PEACOCKE's SELF-KNOWLEDGE Annalisa Coliva an Important Group

JOBNAME: No Job Name PAGE: 1 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 6768DA6D /v2451/blackwell/journals/RATI_v21_i1/rati_381

1 PEACOCKE’S SELF-KNOWLEDGE 2 3 Annalisa Coliva 4 5 Abstract 6 The paper reviews Christopher Peacocke’s account of self- 7 knowledge. His proposal relies on the claim that ﬁrst-order mental 8 states may be given to a subject so as to function as reasons, from his 9 point of view, for the corresponding self-ascriptions. Peacocke’s 10 Being Known elicits two different views of how that may be the case: 11 a given propositional attitude is considered to be conscious if, on 12 the one hand, there is something it is like to have it; and, on the 13 other, if it can occupy a subject’s attention without being an object 14 of attention. I examine both views and conclude that, on the latter, 15 Peacocke’s proposal risks of not offering an independent reason for 16 the self-ascription, and, on the former, of offering no reason at all. 17 I then turn to some ideas from his The Realm of Reason and claim 18 that they can help stabilise his earlier account only at the cost of 19 surrendering internalism in the epistemology of psychological self- 20 ascriptions and of contaminating internalist proposals about 21 knowledge of any subject matter with an externalist base which 22 would betray their point. Unless one doesn’t want to pay this price, 23 then Peacocke’s account offers no solution to the problem of 24 self-knowledge.1 25 26 An important group of the philosophical problems of self- 27 knowledge originate in the immediate and authoritative way in 28 which each of us characteristically knows of his own intentional 29 mental states. After a rather prolonged focus on the problem of 30 the compatibility of such self-knowledge with semantic external- 31 ism, attention has recently been reverting to the older, more 32 general issue of providing a suitable account of self-knowledge per 33 se. Here much of the discussion has been conditioned by the now 34 familiar ‘either by observation, or by inference, or by nothing’2

36 1 I would like to thank Akeel Bilgrami, Carol Rovane, Crispin Wright, Jane Heal, Lucy 37 O’Brien, Elisabeth Pacherie, Jim Pryor, Jerome Dokic, Paolo Faria, Barry Smith, Joelle 38 Proust, Achille Varzi and three anonymous referees for advice on speciﬁc points and 39 helpful comments on previous versions of this paper. 40 2 Christopher Peacocke, Being Known (Oxford: Clarendon Press, 1999), p. 231. JOBNAME: No Job Name PAGE: 2 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 5597481E /v2451/blackwell/journals/RATI_v21_i1/rati_381

14 ANNALISA COLIVA

1 trichotomy which has begun to be perceived by many as a real 2 trilemma, whose horns are each seriously problematical. In par- 3 ticular, the ‘by nothing’ horn, first presented by Crispin Wright 4 and Paul Boghossian in the late ’80s as – ironically – the only 5 viable escape from the earlier ‘either by observation, or by infer- 6 ence’ dilemma,3 has recently come under pressure in its turn. 7 According to the ‘by nothing’ type of view, the characteristic 8 authority granted to a subject’s impressions of his own intentional 9 states is an a priori and constitutive feature of the linguistic prac- 10 tice of ascribing mental states with propositional content.4 11 However, theorists working in this area now seem to be becoming 12 increasingly sceptical of the idea that self-knowledge should best 13 be seen as the result of no cognitive achievement whatever.5 It is 14 indeed natural to think, contrary to constitutivism, that first and 15 second order mental states are distinct existences; moreover, that 16 it is one thing to have first-order mental states and quite another 17 to know what they are and that the latter knowledge may only 18 plausibly be conceived as brought about by being appropriately – 19 that is, cognitively – related to the former.

21 3 See Crispin Wright, ‘Wittgenstein’s Later Philosophy of Mind: Sensation, Privacy and 22 Intentions’, Journal of Philosophy, LXXVI (1989), pp. 622–34 (p. 631 in particular). Paul 23 Boghossian, ‘Content and Self-knowledge’, Philosophical Topics, XVII (1989), pp. 5–26 24 (pp. 5, 17 in particular). The traditional forms of observationalism and inferentialism can 25 be traced back to Descartes and Ryle respectively. Recent supporters of the observational 26 model and of the inferential one, however, are David Armstrong, A Materialist Theory of the 27 Mind (London: Routledge, 1968) – author of a reliabilist version of the former – and Alison 28 Gopnik, ‘How We Know Our Minds: the Illusion of First-person Knowledge of Intention- 29 ality’, Brain and Behavioural Science, XVI (1983), pp. 1–14 – author of a reﬁned version of 30 the latter. For a criticism of these models and their developments see, for example, Sydney 31 Shoemaker, The First Person Perspective and Other Essays (Oxford: Oxford University Press, 32 1986) especially at pp. 201–23; Crispin Wright, ‘Self-knowledge: the Wittgensteinian 33 Legacy’, in Crispin Wright, Cynthia Macdonald and Barry Smith (eds.), Knowing Our Own 34 Minds (Oxford: Clarendon Press, 1998), pp. 13–45; Dorit Bar-On, Speaking My Mind. 35 Expression and Self-Knowledge (Oxford-New York: Oxford University Press, 2004), Chs. 2, 4; 36 and Akeel Bilgrami, Self-Knowledge and Resentment (Boston: Harvard University Press, 2006), 37 Ch. 1. 38 4 Among those broadly in agreement with Wright and Boghossian, beside the already 39 mentioned Shoemaker and Bilgrami, see also Jane Heal, ‘On First-person Authority’, 40 Proceedings of the Aristotelian Society, CII (2001), pp. 1–19 and Julia Tanney, ‘A Constructivist 41 Picture of Self-knowledge’, Philosophy, LXXI (1994), pp. 405–22. 42 5 See Peacocke, Being Known, Chs. 5 (and 6); Richard Moran, Authority and Estrangement 43 (Princeton: Princeton University Press, 2001), pp. 12–27; Bar-On, Speaking My Mind, Ch. 9; 44 Lucy O’Brien, ‘Self-knowledge, Agency and Force’, Philosophy and Phenomenological Research, 45 forthcoming.

PEACOCKE’S SELF-KNOWLEDGE 15

1 In this aporetic context Christopher Peacocke’s recently pro- 2 posed Rationalist account6 of intentional self-knowledge is apt to 3 impress as especially welcome. Peacocke’s model is meant to vin- 4 dicate the intuitive picture according to which self-knowledge is 5 indeed real knowledge, while avoiding both the observational and 6 the inferential stories about how such knowledge is grounded. It 7 is the brief of this note to argue that Peacocke’s proposal unfor- 8 tunately also confronts serious objections of its own, which might 9 perhaps be overcome, but at signiﬁcant costs, and that it isn’t 10 obvious that he or anyone else may want to pay them.

12 Peacocke’s Rationalist account of self-knowledge 13 14 According to Peacocke, ‘conscious thoughts and current attitudes 15 (. . .) can give (. . .) the thinker a reason for self-ascribing an 16 attitude to the content which occurs to the thinker, provided our 17 thinker is conceptually equipped to make the self-ascription’.7 To 18 illustrate: suppose you have (1) an apparent memory that Italy was 19 a monarchy before World War Two. That, according to Peacocke, 20 gives you a (non-conclusive) reason for (2) judging that Italy was 21 a monarchy before World War Two, which, in turn, gives you a 22 reason for (3) self-ascribing the belief that Italy was a monarchy 23 before World War Two (provided you possess the first person 24 concept, the concept of belief and those concepts which are 25 necessary for the specification of the content of your belief). In 26 Peacocke’s view, when the self-ascription is formed in this way, it 27 amounts to knowledge, because the second-order belief would be 28 true – in virtue of the conceptual truth that if one judges that p, 29 one believes it – and justified by the corresponding first-order 30 mental state.8 31 In Peacocke’s view, this proposal avoids inferentialism because 32 ‘to say that (2) is the thinker’s reason for making the judgement 33

34 6 The label is mine but in keeping with Peacocke’s recent pronouncement of being 35 interested in defending a new form of rationalism (see his ‘Three Principles of Rational- 36 ism’, European Journal of Philosophy, X (2002), pp. 375–397, and his The Realm of Reason 37 (Oxford: Clarendon Press, 2003). The earlier account of self-knowledge, offered in Being 38 Known is indeed in line with Peacocke’s later work. 39 7 Peacocke, Being Known, p. 214. 40 8 According to Peacocke, a judgement that p may not always occur, as a matter of fact, 41 but it should always be available to a subject, in order for his psychological self-ascription 42 to be justiﬁed (Peacocke, Being Known, pp. 222–3; 241–2).

16 ANNALISA COLIVA

1 in (3) is not to say that he infers the self-ascription from a premise 2 that he has made such a first-order judgement’.9 It avoids obser- 3 vationalism – both in classic Cartesian and more modern reliabi- 4 list versions – because to say that a first-order mental state is 5 conscious doesn’t mean, according to Peacocke, that it is an object 6 of (quasi-perceptual) attention, but, rather, that it occupies a 7 subject’s attention: being in a certain intentional state contributes 8 to a subject’s phenomenology – to what things are like from 9 his subjective point of view – despite the fact that the state isn’t 10 presented to him as an object.10 Furthermore, contrary to Arm- 11 strong’s reliabilist account, on Peacocke’s view, a first-order con- 12 scious mental state is taken to be a subject’s own reason for making 13 the corresponding self-ascription and not merely its cause.11 14 Finally, Peacocke’s model is not a version of constitutivism 15 because he denies both the constitutivist’s ontological claim that 16 first-order mental states depend for their existence on the corre- 17 sponding self-ascriptions, and the constitutivist’s anti-cognitive 18 claim that subjects’ psychological self-ascriptions are not justified 19 by their first-order mental states.12 In particular, for Peacocke, 20 while it is a conceptual truth that if one judges that p, one believes 21 it and that it is part of the possession-conditions of the concept of 22 belief that one be disposed to judge that one believes that p if one 23 judges that p, this does not at all preclude the possibility that 24 first-order conscious mental states have existence independent of 25 their self-ascription nor, especially, that they can function as a 26 subject’s own reasons for the corresponding self-ascriptions.13 27 It is important to emphasise here that Peacocke’s proposal is 28 offered as a form of epistemologically internalist model: reasons 29 are conscious states of a thinker, which, moreover, are open to 30 first-personal scrutiny. After all, at least in some cases, a subject 31 might realise that he has mistaken either the nature or the 32 content of his first-order mental states, as cases of wishful thinking 33 and of self-deception seem to show, and has thereby been 34 prompted to make an erroneous self-ascription. What is pre- 35 cluded by Peacocke’s model, however, is arrival at the self- 36 ascription on the basis of self-conscious consideration of one’s own

38 9 Peacocke, Being Known, p. 214. 39 10 Peacocke, Being Known, pp. 205–9. 40 11 Peacocke, Being Known, pp. 224–5. 41 12 Peacocke, Being Known, pp. 230–1. 42 13 Peacocke, Being Known, pp. 232–3.

PEACOCKE’S SELF-KNOWLEDGE 17

1 ﬁrst-order conscious mental states. For that would place the very 2 self-knowledge to be accounted for at the foundation of a pur- 3 ported account of how such knowledge is grounded. What Pea- 4 cocke requires is that ﬁrst-order mental states can, without being 5 explicitly judged and thereby made the content of a second-order 6 state, somehow be salient to consciousness and taken at face value 7 in the formation of the relevant self-ascriptions, and subsequently 8 can be offered as reasons for them and, on occasion, scrutinised 9 and assessed.

11 The first two horns of the critique 12 13 Peacocke’s proposal thus relies crucially on the claim that first- 14 order mental states may be given to a subject in such a way as to 15 function as reasons, from his own point of view, for the corre- 16 sponding self-ascriptions. Hence, in order for the proposal to be 17 implemented satisfactorily, he needs a notion of a conscious mental 18 state that can support the claim that the corresponding self- 19 ascription would be rationally justified merely by its occurrence. 20 Concerning what it is for a mental state to be conscious, however, 21 he offers two, not obviously equivalent, proposals. One is that a 22 mental state is conscious if there is something it is like to have it.14 23 This suggestion amounts to the idea that a mental state is con- 24 scious if there is something subjectively distinctive about being in 25 it. The other proposal is that a mental state is conscious if it 26 occupies our attention without being an object of attention.15 This 27 suggestion, in contrast, amounts to the idea that a mental state is 28 conscious if it is such that it can occupy our attention in the 29 following sense: it is something whose nature, content, presuppo- 30 sitions and theoretical and practical consequences we can attend 31 to, without turning the state itself into an object of attention.16 32

33 14 Peacocke, Being Known, pp. 205–6. 34 15 Peacocke, Being Known, p. 207. 35 16 For instance, at pp. 209–10 of his Being Known Peacocke writes: 36 37 I now attempt some further analysis of the occupation of the attention by conscious 38 thought. When you have a thought, it does not normally come neat, unconnected 39 with other thoughts and contents. Rather, in having a particular thought, you often 40 appreciate certain of its relations to other thoughts and contents. You have a thought, and you 41 may be aware that its content is a consequence, perhaps gratifying, perhaps alarming, of 42 another conclusion you have just reached; or you may be aware that its content is evidence for 43 some hypothesis that you have formulated; or that it is a counterexample to the hypothesis

18 ANNALISA COLIVA

1 The second suggestion seems highly problematical. For, pre- 2 sumably, one can attend to the nature, content, presuppositions 3 and consequences of one’s mental states only if one is aware of 4 what kind of mental states they are – judgements, desires, wishes, 5 etc. –, of what content they have, and of how they can be related, 6 as such, to one’s other mental states, with their own respective 7 contents. After all, one can attend to the presuppositions and 8 consequences of, for example, one’s judgement that Italy was a 9 monarchy before World War Two, only if one is aware17 of the fact 10 that it is indeed a judgement and not just a supposition, or a wish, 11 or a suspicion that that was the case.18 Moreover, Peacocke19 main- 12 tains that if a mental state is occupying one’s attention one will 13 probably react to it – feeling gratified, or alarmed, etc. But one 14 can feel alarmed or gratified by a certain thought occurring to 15 him only if one is aware of the kind of thought it is. For instance, 16 in the normal run of cases, I can feel gratified by the belief, or 17 even the certainty, that I will soon have a promotion, but not by 18 the corresponding desire, or fear. (Of course we may complicate 19 the phenomenology and suppose, for instance, that I might feel 20 gratified also by desiring such a coming event, if that showed, say, 21 that I am recovering from a state of dangerous apathy with respect 22 to my work and my career. Still, this would show that for one 23 affectively to react to one’s mental states, one should (at least on 24 occasion) be aware also of their relations. Yet, this would in turn be 25 possible only if one were aware of one’s mental states’ kind as well

27 (italics mine). Now when you think a particular thought, there is of course no 28 intention in advance to think that particular thought. But there can be an intention 29 to think a thought which stands in a certain relation to other thoughts or contents (italics 30 mine). 31 32 This clearly suggests the idea that if a thought occupies one’s attention, then, according to 33 Peacocke, its content must be known to the subject who has it, as well as its practical 34 consequences – like causing one’s feeling alarmed or gratiﬁed –, and theoretical ones – 35 such as its relation to other thoughts. As we shall presently see, however, this requires – for 36 Peacocke too – knowledge of the nature of one’s own mental states as well, viz. of their 37 being beliefs as opposed to wishes, imaginings, etc. 38 17 As a matter of fact, Peacocke talks in terms of ‘rational sensitivity’ to one’s mental 39 states. One way of understanding this (rather vague) expression is in terms of ‘awareness’. 40 But for a different – still problematical – interpretation, see the following paragraph. 41 18 Peacocke himself makes the point at p. 216 of Being Known: ‘Now the thinker who 42 successfully reaches new beliefs by inference has to be sensitive not only to the content of 43 his initial beliefs. He has also to be sensitive to the fact that his initial states are beliefs.He 44 will not be forming beliefs by inference from the contents of his desires, hopes, or 45 daydreams’. 46 19 See fn 15.

PEACOCKE’S SELF-KNOWLEDGE 19

1 as of their content).20 So, the evident problem with Peacocke’s 2 suggestion is that the awareness of the kind and content of one’s 3 own first-order mental state required if it is to occupy one’s atten- 4 tion and inform further attention to its presuppositions and con- 5 sequences seems – on a quite natural reading of Peacocke’s own 6 account of it – to amount to nothing less than knowledge of that 7 very mental state of which, on his view, it is supposed to provide 8 a rational basis! At the least the gap between them seems to 9 be vanishingly small. But unless it can be widened, Peacocke’s 10 account of self-knowledge would bluntly presuppose exactly what 11 it should explain – and indeed, not because some other piece of 12 self-knowledge would be presupposed but, rather, because the 13 very instance of self-knowledge to be accounted for would be.21 14 One may then try to propose different glosses on the notion of 15 a mental state’s occupying one’s attention. According to a first, 16 alternative reading of it, a given mental state could occupy one’s 17 attention if one were attentively engaging in it, without thereby 18 considering it as the particular mental state it is – say a judgement 19 – with the particular content it has – that Italy was a monarchy 20 before World War II, for example –, pretty much as one could be 21 performing attentively a certain activity – like running – without 22 considering it qua the particular activity it is.22 This gloss on the 23 notion of the occupation of the attention wouldn’t presuppose 24 explicit awareness of the very mental state whose knowledge is 25 supposed to be accounted for. Nonetheless, it seems to me that it 26 couldn’t be of avail to Peaocoke because his model requires that 27 one’s first-order conscious mental states be a subject’s own reasons 28 for his corresponding self-ascriptions. But for psychological self- 29 ascriptions to be rational responses to one’s own occurrent con- 30 scious mental states one must actually be aware of them qua the 31 particular mental states they are, with the contents they have. To 32 see this, consider the analogy with running: as much as one could 33 run attentively, without considering it qua the activity it is, if one

35 20 I am grateful to an anonymous referee for bringing this point to my attention. 36 21 Peacocke could reply that what is required is knowledge of one’s judgement, whereas 37 what gets self-attributed is a belief. But, ﬁrst, a judgement is nothing but a mental state (or 38 action). Hence, knowledge of it would be a case of self-knowledge which we are supposed 39 to account for. Secondly, if we applied Peacocke’s model to explain how this latter piece of 40 knowledge could be possible in turn, we would get, on the present proposal, that this very 41 mental state should already be known to a subject who self-ascribed it. 42 22 I am grateful to an anonymous referee for making me think about such an alternative 43 interpretation of Peacocke’s present proposal.

20 ANNALISA COLIVA

1 were to make a rational judgement about what one is attentively 2 doing, one’s action should be manifest to him as the particular 3 action it is. Merely engaging in it, as attentively as it might be, 4 wouldn’t constitute a reason for the corresponding judgement. 5 Similarly, in the case of mental states, their rational self-ascription 6 seems to require switching – however momentarily that might be 7 – from being attentively, yet non-reflectively, engaging in them, to 8 considering them qua the mental states they are, with the contents 9 they have. 10 Another possible way of trying to rescue Peacocke’s present 11 proposal would consist in maintaining that the kind of knowledge 12 of one’s first-order mental state that the occupation of the atten- 13 tion requires is merely tacit or implicit.23 One might then suggest 14 that a subject may have the ability to have certain thoughts, act on 15 them and reason on their basis, thus implicitly knowing what kind 16 of mental states they are, without yet having the necessary con- 17 ceptual repertoire to self-ascribe them. This, however, is what 18 implicit self-knowledge amounts to. And here a problem arises: how 19 is this implicit knowledge achieved? An answer to this question is 20 fundamental if implicit self-knowledge has to be the backbone of 21 an explanation of self-knowledge in general. For, otherwise, 22 Peacocke’s account would only explain, at most, how one can 23 come to self-ascribe mental states that are already implicitly 24 known. Furthermore, even supposing that an account of implicit 25 self-knowledge were forthcoming, how would the conceptualisa- 26 tion of implicitly known first-order mental states proceed? If it 27 depended on a grasp of the conceptual role of first-personal 28 propositional attitudes – in the case of belief, for instance, on 29 grasp of the rule ‘If I judge that p, then I should be disposed to 30 self-ascribe the corresponding belief’ –, wouldn’t that after all 31 require explicit knowledge of one’s first-order mental states? 32 Finally, one could try to revise Peacocke’s own account and 33 propose less demanding readings of the notion of a mental state’s 34 occupying one’s own attention. For instance, following Bonjour,24 35 one might claim that there is something like a ‘constitutive aware- 36 ness’ of the kind and content of one’s intentional mental state 37 which does not coincide with one’s judging that one has that 38

39 23 I would like to thank Barry Smith for bringing this point to my attention. 40 24 Laurence Bonjour, ‘A Version of Internalist Foundationalism’, in Laurence Bonjour 41 and Ernest Sosa, Epistemic Justiﬁcation. Internalism vs. Externalism, Foundations vs. Virtues 42 (Oxford: Blackwell, 2003), pp. 61–8.

PEACOCKE’S SELF-KNOWLEDGE 21

1 mental state with such and such a content.25 But I must confess that, 2 as appealing as it might be, this view seems to me irremediably 3 unclear: if the awareness of the kind and content of one’s mental 4 state is implicit, then the problems just reviewed will re-emerge. If, 5 in contrast, it isn’t, I don’t see how it can avoid collapsing into 6 full-blooded knowledge of one’s first-order mental state. For it may 7 be true that when one is consciously enjoying a first-order inten- 8 tional state one doesn’t actually have to judge that one has that 9 mental state with that content.26 However, if that knowledge has to 10 be somehow explicit, it seems that it should be constitutive of having 11 a conscious first-order mental state that one should at least have the 12 disposition to make that kind of judgement. But, in this latter case, 13 Bonjour’s proposal would just be a mere notational variant of more 14 traditional constitutive accounts of self-knowledge27 a fact that 15 would betray its whole point. Finally, if such constitutive awareness 16 is neither implicit knowledge, nor constituted by the disposition to 17 judge that one has a given first-order mental state, and merely 18 consists in the fact that the kind and content of that state are part 19 of its phenomenology, then the proposal will turn into a version of 20 Peacocke’s first, phenomenological account of what it means for 21 an intentional mental state to be conscious and will be open to the 22 criticisms we shall presently level against it. 23 So, now, let us turn to Peacocke’s other, phenomenal gloss on 24 what it means for a mental state to be conscious. Here too prob- 25 lems arise. The first one is that the work the proposal assigns to 26 the phenomenology of first-order intentional states now restricts 27 it to the case of self-ascriptions of occurrent mental states, and rules 28 it out for standing attitudes that lack any phenomenological 29 impact. Secondly, even in occurrent cases, the capacity of the 30 phenomenology of first-order mental states to carry the resulting

32 25 Bonjour, ‘A Version of Internalist Foundationalism’, p. 62: 33 34 [Awareness of the kind and content of one’s own occurrent mental state is] not in 35 any way apperceptive or reflective in character: [it does] not require or involve a 36 distinct second-order mental act with the propositional content that I have the 37 belief in question. Instead, [it is] partly constitutive of the first-level state of occur- 38 rent belief (. . .). 39 26 According to Bonjour, ‘A Version of Internalist Foundationalism’, pp. 65–8 that 40 would either involve us in a vicious regress, or else never provide an account of how a given 41 mental state (first-order or otherwise) could ever be conscious. 42 27 Such as Bilgrami’s, Self-Knowledge and Resentment, for instance. On his view, it is a 43 necessary condition for having (a certain class of) first-order propositional attitudes that 44 one knows them.

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 10 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 594EEA43 /v2451/blackwell/journals/RATI_v21_i1/rati_381

22 ANNALISA COLIVA

1 epistemological load falls into question as soon as one reflects 2 that, for example, occurrent desires, hopes or wishes with the 3 same content need hardly be different from a strictly phenomeno- 4 logical point of view, yet will be required somehow to rationalise 5 different self-ascriptions. Thirdly, it is widely agreed,28 and indeed 6 by Peacocke himself,29 as it seems, that, even in cases where occur- 7 rent intentional states do have a distinctive phenomenology, it 8 remains that rational and justificatory relations hold only between 9 states with representational content in the first place. The idea is 10 that the kind of phenomenology that, for instance, the judgement 11 that today is a sunny day can have – such as the pleasant and 12 relaxing feeling this thought may produce, together with, 13 perhaps, a strong sense of confidence – will not, by themselves, 14 give one a reason to self-ascribe the corresponding belief. Rather, 15 it is only if the first-order mental state presents itself to one as the 16 type of mental state it is – that is, a belief as opposed to a wish or a 17 desire – with the content it has – that today is a sunny day, as

19 28 This point has been vigorously maintained, for instance, by John McDowell in his 20 Mind and World (Cambridge (Mass.)–London: Harvard University Press, 1994) and by Bill 21 Brewer in his Perception and Reason (Oxford: Oxford University Press, 1999). For opposite 22 views, see, for instance, James Pryor, ‘There Is Immediate Justiﬁcation’, in Matthias Steup 23 and Ernest Sosa (eds.), Contemporary Debates in Epistemology (Oxford: Blackwell, 2005), 24 pp. 181–202. 25 29 At p. 216 of his Being Known Peacocke writes: 26 27 In cases of consciously based self-ascription of attitudes and experiences, a thinker 28 [. . .] makes a transition not only from the content of some initial state, but also 29 makes it because the initial state is of a certain kind (. . .). In the case of consciously 30 based self-ascription, the distinction between those events which are occurrent 31 attitudes of the right kind to sustain the resulting judgement and those which are 32 not is a distinction which is conceptualised by the thinker. 33 34 All this clearly implies that the relevant transitions are made on the basis of how the 35 ﬁrst-order mental states are represented to the subject and not just on the basis of their 36 phenomenology and this seems to ensure their rationality from the subject’s own point 37 of view. So, although, as Jim Pryor has brought to my attention, Peacocke (in his 38 ‘Does Perception Have a Nonconceptual Content?’, Journal of Philosophy, XCVIII (2001), 39 pp. 239–64, pp. 254–5 in particular) maintains that sensations – devoid of any represen- 40 tational content – can immediately justify one’s corresponding self-ascriptions, he does not 41 seem to be inclined to offer an analogous account of intentional mental states and of their 42 self-ascriptions. 43 Furthermore, for reasons of internal coherence with his earlier work, I think Peacocke 44 should acknowledge that only representational contents – let them be psychological or 45 otherwise – can serve as rationalisers of judgements and, in particular, of self-ascriptions of 46 intentional mental states. Indeed it is only on such an assumption that one can understand 47 why, instead of defending the so-called ‘Myth of the Given’ against McDowell’s attacks, he 48 elaborated a notion of nonconceptual, yet fully representational content for experiences (see 49 Christopher Peacocke, A Study of Concepts (Boston (Mass.): MIT Press, 1992), Ch. 3).

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 11 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 49ABD233 /v2451/blackwell/journals/RATI_v21_i1/rati_381

PEACOCKE’S SELF-KNOWLEDGE 23

1 opposed to anything else –, that it can constitute a thinker’s own 2 reason for the corresponding self-ascription. For, otherwise, why 3 should one be more justified in making that self-ascription than in 4 self-ascribing the corresponding hope, say? To see this, consider 5 cases of wishful thinking: certain contents may manifest them- 6 selves to a subject with such an intensity and ‘colouring’, as it were, 7 that, while being merely hopes, can actually be taken for beliefs. 8 So, their phenomenology would be pretty much the same, yet only 9 the self-ascription of the relevant hope would be rational on their 10 basis (not just correct). If so, however, it is unclear how any purely 11 phenomenological account of what it means for a mental state to 12 be conscious can support the claim that a first-order state can 13 stand in a rational, justificatory relation to the corresponding self- 14 ascription.30 To repeat, if the phenomenology can be pretty much 15 the same and no other element is required to account for the 16 rationality of a given self-ascription, why should we think that only 17 the self-ascription of the relevant hope, and not of the corre- 18 sponding belief, would be rational? 19 To recap: on the attention-occupancy account of what it means 20 for a mental state to be conscious, Peacocke’s proposal that a 21 first-order mental state may provide a reason for the correspond- 22 ing self-ascription does not discernibly escape presupposing the 23 very piece of (implicit or explicit) self-knowledge we should be 24 accounting for. On the purely phenomenological account, in 25 contrast, there is a real danger of not being provided with any- 26 thing that could serve as a reason for the corresponding self- 27 ascription. So Peacocke’s account runs the risk either of failing to 28 offer an independent reason for the self-ascription (or of presup- 29 posing implicit knowledge of the very item of self-knowledge to be 30 accounted for), or of offering no reason at all.

32 30 Indeed I think that it is totally unclear how proponents of such a view could solve the 33 problem just mentioned and known as the ‘arbitrariness problem’ (see Pryor, ‘There Is 34 Immediate Justiﬁcation’, pp. 192–193). For instance, Pryor’s own attempt to solve it by 35 appealing to a notion of mental events which are themselves logically structured is dubious 36 both from a metaphysical point of view, as Achille Varzi has remarked to me, and from an 37 epistemological one. For an internalist needs a justiﬁer that is given to the subject and 38 which can play a rationalising role for his self-ascription from his own point of view. The fact 39 that an event might be, unbeknownst to him, logically structured and suited, in principle, 40 to rationalise the transition from its occurrence to its self-ascription, is of no use to the 41 development of a sound internalist epistemology of psychological self-ascriptions.

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 12 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 59FBA81D /v2451/blackwell/journals/RATI_v21_i1/rati_381

24 ANNALISA COLIVA

1 The third horn 2 3 In his most recent work,31 Peacocke presents some ideas which 4 may seem suited to evade the dilemma just reviewed. Key amongst 5 them is the claim that there can be unmediated, non-inferential 6 transitions, viz. movements of thought – in our case they would be 7 from conscious first-order mental states to second-order ones32 8 that are nevertheless rational, because they are truth-conducive and 9 a priori. By this Peacocke means that there can be movements of 10 thought which are truth-conducive not because – or just because 11 – they are underwritten by reliable physical mechanisms, but 12 because they conform to certain a priori principles. An example 13 would be the principle that if one judges that p, one will believe 14 that p. So the truth of ‘I believe that p’ will be guaranteed by such 15 a principle. According to Peacocke, the thoughts in which such 16 transitions culminate can be non-inferentially justified merely by 17 the very occurrence of the relevant anterior states from which the 18 transition proceeds.33 Applying the idea to our present concerns: 19 when one’s psychological self-ascription – ‘I believe that p’–is 20 formed by moving from one’s first-order mental state – a judge- 21 ment that p – to it directly – that is, without the mediation of any 22 other mental state –, one will have a non-inferential justification – 23 an entitlement in Peacocke’s terminology – for the self-ascription.34

25 31 Christopher Peacocke, The Realm of Reason. 26 32 But I need to emphasise that Peacocke in The Realm of Reason does not explicitly 27 consider the case of self-knowledge. Rather, I am freely extending views he develops with 28 respect to the relation between perceptual experiences and empirical beliefs to the case of 29 transitions from ﬁrst-order mental states to second-order ones. He might refuse the exten- 30 sion. However, in a more recent paper (‘ “Another I”: Representing Conscious States, 31 Perception and Others’, in José Luis Bermúdez (ed.), Thought, Reference and Experience. 32 Themes from the Philosophy of Gareth Evans (Oxford: Clarendon Press, 2005), pp. 220–57), 33 Peacocke extends some of these ideas to the case of self-ascriptions of perceptions, i.e. to 34 the transitions from episodes of seeing that p to the judgement ‘I see that p’. However 35 instructive that extension might be, our concern here is with self-ascriptions of propositional 36 attitudes only – which is what is usually understood in the literature by ‘self-knowledge’. A 37 discussion of Peacocke’s treatment of self-ascriptions of perceptions will have to be 38 deferred to another occasion, although one point which will be raised in the following – to 39 anticipate, that his new proposal should best be seen as a form of ‘Rational Externalism’, 40 contrary to Peacocke’s professed internalism – seems to me to apply also to his treatment 41 of self-ascriptions of perceptions. 42 33 Strictly speaking, as Peacocke himself notices (The Realm of Reason, p. 26) the results 43 of these transitions would be only ‘relatively a priori’ since they would be justiﬁed by the 44 occurrence of particular mental states. Still, they wouldn’t be inferred from them. 45 34 See Peacocke, The Realm of Reason,p.11.

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 13 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 47382E12 /v2451/blackwell/journals/RATI_v21_i1/rati_381

PEACOCKE’S SELF-KNOWLEDGE 25

1 This proposal promises relief from the consciousness dilemma 2 by enabling us to deflect its second horn: it would allow first-order 3 mental states to count as justifiers after all, even if their nature 4 and content were no part of their phenomenology, just so long 5 as there are suitable a priori principles, which a subject need not 6 be aware of, which could rationalise the transitions he makes to 7 the corresponding second-order states. So, although someone’s 8 judgement that p need not present itself as such to the subject who 9 makes it, it could still rationalise a self-ascription of the corre- 10 sponding belief, since the two states may be connected by the a 11 priori principle that if one judges that p, one will believe that p. 12 It is well beyond the scope of this note to try to evaluate the 13 general standing, let alone the detail, of the notion of entitlement 14 Peacocke explores in the relevant parts of The Realm of Reason.My 15 remaining point concerns purely what it may have to offer, in the 16 best case, to bolster his earlier Rationalist account of self- 17 knowledge against the problems I have been describing. For this 18 purpose, it is crucial to recall the internalist aspirations of the 19 Rationalist account: its goal was to provide space for, at once, the 20 rationality, from the first person perspective, of the transitions 21 involved and the possibility that a first-order mental state could 22 be a subject’s own accessible reason for his corresponding self- 23 ascription35 so that, if needed, he could scrutinise and assess it. 24 However, under the aegis of Peacocke’s notion of entitlement, the 25 rationality, from the first-person perspective, of the relevant tran- 26 sitions need not be visible. Visibility would require the movement 27 from a first-order to a second-order mental state to be informed by 28 an appreciation of the rational connection between those states, 29 i.e. by an appreciation of the fact that one’s self-ascription is 30 rational because it is arrived at by moving from the appropriate 31 kind of first-order state.36 To illustrate with our by now standard 32 example: one’s self-ascription of a belief that Italy was a monarchy 33 before World War Two would be rational, from a first-person 34 point of view, only if a subject was aware of the fact that it is 35 formed by transition from the occurrence of one’s first-order 36 mental state of judging that that was the case. However, such an

38 35 See Peacocke, The Realm of Reason, pp. 12, 101, 177–8. 39 36 Notice that I am not claiming that the rationalising principles should be self- 40 consciously, or even tacitly employed by a subject, even less that their truth should be 41 appreciated by him in order to have an account of self-knowledge which would be accept- 42 able by internalist lights.

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 14 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 4633F447 /v2451/blackwell/journals/RATI_v21_i1/rati_381

26 ANNALISA COLIVA

1 awareness would presuppose the very knowledge (implicit or 2 explicit, as we saw before) of one’s first-order mental state the 3 model was supposed to dispense with. So there is actually no clear 4 sense in which the refurbished proposal could qualify as internal- 5 ist: although the rationalisers would be internal and even con- 6 scious states of a subject, their rationalising role wouldn’t – and, 7 as a matter of fact, couldn’t37 be manifest to him. Hence, they 8 wouldn’t (and couldn’t) play the role of a subject’s own reason for 9 his judgement. 10 It seems to me that this proposal would better be seen as a form 11 of what may be called ‘Rational Externalism’. Self-ascriptions of 12 intentional states that qualified as knowledgeable under its con- 13 straints would amount to knowledge externalistically construed, 14 simply because the subject would not be required to have any 15 appreciation of the rationalising role his first-order mental states 16 would play with respect to them. Moreover, the relevant self- 17 ascriptions might still qualify as rational because they would be 18 formed in accordance with a priori principles, rather than under 19 merely a posteriori reliable circumstances. Thus, if Rational Exter- 20 nalism was a viable position, it could, in principle, offer escape 21 from the original ‘either by observation, or by inference, or by 22 nothing’ trilemma and vindicate the intuitive thought that self- 23 knowledge is, after all, a genuine cognitive accomplishment. 24 However, the proposal comes at a significant cost, well beyond 25 the price of re-labelling as ‘Rational Externalist’ the view hereby 26 proposed. For the idea that the correct account of the epistemol- 27 ogy of self-knowledge might be Rational Externalist would com- 28 pletely subvert the usual terms of the distinction between 29 internalist and externalist conceptions of knowledge and justifi- 30 cation. For reflect: on an internalist view, knowledge of any 31 subject matter at all is always dependent on internally available 32 reasons – reasons available to one as such as part of one’s psycho- 33 logical self-knowledge (whether knowledge of one’s experiences, 34 judgements, or of other beliefs). Externalism, by contrast, pre- 35 cisely eschews any such dependence. But if self-knowledge in turn 36 now qualifies as genuine knowledge only on an externalist 37 account of it, the effect is that internalism becomes pointless 38 everywhere: the kind of inner discipline which subjecting knowl- 39 edge to internal constraints was meant to impose becomes a

41 37 On pain of falling back into either horn of the consciousness dilemma.

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 15 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 2C82E17A /v2451/blackwell/journals/RATI_v21_i1/rati_381

PEACOCKE’S SELF-KNOWLEDGE 27

1 charade – for satisfaction of the constraints imposed by this disci- 2 pline is now something to be externalistically accounted for! 3 Maybe fans of externalism will welcome this conclusion, but I 4 would expect Peacocke and those who, in following him, were 5 hoping to reach an internistically acceptable epistemology of psy- 6 chological self-ascriptions, not to be among them. 7 So, to summarise: I have claimed that the model of self- 8 knowledge proposed in Being Known risks presupposing that 9 (implicit or explicit) knowledge of one’s own mental states it was 10 meant to explain; or fails to explain how psychological self- 11 ascriptions are justiﬁed by the corresponding ﬁrst-order mental 12 states. However, when buttressed by central ideas from The Realm 13 of Reason the account is stabilised only at the cost not merely of 14 surrendering Peacocke’s professed internalism in the epistemol- 15 ogy of psychological self-ascriptions but of contaminating inter- 16 nalist proposals about knowledge of any subject matter at all with 17 an externalist base which, arguably, betrays their whole point. 18 So long as it is unclear if and why Peacocke, or anyone else, may 19 want to pay this price, it remains that his proposed account rather 20 than resolving the problem of self-knowledge contrives to re- 21 invent it, with a whole new set of uncomfortable alternatives. 22 23 Faculty of Letters and Philosophy 24 University of Modena 25 Largo Sant’Eufemia 19 26 41100 Modena 27 Italy 28 [email protected]

© 2008 The Author Journal compilation © 2008 Blackwell Publishing Ltd JOBNAME: No Job Name PAGE: 16 SESS: 15 OUTPUT: Fri Nov 2 14:50:58 2007 SUM: 042E3C9B /v2451/blackwell/journals/RATI_v21_i1/rati_381 SNP Best-set Typesetter Ltd. Journal Code: RATI Proofreader: Elsie Article No: 381 Delivery date: 2 November 2007 Page Extent: 15