Psychological Arguments for Free Will
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By
Andrew Kissel
Graduate Program in Philosophy
The Ohio State University
2017
Dissertation Committee:
Richard Samuels, Advisor
Declan Smithies
Abraham Roth
Copyrighted by
Andrew Kissel
2017
Abstract
It is a widespread platitude among many philosophers that, regardless of whether we actually have free will, it certainly appears to us that we are free. Among libertarian philosophers, this platitude is sometimes deployed in the context of psychological arguments for free will. These arguments are united under the idea that widespread claims of the form, “It appears to me that I am free,” on some understanding of appears, justify thinking that we are probably free in the libertarian sense. According to these kinds of arguments, the existence of free will is supposed to, in some sense, “fall out” of widely accessible psychological states. While there is a long history of thinking that widespread psychological states support libertarianism, the arguments are often lurking in the background rather than presented at face value. This dissertation consists of three free-standing papers, each of which is motivated by taking seriously psychological arguments for free will. The dissertation opens with an introduction that presents a framework for mapping extant psychological arguments for free will. In the first paper, I argue that psychological arguments relying on widespread belief in free will, combined with doxastic conservative principles, are likely to fail. In the second paper, I argue that psychological arguments involving an inference to the best explanation of widespread appearances of freedom put pressure on non-libertarians to provide an adequate alternative explanation. In the third paper, I argue that the best non-libertarian explanations of a related phenomenon, the apparently widespread belief in ii indeterministic choice, fail to meet their own conditions of adequacy. The appendix
explores an alternative non-libertarian line of explanation of the apparently widespread
belief that choice is indeterministic.
iii
Dedication
To my beloved wife, Teresa.
iv
Acknowledgments
I owe too much to too many people in the course of writing this dissertation to
properly thank all of them adequately in this short space. Nevertheless; thank you to my
advisor, Professor Richard Samuels, whose philosophical and personal guidance shaped
me into the philosopher that I am today as much as it shaped this dissertation. Thank you
to my committee members, Professors Declan Smithies and Abraham Roth, for their
invaluable feedback and support. Thank you to Professor Timothy Schroeder, without whom this project would never even have begun. Thank you also to the numerous
philosophers, including Professor Julia Jorati, Professor Robert Kraut, Professor Tristram
McPherson, Professor Chris Pincock, and Professor Tumar Rudavsky, who have devoted
time and energy to reading drafts, discussing my work, and guiding me in the
philosophical arts. Without all of these people, and many unnamed others, this project
would have fallen apart long ago.
Thank you also to my graduate student colleagues at Ohio State, both for their
philosophical input as well as their valuable support and friendship. My time working on
this dissertation in Columbus, while arduous at times, has also been an extreme pleasure
thanks to all of you.
Finally, thank you to my families in the U.S. and in Canada. My parents, John
and Peggy, my sister, Emily, and brother-in-law, Paul, my brother, Jimmie, and sister-in-
v law Jackie, have all listened patiently and provided advice over drinks when needed, and told me to stop thinking so much when needed as well. My parents-in-law, Brian and
Vikki Kouri, and my brother-in-law Brad, have provided an oasis in Montreal to visit and decompress from philosophical work. I thank all of you from the bottom of my heart.
And of course, thank you to my wife, Teresa, to whom this work is dedicated. She has been my engine, pushing me forward when I was ready to give up, and a constant source of love and support. Teresa, I love you, and look forward to many more years of happiness and philosophy with you.
vi
Vita
June 2006 ...... Worthington Kilbourne High School
2010...... B.A. Philosophy, Georgetown University
2010 to present ...... Graduate Teaching Associate, Department
of Philosophy, The Ohio State University
Publications
Review of A. Mele’s Free: Why science hasn’t disproved free will. (2015). Teaching
Philosophy, Vol. 38 No. 3
Fields of Study
Major Field: Philosophy
vii
Table of Contents
Abstract ...... ii
Acknowledgments...... v
Vita ...... vii
List of Figures ...... xi
Introduction- A Framework for Dealing with Psychological Arguments for Free Will..... 1
1. What is represented? ...... 5
2. What are the psychological states? ...... 9
3. How do these states support libertarianism? ...... 10
4. Looking Backward and Forward ...... 12
Chapter 1: The Cartesian Doxastic Argument ...... 15
1. Doxastic Conservatism ...... 18
2. The Principle of Credulity & the Cartesian Doxastic Argument ...... 21
3. The PoC and the Bootstrapping Problem...... 26
4. Reformulating the Principle of Credulity ...... 31
5. Conclusion ...... 39
viii
Chapter 2: On Libertarianism as an Explanatory Hypothesis ...... 43
1. Appearances of Freedom ...... 46
2. Four Readings of Appearance Claims ...... 51
2.1 The Subjunctive Reading...... 51
2.2 The Epistemic Reading ...... 54
2.3 The Comparative Reading ...... 55
2.4 The Phenomenological Reading ...... 56
3. The Prima Facie Case for Libertarianism ...... 61
3.1 Premise 1 ...... 63
3.2 Premise 2 ...... 63
3.3 Premise 3- Assessing Libertarian Alternatives ...... 68
4. Assessing Non-Libertarian Alternatives ...... 76
5. Conclusion ...... 81
Chapter 3: Indeterministic Intuitions and the Spinozan Strategy ...... 82
1. Setting the Stage ...... 84
2. How Possibly Story 1: The Deductivist Account ...... 93
2.1 The Account ...... 93
2.2 Concerning the Folk Acceptance of (D2) ...... 98
2.3 Objections to the use of IOEDs in the Deductivist Account ...... 103
ix
3. How Possibly Story 2: The Probabilistic Account ...... 107
3.1 The Account ...... 107
3.2 Objections to the Probabilistic Account ...... 110
4. Conclusion ...... 115
References ...... 120
Appendix A: Thinking Reasonably about Indeterministic Choice Beliefs ...... 132
1. The Spinozan Strategy and its Problems ...... 133
2. Choosing on the basis of Reasons ...... 138
3. Constructing RBAs ...... 143
4. RBAs Contrasted with Other Non-Spinozan Views ...... 151
5. Conclusion ...... 156
x
List of Figures
Figure 1- A Menu of Options...... 12
xi
Introduction- A Framework for Dealing with Psychological Arguments for Free Will
In conversations with non-philosophers about how they see themselves deliberating and making choices, people often assert that in paradigm cases, choices are
“made freely.” When pushed as to why they think their choices are free, people often tend to respond with some variation of, “I don’t know…I just do.” When pushed even further, they often say something like, “I just feel free.” Such ruminations on the psychology of free will are not limited to casual conversations over beers at the pub; the history of the free will debate is liberally peppered with supporters and detractors discussing the apparently widespread conviction that free will exists (Descartes CSM II,
342; Spinoza, 1677/ 2005, IIp2s; Reid, 1788/2010, 312; Hobbes, 1656/2005, 41). For
example, Descartes writes in his response to Gassendi:
Your own words [Gassendi], however, establish that you have in fact had
the experience of freedom….Now this would be quite impossible unless
the will had the freedom to direct itself; without the determination of the
intellect, towards one side or the other. (CSM II, 260).1
Thomas Reid also appeals to widespread psychology to argue for free will.
1 This dissertation focuses on contemporary arguments for libertarianism. By using historical examples to motivate the project, I do not mean to assume that they are strictly libertarian in the full contemporary sense. 1
We have, by our constitution, a natural conviction or belief that we act
freely: A conviction so early, so universal and so necessary in most of our
rational operations, that it must be the result of our constitution, and the
work of him that made us. (Reid, 1764/2010, 312).
In recent years, such anecdotal evidence has been supplemented by a flurry of empirical work devoted to the study of folk attitudes towards free will (Nahmias et. al., 2004;
Nichols & Knobe, 2007; Deery, 2013; 2015; Sarkissian, 2011). These studies seek to determine whether there really is a widespread conviction that free will exists, and if so, where these convictions might come from.
In the context of this recent empirical work, numerous philosophers have attempted to revive what might be called psychological arguments for free will, like those found in Descartes and Reid (Campbell, 1967; O’Connor, 1995; Pink, 2004; Searle,
2007; Mawson, 2011; Swinburne, 2013). Broadly, psychological arguments for free will take purportedly common, intuitive mental states, and argue that these mental states support the contention that we have free will.
[T]he agency theory is appealing because it captures the way we
experience our own activity… Such experiences could of course be wholly
illusory, but do we not properly assume, in the absence of strong
countervailing reasons, that things are pretty much the way they appear to
us? (O’Connor, 1995, 106).
2
Freedom, as much as causation, is something that experience represents…
If we still can acquire knowledge of causal power by relying on its fallible
representation by experience, then we can just as well acquire knowledge
of non-causal freedom too, and in the same way. (Pink, 2004, 122).
The principle of credulity… says that things are probably the way they
seem to be in the absence of counterevidence. So we should believe that
in these circumstances where we believe that we are making a choice
without being caused to choose as we do, that we are indeed doing just
that. (Swinburne, 2013, 201).
These arguments are united under the idea that widespread claims of the form, “It appears to me that I am free,” on some understanding of appears, justify thinking that we are probably free.
Generally, these recent psychological arguments have been advanced in support of a libertarian conception of free will; free will that is incompatible with causal determinism. Yet despite their long history in philosophy, psychological arguments are often found lurking in the wings rather than presented at face value. As a result, the details of the arguments are largely obscure in recent presentations. Due to this lack of precision, many opponents of psychological arguments have contented themselves with dismissively pointing out that folk convictions can be (and often are!) mistaken, and so they provide little or no support for libertarianism (O’Connor, 1971, 18; Smilansky,
2000, 59-60; Sommers, 2007). In contrast to these quick dismissals, the present
3
dissertation aims to take psychological arguments more seriously. Working out the
details implicit in arguments that draw on psychological states involves first identifying
which kinds of psychological states support libertarianism, and the way in which those
states are supposed to support the contention that we have free will. As such, this
introduction presents a menu of refinements of psychological arguments for free will,
beginning with the following schema:
Psychological Arguments for Free Will
i. There is some widespread psychological state Ψ
ii. We have reason to think that in virtue of Ψ, libertarianism is probably true.
iii. Therefore, libertarianism is probably true.
Psychological arguments consist of two main moving parts. The first establishes that there is some psychological state(s), which purportedly represents our own free will in libertarian terms. The second moving part is a justification or rationale for thinking that these psychological states in some way justify believing that we have free will of the kind represented by Ψ.
To get clearer on these arguments, in what follows I distinguish libertarianism, a metaphysical thesis regarding free will, from the position that holds that there is widespread folk belief in libertarianism, which I call doxastic libertarianism, as well as the position that holds that there are widespread experiences as of free will in the libertarian sense, which I call experiential libertarianism. I show how these latter two positions are distinct from each other and have each been used in psychological arguments for libertarianism as a metaphysical thesis. I further distinguish between two
4
ways of justifying libertarianism on the basis of these psychological states, one invoking
conservative epistemic principles and one relying on a form of inference to the best
explanation. The resulting picture is a menu of four distinct psychological arguments for
libertarianism, two that rely on doxastic libertarianism and two that rely on experiential
libertarianism, which accommodates most contemporary psychological arguments for
libertarianism. I close by briefly outlining the other papers that compose this dissertation.
Each is motivated by taking seriously some aspect of extant versions of the psychological arguments distinguished in this introduction.
1. What is represented?
The first step in refining psychological arguments is to get clearer on what, in broad strokes, a psychological state would have to represent in order to count as a psychological state about libertarianism. As a first pass, libertarianism can be characterized generally by the following three claims.
1. Free will is incompatible with determinism.
2. Determinism is false.
3. Humans have free will.
According to the first claim, the truth of determinism precludes the possibility of free will. For our purposes here, we can understand the thesis of determinism as causal determinism. That is, the state of the universe at a given time t, along with the laws of nature, are sufficient to necessitate the state of the universe at a later time t+1.
There are numerous motivations for accepting (1). One prominent motivation can be seen in the Principle of Alternative Possibilities (PAP), which states, “a person is
5
morally responsible for what he has done only if he could have done otherwise”
(Frankfurt, 1969). The thought is that free will is a necessary requirement for a person to
be morally responsible. Free will on this reading is at least partially constituted by the
claim that the person in question could have in fact done other than he did. If the thesis
of determinism is true, then it is not true that a person could have done otherwise in the
relevant sense. If determinism is true, it is impossible for a person to have free will, thus
the incompatibility of free will and determinism.
According to (2), the thesis of determinism is false in the actual world. Some find
the adoption of (2) to be in conflict with current scientific practice. They argue that
science is showing that we live in a deterministic world, or at least, a world that is deterministic enough (Caruso, 2012; Harris, 2012). While it may be true that scientists adopt a deterministic attitude in order to aid in making accurate predictions and theorizing, this is not the same as science “proving” the truth of determinism (Nahmias,
2010). The falsity of determinism alone is not sufficient to secure the claim that people
have free will (Strawson, 1986; Van Inwagen, 2000). Libertarians must make the further
claim (3) that we do, in fact, have free will.
One might read (1) as setting constraints on what free will is, (2) secures the
possibility that we have free will, and (3) asserts that we in fact have free will. But such
a minimal characterization says very little about the particular way in which we might
have free will. Thus, in addition to these minimal characterizations, many libertarians
hold one or more of the following three claims to be true:
4. There are true Libertarian Counterfactuals.
6
5. The necessity of (1)-(3) for moral responsibility.
6. The existence of agent causal powers.
What does it mean for Libertarian Counterfactuals to be true? The intuitive idea is that if, having made a choice, we were to run time back and have me choose again without changing anything about the case, I could choose differently. Take the case where you choose chocolate ice cream rather than vanilla. If Libertarian Counterfactuals are true, then were I to face the exact same choice over again, I might choose vanilla even without any change in the antecedent conditions.
The thought can be expressed using standard Lewisian semantics for modal claims. Suppose that Todd freely chooses at time t to spend the last of his allowance on a new video game. If determinism is false and humans have free will, then that means there are possible worlds similar to the actual world where Todd does other than he actually chose to do; perhaps he chooses to save his allowance. According to (4), the closest possible worlds where Todd chooses other than he actually did are worlds where every state of the world leading up to t, along with the laws of nature, are identical to those in the actual world. That is, the difference in Todd’s choice in these closest possible worlds is not in virtue of some prior difference in Todd’s psychology or in virtue of any prior different causal conditions. These Libertarian Counterfactual worlds can be contrasted with counterfactual worlds where Todd makes a different choice at time t, but this world includes some difference in antecedent conditions prior to the choice. While non-libertarians might agree that it is possible for Todd to do otherwise, they will reject
7 the claim that Libertarian Counterfactuals are true at the closest possible worlds.2 That is, if Todd does otherwise, then this is due to some change in the prior antecedent conditions, perhaps a difference in the past or a difference in the laws of nature. The notion of Libertarian Counterfactuals is closely tied to (5) and the PAP, so it is rare to find philosophers who adopt (4) without adopting (5) as well.
Finally, some philosophers are not satisfied with the minimal claim that some human actions are not determined. They attempt to explicate the fact that freely willed actions are different from the normal, deterministic causal chains found in nature via a special power of rational agents. That is, humans can will and bring about actions via agent causation, as opposed to event causation (Taylor, 1992; O’Connor, 2000; Mawson,
2011; Swinburne, 2013). According to the agent causal view, our ability to bring about events is not merely in virtue of some prior chain of causal events. Instead, freely willed actions involve direct causation by a persisting substance, in this case the agent. The fact that an agent who is not determined by prior causal events can nevertheless be causally efficacious is used to explain how it is possible that a person could have done other than he did.
For present purposes, we can summarize libertarianism as the metaphysical position that accepts that (1)-(3), as well as at least one of (4)-(6), obtain in the actual world. However, in what follows, we will need to be able to discuss different attitudes one might take towards (1)-(6), independently of whether they think (1)-(6) actually obtain.
2 I use a Lewisian framework for illustrative purposes only. The concept of Libertarian Counterfactual can be adapted to one’s preferred modal semantics. 8
2. What are the psychological states?
Psychological arguments for libertarianism attempt to identify some widespread psychological states and argue that the libertarianism is justified on the basis of those states. As argued in the previous section, this usually involves the use of psychological states with libertarian contents. Extant versions of the psychological argument for free will largely rely on beliefs and experiences to be the conveyors of those contents.3 To this end, doxastic libertarianism (DL) is the position that there are widespread beliefs in
(1)-(3), or at least beliefs that imply (1)-(3). As a result, according to proponents of DL, belief in libertarianism is in some sense the “default” or intuitive position held by most people. DL is officially neutral on the particular source of the beliefs; they could be held innately, or acquired through testimony or perception, or some other source. However,
DL does assume that folk beliefs regarding (1)-(3) are not necessarily formed on the basis of philosophical argument. Philosophical justification for accepting libertarianism aside,
DL makes only the descriptive claim that beliefs implying (1)-(3) are widespread.
Proponents of DL may additionally hold that there are widespread beliefs in (4)-
(6). In many ways each of (4)-(6) can be seen as implying one of (1)-(3). The existence of agent causal powers as normally defended, for example, would imply that determinism is false. So if there is widespread belief in (6), this would count as support for DL.
Similarly, widespread belief in the indeterminacy of choice could contribute to a folk
3 I restrict this dissertation to considering beliefs and experiences, since the vast majority of psychological arguments for free will (as well as the empirical data) appeal exclusively to one or both of these mental state kinds. The plausibility of arguments involving alternative mental states would depend largely on the states in question. Desires, for example, are often denied as grounds for accepting libertarianism (Nadelhoffer & Matveeva, 2009; Vargas, 2013). In contrast, seemings, understood as mental states independent of beliefs or experiences, could provide a more promising alternative. 9 belief in libertarianism. It is important to reiterate that DL is not the normative claim that belief in any of (1)-(6) is justified. It is simply the descriptive claim that people tend to have widespread beliefs that support a belief in libertarianism.
In contrast to beliefs, psychological arguments for libertarianism have also been forwarded on the basis of widespread experience. I call that position which holds that there is widespread experience of (1)-(3), experiential libertarianism (EL). EL holds that certain widespread experiences are appropriately understood as having libertarian contents. For example, if there were widespread experiences of choice as being both free and uncaused, then a form of EL would be true. The evidence in support of EL is largely phenomenological. As a preview, it involves some of the following claimed experiences: the experience of agency, the experience of being an uncaused cause, the experience that one could have chosen otherwise in an unconditional sense, the experience that one’s choices are not causally necessitated, etc.
Libertarianism, DL, and EL each can be endorsed or denied independently of the other, though they often hang together. One can endorse libertarianism as a metaphysical thesis while holding that either or both of DL and EL are false. In contrast, one can deny libertarianism while holding that either or both of DL and EL are true. Or one could accept or reject all three.
3. How do these states support libertarianism?
Just as the mental states used in premise (1) of psychological arguments have tended to fall into two categories, the rationale for believing in libertarianism in premise
(2) has generally taken two forms. The first form involves the adoption of conservative
10
epistemic principles (Swinburne, 2013; Mawson, 2011). According to these principles, if we find ourselves in the psychological state of believing P or having an experience as of
P, then, merely on the basis of that psychological state, we have some reason to think that
P is true. Descartes’ view concerning clear and distinct ideas is an example of such an
epistemic principle. According to Descartes, it is a general rule that, “…whatever I
perceive very clearly and distinctly is true.” (CSM II, 24). Since Descartes thinks that we
have clear and distinct perceptions of our own free will, he thinks it must be true that we
have free will. While the Cartesian doctrine of clear and distinct ideas is not very popular
in modern epistemology, some contemporary psychological arguments attempt to
motivate and defend independent conservative epistemic principles in place of clear and
distinct ideas, though the arguments remain Cartesian in spirit. For example, Richard
Swinburne adopts the principle of credulity in his defense of libertarianism, according to
which, “…what seems to us to be so probably is so...” (Swinburne, 2013, 42). T.J.
Mawson similarly argues that, in the absence of strong reasons to doubt widespread
beliefs about libertarianism, we should treat them as true on the basis of our convictions
(Mawson, 2011, 159-168). Using these sorts of principles, they argue, for example, that
widespread beliefs regarding libertarianism are true beliefs regarding libertarianism.
The second contemporary form of the psychological argument relies on a form of
inference to the best explanation (IBE) (O’Connor, 1995; Pink, 2004; Searle, 2007). IBE
involves taking the fact that a theory best explains a phenomenon to count in favor of the
truth of the theory (Harman, 1965; Lipton, 2004). On these views, libertarianism
explains the apparent fact that the psychological state Ψ is widespread, in part by making
11
that psychological state veridical. Further, the fact that Ψ is widespread cannot be explained as well by alternative views which hold that Ψ is not veridical, such as those defended in Smilansky (2000) or Nichols (2015). Thus, on the basis of an IBE, we have reason to think that widespread experiences as of being libertarian free are veridical.
4. Looking Backward and Forward
Using the distinctions we have drawn here, we can present a menu of options for running a psychological argument for libertarianism:
Psychological Endorsement Psychological of Libertarianism Arguments
Doxastic Experiential DL vs. EL Formulation Formulation
Cartesian Cartesian IBE IBE Doxastic Doxastic Experiential Experiential Epistemic Support Argument Argument Argument Argument
Figure 1- A Menu of Options
The resulting four arguments, though not exhaustive, correspond to four prominent ways of interpreting psychological arguments for libertarianism. The Cartesian Doxastic
Argument adopts conservative epistemic principles to argue that apparently widespread beliefs regarding libertarianism are probably true, while the Cartesian Experiential
Argument adopts conservative epistemic principles to argue that apparently widespread experiences regarding libertarianism are probably veridical. Similarly, the IBE Doxastic
Argument holds that libertarianism is probably true because it best explains apparently widespread beliefs in libertarianism, while the IBE Experiential Argument holds that 12
libertarianism is probably true because it best explains apparently widespread experiences
of libertarianism. The dissertation consists of three free-standing papers which are united
as part of a larger program to take seriously some of the more popular recent
formulations of these psychological arguments.
The Doxastic Cartesian Argument has received perhaps the most sustained recent defense of the above arguments. In the first paper, I consider various formulations of the conservative epistemic principle deployed in its support. Using Richard Swinburne’s work as a template, I argue that numerous attempts to formulate his (admittedly vague) principle of credulity fail to support a compelling argument. More generally, the epistemic principle needed for a successful Cartesian Doxastic Argument faces a dilemma: the principle must either be formulated in such a way that it supports libertarianism, but also leads to an unacceptable bootstrapping of belief; or it must be weakened enough to avoid bootstrapping, in which case it does not appear strong enough to support an argument for libertarianism.
In the second paper, I focus on the other most popular recent psychological
argument: the IBE Experiential Argument. I argue that appeals to “appearances of
freedom” are best understood as examples of an IBE. When the argument is
appropriately fleshed out, it has the surprising result that current libertarian theories of
free will may be preferable to current non-libertarian theories as an explanation of widespread appearances of freedom. I argue that this puts greater pressure on non- libertarians to provide an alternative explanation of appearances of freedom. The closest non-libertarian alternatives for the purposes of an IBE attempt to explain widespread
13
beliefs, rather than appearances. In particular, they have focused on the belief that choice
is indeterministic. As such, non-libertarian alternatives are best understood as providing a response to the IBE Doxastic Argument, but have yet to provide an adequate response to the IBE Experiential Argument.
In the third paper, I focus on the above mentioned non-libertarian alternatives. I argue that even the best non-libertarian explanations of widespread belief in indeterministic choice fail to meet their own conditions of adequacy. The prominent non- libertarian strategy harkens back at least to Spinoza. According to this Spinozan strategy, people draw an inference from the absence of experiences of determined choice, to the belief in indeterministic choice. Accounts of this kind are historically liable to overgeneralization. The pair of accounts defended in Shaun Nichols’ recent book,
Bound: Essays on Free Will and Responsibility, are the most complete and empirically plausible representatives of this strategy. I argue that both accounts ultimately fail, and overgeneralization worries remain for Spinozan strategists.
In the appendix, I begin an exploration of an alternative program for non- libertarian explanations of the apparently widespread belief that choices are made indeterministically. It starts with the guiding idea that, within the context of deliberation, people tend to conceive of their choices in terms of the reasons that rationalize them, rather than in terms of the prior causes that bring them about. Since people tend to conceive of these choices in terms that are not causal at all, they come to believe that these choices are not subject to the same causal conditions as other phenomena. I explore the strengths and benefits of this line of explanation, while also noting its shortcomings.
14
Chapter 1: The Cartesian Doxastic Argument
It has long been held among philosophers that there exists a widespread belief or
conviction that humans have free will. After trying unsuccessfully to persuade Frans
Burman of his view on free will, Descartes famously argued that all people could uncover a direct, psychological conviction in free will upon close inspection.
Let everyone just go down deep into himself and find out whether or not
he has a perfect and absolute will, and whether he can conceive of
anything which surpasses him in freedom of the will. I am sure everyone
will find that it is as I say… Now we have inner consciousness of our
freedom, and we know we can withhold our assent when we wish.
(CSMK, 342).
The above argument relies on Descartes’ views about clear and distinct ideas. According to Descartes, it is a general rule that whatever is seen very clearly and distinctly is true.4
Since we have a clear and distinct idea of our own freedom, it must be the case that we actually are free.
Although few contemporary thinkers would adopt Descartes’ doctrine of clear and distinct ideas, there is a family of contemporary arguments for free will that appear
4 “I now seem to be able to lay it down as a general rule that whatever I perceive very clearly and distinctly is true.” (CSM II, 24) 15
strikingly Cartesian in spirit. Call this kind of argument for free will the Cartesian
Doxastic Argument. According to the Cartesian Doxastic Argument:
i. There is a widespread intuitive belief that we have free will.
ii. There is some basic epistemic principle according to which our intuitive
beliefs are probably true.
iii. Therefore, it is probably true that we have free will.
The Cartesian Doxastic Argument purports to show that, independent of other philosophical reflection and argumentation, we have good reason to think we probably have free will, merely on the basis of its widespread belief. Paraphrasing Descartes, any person can look inside themselves, find that they believe in free will, and thereby have reason to think the belief is probably true.
The Cartesian Doxastic Argument can be found in the background of numerous philosophical discussions of free will (Reid, 1788/2010; Campbell, 1967; Ginet, 1997;
Mawson, 2011; Swinburne, 2013). Perhaps surprisingly, the first premise has received general acceptance from philosophers throughout history, including those who wish to deny that our beliefs are accurate, as when Thomas Hobbes elaborately compares humans to a sentient spinning top:
A woodden Top that is lasht by the Boyes, and runs about sometimes to
one Wall, sometimes to another, somtimes spinning, sometimes hitting
men on the shins, if it were sensible of its own motion, would think it
proceeded from its own Will, unless it felt what lasht it. And is a man any
wiser, when he runns to one place for a Benefice, to another for a Bargain,
16
and troubles the world with writing errors, and requiring answers, because
he thinks he doth it without other cause than his own Will, and seeth not
what are the lashings that cause his Will? (Hobbes, 41)
Recent empirical work also supports the claim that there is a widespread belief in free
will (Sarkissian et. al. 2010; Nichols 2012; Deery et. al. 2013; Deery 2015). As a result,
recent presentations of the Cartesian Doxastic Argument have focused on articulating and
defending the epistemic principle needed to fill out premise (ii). Although there are
numerous candidates for premise (ii), the epistemic principles invoked usually involve
some kind of doxastic conservatism.5 According to doxastic conservatives, beliefs we already hold enjoy some default justification merely in virtue of being believed
(Chisholm, 1980; McCain, 2008).
In this paper, I focus on one recent (and relatively clear) presentation of the
Cartesian Doxastic Argument forwarded by Richard Swinburne (2013). Swinburne’s
doxastic conservatism takes the form of an a priori epistemic principle he calls the
principle of credulity. According to the principle of credulity, beliefs are probably true
merely on the basis that they are believed. Swinburne uses the principle of credulity to
argue that we ought to embrace libertarian free will. Since there is apparently widespread
intuitive belief in libertarian free will, then according to the principle of credulity, it is
probably true that we have libertarian free will.
I argue that the principle of credulity endorsed by Swinburne is untenable.
Briefly, the principle as formulated leads to an unacceptable bootstrapping of belief; if
5 The position I am calling doxastic conservatism is sometimes called epistemic conservatism. 17 one can bring oneself to hold a particular belief, regardless of the reasons or means of belief formation, then that belief can be held justifiably merely in virtue of my believing it. I then consider three potential revisions to the principle of credulity designed to avoid such bootstrapping. These revisions either fail to avoid bootstrapping, or avoid bootstrapping but are too weak to support a compelling, independent argument for libertarian free will. Because of these problems, I conclude that, if the Cartesian
Doxastic Argument relies on some form of doxastic conservatism, it will likely fail to provide an independent argument for libertarian free will.
The paper proceeds as follows. In section 1, I present doxastic conservatism as a general epistemic position and the traditional motivations for accepting it. In section 2, I show how Swinburne uses doxastic conservatism in the form of his principle of credulity to provide a Cartesian Doxastic Argument for agent-causal libertarianism. In section 3, I argue that the principle of credulity as presented faces unacceptable bootstrapping worries. In section 4, I consider three alternative formulations of the principle of credulity and argue that each fails to support a compelling, independent Doxastic
Cartesian Argument. I conclude in section 5 that, if the Cartesian Doxastic Argument relies on doxastic conservatism, it is likely to fail.
1. Doxastic Conservatism
Suppose that you find yourself consciously considering some belief, say, that there will be a new episode of The Flash on television tonight. You rack your brain to try to figure out where the belief came from but to no avail. You can’t even be sure when the belief came to mind, but it certainly seems to you at this point that there will be a new
18
episode tonight. You have no good evidence suggesting that there isn’t a new episode
tonight (aside from the fact that you believe there is an episode tonight). Furthermore
you have no good evidence to think that the belief was formed in some epistemically
illegitimate way. You then ask yourself, “Is it reasonable to continue to believe that there
will be a new episode of The Flash on TV tonight? Should I withhold judgment? Should
I believe that there is not going to be a new episode?” According to the doxastic
conservative, the reasonable thing to do is to continue to believe that there will be a new
episode. It is reasonable to continue to believe because the thing believed is more likely
than not to be true, in virtue of the fact that you believe it.
Roderick Chisholm gives voice to the idea that continued belief is reasonable
when he writes, “…anything we find ourselves believing may be said to have some
presumption in its favor– provided it is not explicitly contradicted by the set of other things we believe” (1980, 551-552). Doxastic conservatism holds that this intuition reflects a basic, a priori epistemic principle (Chisholm, 1980; Swinburne, 2001; McCain,
2008). According to doxastic conservatism, a person’s merely believing that P gives that person some justification for P, provided the person has no grounds for doubting that belief (Swinburne, 2001, 144). On this view, if I don’t have independent reason to think there will not be a new episode of The Flash tonight, then, finding myself believing that there will be a new episode, I should see the belief in a positive epistemic light.6
Unfortunately, it is far beyond the scope of the current project to resolve disputes within epistemology about the precise meanings of terms such as “evidence” and
6 This understanding of doxastic conservatism may have interesting ties to conservatism in the context of theory choice in science. Lycan (2013) 19
“justification” that are needed to clarify the total import of doxastic conservative principles. Let us say for present purposes that one has justification to believe that P when one is in a position where it would be epistemically appropriate for one to believe that P and to rely on that belief in forming other beliefs (Pryor, 2005). To be epistemically appropriate just means that believing that P follows generally accepted
epistemic standards on belief formation. Furthermore, when discussing evidence, let us
restrict ourselves for the time being (as Swinburne does) to the considerations to which
one has mental access and on which one’s beliefs are actually based.7
Proponents of doxastic conservatism argue that without it, epistemic projects are
non-starters. Swinburne, for example, writes, “Either we must be very sceptical, and
think that almost all our ordinary beliefs are not probably true, or we must endorse the
principle of credulity. But we cannot choose our beliefs at will.” (2013, 44). Any time
that we want to evaluate the truth of our beliefs, Swinburne thinks, we must make
recourse to other evidence. In order to do so, we must consult out own memory. So if
we cannot trust our own memories, we cannot begin to form any justified beliefs about
anything (Swinburne, 2013, 43). Without some form of doxastic conservatism,
proponents like Swinburne argue that we are quickly forced into radical skepticism.
The intuitive plausibility of doxastic conservatism gains further traction when we
consider the fact that in many cases we forget why it seemed justified to form a given
belief in the first place (Harman, 1986; Huemer, 2001; McCain, 2008). Suppose I
7 I further follow Smithies (2016) in using the terms justified, reasonable, and rational interchangeably. Finally, unless otherwise noted, by justification in this paper I mean doxastic justification, the justification to which the believer has access and on which she actually bases the belief. 20
remember that Columbus is the capital of Ohio, but have forgotten that I formed this belief long ago when I visited the state capitol building in downtown Columbus.
According to doxastic conservatism, the inaccessibility of the original justification for
this belief should not prevent me from thinking that my belief is justified. Instead, the
mere fact that I already believe that Columbus is the capital of Ohio provides justification
for the belief. More generally, there is no need to worry about the justificatory status of
beliefs of which I have long since forgotten the origins. Call these sorts of examples
cases of forgotten evidence.
Cases of forgotten evidence are surprisingly common. Many of our everyday
beliefs are held on grounds for which we have long since forgotten the particulars.
According to proponents of doxastic conservatism, it is not only reasonable to continue to
hold these basic beliefs; stronger, they argue that no philosopher should be willing to
reject them. To reject such basic beliefs threatens to give up the philosophical game
more generally and marks a quick slide towards skepticism. If I can’t take my already
held beliefs at face value, they ask, how can I even start to make epistemic headway?
With these motivations in mind, I turn now to Swinburne’s presentation of the Cartesian
Doxastic Argument.
2. The Principle of Credulity & the Cartesian Doxastic Argument
Swinburne’s argument for the existence of libertarian free will depends crucially
on a form of doxastic conservatism. Swinburne presents the principle of credulity as
follows:
21
…any basic belief (that is, the content of that basic belief, the
proposition believed) is probably true (that is, it is more probable
than not that the belief is true) on the believer’s evidence that he
believes it – in the absence of evidence in the form of other basic
beliefs of that believer which makes it probable that he is mistaken.
(Swinburne, 2013, 42).
While Swinburne is refreshingly clear that the principle of credulity is the foundation for
his free will argument, the commitments of the principle are not always clear. As a first
pass, the principle of credulity seems to suggest that the fact that I hold a belief should be
counted as evidence in favor of the truth of the propositional content of that very belief,
in certain contexts. More specifically, the principle only applies to so-called “basic
beliefs.” Basic beliefs are beliefs regarding “…those propositions which seem to us to be
true but not on the basis of other propositions…” Swinburne (2013, 41).8 As examples,
he cites perceptual beliefs, memory beliefs, relatively simple beliefs about arithmetic, and
general beliefs regarding history and geography where we can’t remember why we hold
these beliefs (Swinburne, 2013, 43-44). In order to begin an epistemic journey,
according to Swinburne, we need to be able to rely on these basic beliefs in our further
reasoning, lest we slide into extreme skepticism. So in the case of a basic belief that P where I do not have any evidence supporting P, beyond the fact that I already believe that
P, the principle seems to suggest that I am justified in forming the further belief that P is
8 Although Swinburne presents the principle in terms of basic beliefs, he also seems to apply the principle fairly liberally to experiences, seemings, memory processes, and other mental phenomena. Here, I assume the principle applies only to basic beliefs, and reference to these other mental phenomena is short hand for the basic beliefs they give rise to, as when he writes, “[I]t seems to me (or appears) to me that I am seeing a desk in front of me (that is, I find myself believing that I am seeing a desk)…” (Swinburne 2013, 42). 22
probably true, absent counter-evidence (Swinburne 2013, 43). Notice that this does not
necessarily mean that for every belief we hold, we should form the further belief that the
original thing believed is probably true. However, in the situation where we are
evaluating our own beliefs, we would be justified in holding that our basic beliefs are
probably true.
It is not always clear from Swinburne’s presentation what it means for a proposition to be “probably true.” He sometimes uses the phrase interchangeably with
“more likely than not,” suggesting that at minimum, the probability that the proposition is true is greater than 50%. In an additional note, he writes that he is interested in epistemic
probability, the probability of one proposition on another using correct inductive criteria,
which he contrasts with statistical and natural probabilities. A proposition p has an
epistemic probability of 1 on another proposition q iff q makes p certainly true, and an
epistemic probability of 0 on q iff q makes p certainly false (Swinburne 2013, 230). This
in turn suggests that the basic belief p has an epistemic probability of greater than 50% of
being true merely because of the principle of credulity. This probability is defeasible;
subsequent evidence may be brought to show that the belief is probably not true.
However, merely on the basis that a proposition is believed, and in the absence of such
further counter evidence, we should judge the original belief probably true. Or so the
story goes.
With these clarifications in mind, I present the following as an attempt to
succinctly formulate of the principle of credulity.
23
(PoC) When reflected upon, the content of a basic belief that P should be
judged probably true merely on the basis that one believes it, absent
counter-evidence.
The principle of credulity is at the core of Swinburne’s defense of free will. He
argues that we have basic beliefs about our own free will, which are probably true on the
basis of the principle of credulity, and so we are justified in believing that we probably
have free will. Specifically, he defends a libertarian view of free will, wherein we
exercise free will when we act intentionally without our intentions being fully determined
by prior causes (Swinburne, 2013, 202). Swinburne describes the scenario as follows:
[I]n those circumstances where desires and moral beliefs are in opposition
… we are conscious of deciding between competing alternatives. We then
believe that it is up to us what to do and we make a decision. Otherwise,
we allow ourselves to do as desires and moral beliefs dictate – which is so
often just to conform to habit. The principle of credulity says that things
are probably the way they seem to be in the absence of counter-evidence.
So we should believe that in these circumstances where we believe we are
making a choice without being caused to choose as we do, that we are
indeed doing just this. (Swinburne, 2013, 201).
Swinburne contrasts exercising a power to choose with merely acting out of habit, the latter of which is causally determined by prior states. Since we are free when our choices are “up to us”, and acting out of habit is not “up to us” because it is determined by prior states, it follows that we are not free when we are determined by prior states. Upon
24
reflection, the fact that we believe our choices are uncaused supports the belief that determinism is false. Since we believe that our choices are not determined, and further that they are not merely the result of random mechanisms, we have reason to believe that our choices are up to us; we have reason to believe humans have free will. In virtue of this cluster of basic beliefs revealed through reflection on our own choosing, according to
Swinburne, we have beliefs that imply commitment to libertarian free will.
Using the PoC and Swinburne’s account of our beliefs about our own choices, we can reconstruct the following argument for libertarianism:
Cartesian Doxastic Argument 1
I. There is widespread intuitive belief in libertarianism.
II. (PoC) When reflected upon, the content of a basic belief that P should be judged probably true merely on the basis that one believes it, absent counter-evidence.
III. Therefore, libertarianism is probably true.
The argument relies on the controversial suppressed premise that there is no counter-
evidence to intuitive beliefs that imply free will. I will return to the question of counter-
evidence at the end of the paper, but for now, I grant the suppressed premise in order to
focus on the plausibility of the principle of credulity.
There are a few features worth noting about Swinburne’s presentation of the
Doxastic Cartesian Argument. First, its success does not depend on a particular source/
mechanism that gives rise to the belief in libertarian free will. So long as the belief is
widespread and intuitive, it falls under the purview of the PoC, regardless of the
particular process by which a given individual came to form the belief. Second, the
25
argument purports to provide independent justification for believing in libertarianism.
While more traditional philosophical arguments may strengthen one’s justification for
believing that libertarianism is probably true, the success of the Doxastic Cartesian
Argument does not depend on those philosophical arguments. Thus, the argument can
largely bypass the murky waters of more traditional free will disputes. Finally,
Swinburne’s presentation of the Cartesian Doxastic Argument makes explicit an
argument to which philosophers and non-philosophers often seem implicitly committed;
paraphrasing Samuel Johnson, “We know our will is free, and there’s an end to it!”
According to the Doxastic Cartesian argument, this is not merely a slogan. It is an
independent justification for believing in libertarian free will.
Swinburne’s Cartesian Doxastic Argument relies exclusively on the PoC. As
such, the ultimate success or failure of the argument hinges on the acceptability of the
PoC. In the next section, I argue that, despite what Swinburne and other doxastic conservatives would have us believe, the PoC is not an epistemically modest, skepticism-
blocking principle. Rather, it is a substantial principle with substantial commitments, the
acceptance of which leads to unacceptable consequences. In short, we should reject the
PoC, and so Swinburne’s Doxastic Cartesian Argument fails.
3. The PoC and the Bootstrapping Problem
In the context of the Cartesian Doxastic Argument, the PoC is not an inference
rule that merely codifies typically observed inference patterns. That is, it is not merely
the empirical observation that when there is no introspectively accessible evidence to support a given belief that P, people nevertheless tend to believe that the belief that P is
26
more probably true than not. Rather, the PoC is a substantive, though defeasible,
commitment that enters as a premise in inferential arguments. As such, it must be
defended.
One reason we might accept the PoC is that, generally speaking, we are reliable
belief forming machines. Since we generally form correct beliefs, if we find ourselves
believing something, it is probably a true belief. However, Swinburne argues that we
should not accept the PoC in virtue of our past successes as belief formers (though he
does think that, generally, we get things right more often than not). As mentioned
previously, Swinburne argues that we should accept the PoC in its unrestricted form in
order to deal with cases of forgotten evidence and to avoid skepticism. He thinks this
requires that the principle of credulity be an a priori epistemic principle. So, according
to Swinburne, while our belief that we are generally reliable belief formers may provide additional justification for the truth of our basic beliefs, it requires first that we accept the
PoC as an a priori epistemic principle.
These potential gains from accepting the PoC need to be balanced against other epistemological values. If the PoC is too epistemically permissive or restrictive, licensing unacceptable beliefs and preventing acceptable beliefs, then it should be
rejected. In what follows, I argue that anyone wishing to deploy the PoC faces either of
two problems, depending on what they include among the counter-evidence that bears on a given belief. Either the PoC is too strong and so licenses believing that intuitively unjustified beliefs are more probably true than not, or there do not exist any cases where
27
there is no evidence that bears on the truth of P beyond the fact that P is believed, in
which case the PoC is an empty principle.
The first problem for the PoC is that it is too liberal. According to the PoC, I
should believe that P is more likely true then not, merely on the basis that I already
believe it. Since merely holding some belief justifies that same belief, regardless of the
source and status of the belief, the PoC licenses unacceptable bootstrapping of beliefs.9
Drawing on an example from Foley (1983), suppose that I come to believe that there is an even number of grains of sand on Virginia Beach. Given my evidence, I have no reason to think there is not an even number of grains of sand on Virginia Beach, and so there is no counter-evidence to my even-belief. So, according to the PoC, I should judge that it is probably true that there is an even number of grains of sand on Virginia Beach, even if I can’t come across a single piece of evidence in favor of the claim (aside from the fact that
I believe it, that is).
Intuitively, this consequence is unacceptable. While it is true that I have no reason to think that there is not an even number of grains of sand on Virginia Beach, and so there is no counter-evidence to my even-belief, I also have no reason to think that there is an even number of grains of sand on Virginia Beach. The worry for the PoC is that the absence of counter-evidence does not yet give me a reason to prefer an even- belief to an odd-belief regarding the number of grains. Yet the PoC holds that my even- belief is “probably true.” So, if one accepts the PoC, then intuitively unjustified beliefs would turn out justified. In short: the PoC is too liberal.
9 This argument expands on an argument originally found in Foley (1983) against Roderick Chisholm’s epistemic conservatism. A similar argument can also be found in Lycan (2013) 28
It may be objected that the Virginia Beach example does not show that the PoC is too liberal, since the fact that I “just came to believe” that there is an even number of grains of sand on the beach would itself constitute counter-evidence to my belief. More
generally, the objection goes, if I have no access to the process by which a belief was
formed before it suddenly popped into my head, I have some reason to think that the
belief is not on epistemic good footing. Beliefs that just pop into one’s head for no
apparent reason tend to be arbitrary if they aren’t held on the basis of memory, or
inference, or some other reputable source. The “popping in” of the even-belief about the
number of grains of sand at Virginia Beach supports the contention that the PoC should
not apply, and so I should not believe that it is more probably true than not that there is an
even number of grains of sand at Virginia Beach. Or so the objection goes.
The above objection, when considered seriously, introduces the second problem
for the PoC. The problem is that, if we include among the evidence bearing on the truth
of P, further beliefs about the processes leading to the formation of the belief that P,
those further beliefs (or lack thereof) block the meaningful application of the PoC.
To illustrate this point, return to the original Virginia Beach example. As
formulated, the PoC holds that P is probably true in cases where the only evidence in
favor of P is the fact that P is believed. The Virginia Beach example was meant to
provide just such a case. The objection challenged this presentation by arguing that,
despite appearances, there is available counter-evidence, in the form of other beliefs
regarding the process by which I came to believe that there is an even number of grains at
Virginia Beach. But if beliefs about the opacity of the belief forming process constitutes
29 counter-evidence in this case, then it is entirely unclear how one could ever be in a situation where the only evidence that bears on the truth of P is the fact that one believes it. Either one has evidence that bears on the truth of P beyond the mere fact that P is believed, which includes beliefs about the process by which the belief that P was formed, in which case the PoC should not be applied; or one does not have evidence that bears on the truth of P, which includes a lack of any beliefs about the process by which the belief was formed, in which case the PoC should not be applied on the grounds that the lack of beliefs regarding the process is itself evidence that the belief was not formed in an epistemically legitimate way. Either way, the PoC would fail to support the belief that P.
One might try to avoid the above dilemma by denying that beliefs about the process by which one comes to believe that P is to be included among the evidence that bears on the truth of P. But then one falls back into the original Virginia Beach problem, and the boot-strapping worry re-arises. Another attempt to avoid the dilemma argues that, in virtue of my “good track record” as a belief forming machine, I should be fairly optimistic about the truth of the propositions I find myself believing, even if the particular process by which a given belief was formed in the first place is not immediately apparent. While this may be true, it is not a move which supporters of the
PoC like Swinburne should be willing to make. The PoC blocks skepticism because it confers justification in cases where there would otherwise be no evidence that bears on the matter. But for Swinburne, my good track record only supports the truth of P if I believe I am a reliable belief former. In which case, I have evidence that bears on the
30
truth of P. If the PoC always depends on additional evidence of this kind, then it cannot
provide the foundational justification needed to block the slide to skepticism.
Hopefully, I have said enough here to think that the PoC as formulated by
Swinburne ought to be rejected.10 If the PoC fails, then Swinburne’s presentation of the
Cartesian Doxastic Argument fails as well. We ought not to think that libertarianism is probably true merely on the basis that there is widespread intuitive belief in it.
4. Reformulating the Principle of Credulity
The PoC is a particularly strong principle, as it renders beliefs for which we have no additional evidence probably true. In the previous section, I argued that we should not accept the PoC. Still, it may be possible to adopt an alternative, weaker form of doxastic conservatism that avoids the pitfalls of the PoC while preserving the Cartesian Doxastic
Argument. Consider the following weaker form of doxastic conservatism, based on
Chisholm’s (1980) defense of doxastic conservatism:11
(DC1) For every x, if x accepts a proposition P that is not explicitly
contradicted by any set of propositions accepted by x, then there is some
presumption in favor of the truth of P.
DC1 is much more modest than the PoC. Unlike the PoC, which holds that a belief is
“probably true” merely on the basis that it is believed, DC1 holds merely that there is
“some presumption in favor” of the truth of the proposition believed. That is, beliefs that
we already hold have a default positive epistemic light; there is a small amount of
10 For additional criticisms of an earlier formulation of the PoC, see Martin (1986). 11 This is a version of Chisholm’s principle P3, which reads: “For every x, if x accepts a proposition that is not explicitly contradicted by any set of propositions accepted by x, then that proposition has some presumption in its favor.” (Chisholm 1980, 552). 31
justification for judging that P is true merely on the basis that P is believed, but this does
not necessarily render P probably true. Like the PoC, DC1 is a defeasible principle. If I
have counter-evidence, in the form of other propositions that I accept that would contradict P, then the support offered by DC1 is defeated.
One attractive feature of DC1 is that, since there can be some presumption in favor of a proposition without the proposition believed being probably true, it has an apparent response to the bootstrapping objection. For example, while the fact that I believe that there is an even number of grains of sand on Virginia Beach may provide minimal pro tanto reason to think that the proposition believed is true, it might not cross the threshold of being “probably true”. As a result, DC1 may be able to avoid the bootstrapping problem, since additional evidence would be needed before P is rendered probably true in virtue of DC1.
Although DC1 may permit bootstrapping in fewer cases than PoC, it still permits bootstrapping in some cases. Presumably, if I have equal evidence for two propositions P and ~P, then I ought to withhold believing either proposition. However, if DC1 is accepted, then somehow bringing myself to believe that P also provides an additional presumption in favor of the truth of P. Thus, by coming to believe that P, I also make the belief in P (as opposed to ~P) justified merely in virtue of believing it, if DC1 is accepted
(Foley 1983; Christensen 1994).
There is a further worry for DC1. It is too weak to be inserted into the Cartesian
Doxastic Argument. The conclusion of that argument is that libertarianism is probably true. At best, DC1 can only support the weaker claim that widespread belief in free will
32
provides some presumption in favor of libertarianism. This suggests the following
modified version of the Doxastic Cartesian Argument.
Cartesian Doxastic Argument 2
I. There is widespread intuitive belief in libertarianism.
II. (DC1) For every x, if x accepts a proposition P that is not explicitly contradicted by any set of propositions accepted by x, then P has some presumption in its favor.
III. Therefore, there is some presumption in favor of libertarianism.
On this weaker argument, it remains open to non-libertarians to agree that there is some
presumption in favor of libertarianism merely in virtue of being believed by x, while also
arguing that we should not think that libertarianism is probably true merely in virtue of
being believed. In which case, the Doxastic Cartesian Argument is an argument for
libertarianism only if the small presumption in favor of libertarianism provided by DC1
crosses some “more-likely-true-than-not” threshold.12
For example, the small presumption in favor of libertarianism provided by DC1
could be relevant in the context of an inference to the best explanation argument.
Suppose that we are considering two hypotheses about human decision-making. H1 is
the hypothesis that we have libertarian free will and H2 is the hypothesis that we do not
have libertarian free will. Suppose further that these two hypotheses are each equally
12 It might be the case that the presumption in favor of the proposition believed provided by DC1 is proportional to the confidence with which the belief is held. For example, if I hold a belief with very little confidence, then it is only weakly justified in virtue of my so believing. But a confident belief might have a strong presumption in its favor. It is difficult to see how one could be more confident in a belief without that confidence being related in some way to the strength of the evidence in favor of the proposition believed. Still, it is worth considering this possibility, particularly if my confidence could push the small presumption in favor of a belief into full-blown justification. 33
good at explaining the data; both make equally accurate predictions, both are equally
simple, fecund, etc., and both have some evidence in support of their truth. If H1 is
closer to what we already believe, and one accepts DC1, then we might have some reason
to slightly prefer H1 to H2. In this way, DC1 could support preferring libertarianism
over non-libertarian alternatives, but only on the assumption that there is already some
reason for thinking both H1 and H2 are justified theories on the basis of other evidence.13
In the imagined inference to the best explanation from the previous paragraph, we
already possess independent reasons for thinking that each of H1 and H2 could be true.
Without these independent reasons, DC1 does not provide sufficient support to cross the
“more-likely-true-than-not” threshold, since the apparently widespread belief in
libertarian freedom only provides a small presumption in favor of the truth of
libertarianism. Because of this dependence on other reasons in support of libertarianism,
the Cartesian Doxastic Argument based on DC1 loses much of its original appeal. While
DC1 may provide a way of breaking the stalemate in choosing between two otherwise
equally good hypotheses, it cannot ground an independent argument for libertarian free
will.
The above discussion reveals a serious tension in the Cartesian Doxastic
Argument. It requires a doxastic conservative principle that is strong enough to support
an argument for libertarian free will that is independent of other sources of justification,
while also remaining weak enough to avoid bootstrapping worries. This is an extremely
tall order, and in what follows, I argue that two other attempts to revise the doxastic
13 See my paper “On Libertarianism as an Explanatory Hypothesis” for more discussion about IBE arguments for libertarianism. 34
conservative principle fails to meet both criteria. Consider the following revision of
DC1:
(DC2) for every x, if x accepts a proposition P and there is no counter-
evidence to P in the form of other propositions accepted by x, then it is
epistemically appropriate to believe that P.
DC2 strengthens DC1 by replacing “presumption in favor of the truth of P” with the definition of justification presented earlier; that one has justification to believe that P when one is in a position where it would be epistemically appropriate for one to believe that P. The hope is that this strengthening will support an independent argument for libertarian free will.
At the same time, DC2 weakens DC1 by changing the defeating condition from
“explicit contradiction” to “counter-evidence.” Because of this change, evidence other than explicit contradiction can block the applicability of DC2. For example, suppose I believe that 99% of my beliefs are false. While this does not explicitly contradict P, it might still give me good reason to think that believing that P is not epistemically appropriate. The hope is that this weakening will avoid the bootstrapping worries that plagued DC1 and the PoC.
The attempt is unsuccessful. DC2 allows for the same problematic boot strapping regarding belief that affected the PoC and DC1. If I can bring myself to believe something, then merely in virtue of the fact that I believe it, the belief is justified.
Suppose that I have some evidence in favor of believing that P, but not yet enough to claim that I would be justified in believing that P. Nevertheless, if I can bring myself to
35 believe that P for independent reasons (say, wishful thinking, or a belief forming pill), the fact that I now believe that P itself provides some reason to believe that P, perhaps enough that when combined with my previous evidence, I am now justified in believing that P.
DC2 does not just permit this kind of bootstrapping; it might encourage it. Good epistemic practice involves, at least, forming justified beliefs. According to DC2, taking belief-forming pills can result in justified beliefs. So, if DC2 is correct, then engaging in wishful thinking or taking a belief-forming pill would be good epistemic practice, even if
I already have evidence against believing that P. To illustrate, suppose that I want to believe in the existence of God, but I think that my best evidence supports the conclusion that he does not exist. A close friend presents me with the opportunity to take a pill.
This pill will form in me the belief that God exists, while wiping away all of my memories of taking the pill and the previous evidence I saw as supporting the conclusion that God does not exist. According to DC2, my pill-based belief in God would be justified, so if I ought to form justified beliefs, then I ought to take the pill. This unacceptable result goes against two intuitions. First, a belief formed merely on the basis of a pill is not justified, as the pill does not provide the appropriate reasons to justify the belief. Second, since a belief based on a pill would not be justified, it is bad epistemic practice to intentionally take a pill that wipes away my prior evidence and forms a new belief. Accepting DC2 rejects both of these intuitions.
Some doxastic conservatives may be unmoved by the discussion in the previous paragraph. There is nothing about DC2, they might argue, that says that one ought to
36 engage in bad epistemic practice. The doxastic conservative can respond that in the case where one engages in bad epistemic practice, then one is criticizable on those grounds, while also holding that the belief that one forms on the basis of evidence gathered in an illegitimate way comes out justified. To illustrate, suppose that I have a systematic bias in my evidential gathering system: I ignore every third fact of which I am made aware.
Because of this flaw, a great deal of important information is lost whenever I form my beliefs. In gathering evidence in this way, I engage in bad epistemic practice.
Nevertheless, it may still seem right to say that my beliefs are justified, given that I only have partial information. In short: we should avoid bad epistemic practice, but beliefs made on the basis of bad epistemic practice may nevertheless be justified.
This move attempts to drive a wedge between what constitutes good epistemic practice on the one hand, and justified beliefs on the other. DC2, however, relies on a view where justification is cashed out in terms of what is epistemically appropriate, where that means that believing that P follows generally accepted epistemic standards on belief formation. In this sense, to be justified according to DC2, just is, at least in part, to engage in good epistemic practice. So DC2 avoids neither the bootstrapping objection nor the charge of licensing bad epistemic practices.
According to DC2, in the absence of good grounds for doubting P, we are justified in believing P. But this sort of reasoning involves the fallacy of ad ignoratiam
(Foley, 1983, 175). We can safely interpret the absence of grounds for doubting P as a lack of evidence for thinking that ~P is true. But a lack of evidence for ~P is not the same as evidence that P is true. One commits a similar sort of fallacy when they find
37 there is no counter-evidence to the truth of the proposition that P and so hold that the belief that P is rational.
In normal cases, the fact that there is no evidence to be found for ~P only constitutes a reason for thinking P is justified if there are further background assumptions such that the lack of evidence of ~P constitutes evidence in favor of P. For example, the fact that we haven’t discovered unicorns is evidence for the claim that unicorns don’t exist only because of the background assumption that, were unicorns to exist, they would be in the places we have already looked and found them lacking. But DC2 makes no mention of such principles.
We might try, then, to build the background assumptions into our formulation of doxastic conservatism. That is, we might include a clause that the belief that P is justified on the grounds that I believe it, provided that background assumptions make it such that the absence of counter-evidence gives me reason to judge that P is more probable than not. DC3 below attempts just that:
(DC3) for every x, if x accepts a proposition P and there is no counter-
evidence to P in the form of other propositions accepted by x, then it is
epistemically appropriate to believe that P, provided that background
assumptions imply that the lack of contradictory propositions count in
favor of the belief that P.
Notice that DC3 avoids the bootstrapping worries leveled at DC2, DC1, and the PoC. To see this point, consider once again the Virginia Beach example. I have no idea where the belief came from. These considerations constitute important information when
38
determining whether the belief is epistemically appropriate. It is precisely because I normally do know where my beliefs come from, and because I normally believe that I shouldn’t draw empirical conclusions without first gathering sufficient data, and the fact that the current situation suggests that I have not gathered sufficient data, that it becomes clear that the belief is not epistemically appropriate. The fact that I have no counter- evidence to the belief that there is an even number of grains of sand on Virginia Beach should not count as evidence for the belief that the even-belief is justified, because the background assumptions necessary to imply that the lack of contradictory propositions count in favor of the belief do not obtain. In this way, we can block bootstrapping by appeal to the lack of appropriate background assumptions.
Despite this benefit provided by DC3, it suffers from similar problems to that of the PoC. If the lack of counter-evidence does indeed count in favor of the belief that P, then x is justified on those grounds, and not from the mere fact that P is believed. Thus, it is no longer the case that the belief is justified merely because it is believed and there is no counter-evidence; the belief is justified in virtue of the lack of counter-evidence combined with the background assumption that, were there counter-evidence available, x would be aware of it. In which case, there is no need to wheel in additional a priori principles such as PoC or DC1, DC2, or DC3, since my belief is justified by the evidence already available to me.
5. Conclusion
The difficulties facing DC3 return us once again to the question of whether there is available counter-evidence to the widespread belief that we have libertarian free will.
39
Up to this point, I have granted the point for the sake of argument. It is now worth
considering more carefully. There are, of course, well-worn arguments in the free will debate regarding the plausibility of agent-causal libertarianism, particularly in the face of scientific and naturalist commitments (Strawson, 1986; Searle, 2007; Vargas, 2013, Ch.
2). Settling them all here would be an impossible task.
Swinburne’s general strategy in defense of the “no counter-evidence” claim is to argue that indeterminist theories of quantum physics are consistent with intuitive libertarian beliefs, and that further there is no (in fact, there can be no) evidence in support of the causal closure of physics. Thus, fundamental physics does not constitute counter-evidence to intuitive libertarian beliefs. Furthermore, he argues specifically against the work of Benjamin Libet and Daniel Wegner that sometimes purport to show that we do not have libertarian free will because of the existence of prior physical causes of apparently intentional actions, relying on well known alternative explanations of the data that are consistent with the existence of libertarian free will.
Even if we are inclined to agree with Swinburne that there is no strong reason to prefer non-libertarian over libertarian theories of free will, the claim that there is no counter-evidence to libertarianism is extremely implausible. Swinburne himself acknowledges that counter-evidence may include, “…evidence that other people have basic beliefs inconsistent with [my] basic belief…” (Swinburne, 2013, 44). Surely, hard
determinists, hard incompatibilists, and compatibilists count as people who hold beliefs
inconsistent with his own! Swinburne could allow that these opposing beliefs make his
own intuitive beliefs less plausible, but perhaps not to the extent that the propositions he
40 believes are rendered more probably not true. But he provides no defense for why this should be the case. In which case, even if there is some reformulation of the PoC that I have not considered here that avoids bootstrapping worries, we should be skeptical that it would be applicable to apparently widespread beliefs that we have libertarian free will.
To be clear, I do not think I have put the nail in the coffin for doxastic conservatism, but I hope I have taken some steps to justify thinking that doxastic conservatism misses the mark at its core. Swinburne’s PoC faces unacceptable bootstrapping worries, as do attempts to reformulate the principle in order to preserve the
Cartesian Doxastic Argument. While it of course remains possible to reformulate the
PoC yet again, it currently looks implausible that any formulation can support a plausible, independent argument for accepting agent causal libertarianism.
Furthermore, the discussion of different formulations of the PoC shows that we do not need doxastic conservatism in the first place in order to deal with skepticism or cases of forgotten evidence. Cases of forgotten evidence, for example, can be dealt with by appealing to other background assumptions I already hold. For example, I might think to myself, “I can’t quite remember why I believe this, but the fact that I do believe it is evidence that there was a good reason for my belief at the time. And if there was a good reason for my belief at the time, then the belief must be justified.” Or I might think, “My remembrances are generally pretty justified, so this belief is probably justified too, even though I can’t recall why I believe it.” Both of these explanations are examples of the justified holding of beliefs in virtue of a larger background body of evidence. Neither explanation requires a further, generalized, doxastic conservative principle. When
41 combined with the bootstrapping worries I have argued for here, I see very little reason to accept any generalized, doxastic conservative principle. But without doxastic conservatism, the modern presentations of the Cartesian Doxastic Argument fail.14
14 Although unrestricted forms of doxastic conservatism are problematic, the Cartesian Doxastic Argument could turn on some form of dogmatism. As I understand them, dogmatists are doxastic conservatives who restrict their conservatism to particular domains. For example, perceptual dogmatists think that if one holds a belief on the basis of perception, absent defeaters, then one is justified in holding that belief. The success of these arguments would depend on the particular type of dogmatism, as well as the plausibility that free will beliefs are acquired via the specified domain. 42
Chapter 2: On Libertarianism as an Explanatory Hypothesis
In the context of debates about free will, the oft-quoted Samuel Johnson observed,
“All theory is against the freedom of will; all experience for it.” (Boswell, 1993, 291).
Numerous libertarians (and compatibilists) have been quick to observe that experience speaks in favor of libertarian freedom over compatibilist alternatives. Unfortunately, such appeals to what we might call appearances of freedom tend not to go beyond mere slogans. As such they present as poor arguments (Vargas, 2015, 70). Non-libertarians have interpreted these slogans (rightly, in many cases) as epistemic claims about the status of default beliefs (Nichols, 2015; Vargas, 2015; Smilansky, 2000). On this interpretation, to hold that we appear free is to hold that there is a widespread default belief that we are free. By undermining the epistemic status of these default beliefs, non- libertarians have been able to dismiss libertarian appearance claims out of hand. Thus, non-libertarian philosophers have seen appeals to appearances of freedom as
categorically bad arguments.
I maintain that there is an alternative and more charitable construal of libertarian
appeals to appearances of freedom. In a large swathe of cases, such appeals are best
interpreted as forming the basis for an inference to the best explanation (IBE)
(Swinburne, 2013; Mawson, 2011; Pink, 2004, 2010; O’Connor, 1995; Campbell, 1967;
43
Descartes, CSM II; Reid, 1788). That is, we might charitably interpret these philosophers
as hinting at arguments that conform to the following Libertarian IBE Schema:
The Libertarian IBE Schema
1) There are widespread appearances of freedom in need of explanation.
2) Libertarianism explains these appearances of freedom.
3) There is a range of alternative (both compatibilist and incompatibilist)
explanations of appearances of freedom that do not explain appearances of
freedom as well as libertarianism does.
∴ We have some reason to think libertarianism is true.15
Inference to the best explanation (sometimes referred to as ‘abduction’ or ‘explanatory inference’) occurs when one takes the fact that a hypothesis explains some collection of data better than available alternative hypotheses as reason to infer with some confidence that the hypothesis is true.16 This usually involves consideration of the various
explanatory virtues: conservativeness, elegance, simplicity, fruitfulness, power,
testability, etc. Better explanations exemplify one or more of these virtues to a greater
degree than competing hypotheses.17 The core aim of this paper is to show that non-
15 The framework for this IBE schema is adapted from (Lycan 2002, 413). 16 (Lipton, 2004; Lycan, 2002). In the philosophy of science, there is general debate about whether IBE is ever a justified form of inference, ranging from it never being justified to it being at the core of all forms of inference (Lipton, 2004; Van Fraassen, 1980; Sober, 2013; Harman, 1965). I will assume that IBE arguments as a class are at least sometimes justified, although nothing I say here is meant to commit to a particular view of how IBE works. 17 Ideally, the compared hypotheses should be minimally satisfactory as an explanation of the phenomena. They should predict and account for all of the relevant facts. They should not have obvious lacunas. They 44
libertarians have largely overlooked the fact that, when the above schema is suitably
clarified, it provides prima facie attractive grounds for preferring some form of
libertarianism, other things being equal.
Although IBEs are commonplace in the philosophy of mind, it is unclear how such an argument for libertarianism might be developed. Specifically, there is a range of related issues regarding the scope and nature of the appearances of freedom that, unless addressed carefully, threaten to render the proposed IBE question begging or else entirely otiose. In view of this problem, much of the present discussion will focus on what the appearances of freedom are, and how we must understand “appearance” claims more generally, if they are to provide a basis for a libertarian IBE argument. To be clear, I will not argue here that the libertarian IBE is a decisive argument against the non-libertarian.
I maintain, however, that once the appropriate clarifications are made, it becomes clear
that non-libertarians have seriously underappreciated the force of libertarian appeals to
experience. In particular, dismissing various appearances of freedom as possible
illusions does little to undermine the case for libertarianism if the core of this case is that
libertarianism provides us with the best explanation of these appearances.
The paper proceeds as follows: in section 1, I lay out some of the common
descriptions of experiences that have been taken to provide the appearance of free will. I
argue that these “appearances of freedom” are not a unified experience, but rather a broad
umbrella of disparate experiences. In section 2, I argue that three common uses of the
should make predictions and answer questions to the effect, “What would I expect were this theory true?” Hypotheses that do not meet these minimal criteria are at greater risk of worries and objections found in (Van Fraassen, 1980). 45
term ‘appearance’, drawn from the philosophy of perception and philosophy of science,
fail to support a compelling libertarian IBE argument. If the libertarian IBE is to be
successful, one must adopt a phenomenological reading of appearance claims. In light of
these clarifications, in section 3, I highlight Timothy O’Connor’s case for agent causal
libertarianism as an illustration of the libertarian IBE. Finally, in sections 4 and 5, I consider other potential explanations of appearances of freedom and argue that the burden is currently on non-libertarians to provide a more compelling alternative.
1. Appearances of Freedom
It is not uncommon among libertarian philosophers, even in the modern day, to
hold that apparent experience supports their preferred view. Richard Swinburne provides
one clear statement of the strategy. He writes that, when presented with a choice between
simple alternatives, we:
…become conscious of a significant freedom: ‘it’s really
up to me what I am to do.’…So, in the absence of counter-
evidence (in the form of a deterministic theory of our
behavior in such circumstances, rendered probably true by
much evidence), in those circumstances we probably are
choosing without our choice being caused (Swinburne,
2013, 202).
This suggests a kind of phenomenological argument, where common experience
gives us some reason to believe in certain metaphysical accounts of free will. But it is
46 unclear why having certain experiences should support libertarianism. The first step is to get clear on what such “appearances of freedom” are supposed to be.
The experiences in question revolve around the exercise of agency, and quite often involve cases of choice. The experiences in question are often taken to be commonsensical, in that everyone has had them, and easily illustrated by one key example. Yet even among themselves, libertarians have failed to settle on a univocal description of the putative experience. The variety of descriptions found in the literature and upon personal reflection suggests that there really is no canonical or univocal
“appearance of freedom.” Consider the following (abbreviated) list:
• The experience of agency- It appears that I am active with respect to my
actions (Swinburne, 2013, 201; Moore, 2016; Horgan 2011a, 78).
• The experience of self-as-source- It appears that my choices result from
me, as opposed to resulting from mental states that I have (Horgan, 2011a,
79).
• The experience of being an uncaused cause- It appears that I cause my
actions, but nothing causes me (Mawson, 2011, 151; Swinburne 2013,
201).
• The experience of exercising the power of control- It appears that I
exert an active effort to guide and control my choices (Pink, 2010).
• The experience of a psychological causal gap- There appears to be a gap
in the psychological chain of causation at the point of choice (Searle,
2003, 269).
47
• The experience of indeterminism- It appears that choice is made via an
indeterministic, rather than a deterministic, process (Searle, 2007, 74-75).
• The experience of not being caused by one’s reasons- It appears that the
causes leading up to my choice do not causally necessitate the choice
(Swinburne, 2013, 202; O’Connor, 1995, 200; Descartes CSM II, 260).
• The experience of choosing in view of one’s reasons- It appears that the
consideration and weighing of reasons influences choice in a non-
deterministic way (Swinburne, 2013, 202; O’Connor 1995, 200).
• The experience of being able to choose otherwise in an unconditional
sense- It appears that I could have made a different choice than I did, even
if all facts about me and the world leading up to my choice were the same
(O’Connor, 1995, 200).
• The experience of openness- It appears that, prior to choosing, there are
numerous courses of action available (Horgan, Tienson, & Graham, 2003,
323; Sripada, 2016, 2929-2930).
All of these provide descriptions of how things appear. It (allegedly) appears to
us that we are able do otherwise in an unconditional sense, and this is a different
experience from the experience of being an uncaused cause.18 Yet both descriptions
seem appropriate when discussing appearances of freedom. So when discussing the
“appearances of freedom,” we are actually discussing a patchwork umbrella term
18 This is not to deny that there may be deep conceptual connections between the experiences. However, on the face of it, these experiences express different contents. 48
applying to a collection of heterogeneous experiences, not a term that picks out one
univocal experience.19
It is not even that all appearances of freedom are experiences in the moment of
acting. The experience may be backward looking on the moment of choice, as when it
seems upon reflection that a given choice was not causally necessitated. The experience
may also be forward looking towards a future choice, as when it seems there are
numerous alternative options available to me prior to choosing. The point is that there
are a wide variety of appearance claims that can be considered “experiences of freedom.”
Although this variety of descriptions of the appearances may seem innocuous on
its face, it becomes a problem for developing a libertarian IBE, if libertarians and non-
libertarians are not seeking to explain the same phenomena.20 For example, if one theory
explains the appearance that my reasons do not causally necessitate my choice, and
nothing else, while an alternative theory explains the appearance that my choice is an
uncaused cause, and nothing else, there is no ground for comparing these theories via an
IBE. They lack a common ground. For this reason, there must be some agreement
between libertarians and non-libertarians about which appearances are in need of
explaining. The theory that wins out provides, other things being equal, an explanation
of the largest set of appearances of freedom.
19 (Horgan, 2011; Forthcoming) are notable exceptions that acknowledge the diversity of experiences. It is, of course, always available to opponents to reject these descriptions of the appearances. For the time being, I think there are enough philosophers, libertarian and non-libertarian, that accept these descriptions to use them as a jumping off point. I return to the question of haggling about the appearances later in the paper. 20 See (Nahmias et. al., 2004) for possible differences between libertarian and non-libertarian descriptions of appearances of freedom 49
It is implausible, as some might be inclined to argue, that the experience of my
choice as not causally necessitated by my reasons and the experience of being a totally
uncaused cause are just different ways of describing the same underlying experience. If
that were correct, then libertarians and non-libertarians might actually be picking out the same experience under different descriptions. However, describing my choice as not causally necessitated and describing my choice as being an entirely uncaused cause
provides two entirely different, extensionally non-equivalent descriptions; my choice
could be causally influenced but not necessity, or it could be entirely uncaused but
epiphenomenal.21 It strains credibility to argue that these dramatically different
descriptions are picking out the very same experiential contents.22
This discussion suggests a condition of adequacy on the libertarian IBE. If we are
to compare libertarian and non-libertarian explanations of the appearances of freedom,
then we need to make sure they are picking out some of the same experiences.
Furthermore, other things being equal, the theory of free will that explains the widest set
of experiences is the better explanation. It is worth flagging at this point that not all
libertarian theories will be on equal footing with respect to explaining the wide array of
appearances of freedom.23 I will return to this point, in section 4, when discussing the
merits of various libertarian and non-libertarian theories.
21 Historically, some non-libertarians reject the claim that we experience our choices as not causally necessitated. (Spinoza, 1677/2005, 3IIp2s; Hobbes, 1656/2005, 41.) 22 This is not to say that there is NO overlap in contents, only that the descriptive differences give us some reason to believe there is some difference in experiential contents. 23 For that matter, it may be worth considering non-free experiences as well. Martin Luther famously declared at the Diet of Worms, “I cannot and will not recant anything, for to go against conscience is neither right nor safe. Here I stand, I can do no other, so help me God.” It may be worth considering how 50
2. Four Readings of Appearance Claims
2.1 The Subjunctive Reading
A natural starting place for interpreting ‘appearance’ in the present context is to
look at how the term is used in other IBE arguments. The debate between intelligent
design theorists and evolutionists in explaining apparent design in biology provides a
paradigmatic example. When William Paley presented his version of the teleological
argument in his Natural Theology, he maintained that the best explanation of the fact that
the world appears designed is that it actually is designed by God. At the time, it would
have been difficult to see how the alternative, that the world was not designed by God,
could explain the appearance of design. The publication of Darwin’s Origin of Species
shifted the dialectic by giving full voice to a competing hypothesis. Crucially, he showed
how the appearance of design could arise in the absence of a creator without resorting to
brute chance; apparent design might instead have resulted from evolution by natural
selection. The debate soon became one of haggling over the appearances and
determining which theory best explained apparent design.
Obviously, in seeking to explain apparent design, Darwin and Paley are interested in explaining why the biological world exhibits various characteristics, and not merely why we experience the world as exhibiting these features. As such, a psychological theory will not do. What is required is an account of the etiology of organisms and their traits. Further, the debate between intelligent design and evolutionary theory made
such appearances are related to mechanisms of choice and views of free will just as much as paradigmatic appearances of freedom. 51
headway, in part because it treats claims about appearances subjunctively. Roughly, to
claim that x appears designed is to claim that x possesses those signature properties that one would expect x to possess, were it designed.
When Paley and Darwin agree that the human eye appears to be designed, they are agreeing that the eye exemplifies some signature properties of design. Human eyes are symmetrical with complex interactive parts that work together to perform a crucial function. Paley and Darwin further agree that we are disposed to experience these properties as the products of design. But when it comes to the precise nature of these properties and their etiology, Paley and Darwin disagree. Paley thinks the properties are products of intelligent design. Darwin thinks the properties are products of evolution.
One quickly notices similarities between disagreements about the appearance of design and disagreements about the appearances of freedom. Libertarians and non- libertarians disagree about the etiology of choices; they disagree about whether choices come about via libertarian or non-libertarian processes. This tracks the disagreement about the etiology of organisms and their traits in the dispute between Paley and Darwin.
It is tempting, then, to interpret the appearances of freedom subjunctively. On the subjunctive interpretation, when libertarians claim that choices appear to be free, they would be claiming that choices are the way one would expect, were they the product of a libertarian mechanism.
This temptation should be resisted, as the similarities between apparent design and the appearances of freedom quickly break down. Paley and Darwin agree that organisms appear designed, where apparent design means the organism exemplifies
52 signature properties of design: the organism is symmetrical, the organism is hierarchically complex, etc. These signature properties are theoretically neutral with respect to evolution and intelligent design. Because of this neutrality, Paley and Darwin can provide alternative explanations of the same signature properties. Paley explains the symmetry of the organism using the etiology of design, while Darwin relies on evolution.
The problem for the subjunctive reading of ‘appearance’ in a libertarian IBE is that there are no obvious theoretically neutral signature properties corresponding to appearances of freedom. If one adopts the subjunctive reading of the appearances of freedom, then the non-libertarian is presented with the impossible prospect of explaining why choices have libertarian properties without having a libertarian etiology. For example, it is hard to see how one could show why agents are uncaused causes of their choices without using libertarian resources. As a result, the subjunctive reading begs the question against the non-libertarian and cannot support a compelling libertarian IBE.
The source of the problem for the subjunctive reading is that there is no obvious way, on the basis of the appearances of freedom, of spelling out the signature properties of choices in non-libertarian terms. It might be pointed out that the needed signature properties could be described in response-dependent ways such that they do not beg the question. For example, choices might have the property of “disposed to cause in us the experience as of being uncaused causes.” Perhaps, but now the issue becomes one of explaining the relation between choices and experiencers of those choices such that we are disposed to have the appearances of freedom, rather than mere properties of choices
53
alone. In essence, this move collapses the subjunctive reading into the phenomenological
reading, which I will turn to below.
2.2 The Epistemic Reading
In light of the failure of the subjunctive reading, the next logical step is to look at
how ‘appearance’ claims are used in nearby philosophical domains, starting with the
epistemic reading from the philosophy of perception. I think the epistemic reading can
be dismissed fairly quickly. Briefly, to say that, “x appears F,” is to say that it seems to
me that x is F in the sense that my evidence supports the contention that x is F.24 For
example, to say, “It appears the Falcons are going to win,” is just to say that the evidence
supports the contention that the Falcons will win. In some cases, this evidence may lead
one to form the belief that x is F, though it need not, as there could be other defeating
evidence.25
For the purposes of a libertarian IBE, the epistemic reading will not do, as the first
premise of the Libertarian IBE Schema would beg the question against the non-
libertarian. On this reading, the appearance claims endorsed by O’Connor, Descartes,
and others, would be tantamount to insisting that the available evidence already supports
a libertarian conception of freedom. Yet in the present context, this presumption is up for
grabs. Indeed, non-libertarians deny that the evidence supports libertarianism. In which
case, if claims regarding appearances of freedom are given an epistemic reading, then it is
24 It is possible to distinguish between all-things-considered evidence and strictly visual evidential epistemic senses. (Jackson, 1977, 30). If there is a similar distinction between all-things-considered evidence and strictly agential experiential evidence, I think it is unimportant here, as both readings beg the question against non-libertarians. 25 There may also be doxastic interpretations of appearance claims that involves belief formation but not on the basis of evidence, but I put this interpretation to one side. (Alston, 2002). 54
hard to see how the proposed libertarian IBE could be of any probative value. The very
first premise of the argument supposes that we possess precisely what the IBE is
supposed to provide– a reason to prefer libertarianism to the alternatives.
2.3 The Comparative Reading
We turn now to another traditional reading of appearance claims, the comparative
reading. Roughly, x appears F in the comparative sense if x appears the way Fs appear in
normal conditions. To say, “The hydrant appears to be red,” on the comparative reading
is to say that the hydrant looks the way red things do in normal conditions.
The problem for the libertarian IBE on the comparative reading of ‘appearance’ is that it requires a salient comparison class. In the case of appearances of freedom, however, it is not clear what that comparison class could be. If one says in the comparative sense, “My choice appears to me as not causally determined by prior mental states,” then that means something like, “My choice appears to me the way things not causally determined by prior mental states actually appear, in normal conditions.” But it is not at all clear what these “things” could possibly be in order to compare them to choices. If the comparison class is other actual choices, then the libertarian IBE
straightforwardly begs the question. It assumes that there are actual choices that are not
causally determined by prior mental states. But if it is anything else, it is not clear why
the proposed class would constitute a salient comparison class. Many things are not
causally determined by prior mental states. The motion of the planets is not so caused,
for example. But surely planetary motion does not appear under normal conditions in a
fashion that is relevantly similar to the way my choice now appears. Similarly, one might
55
try to compare the experience of indeterminism to an experience of the indeterministic
movement of quantum particles. But surely these experiences are relevantly different!26
The worries in the previous paragraph resulted from the requirement that an
appropriate comparison class be constituted by actual conditions. Perhaps the
comparison class could include counterfactual conditions. Call this the counterfactual
comparative reading. To say that, “My choice appears uncaused,” on the counterfactual
comparative reading is to say that the choice appears the way one would expect were
there uncaused choices. This reading neither begs the question against the non-
libertarian, nor requires an actual comparison class. The problem with this reading,
however, is that it gives no reason to think these non-actual experiences will be anything
like the appearances of freedom described in section 1. In counterfactual cases, why
should we expect metaphysical libertarianism to give rise to the appearances of freedom
rather than some other, completely alien experiences? Unlike Hume’s missing shade of
blue, there is no continuum surrounding the appearances of freedom from which we
could infer the missing (or in this case, counterfactual) experience (Hume 1748/ 2011).
So the counterfactual appearances imagined can’t possibly adjudicate between libertarian
and non-libertarian theories.
2.4 The Phenomenological Reading
We arrive at the final construal of appearance claims: the phenomenological
reading. Appearance claims in the phenomenological sense are not easy to characterize,
in large measure because they concern the most vexed of philosophical issues:
26 Even if we could view these cases as constituting a comparison class, it is not at all clear what the “normal conditions” of observation could be. 56
consciousness. Nevertheless, the following is, for present purposes, uncontentious.
When I say in the phenomenological sense, “It appears to me that there is a tree in front
of me,” I am describing representational properties of my experience, as in response to
the question, “What are you experiencing?” Appearance claims in this sense describe
representational properties of experiential states– what are sometimes called phenomenal properties.27 The phenomenological reading is not the same as the epistemic reading, as the epistemic status of these phenomenological properties is a further question.
A first point to recognize is that, on the phenomenological reading, libertarianism
alone provides, at best, only a partial explanation of appearances of freedom. If
libertarianism turned out to be true, it would make the contents of experiences of freedom
veridical. While veridicality may be an attractive feature of a theory for some, the fact
that experiences may turn out to be veridical if some version of libertarianism is true does
not explain why there are appearances of freedom in the phenomenological sense. On the
face of it, the truth of metaphysical libertarianism seems consistent with a world in which
agents have all the capacities required of metaphysical libertarianism but no experiences
at all (Strawson, 1986). It also seems consistent with a world in which agents have all the
capacities required by libertarianism and most of the same agential experiences we are
familiar with, but no experiences of freedom. It seems consistent because the collection
of capacities normally elucidated as necessary for exercising libertarian free will do not
include experiential capacities.
27 Not all theorists agree that phenomenal properties are representational, but in the interest of supporting an IBE argument, I limit attention here to representational theories. 57
On the phenomenological reading of appearances of freedom, then, libertarian
metaphysics provides at most a partial explanation. In order to provide a more complete
explanation of the appearances of freedom, one must also provide an account of the
relationship between those appearances and the free choices they purport to be about.
Are the appearances of freedom perceptual states, or cognitive states, or conative states?
Are they veridical? Answering these questions requires what we might call a theory of
choice representation for the appearances of freedom, in addition to a metaphysics of free
choice. As I am using the term here, a “theory of choice representation" provides an
account of the particular way in which the appearances of freedom come to
phenomenally represent our choices as having certain properties, in much the same way
that a “theory of visual perception” explains how visual states comes to represent the
external world as having various properties.28 In the next section, I will argue that the
libertarian IBE is most successful when one adopts a perceptual theory of choice
representation.
The above observation has a number of important consequences. First, it
considerably complicates the IBE for libertarianism by substantially increasing the
number of competing hypotheses. On a phenomenological reading of appearance claims,
it is not libertarianism per se that we are assessing for explanatory adequacy. Rather, we
are assessing libertarianism (L) combined with some theory of choice representation, T.29
28 I take it to be a virtue of the IBE approach that it is neutral with respect to theories of consciousness one deploys in a theory of choice representation: higher order thought theories of consciousness, higher order perception theories, reflexive theories, etc. could all be used in an explanation of the way that the appearances of freedom come to represent purported properties of choices. 29 Even if one argues the metaphysics are important as part of a “difference-maker” explanation, one still must assume some theory of choice representation. 58
In which case, for the argument to support libertarianism, it must be that (L+T) is a better
explanation of the phenomenology than non-libertarianism combined with the same
theory of choice representation, (C+T), as well as non-libertarianism combined with any of the alternative theories of choice representation, (C+T*). Furthermore, it must be that
(L+T) provides a better explanation than any theory of choice representation T* taken in isolation.30 Unless this is so, the IBE will end up providing support for a theory of choice
representation, as opposed to a metaphysics of choice.
Second, once we recognize the above, the proponent of the libertarian IBE must
confront what might be called the Redundancy Problem. As we have seen, because the facts to be explained are purely phenomenal ones, a theory of choice representation is required in order to explain the appearances of freedom. But once the introspective process that gives rise to such experiences is fully specified, one risks rendering the underlying metaphysics explanatorily redundant. That is, because the phenomenal character of appearances of freedom might be largely an artifact of the system by which the experiences are produced, questions about phenomenal character may turn out to be orthogonal to the metaphysics of free will.
Some may balk at the claims of the previous paragraph. The fact that we know how the visual system gives rise to visual experiences does not mean that the fact that there is a tree in front of me is not an important component of a full explanation of the
30 For the explanationist, who interprets “best explanations” in terms of determining likelihoods, then the following must be the case for libertarianism to enjoy any explanatory preference over alternatives (E= experience as of being free): P(E| L)>> P(E| C); P(E| L&T)>> P(E| C&T); P(E| L&T)>> P(E| C&T*); P(E| L&T)>> P(E| T). Granted, it may still provide some slight reason to prefer libertarianism to non- libertarianism on explanatory grounds. But the reason would be weaker. Part of why it will give us weaker justification for believing libertarianism is purely probabilistic. The probability that a conjunction is true cannot be higher than the probability of either of its conjuncts. 59
visual experience as of a tree in front of me. Likewise, the fact that a more worked out
experiential theory would be required to provide a more complete explanation of
experiences of freedom does not mean that the metaphysical theory of free will drops out
of the explanation entirely.
This objection largely misses the point. Depending on what theory of choice
representation one assumes, it is more or less reasonable to suppose that the properties of
my choices explain the appearances of freedom by virtue of the fact that my experiences
accurately represent those properties. The non-libertarian can agree that the existence of choices is relevant to an explanation of appearances of freedom, and even agree that properties of choices are likewise relevant, without taking those properties to be relevant to the explanation in virtue of a veridical or accurate representation of them. An analogy will help clarify. Sailing ships visually appear to get smaller as they move away from the perceiver. This appearance of shrinking, in the phenomenological sense, is explained in part by properties of the way the visual system processes distance, and not merely in terms of properties of ships. While ships and their properties do not drop out of the explanation of the appearance of shrinking, this does not mean that the ship is actually shrinking.
The phenomenological reading allows the same kind of explanation in the context of the libertarian IBE.31 Appearances of freedom may be explained by properties of the
process by which choices come to be represented in combination with properties of
choices themselves without the additional requirement that the appearances be accurate.
31 See (Hume, 1779/2007) & (James, 1902), as well as (Kraut, 2001) for the use of this strategy in the domains of religion and mathematics, respectively. 60
If choices do not have the properties indicated by appearances (as in the ship case), then
we may be inclined to view the appearances of freedom as illusions. But even if choices do have the properties libertarians think they have, the veridicality condition would be explanatorily redundant. In which case there would be no reason to prefer the libertarian account for the purposes of an IBE. This is not to say that veridicality may not be a good-making feature of the explanation; only that veridicality is not a necessary component of an explanation of the appearances of freedom.
The phenomenological reading of appearances raises the possibility that appearances of freedom are causally removed from how we actually make choices. And if this were true, an alternative story could be told about the etiology of those experiences independent of the fact that we make choices (though whether those choices are free in the libertarian sense remains disputed). The question, however, is whether this is a better explanation of phenomenological appearance claims than libertarian alternatives. In the next section, I hope to show that, on the phenomenological reading, some forms of libertarianism explain a broader set of appearances of freedom than non-libertarian alternatives. When combined with other theoretical virtues, this provides a strong prima facie case that libertarianism is the current best explanation of appearances of freedom.
3. The Prima Facie Case for Libertarianism
We are now in the position to begin to fill in the details of the Libertarian IBE
Schema, in light of the above discussion. The revised Schema looks as follows:
61
1) There are widespread phenomenological appearances of freedom in
need of explanation.
2) Libertarianism, in conjunction with an appropriate theory of choice
representation, together explain these appearances of freedom.
3) There is a range of alternative (both compatibilist and incompatibilist)
explanations of appearances of freedom that do not explain appearances of
freedom as well as libertarianism in conjunction with an appropriate
theory of choice representation does.
∴ We have some reason to think libertarianism is true.
Considering the wide variety of libertarian and non-libertarian accounts of free will (or
lack thereof), to fill in this schema could fill numerous volumes. The aim of this section
is merely to sketch the contours of how the libertarian argument might go. Briefly, the
libertarian IBE holds that the appearances of freedom are best explained by the fact that
we actually are free in the way that it appears. Call this the actualist explanation. The
actualist explanation itself relies on the fact that the appearances of freedom are
themselves veridical perceptions that (generally speaking) accurately track how we make
choices. Not all forms of libertarianism can underwrite the actualist explanatory strategy, since some forms of libertarianism would predict different appearances of freedom were we to veridically perceive the process by which we make choices. In what follows, I
argue that agent causal libertarianism in conjunction with a perceptual theory of choice
representation provides the best explanation in that it (1) explains the largest class of
appearances of freedom, (2) with the fewest resources, and (3) is the most conservative in
62
that it makes the appearances of freedom generally veridical. In section 4, I argue that,
for these reasons, and in the current absence of a better non-libertarian alternative, there is a prima facie case to be made for agent causal libertarianism on the basis of the libertarian IBE.
3.1 Premise 1
To fill in the first premise of the Libertarian IBE Schema, we can look back to the list of appearances from section 1. Recall that there is a wide range of appearances of freedom, including the experience of agency, the experience of indeterminism, and the experience of choosing in view of one’s reasons, among many others. These are phenomenological claims about what it is like to exercise agency. Keep in mind that the list in section 1 is not meant to be exhaustive. The list is always open to revision, by adding, dropping, or modifying the descriptions of the appearances to be explained. Still, the provided list is compiled from descriptions collected from libertarian and non- libertarian philosophers alike. Other things being equal, the theory that can explain the largest class of the appearances of freedom is preferable to theories that explain fewer of the appearances.
3.2 Premise 2
The core intuition of the libertarian IBE is that the appearances of freedom can be
explained by the fact that we actually are free in the way that it appears. In this sense,
libertarianism explains the appearances of freedom in much the same way that the
existence of a tree explains my having a visual experience as of a tree in front of me. I
63
have an experience as of a tree in front of me, in part, because there actually is a tree in
front of me. Call this the actualist explanation of appearances of freedom.
In order to make use of the actualist explanation of appearances of freedom in the
context of the libertarian IBE, an appropriate theory of choice representation is needed.
Broadly speaking, if the actualist explanation is to be viable, appearances of freedom
must generally track the properties of our actual decision-making processes. Among others, Timothy Bayne has argued that the appearances of freedom are part of a larger sense modality: the sense of agency. He writes, “Like other perceptual systems, the function of [appearances of freedom] is to generate representations of some domain— in this case, one’s own agency— and make those representations available to the agent’s cognitive system” (Bayne 2011, 6).32 Philosophers who promote actualist explanations
of the appearances of freedom have been drawn to perceptual (or quasi-perceptual) approaches to the appearances of freedom, similar to the approach argued for by Bayne
(O’Connor 1995; Pink 2004; Mawson 2011).
It may be objected that whether appearances of freedom are the result of a perceptual or quasi-perceptual process is an empirical question still under investigation.
Admittedly, if there is no reason to think that the appearances depend, at least partially, on the objects they purport to be about, then the perceptual approach will look extremely unattractive. In response to this worry, we can look to recent work on experiences of agency in the cognitive sciences. Investigations of the source of experiences of agency
32 I substitute my term “appearances of freedom” for Bayne’s “sense of agency” for clarity purposes. Bayne uses the term “sense of agency” to pick out a much broader set of experiences, of which the appearances of freedom are a subset. 64 are ongoing. Nevertheless, explanations in psychology, neuroscience, and other cognitive sciences may provide some early, tentative support for perceptual approaches to experiences of agency, and the appearances of freedom by extension.
One especially promising approach relies on so-called “comparator models.”
Supporters of comparator models of the experience of agency liken our knowledge of our own acting to visual perception (Prinz, 2003; Jeannerod, 1997; Moore, 2016; Carruthers,
2012; Haggard & Tsakiris, 2009). According to these models, experiences of first-person agency arise when forward-looking motor plans match the returning sensory consequences of an action (Blakemore, Wolpert, & Frith, 2002; Moore & Obhi, 2012).
These models suggest that sensorimotor processes that give rise to action also play a substantive role in some appearances of freedom, like the experience of self-as-source.
The comparator model, if correct, could provide fertile ground for developing a perceptual approach to appearances of freedom.
Admittedly, alternatives to comparator models are less supportive of the perceptual approach. So-called ‘theory of apparent mental causation’ models largely divorce appearances of freedom from the processes that give rise to action (Wegner,
2002). On apparent mental causation approaches, appearances of freedom arise from ad hoc inferences that do not track the processes they purport to be about. More recent work suggests that comparator models and apparent mental causation models are not necessarily contradictory, but rather work best when integrated together into a multi- factor system (Moore 2016; Moore and Fletcher 2012; Carruthers 2012). These early models suggest that the perceptual approach to appearances of freedom is not without
65 scientific merit. For these reasons, I think the perceptual approach is at least broadly plausible.
With the perceptual approach in hand, we can now present the actualist explanation of the various appearances of freedom outlined earlier.
• The experience of agency- It appears that I am active with respect to my actions
because I actually am active with respect to my actions, and this property of my
agency is accurately represented in my (quasi)-perceptual experience of my own
agency.
• The experience of self-as-source- It appears that my choices result from me, as
opposed to resulting from mental states that I have, because my choices actually
result from me, and this property of my agency is accurately represented in my
(quasi)-perceptual experience of my own agency.
• The experience of being an uncaused cause- It appears that I cause my actions,
but nothing causes me because I actually cause my actions but nothing causes me,
and this property of my agency is accurately represented in my (quasi)-perceptual
experience of my own agency.
• The experience of exercising the power of control- It appears that I exert an
active effort to guide and control my choices because I actually exert an active
effort to guide and control my choices, and this property of my agency is
accurately represented in my (quasi)-perceptual experience of my own agency.
• The experience of a psychological causal gap- There appears to be a gap in the
psychological chain of causation at the point of choice because there actually is a
66
gap in the psychological chain of causation at the point of choice, and this
property of my agency is accurately represented in my (quasi)-perceptual
experience of my own agency.
• The experience of indeterminism- It appears that choice is made via an
indeterministic, rather than deterministic, process because choices are actually
made via indeterministic processes, and this property of my agency is accurately
represented in my (quasi)-perceptual experience of my own agency.
• The experience of not being caused by one’s reasons- It appears that the causes
leading up to my choice do not causally necessitate the choice because the choice
is actually not causally necessitated, and this property of my agency is accurately
represented in my (quasi)-perceptual experience of my own agency.
• The experience of choosing in view of one’s reasons- It appears that the
consideration and weighing of reasons influences choice in a non-deterministic
way because the consideration and weighing of reasons actually influences
choice in a non-deterministic way, and this property of my agency is accurately
represented in my (quasi)-perceptual experience of my own agency.
• The experience of being able to choose otherwise in an unconditional sense- It
appears that I could have made a different choice than I did, even if all facts about
me and the world leading up to my choice were the same, because I actually could
have made a different choice than I did, even if all facts about me and the world
leading up to my choice were the same, and this property of my agency is
accurately represented in my (quasi)-perceptual experience of my own agency.
67
• The experience of openness- It appears that, prior to choosing, there are
numerous courses of action available because there are actually numerous
courses of action available, and this property of my agency is accurately
represented in my (quasi)-perceptual experience of my own agency.
3.3 Premise 3- Assessing Libertarian Alternatives
Up to this point, I have discussed libertarianism as though it were a single view.
However, there are numerous forms of libertarianism, and it is not obvious that all of them can make equal use of the above actualist explanations. As a first pass, all forms of libertarianism are committed to the falsity of determinism and the existence of genuine alternative possibilities. This suggests that all forms of libertarianism can make use of actualist explanations for the following appearances of freedom: the experience of a psychological causal gap, the experience of indeterminism, and the experience of openness. Not all forms of libertarianism, however, are equally well positioned to explain the other appearances of freedom.
Event causal and non-causal versions of libertarianism have difficulty explaining the agentive aspects of the appearances of freedom. Consider Robert Kane’s (1996) influential libertarian theory, which understands free actions to be non-deterministically event-caused. Briefly, Kane argues that free actions occur when quantum indeterminacies percolate up into chaotic systems in the brain. When presented with a character building choice, these chaotic systems favor different options. Free choice occurs when one of these two fundamentally indeterministic systems wins out. This “winning” does not necessarily involve an active exercise of a power on the part of an agent. Which means
68
that, without a further explication of the way in which an agent is active with respect to
their choice, Kane cannot provide an actualist explanation of the experience of agency.
Kane’s theory is thus unable to explain the full range of the appearances of freedom.
More generally, the experience of agency, the experience of self-as-source, and the experience of exercising a power of control all include the notion of an agent who is active with respect to their actions. But not all event causal and non-causal forms of libertarianism include agents as a fundamental part of their theories. So it is not at all clear that event and non-causal libertarian theories can make use of the actualist explanation to account for the agentive aspects of the appearances of freedom.
Agent causal libertarianism is preferable on IBE grounds in light of the shortcomings of event and non-causal theories. Agent causation just is the active exercise of a power by an agent. Appearances of freedom like the experiences of agency, self-as-source, and exercising a power of control are experiences of actual powers posited by agent causation. In addition, the agent causal libertarian can adopt the actualist explanations of the other appearances of freedom. Since agent causal libertarianism can provides actualist explanations of the entire range of the appearances of freedom, not just a subset, it is explanatorily preferable to other forms of libertarianism.
Let’s take stock. So far in this section I have argued that a perceptual approach to choice representation is empirically plausible, and that this approach, when paired with agent causal libertarianism, supports an actualist explanation of all the considered
69
appearances of freedom.33 Without further argument, contemporary event causal and
non-causal views at best explain only a subset of the appearances of freedom. Recall
that, other things being equal, the explanation that explains the widest range of
phenomena is preferable. I take this to be the greatest explanatory benefit of adopting
agent causation for the purposes of the libertarian IBE: it explains the largest collection of
the appearances of freedom.
A second benefit to adopting agent causal libertarianism is that it can help explain
apparently widespread beliefs about the appearances of freedom. Whether or not there is
a widespread belief in libertarian freedom, people often seem to believe that they
experience their choices in the ways described by the appearances of freedom.
Unfortunately, there is as yet little empirical work on beliefs about experiences of
freedom, though what little work there is increasingly supported the claim that people
think of their experiences of choosing in libertarian terms (Nahmias et. al., 2004; Deery
et. al. 2013, 2015a, 2015b; Knobe, 2014). Casual polling suggests that many
undergraduates describe their experiences using libertarian terms (Holton, 2010, 86-87).
And numerous libertarian, non-libertarian, and agnostic philosophers describe their
experiences in ways that broadly accord with the appearances of freedom canvassed here.
These strands all point towards the tentative conclusion that people tend to believe that
their experiences comport with at least some of the appearances of freedom.
The agent causal libertarian can provide an actualist explanation that reinforces
these beliefs: people often believe that they experience their choices in the ways
33 For ease of exposition, I will use “agent causal libertarianism” to refer to the conjunction of that theory with the perceptual approach to choice representation from here on out. 70
described by the appearances of freedom, at least in part, because their choices actually
appear to them that way. Those who would deny the appearances of freedom take on the additional burden of showing why people tend to believe their choices have the appearances of freedom despite this denial.
Some partial attempts to discharge this burden have been provided by non-
libertarians (Horgan 2011a, 2011b). These responses point out that our belief formation
is often theory laden, and so the appearances of freedom might be “read back” into
experiences of choice. In this sense, appearances of freedom are purely products of
inference or judgment about experiences of choice. In doing so, these theorists deny that
choices have the appearances of freedom. Such haggling about the phenomena to be
explained, in this case the appearances of freedom, is part of the normal course of IBE
debates. To repeat, the abbreviated list of the appearances of freedom provided here is by
no means definitive or settled. Nevertheless, philosophers who deny the appearances of
freedom face an uphill battle. Many non-libertarian philosophers nevertheless agree that many of the appearances of freedom accompany choices (Magill 1997; Horgan, Tienson,
& Graham, 2003; Holton 2009). On the experimental side, in many cases, beliefs about experiences seem to be libertarian, but judgments of freedom and responsibility are a mixed bag of compatibilism and incompatibilism (Feltz, 2017). If folk theories of free will are not clearly libertarian, it is hard to see why reading those theories back into interpretations of the phenomenology would result in comparatively strong libertarian interpretations of the appearances. Once again, the debate is ongoing, but I think agent causal libertarianism currently has the upper hand.
71
It is also worth highlighting the other theoretic virtues enjoyed by agent causal
libertarianism. The explanation of appearances of freedom shows a great simplicity and
elegance. From only two components, a perceptual approach to choice representation
and agent causal libertarianism, the theory accounts for all the appearances of freedom on
our list. As we will soon see, the prominent non-libertarian alternatives, discussed in the next section, require distinct explanations for each of the appearances of freedom, or else fail to explain some of the appearances all together.
Finally, the agent causal explanation of the appearances of freedom is extremely conservative. Other things being equal, we should prefer theories that are consistent with our prior beliefs and experiences. If agent causal libertarianism is correct, then our experiences are largely veridical, and apparently widespread beliefs about those experiences of freedom are largely correct as well. We experience choices the way we would expect, were we libertarian free. All of these considerations taken together provide a prima facie case to prefer a libertarian theory of free will on IBE grounds.
I wish to close this section by providing a few examples of philosophers who can be charitably interpreted as providing the kind of libertarian IBE I have been arguing for here. The agent causal libertarian view defended by Timothy O’Connor provides a paradigmatic example. He begins his article “Agent Causation,” pointing out:
A natural way of characterizing our typical experience of making
decisions and acting upon them – one that would, I think, gain widespread
assent - goes something like this: . . .it seems for all the world to be up to
72
me to decide which particular action I will undertake. (O’Connor 1995,
173).
On O’Connor’s view, event causation and agent causation are subspecies of the larger category of causal production. While event causation is necessitarian, agent causation is an emergent agential power that allows for undetermined causation of intentions and volitions. This view of causation is fundamental to his agent causal libertarian account of free will. On this account, we act freely when we exercise our emergent agent causal powers. O’Connor argues that this view is internally consistent, consistent with broadly naturalistic commitments, and consistent with reasons-based explanations of actions. It suggests that we have free will of a kind that is incompatible with the truth of determinism, but holds that determinism is false (at least in part because of our agent causal powers).
O’Connor suggests that common experience gives us some reason to prefer his agent causal view of freedom. He writes,
I see no reason that explanatory theories invoking the
concept of agent causality should not be adopted. The
alternative – to regard much of behavior as without
explanation (save for the fact that it falls within certain
parameters) – is simply not credible . . . the agency theory
is appealing because it captures the way we experience our
own activity. It does not seem to me (at least ordinarily)
that I am caused to act by the reasons which favor doing so;
73
it seems to be the case, rather, that I produce my decision in
view of those reasons, and could have, in an unconditional
sense, decided differently. (O’Connor 1995, 200).
This passage gestures towards an IBE argument for libertarianism. It is plausible to think
that agent causal libertarianism “captures” the way we experience our own activity in that
it explains the fact that we seem to have experiences of freedom as O’Connor describes
them. We experience that we could have decided differently in an unconditional sense
because we actually could have decided differently than we did. Our experiences are
veridical.
Notice that O’Connor does not necessarily think that agent causal libertarian
conclusions can just be “read off” our experiences. He does not take the fact that we
have experiences of freedom as directly entailing, or providing default entitlement for,
belief in his theory. Other philosophers, such as Richard Swinburne, sometimes argue
that our experiences have a sort of default epistemic status, and so we can justifiably infer
agent causal libertarianism directly from experience without the use of IBE (Swinburne
2013, 142). For O’Connor, in contrast, it is only because his theory explains the fact that
we have certain experiences that we have reason to believe agent causal libertarianism.
This explanation relies on a perceptual approach to the appearances. He writes, “If these
largely similar accounts of the experience of action are, as I believe, essentially on target,
then it is natural for the agency theorist to maintain that they involve the perception of the
agent causal relation” (O’Connor, 1995, 31). Thus, it is plausible to interpret O’Connor
74
as providing a libertarian IBE based on the phenomenological understanding of
appearances of freedom.
O’Connor does not explicitly consider alternative explanations of the appearances
of freedom. Still, philosophers from Descartes and Thomas Reid (1788), up to more
contemporary philosophers like C.A. Campbell (1967), O’Connor (1995), Thomas Pink
(2004, 2010), and T.J. Mawson (2011), appear to suggest that some form of
libertarianism is important for explaining our experiences. Consider two particularly
clear illustrations of the strategy:
Your own words, [Gassendi], establish that you have in fact had the
experience of freedom… Now this would be quite impossible unless the
will had the freedom to direct itself; without the determination of the
intellect, towards one side or the other. (Descartes, CSM II, 260)
Someone who pointed out merely that, given my upbringing, character,
and mental life, it was quite probable that I’d end up telling the truth
would not have offered as complete an explanation as could be offered …
he or she would have to add mention of the agency of the agent– of the
fact that I ended up choosing to tell the truth, that I provided my own bit
of causal ‘oomph’ as we might put it. (T.J. Mawson, 2011, 151)
In this section I have tried to establish a case for the libertarian IBE these philosophers gesture towards. Agent causal libertarianism, combined with a perceptual approach to choice representation, explains the widest collection of
75
appearances of freedom. Furthermore, this explanation is consistent with our best
science, it is extremely simple and elegant, and it is the most conservative. For all
of these reasons, agent causal libertarianism is extremely attractive from the
perspective of an IBE, other things being equal. In the final section, I will argue
that non-libertarians have largely overlooked this argument and, as a result, have
failed to provide a compelling alternative explanation of the appearances of
freedom.
4. Assessing Non-Libertarian Alternatives
Libertarian IBE in hand, I turn now to potential non-libertarian responses. I find extant non-libertarian responses to be largely inadequate. One obvious move for the non-
libertarian is to point out that their theory of free will is consistent with the appearances
of freedom. Unfortunately, merely pointing out that one’s theory is consistent with the
appearances of freedom does not thereby explain why one undergoes the appearances of
freedom. This sort of “explanation” would be bizarre in any other context. It would be
akin to the pre-Darwin opponent of Paley arguing that creatures created by random
chance are consistent with the appearance of design. Consistent, perhaps, but not very
explanatory. There is a burden on the non-libertarian to provide a better and more fine- grained explanation of the appearances of freedom.
Another common response to the libertarian IBE dismisses appearances of freedom as illusions. Call this sort of response the illusionist response. The illusionist fails to seriously take up the challenge pressed by the libertarian IBE. Consider Saul
Smilansky, who writes dismissively, “Such introspection being what it may, this cannot
76
be the final word on whether we do in fact have libertarian free will, since our feelings
themselves might be illusory” (Smilansky, 2000, 60). The illusionist is willing to grant
that we sometimes experience choices as having the appearances of freedom, but they
think we cannot draw any conclusions about the metaphysics of choice on the basis of
appearances, since the appearances could be illusions. Compare D.J. O’Connor, who
argues emphatically, “We cannot distinguish by introspecting our states of mind whether
the feeling of freedom preceding and accompanying a given action is genuine or illusory.
And if that is so, how do we know that any given action is a free action?” (O’Connor
1971, 19). The illusionist raises the possibility that appearances could be mistaken, and
takes this to undermine the libertarian argument from experience (Hobbes, 1656/2005;
Sommers 2007).
The illusionist takes a step in the right direction. By pointing out that appearances of freedom might not be veridical, the illusionist opens up space for the non-libertarian to provide an alternative explanation. Unfortunately, the illusionist fails to provide any alternative. In the absence of an alternative explanation, the illusionist provides no reason to prefer an illusory interpretation of appearances of freedom over libertarian veridical interpretations.
At best, the possibility that appearances of freedom are illusions may weaken the strength of agent causal libertarianism as an explanation by raising worries for the perceptual approach to choice representation. This worry should be taken seriously.
Nevertheless, the mere possibility of illusion is not enough to seriously undermine the perceptual approach when considered in the context of the explanatory benefits enjoyed
77
by the agent causal libertarian explanation, including the scientific plausibility of
comparator models of experiences of agency.
Leaving the illusionists behind, I turn next to a class of non-libertarian responses I call the Spinozan strategy (Magill, 1997; Spinoza, 1677/ 2005; Nichols, 2015). Unlike the illusionist, who provides no alternative account of appearances of freedom, Spinozan strategists attempt to “explain away” appearances of freedom. Once again, we could fill numerous volumes with versions of the Spinozan strategy.34 For present purposes, I
focus on the clearest version of the strategy as defended in the recent work of Shaun
Nichols (2015). Nichols argues that we mistakenly believe we are uncaused causes
because we infer from the lack of an experience of causal necessitation to the conclusion
that we are not causally necessitated by anything. We make this mistaken inference
because we are subject to an illusion of explanatory depth. Nichols seemingly provides
an alternative, non-libertarian explanation of one appearance of freedom.
Although the Spinozan strategy seems to provide an alternative explanation of the
appearances of freedom, it ultimately explains a different phenomenon. The Spinozan
strategist is primarily interested in explaining why people would believe they are not
causally necessitated, but says nothing about why people would experience themselves as
not causally necessitated. One of the upshots of our earlier discussion was that the
appearances of freedom are phenomenological claims. Since the Spinozan strategy
explains the belief that choices are not causally necessitated, but not the phenomenology
of choice, it fails to explain the appearances of freedom at all.
34 I address the Spinozan strategy in more detail elsewhere in my paper “Indeterministic Intuitions and the Spinozan Strategy” 78
Current versions of the Spinozan strategy also lack the other theoretical virtues
enjoyed by libertarian alternatives. The strategy focuses exclusively on explaining away
the belief that choices are not causally necessitated, to the exclusion of all the other
appearances of freedom. Because the strategy is silent on the other appearances of
freedom, it lacks the explanatory generality enjoyed by libertarian alternatives.
Even if the Spinozan strategist could provide explanations for each of the other
appearances, they would be hard pressed to do so without giving up on the comparative
elegance and simplicity enjoyed by the libertarian explanation.35 The problem is that the
Spinozan strategy does not obviously generalize to other appearances of freedom.
Although it seems prima facie reasonable to believe we are not causally necessitated on
the basis of a lack of an experience of causal necessitation, it is not at all clear why we
would believe, for example, that we are active with respect to our choices on the basis of
that same experience. This suggests that in order to account for other beliefs, the
Spinozan strategist will have to provide a different explanation for each belief. And the
problem of not accounting for the appearances of freedom at all still remains.
The illusionist accepts libertarian descriptions of the appearances of freedom but
dismisses them as illusions. The Spinozan strategist denies select appearances of
freedom but tries to explain why we might form beliefs with certain contents. Neither
explains the wide range of appearances of freedom as well as the agent causal libertarian
explanation. Nevertheless, a small minority of non-libertarians has attempted to explain
35 Perhaps the Spinozan strategist could deny that we enjoy ANY of the appearances of freedom. I think this is a substantial claim and hard to justify. While there is still room to haggle about the appearances, denying ALL of them is clearly unmotivated. 79
some of the appearances of freedom without denying them wholesale (Holton, 2009;
Deery, 2014; Horgan, 2011). These theorists provide promising first steps towards addressing the libertarian IBE challenge.
Richard Holton’s compatibilist account of experiences of freedom takes a partial first-step towards addressing the libertarian challenge. On Holton’s view, choices are caused by unconscious mental processes and mechanisms buried deep in the unconscious.
Choices are not, however, causally determined by the beliefs and desires we discover via conscious introspection. When we experience our choices as not causally determined by beliefs and desires, the experiences are veridical. The problem is that we mistakenly infer that our choices are not caused at all on the basis of the experience that our choices are not caused by our beliefs and desires. So while our experiences are not illusions, we do have false beliefs because we infer beyond the scope of experience.
Holton’s account regains some of the elegance seen in libertarian explanations.
Those mental states to which we do have conscious access are not causal determiners of our choice. Instead, choices are causally determined by unconscious states. Given this view of how choices proceed, one would expect to have the experience that choices are not causally determined by consciously accessible beliefs and desires. And that is precisely what we do in fact experience, according to Holton.
Despite this gain I think Holton’s explanation still falls short of the libertarian challenge. Like other non-libertarian accounts, Holton is unable to explain the larger scope of appearances of freedom. Consider the experience of exercising the power of control. Even if my prior consciously considered beliefs and desires are not causal
80 determiners of my choice, it is not clear how this fact could explain why I would experience the power of control. While Holton takes a promising first step, his explanation does not yet match the agent causal libertarian when it comes to breadth of explanation. As a final note on non-libertarian alternatives, if appearances of freedom are taken at face value, then any non-libertarian account will be prima facie less conservative than libertarian alternatives when it comes to beliefs about how we experience our choices. Conservatism is not an overwriting consideration, but a consideration in favor of libertarianism nonetheless.
5. Conclusion
I close with advice for non-libertarians. I have argued here that, as it currently stands, agent causal libertarianism provides the best explanation of appearances of freedom, other things being equal. Because IBE is a form of ampliative inference, this places a high explanatory burden on the non-libertarian. I fear that as it stands, no non- libertarian has come close to addressing the burden. Non-libertarians must approach the appearances of freedom in a systematic and honest fashion. The first step in addressing this burden involves recognizing libertarian appeals to the appearances of freedom as genuine arguments, and not the pig-headed dogmatism sometimes ascribed to it.
Hopefully, I have made some progress here in clarifying the argument and identifying the burden: to provide an explanation of the wide breadth of appearances of freedom in virtue of one’s preferred theory that is comparable to libertarian alternatives with respect to other desirable theoretical virtues.
81
Chapter 3: Indeterministic Intuitions and the Spinozan Strategy
In recent years there has been increasing evidence that people tend to
conceptualize their own freedom in libertarian terms (Sarkissian, 2010; Nichols and
Knobe, 2007; Deery et. al., 2013; Rakos, 2008). This has led to an interest among
philosophers in explaining why libertarian beliefs seems to be the default amongst the
folk. The libertarian position on free will can be minimally characterized as the view that
free will is incompatible with determinism, that determinism is false, and that humans
have free will. If one is to be a libertarian, then, one must reject the doctrine of
determinism.36 In particular, one must hold that free choices are not causally determined
by prior states of affairs.
Since libertarian free will requires that choice not be causally determined, philosophers have sought to explain the component belief that choice is not causally
determined as a partial step towards explaining the belief in libertarian free will. Many
have held that the belief in indeterministic choice is, in turn, somehow based in our own
experience of choosing (Campbell, 1967; O’Connor, 1995; Searle, 2003; Nichols, 2004;
Pink, 2004; Swinburne, 2013). This has led philosophers – libertarian and non-libertarian
36 For present purposes, we can understand determinism as causal determinism, the claim that given a state of affairs of the world at time t, the laws of nature necessarily fix the state of affairs at any subsequent time t+n. For more on determinism and its relationship to the free will debate, see Butterfield 1998. 82
alike – to seek a more complete explanation of the inference from the experience of
choosing to the widespread belief in indeterministic choice.
This paper focuses on a subset of views that attempt to explain the belief in
indeterministic choice by following a strategy that harkens back at least to Spinoza.
According to this Spinozan strategy, people do not believe their choices are
indeterministic because they experience their choices as indeterministic. Rather, people
come to believe their choices are indeterministic because they lack an experience of their
choices being causally determined. That is, according to this strategy, people draw an
inference from the absence of experiences of determined choice, to the belief in
indeterministic choice. The Spinozan strategy can be contrasted with the straight
strategy, which holds that there is widespread experience of choice as indeterministic and that the belief in indeterministic choice is based on this experience.
Although the Spinozan strategy is attractive to many philosophers, its successful
execution is far harder than one might initially suppose, as most extant presentations of
the strategy are liable to overgeneralize. In this paper, I focus on the pair of accounts
defended in Shaun Nichols’ recent book, Bound: Essays on Free Will and Responsibility.
These accounts are the most complete and empirically plausible representatives of the
Spinozan strategy. Nevertheless, I argue that both accounts ultimately fail. Thus,
overgeneralization worries remain a lingering problem for Spinozan strategists.
This paper proceeds as follows: In section 1, I briefly present the Spinozan
Strategy along with a discussion of some widely accepted adequacy conditions one
should meet (all else being equal) if one’s execution of the strategy is to be successful. In
83
section 2.1, I lay out the first of Nichols’ accounts, which I call the Deductivist Account.
In section 2.2, I argue that the limited psychological data proffered in support of a key
premise in the Deductivist Account fails to justify accepting the key premise; in section
2.3, I raise several objections to this account, including the charge that it fails to meet
Spinozan Strategy adequacy conditions, even if the key premise is granted for the sake of
argument. In section 3.1, I lay out Nichols’ other account, which I call the Probabilistic
Account. In section 3.2, I argue that this latter account faces many of the same objections as the Deductivist Account, even after attempts at friendly amendment. I conclude in section 4 by mapping the remaining terrain available to Spinozan strategists who seek to avoid running afoul of their own adequacy conditions.
1. Setting the Stage
In considering why people might think they have free will, Baruch Spinoza famously wrote:
Men believe themselves to be free, simply because they are conscious of
their actions and unconscious of the causes whereby those actions are
determined (1677/2005, IIp2s).
According to Spinoza, we are not aware of any determining causes of our actions, and so think ourselves free on those grounds. Spinoza puts the point in terms of belief in freedom, but the quotation outlines a strategy that, with appropriate modifications, yields a strongly analogous approach to explaining belief in indeterministic choice. Putting the point in terms of indeterministic choice gives rise to the Spinozan strategy. The Spinozan strategy starts by denying that we experience our choices as indeterministic. Rather, it
84
holds that we lack experiences of choice as being causally determined. The widespread
belief in indeterministic choice is then in some way inferred from this dearth of
experiential evidence for the causal determination of choice. The challenge for the
Spinozan strategist is to provide a satisfactory account of this inference.
A first prima facie worry for this strategy is that the inference seems to involve a kind of scope fallacy, whereby people take the absence of evidence as evidence of absence. While inferring evidence of absence from an absence of evidence is not always bad, it requires further background assumptions if the inference is to be even remotely reasonable. Spinozan strategists attempt to provide the background assumptions necessary to bridge this gap (Holton 2009, 2011; Nichols 2015; Horgan 2011b; Dennett
1984; Magill 1998). Prominent examples of Spinozan strategists include Daniel Dennett, who writes of decision-making:
We see the dramatic effects leaving; we don’t see the causes entering; we
are tempted by the hypothesis that there are no causes. The invisibility of
causal paths is not just a matter of the invisibility (to us) of other minds.
From our own first-person ‘introspective’ vantage point the causal paths
are equally untraceable (1984, 77-78).
Richard Holton argues a similar line:
It is easy to see . . . how the experience of choice could be mistaken for the
experience of one’s actions being undetermined: one is mistaking a local
claim of one’s non-determination (one’s action is not determined by one’s
85
prior beliefs, desires and intentions) for a global one (one’s action is not
determined by anything) (2009, 171).37
Again, in a similar vein, Sean Nichols writes:
[A]ssume that people know that what is introspectively accessible doesn’t
fix a decision. If people also presume that they have access to the
proximal influences on their decisions, then they would have the materials
to draw the reasonable inference to indeterminism (2015, 42).
These philosophers all agree that we come to believe our choice is indeterministic on the
basis of an absence of an experience of choice being causally determined.
We might contrast the Spinozan strategy with the straight strategy. According to
the straight strategy, we do experience our choices as causally undetermined and thereby
come to believe, on the basis of such experience, that choice is not causally determined
(Searle 2003, 270; O’Connor 1995, 200; Swinburne 2013, 201-202). Thus, to the extent that inference is involved at all, the inference is a straight one, in the sense that the contents of our experiences and our beliefs about choices are much the same – they both represent choice as indeterministic. This sort of explanation has been particularly attractive to libertarians, who further argue that the belief is justified and provides support for libertarianism (Campbell, 1967; O’Connor, 1995; Pink, 2004; Mawson, 2011; and
Swinburne, 2013).
37 Nichols considers himself to be arguing the same line as Holton, though they do seem to describe the experiences slightly differently. However, both agree that indeterminism, full stop, is not part of the content of the experience of choice. 86
Non-libertarians have been surprisingly reluctant to endorse the straight strategy.
This is curious given that non-libertarians, including compatibilists, hard determinists, and hard incompatibilists, could avail themselves of the strategy with minor addendums.
For example, compatibilism holds merely that free will is compatible with determinism.
It does not further require that determinism be true. In which case, one might maintain
both that determinism is false and compatibilism is true, and accept the widespread belief
in indeterministic choice. Indeed, even if one is a compatibilist committed to the truth of
determinism, like Hobbes or Hume, one might still accept the straightforward
explanation; it would only require that the experience of indeterministic choice be an
illusion (Smilansky, 2000, Sommers, 2007, Caruso, 2015).
Hard determinists, who think determinism holds and rules out free will, and hard
incompatibilists, who think free will is impossible regardless of the truth of determinism,
can similarly avail themselves of the straight explanation with no apparent threat to their
views. In the case of hard determinism, they need only point out that experience is an
illusion. For the hard incompatibilist, either the experience is an illusion, or experience
provides some prima facie evidence to think that we live in an indeterministic world.
Either way, the hard incompatibilist can still maintain their view that free will is
impossible. The straight strategy, then, seems independent of one’s commitments in the
free will debate.
Despite these options, numerous non-libertarians have pursued the Spinozan
strategy as an alternative to the straight strategy (Holton, 2009; Deery, 2013; Horgan,
2011a, 2011b; Magill, 1998; Dennett, 1986; Nichols, 2015). They seem to balk in
87
particular at the kind of raw experience of indeterminism invoked by the straight strategy.
Normally this commitment is based on personal introspection, where philosophers find
themselves unable to find any experience of indeterministic choice (Doris, 2016, 167;
Smilansky, 2000, 59; Horgan, 2011). However, one might also argue that an experience
of indeterminism more generally outstrips the possible bounds of experience, in much the
same way Hume argues we cannot experience causation. As Holton puts it, “[I]t is hard
to think what an experience of that [indeterminism] would feel like. It is too theoretically
loaded (2009, 168).” In rejecting the phenomenological claim, these philosophers must
reject the straight strategy as well.
In addition to being committed to the rejection of this phenomenological claim, the Spinozan strategy is routinely accompanied by a set of adequacy conditions. Like
many explanatory desiderata, these conditions often remain implicit in the structure of the
debates and arguments contained therein. Further, these conditions are perhaps most
plausibly construed as defeasible and ceteris paribus. For all that, I maintain that the
following (no doubt incomplete) list of adequacy conditions is both reasonable and
prevalent in much recent work by Spinozan strategists:
A. Scientific Plausibility: One’s explanations of folk beliefs should be plausible
in light of our best extant science.
B. Reasonableness: All else being equal, to the extent an explanation of folk
belief avoids the attribution of obviously unreasonable inferences, it is to be
preferred over those that attribute less reasonable inferences.
88
C. No Overgeneralization: Explanations of folk beliefs ought not to
overgeneralize.
Let me say a bit more about each of these conditions.
Scientific Plausibility. All else being equal, the explanation of the belief in
indeterministic choice ought to be psychologically plausible (Searle, 2003; Dennett,
1984, 15; Nichols, 2015, 49). This should not be contentious. The Spinozan strategist
makes empirical claims, so we ought to check them against our best empirical data. This
need not require that the account be fully supported by psychological data. Indeed, there
is a general attitude of speculation and armchair psychology involved in these accounts.
Nevertheless, to the extent that one’s account is supported by our best extant science, it is
preferable to merely possible, speculative accounts. If our best scientific picture of human
decision-making conflicts with widespread experience and belief, then explaining away
these experiences and beliefs becomes all the more pressing.38
Reasonableness. This condition calls for a minimal kind of reasonableness. The
requirement is not that the belief be true, or that the grounds for endorsing it be
reasonable, or even that these grounds be prima facie reasonable. For that matter, it does
not even impose the demand that attributions of unreasonable inferences are categorically
impermissible. Rather, the claim is merely that, all else being equal, we should prefer
explanations of folk belief that avoid characterizing a belief’s source as obviously
unreasonable (Dennett, 1986, Ch. 1; Nichols, 2015, 48; Holton, 2009, 171).
38 Many philosophers take studies such as (Libet, 1983; Wegner, 2002; and Soon, 2008) as supporting models of choice that conflict with libertarian views of free will. 89
Belief in libertarian free will, and so indeterministic choice, is widespread and
apparently cross-culturally robust.39 In view of this, there are two independent
motivations for insisting on Reasonableness. The first motivation is relevant to
explaining the apparent entrenchment of belief in indeterministic choice. If the source of
belief in indeterministic choice were obviously unreasonable, it would be hard to see why
the belief would be so widespread. Further, it would be hard to explain why it is so
apparently recalcitrant to reflection. After all, in the history of philosophy, many highly
influential thinkers have perseverated in their belief despite protracted periods of
reflection on such matters. The point is not that explanations of entrenchment that violate
Reasonableness are impossible. (What would an argument for such a claim look like?)
Rather, the point is that entrenchment appears far easier to explain if one supposes the
absence of obvious unreasonableness, and to that extent Reasonableness should be the
default.
The other motivation for insisting on Reasonableness is a minimal notion of
charity. In philosophical practice more generally, it is best to interpret your interlocutor
as making rational inferences. There are numerous examples of charitable principles in
philosophy (Davidson, 1984; Quine, 1960). Preferring explanations that are not
obviously unreasonable falls out of even the most minimal characterization of charity.
The point is not that charity requires that belief in indeterministic choice be rational.
39 (Sarkissian, 2010). Although I find the initial evidence intriguing, the evidence is admittedly in its infancy, and studies such as Nahmias et. al. (2010) may provide some push back. In eastern traditions such as Buddhism and the Samkhya school of Indian philosophy, the idea that agents are independent, original causes of action is presented as an illusion. This suggests they are presenting an alternative to the widespread view, reinforcing the idea that beliefs in a libertarian spirit may not be unique to the Western philosophical tradition. 90
Rather, the point is that even the most minimal accounts of charity prefer attributing not obviously unreasonable inferences to others.
No Overgeneralization. A prohibition on overgeneralization is a normal component of any good explanation. It counts against the acceptability of a purported explanation if the explanation predicts phenomena in places where they do not occur.
What’s true generally holds in the present case. If the Spinozan strategy predicts indeterministic beliefs where people do not hold them, this counts against the Spinozan strategy.
The Spinozan strategy is particularly susceptible to overgeneralization worries.
The lack of an experience of causal determinism is not at all unique to the experience of choosing; I doubt that even paradigmatically causally determined events, like sunrises, are experienced at the time as causally determined. And yet unlike belief in indeterministic choice, people do not infer indeterminism from a lack of an experience of determination in any other cases. No one is going around suggesting headaches are indeterministic. And yet headaches are not accompanied by experiences of determination.
The specter of overgeneralization is of primary concern to the Spinozan strategist.
Shaun Nichols expressly designs his account to address the overgeneralization worry that afflicts other Spinozan strategists. He points out, “[I]gnorance of the causes of these phenomena does not generally breed the belief that the phenomena aren’t determined. So even if Spinoza is right that we are ignorant of the causes of our actions, that doesn’t yet explain why we believe that our actions aren’t determined.” (2015, 38). The challenge
91
for the Spinozan strategist is to posit background assumptions that explain the
indeterministic belief in the case of choice, but also explains why there is no
indeterministic belief in other cases. This is not a detail to be worked out later by
psychologists. It is core to the success of the Spinozan strategy.
Filling in the Spinozan Strategy, then, involves explaining the move from a lack
of an experience of the causal determination of choice to the belief in indeterministic
choice that does not overgeneralize (No Overgeneralization) while invoking inferences on the part of the folk that are not obviously unreasonable (Reasonableness) and are plausible in light of our best extant science (Scientific Plausibility).
Although not a condition of adequacy, I would like to flag one final feature of the
Spinozan strategy before moving on to Nichols’ accounts. As we have seen, the
Spinozan strategist seeks to explain the widespread belief that choice is indeterministic.
Yet in doing so, they invariably ignore another apparently widespread, and closely related, belief – one that, all else being equal, we should also wish to explain It is not merely that people believe that their choices are indeterministic; they seem to believe that they experience their choices as indeterministic (Deery et. al., 2013; Deery, 2015). These studies found that, after making a choice, participants tended to agree that they experienced that they “could have done otherwise” in a way that is incompatible with causal completeness (138, M=5.60). In informal polls of undergraduates, many philosophers find that their students tend to describe their phenomenology in indeterministic terms (Holton, 2010, 86-87). While the empirical evidence here is admittedly in its infancy, it provides early support for an assumption that has long been
92
held among philosophers and non-philosophers: people tend to describe their experiences
of choice using indeterministic terms.
Plausibly, the straight strategy has no special problem explaining this fact.
Perhaps on the basis of introspection or some other means, we routinely take ourselves to
believe that our experiences are a specific way precisely because they are that way. I
believe, for example, that I am having an experience as of a red rose, precisely because I
am having experience as of a red rose. Reasonably enough, some straight strategists ask
why the experience of indeterministic choice should be any different (Pink, 2004). That
is, why don’t we believe that we experience choice as indeterministic simply because we
do experience choice as indeterministic?
In contrast, it is not at all obvious what the Spinozan strategist should say in order to explain this belief about experience. The suggestion, in short, is that where Spinozan strategists typically see one belief in need of explanation, there are in fact two: The belief that choices are indeterministic, and the belief that we experience our choices as indeterministic. Although, in what follows, I focus largely on the former belief, towards
the end of the paper, I will return to the latter.
2. How Possibly Story 1: The Deductivist Account
2.1 The Account
Nichols offers two different “how possibly” stories that attempt to fill in the
Spinozan strategy so as to avoid overgeneralization. I will take each in turn. I call the
first account the Deductivist Account. Nichols presents the Deductivist Account (DA) in
the following argument-form:
93
D1) The [proximal] factors that are introspectively accessible do not determine my choice.
D2) I have introspective access to all the (proximal) factors that influence my choice
D3) My choice is not determined (2015, 42)
The idea is that, if the folk believed (D1) and (D2), then they would have an apparently
valid argument which would support the belief in indeterministic choice. Nichols’ core
task, then, is to support the claim that people actually do believe (D1) and (D2)
Nichols thinks that his libertarian opponents already agree that (D1) is largely accepted, so he thinks the real action is with premise (D2). If there was widespread belief in (D2), according to Nichols, then a good explanation of why (D2) is accepted by the folk would provide a Spinozan strategy explanation that is complete without overgeneralizing. Such an explanation would not rest on an experience of indeterminism and would comport with our best science (Scientific Plausibility). Furthermore, because we assume that introspection reveals all of the proximal influences contributing to choice, the fact that introspection does not reveal choice to be causally determined also supports the belief that nothing causally determines choice. Thus the inference from introspective experience to belief is not obviously unreasonable (Reasonableness).
Finally, it is because we believe we have introspective access to all of the proximal
influences in the case of choice, but not in the case of headaches, that we believe choices
are indeterministic but not headaches (No Overgeneralization). Thus, an explanation of
94
the belief in (D2) provides a complete explanation that doesn’t overgeneralize. Or so the
story goes.
While (D2) may appear fairly benign, it is worth stressing that it is quite a strong
commitment. The Deductivist Account is most naturally read as a person-level
explanation of the widespread belief in indeterministic choice. So (D2) is naturally read
as a person-level folk belief. Further, if (D2) is to justify accepting the claim that choice
is causally indeterministic (as the Spinozan strategy requires), then (D2) must assume
introspective access to the explicitly causal features of the proximal factors influencing
my choice. Finally, even if none of the proximal factors influencing my choice is an
individually determining cause of my choice, the factors could jointly be a deterministic
cause of my choice. So, in order for (D2) to justify the belief that choice is
indeterministic, it must be construed as implying introspective access to the interactions
among these influences as a collective causal system as well.40
As Nichols clearly recognizes, he cannot merely stipulate that (D2) is endorsed by
the folk. This would render the proposal empirically unwarranted, and would also
succeed only in pushing the overgeneralization worry one stage back. In view of this, he
provides a two-staged response. First, he invokes survey-style experiments from Kozuch
and Nichols (2011) to argue that people do in fact accept (D2). Second, he draws heavily
on Leonid Rozenblit and Frank Keil’s research on the illusion of explanatory depth
(IOED), both to explain the (putative) widespread acceptance of (D2), and to avoid
40 This point will be important when discussing the empirical support for (D2). One potential way of expanding this point is to consider the properties of the system as a whole. See Wimsatt (2006) for a helpful discussion of potential criteria for system properties as opposed to merely aggregative properties. 95 overgeneralization worries (Nichols, 2015, 44-4).41 In what follows, I first explain what
IOEDs are, and their proposed role in the Deductivist account. I then argue that neither stage of Nichols’ response is satisfactory.
IOEDs occur when an individual mistakenly believes they have a deeper understanding of a particular phenomenon than they in fact do (Rozenblit and Keil, 2002;
Lawson, 2006; Mills and Keil, 2004). In particular, IOEDs occur when causal interactions in mechanisms that are easily available to perception are mistakenly taken to provide a deep causal explanation of the mechanism. In one experiment, for example, subjects were asked to rate their confidence that they understood how simple mechanisms work– e.g. locks and zippers. Initially, subjects consistently rated themselves quite highly. But when subsequently asked to provide a written explanation, people systematically downgraded their confidence ratings; and then downgraded again when presented with expert explanations of the mechanisms. This downgrading of confidence ratings when faced with the inadequacy of their own understanding is the hallmark of
IOEDs.
As Nichols suggests, it seems possible that we could be subject to an IOED when considering our own choices. When we introspect the factors leading up to our own choices, we plausibly find discrete causal states (thoughts, intentions, etc.). We “see” these discrete states interacting in much the way we see causal states within a physical mechanism like a lock. We use this introspection to explain and make accurate predictions about what we will do. We almost never receive information suggesting that
41The term “illusion of explanatory depth” here is a technical term and should not be confused with the more general notion of an illusion akin to having a false belief. 96
introspection in these cases is misguided. Thus, we could be subject to an IOED in the
case of choice. If we were subject to an IOED in the case of choices, then (D2) would
seem like a plausible assumption to adopt.
Nichols thinks the IOED hypothesis explains why the folk might endorse (D2). It also provides a way to block the overgeneralization problem. If we are subject to an
IOED in the case of choice, but not other psychological phenomena, like headaches, then
we would not assume the analog of (D2) for headaches and the like and, hence, would not
endorse indeterminism about such phenomena. That is, according to Nichols, in the
absence of an IOED:
[T]here is no presumption of broad-scale access to the causally relevant
factors in the generation of headaches. As a result, there is no basis for
drawing indeterminist conclusions about headaches from the fact that we
can’t detect a deterministic set of causes (2015, 46).
Thus, the Deductivist Account promises to furnish us with an elegant explanation of
widespread folk indeterminism about choice: The widespread folk acceptance of (D2)
helps explain indeterminism about choice; the presence of an IOED for choice explains
this widespread acceptance; and the absence of an IOED for most other psychological
phenomena addresses the overgeneralization problem.
Unfortunately, this apparently elegant explanation is subject to serious objections.
First, there is no good reason to suppose that the folk actually accept (D2) – or, at any
rate, not a version sufficiently robust to do the work required of it by the Deductivist
97
Account. Moreover, even if widespread belief in (D2) were assumed, IOEDs fail to
provide a satisfactory explanation of this putative fact.
2.2 Concerning the Folk Acceptance of (D2)
Prima facie, the assumption of (D2)’s widespread acceptance is implausible.
When pressed about one’s choices, “I don’t know why I did that,” and its cognates are
perfectly felicitous utterances in English, sincerely asserted on a seemingly regular basis.
Further, we live in a post-Freudian world where the knowledge that there are unconscious
influences on behavior is a pervasive aspect of popular culture. It would be surprising if
this knowledge had no impact whatsoever on intuitive judgments about choice.
Nevertheless the issue is, of course, an empirical one; and despite these prima facie
considerations to the contrary, it may really be the case that (D2) is widely accepted.
My main concern, however, is the paucity of empirical support for this
contention. To my knowledge, there is exactly one study that purports to bear directly on
the matter – the series of experiments found in Kozuch and Nichols (2011). But contrary
to what Nichols suggests, this study provides little or no support for folk acceptance of
(D2).
Although Kozuch and Nichols (2011) presents a series of five experiments, only
(Experiment 2) investigated the extent to which people assume introspective transparency to their own decision-making, and for this reason, I focus my criticisms on that
98
experiment.42 In Experiment 2, Kozuch and Nichols asked participants the extent to which they agreed with the following prompt:
When I am making a decision about what to do (for example, deciding
whether to go swimming), if I pay attention to my thought processes, I can
usually see what leads me to make the decision I do (146).43
Participants marked their agreement on a 1-7 Likert scale. Agreement with this statement was high, (M = 5.47, S.D. = 1.36) (146). Kozuch and Nichols compared this to agreement in cases where the statements were largely the same but “making a decision”
was replaced by “feeling an urge.”
When I am feeling an urge to do something (for example, deciding
whether to go swimming), if I pay attention to my thought processes, I can
usually see what leads me to feel the urge I do. (146).
Participants showed agreement with this statement as well, (M = 4.69, S.D. = 1.35) (146).
However, since there was greater agreement in the case of decisions than urges, Kozuch
and Nichols take general agreement with the statement, “…I can usually see what leads
me to make the decision I do,” (Em. Added) as evidence for the claim that the folk
believe they, “…are usually able to know the causes of their decisions as opposed to
urges.” (146). It is this result, which according to Nichols, supports the contention that
42 Although I address my criticisms to experiment 2, my worries about the causal interpretation of the probe statements apply to the other experiments as well, mutatis mutandis. 43 Kozuch and Nichols acknowledge that there are some concerns with this kind of explicit-measure testing. Additionally, there was at least one instance where half of the participants failed comprehension checks. While these methodological worries are concerning, my ultimate rejection of Kozuch and Nichols’ conclusions do not depend on them. See Sommers (2015) for more general methodological worries about the course of experimental philosophy in the free will debate. 99
the folk accept (D2), but not analogous claims for urges and other psychological
phenomena.
I have four main concerns regarding the use of Kozuch and Nichols’ study in
developing the Deductivist Account. First, although there is greater agreement in the case of “making a decision” than “feeling an urge”, there is nevertheless agreement with the probe statements in both cases. Indeed, in both cases, the mean is far above the midpoint on the Likert scale. But if this is so, then it is hard to see how the data could support (D2), without also providing significant support for an analogous assumption regarding urges.44 And if this is so, then Nichols would appear to have lost his response to the overgeneralization problem since it is by denying that the folk endorse such analogous assumptions that he seeks in order to resolve the overgeneralization problem.
Second, if Kozuch and Nichols’ study is to support the claim that (D2) is widely endorsed, one must adopt a causal interpretation of expressions like “leads me to decide”.
This is because “influence” in (D2) must mean causal influence. Yet expressions like
“leads me to decide” are clearly amenable to alternative non-causal readings. In
particular, when agreeing that I can “usually see what leads me to make the decision I
do”, I may be agreeing that I appreciate the reasons that speak in favor of performing an
action, without interpreting those reasons as causes of my choice. My point is not, of
course, that we must interpret the probe statements in this way. Rather, the point is that
nothing in the data militates against such a reading, and in favor of a causal
44 Kozuch and Nichols insist that despite agreement in both cases, there is a statistically significant difference in the degree of the effect between the two cases. While this may be true, it would only suggest that (D2) analogs would be slightly weaker than (D2). 100
interpretation. Further, the idea that such locutions are amenable to non-causal construals
is hardly without precedent, either in the philosophy of action or, more recently, in
empirical research on folk judgment. For example, Knobe (2014) defends what he calls
the transcendence vision, on which choices are viewed by the folk as being made “on the
basis” of prior mental states, but not necessarily caused by them (70). On such a view, I
may know the reasons that “lead me to make the decision I do” without necessarily
equating those reasons with the causes of my choice.45 Instead, they could “lead me” to
make a decision by providing a rationalization of my action. In light of interpretations
like those suggested by the transcendence vision, the causal interpretation of the probe
questions adopted by Kozuch and Nichols is in need of far more support.
Third, even if we concede a causal interpretation of the probe statements, the
putative evidence would still fall far short of supporting (D2). (D2) asserts that I have
introspective access to all the proximal factors that influence my choice. But mere
agreement with the claim that I can generally “see” what causes my decisions in no way
suggests such a view. For, even on a causal interpretation, it might indicate no more than
the judgment that I have access to a contextually salient cause of my decision. That is, I
might think I have access to some – even the – contextually salient difference maker, without thereby thinking I have access to all the proximal influences. By way of analogy, consider a case where I take myself to know the cause of an explosion –e.g. that the bomb fuse was lit. Under such circumstances, I may legitimately think I know what
45 This is not meant to refute the claim that reasons are causes, as argued for in Davidson (1963). Rather, it merely suggests that the folk concept of reason may come apart from the folk concept of a mere cause, a point also defended in Malle (1999) & Malle et. al. (2007). 101
caused the bomb to explode without thereby supposing that I have access to all the
proximal influences that brought about the explosion. Mutatis mutandis for decisions. In
short: while a causal interpretation of the probe statements may help support some vague
assumption of transparency, it does not support the stronger contention that (D2) is
widely held by the folk.
Finally, given the above, Kozuch and Nichols’ study also fails to support the
presumption, implicit in (D2), that people think all (proximal) influences on choice are
introspectively accessible mental phenomena. This is because the probe statements in
their experiments are formulated so as not to require participants to make any assessment
regarding the influence of non-introspectable factors. More specifically, by incorporating the phrase “…if I pay attention to my thought processes…”, the probe statements only prompt participants to reflect upon whether they have access to aspects of the thought
processes that lead to their decisions. But the mere fact that people tend to agree that
they can usually tell what aspects of mental processes lead to their decisions, in no way
implies that they deny the existence of additional, non-mental – hence non-introspectable
– influences. For they might think that introspectable mental phenomena and non-
introspectable phenomena influence decisions.46 As a result, the Kozuch and Nichols’
study fails to support the contention that folk thinking about our own decision-making is
typically restricted to introspectively accessible mental events.
46 It strikes me as relatively easy to test whether people accept that non-introspectable factors influence choice. One option would be to run experiments quite similar to Kozuch and Nichols’ own, but where the probe statements concern the influence of extra-mental factors. For example: “When I am making a decision about what to do (for example, deciding whether to go swimming), if I pay attention to the world around me, I can usually see what leads me to make the decision I do.” 102
2.3 Objections to the use of IOEDs in the Deductivist Account
In the previous section I argued that there is little empirical support for the widespread acceptance of (D2), and what support is offered by Nichols requires an unmotivated, strongly causal interpretation of the probe statements. But even if we grant for the sake of argument that (D2) is a widely held, Nichols’ use of the IOED literature is problematic.
First, according to Nichols, we are subject to an IOED in the case of choices but not other mental events because only in the case of choices do we identify plausibly discrete causal states via introspection. But it is hard to see why one would think that such prior causal states are present online in the case of choices. Why not, for example, in the case of headaches as well? For example, thinking hard about the intricate details of the IOED literature is extremely taxing and plausibly plays a causal role in my incurring a headache. External stressors and noises certainly seem to play some causal role in bringing about headaches in normal conditions. I have stared at sheet music while drumming and felt the slow onset of a headache. The page begins to blur, the pounding noise becomes echo-y and more intense, and the pulsing pain in my head slowly ramps up. The strain of focused visual attention and pounding noise are surely proximal causes if ever there were any. If Nichols’ theory predicts an IOED in the case of choices on the basis of the reflective availability of prior discrete causal states, then it ought to predict an
IOED in the case of headaches as well, and so predict an analogue to (D2) in the case of headaches. So, Nichols has not avoided overgeneralization. One of the main demands on his theory fails.
103
One might respond to the above criticism by arguing that many of the discrete causal states I appeal to in the hypothetical reflection on the etiology of headaches are not introspectable mental states. In the case of headaches, but not choices, we look for causes outside of the head. The Deductivist Account thus avoids overgeneralization because (D2), the premise that I have introspective access to all the (proximal) factors that influence my choice, is restricted to introspectable mental states, and so seems plausible in the case of choices, but an analogous principle in the case of headaches does
not.
But this response misses the point. We have already granted that (D2) is widely
accepted for the sake of argument. The purpose of introducing IOEDs is to explain why
people might accept something like (D2) in the case of choices but not in other cases. If
the resources that justify positing an IOED in the first place, i.e. access to plausibly
discrete causal states, are available in the case of headaches as well as choices, then
IOEDs do not provide an adequate response to the overgeneralization worry.
Putting the previous objection to one side, let us grant for the sake of argument
that there is an asymmetry between choices and headaches that justifies thinking there is
an IOED in the former case but not the latter. Even granting that IOEDs resolve the
overgeneralization problem, it is hard to see how the account is to be developed without
running afoul of Reasonableness or Scientific Plausibility.
On its most simple interpretation, the Deductivist Account fails the demand for
Reasonableness because it attributes inconsistent beliefs to the agent. Specifically, it
requires that they think their choices are both deterministic and indeterministic. As
104
Rozenblit and Keil explain, IOEDs occur when we “…mistake shallow, sparse lay
theories of mechanism for deep rich ones (2002, 23),” to provide a causal explanation of
the phenomena. They cite Ahn and Kalish (2000), who write of causal explanation: “If A
causes B, then A makes B happen: B had to happen given A. It was no accident. It is
this sense of necessity that distinguishes ‘real’ causal relations from ‘mere’ correlations.”
(200-201). The lesson: if the phenomenon is not viewed as a causally deterministic
mechanism, then there tends not to be an IOED.47 Rozenblit and Keil’s results support this lesson. IOEDs occurred in the case of devices with clearly visible mechanisms, like zippers and crossbows. In the case of non-devices, such as procedures for folding flags and narratives, IOEDs did not occur (Rozenblit and Keil, 2002, 26).48 If we are to
understand IOED’s in anything like the way that Rozenblit and Keil themselves do, then
we should suppose that they arise against a background of assumptions whereby the
target system is construed as a deterministic mechanism.
In view of the above, if an IOED is to explain the widespread acceptance of (D2),
then that involves subjects construing the target system (in this case, mechanisms of
choice) as deterministic ones. But this assumption is prima facie inconsistent with the
conclusion (D3), that choice is not determined! If IOEDs do underlie the belief in
indeterministic choice, then it would appear that we believe the systems which generate
47 Of course, Nichols could reject this aspect of the Rozenblit and Keil account. However, doing so would come at a cost. This aspect of the account is driven by a pattern in the empirical data, namely: people tend to exhibit IOED’s in the case of (apparently deterministic) mechanisms, and yet fail to exhibit IOEDs in the case of other apparently causal phenomena. If we give up this aspect of the Rozenblit and Keil account, we lose an explanation of why the data exhibits this pattern. Thank you to an anonymous referee for raising this point. 48More specifically, Studies 1-4 indicate an IOED for devices, while Study 10 indicates a smaller IOED for natural phenomena, like the occurrence of earthquakes. While the difference in results for devices and natural phenomena were not significant, they did trend towards significance, with the IOED greater in the case of devices. Of special interest are studies 8 and 9 where no IOED occurred. 105
choices are both deterministic and indeterministic. It may well be that we hold
inconsistent beliefs regarding our choices, but this apparent inconsistency constitutes a
threat to the demand for Reasonableness.
Attempts to ameliorate this apparent inconsistency face a dilemma: either the
account fails to adequately explain the phenomena, or the account fails the demand for
Scientific Plausibility. Suppose the assumption that influences on choice constitute a deterministic causal mechanism were discharged upon accepting (D3), as in a reductio ad absurdum argument. This proposal would meet the demand for Reasonableness. But this move fails to adequately explain the phenomena. For if we discharge the deterministic assumption, then were we to introspect the proximal influences on our choices at some future time, we would not be subject to an IOED. And if we were not
subject to an IOED, then we would no longer accept (D2), the claim that we have
introspective access to all the proximal factors influencing choice. In which case, we
would not accept the conclusion, (D3), that choice is not determined. Discharging the
deterministic assumption thus fails to explain the fact that we perseverate in our belief in
indeterministic choice.
The alternative proposal is to hold that we do have a long-term commitment to conflicting beliefs about the determinacy of choice, but to argue that, depending on how one understands ‘reasonable”, this conflict need not threaten Reasonableness (Sloman and Barbey, 2016). Of course, the mere fact that one holds inconsistent beliefs is not sufficient to count an individual as unreasonable. Nevertheless, I take it to be fairly uncontroversial that if one believes that choices are both deterministic and indeterministic
106
in the context of a single piece of reasoning, then the inference is prima facie
unreasonable. The inference, as it were, wears its unreasonableness on its sleeve.
Still, the beliefs that choice is both deterministic as well as indeterministic could
be compartmentalized from each other, or we could make one (or more) of the beliefs
sub-personal in an attempt to massage away the apparent unreasonableness. Both these moves embed the belief in our cognitive architecture in a way that prevents us from revising (D2) upon reflection. This in turn could explain the perseveration of belief in
(D2) while also allowing subjects to discharge the deterministic assumption.
The above proposals would extend far beyond what is empirically supported by
the IOED literature. Recall, a signature property of being subject to an IOED is that the
agent downgrades their confidence upon further reflection. Yet the above proposals
specifically hold that such downgrading does not occur. In which case we will have lost
much of the scientific plausibility there was for invoking IOEDs in the first place. The present proposals, then, while admittedly possible, no longer meet the demand for
Scientific Plausibility.49
3. How Possibly Story 2: The Probabilistic Account
3.1 The Account
Nichols provides an alternative to the Deductivist Account that relies on methods
common in probabilistic learning models to explain the inference from the absence of
evidence to the evidence of absence. I call this account the Probabilistic Account. In the
49 Still, it is worth further exploring how compartmentalized/ encapsulated closed world assumptions like (D2) could enter into an explanation of the belief in indeterministic choice that meets Spinozan strategy commitments. Thank you to an anonymous referee for pushing this point. 107
Probabilistic Account the inference from the experience of choice to the belief in
indeterministic choice is an induction based on probabilities.
To illustrate, consider a case where a friend shows you two decks of cards: one is
a fair deck and the other contains only black cards. Your friend then removes one of the
decks (you don’t know which one) and destroys it. You draw a card from the remaining
deck, reveal its color (red or black), return the card to the deck and shuffle, and then
repeat the process. Suppose after 100 draws you have only turned up black cards. There
are two possible hypotheses to explain this string of black cards. Either the deck is fair
and you have had an unlikely run of draws, or there are more black cards in the deck than
red cards. It seems reasonable to believe, in light of 100 straight black draws, that the
deck is in fact not a fair deck. The fact that you do not turn up a red card leads you to
believe that this is the all black deck.50 With each draw without a red card, you become
more confident that the deck does not have red cards. The Probabilistic Account uses similar reasoning. Here’s how Nichols presents it:
The class of decisions we’ll focus on are ‘difficult decisions” . . . like
Buridan’s ass cases . . . For each of those decisions, the agent has
knowledge of a set of mental states that influences his decisions . . . The
set of states of which he is aware does not provide a deterministic
explanation of his decision. Given the relevant evidence that he does have
(i.e., evidence of the influence of sets of states) and the evidence that he
50 According to Bayesian reasoning, each additional draw without a red card should make you more confident this is the unfair deck. For further discussion of Bayesian probabilistic reasoning models, see Tenenbaum et. al. (2006) 108
doesn’t have (i.e., evidence of sets of states that determine his decision),
the agent infers that difficult decisions are not deterministically produced
(2015, 48).
Nichols admits that a great deal must be assumed for this to work. It restricts to a class of
decisions, it assumes the accessible states do not provide a deterministic story, it assumes
that there is reason to think that absence of evidence is evidence for absence, and it
assumes the set of sets, and the mental states within each set, are a random and representative sample. This final assumption is of particular importance. It is designed to rule out the possibility of selection bias. That is, it is meant to rule out the possibility that there is a potential influence on choice that, due to the sampling mechanism, consistently slips through the cracks.51
Here I present my attempt to summarize, in premise form, the Probabilistic
Account.
P1) For a class of choices (of which I have made many), I have knowledge of a random and representative sample of mental states that influence the choices in that class.
P2) The samples of mental states of which I am aware do not provide a deterministic explanation of my choices.
P3) The fact that I lack evidence that there are deterministic causes of these decisions provides probabilistic evidence that these decisions are not determined.52
51 Sober (2009) uses the analogy of a fish net with holes so large that fish under a certain length are never caught. 52 Adapted from Nichols (2015, 48-49) 109
In order to reason this way, one need not assume access to all of the mental states that
influence choice. However, it does assume that the mental states to which one has access
are a random and representative sample of the total choice-influencing mental states.53
3.2 Objections to the Probabilistic Account
It is unclear from Nichols’ presentation whether the sample of mental states of
which I am aware in (P1) is a subset of the total mental states influencing a given choice,
or the complete set of proximal influences on a given choice, which is itself a sample of a
wider class of choices. As I will show, both interpretations face problems.
On the first interpretation of (P1), we can introspect a random and representative
sample of causally influential mental states to see if they determine a choice, and then
infer that the total set of causally influential mental states did not determine the choice.
This involves viewing the mechanism of choice as what Richard Levins calls an
“aggregate system.” In aggregate systems, properties of the whole are statistics of the
properties of individual parts of the system (Levins, 1970, 76). In “composed systems,”
in contrast, the properties of the whole depend on the specific ways in which their
component parts are combined. In composed systems, you cannot make justifiable
statistical inferences from parts to wholes because the interactions between parts are
crucial to the properties of the whole, and these interactions are not necessarily identical.
The problem for the Probabilistic Account is that the psychological processes
leading to choices look more like composed systems than aggregate systems. Suppose I
53 Random and representative are not the only assumptions needed for the Probabilistic Account to work. Nichols also points out that (i) it must be assumed that you focus on a restricted set of decisions (he uses difficult decisions), (ii) that you never actually experience this set as determined, and (iii) your introspection actually reveals evidence of the absence of causal determinism. While these are all potentially problematic, in this paper I focus only on the random and representative aspects. 110 want to know whether a long argument with many premises is valid or not. It is not enough for me to take a random and representative sample of the premises in order to check whether the argument as a whole is valid. The problem is that validity involves the relationship between sets of premises and a conclusion, not between individual premises and a conclusion. Sampling individual premises, or even subsets of premises, doesn’t tell us enough about the set of premises as a whole and its relation to the conclusion.
Likewise, sampling a subset of the many mental states that speak in favor or against a particular choice doesn’t tell us enough about the relationship between the mental states taken as a whole and its relation to the choice to justify inferring (P3).
It could be argued that the relevant hypothesis for the Probabilistic Account is that for some choices, introspection reveals a complete set of proximal mental states. 54 This is to adopt the second interpretation of (P1). Because the sample is just the complete set of proximal mental states, there is no concern about the sample failing to reflect properties of the composed system as a whole. Notice, however, that this move collapses the Probabilistic Account into the Deductivist Account. The problems that were raised for (D2) now re-arise. It requires that for any given choice, we assume to have access to the explicitly causal features of a complete set of proximal influences, but there is little evidence that we accept such a strong principle.
Although these interpretative problems are worrisome, there is a further problem for the Probabilistic Account: it fails to accommodate No Overgeneralization. One reason for Nichols’ optimism with regards to the Probabilistic Account largely comes
54 Thanks to Shaun Nichols for pointing out this possibility. 111 from the fact that we have good evidence that human beings learn using probabilistic inferences more generally. In fact, models that invoke probabilistic inference making are pervasive throughout cognitive science (Tenenbaum, 2007; Gerken 2010; Romberg and
Saffran, 2010; Denison, Reed & Xu, 2013). Yet this proliferation of probabilistic explanations also suggests that the overgeneralization worry may once again be around the corner. If we already have good reason to think that probabilistic inference is going on in other domains, then why don’t we see belief in indeterminism popping up in those other domains?
In terms of a probabilistic inference, if we do not find evidence of sets of states that determine choice, then we should not believe that the probability of the effect (my choice) given some prior causal set (proximal mental states) is 1. Instead, we should believe the probability is less than 1. But this is not unique to the case of choice; on probabilistic inferential models we almost never assign probabilities of 1 to causal outcomes.55 Thus, the format of the Probabilistic Account suggests that we should expect folk commitment to indeterminism in a great many areas that people do not obviously think are indeterministic. For example, some people may have beliefs of which they cannot find a deterministic cause among a sample of the mental states that enter into the formation of the belief. And yet there does not seem to be widespread acceptance of the view that belief formation, or even a substantial subset of belief formation, is indeterministic.56
55 Exceptions include logical or mathematical necessities, but these are not clearly causal cases. 56 My experience with the religious beliefs of friends and colleagues are often of this form. They say things like, “I can’t say why, I just believe in God.” 112
What is needed to avoid overgeneralization is some account of an asymmetry such that we probabilistically infer indeterminism in the case of choices but not in other cases. Strangely, despite the significant role overgeneralization plays in framing Nichols’ critiques of opposing views, he makes no attempt to defend the Probabilistic Account from such worries. Acceptance of (D2), supported by an IOED, was used to justify such an asymmetry in the Deductivist Account. Perhaps Nichols sees IOEDs playing a similar role for the Probabilistic Account?
We must proceed with caution when trying to apply the IOED literature to the
Probabilistic Account. Certainly, if I believe that I have access to all the proximal causes of my choice then it is reasonable to believe that the proximal causes to which I have access are a representative sample. But one of the most appealing features of the
Probabilistic Account is that it doesn’t require the implausible belief that I have access to all the proximal causes of my own choice. If IOEDs are to be of any help, it would be by providing support for (P1), the assumption that introspection of the proximal influences on choice is a random and representative sample. Perhaps, as the result of an IOED, I believe that introspection of the proximal influences on my choice provides a random and representative sample. Thus, I believe something like (P1) in the case of choices but not in other cases.
I am suggesting here that the IOED literature could be used in the Probabilistic
Account in much the same way it is pressed into service in the Deductivist Account. But the details are not entirely obvious. More to the point, if my objections to the use of
IOEDs from section 2.3 work, then these problems will re-arise for the Probabilistic
113
Account. So even if Nichols is implicitly assuming that the IOED literature prevents
overgeneralization in the Probabilistic Account, the position remains untenable, given
Spinozan strategy adequacy conditions.
Are there other candidates to mark an asymmetry between inferences regarding
indeterminism in the case of choices and other cases? Terry Horgan suggests that the
very concept of free will biases our interpretation of experiences of choosing (Horgan
2012; 2011a). Thus, although he does not think that we experience our choices as indeterministic, he argues that we interpret our experiences as having indeterministic contents as the result of an interpretative bias. Since we do not bring the concept of free will to bear on our interpretations of other experiences, we do not infer indeterminism in cases other than choice.
Introducing an interpretative bias has the added benefit of addressing a lingering worry for the Spinozan strategy. Even if one successfully explains why people believe their choices to be indeterministic following the Spinozan strategy, it is not obvious why people seem to believe that they experience their choices as indeterministic. Why would so many people describe their experience of choice as indeterministic, if they don’t have an experience of choice as indeterministic? Previously, the straight strategy had a leg up here. According to the straight strategy, people believe they experience their choices as indeterministic precisely because they do experience their choices as indeterministic. In short: the straight strategy can model belief about the experience of choice on belief about our experiences of almost everything else. By introducing an interpretative bias, the Spinozan strategy can explain why people believe they experience choice as
114 indeterministic even though they don’t have an experience of choice as indeterministic.
They believe their experience of choice is indeterministic because there is a bias that affects their interpretation of their experiences.
The proposal of an interpretative bias is intriguing, but it is not of much help to the Probabilistic Account. It is the interpretive bias, and not probabilistic inference, that explains the belief in indeterministic choice. Thus, the Probabilistic Account becomes explanatorily inert. Furthermore, the interpretative bias account is cold comfort for the
Spinozan strategist. Philosophers who use the Spinozan strategy are united in their rejection of a certain phenomenological claim: the claim that we experience choice as indeterministic. This rejection has seemed a powerful reason not to adopt the straight strategy. But this rejection seems unmotivated if, as the result of some bias, we always interpret our experience of choice as indeterministic. On what grounds could one possibly reject the phenomenological claim if we always interpret our experiences in line with that very claim?
4. Conclusion
As normally presented, the Spinozan strategy is incomplete. It holds that we infer that choice is indeterministic from the lack of an experience of choice as causally determined. But the lack of an experience of causal determination is extremely pervasive. Thus the Spinozan strategy is liable to overgeneralize without some account of an asymmetry between the case of choice and other cases. The Deductivist Account and the Probabilistic Account are explicitly crafted to address this overgeneralization worry. I have argued here that even these most sophisticated versions of the Spinozan
115 strategy fail to avoid overgeneralization while remaining faithful to the fairly benign adequacy conditions of Scientific Plausibility and Reasonableness.
In light of this failure, as well as the poor historical track record, there is mounting pressure against the Spinozan strategy. We should be pessimistic that other versions of the Spinozan strategy will fare much better.
Despite this mounting pressure, it is not yet time to give up on the Spinozan strategy. In the eyes of many philosophers, accepting that there is a widespread experience of choice as causally indeterministic is a bitter pill to swallow. To them, the alternative straight strategy gets the phenomenology wrong. By way of conclusion, then, it is valuable to map the remaining options available to these Spinozan strategists if they are to avoid overgeneralization.
The Spinozan strategy is a particular example of a broader form of inference. At the end of the day, the Spinozan strategy seeks to provide grounds that justify taking an absence of evidence (the lack of an experience of determinism) as evidence of absence
(the absence of determined choice). Thus in order to avoid overgeneralization, the
Spinozan strategist must identify some asymmetry such that the inference seems prima facie plausible in the case of choices but not in the case of other phenomena. There are three options for locating such an asymmetry.
The asymmetry could be located in the inferential processes that give rise to the belief in indeterministic choice. That is, it could be the case that there exists a specialized inferential process that is specific to the case of choice. On this option, we believe choices are indeterministic, but not other phenomena like headaches, because the
116 inferential process deployed in the case of choices is different from that in the case of headaches. This possibility was explored briefly in the discussion of an encapsulated
IOED in the context of the Deductivist Account. As I argued in that discussion, there is currently no scientific motivation to think that any such specialized process uniquely obtains in the case of choice. Nevertheless, it is an empirical question worth pursuing.
A second option for the Spinozan strategist avoids overgeneralization by positing background assumptions that specifically constrain a more general inferential process in the case of choice. (D2) is an assumption of this kind; it operates on a general inferential process specifically in the case of choice, but not in other cases like headaches. While I have argued here that (D2) is not very plausible, there could be other assumptions that do similar work. For example, suppose there was a widely held background assumption of the following kind, “For choice, unless determinism is explicitly detected, treat absence of evidence as evidence for absence.” If this kind of background assumption obtained, then a general inferential process could give rise to the belief in indeterministic choice without overgeneralizing. One of the lessons learned in the discussion of Nichols’ accounts is that these assumptions must be fairly strong and domain specific if they are to meet Spinozan strategy adequacy conditions. As such, they tend to collapse into nativist assumptions, and are not currently supported by our best science. Nevertheless, whether a background assumption of this kind obtains is an empirical question worth pursuing further.
There is one more option available to the Spinozan strategist. The previous two options place constraints on the nature of the inferential process involved in order to
117 block overgeneralization. Instead, the Spinozan strategist might change the inputs available to the inferential process in order to block overgeneralization. More specifically, if the experiences available in the case of choices are different from those available in other cases, then this could block the overgeneralization worry.
To flesh the story out in serious detail would go far beyond the scope of this paper, but it is worth making a few comments to indicate how the story might go. The
Spinozan strategist claims that choices are not experienced as causally determined. This claim can be read in multiple ways. On one reading, they may be denying that choice is experienced as determined, while still allowing that choice is experienced as influenced by causal forces. Nichols and others adopt this interpretation. On the other reading, the
Spinozan strategist might deny that choice is experienced as caused at all.
This latter reading opens up interesting possibilities for the Spinozan strategist.
Certainly, we sometimes do seem to experience actions as caused, as when anger boils over, causing one to lash out in rage. However, in the deliberative context normally associated with making choices, it seems that we often lack an experience of choice as caused at all, deterministic or otherwise. That is, the contents of introspected mental states in the deliberative context are often not presented as causally efficacious. Rather, we tend to represent the contents of our mental states in the deliberative context as reasons that favor choosing a particular way. This marks an important difference between choices and other cases where the objects of experience may be presented as candidate recipients of causal force, and so could block overgeneralization. We might infer that the proximal influences on choices revealed via introspection are not causal
118 determiners of choices because choices are not the right kinds of things to be caused at all, at least as presented in experience. More detail is obviously needed to show that this third option can successfully avoid overgeneralization while remaining faithful to
Scientific Plausibility and Reasonableness, though I am optimistic that it is an avenue worth pursuing. Whichever of these three options Spinozan strategists pursue going forward, however, they would do well to keep overgeneralization worries in their sights at all times.
119
References
Ahn, W, and C Kalish. (2000). The Role of Mechanism Beliefs in Causal Reasoning. In
Explanation and Cognition, edited by Frank Keil and R.A. Wilson. Boston, MA: MIT
Press.
Alston, W. P. (2002). Sellars and the “myth of the given.” Philosophy and
Phenomenological Research, 65(1), 69–86.
Anscombe, G. E. M. (1958). Intention (2nd ed.). Oxford: Basil Blackwell.
Bargh, J., Chen, M., & Burrows, L. (1996). Automaticity of Social Behavior. Journal of
Personality and Social Psychology, 71, 230–244.
Bayne, T. (2011). The Sense of Agency. In F. MacPherson (Ed.), The Senses: Classic
and Contemporary Philosophical Lessons. Oxford University Press.
Blakemore, S. J., Wolpert, D. M., & Frith, C. D. (2002). Abnormalities in the awareness
of action. Trends in Cognitive Sciences, 6(6), 237–242.
Boyle, J. M., Grisez, G., & Tollefson, O. (1976). Free Choice: A Self Referential
Argument. Notre Dame, Ind.: University Press of Notre Dame.
Bratman, M. E. (2009). Intention, belief, and instrumental rationality. In D. Sobel & S.
Wall (Eds.), Reasons for Action (pp. 13–36). Cambridge University Press.
Butterfield, J. (2005) Determinism and Indeterminism. In Routledge Encyclopedia of
Philosophy, edited by E. Craig. London: Routledge. 120
Campbell, C. A. (1967). In Defence Of Free Will, With Other Philosophical Essays.
London: Allen &Amp; Unwin.
Carruthers, G. (2012). The case for the comparator model as an explanation of the sense
of agency and its breakdowns. Consciousness and Cognition, 21(1), 30–45.
Caruso, G. (2012). Free will and consciousness: a determinist account of the illusion of
free will. Lexington Books.
Caruso, G. (2015). Free will eliminativism: reference, error, and phenomenology.
Philosophical Studies, 172, 2823–2833.
Chisholm, R. M. (1957). Perceiving: A Philosophical Study. Ithaca, New York: Cornell
University Press.
Chisholm, R. M. (1980). A Version of Foundationalism. In Wettstein, Uehling, & French
(Eds.), Midwest Studies in Philosophy V (pp. 543–564). Minneapolis: University of
Minnesota Press.
Christensen, D. (1994). Conservatism in Epistemology. Nous, 28(1), 69–89.
Davidson, D. (1963). Actions, Reasons, Causes. Journal of Philosophy, 60(23), 685–700.
Deery, O. (2014). The Fall From Eden: Why Libertarianism Isn’t Justified By
Experience. Australasian Journal of Philosophy, 93(2), 319–334.
Deery, O. (2015a). Is agentive experience compatible with determinism? Philosophical
Explorations, 18(1), 2–19.
Deery, O. (2015b). Why people believe in indeterminist free will. Philosophical Studies,
172(8), 2033–2054.
121
Deery, O., Bedke, M. S., & Nichols, S. (2013). Phenomenal Abilities: Incompatibilism
and the Experience of Agency. In D. Shoemaker (Ed.), Oxford Studies in Agency and
Responsibility (pp. 126–150). Oxford University Press.
Denison, Stephanie, Christie Reed, and Fei Xu. (2013). The Emergence of Probabilistic
Reasoning in Very Young Infants: Evidence from 4.5- and 6-Month-Olds.
Developmental Psychology 49 (2): 243–49.
Dennett, D. (1984). Elbow Room. The MIT Press.
Descartes, R. (1984). The Philosophical Writings of Descartes: Volume II. (J.
Cottingham, R. Stoothoff, & D. Murdoch, Eds.). Cambridge: Cambridge University
Press.
Descartes, R. (1991.) The Philosophical Writings of Descartes: Volume III The
Correspondence. (J. Cottingham, R. Stoothoff, D. Murdoch, and A. Kenny, Eds.)
Cambridge: Cambridge University Press.
Doris, J. M. (2015). Talking to our selves : reflection, ignorance, and agency. Oxford
University Press.
Feltz, A. (2017). Folk Intuitions. In K. Timpe, M. Griffith, & N. Levy (Eds.), The
Routledge Companion to Free Will (pp. 468–476). New York: Routledge.
Fischer, J. M., & Ravizza, M. (1998). Responsibility and Control: A Theory of Moral
Responsibility. Cambridge University Press.
Foley, R. (1983). Epistemic Conservatism. Philosophical Studies, 43(2), 165–182.
Frankfurt, H. G. (1969). Alternate possibilities and moral responsibility. Journal of
Philosophy, 66(3), 829–39.
122
Frankfurt, H. G. (1971). Freedom of the will and the concept of a person. Journal of
Philosophy, 68(1), 5–20.
Gerken, L. (2010). Infants use rational decision criteria for choosing among models of
their input. Cognition, 115(2), pp. 362-6.
Ginet, C. (1997). Freedom, Responsibility, and Agency. Journal of Ethics, 1(1), 85–98.
Haggard, P., & Tsakiris, M. (2009). The Experience of Agency: Feelings, Judgments, and
Responsibility. Current Directions in Psychological Science, 18(4), 242–246.
Harman, G. (1965). The Inference to the Best Explanation. The Philosophical Review,
74(1), 88–95.
Harman, G. (1986). Change in View. MIT Press.
Harris, S. (2012). Free Will. Free Press.
Hieronymi, P. (2011). Reasons for Action. Proceedings of the Aristotelean Society,
111(3), 407–427.
Hobbes, T. (1656/2005). The questions concerning liberty, necessity, and chance clearly
stated and debated between Dr. Bramhall, Bishop of Derry, and Thomas Hobbes of
Malmesbury.
http://quod.lib.umich.edu/e/eebo/A44010.0001.001/1:8.3?rgn=div2;view=fulltext
Hobbes, T. (1841). The English works of Thomas Hobbes of Malmesbury- Vol. V. (W.
Molesworth, Ed.). Bohn. Retrieved from
http://onlinebooks.library.upenn.edu/webbin/metabook?id=hobbesworks
Holton, R. (2009). Willing, Wanting, Waiting. Oxford University Press.
123
Holton, R. (2010). Disentangling the Will. In R. Baumeister, A. Mele, & K. Vohs (Eds.),
Free Will and Consciousness. Oxford: Oxford University Press.
Horgan, T. (2011a). Causal compatibilism about agentive phenomenology. In T. Horgan,
M. Sabates, & D. Sosa (Eds.), Supervenience in Mind. Cambridge, MA: MIT Press.
Horgan, T. (2011b). The Phenomenology of Agency and Freedom : Lessons from
Introspection and Lessons from Its Limits. Humana Mente, (15), 77–97.
Horgan, T. (2012). Introspection about Phenomenal Consciousness: Running the Gamut
from Infallibility to Impotence. In D. Smithies & D. Stoljar (Eds.), Introspection and
Consciousness (pp. 405-421). Oxford University Press.
Horgan, T. E., Tienson, J. L., & Graham, G. (2003). The Phenomenology of First-Person
Agency. In S. Walter & H. Heckmann (Eds.), Physicalism and Mental Causation (pp.
323-341). Imprint Academic.
Huemer, M. (2001). Skepticism and the Veil of Perception. Lanham: Rowman &
Littlefield.
Hume, D. (1748/2011). An Enquiry Concerning Human Understanding. (E. Steinberg,
Ed.) (2nd ed.). Hackett Publishing Co.
Hume, D. (1779/2007). Dialogues Concerning Natural Religion and Other Writings.
Cambridge Univ. Press.
Jackson, F. (1977). Perception: A Representative Theory. Cambridge University Press.
James, W. (1902). The Varieties of Religious Experience: A Study in Human Nature.
Dover Publications.
124
Kapitan, T. (1986). Deliberation and the presumption of open alternatives. The
Philosophical Quarterly, 36(14), 230–251.
Knobe, J. (2014). Free Will and the Scientific Vision. In E. Machery & E. O’Neill (Eds.),
Current Controversies in Experimental Philosophy (pp. 69–85). Routledge.
Kozuch, B., & Nichols, S. (2011). Awareness of Unawareness. Journal of Consciousness
Studies, 18(11–12), 135–160.
Kraut, R. (2001). Metaphysical Explanation and the Philosophy of Mathematics:
Reflections on Jerrold Katz’s Realistic Rationalism. Philosophia Mathematica, 9(2),
154–183.
Lawson, R. (2006). The science of cycology: Failures to understand how everyday
objects work. Memory & Cognition, 34(8), 1667–1675.
Lewis, D. (1972). Psychophysical and theoretical identifications. Australasian Journal of
Philosophy, 50(249–258).
Levins, R. (1970). Complexity. In C.H. Waddington (Ed.), Towards a theoretical
biology, Vol. 3, 67-86. Edinburgh: University of Edinburgh Press.
Lewis, D. K. (1973). Counterfactuals. Blackwell Publishers.
Lipton, P. (2004). Inference to the Best Explanation. Routledge/Taylor and Francis
Group.
Lycan, W. G. (2002). Explanation and Epistemology. In P. K. Moser (Ed.), Oxford
Handbook of Epistemology (pp. 408–433). New York: Oxford University Press.
125
Lycan, W. G. (2013). Phenomenal conservatism and the principle of credulity. In C.
Tucker (Ed.), Seemings and Justification: New Essays on Dogmatism and
Phenomenal Conservatism (pp. 293–305). Oxford University Press.
Magill, K. (1997). Freedom and experience: self-determination without illusions. New
York: St. Martin’s Press.
Malle, B. (1999). How People Explain Behavior: A New Theoretical Framework.
Personality and Social Psychology Review. 3 (1): 23-48.
Malle, B. F., Knobe, J. M., & Nelson, S. E. (2007). Actor–Observer Asymmetries in
Explanations of Behavior: New Answers to an Old Question. Journal of Personality
and Social Psychology, 93(4), 491–514.
Martin, M. (2011). The Principle of Credulity and Religious Experience. Religious
Studies, 22(1), 79–93.
Maund, J. B. (1986). The Phenomenal and Other Uses of “Looks.” Australasian Journal
of Philosophy, 64(2), 170–180.
Mawson, T. J. (2011). Free Will: A Guide for the Perplexed. Bloomsbury Academic.
McCain, K. (2008). The virtues of epistemic conservatism. Synthese, 164(2), 185–200.
Mele, A. (2014). Free Will and Substance Dualism: The Real Scientific Threat to Free
Will? In W. Sinnott-Armstrong (Ed.), Moral Psychology Vol. 4: Free Will and Moral
Responsibility (pp. 195–208). MIT Press.
Mills, C. M., & Keil, F. C. (2004). Knowing the limits of one’s understanding: The
development of an awareness of an illusion of explanatory depth. Journal of
Experimental Child Psychology, 87, 1–32.
126
Moore, J. W. (2016). What is the sense of agency and why does it matter? Frontiers in
Psychology, 7(AUG), 1–9.
Moore, J. W., & Fletcher, P. C. (2012). Sense of agency in health and disease: A review
of cue integration approaches. Consciousness and Cognition, 21(1), 59–68.
Moore, J. W., & Obhi, S. S. (2012). Intentional binding and the sense of agency: A
review. Consciousness and Cognition, 21(1), 546–561.
Nadelhoffer, T., & Matveeva, T. (2009). Positive Illusions, Perceived Control and the
Free Will Debate, Mind & Language, 24(5), 495–522.
Nadelhoffer, T., Shepard, J., Nahmias, E., Sripada, C., & Ross, L. T. (2014). The free
will inventory: Measuring beliefs about agency and responsibility. Consciousness and
Cognition, 25(1), 27–41.
Nahmias, E., Morris, S. G., Nadelhoffer, T., & Turner, J. (2004). The Phenomenology of
Free Will. Journal of Consciousness Studies, 11(7–8), 162–179.
Nahmias, E., and D. Murray. (2010). Experimental Philosophy on Free Will: An Error
Theory for Incompatibilist Intuitions. In Jesus Aguilar, Andrei Buckareff, and Keith
Frankish (Eds.), New Waves in Philosophy of Action, 189–215. Palgrave-Macmillan.
Nahmias, E., Morris, S., Nadelhoffer, T., & Turner 1, J. (2005). Surveying Freedom: Folk
Intuitions about free will and moral responsibility. Philosophical Psychology, 18(5), 561–
584.
Nichols, S. (2004). Folk Concepts and Intuitions: From Philosophy to Cognitive Science.
Trends in Cognitive Sciences, 8(11), 514-518.
127
Nichols, S. (2012). The indeterminist intuition: source and status. The Monist, 95(2),
290–307.
Nichols, S. (2015). Bound: Essays on Free Will and Responsibility. Oxford University
Press.
Nichols, S., & Knobe, J. (2007). Moral Responsibility and Determinism: The Cognitive
Science of Folk Intuitions. Noûs, 41(4), 663–685.
Nisbett, R., & Wilson, T. (1977). Telling More than We Can Know: Verbal Reports on
Mental Processes. Psychological Review 84 (3): 231–59.
O’Connor, D. (1971). Free Will. Doubelday Anchor.
O’Connor, T. (1995). Agent Causation. In Agents, Causes, and Events: Essays on
Indeterminism and Free Will (pp. 173–200). New York: Oxford University Press.
O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:
Oxford University Press.
Open Science Collaboration. (2015). Estimating the reproducibility of psychological
science. Science, 349(6251).
Pink, T. (2004). Free will a very short introduction. Oxford; New York: Oxford
University Press.
Pink, T. (2010). Free Will and Consciousness. In T. Bayne, A. Cleeremans, & P. Wilken
(Eds.), The Oxford Companion to Consciousness. Oxford University Press.
Prinz, W. (2003). How do we know about our own actions? In S. Maasen, W. Prinz, & G.
Roth (Eds.), Voluntary Action: Brains, minds, and sociality (pp. 21–33). Oxford
Univ. Press.
128
Pryor, J. (2005). There is Immediate Justification. In M. Steup & E. Sosa (Eds.),
Contemporary Debates in Epistemology (pp. 181–202). Blackwell.
Rakos, R. F., Steyer, K. R., Skala, S., & Slane, S. (2008). Belief in Free Will:
Measurement and Conceptualization Innovations. Behavior and Social Issues, 17(1),
20–39.
Reid, T. (1788/2010). Essays on the Active Powers of Man. (K. Haakonssen & J. A.
Harris, Eds.). Penn State University Press.
Romberg, A. R., & Saffran, J. R. (2010). Statistical learning and language acquisition.
Wiley Interdisciplinary Reviews. Cognitive Science, 1(6), 906–914.
Rozenblit, L., & Keil, F. (2002). The misunderstood limits of folk science: an illusion of
explanatory depth. Cognitive Science, 92, 1–42.
Sarkissian, H., Chatterjee, A., De Brigard, F., Knobe, J., Nichols, S., & Sirker, S. (2010).
Is Belief in Free Will a Cultural Universal? Mind & Language, 25(3), 346–358.
Sarkissian, H., Chatterjee, A., De, F., Knobe, J., Nichols, S., & Sirker, S. (2010). Is Belief
in Free Will a Cultural Universal ? Mind and Language, 25(3), 346–358.
Searle, J. R. (2003). Rationality in action. Cambridge, Mass.; London: MIT.
Searle, J. R. (2007). Freedom and Neurobiology: Reflections on Free Will, Language,
and Political Power. Columbia University Press.
Siegel, S. (2009). The Visual Experience of Causation. The Philosophical Quarterly,
59(236), 519–540.
129
Sloman, S. A., & Barbey, A.K. (2016). The Basic Assumptions of Intuitive Belief: Laws,
Determinism, and Free Will. In L. Macchi, M. Bagassi, & R. Viale (Eds.), Cognitive
Unconscious and Human Rationality (1st ed., pp. 167–180). MIT Press.
Smilansky, S. (2000). Free will and illusion. Oxford: Clarendon press.
Smith, M. (1994). The Moral Problem. Cambridge: Blackwell.
Smithies, D. (2016). Reasons and Perception. In D. Star (Ed.), The Oxford Handbook of
Reasons and Normativity (pp. 453–460). Oxford Univ. Press
Sober, E. (2009). Absence of evidence and evidence of absence: Evidential transitivity in
connection with fossils, fishing, fine-tuning, and firing squads. Philosophical Studies,
143(1), 63–90.
Sommers, T. (2007). The Illusion of Freedom Evolves. In D. Ross, D. Spurrett, H.
Kincaid, & G. L. Stephens (Eds.), Distributed Cognition and the Will: Individual
Volition in Social Context (pp. 61–75). MIT Press.
Sommers, T. (2015). Free Will and Experimental Philosophy: An Intervention. In
Handbook of Neuroethics (pp. 273–286). Dordrecht: Springer Netherlands.
Soon, C. S., Brass, M., Heinze, H.-J., & Haynes, J.-D. (2008). Unconscious determinants
of free decisions in the human brain. Nature Neuroscience, 11(5), 543–545.
Sosa, E. (1980). The Raft and the Pyramid: Coherence versus Foundations in the Theory
of Knowledge. Midwest Studies in Philosophy, 5(1), 3–26.
Spinoza, B. de, & Hampshire, S. (1677/2005). Ethics. (E. Curley, Trans.). Penguin
Classics.
130
Sripada, C. (2016). Free will and the construction of options. Philosophical Studies,
173(11), 1–21.
Stich, S., & Nichols, S. (2003). Folk psychology. In S. Stich & T. A. Warfield (Eds.), The
Blackwell Guide to Philosophy of Mind (pp. 235–255). Oxford: Basil Blackwell.
Strawson, G. (1986). Freedom and Belief. Oxford: Clarendon press.
Swinburne, R. (2001). Epistemic Justification. Oxford University Press.
Swinburne, R. (2013). Mind, brain, and free will. Oxford: Oxford Univ. Press.
Taylor, R. (1992). Metaphysics (4th ed.). Englewood Cliffs: Prentice-Hall.
Tenenbaum, J. B., Griffiths, T., and Kemp, C. (2006). Theory-Based Bayesian Models of
Inductive Learning and Reasoning. Trends in Cognitive Sciences 10 (7). van Fraassen, B. C. (1980). The Scientific Image. Clarendon Press. van Inwagen, P. (2000). Free Will remains a mystery. Philosophical Perspectives, 14, 1–
20.
Vargas, M. (2013). Building Better Beings: A Theory of Moral Responsibility. Oxford
Univ. Press.
Wegner, D. M. (2003). The Illusion of Conscious Will. MIT Press.
Williamson, T. (2004). Philosophical “Intuitions” and Scepticism about Judgement.
Dialectica, 58(1), 109–153.
Wimsatt, W. C. (2006). Aggregate, composed, and evolved systems: Reductionistic
heuristics as means to more holistic theories. Biology and Philosophy, 21(5), 667–
702.
131
Appendix A: Thinking Reasonably about Indeterministic Choice Beliefs
One important component for libertarian views of free will is that free agents are
not causally necessitated by prior states of affairs. Recently, philosophers and cognitive
scientists have sought to explain why the belief that choices are not causally necessitated
by prior conditions seems so prevalent among non-philosopher laypersons (Holton 2009;
Nichols 2015). In doing so, they seek a debunking explanation such that the belief that
choice is not causally necessitated should not provide reason to accept libertarian views
of free will. Many of these philosophers adopt what I have called the Spinozan strategy.
According to the Spinozan strategy, the folk belief in the indeterminacy of choice is
inferred from introspection on the phenomenology of first-person agency.57 More
specifically, Spinozan strategists argue that we do not experience choices as causally
necessitated, but we infer that choices are not causally necessitated from this lack of an
experience of causal necessitation.
I have argued elsewhere that the best extant versions of the Spinozan strategy fail
to satisfy their own conditions of adequacy. Very briefly, Spinozan strategists have
failed to explain why we do not form indeterministic beliefs in cases other than choices
57 In this paper, I use the terms “indeterministic”, “not causally necessitated”, and “not causally determined” interchangeably, though there are some uses of the terms where they come apart. 132
where we also lack an experience of causal necessitation. For this reason, the explanation
provided by the Spinozan strategy overgeneralizes.
I maintain here that extant Spinozan strategies work with an impoverished set of
resources. They have failed to adequately acknowledge that the belief that human
behavior (i.e. choosing) is not causally necessitated arises paradigmatically in the context
of practical deliberation, where people often experience their own behavior as rational in
light of the reasons that speak in favor of a particular action, and not merely in causal
terms. In this paper, I explore alternatives to the Spinozan strategy that rely on the
experience of reasons for acting, which I call the reasons-based accounts (RBAs). Since this phenomenology arises paradigmatically during practical deliberation, it could ground an explanation of the widespread belief that choices are not causally necessitated that avoids overgeneralization.
1. The Spinozan Strategy and its Problems
According to the Spinozan strategy, introspectively available agentive experience of choosing has the following feature:
1. Experience does not present one’s behavior as causally necessitated.58
Thus, when we attend to the experience of first person agency, we do not find causal necessitation as part of its phenomenological contents. There is a subtle distinction
58 This formulation of the Spinozan strategy borrows from Horgan (2012) 133
between this feature of the phenomenology of first-person agency and the following feature:
2. Experience presents one’s behavior as not causally necessitated.
The Spinozan strategy holds that (1) is a prominent feature of the phenomenology
of first-person agency while (2) is not. They argue that it is not even clear what it could
be like to experience one’s behavior as not causally necessitated (Holton 2009, 168).
Nevertheless, in some cases, the absence of evidence can provide evidence of absence, provided one has appropriate background assumptions. On the assumption that, were my behavior causally necessitated then (2) would be a feature of my phenomenology, I could infer that my choices are not causally necessitated on the basis of (1). For this reason,
Spinozan strategists have argued that we assume we have complete access to the influences on our choices, such that the absence of evidence of causal necessitation becomes evidence of absence (Nichols 2015).
The main worry for the Spinozan strategy is that, if we hold such a background assumption in the case of choices, it is hard to see why we wouldn’t hold it in other cases as well. As such, the strategy is liable to overgeneralize. Consider the following analogue to (1).
1.* Experience does not present the phenomena as causally necessitated
134
I submit that (1*) is a common feature of experience in a good many cases. Suppose that
I see a thrown ball land in a potted plant, followed immediately by the lights in the room
turning off. My visual experience presents the events in quick succession, but does not
present them as causally connected (Siegel 2009, 526). In this case, my visual experience
does not present the lights turning off as causally necessitated by the throw of the ball.
Contrast this with the following subtly different feature:
2.* Experience presents the phenomena as not causally necessitated
Despite the apparent widespread occurrence of (1*), and the lack of (2*), people do not
regularly believe that the turning off of the light was indeterministic on the basis of
introspective attention to visual experiences like that of the ball, the potted plant, and the
lights turning off. It may be pointed out that we don’t infer indeterminism in this case
because we already assume that lights only turn on or off when something causes them to
turn on or off. Inferring indeterminism would go against previously held background
causal assumptions, so we assume there must be a hidden cause of which we are
unaware. But this fails to respond to the overgeneralization worry, as it is unclear why
we would have background causal assumptions in other cases, but not when introspecting
first-person agentive phenomenology of choice, unless we already assume that choices
are not causally necessitated.59
59 Despite the overgeneralization worry, the Spinozan strategy is at the same time surprisingly narrow in its scope. The Spinozan strategy is presented in the context of a debunking explanation for why aspects of first-person agentive phenomenology are interpreted as having libertarian satisfaction conditions. Yet the Spinozan strategy only accounts for beliefs resulting from one aspect of first-person agentive 135
The above discussion suggests that, in order to avoid overgeneralization, the
account should explain why indeterministic beliefs tend to arise only in the case of
choices. As a consequence, Spinozan strategists have tended to implicitly adopt the
following condition of adequacy.
A. No Overgeneralization: Explanations of folk beliefs ought not to overgeneralize.
Contemporary Spinozan strategists additionally accept two further conditions of
adequacy:
B. Scientific Plausibility: All else being equal, to the extent an explanation of folk
belief avoids the attribution of obviously unreasonable inferences, it is to be
preferred over those that attribute less reasonable inferences, and
C. Reasonableness: One’s explanations of folk beliefs should be plausible in light of
our best extant science.
These conditions pull in opposing directions, such that even the best versions of the
Spinozan strategy fall short of satisfying all three. While there is scientific evidence for a
vague assumption of introspective transparency, it is not strong enough to block
overgeneralization worries (Kozuch and Nichols, 2011). In contrast, if the assumption is
strengthened to block overgeneralization, it looks increasingly less reasonable and less
scientifically plausible.
phenomenology: the lack of causal necessitation. While the Spinozan strategy may explain the belief that behavior is not causally necessitated on the basis of first-person introspection, it remains silent on other aspects of freedom usually attributed to first-person agentive phenomenology. These other aspects include (but are not limited to): the sense that choices are “up-to-me,” the sense that I am a source of my choice, the sense that I could have done otherwise (all other facts about me held constant), etc. (Nadelhoffer, et. al, (2014); Horgan (2011). By not addressing these other aspects of first-person agentive phenomenology, the Spinozan strategy provides in principle only a partial debunking of apparently widespread beliefs. 136
The Spinozan strategy focuses on how experience does not present the experience of choice: we do not experience our choosing as causally necessitated. It has surprisingly little to say about how we do experience choice. In light of the difficulties faced by the Spinozan strategy, I propose here to explore a collection of alternative approaches that take the phenomenology of choice as a starting point, which I collectively call reasons-based approaches (RBAs). In the deliberative context normally associated with making choices, it seems that we often lack an experience of choice as caused at all, deterministic or otherwise; that is, mental states in the context of practical deliberation are often not presented as causally efficacious. Rather, we tend to represent the contents of our mental states in the deliberative context as reasons that favor choosing a particular way. This marks an important difference between choices and other cases where the objects of experience may be presented as candidate recipients of causal force, and so could block overgeneralization. RBAs try to harness this insight in order to explain the widespread belief that choice is indeterministic.
Unlike the Spinozan strategy, RBAs do not rely on (1), the claim that experience does not present one’s behavior as causally necessitated, as a premise in an inference that concludes with the belief that choices are indeterministic. Instead, the fact that (1) is a widespread feature of first-person agentive phenomenology is merely an enabling condition for inferring that choice is indeterministic, according to RBA. Additionally,
RBAs exploit the fact that first-person agentive phenomenology presents choices as rational in light of the reasons for which they are performed.
137
While RBAs reject the Spinozan strategy, they share many guiding principles.
RBAs agree with the Spinozan strategy that (1), but not (2) above, is a feature of the first-
person agentive phenomenology of choice, and that this fact is important for explaining
the widespread belief in indeterministic choice. RBAs also agree with the Spinozan
strategy that, as much as possible, an explanation of the widespread belief in
indeterministic choice ought to avoid overgeneralisation, it ought to rely on an inference
that is prima facie plausible, and it ought to be plausible in light of our best science.
2. Choosing on the basis of Reasons
The purpose of the following section is to motivate the following common sense
claim: in the context of practical deliberation, we tend to experience, explain, and think
about our choices in terms of the reasons that rationalize them. According to the
traditional Anscombian picture, when we ask the question, “Why are you doing that?”,
we are asking the agent to produce the reasons that rationalize their action (Anscombe
1958, 11). As such, people often (though not always) explain actions in ways that make
the action appear rational on the basis of one’s reasons (even if we never actually choose
and perform complex actions on the basis of rational reflection). On the classic example,
when someone asks, “Why did Teresa open the fridge?”, the answer cites reasons for
acting, often in the form of a belief-desire pair: 1) because she desired a beer and 2) because she believed there was beer in the fridge.60 The core idea here is that we often
explain actions, both our own as well as the actions of others, by appealing to prior
60 Davidson (1963) popularized accounts of this general form, though he did not clearly endorse the reflective element. 138
mental states of the agent; more specifically, we attribute mental states that would rationalize the agent’s action.
According to this traditional characterization, folk explanatory practices often
invoke reasons in the form of prior mental states that rationalize an action. Reasons in
this context are not merely facts or mental states that count in favor of performing an
action from the perspective of a well-informed observer. Rather, they are motivating
reasons: reasons that count in favor of performing an action in the eyes of the agent
(Smith 1994, 95). Thus, when I explain my own action in terms of reasons, I cite the
mental states that I take myself to have consciously considered, and in virtue of which I
acted. Likewise, when we explain Teresa’s behavior in terms of her reasons for acting,
we often (though not always) think that Teresa in some way consciously considered her
desire for a beer, and her belief that there is beer in the fridge, and these considerations
motivate and rationalize her opening the fridge.61
Why would we explain our actions and the actions of others in terms of the
reasons that rationalize those actions? For one, at least in the context of practical
deliberation, we tend to think about our available options in terms of the reasons that
would rationalize them. Choices in the context of practical deliberation are the kinds of
things that are made on the basis of reasons; we judge what we should or should not do
by considering the contents of our mental states. It is a further question, however,
whether we think about choices as caused by those very mental states. To be clear, citing
61 Many philosophers draw a further distinction between motivating and explanatory reasons. (Hieronymi 2011). While this distinction is important, I think it will not affect the present discussion, since the motivating reasons I focus on are also explanatory reasons, so for simplicity I will restrict myself to the normative/ motivating distinction. From here on, I will use the term “reasons” to refer to motivating reasons, unless otherwise specified. 139 prior mental states as influences on behavior does not always involve appeal to reasons.
Prior mental states can explain actions merely by causing them, as when a hypnotized individual is caused to act in otherwise strange and irrational ways. Thus, the influence of prior mental states qua rationalization contrasts with influence qua causation.
Furthermore, reasons-based explanations are not the only way we explain our choices and the choices of others. In addition to reasons and mere causal explanations, we often explain behavior in terms of long-standing character traits (Teresa has always loved beer), in terms of a causal history of reasons (Teresa comes from a family of beer drinkers), in terms of enabling factors available in the environment (All the beer is in the fridge), etc. Nevertheless, in the context of practical deliberation, we think about and explain the influence of our reasons primarily in terms of the degree to which they rationalize our choices.
In addition to explaining and thinking about choices resulting from practical deliberation in terms of the reasons that rationalize them, reasons also play an important phenomenological role in the context of practical deliberation. We often feel the
“weight” of reasons, when choosing between competing options. In the context of practical deliberation, we rarely feel the weight of reasons pulling towards one uniquely rational and required option. Put another way, we feel that our deliberations are guided by ends, or goals, that place constraints on which available options are deemed viable, but within these constraints, there is often more than one option that, if pursued, would bring about one’s ends (Bratman, 2009). Because numerous options would bring about these ends, it often appears that there is no rational requirement to prefer one means of
140
bringing about that end rather than some other means. When there is no means that is
uniquely preferable, given one’s ends, we might say that one’s ends do not rationally
compel a unique course of action.
We might develop an example from Anscombe to illustrate the phenomenology here. Suppose that you have committed to eating wholesome food. In front of you are a
variety dishes that, if eaten, would satisfy your commitment to eating wholesome food.
You must choose which dish to eat. This involves reflection on the various reasons in
favor of choosing each dish. Perhaps Dish A is packed with protein that promotes muscle
growth. Perhaps Dish B contains a variety of unique vitamins. Perhaps Dish C is packed
with easily useable energy. Which particular dish you choose remains open while the
end at which your choice aims remains fixed: to eat wholesome food. In this case, there
is no rational requirement that uniquely picks out one of dishes A-C. Nevertheless, after
some deliberation, suppose you pick a dish (perhaps Dish A).
Now consider the same case, but focus on the phenomenology of the decision-
making process. It appears that, given your commitments and the reasons available to
you, Dishes A-C are all reasonable choices. In light of your commitment to eat
wholesome food, your reasons do not present any of the means available to you as
uniquely necessary means for attaining the end of eating wholesome food. Thus, when
you choose on the basis of your reasons in this case, your reasons do not rationally
compel your choice. This lack of rational necessitation is often expressed in the first-
person agentive phenomenology of choosing. You don’t experience the weight of
reasons pulling you overwhelmingly towards one dish over another.
141
The above description of the phenomenology draws on a long philosophical
tradition of describing the rational influence of ones reasons in causal terms. Hobbes
describes deliberation as a set of scales that lean one way, then the next, when engaging
in practical deliberation (Hobbes 1841, 326). It is common to speak of rational pressure, and the weight of reasons. Although the descriptions involve causal terms, they do not necessarily pick out causal relations between reasons and choices. Rather, they describe the feeling that some options are more rational than others, given one’s reasons. This supports the larger point that our experience tends to present choices as made on the basis of reasons, but remains largely silent when it comes to the causal relations, or lack thereof, that obtain between reasons and choices.
It is important to clarify that I am relying here on folk platitudes regarding practical deliberation, and not providing a philosophy of action per se. Historically, philosophers have had a tendency to exaggerate the degree to which rational reflection guides actions. Indeed, the classic philosophical description of humanity is as the
“rational animal.” This exaggeration bleeds over into philosophical assumptions about folk psychology, where humans are often described as proceeding through life constantly weighing options, considering potential outcomes, and choosing so as to maximize their own goals.62 The thinking is that what makes humans agents is their ability to act on the
basis of reflection upon consciously considered reasons. It sometimes looks difficult to
square this traditional picture of humans as consciously reflective, rational agents with
recent psychological work. Confabulation studies, priming studies, automatisms, and
62 Prominent examples of the kind of view I have in mind are attributable to Aristotle, Mill, Locke, Kant, and the “reflectivist” positions criticized in Doris (2015). 142 many other experimental paradigms in the cognitive sciences have been used to suggest that a good many of our sophisticated actions proceed without the aid of conscious reflection upon reasons (Nisbett & Wilson, 1977; Bargh et. al., 1996; Wegner, 2003).
For present purposes, I remain officially neutral on the metaphysics of reasons and choices in the hopes of bypassing disputes in philosophy and psychology about what it is for an agent to act. It is one question to ask whether the reflective picture of human agency can do the philosophical work it is forwarded to do. It is quite another question to ask whether the reflective picture of human agency resembles how “we,” (“we” the non- philosopher folk and “we” the philosophers not sitting in the metaphysics room gripped by theory), generally tend to think about our own action production (Stich and Nichols,
2003; Lewis, 1972). It is this latter question that is of primary interest for present purposes. According to RBAs, people regularly think about and experience their own choices as being made on the basis of reasons, even if this does not accurately track how choices are actually made. RBAs try to leverage the former fact into an explanation of the widespread belief indeterministic choice.
3. Constructing RBAs
In the previous section, I argued that there is a tendency to experience, think about, and explain, choices resulting from practical deliberation in terms of rationalization by reasons, and not in causal terms. At this point, one might think that
RBAs already have sufficient resources to begin constructing an explanation of the widespread belief in indeterministic choice. The explanation could go as follows:
143
(RBA1) There is a tendency among the folk to experience reasons as prior
mental states that rationalize a given choice. This tendency precludes
experiencing those reasons as causes of a given choice. In which case, in
the context of practical deliberation, where people tend to view choices as
made solely on the basis of reasons, people tend to believe that choices
resulting from practical deliberation are not caused by those reasons, and
so are not caused by anything.
RBA1 suggests that the belief in indeterministic choice results largely from a perceived category mistake. We experience choices as made and influenced by reasons, qua rationalizers, and not caused. So the question of whether choices are causally determined involves a kind of category mistake: choices are not the kinds of things that are the products of causes in the first place. Since choices are not the right sorts of things to be caused at all, they are not causally necessitated. Or so the story goes.
RBA1 faces a version of the overgeneralization worry. According to RBA1, it is because reasons are viewed as standing in a rationalizing relation to the outcome of our choices that it precludes viewing those reasons as causes of our actions. But similarly, theoretic reasons stand in a rationalizing relation to the beliefs that result from theoretical deliberation. Given this similarity, it would seem RBA1 would predict a similar preclusion of causal relations in the case of theoretical deliberation. And yet, there does not appear to be widespread belief in the claim that beliefs are formed indeterministically.
In fact, there seems to be general consensus that one cannot choose what one believes. In
144
some sense, one must believe that which they take themselves to have the most reason to
believe.63
Admittedly, there are many differences between practical and theoretical
deliberation that one might try to appeal to in order to avoid overgeneralization here. For
one, practical deliberation concludes in a choice (or on some accounts, an intention)
while theoretical deliberation concludes in the formation of a belief. By the lights of
RBA1, however, it is entirely unclear why these sorts of difference would matter for the
purposes of forming beliefs about indeterminism. Provided that reasons are interpreted
as influencing beliefs by rationalizing them in the theoretic case, it is hard to see why
RBA1 would not predict, contrary to fact, that there would be widespread belief in the
indeterminacy of theoretic belief formation. Thus, RBA1 may not meet the requirements
of No Overgeneralization.
More worrisome for RBA1, it is hard to see why the mere fact that experience
presents reasons as rationalizing one’s choice should preclude the possibility of causal
relations between reasons and choices. Indeed, though experience in many cases does not present reasons as causally influencing choices, there is empirical support to suggest that the folk concept of practical reason includes a causal component (Malle 1999; 2007).
Even if reasons are not experienced as causal influencers, people seem to think that, at least sometimes, reasons do causally influence some choices. It just doesn’t look scientifically plausible that choices are widely viewed as existing independently of the larger causal order.
63 This is a common response to the implausibility of Pascal’s wager. We can’t choose whether to believe in God. 145
We can look for additional resources by starting where RBA1 faltered. People tend to experience their reasons as merely rationalizing their choices, but they also tend to conceive of reasons as being causally efficacious; they believe that reasons are rationalizing causes. This suggests that people see reasons as playing a dual role in practical deliberation: rationalizing and causing. Perhaps the belief that choice is indeterministic could result from conflating the role of reasons as rationalizers presented in experience with the background assumption that reasons are causally efficacious. As a
result, the more one feels rationally compelled to choose a certain way, the more one is
inclined to judge that they feel causally compelled to choose a certain way.
To flesh out this possibility, return once again to the phenomenology of choosing
between dishes with the end in mind of eating more healthy food. When we reflectively
attend to the decision-making process, we find that the phenomenology often (though not
always) includes the following feature:
A. Experience presents one’s choice as not necessitated by reasons
The phenomenology of choice presents various options as rationally consistent with one’s
prior commitments. Put another way, in view of your goals, any of the options appear
open to you. This is just a way of describing the fact that Dishes A-C each appear
rationally viable, given your commitments.64
64 See Sripada (2016) for discussion of this phenomenology. There may be worries, a la Sartre’s example of Pierre not being in the café, how experience could ever represent a negation. While the story about how such negation is possible may be tricky, I take it as fairly uncontentious that there is a difference between 146
The experience of rational necessitation (and lack thereof) comes in degrees.
Sometimes, when one’s reasons seem to overwhelmingly favor one option over the other, we feel rational pressure to choose that option. That is, were I to fail to choose this option, I would be doing something highly irrational (Consider Martin Luther who says,
“Here I stand, I can do no other.”) Contrasting the experience of rational pressure felt in these cases can help isolate the feeling found in many other cases, where one feels that one’s choices are not necessitated by reasons.
The degree to which one’s choices feel necessitated by reasons depends not only on the options that appear available at the time, but also the goals and ends one hopes to satisfy by way of that act of choice. If the goal is to eat more wholesome food, then perhaps dishes A-C will all appear equally viable options. But when the more specified goal of eating tasty food is considered, then the available options may be reduced, leading to a greater experience of rational necessitation.
Is it ever the case that we feel truly, 100% rationally necessitated when engaging in an act of choice? I suspect not. This is due to the fact that, as humans, we have a great capacity for introducing new and alternative goals that complicate the picture (Sripada
2016). Decisions are rarely made with respect to a single end or goal. Rather, they are made within the context of numerous goal frames (Baars 1993). Given that the “goal posts” can shift, so to speak, it rarely looks like there is one unique, rationally necessary course of action one must choose. But this observation only reinforces the point that, in
the lack of an experience of Pierre being in the café and an experience of the lack of Pierre being in the café, and that both experiences are possible. 147
some cases, reflective attention on the first-person agentive phenomenology of choice presents the choice as not rationally necessitated.
According to (A), the experience of choosing often presents your actions as not necessitated by your reasons. Additionally, people tend to conceive of reasons as causes of their actions. So it might seem appealing to judge that since reasons did not necessitate your choice, and reasons are causes, that causes did not necessitate your choice. This line of reasoning could lead one to conflate (A) above with the subtly different following feature:
B. Experience presents one’s behavior as not necessitated by causes
Since reasons influence choices both rationally and causally, the conflation of (A) with
(B) is surprisingly simple, given prior intuitive commitments. It involves conflating the influence of the contents of a mental state in virtue of which a choice is more or less rational with the causally efficacious mental state in which those contents are embedded.
While this approach may sound attractive, working out the details of the proposal raises numerous obstacles. We can formulate the present proposal more formally as follows:
(RBA2)
1. Experience presents my choice as not necessitated by reasons.
2. Reasons are causes.
3. Therefore, experience presents my choice as not necessitated by causes.
148
4. Therefore, choices are not necessitated by causes.
(1) is supported by reflection on the first-person agentive phenomenology of choice. (2)
is supported by the fact that people tend to conceive of reasons in the context of practical
deliberation as rationalizing causes. However, in order for (3) to seem reasonable on the
basis of (1) and (2), we must interpret (2) as an identity claim to the effect that all and
only reasons are causes. This identity claim is clearly false; there are innumerable kinds
of causes that are not reasons. RBA2 seems to fail the adequacy condition of
Reasonableness on the grounds that (2) is obviously false.
Perhaps RBA2 can be rehabilitated. In the prior context of practical deliberation,
and only in this context, it is conceivable that people constrain their consideration of the
causal influences on choices to those prior mental states that also rationalize behavior. In
which case, premise (2) might appear acceptable, given the context. In which case, one’s
reasons are the only considered influences on choice, causal or otherwise. In the context
of practical deliberation, then, premise (2) appears reasonable, since causal influences
other than consciously considered reasons are ruled out by the context.65
The rehabilitated form of RBA2 suggests that, in the context of practical
deliberation, people tend to view causes as all and only reasons. One upshot of this
strategy is that, outside of the context of practical deliberation, we should not expect
people to accept (2). As a result, we should expect a decreased belief that choices are not
causally necessitated for choices made outside the context of practical deliberation. At
65 This suggestion need not require a standing assumption of introspective access to all of the causal influences on choices. Rather, invoking a reasons-based explanation restricts the explanatory candidates to reasons, such that the further question of whether a given choice is causally necessitated is also restricted to this class of reasons. It is within this prior explanatory context that the degree of rational necessitation is then conflated with causal necessitation. 149
least at first glance, this prediction does not appear unreasonable. Suppose Ashley,
unlike Teresa, is a known alcoholic. We explain her choice to grab a beer from the fridge
by appealing to her longstanding alcoholism, rather than any process of practical
deliberation. Since her behavior is not the product of practical deliberation, we do not accept (2), and so we accept that there are causal influences on beyond the reasons that rationalize her choice to grab a beer. As a consequence, we are less likely to judge that
Ashley’s choice to grab a beer from the fridge was not causally determined. This seems to fit with common sense views about addiction. The choice to drink is no longer in their hands.
The above proposed revision to RBA2 holds that (2) is only accepted in the context of practical deliberation. Even with this restriction, however, the inference fails
Reasonableness on other grounds. Specifically, RBA2 relies on a fallacy of equivocation. The term “necessitated” picks out rational necessity in the first premise, but causal necessity in the third premise. Believing that reasons are causes is not the same as believing that the rational influence of reasons is identical to the causal influence of reasons. So even if (2) is acceptable, the inference fails to be reasonable.66
As a final worry for RBA2, it looks like it will suffer from a problem akin to the
overgeneralization problem, which we can call the Symmetry problem. Since (2) posits
an identity between reasons and causes, inferences from reasons to causes should go both
directions. RBA2 holds that we infer that choices are not necessitated by causes on the
basis that we experience choices as not necessitated by reasons. But if (2) is accepted in
66 There are further worries about the scientific plausibility of the inference generally, as well as the restricted form of (2). 150 the context of practical deliberation, it would also suggest that if we experience choices as necessitated by reasons, we would infer that the choice is causally necessitated, and vice versa. But generally, it does not appear that if we think something is causally necessitated, we think that it is rationally necessitated as well. So RBA2 seems to make implausible empirical predictions.
I have raised here substantial obstacles for pursuing a fleshed out version of the
RBA. However, we can take lessons from the failures of previous hypotheses in order to direct ourselves towards more promising future hypotheses. RBA1 faces overgeneralization worries, and is not sensitive to apparently widespread beliefs about the causal efficacy of reasons. RBA2 fails largely due to problems with (2), the claim that reasons are identical to causes. Rather than thinking that people equate reasons with causes, then, we might weaken the claim such that people tend to take rational influence as evidence for causal influence. We experience the rational influence of reasons when we engage in practical deliberation. We can feel it exerting rational pressure. Despite this rational pressure, cases of akrasia and plain old stubbornness suggest that, in the majority of cases, this rational pressure does not secure an action. If this felt lack of rational compulsion were taken as evidence that the choice similarly lacked causal compulsion, then perhaps we could explain the widespread belief in indeterministic choice while avoiding overgeneralization. However, further exploration of this possibility must be left for another day.
4. RBAs Contrasted with Other Non-Spinozan Views
151
At first blush, there are apparent similarities between the reasons-based explanations I am exploring here and what is sometimes call the transcendence vision of decision-making (Knobe 2014). On the transcendence view, explanations in terms of
reasons are not explanations in terms of causes. Rather, they are explanations in terms of
an agent that transcends the causal talk of the scientific image. On this transcendence
vision, “…an agent can do something for a reason when the resulting action was freely
chosen and not caused by anything at all.” (Knobe, 2014, 78). On the transcendence
vision, it is a fundamental commitment of folk psychology that human agents are viewed
in this transcendent way and not as components of a scientific causal model.
The transcendence view and RBAs appear similar in that they both argue that the
belief that choices are sometimes not causally necessitated depends closely on folk views
about how reasons influence choices. However, the transcendence vision includes the
additional folk commitment that agents are in some sense “distinct from” or “beyond” the
normal causal order. Thus the transcendence vision requires a prior commitment to folk
dualism in order to explain why people believe choices are not causally necessitated. On
the transcendence vision, reasons do not causally necessitate choices because agents
make choices, and agents are not subject to causal influence in the same way as other
entities.
RBAs do not require the prior assumption that agents are inherently distinct from
the larger causal order. Although people sometimes infer that they are not causally
necessitated, this need not entail that the agent making the choice is in some sense “over
and above” the normal causal order. This is a benefit of RBAs, as it is far from obvious
152
that there is very widespread commitment to the requisite kind of dualism needed to get
the transcendence vision off the ground (Mele, 2014, 195). Furthermore, RBAs can
accommodate the folk viewing reasons as causal influences on choices, as seen in RBA2.
RBAs also share some similarities with Terry Horgan’s views on the role of
phenomenology in the formation of free will beliefs (Horgan, 2011a; 2011b). Horgan
presents a rich and complex account of how much of the phenomenology of first-person agency can be interpreted as having satisfaction conditions that are consistent with a deterministic universe. For present purposes, however, we can focus on his account of so-called “recalcitrant data”; that is, his account of widespread beliefs and judgments made on the basis of phenomenology, and the truth of which is NOT consistent with a deterministic universe. The apparent widespread belief that choices are not causally necessitated is a recalcitrant datum.
Horgan offers two resources for explaining away the widespread belief that
choices are not causally necessitated. The first resource is the fact that, according to
Horgan, we cannot experience bodily motion as simultaneously causally necessitated (i.e.
state-caused) as well as “up-to-us” (i.e. actional)
He writes,
“[A]gentive phenomenology and the phenomenology of
state-causation are mutually exclusionary….it is virtually
impossible to simultaneously experience a single item of
one’s own behavior both as actional and as state-caused.
And it is easy to make the mistake of inferring, on the basis
153
of the fact that one cannot experience one’s own behavior
both as action and as state-caused motion, that no item of
behavior can really be both a genuine action and a state-
caused bodily motion.” (Horgan, 2011b, 20).
Horgan thinks that experiencing a choice as an action precludes the possibility of experiencing it as a state-caused motion.67 It is unclear, however, how the mere fact that we cannot experience bodily motion as both actional as well as state-caused explains beliefs about the metaphysics of choice. If we do infer the belief that an event is not state-caused from a lack of an experience of state-causation, then we would expect this inference to arise in domains outside of human action. It wouldn’t matter if I necessarily lack an experience of state-caused motion, as opposed to merely lacking an experience of state-caused motion. So what is important for Horgan’s purposes is that I believe that my choices necessarily lack an experience of state-causation. Stronger, I must believe that I necessarily lack an experience of state-causation because of some fact about the way I actually make choices. It would be very surprising indeed if there were a widespread belief that choices are necessarily experienced as not causally necessitated.
Putting this worry to one side, it is Horgan’s second resource that looks similar to the RBA, as it also relies on contextual parameters that restrict explanatory candidates.
According to Horgan, the concept of “agency” itself is subject to contextual parameters, and asking philosophical question about freedom pushes these contextual parameters towards libertarian interpretations. He writes,
67 Magill (1997) argues a similar line. 154
“[I]n contexts of philosophical inquiry about the
compatibility of the presentational content of agentive
phenomenology with determinism… the very posing of
such philosophical questions tends to drive the contextually
variable implicit parameters governing the judgmental
notion of agency to a maximally strict setting—an unusual
setting in which the freedom dimension of agency is
understood as incompatible with determinism, and in which
the self-as-source dimension of agency is understood as a
matter of metaphysical-libertarian agent causation, as
distinct from state-causation.” (Horgan, 2011b, 20).
Horgan’s proposal is similar to RBAs in that both depend on limiting contextual parameters to avoid overgeneralization. However, the similarities end here. On
Horgan’s view, it is the posing of explicitly philosophical questions about free will that explain interpretations of the phenomenology such that it is libertarian-friendly. On
Horgan’s view, when we ask if the content of an experience of choosing could be veridical if determinism is true, we prime an agent-causal interpretation of the phenomenology; we push the “contextually variable implicit parameters” to the agent-
causal interpretation. So while we normally think of freedom in compatibilist terms,
posing explicitly philosophical questions primes an incompatibilist interpretation of
freedom. Although it is unclear how the details work, Horgan thinks this primed
155
interpretation leads us to interpret first-person agentive phenomenology as presenting choice as not causally necessitated.
On Horgan’s view, the contextually variable implicit parameters in concepts like
“freedom” and “agency” lead us to believe that we experience our choices as not causally
necessitated. In contrast, RBAs rely on the context of practical deliberation to prime
reasons, as opposed to causal, interpretations of experiences of choice. In doing so,
RBAs avoid some of the worries facing Horgan’s view.
The biggest concern for Horgan’s view is that the belief that choice is not causally
necessitated seems to be the default view among people, independently and prior to
philosophical inquiry. Why would first-year undergrads who have no familiarity with
agent-causal libertarian or compatibilist views of free will be subject to implicit
parameter shifts prior to entering a philosophy classroom? As an additional worry, it is
not at all clear how the step from “agent-causal interpretation” to “not causally
necessitated” is supposed to work. While the belief that choices are not causally
necessitated is fairly widespread, the belief that choices are agent-caused is a relatively
niche view among philosophers. If Horgan’s story is correct, we should expect
widespread belief in agent causation. If undergraduate difficulties with the concept of
“agent causation” is any guide, it seems unlikely that belief in agent causation is very
widespread.
5. Conclusion
The positions I have explored here fit into a long tradition of non-libertarian
explanations of the apparently widespread belief that choices are not causally
156
necessitated. While differing from the Spinozan strategy in many respects, RBAs share the assumption that we do not experience our choices as causally necessitated, before inferring both that our choices are not causally necessitated, but also that we experience our choices as not causally necessitated. RBAs also share the commitment to providing a reasonable and scientifically plausible explanation of the beliefs that avoids overgeneralizations. And like Spinozan strategies, RBAs have yet to provide a compelling account that is explanatory while meeting all three constraints.
RBAs differ from the Spinozan strategy in that they take the lack of experience of causal necessitation as an enabling condition, which must be supplemented by experiences of reasons as rationalizing one’s choice in order to explain the belief that choice is indeterministic. By way of closing, it is important to acknowledge one other major flaw that RBAs share with the Spinozan Strategy. While RBAs seek to explain the widespread belief that choices are not causally determined, and even the belief that we experience choices as not causally determined, they fail to explain much of the other beliefs surrounding free choices. They do not explain why we might experience choices as “up-to-us,” or why we might experience that “we could have done otherwise.” While
there is a great deal of work to be done to face the obstacles presented for RBAs earlier in
this paper, I am optimistic that they can be met, and that RBAs can be extended to these
other experiences.
157