussing how propositional and relational faith relate to one another, and how they

84 According to other authors, the claim that faith involves trust has become platitudinous (McCraw 2015). 85 It clearly won’t be the case that trust in God is required by non-theistic religions. However, I’m focussing on faith in the Abrahamic religions, which will involve such trust.

68 are each required for Abrahamic faith. I have claimed that several authors hold that relational faith is a matter of trusting another person, and in Abrahamic religious contexts, this is a matter of trusting God. In order to analyse relational faith, I will begin by giving an analysis of trust.

2.3.2 Trust

In this section I will trace some of the recent literature on trust to outline its central components. I will argue for three necessary components of trust, and these will help us to understand the nature of relational faith. I will argue that trust involves (a) reliance on someone to perform some action, (b) a belief that the trusted party is committed to do what she is relied on for, and (c) a belief that the trusted party is trustworthy with respect to performing the relied upon action. To begin with, consider the example of an employer who trusts his employees not to steal from the till. We might explain this in terms of the employer holding a particular mental state towards his employees with respect to their likely actions concerning the money kept in the till. The content of this mental state will be that when the employees use the till, they won’t take money from it (assuming that their inaction is still a kind of action). In this sense, then, trust is simply a mental state we hold towards something. However, the example could also be interpreted such that a particular action is undertaken by the employer, one where he chooses not to take precautions regarding the money kept in the till or with his employees who are responsible for it. Perhaps he does this by deciding to hand over the keys to the till to his employees, and to not routinely check the CCTV to see if they are stealing from him. In this sense, trust in an action; it is something we do. There seems to be truth in both the claims that trust is a mental state and that it is an action. Are both necessary to our concept of trust, or can they come apart and still count as a kind of trust? For instance, suppose that the employer trusts his employees that they will not to steal by holding the mental state that they will not steal. However, he acts as though they will steal by choosing to hold onto the keys to the till for himself, such that any transactions involving the till requires him to come over and unlock it. Can he really be said to trust his employees? In this case the employer has the mental state appropriate of trust without the appropriate action. Imagine the situation reversed where the employer judges that his employees will most likely steal,

69 but he hands over the keys anyway. When the mental state and action pull in these different directions, can the employer be regarded as trusting his employees? On the account I will give in this section, I will argue that trust requires both a mental state and an action.86 As such, in the two examples above, where, first, the employer had the appropriate mental state but lacked the corresponding action, and second, where the employer lacked the appropriate mental state but acted accordingly, each will fail to qualify as cases of trust. I begin the account where many other writers often do by noting that trust is typically a three-place predicate in which person X trusts person Y to .87 Note that it seems rare, if ever, that someone trusts another person entirely, and so I won’t be discussing a pure kind of trust that is not with respect to something specific. Instead I will concentrate on trust as it applies to a person performing some identifiable action(s). Furthermore, trust needn’t be just between persons but could also be between a person and, say, an institution like a government.88 The majority of my examples, however, will focus on trust between persons as this is particularly relevant to faith in God. There is general agreement amongst accounts of trust in the recent literature that in order for X to trust Y to , X must be reliant in some way on Y -ing.89 Appeal to examples appears to confirm this claim. When the religious individual trusts God, she relies on God to act in some way, for instance, to provide for her needs or to offer her guidance, etc. Let’s examine this relation between reliance and trust more closely to see why trust requires reliance. Some accounts of reliance take it to be a disposition to use a proposition in guiding our behaviour and practical reasoning. For example, according to a recent account given by Alonso, reliance

86 Zagzebski’s (2014) account of trust also includes both components. 87 This relation is noted by Hardin (2002, p.9). For some discussion of the three-place predicate, and alternatives to it, see Faulkner (2015). 88 The situation is slightly different with ‘self-trust’, which is central to research in virtue epistemology. For a general account of this see Zagzebski (1996), and for one applied specifically to religious epistemology, see Zagzebski (2012) and the collection by Callahan and O’Connor (2014). 89 For instance, see Baier (1986), Jones (1996), Hardin (2002), Zagzebski (2012) and Hawley (2014).

70

…involves a reasoning disposition: that is, relying on p involves a disposition to, among other things, deliberate on the basis of p, plan on the basis of p, act on the basis of p and draw conclusions from p, provided the relevant conditions are satisfied. (Alonso 2014, p.166)90

According to this account, to rely on p is to be disposed to act in various ways providing certain conditions hold. Despite having a disposition though, it’s not the case that the disposition will always result in action since dispositions can sometimes be ‘masked’.91 For instance, a glass might be disposed to break when dropped, but if masked by packaging material, when dropped the glass will not break (Johnston 1992). The same will hold for reliance dispositions. For instance, suppose that a husband relies on his wife to remain constant in their marriage, and does so by being disposed to reason practically according to the proposition my wife will remain constant. Given his disposition, the husband is disposed to deliberate, plan, act on the basis of, and draw conclusions from this proposition. Should this disposition be acted upon, it would manifest in certain behaviours for the husband, such as allowing himself to grow a strong loving attachment to his wife, to plan a future family together, and to cease considering other romantic partners. Now, even though the husband is disposed in this way, he may also have competing dispositions, such as a deeply held fear that his wife will cheat on him. This disposition masks his disposition to rely on his wife. Acting upon conflicting dispositions will see the husband keeping his love at arm’s length, never allowing himself to fully fall for his wife, not being willing to consider the prospect of children, and always keeping an eye open for future partners. To take our example further, let’s suppose the fear disposition is stronger than the reliance disposition, to the extent that the husband, without exception, behaves in accordance with it. He is disposed to rely on his wife, but he can’t overcome his fears. Perhaps he has had too many damaging relationships in the past, and finds it incredibly difficult to open himself up to anyone else in a relying way. This unfortunate situation of the conflicted husband is perfectly possible, but is problematic for Alonso’s dispositions account of reliance since, when we reflect on it, we can see that dispositions are insufficient for reliance given that they can become

90 Similarly, Williamson (2000, p.99) claims that ‘using p as a premise in practical reasoning is relying on p’. 91 For an overview see Choi and Fara (2012).

71 masked. What is required for reliance is not dispositions to act, but action. This is clear from the fact that in spite of his disposition to rely on his wife, the husband never in fact actually relies on her because he never performs any of the actions that make him reliant. So, if our behaviour doesn’t line up with our disposition, then we don’t actually rely on others: reliance requires not merely the disposition to act but actions themselves.92 The reliance that constitutes trust requires not merely a disposition to act on the basis that Y will , but to actually act on the basis that Y will . We can develop on the claim that reliance requires action by considering how we entrust goods to other people. I stated at the beginning of this section that trust involves both a mental state and an action. When we consider trust as a mental state that X has towards Y -ing, the mental state of trust is directed towards the future actions of Y. However, if X is to act in a trusting way, X must also perform some relevant action. For instance, when a husband relies on his wife to not cheat on him he acts as though she will not. In doing so, he risks a tremendous amount, and is made extremely vulnerable to her should she decide to act contrary to what he relies on her to do. Likewise, an employer risks losing money, and a religious person risks becoming socially maligned. What is going on in these kinds of cases, some have argued, is one person entrusting something to another, and incurring a risk for doing so. According to one recent discussion, with entrusting ‘[person] x entrusts g to [person] y, where g is some good. With entrusting, all the action goes to x’ (Hieronymi 2008, p.218 n.9). What this means is that for one person to entrust something to someone else, she is required to act in some way, and this action is one where the person gives something valuable to the other. This concept of entrusting, I want to propose, is ideally suited to our concept of reliance. For example, a religious person might rely on God for his physical and spiritual provision by entrusting her physical and spiritual needs – her ‘goods’ – to God. The husband will be relying on his wife to remain committed to him by entrusting his emotional wellbeing and future happiness to his wife (the husbands ‘goods’). Entrusting fits in neatly as an explanation of the active component of reliance: we rely on others by entrusting some ‘good’ to them – something that is of value to us. We can’t merely be disposed to rely, we must actually rely, and to do so, we entrust others with something of value – some ‘good’.

92 For another account of reliance that involves both disposition and action, see MacCormick (1984, p.195).

72

The account of reliance I have proposed is then:

Reliance: To rely on someone to  is to act on the basis of the proposition that she will  by entrusting some good to her, and in so doing, to be at risk given that she may not .

What exactly are we at risk of when we entrust our goods to another person by relying on her? The answer to this question marks one of the two features that distinguishes mere reliance from trust. Are we vulnerable, say, simply to the person not -ing and hence damaging the goods they have been entrusted with? Some examples suggest that this vulnerability is not enough for trust, even if it is for reliance. For instance, suppose we place a valuable object on a shelf in the hope that the shelf will hold the object. So, we have entrusted some good to the shelf. When we do this, are we relying on the shelf, or trusting the shelf? A number of authors have argued that we are reliant but not trusting. The reason for this is because, if the shelf collapsed and damaged the goods we placed upon in, then we may feel disappointed, but we wouldn’t feel betrayed. However, when we rely on something we may feel disappointed if it doesn’t come through for us, but when we trust something we are warranted in feeling not just disappointed, but feeling betrayed.93 Imagine a similar case in which some thieves are relying on you to leave your house alarm off before going away for the weekend. Can we also say that they trust you to be forgetful with the alarm? Plausibly, the answer is no, and again the reason why is that, although the thieves are vulnerable in their reliance, they are not vulnerable to betrayal, which is what would be required for them to trust, rather than merely rely on you. The thieves might feel disappointed in you for keeping the alarm on, but they wouldn’t be warranted in feeling betrayed. This gives us our first pass at an account of trust:

To trust someone to  is to rely94 on her to  and be vulnerable to betrayal should she choose not to .

93 For those who have distinguished trust from mere reliance in this way, see: Baier (1986, p.234); Hawley (2014, pp.1-2); Hieronymi (2008, p.215); Holton (1994, pp.2–3); Jones (1996, p.14; 2004, p.4); McLeod (2015); O’Neill (2002, p.15); Potter (2002, pp.3–4); Pettit (1995, p.205). 94 ‘Rely’ follows the definition of reliance given above.

73

We have thus far been discussing the active component to trust. This analysis states that trust requires reliance, and we can see from the analysis of reliance that reliance requires action – to entrust some good to someone else – and so in virtue of this, trust requires action. Entrusting in this way makes one vulnerable to betrayal. However, the definition thus far raises an immediate question that will bring in the mental state component of trust: why would we be betrayed by someone if she didn’t do what we relied on her to do? Why, for instance, does the house owner not betray the thieves but the wife would betray the husband? A plausible answer to this is that when we recognise a lack of commitment on behalf of the person or thing that we rely upon, to that extent we also recognise that the not doing of what we relied upon that person or thing to do constitutes no betrayal on her part. In short, we can’t be betrayed by someone not -ing if she never committed to  in the first place. This claim explains why some cases look like mere reliance rather than trust. I made no commitment to the thieves to leave the house alarm off before I left, so I didn’t betray them. If I had promised the thieves that I will not set the alarms before I leave for the weekend, perhaps as a ploy to have them caught, then when they break into my house and the alarm goes off then they can rightly feel like I have betrayed them because of my stated commitment to leave the alarm switched off.95 The thieves don’t just merely rely on your forgetfulness in switching on the alarms, but they actually trust you to keep them switched off because you have promised that you will. Despite the plausibility of this commitment view, there are some fairly immediate counterexamples that cause problems for it. For instance, I trust strangers in public areas not to attack me, and I trust my friends not to steal possessions from my house when they come and visit. In neither case, though, must these people have made an explicit commitment not to do these things for me to trust them not to do it. In order to respond to these cases in a way that allows us to retain the commitment account, we need to construe commitments in an especially broad sense. Such a definition is given by Katherine Hawley who is an advocate of including a commitment condition in an account of trust:

95 The view of commitment I am advocating here can be formulated according to John Searle’s theory of speech acts, in which promises, which are a kind of commissive ‘are those illocutionary acts whose point is to commit the speaker…to some future course of action’ (Searle 1979, p.14).

74

commitments can be implicit or explicit, weighty or trivial, conferred by roles and external circumstances, default or acquired, welcome or unwelcome. In particular I will take it that mutual expectation and convention give rise to commitment unless we take steps to disown these. (Hawley 2014, p.11)

Perhaps, then, when I step out into public areas, I trust, rather than merely rely on, the people around me to let me walk unhindered because I recognise their implicit commitment to respect the social conventions of law and the right of free-will. If they abuse either of these conventions by attacking me then I feel they break their implicit commitment and betray my trust. Similar accounts can be given for when people are invited into my home: an implicit commitment is made to respect my property and therefore not to steal it. Taking my possessions would betray my trust since this commitment would be violated. Using this broad notion of commitment, then, we are able to deal with such difficult cases and retain the commitment account. The final point to note on commitment is that in order to feel that our trust has been betrayed we need to have acknowledged that the trusted party has made this commitment to us, be it explicit or implicit. I can’t betray the trust of the thieves even if I have made a commitment to them if they don’t acknowledge this commitment. This is best explained in doxastic terms: we believe that the trusted party has a commitment to act in some way. But do we have to believe it, or can we merely choose to voluntarily accept that someone is committed to -ing in order to trust her? This looks unlikely for the simple reason that if we didn’t believe she was thus committed, but merely went along with it in an accepting way, we wouldn’t feel betrayed if she didn’t . In that case we would lose the explanatory benefit of the commitment account. If the thieves didn’t believe that I made a commitment to leaving off the house alarms, but made a decision to accept that I had, they would have no right to feel betrayed by me. This would be akin to relying on the shelf to hold a valuable object. We might rely on it, but we wouldn’t trust it. That’s because we don’t believe that it’s committed to holding the object, even though we might accept that it is, which would be unusual, given that it’s an inanimate object. The vulnerability to betrayal claim from the initial analysis of trust is now explained by the commitment claim. We are vulnerable to betrayal when we trust

75 someone because when we trust someone we believe that the person is committed to -ing, and if she doesn’t , then she betrays our trust. This claim can be stated as such:

Trusting Commitment: To acknowledge that someone is committed to -ing in a trusting way one must believe that the person has made an implicit or explicit commitment to , and be reliant on her -ing such that if she chooses not to , the trusting party may justifiably feel as though her trust has been betrayed.

A revised formulation of trust will involve both reliance and trusting commitment:

To trust someone to  is to (a) rely on her to , and to (b) believe she has made a trusting commitment to -ing.

This account of trust is almost complete. It still remains, though, to offer a fuller analysis of the mental state constitutive of trust. At present, the account involves practically relying on someone to  and believing she is committed to -ing. However, again when we reflect on certain cases of trust, it also appears that trust requires that you are optimistic about the prospects of the trusted party -ing; optimistic about her ‘coming through’, as it were. Recall the example of the shop owner from the beginning of this section. I claimed that he could be interpreted as holding a particular mental state towards his employees, one with respect to their likely actions concerning the money kept in the till. This attitude was a positive mental state directed at the proposition the employees won’t steal from the till. A number of recent theories of trust (e.g. Jones 1996, 2012) have accounted for this mental state in terms of a recognition of another person as trustworthy. What does this ‘recognition’ amount to, and what is ‘trustworthiness’? Let’s look at the second of these two questions first. By understanding what it is to be trustworthy, we can be more explicit about what it is to recognise another’s trustworthiness. According to Karen Jones, who has produced perhaps the most influential contemporary philosophical work on trust,96 trustworthiness can be construed in the following way:

[Y] is trustworthy with respect to [X] in [her -ing], if and only if she is competent with respect to -ing, and she would take the fact that [X] is

96 For some of her keys works, (1996; 2004; 2012; 2013).

76

counting on her [to ], were [X] to do so…to be a compelling reason for acting as counted on. (Jones 2012, pp.70-71)97

On this account of trustworthiness, then, were the employees counted on by the employer to be honest in leaving the money in the till, then the employees would acknowledge this as a compelling reason to leave the money in the till. The employees exhibit trustworthiness in the fact that they are reliably motivated to act in the relied- upon way. But also, their trustworthiness is manifest in their competence with respect to performing the required action. They need to have some minimal competence in achieving the relied-upon task in order to be considered trustworthy with respect to completing it. This might be a relatively simple condition to fulfil for many cases. It shouldn’t be too difficult, for example, to keep yourself from removing money from a till. However, the same isn’t true with trusting someone to fly an airplane accurately, or perform a complex medical procedure. Such tasks require a much greater degree of skill and experience before the person carrying them out can be considered competent and therefore trustworthy with respect to performing each task. The kind of theory of trustworthiness put forward by Jones is usually referred to as a ‘goodwill and competence account’,98 and it is goodwill and competence that, I propose, we recognise in another when our attitude towards her is one of trust. We recognise the goodwill and competence of another, and this is partly constitutive of the mental state that is a component of trust. One way to have this recognition is, of course, to simply believe that the trusted person is trustworthy with respect to the action in question. Must this mental state be a belief? As I asked with trusting commitments, could this mental state be something other than outright belief? According to the account of trust given by Jones (1996), the state of recognition need only be an affective state of ‘optimism’. This, she claims, can come apart from the belief. However, it would be unusual to think that belief can entirely come apart from optimism since it would allow for the following absurd result: I can be optimistic about

97 I have taken some minimal liberties with paraphrasing here to keep consistency with my discussion thus far. The names are changed from A and B to X and Y. Jones also talks of ‘domains of interaction’ as a broad notion, rather than some particular action ‘’. The account is not changed by this narrowing of scope, but I do consider myself largely sympathetic to the broader-scope approach that Jones adopts. I have only changed it for consistency and simplicity. 98 For an overview, see McLeod (2015). These accounts were first developed in the work of Baier (1986).

77 your trustworthiness with regard to -ing, whilst believing you are not trustworthy with regard to -ing. For instance, I can be optimistic that you would make a trustworthy business partner, whilst disbelieving that you will make a trustworthy business partner. This apparent incongruity gives us a reason, for now at least, to assume that belief is the required state for trust, but to anticipate that it is often, if not always connected to optimism in some way. Perhaps optimism necessarily causes belief, or perhaps belief necessarily causes optimism, or perhaps they are part of the same complex state. I won’t take a position on this for now, but will simply assume that belief is somehow required for the mental state of trust. (In the latter parts of this chapter, we will consider further whether belief is required for trust, and there I will raise some problems for the view that it is not). We are now in a position to give an account of some (or possibly all) of the necessary conditions for trust:

Trust: To trust someone to  is to: (a) rely on her to , (b) believe that she is committed to -ing, and (c) believe (and be optimistic) that she is trustworthy with respect to -ing.

I argued in the previous section that relational faith involves trusting someone. If the analysis of trust I have given here is correct, then we can understand relational faith in terms of this analysis. But there is an important domain within which we trust other people: trusting someone’s testimony. I will discuss shortly how this kind of trust is important to Abrahamic faith, and crucial to motivating the divine testimony response to the PFR. At this point, then, I want to explain what this kind of trust is, before moving on to discuss its role in Abrahamic faith.

2.3.3 Testimonial Trust

An important and fairly common instance of trust is where we trust someone’s testimony. I will argue that with testimonial trust (TT), one has trust in another person’s sincerity and competence with respect to testifying. I will argue for this by explaining what it to testify, and drawing from this an analysis of TT. I will then use this analysis to show that TT requires believing the testimony of speakers.

78

Philosophical discussion of the social dimension of knowledge, by which we acquire knowledge from other people, has bloomed in the past thirty years or so,99 and testimony is seen as the primary way by which we acquire knowledge from others.100 Historically, epistemology has focussed almost entirely on the ways in which people acquire knowledge without reliance on other people.101 This is surprising when we consider just how many beliefs we hold on the basis of the testimony of others. Beliefs concerning scientific claims, the thoughts and feelings of other people, historical facts, and the content of news reports are all formed from the testimony of others. Testimonial beliefs, it seems, are ubiquitous.102 When speakers tell propositions to hearers, speakers are giving their testimony to hearers. A typical instance of testimony involves a single speaker verbally telling a hearer or group of hearers p. If I tell a friend the time, or what I had for lunch today, I am testifying to my friend. Of course testimony also occurs when groups of speakers testify to groups of hearers, or to individual hearers, or when someone writes a letter to someone else, or through a meaningful hand gesture, etc. For simplicity in analysis though, I will address cases with a single testifier reporting via speech. When a speaker S tells a hearer H that p, she performs a particular action – an illocutionary act. The purpose of illocutions is simply to do something with words.103 According to a widely adopted taxonomy developed by John Searle (1979, Chapter 1), there are five types of illocutionary act, each of which is partly defined by the ‘purpose’ of the act: assertives – to commit the speaker to the truth of the expressed proposition; directives – to get the hearer to do something; commissives – to commit the speaker to a future course of action; expressives – to express one’s psychological

99 For some early discussions see Fuller (1988), Goldman (1999) and Schmitt (1994). Some helpful overviews of the topics in social epistemology are Goldman and Whitcomb (2011) and Haddock, et al. (2010). 100 For a recent study of epistemic reliance see Goldberg (2010). 101 The method used by Descartes (1641/1996) in his Meditations is often cited as the classic example of how modern epistemology began with a focus solely on the individual knower. A complement to this approach is Locke’s rejection of testimony as a source of knowledge: ‘The floating of other men's opinions in our brains makes us not one jot the more knowledge, though they happen to be true’ (Locke, 1924, Bk I, Ch VIII, §23). 102 Some important early work that argued for this claim are Hardwig (1985) and Coady (1992). 103 The classic work on speech acts is Austin (1962). This was developed by Searle (1969, 1979).

79 state; and declarations – to bring about a correspondence between the propositional content and reality. What kind of illocutionary act is performed with testimony? Consider some examples of the propositions one person tells to another: ‘It’s sunny outside’, ‘Labour won the vote’, ‘I still have 2 chapters left to read’, ‘Processed meat is bad for your health’. In these examples, the speaker apparently commits herself to the truth of p. When I say that ‘Labour won the vote’, I am reporting that it is true that Labour won the vote. According to Searle’s taxonomy, therefore, when testifying, we perform an assertion. When S tells H that p, S asserts p to H since S is committing herself to the truth of p. Although I have argued that testimony involves performing assertions, it actually requires a specific kind of assertion. Testimony is an interpersonal means of putting propositions forward as true to other people. With testimony, we testify to a person. However, assertions needn’t be directed towards another person. Someone could assert by writing down in her journal facts that she hopes no one will ever read about. She is asserting, but not to another person.104 What would make her assertions testimony would be if the diary was written as a letter intended to be read by a hearer. This would maintain the interpersonal component required for testimony. We can distinguish the kind of assertion required by testimony by use of the expression ‘telling’. With a telling, propositions are put forward by speakers as true to hearers. With testimony then, S makes a particular kind of assertion to H: S tells H that p.105 Given that testimony involves telling, we can explain why testimony has an important role in the formation of the beliefs of other people. When a speaker asserts she presents a proposition as true. If we regard a speaker to be trustworthy, then her presenting p as true gives us a reason to believe p. We regularly believe propositions on the basis of a speaker’s testimony, particularly when we take the speaker to be trustworthy. Listening to a news report from a broadcaster whom we trust will often result in our forming beliefs about current affairs. Moreover, if we believe that the speaker actually knows p, then when we believe p on the basis of the speaker’s testimony, we usually take our belief that p to be knowledge that p. The fact that the assertions of others are used as a basis for forming one’s own beliefs makes assertion normatively significant. We generally want to avoid believing

104 See Lackey (2006) for a similar example. 105 See Fricker (2006) and Pelling (2014) for similar accounts of telling.

80 falsehoods, and so it may be impermissible for speakers to assert something if this could lead other people into believing a false proposition. Accordingly, it has been proposed that assertion is governed by a rule or norm that instructs speakers on when it is permissible to perform an assertion. One popular candidate is the knowledge norm (KN): one may assert p only if one knows p.106 A speaker who knows p when asserting p will avoid leading a hearer into believing a falsehood when the hearer bases her own belief on the speaker’s assertion that p.107 One means of defending KN is by looking at cases in which we can criticise someone for asserting p without knowing p. Suppose someone asks me for directions to the train station. I don’t know the correct way, but I make a guess, and assert that ‘the train station is next left’. Since I don’t know the way to the station I can be criticised for making this assertion because this is misleading for my hearer. The hearer might believe my testimony and follow my directions, but because my directions were given on guesswork rather than knowledge, the hearer might get lost and miss her train. Given the criticism attributed to me in this case, it looks like it supports the KN. Since I didn’t know the directions to the station, I was not permitted to assert directions to the train station, and am criticisable for doing so. Despite the plausibility of KN, there are a number of examples that suggest that KN is too strong. These sorts of cases relate to what Jennifer Lackey (2007) calls ‘selfless assertion’. For instance, she considers the example of a creationist teacher who does not believe that the theory of evolution is true, but sincerely teaches it to the children in her class because she recognises that there is a greater degree of evidence supporting the theory than against it.108 Her reasons for disbelieving evolutionary theory are disconnected from the evidence supporting the theory. She disbelieves the theory due to an emotional attachment she has toward a particular religious view of the world that conflicts with evolutionary theory. Nevertheless, the teacher recognises

106 The most highly cited instance of KN is given by Williamson (2000, chapter 11). The formulation I give here, using the permissibility mode ‘may’, follows various other accounts including Whiting (2015). 107 Following Williamson (2000, chapter 11) it has become commonplace to define assertion by the norms that constitute it. Defining assertion by its constitutive norms, it is argued, is analogous to defining a game by its constitutive rules. For instance, ‘the rule of chess that says you can’t castle if the king is in check is partially constitutive of the move of castling. A move that was not subject to this rule would not be castling’ (Macfarlane 2012, p.84). 108 This example was originally used in Lackey (1999).

81 why she disbelieves the theory, and acknowledges the significant evidence in favour of evolution, and so she teaches the theory as though it is true. Not only is the teacher not criticisable for asserting the main propositions in evolutionary theory without believing them, she is actually worthy of praise since she recognises that there are epistemic reasons to believe the theory, and asserts to others on the basis of these reasons. Lackey’s teacher example suggests that one need not actually believe p for it to be permissible for her to assert p. Assuming that knowledge that p entails belief that p, it follows that one need not know p for it to be permissible that she assert p. This means that KN does not hold. Examples of selfless assertion tell us that assertion is governed by a weaker norm than KN, which does not require that one believe p to assert p. A plausible alternative offered by Lackey is what she calls the ‘reasonable to believe norm’.109 It is reasonable for the creationist teacher to believe the central claims in evolutionary theory, and so she may assert p providing she knows what it is that makes it reasonable for her to believe p. The notion of ‘reasonable’ used here is synonymous with the justification I discussed in Chapter 1. For continuity in terminology, we can say that Lackey defends the justification norm (JN): one may assert p only if one has justification for p.110 Even though it is reasonable for the creationist teacher to believe the claims of evolution – she has justification for believing these claims – she doesn’t believe them for herself. Nevertheless, she may assert these claims in virtue of her justification. From JN we can see that the reason why people typically believe what they assert is because having justification tends to cause beliefs. However, as the example of the creationist teacher shows, one can have justification for p but lack belief that p, and sometimes this is due to non-truth conducive factors, like desire, that influence one’s belief-forming processes. Given what has been said thus far concerning testimony, I can now give an argument for the claim that I proposed to defend at the beginning of this section, that testimonial trust (TT) requires trusting a speaker for her competence and sincerity. The argument can be formulated when we see that for a speaker to satisfy JN requires two

109 For another defence of this see Kvanvig (2009). 110 This account is also consistent with other accounts of assertion, such as Brandom’s (1994, Chapter 3) theory in which the asserter is obliged to offer some justification for the asserted proposition if there is some dispute over it. See Macfarlane (2012) for an overview, and other theories that are also consistent.

82 things of the speaker. First, it requires epistemic competence, which is the competence to acquire the justification for the proposition you are telling. Second, it requires sincerity, which is a matter of not misleading others by being dishonest, and telling p to someone only when you have justification for p. We can now state the argument as follows: (1) speakers are required to tell only what they have justification for (JN). (2) Telling only what you have justification for requires competence and sincerity. It follows from (1) and (2) that (3) speakers are required to tell with competence and sincerity. Add that (4) when we trust what speakers tell us we trust that they are only telling what they have justification for (that they satisfy JN). It follows from (1)-(4) that (5) when we trust what speakers tell us we trust that they are telling with competence and sincerity. The conclusion (5) tells us what testimonial trust is in: the speaker’s competence and sincerity as a testifier. Taken together with my analysis of trust, we can derive the following account of testimonial trust:

TT: When a hearer H trusts the testimony of a speaker S, H: (a) relies on S to tell with competence and sincerity, (b) believes that S is committed in telling with competence and sincerity, and (c) believes (and is optimistic) that S is trustworthy with respect to telling with competence and sincerity.

With this analysis of TT in hand, I want to be more precise about the relationship between TT and the hearer’s beliefs. A hearer is presented with several options when told p: she may believe, disbelieve, or suspend judgment regarding p. I want to show how these three states constitute one’s trusting or distrusting response to the testimony. My aim is to defend the claim that to trust a speaker’s testimony the hearer must believe the propositions the speaker has asserted. I will outline two arguments in favour of this claim from my account of TT. The first argument relies on the claim that when telling someone p, the speaker puts forward p as true. Why would someone want to perform a telling? Why would I want to tell a friend that, for instance, ‘processed meat is bad for your health’? A plausible answer is that I want my friend to come to believe this proposition. The means I have chosen for getting my friend to believe p is by putting forward p as true by testifying that p. Consider how this is put by Elizabeth Fricker:

When a speaker S asserts that P to an audience or hearer H…she thereby vouches for the truth of P to H. She presents P as being so, in an act whose

83

import is that H can form belief that P on her say-so – H’s eventual belief that P will be justified by reliance on S’s word. The ‘bottom line’ of this import of S’s act, is that H can complain to S if P turns out to be false. (Fricker 2006, p.594)

As Fricker puts it, the speaker ‘vouches for the truth’ of p. She is doing so in order that the speaker will come to believe p as well. Suppose the speaker does not come to believe p. I tell my friend that eating processed meat is bad for her health, and she either disbelieves me, or suspends judgment regarding this proposition. Clearly my friend does not trust my testimony.111 Perhaps my friend believes I am being insincere to prevent her from enjoying some of her favourite foods. Or she may think that I lack some epistemic competence with regard to this proposition because I’m not known for being knowledgeable on nutritional issues. In either case, we can explain her lack of trust in terms of a failure to satisfy (c) of TT. Let’s now consider the second argument. According to (a) of TT, H relies on S to testify with competence and sincerity. I argued in §2.3.2 that to rely on someone to  involves entrusting some good to that person, and in so doing, to be vulnerable should the person not . To rely on the sincerity and competence of a speaker, I propose, a hearer must entrust her doxastic states to the speaker. This entrusting is a matter of risking the beliefs the hearer forms on the basis of the speaker’s testimony, and being vulnerable in case the proposition believed turns out false. By judging the speaker to be asserting sincerely and competently, the hearer allows herself to believe what the speaker says on those grounds – on the belief that the speaker is both sincere and competent. If she is, then her beliefs should be in safe hands. However, this is a risk, hence the need for trust. Why does this argument rule out the possibility that one can merely non- doxastically accept the testimony without believing it and have that response be sufficient for trust? Because, if the hearer doesn’t believe, but merely accepts, then she doesn’t entrust anything to the speaker. She doesn’t put her beliefs at risk, she doesn’t make herself vulnerable should things turn out bad. Her acceptance keeps herself at arm’s length from the speaker, not allowing herself to entirely rely on the speaker. In effect, the acceptor fails to meet condition (a) of TT: she doesn’t rely on the speaker. The argument, in short, is this: (1) For H to trust S, H must rely on S; (2) For H to rely

111 See Faulker (2007, p.894) and Keren (2014, p.2598) for the source of this argument.

84 on S, H must believe S (because reliance requires risk and H only risks if H believes); (3) Therefore, for H to trust S, H must believe S. I take these two arguments to support the claim that to trust a speaker’s testimony the hearer must believe the propositions the speaker has asserted. This will be a central feature of my account of Abrahamic faith. With accounts of trust in general and testimonial trust in particular in hand, I want to apply these to a religious context, to explore their role in Abrahamic faith.

2.3.4 Trust and Testimony in Abrahamic Faith

I argued in §2.3.1 that Abrahamic faith requires relational faith, and that relational faith is a matter of trusting others. Abrahamic faith in God, therefore, requires trust in God. There is no reason to think that the trust required by Abrahamic faith should diverge from the general account of trust given in §2.3.2, except in terms of the person trusted (God) and the actions relied upon. These actions will be quite distinctive for religion since God is considered able to perform actions that humans cannot, for instance, the granting of eternal life and the answering of prayers. People rely on God for divine protection, provision of physical and spiritual needs, and their future destiny. But importantly, and I feel this is often overlooked, religious people rely deeply on God for their religious beliefs about, for instance, who God is, and the sorts of things God plans to do in the future. The source of beliefs concerning these matters can be what one perceives to be divine revelation. People perceive such revelation to come in the form of propositional claims made in scripture, for one instance. Since the kinds of propositions believed on the basis of scripture include those concerning the nature of God, the afterlife, and God’s plans for the world, they are not easily, if at all acquirable, by use of reason alone. Hence, people must use what they perceive to be the very words of God to have access to these kinds of propositions. People also take God to speak through various events that happen in their lives. For example, people sometimes take coincidences, such as running into an old friend, as testimony from God that they ought to look out for that friend. It is also commonplace for religious people to feel as though they commune with God through prayer and religious

85 experience, and this gives them guidance on how to live, and practical advice on what to do with certain decisions and situations they face.112 The purpose of these observations is to show that not only is trust in general central to religious faith, but so is TT. This is, in part, the impetus behind the divine testimony theory of faith. The examples of trusting God for what one takes God to have said simply look like examples of TT where God is the speaker. In fact, even general trust where God is trusted to perform some relied upon action seems to presuppose TT. This is because when one trusts God to , presumably that person believes that God has committed to -ing. But how could one presume this if she doesn’t perceive that God has testified in some way to -ing? A similar point is made by William Alston who says that trust in a person is

…reliance on the person to carry out commitments, obligations, promises, or, more generally, to act in a way favorable to oneself. I have faith in my wife; I can rely on her doing what she says she will do, on her remaining true to her commitments, on her remaining attached to me by a bond of love. (Alston 1996, p.13)

We rely on people to do what we perceive them to have committed themselves to doing, and in many cases, this will involve them saying they will do that thing. Could someone trust God to  without taking it that God has ever said that he will ? Such a view can be extracted from the account of religious trust in God given by Richard Swinburne:

[Religious trust] is presumably to act on the assumption that He will do for us what He knows that we want or need, when the evidence gives some reason for supposing that He may not and where there will be bad consequences if the assumption is false. (Swinburne 2005, p.143)113

According to Swinburne, trust in God merely requires that one act on an assumption that God will  where -ing is a matter of God doing for us ‘what He knows that we

112 People can then respond to this perceived communication through commitment and obedience to the claims and commands God has made to them. Some theorists have thought that these two components – commitment and obedience – are features of faith itself. See Bishop (2007) on ‘commitment’, and Evans (1998, p.4) and Wood (2014, p.35) on ‘obedience’. 113 Schellenberg reiterates this definition in his account of trust (2005; p.115).

86 want or need’. But if this action does not involve regarding God to have actually committed to this course of action through telling the trusting party that he will, then this account seems quite unusual. To act on an assumption in this way would then be akin to a wish that a person has that there is some God and that this God will do something, even though the person is unsure as to whether there is a God who has made such a commitment. I might wish that God would make me a millionaire, even though I wouldn’t ever claim that God told me he would. To then say I trust that God will make me a millionaire sounds inappropriate. The reason for this is simple to explain on the account of trust I have given: because it would actually fail to meet condition (b), and hence, would fail to be an instance of trust. This condition was essential for distinguishing trust from mere reliance, and will be required to do the same if we are to account for trust in God. The kind of trust required for religion can be trusting God to perform certain actions. This presupposes that God has committed to performing those actions, and this can be a matter of saying that he will. One can also trust God by trusting certain assertions God makes about, for instance, himself, future events, or the nature of the afterlife. Both trust in general and TT are involved in these two cases of trusting God to perform certain actions and of trusting God’s assertions. Relational faith in God requires trust in God, which in turn requires both that someone have general trust in God to perform certain actions and trust in God’s testimony.

2.3.5 Relational Faith as the Cause of Propositional Faith

The account of Abrahamic faith thus far has focussed largely on relational faith. This has been interpreted through my account of trust. But in §2.3.1, I argued that faith that – propositional faith – is distinct from relational faith, even though Abrahamic faith requires both kinds. Despite this, it’s relatively straightforward to see how they might be connected, and this gives us some notion of what propositional faith is. First, trust contains certain attitudinal components: a belief that the trustee is committed to -ing, and a belief that she is trustworthy with respect to -ing. Hence, when one has faith in someone in such a way that she trusts that person, it follows that she also has faith that (1) the trustee is committed to -ing, and that (2) she is trustworthy with respect to - ing. So, according to the account of trust I have given, having faith in someone involves having faith toward several propositioms: relational faith requires

87 propositional faith. The propositional attitude component to trust is, however, quite limited in scope since it only covers propositional faith toward propositions (1) and (2). But surely we can have faith toward many other propositional contents and it’s worth having a separate account of the connection between relational and propositional faith that allows for this fact. The second way to make the connection is to see relational faith as a cause of propositional faith. Our propositional attitudes are formed in us in numerous ways. For instance, my fear that the aeroplane will crash could be caused in me through watching too many negative TV reports. My belief that our team will win the game could be caused through their excellent performance so far. My desire that I recover from illness could be caused through a combination of emotions, such as a love of life and a fear of losing it. As I argued with TT, one can also acquire belief that p through trusting a person for her testimony. Testimonial trust is another cause of propositional belief. Now, allowing that TT is required by relational faith, it seems plausible that one could form faith that p by trusting – or having faith in – the testimony of another. For instance, if I have relational faith in someone for her testimony, the outcome of this faith is faith that p. Say I have relational faith in a friend when she tells me that she is innocent. My faith could be manifest in trusting her testimony that she innocent, and the outcome – that which relational faith causes – is faith that she is innocent. The claim that relational faith is the cause of propositional faith can allow for propositions across many different contents, and isn’t restricted in the first way of connecting the two kinds of faith. My claim here is not that faith that p is always caused by having faith in someone by trusting what she says, but simply that this cause is commonplace in Abrahamic faith. Consider an abductive argument in favour of this claim. Oftentimes, the PFR is construed as a conflict between beliefs held on reason alone and those adopted from supposed faith in divine revelation.114 Recall that this is how Swindal

114 Some authors have defended a theory of faith according to which faith is believing a person regardless of whether the context is religious or not. In her paper entitled What is it to believe someone?, G.E.M Anscombe remarks that ‘…I might have called my subject ‘Faith’. That short term has in the past been used in just this meaning, of believing someone…This old meaning has a vestige in such an expression as ‘You merely took it on faith’ – i.e., you believed someone without further enquiry or consideration…At one time there was the following way of speaking: faith was distinguished as human and divine. Human faith was

88 characterised the problem in the quoted passage in §1.5.3. Indeed, several of the other accounts of faith discussed in §2.2 also fit into this model, including fideism and the ‘belief without sufficient justification’ model. The latter account was found, in Chapter 1, to have a number of significant supporters. Now, let’s bear in mind another putative feature of propositional faith, namely, that faith involves resistance to counterevidence.115 Take, for instance, such logical problems as The Paradox of the Stone.116 Why doesn’t this logical problem typically cause religious people, who are confronted with it, to refrain from viewing God as omnipotent? One explanation is that these people perceive God to have told them that he is omnipotent, and this proposition is believed by trusting God’s testimony. When counterevidence then presents itself, this is put aside for the reason that God’s testimony is favoured over it. The kind of response the religious believer is involved in with respect to divine testimony is sometimes referred to by epistemologists as believing for preemptive reasons.117 These kinds of reasons function similarly to commands: they give us high- order reasons to act in a way that displaces other reasons we may have. For example, an army private might have a number of reasons to refrain from engaging the enemy in combat, but because he has been commanded by a senior authority to engage them he puts aside these conflicting reasons and acts on the command. Analogously, one might preemptively believe those propositions one perceives to come from God because she considers God to be an epistemic authority on the proposition in question. This would involve discounting evidence contrary to the proposition thought to be testified to by God as merely misleading. 118 We can summarise the abductive argument I’m giving here as follows: propositional faith is resistant to counterevidence, and one explanation for this is that faith that p is often caused by trust in God’s own testimony. When people trust God’s testimony they recognise that God is an epistemic authority, and take this into account believing a mere human being; divine faith was believing God.’ (Anscombe 1979, pp. 141- 42) 115 Some recent defences of this claim include Buchak (2017b), McKaughan (2017) and Howard-Snyder (2017). 116 For instance, see Mavrodes (1963). 117 See Raz (1978), Zagzebski (2012) and Keren (2014). 118 Compare this with simply viewing faith as self-deceptive (see Bayne 2009, and Mele 2009, for accounts of what this might be like). The preemptive reasons account gives a normative epistemic reason for discounting contrary evidence, whereas in self-deception, the reasons for putting aside conflicting evidence are always epistemically vicious.

89 when retaining their propositional faith when faced with counterevidence. I take this argument to support the claim that when relational faith involves trusting God’s testimony, then faith that p that can be caused by, and sustained through this trust. By viewing faith this way, we can also see why many regard faith to be in conflict with reason – because faith is apparently resistant to counterevidence. I will revisit the issue of preemptive reasons and their role in religious epistemology in Chapter 5. There is another important explanation, though, for why propositional faith in the religious domain is considered epistemically dubious: because the content of the proposition is, in many cases, not easily believable. Without this additional explanation, it’s difficult to see how relational faith in God and propositional faith can come apart for religious faith since we would lack an explanation for why religious faith has received so much criticism. For instance, if propositional faith is merely belief plus perhaps some evaluative component (to be addressed presently), why should it be so commonly thought of as irrational? If it’s because faith is grounded in testimony, that suggests that the cause of the propositional attitude is questionable. But it needn’t be grounded in this source, so we need the further explanation to contend with the widespread concern that religious faith is not epistemically justified. This also lends understanding for why faith that, say, my spouse will remain constant, or my team will win the match, are not propositions targeted for regular criticism, where religious propositions are: it isn’t so much the attitude of faith that is criticised, but the content of the attitude. Nevertheless, this has naturally turned attention towards the source of such content, which as I suggest, is regularly grounded in revelation. So, whilst faith that p need not be grounded in trust in God’s testimony, it commonly is for Abrahamic faith, which has naturally led to criticism over the source of propositional faith. This takes us to the final question for this chapter: What is propositional faith?

2.3.6 Propositional Faith

Many philosophers and theologians have taken propositional faith to be an attitude that is at least partly constituted by belief.119 Faith that p would then require belief that

119 For instance, Augustine (1999), Aquinas (1948, II.II), Locke (1924), Berkeley (2007), MacDonald (1993), Evans (1998), Plantinga (2000, chapters 8 and 9) and Swinburne (2005, pp. 138-148).

90 p. This view is both historically orthodox, and widespread in folk intuition.120 There are several arguments in favour of this claim. First, it seems plausible that faith that p involves representing p as though it is the case that p. For instance, if I have faith that Theresa May will make a good Prime Minister, intuitively I represent the world as being a certain way – one in which it is true that Theresa May will make a good Prime Minister. This implies that propositional faith is constituted by a cognitive attitude. Clearly, belief can be the cognitive attitude that partly constitutes propositional faith. When I have faith that Theresa May will make a good Prime Minister I can believe this proposition. Of course, there are other attitudes that are cognitive but not doxastic, and Chapter 3 will evaluate whether this attitude can be non-doxastic. At that point I will discuss arguments in favour of the claim that the cognitive component to propositional faith need not be satisfied only by belief, and so defer discussion of this claim until then. A second argument in favour of a belief requirement is derived from my reflections in the foregoing, that faith is often seen to be epistemically defective, but for this to be the case, the content must be believed. If propositional faith does not require belief, why have we had two thousand years of the problem of faith and reason? Well, perhaps all that we need is the weaker condition that faith typically involves belief. But then we will return to the previous issue of whether a non-doxastic cognitive attitude can stand in for the cognitive component of faith, where this will be addressed in Chapter 3. The third argument has been discussed in the latter parts of this chapter: propositional faith can be caused by TT, but the resulting attitudes of TT are doxastic, so propositional faith, in the context of Abrahamic faith, is at least partly constituted by belief. The force of this argument stands or falls with the plausibility of its premises which have been discussed at length, so I won’t address it any further. There are a number of other arguments in the literature that favour this doxastic condition,121 but I take these three to suffice for now until the next chapter. According to this orthodox account of faith, then, S has faith that p only if S believes that p. Although belief is then necessary for faith, it is rarely, if ever, thought to be sufficient for faith. Propositional faith is typically regarded to be a complex

120 A similar point is made by Howard-Snyder (2013a) who labels this ‘the common view’. 121 Howard-Snyder (2016) evaluates seven arguments for the doxastic condition.

91 propositional attitude with a positive evaluative component. A standard religious argument for this takes a cue from the biblical passage ‘You believe that God is one; you do well. Even the demons believe—and shudder!’ (James 2:19): demons believe that there is a God without the appropriate preferences and feelings needed for faith. This passage led Thomas Aquinas to say that faith involves ‘a certain affection for the good’ even when faith is ‘lifeless’ (1948, II.II.5.2). Alvin Plantinga elaborates: the person of faith

…not only believes the central claims of the Christian faith; she also (paradigmatically) finds the whole scheme of salvation enormously attractive, delightful, moving, a source of amazed wonderment. She is deeply grateful to the Lord for his great goodness and responds to his sacrificial love with love of her own. The difference between believer and devil, therefore, lies in the area of affections: of love and hate, attraction and repulsion, desire and detestation. (Plantinga 2000, p. 292)122

On this account, propositional faith involves a positive evaluation that is grounded in the affections. Perhaps these affections, like love and gratitude, are part of what it is to have relational faith, and the positive evaluative component to the attitude is held in virtue of these affections. However, I have been claiming that relational faith and propositional faith can come apart, so we would then require an explanation for the evaluative component when they do come apart. Moreover, one needn’t have affection for the object of the proposition to evaluate it positively. For instance, I needn’t have any affection for politics to think that politics can be a force for good, I could simply regard politics as a good thing. So, whilst Plantinga’s account of the positive component to faith is a likely description of the life of many who are religiously committed, propositional faith can have a positive evaluation without the need to draw from the affections. It could simply involve an evaluative belief. There are other prominent arguments for the evaluative component as well that apply to faith regardless of the content. The first of these are often developed from language. To some, it seems incongruous to talk of ‘faith’ without there being a positive evaluation of its object. As Walter Kaufman notes, ‘One cannot say, without doing violence to language: “I have faith that I have cancer”’ (1958, p. 113). The same seems true if I say I have ‘faith that terrorism will succeed’ or ‘faith that we will lose

122 See also MacDonald (1993, p.44).

92 the battle’. This is because our use of ‘faith’ in these expressions typically conveys a positive evaluation if its object, and these are not propositions we regularly evaluate positively. According to another argument, given by Howard-Snyder, faith must be more than belief because

…one can believe something even though one has no tendency at all to feel disappointment upon learning that it’s not so, but one cannot have faith that something is so without at least some tendency to feel disappointment upon learning that it’s not so. That’s because one can have faith that something is so only if one cares that it is so. (Howard-Snyder 2013a, p.360)

We could formulate this argument as follows: (1) If one merely believes p then it’s not the case that one tends to feel disappointed upon learning not p; (2) If it’s not the case that one tends to feel disappointed upon learning not p, then one does not have faith that p; (3) So, if one merely believes p, then one does not have faith that p. The first premise is highly intuitive. There are all sorts of propositions I believe that I don’t care at all whether they are true or false. I believe the book in front of me is hardback, but I don’t care that it’s hardback. I would be just as happy with a paperback version. The second premise relies on two claims that are also equally plausible, first, that one can be disappointed that not p only if one cares that p, and second, that if one has faith that p then one is disappointed to hear that not p. From here you derive the claim that faith that p requires caring that p. This seems correct when we think that to have faith that Theresa May succeeds, that we win the race, that our spouse remains faithful, and that God exists, we care about these things. Finally, caring that p seems sufficient for evaluating p positively. It follows that propositional faith requires a positive evaluation of the proposition, but belief is not sufficient for evaluating a belief positively. On the basis of these arguments it seems that S has faith that p only if S evaluates p positively. This may be held in virtue of one’s affections, one’s general caring attitudes, or it may simply be an evaluative belief. According to the two conditions thus far defended, then, we can formulate what I will call the belief plus (B+)123 theory of propositional faith:

123 This expression is borrowed from McKaughan (2013).

93

B+: A person S has faith that p iff S (a) believes that p, and (b) evaluates p positively.

This concludes the account of propositional faith and its potential connections to relational faith. I want to finish this chapter by summarising what has been said so far. From there we can go on to evaluate DTT and NDT as approaches to responding to the PFR.

2.4 Conclusion

In §2.2 I argued that in order to analyse faith, one must first specify the kind of faith to be analysed. I stated that I wanted to focus on what I called Abrahamic faith, in which we found that faith has both a propositional and a relational dimension. The latter is largely a matter of placing trust in God concerning various actions that we believe that God is committed to performing. One of these, of course, is to speak to us concerning various things with both competence and sincerity. To do so is for God to engage in divine testimony, and this provides the connection to the divine testimony response to the PFR. So, relational religious faith is simply to trust in God. Propositional faith is a complex attitude constituted by both belief and a positive evaluation of its object. In §2.3.5 I attempted to draw together the account of relational faith with propositional faith. I argued that a commonplace way by which one comes to have faith that p is by trusting God’s testimony. In this case, relational faith is the cause of propositional faith. This connection is not necessary, but I propose that it’s widespread within Abrahamic faith, and that it offers an explanation for the motivation behind the PFR. The account of Abrahamic faith that I have proposed, in which there is a causal connection between relational and propositional faith, is one way of arguing for the divine testimony approach to addressing the PFR outlined in Chapter 1. The epistemic status of DTT will be the subject of Chapters 4 and 5. For now, though, let’s turn to the NDT approach of responding to the PFR since, as we will see, NDT does so by rejecting condition (a) of B+.

94

3. NON-DOXASTIC FAITH

3.1 Overview

This chapter outlines and evaluates the theory that provides the first of the two responses to the problem of faith and reason (PFR) – the non-doxastic theory of faith (NDT). This theory maintains that an individual need not believe those propositions she has faith toward. That is, she can have faith that p without belief that p. When used as a response to the PFR, the advocate of NDT can avoid the charge that propositional faith is irrational by allowing that degrees of belief can adjust in a way consistent with what is rationally required, even though this may cause confidence in a proposition to drop below what is required for belief. When this happens and someone with faith that p does not believe that p, NDT maintains that she need not lose faith that p. Using NDT to respond to the PFR will only be an option if the theory itself is defensible. I will argue that NDT is not defensible, and so it fails as a response to the PFR. This is because the arguments used to support NDT are untenable, and that NDT allows for a kind of pretence faith. I will argue that since NDT is consistent with pretend faith, this shows that NDT is false or at best incomplete. In §3.2 I will give a brief overview of NDT, stating a set of necessary and jointly sufficient conditions for the theory, before moving on in §3.3 to evaluate the four most widely used arguments in favour of NDT. Each of these, I will propose, fails to persuasively establish that faith can be non-doxastic. In §3.4 I will consider the important objection against NDT that it is compatible with a kind of pretend faith. The most promising response that can be given to this objection, I will argue, grants the advocate of NDT a mere Pyrric victory: it does significant damage to NDT itself, even though it overcomes the objection. Given this, and the failure of the arguments in favour of this theory, I will conclude that we have sufficient reason to endorse a B+ account of propositional faith, and that we cannot overcome the PFR by appeal to non- doxastic accounts of faith.

3.2 The Non-Doxastic Theory124

124 Parts of this section were first published in Malcolm and Scott (forthcoming) and Malcolm (forthcoming). For this thesis I have rewritten and developed on this earlier work.

95

Non-doxastic theories of propositional faith are committed to the claim that faith that p does not require belief that p.125 126 This claim entails that B+ is incorrect since condition (a) is false. To see how this claim underpins NDT, let’s begin with a few theories that state their commitment to it. According to John Bishop’s (2007) ‘doxastic venture model’, faith requires ‘an active venture in practical commitment to the truth of faith-propositions’ (p.106). On Bishop’s theory, therefore, to have faith that, say, Jesus rose from the dead, or I will make it through my exams, or my marriage will last, is to be practically committed to the truth of one of these propositions. This will involve acting as though the proposition is true, mentally assenting to it, saying that it is the case that p when asked whether p, and so forth. Bishop argues that if people can be committed to faith propositions ‘without actually believing the truth of the faith‐ propositions on which they act, then it ought to be conceded that they may indeed be making an authentic venture in faith’ (p.119).127 According to Bishop, then, faith that p requires a practical commitment to the truth of p, which one can make if one believes p or if one does not believe p. But how can one make this commitment to p without believing p? Let’s consider a second example of NDT for an answer to this question. Richard Swinburne has defended a version of NDT, for which he argues that

…the faith needed for religion is basically a commitment to seek a goal…If you want the love of God for yourself and your fellows enough, you need to believe that there is quite a chance that there is a God and that it is more probable that you and your fellows will reach Him by following the Christian way (and assuming, not necessarily believing, the claims of the Creed) than by following any other way. (Swinburne 2001, pp. 211-212)

For Swinburne, religious faith is a commitment to pursue certain goals, such as the attainment of the love of God. Part of reaching that end, he claims, is by assuming propositions concerning God as outlined in a religious creed. For example, one could

125 Perhaps the accounts that most straightforwardly deal with propositional faith are Howard- Snyder (2013a; 2016b), Buchak (2012; 2014a), Schellenberg (2005; 2014), Alston (1996) and Audi (2011), although this list is not exhaustive. 126 There are also non-doxastic theories of other kinds of faith. For instance, Jonathan Kvanvig’s account defines faith as ‘a disposition to behave or respond to the circumstances of life in service of an ideal’ (forthcoming, p.14). He maintains that this need not involve belief, but also, that it need not involve a propositional attitude. 127 For similar proposals see the fiducial accounts given by Bishop (2014) and McKaughan (2013).

96 assume that God is all loving, which is found in several Christian creeds, and assuming this proposition may be able to help you attain the love of God for yourself. Faith that p, for Swinburne, can be to assume that p in order to attain some goal. Note that Swinburne does not require that belief constitute propositional faith; it is only required that one assume p to have faith that p. We can now say how one can make a commitment to p without believing p: one can commit to p by assuming p even if one does not believe p. Assuming is one kind of cognitive attitude, along with belief, that I discussed in §1.2. Some defences of NDT have stated explicitly that faith that p requires a cognitive attitude, but does not require that this attitude be any specific kind of cognitive state. For instance, Daniel Howard-Snyder’s account maintains that faith that p is constituted by ‘a positive cognitive stance toward p’ (2013, p.367).128 Both assumption and belief are positive cognitive stances one can take towards a proposition. Some other suggestions for this positive cognitive stance have included imagination (Schellenberg 2014), acceptance (Alston 1996), acquiescence (Buchak 2012), and planning states (Kvanvig 2013). According to each of these accounts, propositional faith requires commitment to the truth of a proposition, in the way that Bishop and Swinburne have proposed, and one can make this commitment by holding or taking on one of these cognitive attitudes towards the proposition. A typical account of NDT will state that faith that p requires a positive cognitive attitude toward p. One way to characterise the nature of a positive cognitive attitude such as acceptance, assumption, or assent, is that to accept that p, for instance, is to take p as a premise in one’s practical reasoning. As such, one acts as if p, adopting a policy of going along with p in some or all of one’s deliberations (see Cohen 1992, p.4). For instance, if I have a positive cognitive attitude toward the proposition that I will win the race, then I will view this proposition positively, and by accepting, or assuming, or supposing that I will win the race, I will be disposed to work the proposition into my practical reasoning by going along with it in my deliberations. This ought to lead me to act as though I will win the race by, for instance, signing up to compete, expecting I will win, and competing as hard as possible.

128 For other explicit endorsements of faith that p requiring ‘a positive cognitive attitude’ see McKaughan (2013) and Schellenberg (2005).

97

Characterising cognitive attitudes in terms of their role in practical reasoning is helpful for understanding non-doxastic attitudes that are adopted voluntarily. In §1.2 I explained how acceptance, assumption and the other non-doxastic attitudes share the same features of belief, where the difference between them and belief is that belief is non-volitional, whereas non-doxastic cognitive attitudes are volitional. The volitional nature of the non-doxastic cognitive attitudes, however, makes them importantly different from belief in terms of their relationship to practical reasoning. Accepting p requires working p as a premise into some or all of one’s practical reasoning. The same isn’t true of believing p. I can believe that the time is 10.30am but act on the basis that its 2pm should I wish to. Even though belief as a doxastic state is a positive cognitive attitude, it is not so closely tied to practical reasoning as the volitional non-doxastic cognitive attitudes. From what has been said of cognitive attitudes, we can see that non-doxastic theories of faith maintain that (a) in B+ is too narrow. They claim that (a) should be replaced by a broader cognitive condition according to which a person S has faith that p only if S has a positive cognitive attitude toward p. This condition can be satisfied by belief, but it can also be satisfied by a variety of non-doxastic attitudes as well.129 Since §3.3 will outline the arguments in favour of this first condition, I will defer discussion of them until then. Although NDT defends the legitimacy of propositional faith without belief, advocates of NDT maintain that disbelief that p is incompatible with a positive cognitive attitude that p.130 If one suspends judgment concerning p or one believes p, one can have faith that p, but one cannot have faith that p if one disbelieves that p. Consider this as stated by Audi:

If I believe that not-p, surely I cannot have faith that p, just as I cannot (at least normally) believe both that p and that not-p. I can have such faith compatibly with an absence of any feeling of confidence regarding p, and even with a belief that p is not highly probable. But if I disbelieve p, I do not have faith that it is so. (Audi 2011, p.73)

129 One exception to this was the early work of J. L. Schellenberg (2005, p.132). Given objections from Howard-Snyder (2013b), Schellenberg has since qualified his view on this (2013). I will discuss this further in §3.3.3. 130 Howard-Snyder (2013a) discusses this in moderate detail.

98

Audi seems correct to say that belief that p is incompatible with belief that not-p. But what kind of incompatibility is he referring to? Presumably, he means rational incompatibility. It would not be rational to believe p whilst concurrently disbelieving p. Is faith that p incompatible with belief that not-p in this same way? This would be the case if faith that p required belief that p, as B+ maintains. But Audi is a defender of NDT, as is made clear when he claims that ‘propositional faith need not be doxastic’ (2011, p.73). He endorses the view that accepting p is sufficient for the positive cognitive attitude required by propositional faith. How might, then, accepting p be incompatible with disbelieving p? Presumably, the incompatibility referred to here is pragmatic rather than rational. The reasoning behind this could be that non-doxastic cognitive attitudes are adopted in order to accomplish your practical aims, but it’s not possible to achieve those aims if you disbelieve that they can be accomplished. Suppose I accept that I will do well in an interview to motivate myself to do well. One might argue that if I disbelieve that I will do well, then I cannot accept that I will do well. This example would then suggest that acceptance and other non-doxastic attitudes are practically incompatible with disbelief. However, this claim seems dubious. For instance, imagine that a lawyer believes her client is so obviously guilty that he stands no chance of being found innocent. In order to put her doubts aside and make a compelling case for his innocence she accepts that he’s innocent as a way of motivating herself to defend him passionately in court and to pursue all the available avenues of evidence proving his innocence. This looks both psychologically possible and prudentially advisable.131 It looks as though a non-doxastic cognitive attitude is practically compatible with disbelief in the same proposition. One can accept, assume, imagine, or assent to p as a means to achieve one’s aims, even if one disbelieves p. However, it’s regularly stated by advocates of NDT that propositional faith excludes disbelief. Given that my aims here are to faithfully represent this class of theories, I suggest that this claim is stated as an explicit condition within the analysis, rather than being entailed by the first cognitive condition. So, the second condition advocates of NDT ought to state is that a person S has faith that p only if S has no outright disbelief that p. Since I have rejected one argument in favour of this condition, it currently stands undefended, and

131 This objection is also raised by Malcolm and Scott (forthcoming), and Malcolm (forthcoming).

99

I am aware of no further argument for it that exists in the NDT literature, despite the fact that this condition has received widespread endorsement.132 As we will explore further in §3.3, whether the condition is defensible is crucial for determining who has faith. The two conditions outlined thus far are distinctly cognitive in nature, but propositional faith, as we saw in Chapter 2, is also thought to involve various non- cognitive components as well. The arguments for this are used by both advocates of NDT and B+ alike and so need not be reiterated here (see §2.3.6 for discussion of this argument). Suffice it to say that NDT is committed to the condition that a person S has faith that p only if S has a positive evaluative/affective attitude toward p. I take the three conditions thus stated to be a fair representation of the conditions stated for propositional faith in a broad range of non-doxastic theories. According to NDT, then, a person S has faith that p iff S has:

(a) a positive cognitive attitude toward p; (b) no outright disbelief that p; (c) a positive evaluative/affective attitude toward p.

Note that this analysis accounts for propositional faith regardless of the domain, be it religious or not. So, one’s faith that democracy will succeed, faith that the marriage will last, or faith that we will make it home safely are all instances of the same kind of propositional attitude. NDT is therefore theoretically parsimonious – something often regarded as a theoretical virtue. When outlining (a) we saw that it can be satisfied either by belief, or by a volitional non-doxastic state. In the former case, if we want to know why someone has come to have the belief, we might look to evidential reasons for an explanation. With a non-doxastic attitude, though, one’s practical concerns also give reasons bearing on why someone would choose to adopt a positive cognitive attitude towards p, even when she does not believe p. Consider the earlier example of the lawyer. She accepts that her client is innocent as a means to achieving the end of effectively defending his innocence. It’s widespread amongst accounts of NDT to draw analogies between examples of means-ends reasoning and propositional faith. For instance, Alston (2007,

132 In fact, the recent article by Howard-Snyder (2016b) reiterates his support for this condition, without offering argument in its favour. He does give some arguments for it in his (2013a) though, which are similar to the argument addressed in this section.

100 p.133) famously draws comparison between his account of faith and an army general who, without having all the necessary information to hand to accurately disperse his troops, accepts the locations of the incoming enemy soldiers as a means to direct their movements. In the same way, Howard-Snyder (2013, p.364) imagines a backpacker, set out on a 2500 mile hike, who assumes that if he keeps going, he will eventually reach home, and does so as a means to motivate himself to get there. Both cases involve means-ends reasoning, and hence are pragmatic in nature and are supposed to display examples of propositional faith. The role of practical, means-ends reasoning in NDT is not explicitly stated in any of (a)-(c), but is in the background to how faith is construed on these accounts: it is part of the description of propositional faith, rather than being a prescription on it. In §3.3.4 I will discuss further how the notion of having faith without belief in this pragmatic way is both used as an argument in favour of NDT, and yet causes serious problems for the theory. Stating this condition clearly within the analysis will mark one attempt to rescue NDT from an important objection. I want to move now to consider the leading arguments in favour of condition (a). If this condition can be defended, then NDT can offer a legitimate alternative response to the PFR. If not, this response will not be an option. As things currently stand, I don’t think that it will be, and I will try to show why by considering and rejecting four arguments in favour of NDT.

3.3 Four Arguments for the Non-Doxastic Theory133

3.3.1 Doubt

Many people with faith suffer from some sort of doubts regarding the object of their faith at some time in their lives. Someone with faith that the labour party can bring about positive change may be led to doubt this proposition at some time or other. The same is true for someone with faith that their business will stay afloat, or faith that their spouse will not leave them. Doubt towards the object of our faith simply seems to be part of what it is to have faith. This claim seems equally true of faith towards religious entities. Observing the suffering and apparent evil in the world, or the

133 Discussion on three of these arguments was first published in Malcolm and Scott (forthcoming). For this thesis they have been significantly expanded upon, and a fourth argument has been added and evaluated.

101 supposed silence of God, or the great deal of religious diversity, often leads people to question their faith. This is part of the reality of those who are religiously committed. This data is sometimes used as a premise in an argument against B+ that I will call the argument from doubt. To my knowledge, this argument goes back at least as far as the work of Louis Pojman (1986),134 who, defining doubt as ‘the absence of belief’ (2001, p.137), maintained that doubt is compatible with faith. If Pojman is correct, and doubt is compatible with faith but that doubt involves a lack of belief, then it follows that faith can be held without belief.135 To give this argument a little more context, consider the way it is put in a recent article by Daniel McKaughan:

Faith is clearly not incompatible with a persistent sense of uncertainty, dark nights of the soul, or a pervasive sense of the hiddenness of God…If deep, sincere, and wholehearted faith coexists with doubt in the lived experience of many religiously committed persons and can do so in a relatively stable way despite fluctuating levels of confidence, surely this fact is one that any adequate account of faith ought to be able to accommodate. Belief-plus models define faith in such a way as to preclude significant doubt, yet faith appears to be compatible with doubt in a way or to an extent that belief is not. (McKaughan 2013, pp.106-07)

These sentiments are influenced directly by the work of Pojman, but Pojman’s and McKaughan’s remarks merely give us a glimpse into the argument itself and leave a number of questions unanswered. For instance, why regard doubt as the absence of belief? Why assume that faith can be held concurrently with doubt? What form does this argument take when trying to establish NDT as a credible position? To address these questions, let’s begin by taking a brief look at the phenomenology of doubt, before attempting to construct the argument itself. There are at least three plausible ways that we might construe the nature of doubt. First, in some cases one doubts p by virtue of having grounds for thinking that p is untrue, or one loses confidence that p with or without such reasons. For a religious context these sorts of reasons were highlighted in the foregoing: the problem of evil, religious disagreement, divine hiddenness, etc. It is typical to express this kind of

134 Versions of this argument also appear in Howard-Snyder (2013a), McKaughan (2013), Audi (2008; 2011) and Schellenberg (2005). 135 Howard-Snyder summarises the argument as follows: ‘unlike faith that p, belief that p is at odds with being in doubt about it’ (2013a, p.361).

102 doubt in terms of count nouns –136 ‘S doubts that p’ – and to give the reasons for why

S has these doubts. We can call this doubt1. Although this kind of doubt looks pervasive, it does not look as though it will support the argument at hand, since it is evident that one can recognise reasons for doubting that God exists while nonetheless believing that God exists. One may be unsure about whether God exists, because of the problem of evil, say, whilst concurrently believing that God exists, provided we allow that belief can come in degrees. Doubts of these sorts – doubt1 – therefore, are compatible with faith on the B+ account. Two other kinds of doubt, however, do seem incompatible with belief. To doubt that p is sometimes simply to disbelieve it: ‘I doubt that there is a God’ may imply one’s belief that there is no God. This kind of doubt – doubt2 – is also unhelpful to the argument from doubt. Since faith is incompatible with disbelief according to both B+ and NDT, then doubt that amounts to disbelief will be a loss of faith.137 So, there needs to be a further kind of doubt if doubt can be used to support Pojman’s argument. A third kind of doubt, which is cited as the kind authors have in mind when endorsing the argument from doubt, is when an agent does not believe p whilst not disbelieving p. This kind of doubt – doubt3 – is what is often thought to take one to a state of suspension of judgment. It characteristically involves

…a complete absence of confidence with respect to p and also the absence of belief that p, given what appears to the doubter to be inconclusiveness in the relevant evidence. And precisely because of this apparent inconclusiveness, we do not find the belief that not-p being substituted for the belief that p. What we have here instead is a state in which one believes neither p nor not-p. (Schellenberg 2005, p.96)

When one is in this state of doubt3, we often say that this person is in doubt about whether p. According to Howard-Snyder, to be in doubt ‘is for one neither to believe nor disbelieve p as a result of one’s grounds for p seeming to be roughly on a par with one’s grounds for not-p’ (2013a, p.359). Since Pojman takes doubt to be the absence

136 Moon (forthcoming) considers types of doubt and the form they take in speech. Examples of doubt1 are related to count nouns, doubt2 to verbs and doubt3 to mass nouns. 137 As Schellenberg points out, a doubt that entails ‘that God does not exist…takes us past doubt of the sort with which we are concerned’ (2005, p.96).

103 of belief, it seems as though doubt3 is the most relevant kind of doubt that he would take to support his argument. With these distinctions at hand, we are now in a place to offer a statement of the argument from doubt. The version that I will give, that seems like the strongest version available, takes the form of a reductio ad absurdum. In the following, ‘Dp’, ‘Bp’ and ‘Fp’ stand for doubt, belief and faith towards a proposition p, and S is an agent with the attitudes specified in brackets. To begin the argument, we can see that it follows from the account of doubt3 (D3) that:

(1) S (D3p)  S (~Bp ˄ ~B~p).

The B+ account of faith is committed to the following claim which NDT disputes:

(2) S (Fp)  S (Bp).

Now, the doubt argument posits someone who doubts3 that p and has faith that p:

(3) S (D3p ˄ Fp)  S (~Bp ˄ ~B~p ˄ Fp).

How does this present a problem for B+? Well, it follows from (2) (the statement of B+) and (3) that:

(4) S (D3p ˄ Fp)  S (~Bp ˄ ~B~p ˄ Bp).

The reason for this is because, on B+, faith requires belief. Now, assume for reductio that S both doubts3 that p and has faith that p:

(5) S (D3p ˄ Fp). It follows from (4) and (5) that:

(6) S (~Bp ˄ Bp).138 And this looks like an absurd result: someone cannot both believe p and not believe p. So we must reject one of the two principal premises (2) or (5): either the faithful do not have doubt3 (which would reject (5)), or B+ is incorrect (which would reject (2)). But as we have seen, it is extremely plausible to suppose that people can doubt what they at times have faith toward, especially for those who are religious. So, it seems to follow that (2) – the B+ account – is what ought to be rejected. The problem with this argument is that the supporter of B+ can opt to reject (5): someone who is in a state of not believing that p does not have faith that p. That, after all, is precisely what B+ is committed to. To assume (5) is just question-begging

138 S also ~B~p, but this is irrelevant to the reductio argument.

104 in favour of NDT. Wheeling out the generic observation that people of faith undergo periods of doubt is by itself ineffective against B+. What is needed is evidence of doubt3 (i.e. this specific variety of doubt) going along with faith. We can fully recognise the reality of the crises of faith referred to by supporters of the doubt argument, whilst still requiring that for these cases to present prima facie counterexamples to B+, the advocate of the argument from doubt still needs to give further evidence for why doubt3 can be combined with faith without one losing faith. How might one proceed to give such evidence? It is difficult to see how this could be forthcoming, but an appeal to faith-exemplars is often seen to hold some intuitive appeal in favour of premise (5). Consider the following comments from Daniel Howard-Snyder:

[O]ne might simply witness, and listen to the testimony of, those whom one takes to be paradigm cases of faith, and see for oneself that faith and doubt are compatible. Take Mother Theresa, for example, or Bill Alston or T.S. Eliot, all of whom experienced doubt of a sort and to such an extent that, during their years as Christians, they lacked belief of the propositions constitutive of the basic Christian story, but still accepted them or acted on the assumption that they were true or believed they were the least false of the options they found credible, and the like… If you can’t see Mother Theresa or Alston or Eliot or any number of others who suffer from doubt as exhibiting genuine faith, what good would an argument do you? (Howard-Snyder 2016a)139

These points do seem to lend some intuitive support to (5). However, their intuitive pull need not be interpreted as supporting a theory of faith as such, but, rather, they give a sympathetic interpretation of who we should attribute with faith. From the examples, it seems that Howard-Snyder is saying that we should attribute faith to those people who show such a sincere commitment to something. But if this is correct, then

‘faith-exemplars’ who exhibit commitment with doubt3 will not conclusively favour the argument from doubt. To see why, we can ask the following question: should cases involving doubt3 be understood as faith with doubt or as (at least for the duration of

139 This is taken from formal written comments given to me from Daniel Howard-Snyder at The Value of Faith conference in San Antonio, January 2016, in response to my paper ‘How to Defend the Doxastic Theory of Faith’. Thank you to Professor Howard-Snyder for taking the time to provide me with these comments. They are also used at later points in this chapter.

105 those doubts3) a loss of faith? Well, the problem for advocates of NDT is that your view on this will depend on whether you are already sympathetic to B+ or NDT. If the former, you won’t find the appeal to these exemplars ultimately persuasive, and if the latter then you will. As such, it just doesn’t look as though the evidence of people experiencing doubt offers support to one theory over another. Not only is evidence of doubt not clear evidence in favour of NDT, but there are several other issues that appear problematic for this argument. First, looking back at the remarks made by McKaughan, it doesn’t seem correct to say that B+ cannot make way for ‘significant’ doubt. For B+ the religious beliefs that form part of faith can go along with deep and on-going doubts1 that are more serious and long-standing than could be sustained for beliefs that do not form part of faith. This is because of the agent’s deep attachment to the beliefs that constitute faith: they are matters of ultimate importance to her and play a critical role in her thinking and decision-making. This point is easy to overlook because B+ is sometimes treated as if it were a belief only theory about faith. However, according to B+, faith is constituted by various commitments and non-cognitive attitudes; these can make faithful people disinclined to give in to doubt and jettison their beliefs. For example, consider a father who is devoted to his daughter and has invested a great deal of his life into her upbringing and her ongoing and future wellbeing. He treasures her and believes that she is a good person. However, he hears stories from reputable sources about her behaviour that cast doubt on the truth of this belief. This causes him a great deal of worry and he questions but does not lose his belief that she is a good person. If his belief concerned someone in whom he was less invested, he would be much more susceptible to the contrary evidence undermining his belief. Of course, there comes a point where to maintain belief against mounting contrary evidence becomes delusional. But we needn’t regard the father’s holding fast as merely delusive; to the contrary it seems, at least up to a point, to be admirable. Emotions and practical commitments can help us overcome crises of confidence and resist reasons to question what we believe. Additionally, it seems odd to draw attention to crises of faith as reasons for supporting NDT over B+. The only crisis that would lend support to NDT is where one has a loss of belief in p, retains a positive cognitive attitude towards p, and does not lose one’s faith that p. Consider, however, that according to NDT, belief is not in any case required for faith. So, by NDT’s own lights, this can’t be a crisis of one’s faith. In fact, to someone suffering from doubts3, the supporter of NDT can offer the 106 following advice: you aren’t suffering from any kind of crisis of faith – keep up your practical commitments and don’t worry about believing! With the argument from doubt looking unpersuasive in defence of NDT, let’s move to consider a second argument in its favour.

3.3.2 Language

Linguistic evidence about the use of ‘faith’ is used by some supporters of NDT to argue that faith need not involve belief. The basic strategy is to find cases in which we attribute to S faith that p without it being a truth condition of that utterance that S believes that p. I will call this the argument from linguistic data. This sort of strategy for establishing philosophical theories by appealing to philosophical intuitions about ordinary language use is a method sometimes employed by other analytic philosophers. Thus, the reservations I will raise concerning whether linguistic data could substantiate a theory about the nature of faith will also affect philosophical theorising more generally. Let’s begin by looking at the supposed evidence before assessing its merit in defending NDT. Consider the following example from Howard-Snyder:

[I]magine that I disclosed to you in a heart-to-heart exchange: “I can’t tell whether what I’ve got to go on favors the existence of God, but I have faith that God exists nonetheless.” You wouldn’t be perplexed, bewildered, or suspicious at all about what I said; at least you need not be. What I said wasn’t weird, or infelicitous; there’s nothing here that cries out for explanation. (2013a, pp.362-63)140

The claim being made is that the utterance from the quotation expresses faith without belief, but that the utterance is used felicitously. From this, Howard-Snyder goes on to conclude, the utterance supports NDT. We might expect that the argument from linguistic data would therefore run something like this:

(i) If utterance U is used felicitously then U supports theory T; (ii) Utterance U, in which one claims to have faith that p without belief that p, is used felicitously;

140 For other similar cases, see Himma (2006) and Diller (2013).

107

(iii) Therefore, utterance U, in which one claims to have faith that p without belief that p, supports theory T.

Given what U apparently shows, it follows from (iii) that T is NDT not B+. In order to get this argument off the ground, the linguistic data first needs to establish premise (ii). Does Howard-Snyder’s data do this? Well, utterances of the sort he has given seem linguistically felicitous, but how should we interpret them?

Howard-Snyder takes the speaker to be expressing doubt3 about the existence of God compatibly with having faith. However, on the face of it, the utterance is about the evidence available to the speaker for the existence of God and not belief in it. Believing that something is true while lacking compelling evidence for it is commonplace. It is also something that believers can recognise in themselves. I might say, for example, ‘I believe Brazil will win the 2018 World Cup but I can’t say that I have persuasive evidence that they will beat Germany’. If we can speak of belief without solid evidential support, then speaking of faith without evidential support is not a reason for thinking that faith is not partly constituted by belief. This objection also applies to other evidence Howard-Snyder offers:

[I]magine that Anne disclosed to us in a heart-to-heart exchange: “I used to think my evidence clearly supported the existence of God, but when I take into account what I’ve learned over the past ten years, I can’t tell anymore, I can’t tell whether it does or not; still, I have faith that He’s there.” (Howard- Snyder 2016a)

Again, even though Anne’s remark appears felicitous, it doesn’t look at all as though she is clearly expressing faith without belief. She might yet believe that God is there, even though she can’t tell exactly where the evidence leads. If the evidence leads away from belief in God, she can still believe in God (although her belief would then be epistemically unjustified). There are issues of interpretation over other cases as well. Consider the following example:

Suppose Tyler is on the job market and we’re at our favorite coffee shop talking about his prospects. Things look grim; a couple of interviews but no campus visits, and now it’s nearing the end of job-hunting season. As we get up to leave, he says to us: “Well, although I’d be a fool to think I’m going to

108

get a job, I still have faith that something’s going to open up for me.” (Howard-Snyder 2016a)

Although this example also seems felicitous, Tyler needn’t be expressing faith without belief. Instead, he could simply be claiming to have a doxastic faith in the teeth of contrary evidence. Indeed, oftentimes this is just how faith is construed, as we saw with the PFR in Chapter 1. And something similar is true of the following:

Suppose John confides in us, his friends, after a long talk about the state of his marriage: “To tell you the truth, things have gotten so bad, I don’t know what to believe, whether she’ll stay with me or move out on her own or go live with a friend or what; even so, I still have faith in her, faith that she’s not going to leave me.” (Howard-Snyder 2016a)

John might just as well be expressing uncertainty over how his beliefs should respond to the evidence he has than expressing faith without belief. His remark is felicitous, but doesn’t clearly express a lack of belief, but, rather, as with the previous cases, he appears to retain a believing faith even when the evidence opposes his belief. Are there any instance of linguistic data that unambiguously favour NDT? The following example, provide by Lara Buchak (2012, n.1), is slightly more promising:

(7) ‘I don’t know whether X – I have no idea whether I believe that X or not – but I have faith that X.’

Buchak takes (7) to be linguistically felicitous, and thus to support the view that faith does not require belief. Again, whilst (7) looks felicitous, it’s notable that the claim about the speaker’s belief in (7) is used to hedge a claim about the speaker’s knowledge that X rather than straightforwardly expressing a lack of belief that X. Perhaps this is relatively easy to rectify though. Instead of using (7), one could simply give examples of the form, ‘I don’t believe that p but I have faith that p’, and test these for linguistic felicity. For instance, ‘I don’t believe England will win in the cricket, but I have faith that they will’, or ‘I don’t believe my marriage will last, but I have faith that it will’, or ‘I don’t believe that God exists, but I have faith that he exists’. These statements remove any ambiguity from the data and present us with straightforward statements that could be used to support NDT, and importantly, premise (ii) in the argument from linguistic data.

109

Are these statements linguistically felicitous? Well, from what I can see they don’t appear to present a clear linguistic mistake, and so perhaps they are felicitous. This is, of course, just an intuition, and as with the faith-exemplar cases discussed in the previous section, someone else’s intuitions may simply differ. They might well be within their rights to demand a further argument for construing such statements as felicitous. I won’t pursue this line of objection any further, however, since the main problem as I see it, rests with premise (i). Assuming that (ii) is correct, which is not uncontroversial, there are specific reasons for thinking that this does not support any particular theory of faith, and general reasons for taking this to be an implausible strategy for establishing philosophical theories. Consider that if (ii) is correct, it should be easy to find examples of felicitous affirmations of faith that p along with unqualified denials of belief that p. For example:

(8) I don’t remotely believe that God exists but I have faith that God exists nonetheless

While it is an unusual claim to make, (8) does not seem to involve a linguistic mistake. The same seems to be true of the following:

(9) It’s clear that we’re the underdogs. In fact, if I’m being honest, I believe that we are going to lose. But miracles can happen, and I still have faith that we are going to win.

While (9) seems to express an odd mix of attitudes, it does not seem to involve any linguistic blunder, especially if we accept as felicitous the kinds of cases I proposed could support (ii). However, for the defender of the argument from linguistic data, (8) and (9) are problematic because they appear to express faith with disbelief, which NDT rejects. Could the defender show that the examples I used for (ii) are linguistically felicitous, whilst (8) and (9) are infelicitous? This looks like a tall order. Part of the problem is that our linguistic intuitions, for all but the most straightforward cases, are notoriously unreliable. It is also a problem that one’s judgements about linguistic felicity may be affected by one’s philosophical views, making such judgements a shaky foundation on which to resolve substantive philosophical issues. It’s unlikely, for instance, that someone who holds a belief-only view of faith will agree that any of the above cases pass the test for linguistic felicity.

110

There are other problems associated with using linguistic data to substantiate a theory of faith. The first is that any such argument must contend with the widespread phenomenon of loose talk. Consider the following utterances:

(10) The audience fell silent. [Said in reference to a rock concert] (11) The fridge is empty. [Said in a discussion of what should be done for dinner] (12) The lawn is square. [Reporting the shape of the speaker’s garden]

In each of these cases what is meant by the speakers by ‘silent’, ‘empty’ and ‘square’ does not strictly accord with how these expressions might be defined. The speaker of (10) is not claiming that the audience is, strictly speaking, silent. They were certainly breathing; perhaps some of them were talking quietly. Similarly, the speaker of (11) is not proposing that the fridge contains no air or that the shelves are missing; there may even be some food in the fridge but not enough to make a satisfactory meal. The speaker of (12) is clearly not committed to the lawn having sides of precisely the same length or internal angles of precisely ninety degrees. So we can use expressions loosely to communicate something different from their standard lexical content. We do not need to settle the correct linguistic theory of loose talk to recognise that the phenomenon is a standard component of communication.141 Similarly, we can talk loosely of faith, belief or disbelief. Suppose that belief that p is one of the truth conditions of faith that p. We can say that we have faith that p, even if we don’t believe that p, loosely to communicate (say) that we support p or are enthusiastic about p. For this reason, (8) does not show that we can have faith without belief any more than (12) shows that we can have squares that do not have sides of the same length. Both can be understood as loose talk. (7), (8) and (9) do not, therefore, provide linguistic evidence that clearly challenges B+; nor, for the same reason, does (9) constitute a worry for NDT.142 This concern also extends to other cases as well. Take, for instance, the following:

(13) I had to make a choice what to believe, so I decided to believe that God must exist.

141 See Carston (2002). 142 In §2.3.6, I discussed the role that language plays in establishing a non-cognitive component to faith. Given the worries raised here against such a method, I utilised other arguments to bolster linguistic intuitions so that we needn’t rely on them.

111

(14) It’s time for you to settle on what you believe on this matter. (15) You should not believe that the Holy Spirit is a person.

Each of these examples seems to suggest that belief is voluntary. There’s nothing clearly linguistically infelicitous about (13)-(15) but wide agreement on philosophical analysis tells us that beliefs are not directly within our control. Now, perhaps the advocate of NDT wishes to use these examples to establish the claim that belief is voluntary. But assuming that she doesn’t, I would suggest that she abandons the linguistic strategy altogether. It looks like the argument from linguistic data fails. Even if (ii) is true, which is hardly uncontraversial, (i) looks entirely implausible. As such, the advocate of NDT needs further arguments to support her theory. Let’s move on to consider a third option.

3.3.3 Volition

The notions of the voluntariness of faith and the involuntariness of belief have emerged in other arguments defending NDT. For instance, the work of J. L. Schellenberg (2005, 2007, 2009) uses these notions to defend an account of faith that is entirely non-doxastic since he claims that ‘faith is positively incompatible with belief’ (2005, p.132). This claim seems shocking given that many, if not most people claiming to have faith would also claim that their faith is constituted by belief in its object. Upon what arguments does Schellenberg base such a strong claim? The answer is that faith is voluntary, or within one’s volitional control. He claims that ‘faith must be understood as voluntary’, and goes on to make these remarks:

[T]he link between faith (or its absence) and our actions is direct: it is not just that how we live may over time have some influence on whether we have faith or lack it but rather that, if I try, I can have (or lose) faith right now through an act of will and that without such exertion faith will definitely be lost. (Schellenberg 2005, p.147)

How can Schellenberg’s claims be turned into an argument for NDT? It seems to require the premise that belief is involuntary, which is a claim I defended in §1.2.1. Hence, we can construct the following argument:

(iv) Faith is voluntary;

112

(v) Belief is involuntary; (vi) Therefore, faith cannot be constituted by belief.

I will refer to this as the argument from voluntariness. Despite its simplicity, I will show that this argument is not just unsound but also invalid. To see why, let’s briefly review the three arguments that Schellenberg offers in its defence, focussing solely on premise (iv). The first argument begins with the dual claims that religion is accessible to all, and that committing oneself to a religion suffices for faith:

The religious life is open to all who are genuinely interested: those who seek will find; the Way, though easy to miss, is always present and accessible…If we take seriously this open-door policy of religion…faith is to be understood as the sort of thing that can be intentionally produced and that the class of those who may thus produce it is wide and inclusive. (Schellenberg 2005, pp.148-149)

Now, I agree with Schellenberg that religion is something that can be pursued voluntarily – indeed, one ought to be free to pursue the religion one sees fit. The argument is problematic, though, for trying to establish (iv). First, faith is held toward many different objects, and it isn’t clear that we are free to access anything we have faith toward in the same way as we are with religion. Second, why should we think that having access to something, and thus committing ourselves to it, is sufficient for having faith towards it? There is a distinction behind what is argued here that has not been articulated. Some who pursue religion claim to lack faith toward its object, even though they desire to have faith. This indicates that commitment to a way of life does not entail faith towards the object of that way of life. (More on this shortly). Even if the previous two objections fail though, such that voluntary commitment to a religion entails faith toward it, this still causes no problems for the advocate of B+. This is because, as we saw in Chapter 2, there are different kinds of faith. We can accept that (iv) Faith is voluntary, without accepting the conclusion (vi), since (iv) is ambiguous. The argument equivocates on ‘faith’ by failing to properly distinguish what kind of faith. If Schellenberg is actually talking about relational faith, which seems most appropriate given the statement above, then the argument would run as follows: (iv*) Relational faith is voluntary; (v) Belief is involuntary; (vi) Therefore, relational faith cannot be constituted by belief. But this would therefore be

113 no argument against propositional faith. Hence, the argument has no force against B+. And what this reveals is that even if there is a voluntary component to faith, it doesn’t follow that there can’t be an involuntary component to faith. These can be distinct parts of the same kind of thing.143 A second argument Schellenberg uses in favour of (iv) comes from virtue theory. He claims that

…theologians and ecclesiastics have often spoken of faith as meritorious and virtuous – notions that go well with an understanding like my own: how can I be credited with or praised for something I do not take a hand in myself? (Schellenberg 2005, p.148)144

According to this argument, then, if a person’s faith is a virtue then we must be able to credit that person’s faith as meritorious, but we can only attribute meritorious status to a person’s faith if that faith is voluntary, and hence, if a person’s faith is a virtue then it must be voluntary. This argument faces two objections. First, faith can be partly voluntary and partly involuntary. Someone with Abrahamic faith could trust in a voluntary way and hence exhibit relational faith, whilst holding involuntary propositional faith. If this is true, then we can attribute meritorious status to the voluntary part of a person’s faith, and hence recognise it as virtuous, whilst allowing that there can also be involuntary components to a person’s faith, such as belief. Second, we can attribute meritorious status to a person’s attitudes, including her faith and beliefs, even when they are not voluntary. For instance, we often think that someone’s liberal attitudes, open-mindedness, intellectual humility, or beliefs favouring egalitarianism are worthy of merit, and hence are an excellence of a person, despite not being directly voluntary. In the same way, one’s faith need not be voluntary to be virtuous. The third argument Schellenberg appeals to as a means of establishing (iv) is, somewhat unfortuitously, linguistic use of ‘faith’. He appeals to linguistic data to show ‘that faith is a matter of action and is voluntary’ (2005, p. 150). These include such locutions as:

(16) Just step out in faith, and God will provide.

143 For further discussion on a related point, see Howard-Snyder (2013b, p.187) 144 Richard Swinburne makes a similar remark when analysing Aquinas’ theory of faith (2005, p.140).

114

(17) You should take it on faith that what the Koran says is true.

This strategy faces the same difficulties I have raised so far. First, intuitions about linguistic utterances are a shaky foundation upon which to substantiate a theory of faith. Second, even if utterances such as (16) can lend support to the claim that faith is voluntary, it doesn’t follow that all components of faith must be voluntary: (16) might simply be true of relational faith, without entailing anything at all about propositional faith. So, the three principle supporting arguments Schellenberg offers in favour of the argument from voluntariness appear to fail. But there is also a problem with Schellenberg’s claim that faith is voluntary that stems from his own account of faith. One of the necessary conditions he states for propositional faith is that, for a person S, S has faith only if ‘S considers the state of affairs reported by p to be good or desirable’ (2005, p.139). This condition is shared by both B+ and NDT. However, one cannot choose, at will, to consider something to be good or desirable. This is something that comes through experience and the cultivation of one’s pro-attitudes. Hence, even by Schellenberg’s own lights, faith cannot be entirely volitional. So it appears that we will need to look to other arguments to support NDT.

3.3.4 Pragmatism

In §3.2 I briefly addressed the issue that non-doxastic cognitive attitudes are adopted for practical, or means-ends reasons. However, if someone were to ask an advocate of B+ what reasons count in favour of having propositional faith, she will likely give reasons for belief to cover the doxastic component, and reasons for positive evaluation, to cover the evaluative component. As we saw in Chapter 1, there are more than just epistemic reasons for having a belief; there are also pragmatic and moral reasons too. Even though there are pragmatic reasons in favour of believing some proposition, these are still reasons for believing. The kinds of pragmatic reasons bearing on non- doxastic cognitive attitudes, though, are pragmatic reasons for accepting-without- believing. So, there can be reasons that count in favour of having faith that are purely pragmatic for both B+ and NDT. The difference is that B+ may advise faith for means- ends reasons, and achieve these by believing, whereas NDT may advise faith for means-ends reasons, and achieve these without believing.

115

Now, given that faith can be adopted for means-ends reasons, advocates of NDT may simply regard the doxastic component as superfluous. If you can have faith for purely pragmatic reasons, what difference does belief make to the faith of the individual? Consider the quotation made earlier from Swinburne. He sees faith as a ‘commitment to seek a goal’. The kinds of things he has in mind are typically eternal goals, or those relating specifically to relationship with God, such as salvation, eternal life, and answers to prayer. If faith in religion is the pursuit of these sorts of goals, then you may be able do this with either the aid of belief or some non-doxastic state. Now, there are certainly reasons for why believing the propositions in question is more preferable to accepting them. As William Alston notes, ‘the accepter will receive less comfort from her faith than the unquestioning believer, and she will be more troubled by doubts and waverings’ (2007, p.136). But this psychological benefit would not show that belief is required for faith; merely that it is desirable for faith. For faith, according to NDT, one simply requires what will help one to achieve one’s practical goals. This claim is appealed to for distinctively religious cases by Jonathan Kvanvig, who addresses the case of Abraham:

Consider Abram, who is told to leave Ur and go to a foreign land. He does so in commitment to a certain way of life, and it counts as an expression of saving faith. What propositional contents must he have believed in order for this story to make sense? Did he have to believe that God exists? I suppose he did so believe, but the same story could have been true if he had only been disposed to believe such, or disposed not to believe the denial, or if he had merely mentally assented to the claim and was determined to behave in accord with the assumption or presupposition that God exists. (Kvanvig 2013, p. 124)

Not only do Kvanvig’s remarks here sound reminiscent of Swinburne’s, but also of the earlier cases of Alston’s army general and Howard-Snyder’s backpacker. Moreover, Kvanvig’s is an argument for the view that non-doxastic mental states are sufficient for faith (or, ‘saving faith’ for this religious case) because they are sufficient to help one achieve one’s practical aims. In order to present this as an argument in favour of NDT, let’s refer to instances of supposed faith where one accepts p for purely pragmatic reasons without believing, pragmatic faith. A straightforward argument from pragmatic faith for NDT, which I will refer to as the argument from pragmatic faith, is as follows:

116

(18) Pragmatic faith is faith. (19) Belief is not essential to pragmatic faith. (20) Therefore, belief is not essential to faith.

The problem with this argument is that, as with the argument from doubt, (18) looks like a question-begging premise since the nature of faith and its relationship with belief is the point at issue. Supporters of B+ may see pragmatic faith as faux or only superficial examples of faith because they lack belief. An argument for (18), like that given by Kvanvig, that leads to trading intuitions about disputed examples of what should be considered genuine faith is unlikely to be fruitful. We saw this same problem when Howard-Snyder appealed to faith-exemplar cases. It was open to the defender of B+ to simply reject those cases as putative examples of faith because the agent in each case does not believe the proposition in question. Another way that an advocate of this argument might support (18) is suggested by some of Swinburne’s and Kvanvig’s remarks. They could employ what is sometimes called the ‘duck test’:

(21) If an agent talks and acts as if she has faith that p, then she probably has faith that p. (22) Agents with pragmatic faith that p talk and act as if they have faith that p. (23) Therefore, pragmatic faith that p is probably faith that p.

If (23) is correct, then it follows that B+ is probably wrong in proposing that belief is essential to faith. The duck argument is only effective, however, in contexts where there aren’t duck impostors – animals that look and quack like ducks but are a different species. There is a comparable worry about pragmatic faith: without the belief condition, we are unable to adequately distinguish faith from pretend faith. Pretence has enough functional similarities to belief to generate the appearance of genuine faith. In pretending that there is a God, for instance, one enters into a positive cognitive stance towards the proposition that there is a God – it is something that one takes on board as if it were true – and acts as if the proposition is true by, for instance, forming plans as if there were a God, saying (without believing it) things like ‘Yes, there is a God!’, making practical decisions on the basis of there being a God, and so on. The pretence is pursued for reasons unrelated to its truth, which make it sufficiently desirable to keep up appearances or undesirable to stop. Perhaps social and family pressures lead the agent to pretend that there is a God to avoid the unpleasantness of

117 leaving the religious community, or perhaps the agent pretends in order to secure the advantages of membership of the community.145 While some may welcome the inclusion of pretend faith as a variety of faith, it seems that pretend faith is no more faith than pretending to be in love with someone is to be in love with someone. Presumably many supporters of NDT will agree that a theory of faith that lacks the resources to distinguish faith from pretend faith incurs a significant theoretical cost. Why might this be? Well, according to NDT the person who satisfies (1)-(3) is not pretending to have faith, but actually has faith. The person with pragmatic faith satisfies (1)-(3) and so it follows that, on NDT, the person with pragmatic faith actually has faith. Her faith is genuine. However, given the discussion in the previous paragraph, the person with pragmatic faith is also indistinguishable from the person who is pretending to have faith. Her faith is also, therefore, not genuine. She is just pretending to have faith. What this leaves us with is an incongruous result. The pragmatist is both pretending to have faith, and actually has faith. Her faith is both real and faux; genuine but pretence. The fact that NDT allows for this incongruity is reason to suppose that it at best states necessary, but fails to state jointly sufficient conditions for propositional faith. And this seems to be the problem that pragmatic faith reveals which faces NDT. Having admitted cases of pragmatic faith, on what grounds are cases of pretend faith excluded? Before we consider some possible responses to this question, it’s worth noting that B+, of course, has a very neat solution: pretend faith is not faith because it does not involve belief. If NDT were to include a doxastic condition, and hence reject its all-encompassing cognitive condition, it would rule out the incongruity caused by instances of pragmatic faith. This would, though, simply be a statement of B+. As we can see, not only does the argument from pragmatic faith fail due to the problem of pretence, but drawing out this problem presents an objection against NDT itself: if we are unable to distinguish genuine from pretend faith on NDT, then we are theoretically required to endorse B+ since it lacks this incongruity. The advocate of NDT needs to find a way of distinguishing the two positions. There may be several ways in which she could do this. First, perhaps she could show that the positive cognitive attitude that NDT posits can be clearly distinguished from pretence. But

145 This is not to say that the reasons for the pretence have to be thought through; nor need a pretender, immersed in the pretence, be aware that they are engaged in a pretence.

118 how? One argument comes, again, from Daniel Howard-Snyder, who claims that assumption is different:

One can act as if p while disbelieving p, but one cannot assume p while disbelieving p. For when one assumes p, one has not settled on not-p; but when one disbelieves p, one has settled on not-p, even though one might dissemble and act as if p (Howard-Snyder 2013a, p. 366).

Now, if we grant that we can pretend that p either if we do not believe that p or if we disbelieve that p, which clearly we should, and we agree with Howard-Snyder that we cannot assume p while disbelieving p, then we can mark a clear difference between assumption and pretence. Although this argument is valid, it looks false since we commonly assume propositions that we believe or even know to be false. For example, in explaining a reductio ad absurdum to someone we assume a premise that we already know from the argument is false. We also make non-deceitful assumptions that we believe to be false as a matter of pragmatic requirement. Suppose Alston’s army general knows that at least some of the reports he has been given are misinformation. However, he may assume the hypothesis – which he believes is false – because he has no acceptable way of selecting an alternative and standing still is not an option. He may put out of his mind all consideration of alternative hypotheses and his subsequent behaviour and apparent commitments will be the same as in the case that Alston describes. The relationship between assumption and disbelief does not, therefore, exclude pretence. Is there a different non-doxastic cognitive state that could be posited in NDT that is incompatible with disbelief? Howard-Snyder argues as follows:

[I]f I disbelieve that my marriage will last, I’ll tend to say it won’t, when asked; I’ll tend to feel it to be the case that it won’t when I consider the matter; I’ll tend to use the proposition as it won’t as a premise in my practical reasoning...The incongruity of faith and disbelief suggests that faith requires a more positive cognitive stance towards its object precisely because the dispositional profiles of negative stances like disbelief are incongruent with faith’ (Howard-Snyder 2013a, p.361).

The claim here is that faith requires a positive disposition towards its object, but that disbelief does not provide one with a positive disposition. If pretend faith is compatible

119 with disbelief, then it looks like this is what distinguishes genuine faith, which is dispositional, from pretend faith, which presumably for disbelieving cases is not. This argument also looks problematic for establishing the distinction, for, if I believe that my marriage will not last I may nevertheless, for the sake, let’s say, of our children and my social status, pretend that it will. In this case, despite my disbelief that my marriage will last, I will not be disposed in any of the ways that Howard- Snyder suggests. My care for my children’s wellbeing and concern about how I am perceived and treated by society at large will motivate me to talk and act as if my marriage will continue. I may also very strongly hope that the marriage will continue and do my utmost to ensure that it does and pretend (or assume or accept) that it will do so; putting aside thoughts of marital failure may be crucial to this. More generally, disbelief and pretence seem compatible with ebullient and positive attitudes towards the subject of the engagement. The participants in a historical re-enactment of an American Civil War battle believes that they are not confederate troops but may be fully and enthusiastically engaged and ‘go along with’ or say ‘yes’ to the pretence that they are. So far we have not found a way to distinguish the non-doxastic cognitive attitudes suggested by NDT from pretence. However, there is another option open to the advocate of NDT – one that I believe does have much more potential to be successful. The best way to see this, I propose, is by looking at a particular strand of the pretence objection. This comes about by appreciating that certain pretence theories, and in particular a theory known as fictionalism, appear to be granted faith on NDT. Let’s defer the next response, then, to the latter parts of the next section, and use it as a response to both the particular case of fictionalism, and to pretence positions more generally.

3.4 The Fictionalist Objection146

3.4.1 The Objection

146 This objection was first published in Malcolm and Scott (forthcoming) and extended discussion was then published in Malcolm (forthcoming). For this thesis I have developed significantly on those earlier ideas.

120

In this section, I want to look at an objection against NDT that follows directly from the previous discussion of pretence. I will refer to this as the fictionalist objection. The objection runs as follows:

(vii) According to NDT, fictionalists can have faith; (viii) If a theory of faith permits fictionalists to have faith then that theory must be false or incomplete; (ix) So, NDT must be false or incomplete.

The task of the section will be to, first, explain why both (vii) and (viii) are plausible premises, and second, to show how the advocate of NDT can reject (vii). By doing so, I will explain how this response also shows that NDT can distinguish pragmatic faith from pretence. The response I will give that resolves this problem, however, will be shown to incur a number of theoretical costs for NDT. If these make NDT unfeasible, as I suggest they will, then I propose we have some theoretical reason to adopt B+ over NDT. In order to see why (vii) is plausible we need to begin with a definition of religious fictionalism. Religious fictionalism is a group of theories that principally concern our linguistic use of claims that have a religious subject matter.147 Each theory is committed to the view that a speaker who utters a sentence with some religious content does not believe the content of these sentences, i.e., the speaker believes that these sentences are not true, or are in error in some way or other. The religious fictionalist who says that ‘Jesus rose from the dead’ does not believe this proposition. Although the speaker does not believe the claims from the discourse, fictionalism maintains that she non-doxastically accepts these claims. As I claimed in §3.2, with acceptance and other positive cognitive attitudes, one acts on the basis that p, using p as a premise in some or all of one’s reasoning. Hence, by accepting religious claims, the speaker doesn’t only use them in discourse, but behaves as if these claims are true. She might, for instance, say that God exists when in the company of friends and family, engage in worship and prayer as though God exists and is worthy of worship and can grant answers to prayer, and plan her life around the proposition that God will guide her actions if she is faithful to his commands.

147 For a general discussion of fictionalism regardless of the domain see Eklund (2015).

121

There are notable theories of fictionalism across a number of different domains, for instance, in ethics, mathematics and science,148 and these theories share their traits with each other, and with religious fictionalism. So, for the ethical fictionalist, even though she doesn’t believe that, for example, there is such a thing as moral wrongness, she accepts that there is, and so she continues to use ‘wrong’ in her discourse, and acts as though certain actions are wrong. Likewise the religious fictionalist who accepts certain Christian claims won’t believe that Jesus is the path to salvation, but will continue to assert this claim, and act consistently as though it is true. There are two predominant kinds of fictionalism: hermeneutic and revolutionary.149 The former is largely a descriptive thesis, which aims to tell us what our attitudes are towards the propositions in a particular domain of discourse. The latter is normative, advising what stance one ought to adopt toward the propositions in that discourse. Hermeneutic fictionalism in general aims to show that our doxastic attitudes towards the claims in a discourse are non-believing ones. Revolutionary fictionalism advises that one should not believe these claims, but that it’s worthwhile continuing to accept and use them. The argument at hand will concern both kinds of fictionalism, depending on the non-doxastic theory of faith. Some are more descriptive, others are more prescriptive. The main reason that the fictionalist accepts and uses claims from a particular discourse is due to their instrumental value. With moral fictionalism for instance (Nolan et al, 2005), one won’t believe that moral claims posit moral entities that exist, perhaps because she has been persuaded by arguments in favour of their non-existence (see Mackie 1977). However, this person will nevertheless regard it as useful to continue to use moral discourse because, for instance, doing so has a reinforcing effect on one’s resolve toward behaving morally. Given this, the moral fictionalist adopts an attitude of acceptance towards those claims and positively applies them in everyday discourse. In the case of mathematics, although one believes that there are no such things as mathematical entities, she will retain the use of mathematical language due to the indispensability of quantification over numbers to our scientific theories (Leng

148 Examples include Joyce (2002) for ethics, Field (1980) for mathematics, and van Fraassen (1980) for science. 149 For a discussion of the distinction see Kalderon (2005).

122

2005). Similarly, a religious fictionalist will not take her lack of belief of religious propositions as a reason to disengage her use of them. Instead, she will see the benefits of continuing to use religious propositions in her discourse, and may even find it beneficial to fully immerse herself in a religious tradition or set of practices in order to fully attain these religious benefits.150 These benefits may be psychological, in the form of, say, existential comfort, or social, for instance, to retain one’s personal identity within a religious community. It might be objected that, contrary to premise (vii), fictionalists cannot have faith on NDT since they disbelieve the claims from the discourse, and NDT requires that one does not disbelieve the proposition one has faith toward given condition (2). There are two ways to respond to this objection. First, we could attempt to reject (2). This may be fairly straightforward since, as we saw in §3.2, there is a lack of persuasive arguments in favour of this condition. If we simply reject (2), then disbelieving fictionalists would not be denied propositional faith on the grounds of (2). However, there is a second response available that shows how fictionalists can meet the conditions outlined by advocates of NDT without rejecting any of those conditions. This response denies that all fictionalists are committed to disbelief. Although some fictionalists will disbelieve the propositions they accept, others will merely lack belief, and hence will suspend judgment concerning p. Some theories of fictionalism defend an agnostic position, as is the case with van Fraassen’s revolutionary fictionalism.151 Similarly, not all religious fictionalists must be atheists; some could merely be agnostic. On the basis of what I have said about fictionalism, we can define the theory as follows:

Fictionalism: a fictionalist does not believe the claims from a particular discourse, but accepts and utters them for their instrumental value.

The fictionalist who meets this definition will also meet condition (1) of NDT given that she accepts p. Despite this, not every fictionalist will have propositional faith.

150 For an early account see Le Poidevin (1996). A useful recent article that contrasts several different approaches is Deng (2015), and some current theoretical development on fictionalism is given in Jay (2014 and 2016 – the latter of which also addresses NDT). For discussion of ‘immersion’ in a religious fiction see Lipton (2007). 151 van Fraassen (1980) calls his theory ‘constructive empiricism’.

123

Importantly, the fictionalist must not disbelieve the propositions in question (given (2)), and must evaluate them positively (given (3)). Yet there’s no reason to suppose that a fictionalist couldn’t meet these two requirements. In fact, it will be the case that all fictionalists satisfy (3) since the reason they accept and utter claims from the domain is because they regard them as having instrumental value, and to regard something as instrumentally valuable is to evaluate that thing positively. I regard money as instrumentally valuable, and so I evaluate money positively. Likewise, when we regard our acceptance of a proposition as instrumentally valuable, we will evaluate the acceptance positively. Therefore, it looks like all fictionalists satisfy condition (3), just as they all satisfy (1). Only some fictionalists will satisfy (2) given that some will disbelieve, whilst others will have suspension of judgment. Where a fictionalist satisfies all three conditions for NDT, we would have, according to NDT, a fictionalist with faith. Now, should we go further and reject (2), which is a condition that I see no reason to retain, then all fictionalists will have faith on NDT. One response that can be made to the claim that fictionalists can have faith is to argue that even though this appears prima facie to be the case, perhaps NDT excludes fictionalists from faith on some particular accounts of NDT. For instance, a non-doxastic faith theorist might wish to argue that the kind of acceptance required by her theory must take into account the epistemic reasons for engaging in the domain.152 For instance, one might argue that to accept p one must have evidence for p. One wouldn’t accept p unless one had some epistemic justification for p, albeit insufficient justification to permit belief that p. This might rule out fictionalists from having faith who appear to only take into account instrumental reasons into their acceptance, and so may have no epistemic justification for p. There are two possible responses to this objection. First, recall that I am only claiming that, according to NDT, fictionalists who lack belief that p have faith that p, not that fictionalists who disbelieve p have faith that p. It’s likely that non-believing fictionalists will have some epistemic reasons for p, and that they wouldn’t engage in their acceptance without having some epistemic reasons for thinking that the propositions they accept are at least plausible. The reasons for this are twofold. First, our fictionalists have not yet been led to disbelief: they are agnostic, and this will

152 The case of acceptance for epistemic reasons is due to Michael Bratman’s (1992) version of acceptance.

124 probably be the case because they see the evidence for p as roughly balanced. Second, it might be quite unattractive to accept propositions that we expect will be completely false, fictionalists included. Consider a scientific fictionalist who accepts the existence of quarks. If she had no epistemic reason for thinking that quarks might exist then she probably wouldn’t do so. She needn’t have conclusive evidence that quarks exist to make accepting their existence advisable, but if she believed outright they do not exist, perhaps we wouldn’t recommend that she keeps using them in her discourse. The second response is that the account of acceptance for epistemic reasons might be too particular for many plausible cases of propositional faith. Take Howard- Snyder’s hiker for instance. She might not assume that she will complete the hike due to any epistemic reasons she has for thinking she will complete it. Her reasons might be entirely instrumental. Nevertheless, she is a putative case of faith according to NDT. So, specifying particular kinds of acceptance that exclude fictionalists may also exclude many cases that NDT considered genuine faith. To avoid this undesirable outcome, I suggest it would be preferable that NDT be stated in broad terms to encompass many different kinds of acceptance that people may be involved in. That will allow for accepting that p only if one has justification for p, and accepting that p only when one does so for purely instrumental gain.

3.4.2 The Problem with Fictionalist Faith

I have argued that according to NDT, fictionalists can have faith. In doing so I have tried to establish premise (vii). But what about (viii)? Why is it that fictionalist faith gives us a reason to suppose that NDT must be false or incomplete? Why not simply accept that some fictionalists have faith? One reason is that some advocates of NDT do not regard fictionalists to be genuine cases of faith. For instance, in a recent article, Daniel Howard-Snyder applauds Alvin Plantinga for distinguishing his account of belief in God – which Howard-Snyder takes to offer an argument in favour of a doxastic account of faith – from various radical approaches to religion, including fictionalism:

Plantinga writes these words in the context of a lament over the state of contemporary theology, which he finds steeped in the deplorable influence of religious non-cognitivists such as Richard Braithwaite and religious fictionalists like Gordon Kaufmann, John Hick, and Don Cupitt. I join him in

125

that lament. However, non-cognitivism and fictionalism couldn’t be further from our concerns. (Howard-Snyder 2016b, p.15)

As Howard-Snyder’s remarks make clear, he does not consider fictionalists to be attributed with faith on his version of NDT. However, simply denying that fictionalists have faith is not sufficient for showing that they do not. Perhaps there are problems posed by fictionalist faith that NDT needs to avoid. In the ensuing I will propose two reasons for why the possibility of fictionalists with faith is problematic for the advocate of NDT. The first reason we might have to tell the two positions apart is theoretical. These are, after all, two quite different approaches to one’s engagement within a particular domain. Fictionalism is primarily a theory about the meaning of our utterances within that domain, although, of course, fictionalists also behave in certain ways as though a claim is true. NDT, on the other hand, is concerned with our attitudes towards various propositions. It would certainly be theoretically rewarding to determine what distinguishes the two positions. Given that, as Howard-Snyder claims, fictionalism ‘couldn’t be further from’ the concerns of advocates of NDT, then, perhaps the appearance of fictionalists with faith is no more than a mere appearance, and it would be nice to know how fictionalism is distinguishable from propositional faith if it is only a mere appearance. However, there is a more pressing issue, which ought to motivate the advocate of NDT to ensure that the two theories are actually distinct. For, if the appearance is not simply a mere appearance, but an actual overlap between NDT and fictionalism, then what prevents the accusation that advocates of NDT are simply describing fictionalism by some alternate route? Non-doxastic faith might just as well be called ‘fictionalism’. Do we actually just have on our hands two ways of naming the same phenomenon? If so, this threatens to make NDT redundant and to collapse into a kind of agnostic fictionalism. Instances of faith without belief will be instances of agnostic fictionalism, and all that will be left are instances of doxastic faith. But this simply leaves us with a doxastic account on one hand, and fictionalism on the other. The threat of redundancy for NDT gives the advocate of NDT a reason to determine what makes NDT distinct from fictionalism. In addition to this concern, we are still left with the question over why fictionalists can satisfy (1)-(3) without seeming like cases of genuine faith. Why

126 shouldn’t they? What guides the intuition that fictionalists do not have faith? Well, for one thing, fictionalists are engaged in a pretence.153 With her affirmation of claims from a discourse, the fictionalist is merely pretending in her attitudes towards such claims. A fictionalist with faith, then, would simply be pretending to have faith. As we saw in §3.3.4, this produces an incongruity that needs to be resolved. The second reason we have to tell the two positions apart is practical. The advocate of NDT might not want her theory to grant faith to certain radical positions. She may feel this gives faith too broad a scope. From the remarks quoted above, we can see that Howard-Snyder does not think that fictionalism is a positive influence on theology. Unfortunately, the reasons for this are not stated. So, whilst we can only speculate as to why this is the case for Howard-Snyder, for some, faith in the religious domain at least carries certain soteriological implications. In Christianity for instance, there is the concept of saving faith. One who has saving faith is thought to have been granted eternal life. It might be that someone wouldn’t want to regard the fictionalist position as a sufficient means by which to attain saving faith. These theoretical and practical concerns jointly motivate an attempt to overcome the possibility of fictionalist faith because if they cannot be overcome, then a purely doxastic account of faith seems more favourable than a non-doxastic account. Note that these concerns apply equally even if the fictionalists I claim have faith on NDT are only of the non-believing, rather than of the disbelieving variety. Non- believing fictionalists still pose the theoretical problem of NDT being assimilated under fictionalism, and they are also problematic for the Christian conception of saving faith.

3.4.3 The Importance of Truth

There are at least two possible ways to distinguish faith from fictionalism. First, and quite simply, we can adopt a doxastic account of faith by rejecting (a). In that case, (a) would read: ‘a belief that p’. Believing the proposition in question would categorically exclude all fictionalists. This option will not be of interest to advocates of NDT since

153 For instance, Nolan et al. construes fictionalism about a discourse as that which ‘takes certain claims in that discourse to be literally false, but nevertheless worth uttering in certain contexts, since the pretence that such claims are true is worthwhile for various theoretical purposes.’ (2005, p.308, emphasis added)

127 it amounts to an endorsement of B+. Second, we can add something to the analysis of NDT that holds for faith but not for fictionalism. We have already seen how acceptance for epistemic reasons only is a possible way of proceeding with the second option. Here I want to address what I take to be a more promising option, which is to argue that the truth of the proposition in question matters to the person with faith, whereas it does not for the fictionalist. In the ensuing I will explain how this responds to the claim that fictionalists have faith, and how it changes the analysis of NDT. How might truth distinguish faith from fictionalism?154 It’s certainly part of what it is to be a fictionalist that the truth of the propositions she accepts bears no consequence on whether she accepts them. The value that the fictionalist sees in accepting and applying those propositions in her discourse in no way depends upon whether or not they are true. This same point might not hold for the person with faith. Suppose that someone badly wants to win in a race and has ‘non-doxastic faith’ that she will win, and so does not believe that she can win, but accepts that she can win the race, and does so in order to spur herself on to win it. If she was able to learn, say, by somehow seeing into the future, that she will not win the race no matter how hard she tries, this fact seems important to her, and it may well affect whether she continues to accept that she will win it. She may even cease to engage in the race altogether. In the religious domain, suppose someone is able to learn that the proposition ‘God exists’ is categorically false. Presumably, for the person of faith, this would radically alter her attitudes and behaviour, such that she would no longer adopt a positive cognitive attitude towards that proposition, and would cease to commit to acting as though it were true.155 The argument being made here is as follows: if the person with faith that p discovered that not-p, this discovery would be critical for whether she would continue with her cognitive commitment to p, whereas for the fictionalist, the truth or falsity of

154 A brief mention of the approach is given in a presentation by Daniel McKaughan (2014). With the exception of the reference to Buchak (see n.156 below), I am not aware of this option being directly discussed elsewhere in the literature. However, one could indirectly appeal to this claim from the account given by Howard-Snyder, who says that ‘one can have faith that something is so only if one cares that it is so’ (2013, p.360). It could be argued that one cares that p is so only if one cares p is true. There is agreement on this condition from Schellenberg (2013). 155 We could see this as equivalent to asking someone the counterfactual: if it were the case that God does not exist would you continue with your commitments to God?

128 p has no bearing on whether or not she accepts and utters p.156 Call this the importance of truth argument. Although the significance of truth does appear to mark a clear distinction between the two positions, it’s hard to say for certain whether the importance of truth argument is sound because it involves an empirical claim that we currently lack any evidence for. Who’s to say, for example, that many people with faith in the religious domain wouldn’t simply carry on as normal if they found out that God does not exist? The benefits of religious engagement might outweigh those of disengagement, even upon discovering that the religion’s major propositions are in fact false. The same applies to the runner in the race. We can’t say for sure whether she would cease her cognitive commitment. She may be motivated to retain it for other reasons, say, due to the support she has received from family and friends. In both cases, it’s not necessary that one’s cognitive commitments will turn on the truth or falsity of the propositions one has faith toward. Perhaps what needs to be done to defend the importance of truth argument is to show why someone with faith would cease her cognitive commitment to p upon learning not-p. One reason for this may be due to instrumental gain. If propositional faith involves a pragmatic component in which one adopts a cognitive attitude in order to gain something of value – some good that is worth attaining – then the truth of the proposition may be necessary for this aim to be achieved. For instance, suppose a runner adopts faith that she will win the race entirely as a means to win the race. Winning the race is an instrumental good she wants to attain. If she learnt that she will lose the race, then this will result in her choosing not to have faith that she will win the race, the reason being that she knows that she cannot acquire the good. Supposing this is true, we could say that if the runner still does accept that she will win the race, then perhaps she doesn’t have faith after all, she is actually just pretending that she will win it. If we adopted this approach, then a restatement of the first condition would be:

(1*) a positive cognitive attitude toward p, adopted instrumentally to attain certain goods.

156 A similar claim is made by Lara Buchak: ‘We do not attribute faith to a person unless the truth or falsity of the proposition involved makes a difference to that person.’ (2012, p.226)

129

The attainment of goods is the reason why the truth matters to the person with faith. If she knew the proposition were false, she wouldn’t adopt the attitude because she would then know that by doing so she wouldn’t be able to acquire the relevant goods. There is also some support for interpreting NDT as requiring (1*) rather than (1) from prior examples of propositional faith. Consider the earlier examples given by Swinburne, Alston and Howard-Snyder. Each example involved an individual who adopted the attitude in order to reach some goal or acquire some good. (I mentioned at the end of §3.2 that in the background to NDT is a description of propositional faith as involving instrumental reasoning. (1*) now states this description as a prescription on NDT). Despite this modification to NDT, fictionalists will still have faith since they also hold a positive cognitive attitude toward p to attain certain goods. The religious fictionalist, for instance, accepts and affirms religious claims for their instrumental value: she wants to use the discourse as a means to, for example, retain a sense of spiritual comfort and personal identity. So, (1*) does not exclude fictionalists from faith. However, there is a response open to the advocate of NDT that does exclude fictionalists, but only for instances of faith toward propositions with a religious subject-matter. At this point I will branch off from discussing faith regardless of the domain to faith solely in religious content. The reason why will be made clear in the ensuing, and the costs of doing so will affect the overall account. To see the religious response, let’s begin by supposing that, according to (1*), the proposition that God exists is necessary for the person with faith’s instrumental aims to be achieved. Some of these aims will be restricted to life in this world. For instance, she might accept that God exists, and that God guides her actions, in order to feel as though the decisions she makes in life are significant. This might be the case when she accepts a particular job or dedicates her time to supporting a charitable cause. The fictionalist’s aims are also restricted to life in this world. For example, she might accept that God exists in order to have a moral framework within which to make moral decisions. In this case, if the fictionalist discovered that God does not exist, she may feel a sense of disappointment, but it needn’t stop her from engaging in her religious practice. This is because she doesn’t actually need the proposition God exists to operate within the moral framework. As I have mentioned, some accounts of faith do endorse engagement with religion purely for their benefits in this world. For example, the famous pragmatist 130

William James claimed that ‘religion offers itself as a momentous option. We are supposed to gain, even now, by our belief, and to lose by our non-belief, a certain vital good.’ (1919, p.26). If we take these ‘vital goods’ that we can gain ‘even now’ to be the kinds of things I have said are benefits to be attained in this world, then it looks like James endorses the faith of the fictionalist. After all, the fictionalist is looking to gain vital goods by her acceptance of religious claims. However, this is not all James has to say on the matter. He adds shortly after that ‘if religion be untrue, we lose the good’. But how could we lose goods in this world, such as the usefulness of a rigid moral framework, if it turned out that that God does not exist? Presumably, James means to say that faith involves not only engaging in religion to attain goods in this world, but to attain eternal goods as well, most significantly among these being eternal life and relationship with a divine being.157 It is the attainment of eternal goods, i.e. goods that have their value beyond life in this world, or that require the existence of divine beings, that do require the proposition ‘God exists’ to be true, and if it’s not true, then we cannot attain the eternal goods that our faith pursues. So, according to the pragmatists, faith involves pursuing both temporal goods in this world, and eternal goods that go beyond this world, the latter of which requires that God exist for their attainment. We can now say why truth is crucial to a person of faith: because faith requires pursuing eternal, and not merely temporal goals, whose attainment requires certain propositions to be true. We could modify (1*) to reflect this as follows:

(1**) a positive cognitive attitude toward p, adopted as a means to attain both temporal and eternal goods.

This condition is one that no fictionalist can meet since she does not pursue eternal goods through her engagement with a particular discourse. This is simply not the position the fictionalist takes. She is looking for an alternative position to take on religious claims given that she does not believe certain claims to be true, or she recognises that she should not believe them to be true. This is one of using those claims to extract their temporary benefits. She is not engaging in the discourse in the hope that its claims are true, and that she might acquire something like eternal life. Engaging in a discourse in the hope that God would grant eternal life or answers to prayer

157 This is also the main motivating feature in Pascal’s Wager (see Jordan 2006, for an overview).

131 involves doing so in the hope that the claims of the discourse are true (rather than as you would a useful fiction), which stands in clear contrast to fictionalism. So, it looks like we have one plausible way to defend the importance of truth argument. If that argument is defensible, then we have a defensible means of rejecting the claim that fictionalists can have faith. However, endorsing this means of defending the importance of truth argument opens up two serious problems. First, the solution does not work for faith toward non-religious content. Presumably there are many propositions one has faith toward that do not have anything at all to do with pursuing eternal goods. For instance, faith that I will win the race or faith that my marriage will last. Moreover, there are even some religions that do not seek eternal goods, but merely pursue temporal benefits, as is the case with some forms of Buddhism. This may be of little consequence to those who consider religious faith to be distinct from non- religious faith, but we then lose the virtue of parsimony – of having a unified account of the propositional attitude. As such, faith in the religious domain will have different conditions from faith in other domains, and the latter will still be a faith of which fictionalists may be entitled. This issue may not be decisive for religious faith, but it certainly marks a drawback from using this solution for those who wish for a parsimonious theory of our propositional attitudes. (I will revisit this issue shortly). A second issue is that, according to this new condition, someone can have absolute confidence and outright belief that God exists, and have a genuine and deep affection for God, yet not have faith. However, this would seem absurd. Consider, for instance, the person who deeply loves God and believes that God exists with complete confidence. Nevertheless, she is temporarily unmotivated to pursue her religious commitments, perhaps due to depression or akrasia. Despite her temporary despondency, she retains both strong belief and passionate affection for God. We might think of her faith as dormant, for the time being at least, rather than lost.158 Yet, if we include a pragmatic condition in our analysis whereby faith requires the pursuit of any kind of good, be it temporal or eternal, then these kinds of people will be classified as faithless, where surely they should not be. This latter problem can be overcome, though, by specifying the analysis in the correct way, to only impose a pragmatic constraint on those who adopt a cognitive

158 If we thought of faith as motivational (Scott 2013, chapter 5) then we could say that she has faith but is simply suffering from a motivational disorder.

132 attitude voluntarily. There will, then, be a disjunction between those who believe, and those who hold a positive cognitive attitude for pragmatic reasons. As such, condition (1**) splits into

(1a) a belief that p; or (1b) a positive cognitive non-doxastic attitude towards p, adopted as a means to attain both temporal and eternal goods;

This final analysis overcomes the problem of the unmotivated believer in virtue of (1a), and shows how faith is to be distinguished from fictionalism by either condition. Despite the resolution the new analysis brings to the problems at hand, it represents a mere Pyrrhic victory for NDT. The main reason for this is because, as was mentioned in the foregoing, it lacks the virtue of having a unified account of propositional faith. We will need other solutions to overcome the fictionalism problem for faith toward non-religious content. But this appears to be a significant cost for the solution proposed by (1a) and (1b). In fact, given that the revised solution only solves the fictionalist problem for religious cases, it makes this solution look untenable. What we ideally require from our metaphysical theories is that they cover all instances of a phenomenon. If faith is of one kind, then the theory should cover all instances of faith. The solution proposed by (1a) and (1b) does not do this, and so looks frail from a theoretical perspective. In contrast, B+ simply covers all instances of faith, without requiring the additional condition (1b) as it sees this as superfluous. For B+, fictionalists and pretenders are excluded by (1a), without the need for recourse to some further exclusionary condition. B+ is therefore simpler, more parsimonious, and hence, more theoretically virtuous. Using the importance of truth argument is the most promising means I am aware of to show that fictionalists cannot have faith. However, as I have shown, the only defensible way to cash it out leads to a far too restrictive definition of propositional faith. Without another solution to the fictionalist objection, it seems that, for now at least, B+ is the most defensible theory of propositional faith, and one that I will endorse for the remainder of this thesis.

3.5 Conclusion

133

This chapter has argued at some length that, despite their ever increasing popularity in the philosophical literature, non-doxastic theories of propositional faith are indefensible. I began by giving a broad statement of these theories, and found numerous examples that fit with this account in the current literature. I then examined the four leading arguments in favour of NDT – those involving doubt, language, virtue and pragmatism. Each argument was found to be lacking in some way, and so each failed to adequately provide a defence of NDT. I then considered the objection to NDT that it is consistent with certain pretence positions, and used religious fictionalism to highlight this point. I proposed that this gives us a reason to think that NDT is false. Although there are ways that advocates of NDT can respond to this objection, none of these are particularly appealing, and incur certain theoretical costs that give us a reason to favour B+ as a theory of propositional faith. Overall, the prospects for NDT appear to be quite poor, and my proposal is that we leave aside NDT as a response to the PFR unless a more persuasive case can be made for it. For the remainder of this thesis, then, I will examine the second response to the PFR – the divine testimony theory – which, as a doxastic theory, utilises the B+ account of propositional faith.159

159 Granted that if NDT is defensible, one could have a divine testimony theory that combines with NDT, but since it is not defensible, I will put this option aside.

134

4. DIVINE TESTIMONY

4.1 Overview

In this chapter and in Chapter 5 I explain and evaluate the second of the two responses to the problem of faith and reason (PFR) – the divine testimony theory (DTT). This response maintains that when someone’s propositional faith is the result of trust in divine testimony, this propositional faith can be epistemically justified, and we can show how by looking to general theories in the epistemology of testimony. In Chapter 2 I explained how propositional faith may be related to trust in testimony. This chapter outlines what it is for God to be the speaker of the testimony, and Chapter 5 explores how one is justified in trusting divine testimony. At the end of Chapter 1 I considered several recent authors who defend a conceptual connection between religious faith and believing divine testimony. I developed this in §2.3.5 where I argued that relational faith can cause propositional faith, and how this is sufficient for Abrahamic faith. One way to look at faith is as a hearer trusting what a speaker reports to her. Abrahamic faith could then function as the hearer’s trust in what is said, with God acting as the speaker. Recall in §2.3 that I discussed how, with testimony, a speaker S tells a hearer H that p. There are two sides to this. The speaker’s address and the hearer’s response. We could say that Chapter 2 covered part of what it is to respond to testimony. Much of this chapter is dedicated to explaining things from the speaker’s perspective – of what it is for God to speak. I begin this chapter in §4.2 from the perspective of faith as trust in God’s testimony by reviewing some of the historical treatments of this claim. Subsequently, in §4.3 I discuss faith from the alternative perspective, by giving an account of what it is for God to testify to people. I do so by drawing largely on the work of Nicholas Wolterstorff, and in so doing, outline three ways in which God speaks. This will give us the whole picture of God, as a speaker, addressing human people, whilst people, as hearers, receive the address from God. The defence of the epistemic justification of Abrahamic faith as defended in Chapter 2 will then be discussed in Chapter 5 by utilising the accounts of divine testimony developed in this chapter.

4.2 Faith: ‘The Ear of the Soul’

135

The claim that faith requires believing what God has told us has a long history in theology, particularly within the Christian tradition.160 Thinking back to the narrative of Abraham discussed in §2.3.1, he believes what God tells him and in doing so, I have proposed, he has relational faith in God that causes propositional faith that p. The early Greek Christian thinker Clement of Alexandria, writing in the 3rd Century A.D. concerning the faith of Abraham says:

[I]f to Abraham on his believing it was counted for righteousness; and if we are the seed of Abraham, then we must also believe through hearing. For we are Israelites, who are convinced not by signs, but by hearing. (Strom 2.6)161

Clement’s claim here cuts right to the heart of faith and reason. He is saying that religious people ought to believe propositions concerning God not through outward ‘signs’ that convince their perceptual faculties, for this does not produce ‘righteousness’. Rather, they ought to believe by trusting what God has said without the aid of such signs. There is something pleasing to God when people trust him in this way, according to Clement. He intimates as much when he goes on to tie this concept of believing by hearing to the nature of faith:

Happy is he who speaks in the ears of the hearing. Now faith is the ear of the soul. And such the Lord intimates faith to be, when He says, "He that hath ears to hear, let him hear;" so that by believing he may comprehend what He says, as He says it. (Strom 5.1. my emphasis)

This poetic and striking metaphor, that faith is ‘the ear of the soul’, is a metaphorical way of endorsing the divine testimony account of faith given in Chapter 2. Abrahamic faith requires hearing what God says, and trusting this as truth by believing God. In §1.5.3 we saw how Locke and Kenny held a similar view of faith, and in §2.3.5 the account was also endorsed by Anscombe and Zagzebski. These authors are writing in a particular tradition of Christian theology, and this tradition can be thought of as beginning with authors like Clement. Since Christian faith is one instance of Abrahamic faith, the writings of Christian theologians on faith is a useful resource for understanding what this kind of faith is.

160 A helpful overview of this notion of faith in the early church fathers is Lamont (2009). For a more extended discussion see Lamont (2004). 161 The Stromata, or Miscellanies (1971).

136

Perhaps the most influential author, though, to endorse this theory of faith is St. Thomas Aquinas (ST 1947; DV 1953),162 who is in many ways a contemporary of Clement. In his summary of Aquinas’ definition of faith, Richard Swinburne claims that, for Aquinas, ‘to have faith in God is simply to have a belief-that, to believe that God exists’, in addition to ‘certain other propositions as well’ (2005, p.138). These other propositions concern ‘what God is like and what acts He has done, and you have to believe these latter propositions on the ground that God has revealed them’. The claim that one must believe propositions about God on the grounds of revelation is what makes faith a virtue for Aquinas. He regards faith as a mean between science (scientia) and opinion (opinio). For Aquinas, science is a kind of knowledge of propositions which we have clear reasons for believing because ‘science is derived from self-evident and therefore "seen" principles’ (ST II.II.1.5). So, scientific knowledge is a true belief where there is an understanding for why the belief is true, namely, something in the world that is evident to the senses.163 In contrast, opinion does not have so close a connection to truth as science since ‘it is possible for a believer to have a false opinion’ (ST II.II.1.3) – an opinion can be false. Whilst faith, for Aquinas, might lack the certainty of knowing what makes it true, thus giving us less confidence in the object of our faith, it is firmer than opinion since whilst ‘it is possible for a believer to have a false opinion through a human conjecture…it is quite impossible for a false opinion to be the outcome of faith’ (ST II.II.1.3). So, faith differs from opinion since it is infallibly true. The reason it is infallibly true is because ‘it must derive this infallibility from its adherence to some testimony in which the truth is infallibly found’ (DV XVI.8). This is the case, for Aquinas, because God himself revealed its propositional content. We can see, then, why for Aquinas, Christian faith is belief in propositions as revealed by God. Because the propositions one has faith toward are true, though not self-evidently, as is the case with science when we directly perceive some object. The

162 These abbreviations refer to Aquinas’ works Summa Theologica (ST) and De Veritate (DV). 163 We could interpret Aquinas’s views on knowledge as epistemically internalist. Some accounts of epistemic internalism (Bonjour 1985, Chisholm 1988) require that for epistemic justification regarding p, that which justifies p must be accessible on internal reflection. Put another way, to know p we must know how we know p (see Poston 2007).

137 truth, though, is guaranteed because these propositions come from God who, presumably, does not lie when testifying,164 as Aquinas notes:

[T]o assent to the testimony of a man or an angel would lead infallibly to the truth only in so far as we considered the testimony of God speaking in them. Consequently, faith, which is classified as a virtue, must surpass the truth of man’s own understanding and thus make it embrace that truth which is in the divine knowledge. (DV XVI.8)

I propose that we can see in this passage how Aquinas likens faith to believing what God is saying. Faith, Aquinas claims, must go beyond our understanding – the kind of understanding we gain from science – for the reason that faith gives one access to knowledge that must come from God. It’s understandable why Aquinas would think that many of the propositions believed by religious people are only accessible through trust in what God says. Propositions concerning the afterlife, the future, or non- tangible spiritual beings cannot be known through science. We must rely on God to tell us about these things, and that, for Aquinas, is what makes faith important, and why faith requires trusting God’s testimony. One final, and often overlooked writer who endorses this theory of faith is the 17th century theologian John Owen. I introduce his work at this stage because, first, it is perhaps the most explicit articulation of the divine testimony account of faith, and second, because I will discuss his way of overcoming certain epistemological problems that arise from this theory in Chapter 5. Owen regarded faith as a source of knowledge, according to which he says:

This is our assent upon testimony, whereon we believe many things which no sense, inbred principles, nor reasoning of our own, could either give us an acquaintance with or assurance of. (Owen 1852, p.83)

By way of this definition, Owen appears to view faith as a faculty for acquiring knowledge regardless of the content. Testimony gives knowledge of innumerable propositions, and is quite a typical means of acquiring knowledge. Such a view appears identical to the claims made by Anscombe that were stated in Chapter 2. As Owen sees it, it is by way of this capacity for knowledge acquisition, rather than by sense perception or a priori reasoning, that we come to know things from and concerning

164 I will address this issue further in §5.3.1.

138

God: ‘This, therefore, being the most noble faculty of our minds is that whereunto the highest way of divine revelation is proposed’ (1852, p.88). According to Owen, then, faith is believing testimony, and divine revelation comes by way of testimony. We will revisit Owen later to review what he has to say concerning one’s justification for believing divine testimony. When I discussed the nature of assertion in §2.3.5, I claimed that a common function of assertion is to transmit propositions from speakers to hearers. Faith may be thought of as the hearer’s side in the transmission process of believing the propositions told – or as Clement puts it, faith is ‘the ear of the soul’. The speaker side of the transmission process is then undertaken by God who intends to transmit his divine knowledge through speaking in some way. Although I have shown that there are a number of authors who have proposed accounts of faith as believing divine testimony, it is uncommon to find philosophers discussing what it is for God to speak. That is, there are philosophical accounts of faith as the hearer side in a testimonial exchange, but not of God’s speech as the speaker side. Often such accounts are left to theologians who have developed theories of divine revelation.165 However, one philosopher who has reflected on the nature of divine speech is Nicholas Wolterstorff (1996), and given that his theory is the most advanced philosophical treatment of divine speech available, it is his account of divine speech that I discuss in the ensuing.

4.3 Divine Speech

4.3.1 Authorial and Deputised Discourse

Whilst Wolterstorff has developed a thoroughgoing account of divine speech, his theory is not particularly systematic. In this section I shall bring together different claims that he makes about the nature of divine speech to produce three distinct ways of understanding divine discourse. Each of these three ways by which God is proposed to speak will then be utilised in Chapter 5 when I discuss how one is justified in believing God has spoken. Such justification changes depending on the account of divine speech, so it’s important at this stage to have a clear account of what it is for God to speak.

165 For some philosophical treatments of revelation see King (2012 and 2013).

139

I will begin with an example of divine discourse that Wolterstorff (1996) discusses himself, which is found in the conversion narrative of St. Augustine in his Confessions (1998).166 Prior to his conversion to Christianity, Augustine was conflicted by two strong but competing desires. On the one hand, he wanted to commit his life to God by following God’s commands and the plans that God had for his life. On the other, Augustine was driven by his natural desires for certain pleasures, like promiscuity and the drinking of alcohol, which he believed led him away from God’s commands and plans for his life. In one narrative Augustine records, he is experiencing some inner turmoil caused by this conflict of his desires, when he hears the voice of a child nearby to him saying ‘Take it and read, take it and read’. Unsure of where the voice is even coming from, Augustine takes this to be an instruction to read from the Christian scriptures. His reasoning comes from a story he had previously heard:

For I had heard the story of Anthony, and I remembered how he had happened to go into a church while the Gospel was being read and had taken it as a counsel addressed to himself when he heard the words Go home and sell all that belongs to you. Give it to the poor, and so the treasure you have shall be in heaven; then come back and follow me. By this divine pronouncement he had at once been converted to you.

Augustine gives some clues here for his own view of divine speech. He considers the words spoken here to Anthony, taken from the Gospels,167 to be ‘divine pronouncement’ – God speaks to Anthony through the medium of the biblical gospels. Moreover, he wants to respond to his perceived instruction to ‘Take it and read’ by doing the same – reading the Bible as a means of hearing from God. He then concludes the story:

I hurried back to the place where Alypius was sitting, for when I stood up to move away I had put down the book containing Paul’s Epistles. I seized it and opened it and in silence I read the first passage on which my eyes fell: Not revelling and drunkenness, not in lust and wantonness, not in quarrels and rivalries. Rather, arm yourselves with the Lord Jesus Christ; spend no more thought on nature and nature’s appetites. I had no wish to read more and no need to do so. For in an instant, as I came to the end of the sentence, it was as

166 The references from The Confessions in this section are from Chapter VIII, Section 12. 167 (Matthew 19:21; Luke 18:22; Mark 10:21).

140

though the light of confidence flooded into my heart and all the darkness of doubt was dispelled.

The same experience that the character Anthony had had was now being experienced by Augustine. In the reading of the Bible – this time in Paul’s letters, rather than in the Gospels – Augustine took himself to be hearing God speak to him. In particular, Augustine was hearing God command him to relinquish precisely what he had been in conflict over, namely, his natural desires. It was all he needed to decide which of his conflicting desires to follow, and he chooses to convert to a life that follows God’s commands and plans for his life. In order to interpret Augustine, Wolterstorff considers two possible ways of reading what Augustine has said. The first of these yields two distinct ways that God might speak, and I will discuss these two ways in the remainder of this section. The second interpretation yields a third way that God might speak, and I will discuss that in the next section. According to the first reading, Wolterstorff claims that we could understand Augustine as saying,

…back several centuries God had said, by way of Paul’s writing his letter to the Roman Christians: arm yourselves with the Lord Jesus Christ and spend no more thought on nature and nature’s appetites; and he, Augustine, recognised this command as addressed (or applying) also to himself, and found himself obeying it. (1996, p.7)

Wolterstorff is proposing to interpret Augustine as follows: Augustine has read a section from the Bible from the Book of Romans written by St. Paul that records a command made to the church in Rome in the 1st century A.D. Augustine takes this command to have been given directly from God to Paul and recorded in Romans, and then takes this command to also apply to himself. There are two possible accounts that can be given for the supposed discoursing between God and Augustine in this example. The first I will call authorial discourse,168 the second Wolterstorff calls deputised discourse. With authorial discourse, God speaks directly to the authors of scripture, who write down God’s words. In the above example, God would speak directly to Paul and

168 Wolterstorff doesn’t outline this possibility in his work, nor does he use the term ‘authorial’ as I am doing here. This is interpretive of some of Wolterstorff’s remarks.

141

Paul would write down what God said verbatim. Other people would then read what God had said by reading the writings of Paul, or some other biblical author. The distinctive nature of authorial discourse, therefore, is that God speaks, and biblical authors write down what he says, and other people read what God says by way of these authors. In analogical terms we can liken authorial discourse to a letter dictated and then given to other people for whom it may be relevant. So, the author dictates word for word what she would like to say to a scribe who writes the letter out and who subsequently passes this letter on to the audience that the letter is intended for. In the case of Augustine’s experience, authorial discourse would be the appropriate kind of divine speech if, in Augustine’s reading of the writings of Paul, he takes Paul to have recorded the direct words of God, and takes these words to have relevance for himself in his context and situation. Before proceeding to explain deputised discourse, it’s worth highlighting three issues with authorial discourse. First, there are questions concerning the reliability of any writing, be it from a religious scripture or from a historical text, which goes back as far as 2000 years. For someone today who would like to hear God’s very words, and holds an authorial discourse view of divine speech, she must accept that some of the writings may have been changed, lost or elaborated upon in the time since they were first written. To claim to be reading God’s words as they were originally spoken to an author in the 1st century A.D. one faces this significant challenge. Second, since authorial discourse maintains that all biblical writings are the words of God dictated to someone who writes those words down, authorial discourse doesn’t appear appropriate for most biblical writings. Much of the Psalms, history books and even Paul’s epistles express personal views directed towards God.169 In many cases God is directly addressed by the writer. Even though in some cases, such as with the prophetic genres, it might be apt to say that the author records God’s own words, this view looks implausible for most of the remainder of the Bible. Something similar can also be said of the Koran. Third, Augustine seems to have had a personal experience of hearing from God, but this account of authorial discourse removes any interpersonal or relational

169 That isn’t to say, though, that God wouldn’t be involved with saying fairly mundane things in the clearer cases of God speaking. Mavrodes (1988, pp.123-124) considers several cases in which fairly mundane testimony is attributed to God.

142 component. The letter was written to a particular community at a particular time, namely, the church in Rome, and so whilst it would have been personal for them, reading a letter intended for someone else is not quite so personal. We will see as we proceed how these three worries are problematic for, or are overcome by, the two remaining accounts. The second way we could understand Wolterstorff’s interpretation of Augustine is as deputised discourse.170 With deputised discourse the authors of scripture speak for God with an authority that has been deputised to them by God. So, they write portions of the Bible, and attribute this writing to God himself, because they have been given the deputised authority to do so. In this way, God speaks by way of the claims made by the authors of scripture. For example, when Paul writes to the church in Rome, ‘there is now no condemnation for those who are in Christ Jesus’ (Romans 8:1), according to deputised discourse, Paul writes this as someone who has the authority to speak for God, and so what has been written by Paul is effectively what God would have said had God himself been the writer or speaker. An analogy for deputised discourse, and one that Wolterstorff uses frequently,171 is the secretary preparing a document for an official. For instance, a secretary of the President will prepare certain documents on the President’s behalf. This could include, say, an executive order to appoint a new member of staff. The secretary will prepare a document that officially confirms this appointment, and the secretary will have the deputised authority to sign the document on behalf of the President. When the secretary signs the document, the secretary speaks for the president in virtue of deputised authority. By doing so the President performs illocutionary acts by way of what someone else says. So, it isn’t that the secretary performs an illocutionary act when writing and signing the executive order. Rather, the secretary merely performs a locution, and because of the authority given to the secretary, the secretary is able to assign an illocution to the President by way of the executive order. Wolterstorff says this explicitly with regard to Anthony – the character who inspired Augustine to seek out God’s speech:

170 This term and concept is taken directly from Wolterstorff (1996). He develops the account in Chapter 3. 171 See 1996, Chapter 3 (pp.38-51).

143

It was by way of that lector’s locutionary act of uttering those words that God performed the illocutionary act of speaking to Anthony. It was not divine discourse mediated by human discourse, but divine discourse mediated by human locution. (1996, p.189)

Deputised discourse is one kind of what Wolterstorff calls ‘double agency discourse’ (1996, p.38) since there are two agents involved in the speaking. One agent performs a simple utterance, but by way of that utterance a speech act is performed by another agent.172 Now, clearly not everything that’s said or written by the President’s secretary will count as the President’s own discourse. It will only count as the President’s own speech when the writing is officially attributed to the President, presumably through some stamp or seal. The same should also be true with writers of scripture. Sometimes what is written will merely express the author’s own views, and at other times it will attribute speech to God. We can then see how deputised discourse is able to avoid the problem noted in the foregoing concerning the various genres at play in the Bible. If God’s speech could be considered deputised in those instances where it appears that God is speaking but not in others then we needn’t consider the whole of the Bible to be God’s own discourse, but as a mixture of human and divine. Where, authorial discourse is committed to the claim that God speaks through all of the Bible, deputised discourse is not so committed. Nevertheless, a new problem does arise for this account that makes it slightly unappealing, for, we are left with a hermeneutical problem of interpretation. Thinking back to the example of Augustine, it isn’t absolutely clear why what is said should be taken as coming directly from God rather than from St. Paul. The same is true from the sentence I quoted in the foregoing. When Paul writes that ‘there is now no condemnation for those who are in Christ Jesus’, should this be taken as Paul’s words, or as Paul speaking for God? Granted interpretation may be simpler when it comes to the prophetic books. In those cases it often explicitly states that the prophet is speaking on behalf of God in a deputised way, or recording God’s own speech, as with authorial discourse. For most scripture, though, who should be identified as the speaker in any case is not obvious.

172 Deputised discourse is central in Wahlberg’s (2014) account, who takes Jesus to have the authority of speaking for God. I will explore parts of his account in the next chapter.

144

There is a separate but related problem with deputised discourse. It seems plausible that someone could hear from God even when some writing is not attributable to God. For instance, when St. Paul says that ‘there is now no condemnation for those who are in Christ Jesus’, even if Paul does not speak as a deputy on behalf of God, people often feel as though God is communicating to them via St. Paul’s words. It would be beneficial to have an account of divine speaking that allows for this fact, but only authorial discourse does allow for it. Although this problem is perhaps not decisive in ruling out the plausibility of the deputised model, it does motivate the need for a more explanatorily satisfying account. The third theory of divine speech gives us just such a theory.

4.3.2 Appropriated Discourse

We can begin by considering a second interpretation that Wolterstorff offers of the example given by Augustine. He proposes that

…by way of his now reading those words from Paul’s letter, God was here and now saying to him: Augustine, arm yourself with the Lord Jesus Christ and spend no more thought on nature and nature’s appetites.’ (1996, p.7)

As with our two previous accounts, the way that Wolterstorff now proposes that God speaks is by using the words of someone else. The difference is that with both authorial and deputised discourse, God’s words have been recorded in the past, and are then read by people at some later time. Now though, God is speaking in the present – the ‘here and now’. The way that God does this, Wolterstorff proposes, is by appropriating what someone else has said.173 He offers the following description of appropriation:

Sometimes one person says something and another remarks, “I agree with that” or “She speaks for me too” or “Those are also my convictions” or “I share those commitments”…Or a person says, in a parliamentary session, “I second the motion.” In all such cases, one is not just appropriating the text of the first person as the medium of one’s own discourse; one is appropriating the discourse of that other person…What the second person says is determined, in good measure, by what the first person said. (1996, p.52)

173 As with ‘deputised discourse’, the term ‘appropriated discourse’ is also Wolterstorff’s own.

145

With appropriated discourse, a first speaker speaks for herself, and a second person says precisely what the first person says by, for instance, saying she agrees with the first person, or saying that her thoughts are the same as the first person, or ‘seconding’ the first person’s remarks in some formal way. For example, in conversation with a friend she might say to me ‘I think that the new Labour policy on education is a farce’, and I might say precisely the same sentence by responding with a vigorous nod of the head, or by saying ‘I couldn’t agree more’, or ‘my views on the matter are just the same as yours’. When I do this, I appropriate my friend’s discourse and say what my friend said by way of appropriating what she said. Appropriated discourse is similar to deputised discourse for the reason that both are double agency discourse. However, rather than God’s words being authorised by a speaker who speaks for God, with appropriated discourse the speaker speaks for herself and God appropriates the words of the speaker and uses those words as an instrument of his own discourse. For instance, when Paul says ‘there is now no condemnation for those who are in Christ Jesus’, these words are Paul’s. They aren’t dictated to him, as with authorial discourse, nor are they attributed to God, as with deputised discourse. However, God may speak the exact same words by agreeing with, or ‘seconding’ them in some way. How God actually appropriates the words of another will be discussed extensively in §5.4.2. Suffice it to say for now, though, that one way in which God appropriates is by giving people a sense that God is speaking to them through what they are reading or hearing. In Christian theology, one can appeal to the work of the Holy Spirit to explain how God does this, as I will explain later. Perhaps what is most important to acknowledge for now, though, is that with appropriated discourse, even when someone reads scripture, it doesn’t follow that they are hearing from God, or even that they will have a sense that God is speaking. Whilst the primary speech that God uses to appropriate may be religious scripture, there may be other kinds of speech that he will use as well. For instance, God could appropriate the words of a preacher, a story with a poignant message, or even a person’s life events and experiences. For instance, somebody may be applying for a job and when reading the job description might feel as though God is saying to her that this is a job he wants her to have. This is a particular virtue enjoyed by the appropriation account that is lacking in the earlier two, but which helps to explain the

146 common claim that people hear God speaking to them in many different situations, and not merely through the reading of scripture.174 Despite the explanatory utility of the appropriation account, it fails to explain why it is that scripture in particular has priority as a source of divine discourse over and above any other discourse God might appropriate. Whilst people do claim that God speaks to them through some major life event, the religious scriptures are thought to hold priority as a means of divine discourse over these other kinds. Wolterstorff makes a similar claim to this and points to a possible solution:

[I]t would be bizarre to think of God as just finding books lying about and deciding to appropriate them; the appropriation model calls for supplementation with some doctrine of inspiration. (1996, p.187)

So, the reason that scripture enjoys priority as a particular kind of revelation is that it is inspired in some way by God. But what do we mean by ‘inspiration’? Do we simply mean that those who wrote the scriptures had encounters with God that inspired them to write about him? In that case we could mean that the authors of scripture were inspired by their experiences of God. Or is it more similar to the authorial discourse model in which God speaks to people and they record his words? In that case we could mean that scripture itself is inspired by God telling people things to write down. I propose that we needn’t actually settle what is meant by ‘inspiration’. This is because appropriation as a kind of divine discourse can work alongside one or many theories of inspiration. The point of appropriation is that God uses various means, including the writings of the religious scriptures, to speak to people in the present. That doesn’t change on either of the accounts of inspiration I proposed in the previous paragraph. All that an account of inspiration offers is why scripture has priority as a means by which God communicates to people. The appropriation theory of divine speech can accept that scripture is the primary means that God uses to communicate, without needing to commit to why this is the case. Wolterstorff remarks along similar lines when he says,

…it may not only be the case that [the Bible] was a medium of divine discourse which occurred back in antiquity – when Paul wrote it and sent it off, or when it became part of that single book which is the Bible. It may be

174 See Mavrodes’ (1988, Chapter 2) ‘Causation Model’ for a similar claim. This notion is also central in accounts of Reformed Epistemology (Plantinga and Wolterstorff, 1983).

147

a medium of contemporary divine discourse, a medium of God’s here and now addressing you and me, the generating event being some such thing as our now being confronted with the text or God’s now presenting the text to us. It appears to me that it was along these lines that Augustine interpreted what happened to him in the garden and what happened to Anthony in the Coptic church. (1996, p.56)

According to the appropriation theory, it may be the case that God discoursed with the writers of the Bible and they were inspired to write down letters that became scripture, but what is important is that God speaks now to people by appropriation, and in a way that is not limited to what we now call the Bible. As the previous quote also brings out, Wolterstorff declares that this is actually how we ought to interpret the experiences of Augustine and Anthony. On the appropriation model we also have a way out of the three problems I introduced with the account of authorial discourse. First, the reliability of the scriptures is not essential. What is important is that people can be justified in believing what God is saying to them in the present. For instance, if someone comes to believe that there is an eternal heaven and does so by reading the Bible, we needn’t check the reliability of the texts that make claims about heaven. This is because, on the appropriation model, someone believes that there is an eternal heaven because she takes God to be telling her in the present through reading the Bible that there is an eternal heaven. If she can be justified in believing God has told this to her in the present, then the reliability of the original speech that has been appropriated by God becomes less important. Second, the appropriation account allows that human authors are the writers of the Bible and other scriptures, which avoids instances where God is clearly not the speaker. Nevertheless, appropriation allows that God inspired these authors, and appropriates their discourse for his own means by speaking directly to people in the present. Third, since God speaks directly to people when he appropriates the speech of someone else, this is more intimate and interpersonal in a way that makes sense of many peoples’ experiences of hearing from God. These kinds of experiences tend to go along with trusting relationships, which is a critical component of Abrahamic faith.

4.3.3 Appropriated Discourse in Theology 148

Before moving on, it’s worth noting the support that the appropriation model has received from contemporary theology through the work of Karl Barth.175 Barth can be interpreted as sympathetic to a theory of appropriated discourse for divine speech. His support for the theory is beneficial since he offers a non-revisionary account of religious practice that aims to describe how people actually take themselves to hear from God. To begin with, Barth rejects both the authorial and deputised theories of discourse. He does so by way of a distinction between revelation and witnessing to revelation. For Barth, God revealed himself directly to those who would author the Bible, but the Bible itself is merely a human record of this revelation – a witness to the revelation – rather than a directly scribed record of what God had said:

In the Bible we meet with human words written in human speech…Therefore when we have to do with the Bible, we have to do primarily with this means, with these words, with the witness which as such is not itself revelation, but only…the witness to it’ (I.2. p.463)

This implies that the authorial model is false because the Bible is human speech not God’s own words, and the deputised model is false because the human authors do not speak for God with deputised authority, but merely record their own experiences. Nevertheless, the importance of the Bible is preserved since it records first-hand experiences of God’s revelation, through which, as we will see, God then speaks directly to others. Barth does not see the words written down in scripture as the Word of God – as God’s speech recorded directly: ‘the presence of the Word of God itself, the real and present speaking and hearing of it, is not identical with the existence of the book as such’ (I.1. p.530). Nevertheless, there is something ‘real and present’ about hearing God speak when reading scripture. He goes on to say that

…the Bible as it comes to us in this or that specific measure is taken and used as an instrument in the hand of God, i.e., it speaks to and is heard by us as the authentic witness to divine revelation and is therefore present as the Word of God. (I.1. p.530)

175 The quotations in the ensuing are from Barth’s Church Dogmatics (1956).

149

Here we see Barth’s implicit advocacy of the appropriation model since he claims that we hear God speak through the Bible by way of God using it as an instrument of his own speech. On this account, as we have seen, the Bible is used at particular moments and with particular people to speak directly to them, rather than being a direct record of divine speech. Barth adds to this point by referring to the hearing of God through reading scripture as an ‘event’. He takes events to be limited in duration, and thus by referring to divine speech as an event, it would follow that it has limited duration. This view is consistent with appropriation, where God’s speech is present for a fixed time, and inconsistent with the two other theories which treat God’s speech as recorded unchanging in the same state for all time. Interestingly for my purposes, Barth refers to faith as being identical with this ‘event’ of hearing from God: ‘faith in the promise of the prophetic and apostolic word is…an event, and is to be understood only as an event. In this event the Bible is God’s Word’ (I.1. p.109).176 So, to bring the discussion from this chapter full circle, it appears as though Barth also views faith as hearing from, and believing God.

4.4 Conclusion

We have found some historical support for the view that Abrahamic faith requires trusting what God has said – trusting divine testimony. This support is in addition to a number of recent authors who have endorsed the same or a similar view. This account of faith has two sides, as it were. First, there is the hearer of the speech, who receives what has been said in some way. In Chapter 2, I argued for an account of faith according to which faith requires trusting what God has said, and that this, in turn, requires believing what God has said. So, believing what God tells us captures one side to the notion of faith as believing divine testimony – the hearer side. Second, there is the speaker of what has been said – namely God – who can be seen as testifying to, or revealing certain claims or truths. In this chapter, I discussed three possible ways for God to speak, as drawn from the work of Nicholas Wolterstorff: authorial discourse, deputised discourse, and appropriated discourse. Depending on which

176 Barth emphasises in other places too that he views faith as part of the ‘event’ of hearing from God: ‘The man who so hears their word that he grasps and accepts its promise, believes. And this grasping and accepting of the promise: Immanuel with us sinners, in the word of the prophets and apostles, this is the faith of the Church’ (I.1. p.108)

150 model of divine speech we use, there will be implications for the epistemic status of the beliefs of the hearer. In the next chapter, I will draw out what these implications are, and defend the justification of DTT as a response to the PFR.

151

5. JUSTIFYING FAITH IN DIVINE TESTIMONY

5.1 Overview

In this chapter I will evaluate several ways by which the divine testimony theory (DTT) of faith can be epistemically justified, and hence, how DTT can be used to respond to the problem of faith and reason (PFR). I will situate this discussion within the context of leading theories from the epistemology of testimony. Doing so provides several novel ways of justifying someone’s faith when this is based on trusting God’s testimony. I will consider and reject two attempts to justify this kind of faith, before augmenting the work of Alvin Plantinga to develop a third way that I will argue can be used successfully to respond to the PFR. The chapter begins in §5.2 by looking at when someone is justified in believing the testimony of another person. I will address this by introducing the two central theories in the epistemology of testimony – reductionism and non-reductionism.177 In §5.3 I consider whether it is possible to use these theories to justify one in believing what God has said. I will argue that there’s a disanalogy between human and divine testimony that gives rise to an important objection: in human testimony, the existence of a speaker is a given, whereas in divine testimony, the existence of the speaker requires justification before one can believe what one takes oneself to have been told. The remainder of §5.3 considers a kind of reductionist approach to addressing this problem found in the work of John Locke, which I argue amounts to an attempt at natural theology. In §5.4, I consider two non-reductionist approaches that develop on the work in Chapter 4 where one approach is based on deputised discourse and the other on appropriated discourse. Although deputised discourse approaches to justification are possible, I will argue that they face significant challenges if they are to be defensible. §5.4.2 then defends a theory of justification for appropriated discourse that I believe offers a tenable response to the PFR.

5.2 Reductionism and Non-Reductionism in Testimony

177 For a ground-breaking study of these two contrasting positions, including a defence of non- reductionism, see Coady (1992). A brief and accessible overview is Lackey (2011), whilst a more substantial collection since Coady is Lackey and Sosa (2006).

152

5.2.1 Non-Reductionism

How are beliefs formed on the basis of trusting someone’s testimony justified? The first theory I will evaluate is often referred to as non-reductionism.178 Non- reductionism is committed to the claim that testimonial belief is basic. As such, one’s justification for testimonial belief does not come from epistemic sources, like induction and perception, which are prior to the testimony itself. Rather, one is justified in taking the testimony as basic without support from these sources:

[T]he essence of a nonreductionist theory is that testimonial knowledge is not ultimately reducible to purportedly more basic sources of knowledge and, accordingly, that knowledge can be acquired merely on the basis of a speaker’s testimony. (Lackey 2003, p.717)

The justification of testimony according to non-reductionism does not, therefore, depend upon any prior positive evidence a person has that demonstrates the competence and honesty of the speaker. Suppose someone tells you that the time is 2pm. According to non-reductionism, if you come to believe this proposition on the basis of the testimony, your belief can be justified without you first being justified that this speaker is competent and sincere. If it can be defended that testimonial beliefs are basic, it would then follow from the mentalist theory of evidence given in Chapter 1 that testimonial beliefs are sources of evidence, just like perception, memory and a priori intuition. It might seem that non-reductionism defends the justification of testimonially acquired beliefs regardless of the reliability or trustworthiness of the speaker. This could mean that non-reductionism defends gullibility as a justified response to testimony. For instance, if I don’t need prior evidence about the sincerity and honesty of the person I meet in the street who tells me that the earth is flat, then presumably I can be justified in believing this proposition. However, there is more to non- reductionism than not being required to have positive reasons for a speaker’s trustworthiness. Non-reductionism also states conditions on when someone is justified in trusting a speaker. These conditions have been adapted from the work of Thomas Reid (1983) who, it is often claimed, is the forerunner to non-reductionism. Reid

178 Supporters of non-reductionism include Audi (2001), Burge (1993), Coady (1992), Dummett (1994), Geach (1977, pp. 33-34), Goldberg (2007, 2010), Lackey (2003, 2008), and McDowell (1998).

153 maintains that ‘in the matter of testimony, the balance of human judgment is by nature inclined to the side of belief; and turns to that side of itself when there is nothing put into the opposite scale’ (p.95).179 180 Reid’s remarks appear to defend a condition that is central to non-reductionism: if a hearer has no all things considered reason to reject the testimony of the speaker, then she is justified in believing him.181 Let’s explore this proposed condition in a little more detail. One way to explain someone’s reasons to reject testimony is in terms of epistemic defeat. There are a number of types of defeaters, but perhaps the one most relevant for testimony is what we can call doxastic defeat.182 A doxastic defeater is a believed proposition D, which, when believed by a hearer H, conflicts with a speaker S’s testimony p. For instance, suppose S tells H that the football match starts at 7.30pm. H might already believe that the match starts at 7pm. Hence, H’s believed proposition conflicts with S’s testified proposition. Doxastic defeaters come in at least two kinds. There are those that are believed without sufficient justification – unjustified doxastic defeaters (UDD) – and those believed with sufficient justification – justified doxastic defeaters (JDD). In the prior example, if H believes D because she read it in the newspaper but she has terrible eyesight and wasn’t wearing her reading glasses at the time, then this can undercut the justification she has for D. This will result in D merely qualifying as a UDD. Given this, the defeater does not actually defeat the justification a hearer has for p, but may cause her to withhold belief. In that case, she has failed to acquire some knowledge that, in order to have been fully rational, she ought to have acquired.183 However, if she was wearing her glasses then she is in an ideal epistemic situation to form the belief and D will qualify as a JDD against p. We can see from this that the kind of defeater at work in non-reductionism

179 Reid defends this with his two principles: first, the ‘principle of credulity’, according to which God has given us a ‘disposition to confide in the veracity of others, and to believe what they tell us’ (p.95); second, the ‘principle of veracity’, whereby speakers have ‘a propensity to speak truth’ (p.94). 180 For more on Reid’s account of non-reductionism, see Wolterstorff (2001) and Audi (2006). 181 An early formulation of this claim is Burge’s ‘Acceptance Principle’: ‘A person is entitled to accept as true something that is presented as true and that is intelligible to him unless there are stronger reasons not to do so.’ (1993, p. 467) 182 For an overview of several kinds see Lackey (1999; 2003), and on doxastic defeaters see Bergmann (2005). 183 This implies that she was culpably ignorant in an epistemic sense. See Goldberg (forthcoming) for an account of epistemic culpable ignorance.

154 about the justification of testimony are JDDs since non-reductionism is a theory about testimonial justification, not about the sorts of reasons a person may have for failing to believe something she should have believed. Note that doxastic defeaters as I am construing them do not function in virtue of the truth-value of D but whether one is justified in believing D. That is, the testimony is not defeated just in case the hearer believed D and D is true, but believes D with justification, regardless of whether it is true or false. It is, then, their normative rather than their veridical status that makes them defeat the justification for an item of testimony. Another point worth noting is that the conflicting proposition believed need not be in regard to the testimony, p, but could be defeating of the testifier. That is, D might be a proposition believed about the speaker concerning her lack of credibility. One might believe, with justification, that the speaker has a tendency to fabricate the truth. For instance, we might have a friend whom we know enjoys telling us falsehoods to mislead us. In this case we would have a defeater that still undercuts our friend’s testimony, but rather than defeating the proposition the friend might tell us, it would simply undercut the credibility of our friend as a reliable testifier. Finally, defeaters themselves can be defeated. For instance, where H believes that the game starts at 7pm because she read it in the newspaper, and S tells her that the game starts at 7.30pm, she has a defeater for S’s testimony. However, she might check the online TV guide and discover the time stated there is 7.30pm, and hence she would have a defeater for her own defeater (belief). And we can extend the example even further such that one’s defeater-defeaters can themselves be defeated by other beliefs one comes to hold. In order for the speaker’s testimony to be defeated by one’s own JDDs, one must have no undefeated JDDs for the testimony or against the testifier. If one has no undefeated JDDs for p or against the speaker S then one is justified in believing the testimony according to non-reductionism. Another way that this could be put is that defeaters give one a normative reason to reject the testimony. Defeaters for defeaters are taken into consideration as one’s reasons to reject or believe the testimony. Once all defeaters are considered, someone will have an all things considered reason to reject or believe the testimony. The discussion of defeaters tells us what these reasons are and how they are weighed and how they interact. On this fairly simple account of non-reductionism, the following conditions hold: 155

Non-Reductionism: When a speaker S tells a hearer H that p, H is justified in believing S that p iff (a) H has no undefeated justified doxastic defeaters that are sufficient to count against p or against the credibility of S.

Let’s move on now to consider what conditions are required by the rival theory of reductionism, and which of the two theories is the most plausible.

5.2.2 Reductionism

The central claim of reductionism in testimony is that testimonial knowledge is non- basic. As such, one’s justification for testimonial belief comes from epistemic sources that are prior to the testimony itself, in particular, the role of induction. The reductionist claim is that for testimonial beliefs to be vindicated as a source of knowledge, they must be grounded in more basic sources. This view is often attributed to Hume, when he says, for instance:

[O]ur assurance in any argument of this kind [from testimony] is derived from no other principle than our observation of the veracity of human testimony, and of the usual conformity of facts to the reports of witnesses…The reason, why we place any credit in witnesses and historians, is not derived from any connexion, which we perceive a priori, between testimony and reality, but because we are accustomed to find a conformity between them. (Hume 1975, pp.112-113, §88-89)

We can see here how Hume clearly contrasts with Reid and non-reductionism in requiring that we have observed a conformity between speakers and the truth of what they say. The implications of Hume’s view are that one must have positive reasons to accept a testimonial report, and, as Lackey explains,

…they are typically the result of induction: for instance, we observe a general conformity between facts and reports, and, with the aid of memory and reason, we inductively infer that certain speakers are reliable sources of knowledge. In this way, the justification of testimony is reduced to the justification we have for sense perception, memory, and inductive inference.’ (Lackey 2003, n.1)184

184 For further discussion of this view see Fricker (1995, p.394)

156

According to reductionism, then, for any instance of testimony, the hearer must have positive reasons for thinking the speaker is both epistemically competent and reliable as a speaker for her to have justification for believing the testimony. These reasons might come locally, as when we have prior experience of this particular speaker. There is a strong correlation between what my spouse tells me and the truth of what she tells me, so I can infer from this that when she tells me p, it will likely be the case that p, and the positive reasons I have for trusting my spouse justify this inference and my belief in the testimony. Positive reasons might also come globally, as when we see a general conformity between speakers and their competence and sincerity. On the whole, I find that speakers tend to tell me the truth, and this inductive evidence can be used to draw the inference that when someone tells me p, it will likely be the case that p, and again, the evidence I have justifies the inference and my belief in the testimony. Although reductionists require that a further condition be met before one is justified in believing testimony, they are still committed to the first requirement set by non-reductionists. If they weren’t then someone could be justified in believing testimony in virtue of some positive epistemic reasons, even though these are undercut by defeaters that a person also has. To avoid such a contradictory theory, reductionism simply requires that both conditions be fulfilled for testimonial justification. A straightforward way to construe reductionism, then, is in terms of reductionism requiring an extra necessary condition over and above that of non-reductionism:

Reductionism: When a speaker S tells a hearer H that p, H is justified in believing S that p iff (a) H has no undefeated justified doxastic defeaters that are sufficient to count against p or against the credibility of S, and (b) H has positive reasons in favour of S’s credibility.

We can see here that it is condition (b) that distinguished the two theories.185 Before moving on to evaluate which theory is the more credible, I want to briefly note two things. First, reductionism is not particularly popular, especially given the attacks that it received in the work of Coady (1992, Chapter 4), which will be addressed presently. Later in this chapter I will outline a modified version of reductionism that is very different from the theory discussed in this section. I will

185 As Lackey puts it, ‘What divides nonreductionists and reductionists…is whether…other sources are needed in order to ground the justification of beliefs acquired via testimony’ (2003, p.717).

157 make it explicit in what ways it differs, and why that version is in fact unaffected by Coady’s objections. The reasons why will become clearer later. The second point I want to raise is that according to reductionism, when taken together with the mentalist theory of evidence assumed in Chapter 1, testimony does not count as a source of evidence. This is because only basic sources of knowledge are evidential. This will apply even in my later formulation of modified reductionism, and hence, divine testimony is not a source of evidence for either reductionist account. It will require other forms of evidence to establish its own justification. There are several arguments that have been raised against reductionism. If reductionism looks untenable, then we have at least some reason to favour non- reductionism. I will briefly present three objections to reductionism that are worth bearing in mind when I consider versions of reductionism and non-reductionism for divine testimony. The first objection to reductionism is that it looks like there are many situations in which we are unable to acquire the relevant positive reasons required to tell that a speaker is credible, and yet we are able to acquire knowledge from such people. Consider the following example:

[U]pon arriving in Chicago for the first time, I may receive accurate directions to Navy Pier from the first passerby I see. Most agree that such a transaction can result in my acquiring testimonial knowledge of Navy Pier’s whereabouts, despite the fact that my positive reasons for accepting the directions in question—if indeed I possess any—are scanty at best. (Lackey 2011, p.77)

This kind of example seems to show that the standards set by reductionism are too strong. We often acquire knowledge from people that we have only just met, without knowing a thing about the speaker. For the second objection to reductionism, recall that Hume claims that the assurance we have that comes from testimony is given by ‘our observation of the veracity of human testimony, and of the usual conformity of facts to the reports of witnesses’. This might be true for some cases, say, when a friend regularly reports to me the location of the meeting room and is always correct in his report. However, it looks like for the vast majority of cases, I must rely on others’ testimony to establish the reliability of the speaker.186 I lack the conceptual resources to check whether a

186 It is also problematic that such a demand will make the testimonially acquired beliefs of children largely unjustified who are unable to check the conformity of facts to such reports.

158 doctor, physicist or neuroscientist is reporting in a way that conforms to the facts. Only people who possess such abilities can undergo the checking process. In order for me to know whether such people are credible testifiers, I must myself rely on the testimony of those with the ability to do the necessary cross-checking. What’s worse is that this appears to be a fairly widespread phenomenon. I cannot tell whether almost any of the testifiers I acquire knowledge from are generally reliable without relying on the testimony of others to do so. But if this is the case for the reductionist, then, as Coady observes, she is

…involved in vicious circularity, since the experience upon which our reliance upon testimony as a form of evidence is supposed to rest is itself reliant upon testimony which cannot be reduced in the same way. (Coady 1992, p.81)

The problem of circularity for the reductionist is that the positive reasons that count in favour of believing a speaker ‘cannot themselves be ineliminably based on the testimony of others’ (Lackey 2003, n.1).187 One way that the reductionist could respond to the circularity problem is by setting up reductionism as a thesis about people in general, rather than any individual speaker. Reductionism would then only be a thesis about global justification, rather than local justification. Providing one observes a conformity between facts and reports in one’s everyday experience of testifiers, then one could gain justification for believing people in general in a way that does not rely upon the testimony of others. This should get around the problem of circularity. However, there is a further problem afflicting reductionism that a theory of global reductionism is unable to overcome. For, the reductionist theory apparently allows that there can be communities of people for which we can never be justified in believing what they tell us. This would be the case if there was no conformity between what was reported and facts reported to. To see this, Coady (1992, pp.84-90) imagines a community of ‘Martians’, who

…constantly misinform each other…They always, for instance, tell each other the wrong time and date, give their names and addresses wrongly, say falsely

187 For a response see Shogenji (2006) and Fricker (1994, 1995).

159

what the weather is like outdoors, and give false information about where they have been and what they have been doing. (pp.85-86)

In such a situation, there would be no conformity between facts in the world and the reports of the Martians, be it locally or globally, and so one could never be justified in believing the testimony of the Martians. If this is the case though,

…[i]t is therefore very hard to imagine the activity of reporting in anything like its usual setting with the Martians, for there would surely be no reliance upon the ‘reportive’ utterances of others. (p.87)

This is problematic, though, for any account of testimony since the reports given in testimony, as we saw in §2.3.3, are subject to norms that constitute the speech act involved in testifying. Such norms involve only testifying what you have justification to believe, and our social awareness of such norms is what makes us more disposed towards relying on others’ testimony. And so, ‘With no reliance…on the utterances of others, the Martian community cannot reasonably be held to have the practice of reporting’ (p.87). At least some adherence to the norms governing testimony and some reliance on the testimony of others is a requirement for testimony to even be possible. Where these are absent, there is no testimony. What this means, then, is that if reductionism allows for communities like the Martians in which there can be no conformity between reports and facts, then these communities are not even involved in the practice of testifying. This situation is possible due to the reductionist requirement to have positive reasons for regarding the speaker as a credible testifier. Since non-reductionism is not committed to this condition, then it is not committed to this unpalatable scenario. So it looks like we have several good reasons to be dubious of the reductionist constraint to have positive reasons to believe a speaker before one’s testimonial beliefs are justified. Reductionism denies testimonial knowledge where it seems we have such knowledge, and it leads to circularity and scenarios in which testimony is not even given. We are now at the stage where, from Chapter 4, we have three theories of divine testimony, and from this section, we have two theories for the epistemic justification of testimony. Although the authorial model of divine testimony, and reductionism as an epistemological theory of testimony are perhaps the more unfavourable theories, I will keep them in play for now as rhetorical devices; they will help to more clearly define the theories that we will see are more plausible.

160

With these theories in hand we can turn to the question of the justification of believing what God tells people to assess whether such a view can gain support from theories in the epistemology of testimony.

5.3 Modified Reductionism in Divine Testimony

5.3.1 The Absent Speaker

How do we evaluate the beliefs a person holds that are grounded in alleged divine testimony? One option is to draw from one of the theories of epistemic justification. Let’s assume, for instance, that non-reductionism about testimony is correct. Next, suppose that the claim that God speaks can be understood in terms of divine testimony. If one has no undefeated defeaters for God’s credibility as a speaker, nor for the propositions God asserts, then one can be justified in believing divine testimony. Moreover, such beliefs will be basic, and will hence qualify as evidence. Beliefs held in this way, if true, will subsequently qualify as knowledge, can justify beliefs that are inferred from it, and will show the PFR to be false. Call this approach, non-reductive divine testimony (NRDT). There is a fairly immediate problem facing any attempt to establish NRDT. The problem is that it’s not obvious that what someone alleges to be divine testimony is actually God’s speech. When someone claims that God spoke to her when she was reading scripture and told her that Jesus died for her sins, it’s not obvious that God has actually spoken to her. As such, any comparison between human and divine testimony is disanalogous. In cases of human testimony there is no question over whether one has in fact been testified to. The existence of the speaker is never called into question. However, in the case of divine testimony, God is not clearly identifiable as the speaker in the discourse. It is not obvious that God discourses through scripture, speaks through an experience, tells us something through a song, or even that prophets speak for God. There appears to be a prior problem when it comes to attributing speech to God, according to which one must identify God as the speaker before one can consider whether or not one is justified in believing what has been said. The issue of being required to identify the speaker first is unique to the context of divine testimony. The point at issue here is that if one wants to appeal to theories of testimony to justify religious belief grounded in supposed divine speech, then one must first be

161 justified in believing that the divine has spoken. This is what we may call the problem of the absent speaker. It has been articulated by John Lamont:

On the non-reductionism view, testimony gives knowledge when the speaker who is believed is honest and knowledgeable. In the case of God, there is no difficulty about the speaker’s honesty and knowledge, since these are possessed necessarily. Instead, the difficulty lies in identifying divine speaking. In most of the Christian tradition this speaking is seen as being done by human instruments…and as occurring at least in the scriptures, and for many Christians in the teaching of the Church as well. (Lamont 2009, p.115)188

The disanalogy between divine and human testimony makes it seem as though NRDT cannot even get off the ground since it looks necessary that one is justified in believing that God has spoken to her before she is justified in believing what has been said. As such, the justification one has for believing what has been said is mediated via the justification one has for the prior belief that God has spoken. This means that the justification of propositions believed on alleged divine testimony reduces to the justification for believing God has spoken. This kind of structure to justifying belief in revelation goes back at least as far as the work of John Locke, who offers an alternative to NRDT. Let’s look at Locke’s theory to see if it is credible before returning, in §5.4, to consider two possible versions of NRDT, and their solutions to the problem of the absent speaker.

5.3.2 Locke’s Modified Reductionism

John Locke was highly critical of what he referred to as ‘enthusiastic’ believers in revelation who, if they did not know that ‘God is the revealer’ of what they take to be revealed by God, had a mere ‘groundless’ assurance in the claims revealed, and that ‘whatever light [they] pretend to, it is but enthusiasm’.189 Having knowledge that God

188 The claim that God possesses honesty and knowledge necessarily looks uncontroversial, but some recent research has shown that in the Islamic tradition, the Koran claims that Allah is occasionally deceptive (Baldwin and McNabb, forthcoming). This gives the Muslim a defeater for her justification for believing God as a speaker, and so a reason why NRDT might not be open to her. 189 The quotations in this section are taken from Locke (1689/1924). His discussion of the distinctions between faith and reason, as how these apply to divine revelation are in Book IV, Chapters XVIII and XIX.

162 is ‘the revealer’ of any proposition is necessary, for Locke, before one can be justified in believing something they take to be revealed. This is right at the heart of Locke’s own views of the PFR given his definitions of both faith and reason. Locke ‘contradistinguishes’ faith from reason by, as we saw in Chapter 1, espousing the view that faith is simply believing propositions as though they have come from God as the speaker (or revealer). In contrast, Locke defines reason as

…the discovery of the certainty or probability of such propositions or truths, which the mind arrives at by deduction made from such ideas, which it has got by the use of its natural faculties, viz. by sensation or reflection. (Ch XVIII, §2)

There is some comparison to be made here between Locke on ‘reason’ and Aquinas on ‘Scientia’, as discussed in §4.2. Both are taken to yield a truth of which we are certain, and both appear to endorse a kind of epistemic internalism, since there is some awareness from the subject of why she has arrived at truth. It is either by way of some perception, or through a priori deduction. Contrastingly, faith does not enjoy such certainty, being simply belief held on presumed divine revelation. But Locke does not take faith to be incapable of epistemic justification. In fact, in a way similar to Aquinas, Locke first grants that ‘Whatever God hath revealed, is certainly true; no doubt can be made of it’ (Ch XVIII, §9). So, the propositions held on faith, if actually coming from God, are just as certain, if not more certain, than those held on reason. However, Locke goes on to repeatedly claim that whether what is thought to have been revealed is ‘a divine revelation, or no, reason must judge’:190

Because the mind, not being certain of the truth of that it does not evidently know, but only yielding to the probability that appears in it, is bound to give up its assent to such a testimony, which, it is satisfied, comes from one, who cannot err, and will not deceive. But yet, it still belongs to reason, to judge the truth of its being a revelation, and of the significance of the words, wherein it is delivered. (Ch XVIII, §8)

The central argument that Locke appears to be leaning on here (though there are others) is something akin to the problem of the absent speaker. We lack certainty over

190 The claim that ‘reason must judge’ whether a particular source is divine revelation or not is made repeatedly throughout the two chapters.

163 whether God has actually spoken in a way we do not when we receive the testimony of another person. The position he is keen to avoid is where someone is simply convinced by the certainty she feels over her own convictions. In order to avoid this, one must judge any claim to have heard from God by one’s natural faculties of reason. Locke’s final dictate on this matter is that ‘God, when he makes the prophet does not unmake the man’ (Ch XIX, §14) – we are still under our own faculties of reason to judge whether any supposed instance of divine revelation is in fact such a thing, and only once this reason is satisfied will we be justified in believing that it is. In the next section I will look briefly at what Locke’s own views were on how one might use reason to determine whether God had revealed something to someone. Before we move on to that issue though, it’s worth having a clear statement of the epistemic structure of Locke’s account. Locke is insisting that before we can be justified in believing p that we take to have been spoken to us by God, we must first be justified in believing that God has spoken. Belief in that proposition, i.e. God has spoken to me, Locke seems to have suggested, is non-basic, since it requires evidence from more basic sources for its own justification. (I will be exploring the non-reductive view that it is basic in §5.4.) So, Locke’s account holds that belief in any proposition inferred from the belief that God has spoken and told me p will derive its justification from prior sources of justification. As such, justification for beliefs held on sources of divine testimony supervene on the justification one has for believing one has been testified to by the divine: the justification is dependent and asymmetric.191 Given these claims concerning the structure of justification, it follows that beliefs held on divine testimony, on Locke’s model at least, are non-basic. But it doesn’t follow that Locke is committed to reductionism about testimony. Reductionism in testimony is a thesis concerning the credibility of a speaker and what it takes for one to be justified in believing her testimony. Reductionism in divine testimony, on Locke’s account, is a thesis concerning the very existence of a speaker and what it takes for one to be justified in believing God has testified. I propose we call this alternative thesis Modified Reductionism, since it uses a non-reductionist

191 If the relation between the beliefs is non-causal, then it looks like a kind of grounding relation (see the collection in Correia and Schnieder 2012). It isn’t certain that this is the case though. The higher-level justification may well be caused by that at the lower level.

164 strategy for assessing the credibility of the speaker, and therefore rejects reductionism about testimonial justification, but requires a reductionist treatment of believing one has been testified to by God. We can call Locke’s account a modified reductionism for divine testimony (MRDT). As an account of speaker testimony, MRDT will still be subject to at least those epistemic demands as imposed by non-reductionism. Hence, one of the constraints on being justified in believing that p on the basis of divine testimony is that one have no undefeated defeaters sufficient to count against p or God’s credibility as a speaker. Should we also require the stronger claim made by reductionists that we also require positive reasons in favour of God’s credibility? The prima facie response to this is ‘no’. This is because, as Locke notes, God is a speaker who ‘cannot err’, whose knowledge is ‘certainly true’. It wouldn’t seem plausible to require extra demands on hearer’s to be justified in believing that God is credible. I suggest, then, that MRDT only borrow from the weaker, non-reductionist conditions, rather than those from reductionism.192 The second and more pertinent issue addresses what the requirements will be for one to have more basic evidence in favour of the belief that God speaks. Although I will make some suggestions for this in the ensuing, at this stage we simply need to set out a condition for such approaches to be met. We needn’t devise anything new here. In Chapter 1 I set out an analysis for epistemic justification in terms of evidence. One’s non-basic belief D that p is epistemically justified if and only if D fits the evidence for p. I suggest, then, that with these two issues in mind, we state MRDT as follows:

Modified Reductionism: When a hearer H believes herself to have been told by God that p, H is justified in believing God that p iff: (a) H has no undefeated justified doxastic defeaters that are sufficient to count against p or against the credibility of God (non-reductionism); and (b) H has sufficient evidence to believe that God has testified to her that p (epistemic justification).

From this account we can see that condition (b) sets a requirement on what it is to believe that God has testified at all, and (a) is the standard condition, held by both

192 Although see n.188 for problems with this view.

165 reductive and non-reductive accounts of testimony, for when a hearer is justified in believing a speaker. For the remainder of this section, I will only discuss issues relating to condition (b). With regard to divine speech, MRDT is best suited to the appropriation model. This is for two reasons. First, because MRDT can take many sources to be instances of divine speech given that it confronts the problem of the absent speaker head on. It accepts that God, as a speaker, appears absent, and attempts to justify belief that God has spoken. MRDT can allow that God speaks to people, in the here and now, through experience and personal scripture readings, in a way consistent with the experiences of everyday believers. Second, MRDT is a theory of the justification of an individual who is experiencing God speaking to her, rather than a record of previously spoken- to individuals. Being so compatible with the appropriation model is beneficial given the very notion of faith as believing divine testimony. Since relational faith is interpersonal, the appropriation model captures the notion of faith as God speaking to me, rather than God speaking to person S and S communicating it to me (as with deputised discourse). This consistency with both people’s individual experiences of divine speech, and with our account of faith looks like a bonus for MRDT. So how did Locke think that (b) could be satisfied? As we will see, the very requirement to have (b) satisfied before belief is justified is part of a more general problem in the philosophy of religion.

5.3.3 Justifying Faith Reductively

In order for one to be justified in believing that God has testified, Locke draws from early biblical characters to see how they took themselves to hear from God: ‘the holy men of old…were not left to their own persuasions alone, that those persuasions were from God; but had outward signs to convince them of the author of those revelations.’ (Ch XIX, §15). And what were these ‘outward’ signs that authorised the revelations as coming from God? In a word, miracles:

Moses saw the bush burn without being consumed, and heard a voice out of it. This was something besides finding an impulse upon his mind to go to Pharaoh, that he might bring his brethren out of Egypt: and yet he thought not this enough to authorise him to go with that message, till God by another

166

miracle, of his rod turned into a serpent, had assured him of a power to testify his mission by the same miracle repeated before them, whom he was sent to. (Ch XIX, §15)

In the narrative appealed to here, Moses believes himself to have been told by God to instruct the Pharaoh of Egypt to release the Israelite people from their slavery. According to Locke, to demonstrate to Moses that this testimony does in fact come from God – to justify his belief – and show that it is more than a mere ‘impulse upon his mind’, God performs several miracles. He speaks through a burning bush, performs miraculous acts on Moses’ staff, and repeats these acts in front of Pharaoh. Presuming that Locke expects that a similar standard must be met for all people’s beliefs that God has spoken to them if they are to be justified, this seems like a high standard indeed. What is the exact structure to the justification here, and what does this tell us about the kind of problem Locke’s reductionism is addressing? To begin with, Moses has a basic perceptual belief, (P1) ‘There is a bush burning in front of me that is not being consumed’. Add to this a second basic perceptual belief, (P2) ‘A voice is coming from the bush telling me that it is from God and I must tell the Pharaoh to release the

Israelites’. Now add to this repeated incidents in which Moses believes that, (P3/P4/P5)

‘My staff has turned into a serpent’. Provided that (P1) and (P3/P4/P5) are connected to authorising the statement made in (P2), then Moses seems entitled to make the inductive inference that, (P6) ‘God has told me that I must tell the Pharaoh to release the Israelites’. As we can see belief in proposition P6 is not basic, but inductively inferred from more basic propositions. Hence, the justification it receives is derived from the justification one has for the prior propositions. There is more, though that can be derived from this chain of reasoning.

Suppose one believes, and perhaps quite rightly so, that, (P7) ‘If a person speaks then that person exists’.193 It seems to follow from (P6) and (P7) that, (P8) ‘God exists’. Now, this proposition is a deductive inference: God’s existence here is deductively entailed by other propositions one believes, and these are believed inductively, from inference based on prior evidence and experiences one has. So ultimately, Locke’s model for justifying divine speech is simply a special instance of an argument for the existence of God that we might call the argument from divine speech. Theologian Mats

193 We could view this as an alternative to Decartes “cogito”. Rather, the claim would be ‘I speak therefore I am’ (loquor ergo sum).

167

Wahlberg has also drawn this connection: ‘To identify divine speech as divine (i.e., as God’s speech) means to acquire knowledge that God speaks, knowledge entailing, of course, that God exists’ (2014, p.144). As such, the NRDT model, as a means of overcoming the problem of the absent speaker, is directly relevant to natural theology’s aims of providing arguments for the existence of God. Evidence for divine testimony is also evidence for the existence of the divine.194 The problem with this method is that if one wants to appeal to modified reductive theories of testimony to justify religious belief grounded in divine speech, one needs to recognise that there is a danger that this approach simply collapses into the general issue of establishing the existence of God. This issue generalises to other means of using the reductive approach too. Consider another argument given, this time, by the theologian John Owen, who appeals to the Bible as God’s means of speaking to people:

The principal divine effect of the word of God is in the conversion of the souls of sinners unto God…of this great and glorious effect the word is the only instrumental cause, whereby the divine power operates and is expressive of itself…The work which is effected by it, in the regeneration, conversion, and sanctification of the souls of believers, doth evidence infallibly unto their consciences that it is not the word of man, but of God. (Owen 1852, pp. 94- 95)195

Perhaps similarly to Locke, Owen here is appealing to ‘divine effects’ from reading the Bible as authorising it as coming from God. Rather than outward miracles though, the idea is that inward changes in a person demonstrate that the Bible is God’s own speech.

194 John Lamont has remarked similarly, ‘How are we to identify the assertions of these [biblical authors] as God’s assertions without appeal to [natural theology]?’ (2009, p.115) 195 Could this be interpreted as a non-reductive approach to justifying divine speech (as in Lamont 2009, pp.115-116)? This might depend on whether you are an internalist or externalist about epistemic justification. One might appeal to an externalist notion of reliability of the Word in converting, and hence coming from God. In that case, beliefs held on this basis might be properly basic even if the believer is unaware of what makes them justified. Alternatively, if we must be aware of this converting effect, as would be the case with internalism, then it looks like we need prior evidence of this before we are justified in believing it has come from God. However, the externalist approach will hardly meet reliabilist standards compared with, say, visual perception. For instance, many have read the bible without being converted, so it doesn’t look particularly reliable.

168

Now, Owen’s argument doesn’t seem particularly strong. Plenty of people have read the Bible without having a conversion experience. However, this is not the issue I want to address. What I am drawing attention to is the appeal to the modified reductive method. Owen is looking to draw an inference between evidence of conversion upon reading the Bible, and the Bible being God’s actual speech given to people. The inference is inductive, and if the evidence were strong enough or particularly convincing, one might be justified in believing that God has spoken to her through the Bible. And again, we can form the deductive inference that God exists from this. There will be other means, too, of constructing a similar justificatory structure, drawing on other examples from natural theology. The point is that using the modified reductive method to establish that the alleged testimony comes from God is ultimately a special instance of an argument for the existence of God – an instance of natural theology. There is nothing, in principle, wrong with attempting to follow this method. However, for my purposes, appealing to this method would be a somewhat disappointing result. Thinking back to the PFR as given in Chapter 1, we defined the evidentialist objection as the claim that, ‘The beliefs that constitute a person S’s faith do not fit the evidence’. The reductive response attempts to address this objection by giving evidence for any belief held on alleged divine testimony. For instance, say a person believes that Jesus is the son of God and does so because she believes that God has told her this proposition. The modified reductionist then attempts to give evidence that God has told her the proposition, and if that can be established, then the proposition can be believed with justification. As I have been arguing though, giving evidence for the claim that God has spoken amounts to evidence that God exists. As such, this approach looks very much like a special instance of natural theology. The non-reductive response to the evidentialist objection is somewhat different though. When a person believes that Jesus is the son of God because she believes that God has told her this proposition, rather than trying to give evidence for this the non- reductionist strategy is to turn the structure of justification on its head. So, instead of giving evidence that God has spoken, it claims that the believed proposition, and the belief that God told the hearer this proposition, is properly basic, from which it would follow that God has spoken. What this means is that the testimony itself is evidence for, and justifies belief in the claim that God has spoken, rather than the other way round in which evidence is required that God has spoken before the testimony itself is 169 justified. To put it another way, the question I wanted to ask is, whether, rather than requiring other basic beliefs as evidence that the testimony comes from God, can one be justified in holding the testimony as properly basic, without prior evidence that it comes from God?196 If such an approach is feasible, then there will be more than one means of justifying beliefs that come from supposed sources of divine speech. I want to leave modified reductionist accounts behind at this stage. They aren’t the kinds of theories that are relevant to my aims in this thesis. I don’t deny that they can be defended, but they don’t appear to be particularly unique, or independent of the discussions surrounding natural theology. In the final section, then, I want to show how it is possible to have a non-reductive theory of divine testimony, according to which divine testimony is a basic source of evidence, and, in particular, how the problems of the absent speaker can be overcome whilst still adhering to a non- reductionist account.

5.4 Non-Reductionism in Divine Testimony

5.4.1 Deputised Discourse: Wahlberg

The details of a non-reductive approach to justifying beliefs held on divine testimony depend in large part on the nature of the discourse being used. I want to begin first by assessing one recent attempt at NRDT that uses the deputised mode of discourse. The model has been recently proposed by Mats Wahlberg. I will give a fairly basic overview of his argument without paying great attention to detail. The reason for this, as will become clearer shortly, is because Wahlberg includes many theological and historical components in his theory, and these aren’t the sorts of things that I want to dedicate sustained philosophical reflection to. Wahlberg’s account begins by proposing the following:

Suppose that the prophet P speaks in the name of God. This means that (some of) P’s utterances constitute divine speech-acts. When P speaks, God speaks. Suppose, furthermore, that P – speaking in the name of God – says that she speaks in the name of God. Since God knows what he is talking about, and

196 This strategy is closely association with the work of Alvin Plantinga and Nicholas Wolterstorff (1983), and has come to be known as ‘Reformed Epistemology’. Later in this chapter I will look at modelling a theory of justification on Plantinga’s more recent theory (2000).

170

since P’s utterance constitutes (“counts as”) a divine assertion, it follows that P’s utterance makes knowledge of the proposition that P speaks in the name of God publicly available. (Wahlberg 2014, p.145)

This approach to justifying belief in divine speech assumes two of the things already discussed thus far. First, it uses the deputised model of divine speech in which a person has been given the authorisation to speak for God. This person is a prophet, and so this could be someone like Jesus or Mohammed. Second, Wahlberg works within non- reductionism in testimony in that one is justified in believing divine testimony providing one has no undefeated defeaters against the speaker or what she asserts. With the components of the theory in place, we can formulate Wahlberg’s argument, including his solution to the absent speaker problem, as follows:

(1) (Prophet) P claims to speak for God. (2) If (hearer) H is justified in believing that P speaks for God then H is justified in believing that what P says is from God is in fact from God. (3) H is justified in believing that P speaks for God iff H has no undefeated justified doxastic defeaters that are sufficient to count against the claim that P speaks for God or against the credibility of P. (non-reductionist assumption) (4) H is justified in believing that P speaks for God. (5) Therefore, H is justified in believing that what P says is from God is in fact from God.

If this argument is sound then it follows from (5) that if H does in fact believe those propositions that P asserts and that P’s claims are from God, then providing that those propositions are true, H gains basic knowledge via testimony from God. Looking at the argument closely, we can see that several of the premises are fairly plausible. (1) fits with the deputised model of divine discourse as already outlined. (2) also appears acceptable, particularly when we consider analogies to human deputised discourse. For instance, it seems reasonable that if I am justified in believing that the secretary speaks for the manager, then when the secretary says that the manager says p, it follows that I am justified in believing that the manager says p. There will only be controversy surrounding (3) if we wish to reject non-reductionism about testimony. Given the weight of support for non-reductionism, (3) is fairly safe to assume for the time being. Finally, (4) represents the main point of contention. The problem with (4) is that one can easily imagine defeaters for someone’s claim to speak

171 for God. In fact, regularly we might imagine that someone who claims to have the deputised authority to speak on God’s behalf is either incredibly special – as in the case of The Pope – or a madman – as is the case when ordinary folk make this claim. Either case seems to present one with enough of a sceptical problem to defeat one’s justification for holding that belief, and hence, enough to prevent the belief from yielding knowledge. Wahlberg does not fail to recognise this problem. Rather, he attempts to show, in a similar way to Locke, that such defeaters can be overcome by appeal to miracles. It should be stated that at this point his account narrows its focus substantially to focus singularly on Jesus as the prophet P. He claims that a first-century Jew who is told by Jesus that he speaks for God is justified in believing that claim providing that the Jew actually witnessed the miraculous events the Bible reports to have been performed by Jesus.197 As such, miracles perform a particular epistemic function in justifying the claim that Jesus spoke on behalf of God, and so what Jesus said in this capacity was God’s own speech.198 This epistemic function was also affirmed by Thomas Aquinas, who claimed that miracles ‘are rightly appealed to in proof of the Faith…For a pronouncement issued by a man with a claim to divine authority is never more fittingly attested than by works which God alone can perform’ (2002, p.153). Now, even though it seems plausible that someone living at the time of Jesus who witnessed numerous miracles, including a bodily resurrection, would be warranted in his belief that Jesus speaks for God, why should we expect that someone living today can receive the same degree of justification? Surely there is quite a difference between observing the claims made by an individual to speak for God and various miraculous actions that work to corroborate these statements, and hearing about those claims and events some 2000 years later. In the latter case – the situation with religious believers today – people hear of these miraculous events through biblical record and the testimony and tradition of the Church. This is the situation as Wahlberg sees it, but he attempts to defend the view that one can still be justified even in this case. His argument for this is both theological and historical. Wahlberg makes a concerted effort to establish that the Gospel records are accurate, that they have been

197 The central miracle being, of course, that Jesus rose from the dead (see pp.152-158). 198 Other theologians have made similar claims: ‘it is reasonable to think that [Jesus], as well as his followers, saw his miracles as testifying to his being a true messenger from or agent of God’ (Sanders 1985, p.173).

172 preserved throughout history, and that the Church and its traditions attest to this fact.199 Knowledge from God is then obtained from Jesus as the prophet speaking in God’s name, who is recorded in writing, which is preserved and passed down through biblical record. Is Wahlberg’s approach a plausible approach to NRDT? I am sceptical, and I have three reasons for why this is the case. First, it would be an impressive feat if someone could demonstrate that the Bible, as we have it today, accurately records the precise words said by Jesus. There are a number of problems with achieving this task. For instance, there are many different translations in which Jesus says things in different ways. Some of the Gospel accounts appear to offer conflicting versions of what has been said. They have been handed down through so many generations that it seems likely that something has been lost or changed along the way, especially when we consider that people may have altered things to fit their own interests. Now, this is not to say that they are not reliable. However, one’s awareness of these issues act as defeaters for the reliability of what has been said, and consequently, for the justification one has for one’s beliefs. One would need to have an impressive amount of evidence to defeat these defeaters – so much so that it seems unlikely that anyone could have it. If one does not, then one’s beliefs can be luckily true, as when one guesses the correct answer, but not justifiably true. I noted at the beginning of this section that I would not cover Wahlberg’s arguments in too much depth and the reason why is because they are really the province of historical-biblical studies, and it is not the role of the philosopher to appraise them. It is sufficient, for now, to state my scepticism of this project, whilst accepting that it is possible, though seemingly unlikely, for Wahlberg to demonstrate the reliability of the biblical accounts. Second, many of the aforementioned issues arise because of Wahlberg’s use of the deputised model of divine testimony. Using this model is open to objections concerning the overall aptness of deputised discourse for describing how people typically hear from God. One problem that emerged in §4.3.1 is that deputised discourse is not particularly personal, whereas people often feel that they hear from God personally. The model of faith I have been using involves a component in which one trusts God for what God says, but on Wahlberg’s account, one is more involved in trusting the Bible than God. This appears to miss the target for the way that many

199 For incisive criticism of this approach see Plantinga (2000, pp.268-280).

173 people actually hold their beliefs, and so even if people are justified on his model, people may not be believing that way, i.e., through Jesus and via a reliable record of the Gospel account, but by another way altogether. On a related, and third point, Wahlberg’s account might not overcome the PFR for many of the beliefs people hold. Although people will believe some things because they have been said by Jesus, they also believe many other things, either contained in other parts of scripture, or from experiences where they believe that God has spoken to them. It would be helpful to have a model of discourse that was able to account for all of the beliefs that people hold on alleged divine testimony, rather than just a select few. These three issues may not be decisive to show that Wahlberg’s account fails, but they are enough to make it an unappealing approach to NRDT, or indeed, a solution that is appropriate for philosophical critique. I propose that these reflections show that in order to make a plausible case for being justified in believing that God has told you some proposition, we need to utilise a different approach to divine discourse. As I will outline in the ensuing, the most useable kind of divine discourse, and the kind that best suits the experiences of many people today, is the appropriation model of discourse.

5.4.2 Appropriated Discourse: Plantinga

In this final section, I turn to the work of Alvin Plantinga (2000), who is perhaps the most well-known contemporary philosopher of religion. In his work Warranted Christian Belief,200 Plantinga has developed a thorough and extensive epistemology of Christian belief, and situated this within his own account of Christian faith. I have two aims in examining Plantinga’s account. First, to show that Plantinga’s theory can be interpreted as utilising the appropriation model of divine discourse. Second, to outline how, if his account endorses non-reductionism about divine testimony, it can overcome the problem of the absent speaker, and can give one justification for believing God as a speaker. Given these two claims, we have a theory that accounts for beliefs held on divine discourse that is consistent with my own theory of faith, and with the experiences of many people, and that can meet the conditions specified by the non-reductive model. Interpreting Plantinga in this way, then, looks like the best

200 Quotations from this section refer to his (2000). This work has also recently been shortened into a more manageable volume – Knowledge and Christian Belief (2015).

174 theoretical response to the PFR. Throughout this section I will give some reasons for thinking that interpreting Plantinga through the perspective of non-reductionism in testimony and the appropriation model of discourse is an improvement on Plantinga’s theory as it currently stands. Let’s begin by looking at what Plantinga sees as the source of Christian beliefs – divine testimony. His view of such testimony is generally confined to the Christian Bible. His account of this, which he largely just refers to as ‘scripture’, is as follows:

Scripture is most importantly a message, a communication from God to humankind; Scripture is a word from the Lord. But then this just is a special case of the pervasive process of testimony, by which, as a matter of fact, we learn most of what we know. From this point of view, Scripture is as much a matter of testimony as is a letter you receive from a friend. What is proposed for our belief in Scripture, therefore, just is testimony — divine testimony. So the term ‘testimony’ is appropriate here. (p.251)

It might seem at first that here Plantinga views divine discourse in accord with the deputised or authorial models. After all, he appears to be favouring a narrow view of divine discourse according to which God only speaks to people through the Bible, and that what is said in the Bible is like a letter written to all people – written just once and not addressed to any particular individual. He appears to confirm this view in other places as well:

So Scripture is, indeed, testimony, even if it is testimony of a very special kind. First, the principal testifier is God. It also differs from ordinary testimony in that in this case, unlike most others, there is both a principal testifier and subordinate testifiers: the human authors. (p.252)

This is precisely what both the authorised and deputised models of discourse are committed to. That God, in the former sense, dictates to people precisely what to say, or in the latter, that God is the speaker but deputises people to speak on his behalf. However, Plantinga’s account is reinforced by a second component, which suggests that he favours neither of these accounts. He adds the following:

There is still another difference: it is the instigation of the Holy Spirit, on this model, that gets us to see and believe that the propositions proposed for our beliefs in Scripture really are a word from the Lord. This case also differs from the usual run of testimony, then, in that the Holy Spirit not only writes

175

the letter (appropriately inspires the human authors) but also does something special to enable you to believe and appropriate its contents. So this testimony is not the usual run of testimony; it is testimony nonetheless. (p.252)

What Plantinga is saying is that God is directly involved at the time that the person who is reading scripture comes to believe what she does concerning the contents of scripture. The ‘part’ of God, according to Christian theology, which is involved in this process is the Holy Spirit.201 The work done here by the Holy Spirit is twofold. He causes belief in people, and causes people to believe that scripture is God’s own testimony. In this latter sense, God is involved in appropriating scripture. Someone reads it, and God gets her to believe, at that time, there and then, that God is speaking to her by what she is reading.202 In so doing, God is agreeing with, or ‘seconding’ the scriptures being read. It is by way of the Holy Spirit’s work in the process of people believing God’s testimony that I conclude that Plantinga can be seen as endorsing a version of the appropriation model. If this is how Plantinga thinks God speaks to people, just what do people come to believe when hearing from God in this way? Plantinga’s usual claim here is that people come to believe ‘the great things of the gospel’, by which he means certain claims including Jesus is the son of God, Jesus came to bring salvation, etc. Usefully for my purposes, Plantinga then ties his account of this process of believing divine testimony to his own theory of faith: ‘faith is belief in the great things of the gospel that results from the internal instigation of the Holy Spirit’ (p.252).203 So, Plantinga can be seen as yet another author who endorses the view that faith, in some Abrahamic traditions, requires believing divine testimony. Although Plantinga’s account of divine testimony appears to utilise the appropriation theory, it is much more restrictive than the theory I described in Chapter

201 For philosophical reflection on the witness of the Holy Spirit see Dougherty (2014) and Mavrodes (1988, p.150). Plantinga himself is influenced in large part by the theology of John Calvin (1960). 202 We saw in §4.3.1 that we need an explanation for what makes the Bible the sort of text that God tends to appropriate. This is because the authors of it were either inspired in the sense of deputised or authorial discourse, or simply by their own experiences. When God appropriates the text, though, it doesn’t matter which of these is correct: the divine testimony comes to the person at the time of reading, rather than at the time of it being written. 203 This is merely necessary, but not sufficient for Plantinga’s theory of faith. In Chapter 9 of his (2000), he goes on to say that faith also involves trust, commitment and an array of affections like love, awe and wonder.

176

4. This is because he limits the events that are appropriated to instances in which someone takes herself to have had the main claims of the Christian narrative asserted to her:

We read Scripture, or something presenting scriptural teaching, or hear the gospel preached, or are told of it by parents, or encounter a scriptural teaching as the conclusion of an argument (or conceivably even as an object of ridicule), or in some other way encounter a proclamation of the Word. What is said simply seems right; it seems compelling; one finds oneself saying, “Yes, that's right, that's the truth of the matter; this is indeed the word of the Lord.” (p.250)

So, although he broadens the scope beyond merely a person’s individual reading of scripture, he narrows it to the main claims made as part of the Christian narrative of salvation. And yet even the scope of when an event is appropriated is narrower than I had previously described. I suggested that God is sometimes experienced as speaking through certain life events, the advice of a friend, or through watching a movie or listening to music. This certainly would suggest that what is said is also far too narrow given appropriation, and in particular, people’s experiences of God speaking to them. People often believe that God is involved in advising them on the small details of their lives – ‘take that job’, ‘marry that person’ – as well as presenting himself as loving and caring toward people. I suggest, therefore, that Plantinga’s account can be improved upon if we take a broader scope to both the issues of the kinds of things God appropriates, and what God says when he appropriates these things. God can appropriate many more things than the Bible and can say many more propositions than the gospel narrative. In the ensuing I will assume that this broader approach is correct. As we will see, taking a broader approach to divine speech won’t alter the epistemic component to Plantinga’s overall theory. What Plantinga does offer us, though, that needs to be retained, is a theory of divine testimony that is fit for use as part of a non-reductive approach to justifying religious beliefs. The function of the Holy Spirit is integral to this, and will be the means of addressing the problem of the absent speaker. From what has been said of Plantinga’s account, the theory states that a person H takes herself to hear from God that some proposition p is the case, and does so through some medium m, for instance,

177 reading the Bible, hearing a sermon, or (if we are happy to extend the mediums available) through an experience or conversation with a friend. So, H takes the proposition to be the case, and to come from God. There are two beliefs in play here – (1) belief in the proposition p testified to, and (2) belief that God testified that p. Plantinga explicitly says something very similar:

[U]pon reading or hearing a given teaching—a given item from the great things of the gospel—the Holy Spirit teaches us, causes us to believe that that teaching is both true and comes from God. So the structure here is not: what is taught in Scripture is true; this (e.g., that in Christ, God was reconciling the world to himself) is taught in Scripture; therefore, this is true. It is rather that, on reading or hearing a certain teaching t, one forms the belief that t, that very teaching, is true and from God.’ (p.260, emphasis added)

This is precisely the way that ordinary testimony is believed as well. When we hear some proposition p told to us by some speaker S, we believe both that p and that S told us p. In this instance of divine testimony, we believe both that p and that God told us p. The former instance of testimony in ordinary contexts is taken by non-reductionists to be a basic belief, and so, it seems that the latter has at least the potential to be basic too, providing some standard of non-reductionism can be met for divine testimony. Wouldn’t that, however, be to ignore the outstanding issue of the problem of the absent speaker that faces divine testimony but not ordinary testimony? On what basis does Plantinga think that we can get around this problem, and hence treat beliefs formed on divine testimony in a basic way? The answer to this question has three components. Let’s take them each at a time. The first component is, as we have already seen, that God is directly involved in the process of bringing about both the belief in the proposition testified to, and the belief that he is testifying. In standard cases of testimony, the latter proposition is simply something you are aware of through sensory perception. You hear a voice or see a person and believe that someone is speaking. But suppose there is no speaker, and the ‘hearer’ is simply hallucinating the speaker. It would follow that the hearer is not justified in believing either that there is a testifier, or the proposition testified to. So, it seems as though there are necessary conditions in testimony for the existence of an actual speaker. These are never stated, though, in the literature on testimony for the reason that the presence of the speaker is simply presumed. This is true of the accounts

178 of reductionism and non-reductionism given above. They say that ‘When a speaker S tells a hearer H that p…’ not ‘When a hear H believes herself to hear a speaker tell her p…’. Only in the latter case do we require a necessary condition on the actual existence of the speaker since it is not presupposed. In the former case, though, it is. One way to address the problem of the absent speaker is, then, to put it in terms of the second option, and state necessary conditions on the speaker. Taken together with the foregoing discussion of Plantinga’s account, we can state the conditions on justification as follows: “Hearer H is justified in believing that God has told her p via medium m if and only if…”. On the supposition that whatever will be on the right- hand side of this biconditional can be met, it would logically follow that H is justified in believing that God has told her p. Now, as I mentioned above, Plantinga’s account works on the basis that the Holy Spirit has in fact testified to a hearer, making it the case that she sees that the testimony comes from God. For his theory, then, the hearer is not having an illusory experience – there is a speaker who is perceived to be speaking, and in fact, the speaker that is perceived is actually speaking. If that is the case, then the hearer is partially justified in her belief. The speaker condition is met only if there is a speaker, which there is in the form of the Holy Spirit, so the speaker condition is met. So, for H to be justified in believing that God has testified to her p, God must have actually testified to her p. And not only that, but through the medium in which H takes herself to have been testified to, either, say, through scripture, sermon, experience, etc. The first component of a hearer’s justification is, therefore, a conditional clause that depends on the existence of God for its satisfaction. This may seem slightly disappointing for the religious believer since she would like the higher-order knowledge that her beliefs are justified, not that if God exists and has actually spoken to her, only then are they justified. Still, this distinction has some rhetorical force as part of Plantinga’s overall strategy. His work is aimed at resolving what he calls the de jure challenge to Christian belief, which he defines as ‘the claim that such belief is irrational or unreasonable or unjustified or in some other way properly subject to invidious epistemic criticism’ (p.167). This he contrasts with the de facto challenge to belief in God which is simply that ‘the belief in question is false’ (p.167). The objections that motivate the PFR can be seen as de jure challenges, and hence, responses to it as responses to the de jure challenge. However, if religious belief is

179 justified partly on the condition that God exists, then it seems to follow that, as Plantinga says,

…There isn't a sensible de jure question or criticism that is independent of the de facto question. There aren't any de jure criticisms that are sensible when conjoined with the truth of theistic belief. (p.191)

In other words, objections to the rationality or justification of religious belief are not independent of objections to the truth of those beliefs. If God exists, then one’s religious beliefs are rational and justified. This follows from Plantinga’s view of God from the perspective of Christian theology as the kind of God who speaks to people. (I will give this claim more attention shortly). I have only discussed this claim so far in the context of the condition of God actually speaking to people. Plantinga actually derives it from elsewhere. For him, it follows from his views of the cognitive constitution of human beings, and to this issue I now turn to address the second component of how Plantinga’s account can be interpreted as a plausible approach to NRDT. In doing so, we will come back to the issue of the de jure and de facto challenges. Plantinga’s account of the cognitive mechanisms that humans have is influenced largely by his reading of St. Thomas Aquinas (1947) and the theologian John Calvin (1960). He takes them both, and Calvin in particular, to claim that humans have ‘a kind of faculty or a cognitive mechanism, what Calvin calls a sensus divinitatis or sense of divinity, which in a wide variety of circumstances produces in us beliefs about God’ (p.172). He is assuming that this so-called ‘sensus divinitatis’ is part of our natural cognitive architecture, and its purpose is to produce in us beliefs about God when in certain circumstances. He adds:

These circumstances, we might say, trigger the disposition to form the beliefs in question…Under these circumstances, we develop or form theistic beliefs—or, rather, these beliefs are formed in us; in the typical case we don't consciously choose to have those beliefs. Instead, we find ourselves with them, just as we find ourselves with perceptual and memory beliefs. (pp. 172- 73)

The kinds of circumstances that Plantinga takes to be appropriate for this cognitive mechanism are, for instance, where someone sees ‘the marvelous, impressive beauty of the night sky; the timeless crash and roar of the surf that resonates deep within us;

180 the majestic grandeur of the mountains’ (p.174). In these circumstances, he claims, the sensus divinitatis is disposed to bring about in the individual beliefs concerning God, such as that God exists, that God created beautiful things, that God is immensely powerful, etc. Plantinga draws analogy between beliefs formed by the sensus divinitatis with perceptually formed beliefs. He claims that the former beliefs are basic since, as we saw in Chapter 1, this is the case with the latter. He says, for instance:

It isn't that one beholds the night sky, notes that it is grand, and concludes that there must be such a person as God…It is rather that, upon the perception of the night sky or the mountain vista or the tiny flower, these beliefs just arise within us. They are occasioned by the circumstances; they are not conclusions from them. The heavens declare the glory of God and the skies proclaim the work of his hands: but not by way of serving as premises for an argument. (p.175)

So, on this theory, beliefs formed by the sensus divinitatis are not inferred from other beliefs or premises, but are basic. This is the case because the sensus divinitatis produces beliefs in us in the same kind of way that beliefs are produced in us by our other perceptual faculties like sight, hearing and touch. Now, the question that most naturally follows from this is are they justified? These beliefs might be basic, but are they properly basic? Plantinga’s entire project rests on demonstrating that beliefs formed in this way are in fact justified. But how can he do this? The method he uses is by appeal to the externalist theory of ‘proper functionalism’ that is connected to several theories that we briefly looked at in §1.3.5. According to this theory, a basic belief is justified, roughly speaking, providing it is formed by cognitive faculties functioning properly in an environment that is truth- conducive. We can see how this might be suitable for other basic sources of belief, such as visual perception. Providing our visual faculties are functioning properly, i.e. they are not significantly damaged or under the influence of intoxicating substances, and providing we are in an environment that is conducive to forming true beliefs, i.e. there is plenty of good light and we aren’t wearing tinted spectacles, etc., then it would seem that what we come to believe through these faculties is justified. The visual faculties are properly functioning in a reliable way, and so will give us, as an output, reliably formed beliefs.

181

The proper functionalist theory of justification is fairly straightforward to transpose into the context of one’s religious beliefs on Plantinga’s theory. For Plantinga, for basic beliefs to be justified,204 they need to be ‘produced by cognitive faculties functioning properly in a congenial epistemic environment according to a design plan successfully aimed at truth’ (p.178). But, as we have seen, he has already proposed a cognitive faculty that works in a way that is similar to our other perceptual faculties, namely, the sensus divinitatis. Plantinga says that according to his model,

…our cognitive faculties have been designed and created by God; the design plan, therefore, is a design plan in the literal and paradigmatic sense. It is a blueprint or plan for our ways of functioning, and it has been developed and instituted by a conscious, intelligent agent. The purpose of the sensus divinitatis is to enable us to have true beliefs about God; when it functions properly, it ordinarily does produce true beliefs about God. These beliefs therefore meet the conditions for warrant. (p.179)

So, where we want to look to the proper functioning of our cognitive faculties to justify our basic beliefs, the sensus divinitatis provides us with just this faculty that can justify those beliefs concerning God that are formed in a basic and immediate way. Notice, though, that this special faculty, which is part of our proper functioning, only produces reliable, true beliefs because it has been designed in us by God. It is God who has endowed us with this cognitive faculty, and hence, not merely its reliability, but even its very existence depends upon the prior existence of God to design and endow people with it. And here we return to the distinction between de jure and de facto challenges to Christian belief. If there is no God, then there is no sensus divinitatis, and so beliefs concerning God will indeed be unjustified and irrational. But if God exists, then perhaps he has endowed us with this faculty, at which point such beliefs may yet be justified and rational. An answer to the de jure challenge depends upon the answer to the de facto challenge. He argues that

204 He uses the technical term ‘warrant’ to define this, and takes this to differ from ‘justification’, which he argues is a deontological notion. He claims that when someone’s basic belief is justified, ‘he is within his epistemic rights, is not irresponsible, is violating no epistemic or other duties in holding that belief in that way’ (p.178). I am conflating the two notions for parsimony, and because if one is warranted, in Plantinga’s technical sense, then one is justified, in the sense that I have been using the term, and presumably Plantinga’s sense too. Warrant, therefore, entails justification.

182

…if theistic belief is true, then it seems likely that it does have warrant. If it is true, then there is, indeed, such a person as God, a person who has created us in his image …who loves us, who desires that we know and love him, and who is such that it is our end and good to know and love him. But if these things are so, then he would of course intend that we be able to be aware of his presence and to know something about him. (p.188)

It is the endowing of us with the sensus divinitatis that Plantinga assumes would be the partial means by which God would have us know things about him. A fairly immediate objection that seems to face this account is that not all people, when they have the sorts of experiences that Plantinga is referring to, find themselves forming beliefs concerning God. Not everyone perceiving a sunset would form the belief that God is supremely powerful. However, most people perceiving a sunset would form the belief that there is a sunset, and those who don’t would be likely suffering from some sort of cognitive malfunction. So surely Plantinga is not licenced to draw an analogy between the sensus divinitatis and our other perceptual faculties. However, Plantinga’s strategy to overcome this issue is simply to accept that people are suffering from some kind of cognitive malfunction. The response he gives, which is taken directly from the Christian salvation story, is that people have fallen into sin, and that this has had damaging consequences, amongst which include damage to the sensus divinitatis:

the sensus divinitatis has been damaged and deformed; because of the fall, we no longer know God in the same natural and un‐problematic way in which we know each other and the world around us. Still further, sin induces in us a resistance to the deliverances of the sensus divinitatis…we don't want to pay attention to its deliverances. (p.205)

Because of the cognitive damage caused by sin, people are unable to form beliefs concerning God in any reliable and systematic way, as we are with perception. Some people form belief concerning God, but not everyone. And this claim, again, is dependent not only on the de facto question of whether God exists being answered in the affirmative, but also the truth of the basic Christian narrative.

183

Must we posit the existence of the sensus divinitatis for Plantinga’s account to work? Some authors have suggested that we do not.205 All we need is that we are creatures who have been created such that under the right conditions we would form beliefs about God. But these could arise through mechanisms that are constituted by a number of our natural cognitive faculties working together. Something like this could be derived from recent work in the cognitive science of religion.206 So there could be alternative ways of meeting the proper functionalist condition without the need to posit the sensus divinitatis. Where does this discussion leave us? Well, we now have a basic sketch of what a plausible account of properly basic belief in God could look like. This can be augmented to account for properly basic belief in divine testimony. For, suppose that God has designed us in such a way that when we are in circumstances in which God actually speaks to us and tells us p we form the belief that God has spoken to us and told us p. It would follow, if proper functionalism is correct, that we would be justified in holding these beliefs. And not only that, but holding them as both basic and justified. We now have two necessary conditions for the non-reductive justification of beliefs formed on divine testimony:

Non-Reductive Divine Testimony: H is justified in believing that God has testified to her p if and only if: (i) God has actually testified to H that p; and (ii) God has designed H in such a way that when H is in circumstances in which God actually speaks to her and tells her p, then H would form the belief that God has spoken to her and told her p.

The account relies on the de jure/de facto distinction in the sense that H is only justified on the condition that God actually exists and that God has created her in a certain way. Effectively, to demonstrate that H’s beliefs would be unjustified would be tantamount to demonstrating that the God we have been describing does not exist. Moreover, the problem of the absent speaker has been evaded. The claim we are justifying is that ‘God has testified to H that p’, not merely ‘when God testifies to H that p’. In the latter case we were seeking to demonstrate whether we are justified in

205 See Moreland and Craig (2003, p.167) and Moon (2016, p.884). 206 See Lyons (2009), and Clarke and Barrett (2010, 2011).

184 believing p. In the former, we seek to justify both the testimony p, and the belief that God is the testifier. Since, on Plantinga’s account of divine testimony, we don’t merely believe p but that God is the speaker of p, if we are justified in believing the testimony then we are also justified in believing that the testimony comes from God, the latter belief being secured under both (i) and (ii). The theory is almost complete, but at present it seems open to the following sorts of consideration. Suppose someone, hearing a sermon, believes that God speaks to her through the message telling her that he is all loving and that she comes to believe this proposition. So, she believes that God is all loving, and that has God spoken to her to tell her this. Now assume that (i) and (ii) are actually true – God actually has spoken, and has actually designed her such that when he speaks, she will form these beliefs. Add to this, though, that she can be justified in believing that God, is in fact, not all loving. Her reason for this could quite fairly be the evidence she sees of the pain and suffering in the world.207 She takes this to count against her belief that is coming from God’s own testimony. What we have here is a simple case of a doxastic defeater for the testimony that is coming from God, and not only that, but a justified doxastic defeater. Such a consideration hardly seems irrelevant to the justification of the hearer’s belief. If this is so then there is a problem for the de jure/de facto argument that Plantinga relies upon. It seems as though God could exist, and conditions (i) & (ii) could be met, but without the hearer being justified in believing what has been testified to. What we are doing here is assuming that the non-reductionist condition plays a role in the justification of the hearer’s beliefs in addition to the two conditions already supplied. For NRDT, then, H is justified in believing that God has testified to her p if and only if, (i), (ii) and

(iii) H has no undefeated justified doxastic defeaters that are sufficient to count against p or against the credibility of God as a speaker. (non-reductionism)

It is, then, the full set of (i)-(iii) that state the necessary and jointly sufficient conditions for non-reductive divine testimony on this augmented version of Plantinga’s account.

5.4.3 Defeaters and Epistemic Authority

207 This could be viewed as a kind of evidential argument from evil (see Adams 1999).

185

There is a further advantage to endorsing (iii) as a necessary condition on justification that I believe to be an improvement on Plantinga’s theory. For, if the only specified conditions were (i) and (ii), then the hearer in the circumstances of believing herself to have heard from God really has no second-order knowledge over whether or not her beliefs are justified. This is due to the externalist theory of justification that Plantinga is working with. To see this more clearly, consider a contrast with an internalist theory. On that theory, that which justifies belief in the proposition in question must be accessible to the believer. But if one can access what justifies the belief, say, for instance, the evidence available to the individual, then one can reflect on one’s justification for this first-order belief and hence form a second-order belief about one’s justification: one can believe that one is justified in believing. This can bring a degree of confidence and assurance to the individual; she can believe with confidence or even know that her beliefs are on a firm foundation.208 In contrast, if (i) and (ii) are the only conditions specified for H to satisfy, it looks like she can only have second-order knowledge that she is justified providing she can first have knowledge that God exists. Acquiring such knowledge is, of course, both problematic and controversial. To do so she would need to appeal to some form of natural theology – to religious experience, arguments for God’s existence, etc. How are things different when (iii) is included? Well, H still incurs the problems of lacking certainty that follow from the foregoing discussion, but she is able to take some confidence from that which she already has awareness of, namely, if she lacks defeaters that are sufficient to count against the credibility of the speaker or against the proposition. This at least she can take some confidence from. This particular benefit that comes from the inclusion of (iii) has further implications too when we reflect on the sorts of things that can be defeaters. Typically, defeaters against speakers or their testimony are usually of the epistemic kind. And yet it seems as though there should be other sorts of relevant defeaters too. In the religious case for instance, moral defeaters also appear relevant. Consider, for example, someone who believes that God has told her to commit atrocities in his name. If only (i) and (ii) are conditions on justification, this person is unable to know whether or not

208 Alston (1988) argues that this feature of internalist justification is what makes it so attractive. People don’t only benefit from knowing what justifies their own beliefs, but can tell others what justifies them too. Of course this doesn’t make internalism true, but does make it appealing.

186 she is justified in believing that God has in fact told her this. However, if (iii) is included, there is a means of showing such beliefs to be unjustified. Recall in §1.3.6 that we saw that moral and pragmatic reasons for belief are also pro-tanto considerations that bear on what we should believe. With the atrocities example, it looks like moral considerations should weigh in and undermine one’s all things considered reason for holding the belief in the alleged divine testimony. Without this condition, as is the case with Plantinga’s own account (rather than my modified version), it is very challenging to defeat the testimony – one would need to demonstrate the non-existence of God. With condition (iii), though, we have a plausible way to defeat it by allowing that moral reasons for belief defeat belief in the proposition that one should commit atrocities in God’s name. There is a final problem to consider before summarising the account. Note that the first example above, pointing to the pain and suffering in the world, appears to have left us with a problem since it looks as though H fails to satisfy condition (iii): H does have an undefeated justified doxastic defeater that counts against p, namely, the pain and suffering in the world. So, it would seem that H is not justified in her religious beliefs held on divine testimony. There is, though, a fairly plausible way out of this situation, and other situations like it. It comes from appreciating that the testifier in question has a degree of authority on a given proposition that would grant his testimony a greater evidential weighting compared with other testifiers. We can recall from an earlier quote by John Lamont that we commonly take God’s knowledge and honesty to be necessary. If this is true, then it seems that propositions testified to by God should outweigh any other considerations against the testimony. One way to understand this claim is in terms of epistemic authority. The work of Linda Zagzebski is pertinent to this topic,209 who has set out conditions on what it is to defer to an authority on belief. Drawing on the political theorist Joseph Raz (1978), she considers cases where having a reason to  that comes from an authoritative source is a reason that replaces other reasons that count in favour of or against -ing. For example, suppose you are driving home and you stop at a red light. You do so because the law, which is one kind of authority, requires you to stop. You may have other reasons to keep going, say, because you want to get home quickly, but

209 See Zagzebski (2012) and Keren (2014).

187 the authority of the law gives you a reason to stop at the light in a way that replaces those other reasons you may have. An epistemic example of authority is when we regard an individual as an authority on a matter to the extent that what the individual believes on this matter gives us a reason to believe the same thing, regardless of conflicting reasons. For instance, it might seem against common sense to take seriously the possibility of multiverse theory, but the very fact that Stephen Hawking is willing to entertain it would give many people a reason to believe that it’s true. This sort of example is used to support what Zagzebski, following Raz, terms ‘the preemption thesis for epistemic authority: The fact that the authority has a belief p is a reason for me to believe p that replaces my other reasons relevant to believing p and is not simply added to them’ (p.107). Now, in examples in which one has what appears to be defeaters for the proposition testified to by God, according to the preemption thesis, since God is the speaker, one is still justified in believing the testimony. Although this looks like an appealing solution, as it currently stands the preemption thesis looks a little too strong. Recall the example of the individual who believes God has commanded him to perform atrocities. We don’t want the person to simply appeal to the authority of God as a reason for holding this belief. It should be the case that other considerations come into play too, namely, moral considerations. So, perhaps saying that authority is a reason that replaces other reasons is too strong. Authority might count as having greater weight, but not an overriding weight. To put it another way, authority is not an ultimate reason for believing the authority’s testimony, but a pro-tanto reason whose weight is relatively larger than other pro- tanto reasons a person has, and this weight is taken into consideration concerning the most rational course of action, all things considered. The kind of epistemic authority coming from experts is quite common. My friend is an expert on what he had for lunch, and is thus an epistemic authority on this matter. His testimony on what he had for lunch carries greater weight than someone else’s who wasn’t there. Of course, epistemic reasons aggregate, such that if my friend told me he had a sandwich in the cafeteria, but three friends I trust disagree and say they saw him having a burger at a fast-food outlet, the weight of their testimony could outweigh my friend’s self-report. This is the kind of authoritative reason I have in mind for divine testimony. God’s testimony carries a relatively greater weight on certain matters, but can still be outweighed by conflicting reasons. 188

There is one clear benefit to bringing the notion of authoritative reasons into the account of divine testimony – an explanatory benefit. For consider, as we saw in Chapter 1, that faith is often viewed as irrational, or irresponsive to evidence. For example, pain and suffering in the world, religious disagreement or a lack of positive evidence in favour of God’s existence function to make faith in God unjustified. Moreover, paradoxes that arise when considering theological doctrines like the incarnation, trinity, omnipotence, and divine foreknowledge should cause people to reconsider their faith. In many instances though, this is not the case. People are often aware of these issues but retain their faith anyway. How do we explain this? Is it simply the result of wishful thinking that people do not regularly relinquish their faith against this tide of counter-evidence? Perhaps people feel as though the evidence is equally weighted. One option to explain all of this is to appeal to authority – people see God’s testimony as justifying them in holding on to their faith in spite of reasons to relinquish it. If that is so, then the authority and the preemption thesis performs a neat ‘rounding- off’ task for the arguments I have made in this thesis. For, I have argued that faith in divine testimony can be one way of manifesting faith in God. But faith in divine testimony is justified partly on the basis of the preemption thesis, which itself explains why faith is resistant to conflicting reasons. So, epistemic issues that surround one’s faith in God are neatly explained by appeal to faith in divine testimony, thanks to the preemption thesis. It looks like believing divine testimony is connected to faith in a nice explanatory way after all. This concludes my discussion of justifying faith for appropriated discourse. I will briefly recap on some of the benefits we can bring to Plantinga’s theory by augmenting it as I have done. First, by interpreting him as adopting a theory of appropriated discourse, we can neatly transition from discussing divine speech purely in terms of biblical, or broadly scriptural revelation, to divine speech through many further means. This is consistent with the experiences people have today, and the nature of faith as interpersonal. Second, by taking his account of proper function that is delivered by the sensus divinitatis to apply in the context of divine testimony, rather than simply through everyday experiences. Third, by omitting an ontological commitment to the sensus divinitatis by requiring that we are simply designed to form the belief under the right conditions. Fourth, by introducing the requirements set by non-reductionism in testimony, thereby having the means to overcome various 189 counterexamples that could come against the theory. Moreover, and fifth, doing so allows us to introduce the notion of epistemic authority that not only looks like it’s of clear and obvious relevance to the issue of divine testimony, but is also connected to religious faith in a helpful, explanatory way.

5.5 Conclusion

We have now explored the account of faith I argued for in Chapter 2 in light of recent developments in the epistemology of testimony. Faith in divine testimony faces a problem that trust in testimony in non-religious contexts does not face, namely, that the speaker is not obviously present in these contexts in the way that the speaker is in cases of human testimony. One way to overcome this is to find ways of justifying the belief that God has spoken before justifying belief in what God has said. However, this looks like a straightforward case of natural theology, and so is less interesting for my purposes in this thesis. One can overcome this problem by augmenting the work of Alvin Plantinga, in particular by drawing on the distinction between de jure objections to rational belief in God and de facto objections to the existence of God. This allows us to state the conditions for being justified in believing testimony in a conditional way that evades the absent speaker problem. By adding the non- reductionism conditions from general testimony to the particular account developed from Plantinga’s work, we also have a means of evading several counterexamples that come against his theory as it currently stands. I believe that the final theory, developed from Plantinga’s work, is an improvement on his overall account, and the most viable option for addressing the general problem of faith and reason on the account of faith I have defended in this thesis. The conditions stated in §5.4.3 for a non-reductive divine testimony show how one’s faith can be epistemically justified, hence rejecting premise (1) of the PFR from Chapter 1.

190

6. CONCLUSION

6.1 Overview

In this thesis I have defended numerous theories, including giving accounts of the nature of trust, epistemic justification, reasons for belief, assertion, and the nature of philosophical analysis itself, among others. However, an account of each of these concepts was either adopted or defended for the primary purpose of exploring the nature of faith, and the epistemological implications of my own account so as to offer a response to the perennial problem of faith and reason (PFR). To this end, I have defended three central claims: (1) Abrahamic religious faith requires trusting divine testimony; (2) Faith in divine testimony can be epistemically justified; (3) Non- doxastic theories of faith fail as accounts of faith, and so responding to the problem of faith and reason by using these theories is not available. In this conclusion, I will briefly reiterate what is most salient to these three claims, taking them one at a time respectively, and will propose ways in which research on these issues can be taken further.

6.2 Faith in Testimony

I argued in Chapter 2 that one of two conditions required for Abrahamic faith is trust in God. This trust can be manifest in many different ways, but one that I believe to be widespread, and essential to the trusting response required by Abrahamic faith is trusting what God says – trusting God’s testimony. In fact, I argued that it makes little sense to trust God in any way without first trusting something God has said. Testimonial trust in what God says is therefore more basic than trust in God in general, and not merely an instance of such trust. It is on the basis of this conclusion that I motivated further discussion of the epistemological implications of such faith, particularly given the resources available from the epistemology of testimony. Surely though, one might respond, one can have faith without trusting God’s testimony. For instance, I can clearly have faith that God exists, without believing such faith to have come from discoursing with God himself. I agree entirely with this response. However, faith of this kind, as a propositional attitude, is not sufficient for faith in God required by Abrahamic faith. There are other conditions required too. Importantly, one must entrust God with something valuable. But, when one does this,

191 the trust one shows implies that God has committed himself in some way to performing some future action, and our beliefs concerning this future action come by way of trusting what God has said, either implicitly or explicitly. Although I have tried to defend the account of Abrahamic faith with my own theory of trust, and with the support of numerous sources who have said something similar, the claim that faith requires trust in God’s testimony is not essential to motivating the epistemology that my thesis subsequently explores. All that is needed is the less strenuous claim that trust in God’s testimony is sufficient for the trust required by Abrahamic faith. Since there has been very little work on how this claim affects the epistemology of faith from the perspective of the reductionism/non- reductionism debate in epistemology, it is still worth exploring. What is perhaps more controversial, and what I have explored only to a limited extent, is whether divine speech is analogous to human speech. For the sake of adopting a particular epistemological framework, I have assumed that a comparison can be drawn between the two. Here is another place where someone might want to push back against my theory. It could be objected that there is no comparison to be drawn, and hence we are not warranted in using the same epistemological framework. To see this, consider that when I claim that ‘John said that p’, I would most likely mean, quite literally, that John said that p, as in, John spoke the words, and in so doing, said p.210 However, when I claim that God said that p, is there any literal sense in which I can really mean this? Can God literally speak the words, and thus say p? What people typically mean is that God communicated p in some figurative sense, using the concept of speaking as a metaphor to aid our understanding. God’s act of communication, however, might be very different from human speech. For example, a feeling that God has affirmed one’s actions; an internal sense that God has communicated a proposition; a reading from some scripture and the impression that this was appropriated by God; a perceiving of some event and believing God to be asserting some moral command by it. In some of these cases, the alleged divine speech is done indirectly by some other action or event – what Wolterstorff calls double agency

210 I might also mean that John wrote p in a letter, and this might not be so problematic when attributed to God. But since I would often mean John spoke p, the problem of drawing an analogy between the two communicators (people and God) persists for these common cases.

192 discourse. In others there is a vague impression or internal sensation that one experiences. This is closer to the deliverances of Plantinga’s sensus divinitatis. Here is the problem that these observations leave open: general theories of testimony work on the model of human speaking in quite literal terms. Testimony in religion, when God is the speaker, works with a model of speaking in metaphorical terms, and usually in many different ways. Since the epistemology supervenes on a theory of the nature of testimony, then presumably the epistemology changes too if the nature of the speech changes. I have assumed that we can talk about divine speech in the same way that we can for human speech, but perhaps we cannot, and if we cannot then perhaps the epistemology cannot be co-opted for the religious context. I believe further philosophical attention needs to be given to what it is for God to speak, and how this affects the epistemological framework for religious beliefs grounded in this source.

6.3 Justified Religious Faith

The account of faith I defended in which Abrahamic faith requires trust in divine testimony was explored from two perspectives: first, from the point of view of people as hearers in a testimonial exchange, and second, from the point of view of God as the speaker in a testimonial exchange. Whilst Chapter 2 dealt with the first perspective, Chapter 4 addressed what it is for God to speak, and I have just mentioned one way in which this research could be developed further to address some outstanding questions left by my account. It was the task of Chapter 5 to place my account of faith within the current framework of the epistemology of testimony to see what the results would be. How did the account fare, and does it help to overcome the problem of faith and reason? I found examples of non-reductionism in recent and older literature, and also proposed another way – modified reductionism – which I drew from the theory of John Locke. I eventually found that the most promising approach was to see whether my account of faith can be justified according to non-reductionism, and found that we can modify the popular and influential religious epistemology of Alvin Plantinga so as to regard him as giving a non-reductionist account. His recent work can be regarded as endorsing a particular view of divine speech known as appropriated discourse. In addition to this, Plantinga is known for his defence of the view that religious belief

193 can be properly basic. With these two claims, it is possible to use Plantinga’s account to explore a non-reductive approach to divine discourse, and hence a non-reductive approach to justifying the account of faith I have defended. The main problem facing a non-reductive account in a theistic context is the apparent absence of a speaker. How can we be justified in believing God’s speech non-reductively if we must first be justified in believing that God has spoken? (This problem is clearly related to that of the previous section where our speech is direct and we mean that we speak literally, whereas God’s speech appears less direct, and we mean, metaphorically, that God speaks). By using Plantinga’s distinction between the de jure and the de facto objections to theistic belief, it appears that we can evade this problem by endorsing a conditional view of religious knowledge: we are justified providing God has actually spoken and designed us in a certain way. This, from what I can tell, looks to be the most promising way to have a non-reductionist approach to justified faith in divine testimony. It comes at certain costs though, which are not uncommon to externalist accounts of epistemology in general. The main issue is that one cannot have second-order knowledge of whether or not one is justified in what one believes. On this account, one might never really know whether one’s particular religious beliefs are justified or not, but this is perhaps part of what it really means to have faith in God. This route to granting positive epistemic appraisal to religious beliefs is well- trodden territory in the sense that most of the work in religious epistemology has been devoted to deontological concepts like justification, warrant and rationality. My thesis has been written in this same tradition. However, the account of faith I have given could be interpreted according to concepts in virtue epistemology. One option is the following: supposing that faith in God requires trust in God’s testimony, or at least that this trust is sufficient for Abrahamic faith. Now, imagine someone who believes that God is all loving, and does so on the basis that she believes God to have told her that he is all loving. This person observes the incalculable weight of evil in the world, acknowledges that this is evidence against the proposition that there is an all loving God, but continues to believe that God is all loving anyway. I suggested at the end of Chapter 5 that this might be a reasonable response if this person is treating God as a testifier whose testimony comes with a significant degree of authority. It’s also imaginable that this person accepts the testimony because she acknowledges her own intellectual limitations. Her understanding of God is that he is omniscient, whilst she 194 is not, and so she ought to put aside her intellectual doubts and trust him because he knows what he is talking about. The trusting response I am describing here appears to be a result of the intellectual humility of the agent. At least two current accounts in the philosophical literature associate intellectual humility with acknowledging one’s intellectual limitations, and deploying them in one’s practical and theoretical reasoning.211 In fact, the account by Whitcomb et al. (2017) says explicitly that intellectual humility ‘increases a person’s propensity to defer to others who don’t have her intellectual limitations’ (p.522). It seems to me that the character I described – a fairly common example of religious faith, in my view – is exhibiting intellectual humility with her deference to God’s testimony, and this is the case whether or not she has second-order knowledge of whether or not she is justified in believing that God has spoken to her. Evaluating faith in terms of epistemic virtue is distinct from evaluating it in terms of justification. This provides at least one way of seeing faith as the deliverance of epistemic virtue, and falls directly out of the account of faith I have defended. I believe that more research should be dedicated to appraising faith in terms of epistemic virtue, rather than solely in deontological terms, and my proposal here marks one means of doing so.

6.4 Faith without Belief

In Chapter 3 I discussed and evaluated numerous recent theories of propositional faith that are each committed to the claim that faith that p does not require belief that p. I group together these accounts under the title non-doxastic theories of faith, or NDT. It is evident from the widespread support for NDT in the current philosophical literature that NDT is the most popular current theory of propositional faith. Since the account of faith I defended in Chapter 2 stated that faith that p requires belief that p then NDT is a partial rejection of that account. A thorough discussion of NDT therefore provided a richer evaluation of my own account of faith by treating NDT as an objection to the doxastic component of my theory. NDT also stands on its own as a response to the PFR as I argued in §1.5.2. Despite the popularity of NDT, there has been very little published work critiquing the theory. I hold that NDT fails on two fronts. First, the leading arguments

211 See Whitcomb et al. (2017) and Tanesini (forthcoming).

195 in favour of NDT do not successfully establish it as a tenable theory. I considered four such arguments, and showed where each of these cannot defend the plausibility of NDT. Second, NDT lacks the resources to distinguish genuine from pretend faith as can be seen by the fact that fictionalists can have faith according to the conditions set out by NDT. I have proposed how NDT can respond to this problem by distinguishing faith from pretence in terms of the importance that each give to truth. Although I grant that truth can distinguish faith from pretend faith, taking that approach is only possible for faith toward religious propositions. This way of adapting NDT to respond to the pretence problem yields NDT untenable since propositional faith is an attitude that can be held towards many propositions, including those with non-religious content. A theory of propositional faith towards solely religious content is too uneconomical to be considered plausible. To make progress with NDT I believe that its advocates need to provide stronger arguments in its favour, and to produce clear ways by which to distinguish faith from pretend faith. From my research on the topic, the arguments stated in Chapter 3 are the most widely defended, and so a response to the objections made there would be a good starting point.212 A response to the worries concerning pretence would also be helpful, not merely as a way of making NDT a more established theory, but to clarify how its advocates see NDT as uncommitted to pretence or fictionalist faith.213 On a final point concerning NDT, it’s worth noting that as a response to the PFR, NDT is in fact quite a limited option. Consider this in the religious context, where someone has faith, say, that God is all loving. NDT offers no response to the claim (made in premise (1) of the PFR stated in Chapter 1) that when someone’s faith that God is all loving involves belief that God is all loving, that belief in that proposition cannot be justified. If NDT grants to the advocate of the PFR that faith involving belief is unjustified then it follows that, on NDT, one cannot know p when one has faith that p. The same is true for the response NDT makes to the PFR by claiming that someone’s faith that God is all loving does not require belief that God is all loving (a rejection of premise (2) of the PFR from Chapter 1). In that case, on NDT, one also does not know

212 Three of these arguments are discussed in Malcolm and Scott (forthcoming). 213 The objection is introduced in Malcolm and Scott (forthcoming) and ‘the truth response’ is considered in Malcolm (forthcoming).

196 p when one has faith that p since she lacks belief, and belief is required for knowledge. Without supplementing NDT with a defence of the justification of religious beliefs held on religious faith NDT cannot defend the possibility of knowledge of propositions held on faith. The inability for NDT to defend knowing propositions one has faith toward presents a drawback for NDT as a response to the PFR since religious people would like to know the propositions they have faith toward. NDT cannot deliver on this. The only thing that NDT can deliver on (if it were correct) is the claim that religious faith involves epistemically justified doxastic attitudes in cases where somebody lacks belief. Moreover, since NDT denies that one can have faith that p when disbelieving p, NDT can only defend the justification of agnostic faith – when the doxastic attitudes constituting someone’s faith are suspension of judgment. How concerned should someone be who is undecided about religious claims that her ‘faith’ is epistemically justified? This may be of concern to some people, but I suggest that most people will be more concerned that their faith towards propositions concerning God can be known rather than justified due to lack of belief. Without belief or justification there can be no knowledge, and so NDT cannot deliver this. As far as religious epistemology is concerned, NDT is a limited option and lacks the benefits of the divine testimony theory, which seeks to epistemically justify religious faith. The benefit enjoyed by the theory of faith I have defended in this thesis, according to which faith can be epistemically justified, adds to its attractiveness and gives a further reason to explore doxastic, rather than non-doxastic theories of propositional faith in future faith- research.

197

References Adams, M. M. 1999. Horrendous Evils and the Goodness of God. Ithaca: Cornell University Press. Alonso, F. 2014. ‘What is reliance?’ Canadian Journal of Philosophy 44, 2, pp. 163- 183. Alston, W. 1976a. ‘Two Types of Foundationalism,’ The Journal of Philosophy 73, pp. 165-185. Alston, W. 1976b. ‘Has foundationalism been refuted?’ Philosophical Studies 29, pp. 287-305. Alston, W. 1988. ‘An Internalist Externalism,’ Synthese 74, pp. 265-283. Alston, W. 1991. Perceiving God: The Epistemology of Religious Experience. Ithica: Cornell University Press. Alston, W. 1996. ‘Belief, Acceptance, and Religious Faith,’ in J. Jordan and D. Howard-Snyder (eds.) Faith, Freedom, and Rationality. Lanham: Rowman & Littlefield, pp. 3-27. Alston, W. 2007. ‘Audi on Nondoxastic Faith,’ in J. Greco, A. Mele and M. Timmons (eds.) Rationality and the Good: Critical Essays on the Ethics and Epistemology of Robert Audi. Oxford: Oxford University Press, pp. 124-138. Anscombe, G.E.M. 1979. ‘What is it to Believe Someone?’ in C. F. Delaney (ed.) Rationality and Religious Belief. Notre Dame: University of Notre Dame Press. Aquinas, T. 1947. Summa Theologica. The Fathers of the English Dominican Province (trans.). Aquinas, T. 1953. De Veritate. J. McGlynn (trans.). Chicago: Henry Regnery Company. Aquinas, T. 2002. Aquinas’ Shorter Summa. Manchester: Sophia Institute. Aristotle. 1994. Posterior Analytics. Jonathan Barnes (trans.), 2nd edition. Oxford: Clarendon Press. Audi, R. 1991. ‘Faith, Belief and Rationality,’ Philosophical Perspectives 5, pp. 213- 239. Audi, R. 1994. ‘Dispositional Beliefs and Dispositions to Believe’. Noûs, 28 04, pp. 419-434. Audi, R. 2001. Epistemology: A Contemporary Introduction to the Theory of Knowledge. 2nd edition. London: Routledge.

198

Audi, R. 2003. ‘Contemporary Modest Foundationalism,’ in Pojman, L. (ed.) The Theory of Knowledge: Classical and Contemporary Readings, 3rd edition. Wadsworth: Belmont. Audi, R. 2006. ‘Testimony, Credulity, and Veracity,’ in J. Lackey and E. Sosa (eds.), The Epistemology of Testimony. Oxford: Oxford University Press, pp. 25-45. Audi, R. 2008. ‘Belief, Faith and Acceptance,’ International Journal for Philosophy of Religion 63, pp. 87-102. Audi, R. 2011. Rationality and Religious Commitment. Oxford: Oxford University Press. Augustine. 1998. The Confessions. H. Chadwick (translated and edited). Oxford: Oxford University Press. Austin, J. L. 1962. How to Do Things With Words (2nd ed.). London: Oxford University Press. Baier, A., 1986. ‘Trust and Antitrust,’ Ethics 96, 2, pp. 231-260. Baier, A., 1995. ‘Trust and Antitrust,’ in Moral Prejudices: Essays on Ethics. Cambridge: Harvard University Press, pp. 95–129. Baldwin, E. and McNabb, T. Forthcoming. ‘An Epistemic Defeater for Islamic Belief?' International Journal of Philosophy and Theology. Barth, K. 1956. Church Dogmatics. G. T. Thomson (trans.). New York: Scribner. Bayne, T. 2009. ‘Delusion and Self-Deception: Mapping the Terrain,’ in T. Bayne and J. Fernandez (eds.) Delusion and Self-Deception: Affective and Motivational Influences on Belief-Formation. East Sussex: Psychology Press, pp. 1-22. Bergmann, M. 2005. ‘Defeaters and Higher-Level Requirements,’ The Philosophical Quarterly 65, 220, pp. 419-436. Berkeley, G. 2007. Alciphron: or the Minute Philosopher. A Defence of the Christian Religion against the So-called Free-thinkers. J. Bennett (ed.). Published on: earlymoderntexts.com. Bishop, J. 2007. Believing by Faith: An Essay in the Epistemology and Ethics of Religious Belief. Oxford: Oxford University Press. Bishop, J. 2016. ‘Faith,’ Stanford Encyclopaedia of Philosophy, Online at: https://plato.stanford.edu/entries/faith/. Bonjour, L. 1985. The Structure of Empirical Knowledge. Cambridge: Harvard University Press.

199

Bonjour, L. 2003. ‘A Version of Internalist Foundationalism,’ in L. Bonjour and E. Sosa, Epistemic Justification. Oxford: Blackwell. Brandom. R. 1994. Making it Explicit. Cambridge: Harvard University Press. Brandom, R. 2000. Articulating Reasons. London: Harvard University Press. Bratman, M. 1992. ‘Practical Reasoning and Acceptance in a Context,’ Mind 101, pp. 1-15. Bricke, J. 2000. Mind and Morality: An Examination of Hume's Moral Psychology. Oxford: Oxford University Press. Brogaard, B. 2013. ‘Phenomenal Seemings and Sensible Dogmatism,’ in C. Tucker (ed.) Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, pp. 270-289. Buchak, L. 2012. ‘Can It Be Rational to Have Faith?’ in J. Chandler & V. Harrison (eds.) Probability in the Philosophy of Religion. Oxford: Oxford University Press, pp. 225-246. Buchak, L. 2014a ‘Rational Faith and Justified Belief,’ in L. Callahan and T. O’Connor (eds.) Religious Faith and Intellectual Virtue. Oxford: Oxford University Press, pp. 49–72. Buchak, L. 2014b. ‘Belief, Credence and Norms,’ Philosophical Studies 169, pp. 285– 311. Buchak, L. 2017a. ‘Reason and Faith,’ in W. Abraham and F. Aquino (eds.) The Oxford Handbook of the Epistemology of Theology. Oxford: Oxford University Press. Buchak, L. 2017b. ‘Faith and Steadfastness in the Face of Counter-Evidence,’ International Journal for Philosophy of Religion, 81, pp. 113-133. Burge, T. 1993. ‘Content Preservation,’ The Philosophical Review 102, pp. 457–488. Callahan, L. and O’Connor, T. 2014. Religious Faith and Intellectual Virtue. Oxford: Oxford University Press. Calvin, J. 1960. Institutes of the Christian Religion. F. L. Battles (trans.) and J. T. McNeill (ed.). Philadelphia: Westminster Press (originally published in 1555). Carston, R. 2002. Thoughts and Utterances. Oxford: Blackwell. Chan, T. 2013. ‘Introduction,’ in T. Chan (ed.) The Aim of Belief. Oxford: Oxford University Press, pp. 1-14. Chisholm, R. 1977. Theory of Knowledge. 2nd edn. Englewood-Cliffs: Prentice-Hall.

200

Chisholm, R. 1988. ‘The Indispensability of Internal Justification,’ Synthese 74, pp. 285-296. Choi, S. and Fara, M. 2012. ‘Dispositions,’ Stanford Encyclopaedia of Philosophy. Online at: https://plato.stanford.edu/entries/dispositions/. Churchland, P. 1981. ‘Eliminative Materialism and the Propositional Attitudes,’ Journal of Philosophy, 78, pp. 67–90. Clark, K. J. & Barrett, J. 2010. ‘Reformed Epistemology and the Cognitive Science of Religion,’ Faith and Philosophy 27, pp. 174–189. Clark, K. J. & Barrett, J. 2011. ‘Reidian Religious Epistemology and the Cognitive Science of Religion,’ Journal of the American Academy of Religion 79, pp. 639– 675. Clement of Alexandria. 1971. ‘The Stromata, or Miscellanies,’ in J. Donaldson and A. Roberts (eds.) The Ante-Nicene Fathers: Translations of the Writings of the Fathers Down to A.D. 325, Volume II. Grand Rapids: WM. B. Eerdmands Publishing Company. Clifford, W. K. 1879. ‘The Ethics of Belief,’ in Lectures and Essays. London: Macmillan. Coady, C. A. J. 1992. Testimony: A Philosophical Study. Oxford: Oxford University Press. Cohen, L. J. 1992. An Essay on Belief and Acceptance. Oxford: Oxford University Press. Conee, E. and Feldman, R. 2004. ‘Introduction,’ in E. Conee and R. Feldman (eds.) Evidentialism: Essays in Epistemology. Oxford: Oxford University Press, pp. 2-8. Conee, E. 2013. ‘Seeming Evidence,’ in C. Tucker (ed.) Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, pp. 52-68. Correia, F. and Schnieder, B. (eds.). 2012. Metaphysical Grounding: Understanding the Structure of Reality. Cambridge: Cambridge University Press. Craig, W. L. 1979. The Kalām Cosmological Argument. London: MacMillan. Davidson, D., 1963. ‘Actions, Reasons and Causes’. The Journal of Philosophy 60, 23: pp. 685-700. Davies, B., 2014. Thomas Aquinas’s Summa Theologiae: A Guide and Commentary. Oxford: Oxford University Press.

201

Deng, N. 2015. ‘Religion for Naturalists,’ International Journal for Philosophy of Religion 78, pp. 195-214. Deonna, J., and Teroni. F., 2012. The Emotions: A Philosophical Introduction. Oxford: Routledge. DePaul, M. 2009. ‘Phenomenal Conservatism and Self-Defeat,’ Philosophy and Phenomenological Research 78, pp. 205-212. DePoe, J. 2011. ‘Defeating the Self-defeat Argument for Phenomenal Conservativism,’ Philosophical Studies 152, pp. 347–359. Descartes, R. 1996. Meditations on First Philosophy. J. Cottingham (trans.). Cambridge: Cambridge University Press. Dewey, J. 1934. A Common Faith. New Haven: Yale University Press. Diller, J. 2013. ‘The Conceptual Focus of Ultimism: An Object of Religious Concern for the Nones and Somes,’ Religious Studies 49, 2, pp. 221-233. Dougherty, T. 2011. ‘Introduction,’ in T. Dougherty (ed.) Evidentialism and Its Discontents. Oxford: Oxford University Press, pp. 1-13. Dougherty, T. 2014. ‘Faith, Trust, and Testimony,’ in L. Callahan and T. O’Connor (eds.) Religious Faith and Intellectual Virtue. Oxford: Oxford University Press, pp. 97-121. Dougherty, T. and Tweedt, C. ‘Religious Epistemology,’ Philosophy Compass 10, 8, pp. 547-559. Dougherty, T. and Walls, J. forthcoming. Two Dozen (or so) Arguments for God: The Plantinga Project. Oxford: Oxford University Press. Dummett, M. 1994. ‘Testimony and Memory,’ in Knowing from Words, B. Matilal and A. Chakrabarti (eds.). Boston: Kluwer, pp. 251-272. Eklund, M. 2015. ‘Fictionalism,’ Stanford Encyclopaedia of Philosophy. Online at: https://plato.stanford.edu/entries/fictionalism/ Evans, C. S. 1998. Faith Beyond Reason: A Kierkegaardian Account. Cambridge: Wm. B. Eerdmans Publishing Co. Faulkner, P. 2007. ‘On Telling and Trusting,’ Mind 116, 464, pp. 875–902. Faulkner, P. 2015. ‘The Attitude of Trust is Basic,’ Analysis 75, 3, pp. 424-429. Feldman, R. 1988. ‘Epistemic Obligations,’ Philosophical Perspectives 2, pp. 235- 256. Feldman, R. 2003. Epistemology. Upper Saddle River: Prentice Hall. Field, H. 1980. Science Without Numbers. Princeton: Princeton University Press. 202

Foley, R. 1993. Working Without a Net. New York: Oxford University Press. Fricker, E. 1994. ‘Against Gullibility,’ in Knowing from Words, Matilal and Chakrabarti (eds.), Boston: Kluwer, pp. 125–161. Fricker, E. 1995. ‘Telling and Trusting: Reductionism and Anti-Reductionism in the Epistemology of Testimony,’ Mind 104, 414, pp. 393-411. Fricker, E. 2006. ‘Second-Hand Knowledge,’ Philosophy and Phenomenological Research LXXIII, 3, pp. 592-618. Frost-Arnold, K., 2014. ‘The Cognitive Attitude of Rational Trust’, Synthese 191, pp. 1957–1974. Fuller, S. 1988. Social Epistemology. Bloomington: Indiana University Press. Geach, P. 1977. The Virtues. Cambridge: Cambridge University Press. Gettier, E. 1963. ‘Is Justified True Belief Knowledge?’ Analysis 23, 6, pp. 121-123. Goldberg, S. 2007. Anti-Individualism: Mind and Language, Knowledge and Justification. Cambridge: Cambridge University Press. Goldberg, S. 2010. Relying on Others: An Essay in Epistemology. Oxford: Oxford University Press. Goldberg, S. Forthcoming. ‘Should Have Known,’ Synthese. Goldman, A. 1976. “Discrimination and Perceptual Knowledge,” Journal of Philosophy 73:20, 771-91. Goldman, A. 1979. ‘What Is Justified Belief?’ in G. Pappas (ed.) Justification and Knowledge. Dordrecht: D. Reidel, pp. 1-23. Goldman, A. 1989. ‘BonJour’s The Structure of Empirical Knowledge,’ Philosophical Studies 44, pp. 105-114. Goldman, A. 1999. Knowledge in a Social World. Oxford: Oxford University Press. Goldman, A. 2008. ‘Immediate Justification and Process Reliabilism,’ in Q. Smith (ed.) Epistemology: New Essays. Oxford: Oxford University Press, pp. 63-81. Goldman, A. and Whitcomb, D. 2011. Social Epistemology: Essential Readings. New York: Oxford University Press. Haddock, A. Millar, A. and Pritchard, D. 2010. Social Epistemology. Oxford: Oxford University Press. Hardin. R. 2002. Trust and Trustworthiness. New York: Russell Sage Foundation. Hardwig, J. 1985. ‘Epistemic Dependence,’ The Journal of Philosophy 82, 7, pp. 335- 349. Hawley, K. 2014. ‘Trust, Distrust and Commitment,’ Nous 48, 1, pp. 1–20. 203

Heil. J. 1984. ‘Doxastic Incontinence,’ Mind 93, 369, pp. 56-70. Helm, P. 1999. Faith and Reason. Oxford: Oxford University Press. Helm, P. 2001. Faith with Reason. Oxford: Oxford University Press. Hieronymi, P. 2008. ‘The Reasons of Trust,’ Australasian Journal of Philosophy 86, 2, pp. 213-236. Himma, K. 2006. ‘Christian Faith without Belief that God Exists: A Defense of Pojman‘s Conception of Faith,’ Faith and Philosophy 23, 1, pp. 65-79. Holton, R. 1994. ‘Deciding to Trust, Coming to Believe,’ Australasian Journal of Philosophy 72, 1, pp. 63-76. Howard-Snyder, Daniel. 2005. ‘Foundationalism and Arbitrariness,’ Pacific Philosophical Quarterly 86, pp. 18-24. Howard-Snyder, D. 2013a. ‘Propositional Faith: What It Is and What It Is Not,’ American Philosophical Quarterly 50, 04, pp. 357-372. Howard-Snyder, D. 2013b. ‘Schellenberg on Propositional Faith,’ Religious Studies 49, 02, pp. 181-194. Howard-Snyder, D. 2016a. ‘Comments on Finlay Malcolm’s “How to Defend the Doxastic Theory of Faith”,’ The Value of Faith Conference, San Antonio. Howard-Snyder, D. 2016b. ‘Does Faith Entail Belief?’ Faith and Philosophy 33, 2, pp. 142-162. Howard-Snyder, D. 2017. ‘Markan Faith,’ International Journal for Philosophy of Religion, 81, pp. 31-60. Howard-Snyder, D. & Coffman, E. J. 2006 ‘Three Arguments Against Foundationalism: Arbitrariness, Epistemic Regress, and Existential Support,’ Canadian Journal of Philosophy 36, 4, pp. 535-564. Huber, F. 2009. ‘Belief and Degrees of Belief,’ in Franz Huber and Christoph Schmidt-Petri (eds.) Degrees of Belief. Dordrecht: Springer, pp. 1-36. Huemer, M. 2001. Skepticism and the Veil of Perception. Lanham: Rowman & Littlefield. Huemer, M. 2006. ‘Phenomenal Conservatism and the Internalist Intuition,’ American Philosophical Quarterly 43, pp. 147-158. Huemer, M. 2007. ‘Compassionate Phenomenal Conservatism,’ Philosophy and Phenomenological Research 74, pp. 30-55. Huemer, M. 2009. ‘Apology of a Modest Intuitionist,’ Philosophy and Phenomenological Research 78, pp. 222-236. 204

Hume, D. 1975. Enquiries Concerning Human Understanding and Concerning the Principles of Morals. 3rd edition. Oxford: Clarendon Press. Hume, D. 1978. A Treatise of Human Nature. 2nd edition. Edited by L. A. Selby-Bigge and P. H. Nidditch. Oxford: Clarenden Press. Jay, C. 2014. ‘The Kantian Moral Hazard Argument for Religious Fictionalism,’ International Journal for Philosophy of Religion 75, pp. 207-232. Jay, C. 2016. ‘Testimony, Belief and Nondoxastic Faith: The Humean Argument for Religious Fictionalism,’ Religious Studies 52, pp. 247-261. James, W. 1896. The Will to Believe: And Other Essays in Popular Philosophy. London: Longmans, Green & Co. Johnston, M. 1992. ‘How to Speak of the Colors,’ Philosophical Studies 68, pp. 221– 263. Jones, K. 1996. ‘Trust as an Affective Attitude,’ Ethics 107, 1, pp. 4-25. Jones, K. 2004. ‘Trust and Terror,’ in P. Desautels and M. Walker (eds.) Moral Psychology; Feminist Ethics and Social Theory. Oxford: Rowman & Littlefield Publishers, pp. 3-18. Jones, K. 2012. ‘Trustworthiness,’ Ethics 123, 1, pp. 61-85. Jones, K. 2013. ‘Trusting Interpretations,’ in P. Mäkelä and C. Townley (eds.) Trust: Analytic and Applied Perspectives. New York: Rodopi, pp. 15-30. Jordan, J. 2006. Pascal’s Wager: Pragmatic Arguments and Belief in God. Oxford: Oxford University Press. Joyce, R. 2002. The Myth of Morality. Cambridge: Cambridge University Press. Kalderon, M. E. 2005. Moral Fictionalism. Oxford: Clarendon Press. Kant, I. 1998 [1781/87]. Critique of Pure Reason. P. Guyer & A. Wood (trans.). Cambridge: Cambridge University Press. Kaufman, W. 1958. Critique of Religion and Philosophy. Princeton: Princeton University Press. Kaufman, W. 1978. The Faith of a Heretic. New York: New American Library. Kawall, J. 2013. ‘Friendship and Epistemic Norms,’ Philosophical Studies 165, 2, pp. 349-70. Keller, S. 2004. ‘Friendship and Belief,’ Philosophical Papers 33, 3, pp. 329-51. Kenny, A. 1992. What is Faith? Oxford: Oxford University Press. Keren, A. 2014. ‘Trust and Belief: A Preemptive Reasons Account,’ Synthese 191, pp. 2593–2615. 205

Kierkegaard, S. 1846/1941. Concluding Unscientific Postscript. D. F. Swenson (trans.). Princeton: Princeton University Press. Kim, J. 2008. ‘What is “Naturalized Epistemology”?’ in E. Sosa, J. Kim, J. Fantl and M. McGrath (eds.) Epistemology: An Anthology. Oxford: Blackwell, pp. 538-551. King, R. 2012. ‘Divine Revelation,’ Philosophy Compass 7/7, pp. 495-505. King, R. 2013. ‘Divine Self-Testimony and the Knowledge of God,’ International Journal for Philosophy of Religion 74, pp. 279-295. Klein, P. 1999. ‘Human knowledge and the regress of reasons,’ Philosophical Perspectives 13, pp. 297-325. Klein, P. 2000. ‘The Failures of Dogmatism and A New Pyrrhonism.’ Acta Analytica 15, pp. 7-24. Klein, P. 2004. ‘What is wrong with foundationalism is that it cannot solve the epistemic regress problem,’ Philosophy and Phenomenological Research LXVIII, pp. 166-171. Kvanvig, J. 2003. The Value of Knowledge and the Pursuit of Understanding. Cambridge: Cambridge University Press. Kvanvig, J. 2009. ‘Assertion, Knowledge, and Lotteries,’ in D. Pritchard and P. Greenough (eds.) Williamson on Knowledge. Oxford: Oxford University Press, pp. 140-160. Kvanvig, J. 2010. ‘The Swamping Problem Redux: Pith and Gist,’ in A. Haddock, A. Millar and D. Pritchard (eds.) Social Epistemology. Oxford: Oxford University Press, pp. 89-111. Kvanvig, J. 2013. ‘Affective Theism and People of Faith,’ Midwest Studies in Philosophy XXXVII, pp. 109-128. Kvanvig, J. Forthcoming. ‘Faith and Humility.’ Oxford University Press. Lackey, J. 1999. ‘Testimonial Knowledge and Transmission,’ The Philosophical Quarterly 49, pp. 471–490. Lackey, J. 2003. ‘A Minimal Expression of Non-Reductionism in the Epistemology of Testimony,’ Noûs 37, pp. 706-723. Lackey, J. 2006. ‘The Nature of Testimony,’ Pacific Philosophical Quarterly 87, pp. 177–97. Lackey, J. 2007. ‘Norms of Assertion,’ Noûs 41, 4, pp. 594-626. Lackey, J. 2008. Learning from Words. Oxford: Oxford University Press.

206

Lackey, J. 2011. ‘Testimony: Acquiring Knowledge From Others,’ in A. Goldman and D. Whitcomb (eds.) Social Epistemology: Essential Readings. Oxford: Oxford University Press, pp. 71-91. Lackey, J. and Sosa, E. 2006. The Epistemology of Testimony. Oxford: Oxford University Press. Lamont, J. 2004. Divine Faith. Aldershot: Ashgate. Lamont, J. 2009. ‘A Conception of Faith in the Greek Fathers,’ in O. Crisp and M. Rea (eds.) Analytic Theology: New Essays in the Philosophy of Theology. Oxford: Oxford University Press, pp. 87-116. Le Poidevin, R. 1996. Arguing for Atheism. London: Routledge. Leiter, B. 2012. Why Tolerate Religion? Princeton: Princeton University Press. Leng, M. 2005. ‘Revolutionary Fictionalism: A Call to Arms,’ Philosophia Mathematica III, pp. 277-293. Lipton, P. 2007. ‘Science and Religion: The Immersion Solution,’ in A. Moore and M. Scott (Eds.) Realism and Religion: Philosophical and Theological Perspectives. Aldershot: Ashgate, pp. 31-46. Littlejohn, C. 2011. ‘Defeating Phenomenal Conservatism,’ Analytic Philosophy 52, pp. 35-48. Littlejohn, C. 2012. Justification and the Truth Connection. Cambridge: Cambridge University Press. Locke, J. 1694. The Works, vol. 6 (The Reasonableness of Christianity). The Online Library of Liberty. Locke, J. 1689/1924. Essay Concerning Human Understanding. A.S.Pringle-Pattison (ed.). Oxford: Clarendon Press. Lord, E and Maguire, B. 2016. Weighing Reasons. Oxford: Oxford University Press. Luther, M. 1907. Luther's Catechetical Writings, J. N. Lenker (trans.), 2 vols. Minneapolis: Lutheran Press. Lyons, J. 2009. Perception and Basic Beliefs. New York: Oxford University Press. MacCormick, N. 1984. ‘Voluntary Obligations,’ in Legal Right and Social Democracy: Essays in Legal and Political Philosophy. Oxford: Clarendon Press, pp. 189-214. MacDonald, S. 1993. ‘Christian Faith,’ in E. Stump (ed.) Reasoned Faith: Essays in Honour of Norman Kretzmann. Vermont: Echo Point Books, pp. 42-69.

207

Macfarlane, J. 2012. ‘What is Assertion.’ in J. Brown and H. Cappelan (eds) Assertion: New Philosophical Essays. Oxford: Oxford University Press, pp. 79-96. Mackie, J. L. 1977. Ethics: Inventing Right and Wrong. London: Penguin. Mackie, J. L. 1983. The Miracle of Theism. Oxford: Clarendon Press. Malcolm, F. Forthcoming. ‘Can Fictionalists Have Faith?’ Religious Studies. Malcolm, F. and Scott, M. Forthcoming. ‘Faith, Belief and Fictionalism,’ Pacific Philosophical Quarterly. Markie, P. 2005. ‘The Mystery of Direct Perceptual Justification,’ Philosophical Studies 126, pp. 347-373. Markie, P. 2006. ‘Epistemically Appropriate Perceptual Belief,’ Noûs 40, pp. 118- 142. Markie, P. 2013. ‘Searching for True Dogmatism,’ in C. Tucker (ed.) Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, pp. 248-268. Mavrodes, G. 1963 ‘Some Puzzles Concerning Omnipotence,’ Philosophical Review 72, pp. 221–3. Mavrodes, G. 1988. Revelation in Religious Belief. Philadelphia: Temple University Press. McCraw, B. 2015. ‘Faith and Trust,’ International Journal for Philosophy of Religion, 77, pp. 141-158. McDowell, J. 1998. ‘Knowledge by Hearsay,’ in Meaning, Knowledge and Reality. Cambridge, MA: Harvard University Press, pp. 414-444. McGeer, V., 2008. ‘Trust, Hope and Empowerment’, Australasian Journal of Philosophy, 86:2 pp. 237 – 254. McGinn, C. 2011. Truth by Analysis: Games, Names, and Philosophy. Oxford, Oxford University Press. McGrath, Matthew. 2013. ‘Phenomenal Conservatism and Cognitive Penetration: The ‘Bad Basis’ Counterexamples,’ in C. Tucker (ed.) Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, pp. 225-247. McKaughan, D. 2013. ‘Authentic Faith and Acknowledged Risk: Dissolving the Problem of Faith and Reason,’ Religious Studies 49, 01, pp. 101-124.

208

McKaughan, D. 2014. ‘Faith as Active Commitment,’ The Nature of Faith Conference, St. Louis. Online here: https://www.youtube.com/watch?v=YrgVd96h2WA. McKaughan, D. 2017. ‘On the Value of Faith and Faithfulness,’ International Journal for Philosophy of Religion, 81, pp. 7-29. McLeod, C. 2015. ‘Trust,’ Stanford Encyclopaedia of Philosophy. Online at: http://plato.stanford.edu/entries/trust/. McMyler, B. 2011. Testimony, Trust and Authority. Oxford: Oxford University Press. Mele, A. 2009. ‘Self-Deception and Delusions,’ in T. Bayne and J. Fernandez (eds.) Delusion and Self-Deception: Affective and Motivational Influences on Belief Formation. East Sussex: Psychology Press, pp. 55-70. Mittag, D. 2004. ‘Evidentialism,’ in Internet Encyclopaedia of Philosophy. Online at: http://www.iep.utm.edu/evidenti/. Moon, A. 2016. ‘Recent Work in Reformed Epistemology,’ Philosophy Compass 11, pp. 879–891. Moon, A. Forthcoming. ‘The Nature of Doubt and a New Puzzle About Belief, Doubt, and Confidence,’ Synthese. Moreland, J. P. & Craig, W. L. 2003. Philosophical Foundations for a Christian Worldview. Madison: InterVarsity Press. Mugg, J. 2016. ‘In Defense of the Belief-Plus Model of Faith,’ European Journal of Philosophy of Religion 8, 2, pp. 201-219. Murty, S. K. 1959. Revelation and Reason in Advaita Vedanta. New York: Columbia University Press. Muyskens, J. L. 1979. The Sufficiency of Hope: The Conceptual Foundations of Religion. Philadelphia: Temple University Press. Nolan, D., Restall, G. & West, C. 2004. ‘Moral Fictionalism Versus the Rest,’ Australasian Journal of Philosophy 83, pp. 307-330. Nottelmann, N. 2011. ‘Belief-Desire Explanation,’ Philosophy Compass, 6, 12, pp. 912-921. O’Neill, O. 2002. Autonomy and Trust in Bioethics. Cambridge: Cambridge University Press. Olson, J., 2014. Moral Error Theory: History, Critique, Defence. Oxford: Oxford University Press.

209

Owen, J. 1852. ‘The Reason of Faith,’ in W. Goold (ed.), The Works of John Owen, vol.4. Edinburgh: Johnstone and Hunter. Park, S. B. 1983. Buddhist Faith and Sudden Enlightenment. New York: SUNY Press. Pascal, B. 1660/1962. Pascal's Pensées. Martin Turnell (trans.). New York: Harper and Brothers. Pelling, C. 2014. ‘Assertion, Telling, and Epistemic Norms,’ Australasian Journal of Philosophy 92, 2, pp. 335-348. Penelhum, T., 1989. ‘The analysis of faith in St. Thomas Aquinas’ in T. Penelhum (ed.). Faith. London: Collier Macmillan. Pettit, P. 1995. ‘The Cunning of Trust,’ Philosophy and Public Affairs 24, 3 pp. 202– 25. Pinker, S. 2006. ‘Less Faith, More Reason,’ Harvard Crimson, October 27. Online at: http://www.thecrimson.com/article/2006/10/27/less-faith-more-reason-there-is/. Plantinga, A., and Wolterstorff, N. 1983. Faith and Rationality: Reason and Belief in God. Notre Dame: University of Notre Dame Press. Plantinga, A. 1983. ‘Reason and Belief in God,’ in A. Plantinga & N. Wolterstorff (eds.) Faith and Rationality: Reason and Belief in God. Notre Dame: University of Notre Dame Press, pp. 16-93. Plantinga, A. 2000. Warranted Christian Belief. New York: Oxford University Press. Pojman, L. 1986. ‘Faith without Belief,’ Faith and Philosophy 3, pp. 157-176. Pojman, L. 2001. Philosophy of Religion. New York: McGraw-Hill. Poston, T. 2007. ‘Internalism and Externalism in Epistemology,’ Online at: http://www.iep.utm.edu/int-ext/. Poston, T. 2016. ‘Foundationalism,’ Internet Encyclopaedia of Philosophy. Online at: http://www.iep.utm.edu/found-ep/. Potter, N. 2002. How Can I Be Trusted? Lanham: Rowan and Littlefield. Price, H. H. 1965. ‘Belif ‘In’ and Belief ‘That’,’ Religious Studies 1, 1, pp. 5-27. Ramsey, F. 1926. Truth and Probability. Reprinted from The Foundations of Mathematics and Other Logical Essays, by R. B. Braithwaite (ed.) 1950, New York: The Humanities Press. Ratcliffe, M. 2008. Feelings of Being. Oxford: Oxford University Press. Raz, J. 1978. ‘Authority and Consent,’ Virginia Law Review 67, pp. 103–17. Reid, T. 1983. Thomas Reid's Inquiry and Essays, R. Beanblossom and K. Lehrer (eds.). Indianapolis: Hackett. 210

Reisner, A. 2008. ‘Weighing Pragmatic and Evidential Reasons for Belief,’ Philosophical Studies, 138, pp. 7–27. Reisner, A. 2009, ‘The Possibility of Pragmatic Reasons for Belief and the Wrong Kind of Reasons Problem,’ Philosophical Studies 145, pp. 257–272. Sanders, E. P. 1985. Jesus and Judaism. London: SCM Press. Schellenberg, J. L. 2005. Prolegomena to a Philosophy of Religion. Ithaca: Cornell University Press. Schellenberg, J. L. 2007. The Wisdom to Doubt: A Justification of Religious Skepticism. Ithaca: Cornell University Press. Schellenberg, J. L. 2009. The Will to Imagine: A Justification of Skeptical Religion. Ithaca: Cornell University Press. Schellenberg, J. L. 2013. ‘Replies to My Colleagues,’ Religious Studies 49, pp. 257- 285. Schellenberg, J. L. 2014. ‘How to Make Faith a Virtue,’ in L. Callahan and T. O’Connor (eds.) Religious Faith and Intellectual Virtue. Oxford: Oxford University Press, pp. 74–92. Schmitt, F. 1994. Socializing Epistemology: The Social Dimensions of Knowledge. Lanham: Rowman & Littlefield. Schroeder, T. 2006. ‘Propositional Attitudes,’ Philosophy Compass, 1, pp. 65-73. Schwitzgebel, E. 2002. ‘A phenomenal, dispositional account of belief,’ Noûs, 36, pp. 249–275. Schwitzgebel, E. 2013. ‘A dispositional approach to attitudes: Thinking outside the belief box,’ in N. Nottelmann (ed.) New Essays on Belief. New York: Palgrave Macmillan, pp. 75–99. Scott, M. 2013. Religious Language. London: Palgrave Macmillan. Searle, J. 1969. Speech Acts. Cambridge: Cambridge University Press. Searle, J. 1979. Expression and Meaning. Cambridge: Cambridge University Press. Searle, J. 1983. Intentionality. Cambridge: Cambridge University Press. Sessions, W. L. 1994. The Concept of Faith. Ithaca: Cornell University Press. Shogenji, T. 2006, ‘A Defense of Reductionism about Testimonial Justification of Beliefs,’ Noûs, 40, pp. 331–346. Siegel, S. 2013. ‘The Epistemic Impact of the Etiology of Experience,’ Philosophical Studies 162, pp. 697-722.

211

Skene, M. 2013. ‘Seemings and the Possibility of Epistemic Justification,’ Philosophical Studies 163, pp. 539-559. Slote, M., 2014. A Sentimentalist Theory of the Mind. Oxford: Oxford University Press. Smith, M. 1994. The Moral Problem. Oxford: Blackwell. Smith, W. C., 1979. Faith and Belief: The Difference Between Them. Oxford: Oneworld Publications. Sosa, E. 1980. ‘The Raft and the Pyramid,’ Midwest Studies in Philosophy 5, pp. 3- 26. Sosa, E. 1991. Knowledge in Perspective. Cambridge: Cambridge University Press. Sosa, E. 2007. A Virtue Epistemology. Oxford: Oxford University Press. Steglich-Petersen, A. 2013. ‘Truth as the Aim of Epistemic Justification,’ in Timothy Chan (ed.) The Aim of Belief. Oxford: Oxford University Press. Steup, M. 2000. ‘Doxastic Voluntarism and Epistemic Deontology,’ Acta Analytica 15, pp. 25–56. Stich, S. 1983. From Folk Psychology to Cognitive Science, Cambridge: MIT Press. Stump, E., 2003. Aquinas. Abingdon: Taylor & Francis. Swinburne, R. 2001. ‘Plantinga On Warrant,’ Religious Studies 37, pp. 203-214. Swinburne, R. 2004. The Existence of God. 2nd edition. Oxford: Oxford University Press. Swinburne, R. 2005. Faith and Reason. 2nd edition. Oxford: Oxford University Press. Swinburne, R. 2011. ‘Evidence,’ in T. Dougherty (ed.) Evidentialism and Its Discontents. Oxford: Oxford University Press. Swindal, J. 2001. ‘Faith and Reason,’ Internet Encyclopaedia of Philosophy. Online at: http://www.iep.utm.edu/faith-re/. Suits, B. 1967. ‘What is a Game?’ Philosophy of Science 34, 2, pp. 148-156. Sylvan, K. 2016. ‘Epistemic Reasons I: Normativity.’ Philosophy Compass 11, 7, pp. 364–376. Tanesini, A. Forthcoming. ‘Intellectual Humility as Attitude,’ Philosophy and Phenomenological Research. Tucker, C. 2010. ‘Why Open-Minded People Should Endorse Dogmatism,’ Philosophical Perspectives 24, pp. 529-545. Turri, J. 2009. ‘The Ontology of Epistemic Reasons,’ Nous 43, pp. 490–512. van Fraassen, B. 1980. The Scientific Image. Oxford: Oxford University Press. 212

Wahlberg, M. 2014. Revelation as Testimony. Cambridge: Eerdmans Publishing Company. Watson, J. 2016. ‘Epistemic Justification,’ Internet Encyclopaedia of Philosophy. Online at: http://www.iep.utm.edu/epi-just/. Whitcomb, D., Battaly, H., Baehr, J. and Howard-Snyder, D. 2017. ‘Intellectual Humility: Owning Our Limitations,’ Philosophy and Phenomenological Research, XCIV, 3, pp. 509-539. Whiting, D. 2015. ‘Truth is (Still) the Norm for Assertion: A Reply to Littlejohn,’ Erkenntnis 80, pp. 1245–1253. Williamson, T. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. Wittgenstein, L. 1953. Philosophical Investigations. G.E.M Anscombe (trans.). Oxford: Blackwell. Wolterstorff, N. 1996. Divine Discourse: Philosophical Reflections on the Claim that God Speaks. Cambridge: Cambridge University Press. Wolterstorff, N. 2001. Thomas Reid and the Story of Epistemology. Cambridge: Cambridge University Press. Wood, W. 2014. ‘Faith’s Intellectual Rewards,’ in L. Callahan and T. O’Connor (eds.) Religious Faith and Intellectual Virtue. Oxford: Oxford University Press, pp. 29– 48. Zagzebski, L. 1996. Virtues of the Mind. Cambridge: Cambridge University Press. Zagzebski, L. 2012. Epistemic Authority: A Theory of Trust, Authority, and Autonomy in Belief. Oxford: Oxford University Press. Zazebski, L. 2014. ‘Trust,’ in K. Timpe and C. Boyd (eds.) Virtues and Their Vices. Oxford: Oxford University Press, pp. 269-283.

213