How to commit Moore\'s paradox

How to commit Moore’s paradox*

In the literature on the topic, it is widely acknowledged that Moore’s paradox comes in two forms: (1) I believe that P, but it is not the case that P;1 and (2) I do not believe that P, but it is the case that P. The paradox arises from the fact that the conjuncts in (1) and (2) have independent truth-conditions and can therefore be both true (or false) at once. If so, then, (1) and (2) – i.e. these conjunctive propositions – could be true and yet judging or asserting them would be so deeply absurd as to result in self-defeat. We sense that a subject cannot really believe that it is raining, say, while also judging that it is not; or that he cannot disbelieve that it is raining while also judging that it is. Yet, the nature of Moore’s paradox remains elusive in many ways and several different explanations of it have been proposed. Despite its elusiveness, study of this paradox is judged to have extremely wide implications (Green & Williams 2007: 3-4) and Richard Moran (1997: 143) insists that Moore’s paradox is “an emblem for peculiarities in the first person point of view, specifically how the possibilities for thinking and talking about oneself are systematically different from the possibilities of thinking and talking about other people”. In a recent paper, Tom Baldwin (2007) has argued that, in order to provide a satisfactory account of Moore’s paradox, we should bring in a normative notion of belief – the notion of belief as commitment. For, if we stick to purely functionalist accounts of belief, nothing would prevent a subject from sensibly

*

Acknowledgments: …

1

Sometimes (1) is phrased differently, viz. (1*) “P and I believe that not-P” (cf. Green and Williams 2007, p. 5).

Notice that (1*) can be obtained from (1) through uniform substitution of P with not-P and commutation, provided that double negation elimination held. (1) (or 1*) is called the commissive form of the paradox and (2) the omissive one (cf. Green and Williams 2007).

1

judging (or asserting) either (1) or (2).2 I concur with Baldwin’s judgment and further motivations in its favor will be offered in the sequel. Yet, it will be argued that it is not enough merely to introduce the notion of a belief as a commitment in order to rescue Moore’s paradox. Rather, to avoid its disappearance – i.e. its being perfectly legitimate to judge (or assert) either (1) or (2) at least on occasion – we should stick to a rather resolute notion of commitment. The final sections of the paper will spell out such a notion and will aim to establish this conclusion; or, at least, to show how, unless we are prepared to embrace it, we would be in danger of actually losing Moore’s paradox. Given the high stakes at issue, such a result would seem to entail rather unpleasant philosophical consequences. Furthermore, it would become a total mystery why we intuitively find either (1) or (2) so defective. Hence, the loss of Moore’s paradox would cause a severe blow both to our philosophical and to our self-understanding.

1. Moorean and Wittgensteinian analyses Two main kinds of analysis of the paradox have been proposed since George Edward Moore first discovered it. They can be traced back to Moore himself and to Wittgenstein. We shall therefore call them the Moorean and the Wittgensteinian analysis, respectively. The details of the developments they have been subject to over the years will not be discussed. For, although these developments are interesting and insightful, it is not clear that they will be able to eschew the problems that afflict their respective

2

An analogous suggestion can be found in Millar 2004 (p. 125). At least for the sake of this paper, it will be

assumed that broadly functionalist accounts of beliefs and normative ones are the main options on the table. Eliminativism, besides being problematical in its own right, would be a non-starter in connection with Moore’s paradox, for it would actually condone “It is raining but I do not believe it”, since, by its lights, there would be no beliefs (cf. Turri 2010). As will become apparent, in some cases (1) and (2) can be coherently judged/asserted. This, however, does not mean that there are never cases in which those judgments/assertions would result paradoxical. The task will be to spell out the conditions in which judging or asserting either (1) or (2) would result in self-defeat.

2

predecessors.3 Nevertheless, the aim of this section is only to highlight the main difficulties these analyses face and thereby provide at least prima facie motivation to look for an alternative. We will start by giving an outline of Moorean and Wittgensteinian analyses and by exposing their shortcomings. It will be argued that the major difficulty with the former is that it does not explain the oddity of holding either (1) or (2) in thought.4 For it is committed to the view that the paradox arises only because the assertion of either (1) or (2) violates some pragmatic norm that governs that linguistic practice. By contrast, the main problem with the latter is that it involves a loss of subject-matter. For, by reducing Moore’s paradox to an outright contradiction, it dissolves, or explains away what needed explaining in the first place: how it is possible for two independent propositions, once they are conjoined, to give rise to selfdefeating assertions or judgements.5 The proposal is to take these points quite seriously, and hence to steer away from any analysis of Moore’s paradox which either claims that, by application of whatever principle, the very content of (1) and (2) comes down to a contradiction of the form “p and not-p”, or appeals to some pragmatic norm governing the speech act of assertion. At the core of all Moorean analyses lie two main claims:

3

Moore 1942, pp. 540-3 and 1944, p. 204; and Wittgenstein 1953, pp. 190-2 and 1980, §§ 90-6. Among

Moorean analyses, see Baldwin 1990 and Shoemaker 1995; among Wittgensteinian ones see Collins 1996, Heal 1994, Lee 2001, Linville and Ring 1991, Malcolm 1995. 4

Heal 1994, p. 6, Shoemaker 1995, p. 213, Moran 1997, p. 144 and Baldwin 2007, pp. 78-9 emphasise this

point. Green and Williams (2007, p. 6), date it back to Sorensen 1988. Chan 2010 tries to make sense of the idea that there might be pragmatic paradoxes at the level of thought. Yet, he concludes that, in the case of Moore’s paradox, this would nevertheless lead to an unsatisfactory account. 5

Linville and Ring (1991) are committed to this claim. Heal (1994, p. 6) places as a requirement upon any

satisfactory analysis of Moore’s paradox that “the solution must identify a contradiction, or something contradictionlike, in the Moorean claims”. Both requirements will be criticized at length, for different reasons, in the text.

3

(a) Neither (1), nor (2) are contradictions. For they are conjunctions of propositions with different truth conditions. Hence, they can both be true at the same time. (b) Given (a), there is nothing wrong with (1) and (2) as such. Thus, there is nothing wrong with believing or entertaining their respective contents in thought. For one can perfectly well think that one has a false belief – (1) – and that is ignorant with respect to P – (2). Rather, the oddity of Moorean contents – what, in effect, entitles us to talk about a paradox in their connection – is the fact that when one asserts either (1) or (2), then one is doing something pragmatically inconsistent. Assume for the sake of argument that the assertion of “p” – let it be sincere or not – presents the speaker as believing that p. By then adding in one breath, as it were, that one does not believe that p, one is presenting oneself, at one and the same time, as both believing and not believing that p. This, however, seems to be pragmatically inconsistent. For it appears that one cannot consistently perform a speech act whose (non-deductive) implication is that one believes what one is saying while at the same time cancelling that implication by saying “But I do not believe it”. However, this account is no longer available in the case of (1).6 For by asserting “not-p” one presents oneself as believing that not-p is the case. Yet by asserting “I believe that p” one presents oneself as believing the opposite. Hence, by asserting (1) one presents oneself as believing contradictory propositions. Yet, where exactly is the pragmatic inconsistency in presenting oneself as believing contradictory propositions? After all, it seems possible even to form the intention of informing one’s audience that in a process of self-scrutiny, for instance, one has found out that one has contradictory beliefs (as we shall see in the following). Therefore, it is not clear what pragmatic norm one would be violating by conveying the thought that one has contradictory beliefs, if any at all. Whether or not a Moorean theorist may be able to make a convincing case for the view that to present oneself as believing contradictory propositions is a kind of pragmatic inconsistency, the main problem with the Moorean analysis is as follows. If the paradox can be made sense of only by appealing to pragmatic

6

Heal 1994, pp. 11-2.

4

norms governing communication and speech acts, this shows by itself that there should be no analogue of Moore’s paradox at the level of thought. Still, judging (1) or (2) would indeed be an odd thing to do. Yet, on the Moorean analysis, we are deprived of any means to make sense of this very possibility. Hence, whether or not the Moorean analysis is successful as an analysis of Moore-paradoxical assertions, it does not have the resources to explain Moore-paradoxical judgements. Any Moorean analysis, therefore, is at least wanting for a lack of generality. Wittgensteinian analyses of Moore’s paradox are united in holding the following:7 (a’) Moore’s paradox is not really a paradox, because, in effect, it is an outright contradiction. For asserting “I believe that p” is tantamount to asserting “p” in a tentative voice, so to speak; and asserting “I do not believe that p” is tantamount to asserting “not-p”. Hence, both (1) and (2) come down to an assertion of “p and not-p”. (b’) For this very reason, it is neither all right to assert nor to judge (1), or (2). (b’) is the only feature which, at least prima facie, seems to speak in favour of Wittgensteinian analyses. For (a’) has the well-known and unpalatable consequence of depriving one of the means of explaining the following platitudes. Embedding: (1) and (2) can be embedded in wider contexts, e.g. ”Suppose that I believe that p, but it is not the case that p” and “Suppose that it is the case that p, but I do not believe it”. The possibility of embedding (1) and (2) in suppositions, in turn, explains why they can occur as antecedents in correct conditional statements.8 Past: (1) and (2) can be turned into the past. “I believed that p but it was not the case that p” or “I did not believe that p but it was the case that p” are perfectly fine.

7

Moran 1997 denies that this is actually Wittgenstein’s position. This, however, is the vulgata and constitutes

the bulk of what I call “Wittgensteinian” analyses of Moore’s paradox. 8

Wittgenstein himself was aware of this. See in the following for a (failed) attempt at reconciling this platitude

with a Wittgensteinian account of Moore’s paradox.

5

Third-Person: (1) and (2) can be put into the third person. “She believes that p, but it is not the case that p” and “It is the case that p, but she does not believe it” can both be used to make accurate statements. However, no formal contradiction – that is, no contradiction of the form “p and not-p” – can be the content of a coherent supposition,9 nor does it turn into a non-contradiction once put into the third-person, or when changed of tense. Yet, on the Wittgensteinian analysis we are asked to take (1) and (2) as formal contradictions, while, at the same time, granting the platitudes. How can these two claims be reconciled? The Wittgensteinian strategy consists in claiming that the verb “to believe” has a different meaning depending on whether it is used in the first or in the third person and on whether it is used either in the present or in the past. In particular, we should concede that no use of the verb “believe” in the first person, present, is such as to describe a subject’s belief.10 Yet, this seems odd, for there is such a use. Consider “I do not know who did it, but I believe it was the butler” said by a subject who has no warrant for charging the butler with the mischief, but is totally subjectively convinced that he is the culprit. Now, if the speech act of making an assertion is governed by the epistemic norm that one should have a warrant for one’s claim, then a subject who asserts that sentence would precisely not be in a position to assert “The butler did it”. Hence, a subject’s assertion of “I believe the butler did it” cannot be taken as equivalent to the assertion of its embedded content. Thus, it must be taken as a description of the subject’s doxastic state, viz. of her conviction that the butler is the

9

One might object that one can suppose anything. For instance, opponents of dialetheism suppose that p and

not-p in order to show that any consequence follows. However, exactly this case shows that the supposition was not coherent in the first place. At any rate, formal contradictions at least do not cease to be such when changed of tense or person. 10

By contrast, the meaning of “to believe” both in the first-person past and in the third person is descriptive of

a mental state, which one is either ascribing on the basis of behavioural criteria (sometimes even in one’s own case) or on the basis of memory.

6

culprit, which can be in place no matter whether she has a warrant for “The butler did it” and, hence, independently of whether she would ever be in a position warrantedly to assert it.11 When we turn to the use of the phrase “I do not believe that p”, the case for not taking it to be equivalent to the assertion of “not-p” is even more obvious.12 For it is not always the case that someone who is in a position sincerely to assert “I do not believe that p” is also and automatically in a position to assert “not-p”. If a subject has suspended judgement on whether the butler is the culprit, because there is equal evidence pro and against him, her assertion of “I do not believe the butler did it” is consistent with her epistemic situation in a way in which “The butler did not do it” is not. Said otherwise, agnosticism with respect to P is a coherent position. Hence, it is coherent for someone to assert “I do not believe that p and I do not believe that not-p”. Yet, if we applied the Wittgensteinian analysis to that assertion it would turn it into an assertion of “p and not-p” (provided double negation elimination held). Agnosticism – allegedly a respectable position – would thus amount to an outright contradiction. Hence, while granting that sometimes the locution “I believe” is used to weaken one’s assertion of “p”, and that the locution “I do not believe” is used to express the negation of the embedded content, some other times those locutions are used to describe one’s belief or one’s absence of belief. Therefore, even if Wittgensteinian analyses could apply to one kind of use of “I (don’t) believe”, they could not apply to the other. Thus, Moore’s paradox could still arise for the latter use of “I (don’t) believe” and this is what needs explaining.13 11

Cf. de Almeida 2001, p. 38.

12

Heal 1994, p. 7, who favours a Wittgensteinian analysis of Moore’s paradox, acknowledges this point, but,

surprisingly, goes on to offer a uniform account of it, where both (1) and (2) come down to contradictions of the form “p and not-p”. However, such a result can be obtained only by taking “I do not believe that p” as equivalent to “I believe that not-p” and thus by taking its assertion as equivalent to the assertion of “not-p”. 13

This point (cf. de Almeida 2001, p. 39) is not sufficiently appreciated by Heal 1994, pp. 20-24 who puts

forward the view that self-ascriptions of belief have a double role: that of describing one’s attitude and that of expressing the belief in p. Only the latter aspect of the concept of belief would lend itself to a Wittgensteinian

7

Finally, Wittgensteinian analyses, like Moorean ones, start by considering the assertion of (1) and (2). In particular, it is only because sometimes the words “I (don’t) believe” are used to weaken one’s assertion of “p”, or to assert the negation of the embedded content, that Wittgensteinians have the resources to claim that (1) and (2) come down to outright contradictions. Moreover, it is only via considerations regarding the use of those words that Wittgensteinians can claim that the content of Moore-paradoxical sentences can be neither coherently asserted, nor believed or judged. However, for Wittgensteinians too the problem arises of explaining Moore-paradoxical judgements – that is judgements that are not verbally expressed (although they may be so expressed) whose content is given by (1) and (2). In such a case the characteristic appeal to the use of the phrase “I (don’t) believe” – at least in certain contexts – is blocked and it is not easy to see what else they could appeal to in order to produce an outright contradiction of the form “p and not-p” at the level of thought.14 Notice in fact that the so-called Evans’ point – according to which in order to answer the question whether one believes that p one should put into operation whatever procedure one has for answering the question

treatment of Moore’s paradox. Yet, even if Wittgensteinian analyses worked for certain instances of Mooreparadoxical sentences they would not work for others (or, on Heal’s account, even if they worked for the latter reading of Moore-paradoxical sentences, they would not work for the former). Thus, Wittgensteinian analyses do not explain away Moore’s paradox and, when faced with genuine instances of it, are left with no means to address it. 14

This would be obvious if one held the principle that “what can be (coherently) believed constraints what can

be (coherently) asserted” but not vice versa (Shoemaker 1995, p. 227, 1n holds the quoted principle but does not explicitly commit himself to the negation of its converse. In contrast, de Almeida 2001, p. 33 holds the whole principle). Wholehearted Wittgensteinians would probably reject this principle, but more recent proponents of the Wittgensteinian analysis of Moore’s paradox, such as Jane Heal, do not explicitly reject it. Therefore, if one were to defend the Wittgensteinian analysis of Moore’s paradox one had better either motivate the rejection of the principle, or find a way of extending the Wittgensteinian analysis to the case of Moore-paradoxical judgments. As we shall see, this is no trivial matter.

8

whether p – cannot be used in this connection.15 For, although it may well be true that whenever I am in a position (warrantedly) to judge that p is the case, I would also believe that p and would automatically be in a position to judge that I believe that p – provided I have the relevant concepts – this would not produce the intended result. For Evans’ point is a belief-introduction rule – that is, a rule which licenses inferences from [p]16 to [I believe that p] – while what is needed, in order to generate a contradiction of the form [p and not-p], is a belief-elimination rule – that is, a rule which allows inferences from [I believe that p] to [p]17 – and thus allows one to eliminate the belief-operator in the doxastic conjuncts occurring in (1) and (2). It may be objected that, despite there being no contradiction of the form [p and not-p], one cannot have a justification for “I believe that p” as well as for “not-p”. For, given Evans’ point, one would have to have a justification for “p” and also for “not-p”. As we shall see at length in the following, however, one can indeed have justification for “I believe that p” and also for “not-p”. This entails that Evans’ account of our knowledge of our own beliefs as based on the very same grounds we may have for their embedded contents is highly problematical and cannot provide, as such, a satisfactory account of Moore’s paradox.

15

Evans 1982, pp. 225-226. The idea is that self-ascriptions of beliefs are not based on looking inwards and on

finding out that one has a given mental state. Rather, they draw on the same epistemic resources that would license the mere judgment (or assertion) that p. While there is much to applaud in the attempt to avoid the Cartesian epistemology of the mind, Evans’ point does not help the cause of Wittgensteinians, nor is very convincing as an account of self-knowledge. For a sustained application of Evans’ point to self-knowledge, see Gallois 1996 and Fernández 2013. For a critique, see Gertler 2011a, pp. 174-177, Ashwell 2013 and @ forthcoming. In response to the suggestion that one may use Evans’ point to explain Moore’s paradox independently of developing it into a viable account of self-knowledge, it should be noted that it would be an extremely implausible move to make. For what is under attack is exactly the distinctive claim that would be used to explain the paradox, viz. that the very same grounds would support both “p” and “I believe that p”. 16

Square brackets will be used for propositional contents.

17

This is a difficult rule to motivate, since it would imply omniscience.

9

One may then try to apply Evans’ point to generate a contradiction of the form [I believe that p and I do not believe that p]. This strategy may be successful with respect to (2). For, if a subject with the relevant conceptual repertoire is able (warrantedly) to judge that p, then she is ipso facto able to form the belief that she believes that p. Hence, in (2), by application of Evans’ point, one may derive a contradiction of the form [I believe that p and I do not believe that p]. Yet the same strategy cannot succeed with respect to (1). For, by application of Evans’ point, one derives only [I believe that p and I believe that not-p]. This, however, is not a contradiction nor is – as such – an absurd judgement.18 Let me stress one final point against Wittgensteinian analyses. If Wittgensteinians are right about the use of the phrase “I (don’t) believe”, what one entertains in thought when one utters a Moore-paradoxical sentence is the contradictory content [p and not-p]. Hence, either there are no Moorean-paradoxical thoughts – just contradictions –; or else Wittgensteinians too are forced to allow that there is a genuine occurrence, in thought, of [I (don’t) believe (that p)], which can’t be reduced to an occurrence of [p] (or of [not-p]). Thus, despite initial appearances to the contrary, Wittgensteinian analyses too might be wanting for a lack of generality, as well as for either failing to explain the platitudes, or for doing so only by denying that the verb “to believe” in the first person present can be used to describe a subject’s belief. To conclude: Moorean and Wittgensteinian analyses face at least prima facie serious difficulties. Although, no doubt, their supporters could try to finesse them, the preceding criticisms offer enough motivation to attempt to find an alternative explanation.

2. The constraints on any feasible account of Moore’s paradox

18

Heal 1994, p. 11. Williams 2007 provides a sustained defense of the application of Evans’ point (see also

Williams 2004). However, it makes use of two principles – that he calls “Evans’ Principle” and “Analogue of Evans’ Principle” (at p. 95 and 101 respectively) – that are not clearly Evans’ and relies on the problematical idea that Evans’ point can be developed into an adequate theory of self-knowledge (cf. fn. 15). For a discussion, cf. Vahid 2005, Brueckner 2006, 2009 and Williams 2009 for a response.

10

The previous analysis of both Moorean and Wittgensteinian accounts helps us see what the constraints on any feasible explanation of Moore’s paradox are. On the one hand, we do not want to lose our subjectmatter. That is to say, we do not want to lose sight of the idea that (1) and (2) can be taken to be paradoxical. Hence, we cannot put forward an analysis according to which the content of (1) and (2) is of the form [p and not-p], for there is nothing paradoxical in formal contradictions. Moreover, avoiding turning the paradox into a contradiction will allow us to respect the platitudes while holding that the verb “to believe” can be used uniformly across changes of person and tense to describe a subject’s mental state. On the other hand, we do not want just an explanation of Moore-paradoxical assertions, but also of entertaining Moore’s paradox in thought, from which an explanation of the former can either be derived, or to which it can be added, depending on one’s own theoretical preferences. Hence, the explanation of the paradox cannot proceed by appealing (just) to norms governing the pragmatics of communication. Let us then take stock and put down clearly what we are looking for. We want to be able to explain why holding in thought [I believe that p, but it is not the case that p], or [I do not believe that p, but it is the case that p] would be so absurd as to result in self-defeat. Yet, we do not want to do so by saying that those contents are equivalent to [p and not-p].

3. What Moore’s paradox isn’t about: Jane’s odd case In order to come to grips with Moore’s paradox it is instructive to realise that not all kinds of selfascriptions of beliefs (in conjunction with the relevant non-doxastic contents) would give rise to it. Consider the following story, which involves the self-ascription of a previously unconscious belief. Jane is married to Jim. They have been married for several years and have a daughter. Jane is often at home, on her own, attending to domestic chores. From time to time, she feels lonely and wishes that she had pursued her own career. More often than not, however, she feels much rewarded by the fact that her family is so serene. Indeed, when she meets with her friends, who sometimes complain

11

about their husbands, she cannot help remarking that her life makes her happy and that her husband is adorable and completely trustworthy. Still it often happens that, while preparing for the laundry, Jane carefully searches Jim’s pockets. While tidying up his studio, she opens and examines all the drawers. While dusting the furniture, she indulges on the screen of his laptop, left open on the incoming messages. One day, Ann, a psychoanalyst friend of Jane’s, approaches her and tells her about Freud’s theories concerning the unconscious. Little by little, the deep significance of a whole series of previously meaningless actions is disclosed to Jane. Ashamedly she realises that all that attention spent over the content of her husband’s pockets was a sign of her being insecure about him. All that dusting the screen of his laptop, a symptom of her thinking that he might have some intimate correspondence with another woman. Still, Jane knew all too well that Jim had always been the most truthful of men. The thought popped into her head: “I do believe that Jim is unfaithful to me, but he is not”. The moral to draw is that if one ascribes to oneself a previously unconscious belief, on the basis of the scrutiny of one’s own behaviour (together with some general background theory) and also holds the negation of its embedded content – as is the case with Jane’s final assertion – then Moore’s paradox is dissolved. In other words, in the context envisaged, judging (or asserting) [I do believe that Jim is unfaithful to me, but he is not] amounts to a correct description of one’s own mind. This is not to say that Jane would not display some kind of irrationality (cf. §4). The point, however, is that her judgement or assertion would be perfectly legitimate, while certainly depicting an odd state of mind.19

4. What Moore’s paradox is about—first pass

19

Lee 2001, pp. 366-7 has a similar story. Moran 1997, Martin 1998, Williams 2006 and Gallois 2007 make

remarks that are congenial to the gist of Jane’s story. Chan 2008 presents a case of wishful thinking in order to make a similar point. Gertler 2011b contends, contra Gendler 2008, that assertions such as Jane’s are, in the relevant context, perfectly legitimate and such that the doxastic conjunct expresses full and real belief.

12

Paying attention to Jane’s odd case actually helps us put forward a diagnostic suggestion and also identify the constraints that any adequate account of Moore’s paradox must meet in order to save the paradox and hence be an account of what it is supposed to explain. The diagnostic suggestion is this: any satisfactory account of Moore’s paradox will have to unravel the complexity of our concept of belief.20 For the mere self-ascription of a belief (or of a lack of it) is not enough invariably to generate a paradox (if accompanied by the negation of its embedded content or by the assent to it).21 Here are the constraints. (I)

One’s first-order belief must be self-known.22

For there is nothing paradoxical in unconsciously holding a first-order belief which contradicts other beliefs that one explicitly entertains. Jane, before becoming aware of the fact that her actions were partly

motivated by the belief that her husband was unfaithful to her, would openly claim that her husband was trustworthy. In such a case, one could say that Jane was self-deceived. Now, on one very plausible account of self-deception,23 this would mean that while Jane had the explicit belief that her husband was a faithful man, she also had the unconscious belief that he was not.24 No 20

Other kinds of analysis of Moore’s paradox, which hinge on the idea of unravelling the complexity of the

notion of belief, are Heal 1994 and Baldwin 2007. They stand opposed to other strategies, such as Shoemaker’s, which try to derive an explanation of Moore’s paradox from an account of self-knowledge. Self-knowledge is a necessary condition for giving rise to the paradox, but it is not sufficient for it. 21

The reader will make the necessary adjustments to apply the same kind of claim to (2).

22

Gallois 2007 defends this point, although he prefers to say that the belief must be consciously held. Williams

2006 talks of consciousness in this connection, but he is happy to endorse Rosenthal’s higher-order account of consciousness, in which the latter comes down to self-knowledge. 23

Cf. for instance Bilgrami 2006 and @ 2012.

24

Bilgrami’s account of self-deception interprets the former belief as a commitment and the latter as a

disposition. Self-deception would then be a case in which one’s (unconscious) dispositions aren’t in keeping with one’s explicitly held commitments and this creates a conflict in one’s outward behaviour (contra what Moran 1997 maintains). There will be more about this distinction in the following footnote.

13

doubt, Jane would be irrational, in some sense, 25 but there would be nothing paradoxical in her situation. Hence, a necessary condition upon any genuine instance of Moore’s paradox is that a given first-order belief be self-known. The second necessary condition we should list is this: (II)

The self-ascription of belief, which constitutes the doxastic conjunct of either (1) or (2), is not merely the self-attribution of a disposition.26

For there is nothing paradoxical in finding out and, therefore, in self-ascribing a given belief, which contradicts what one explicitly judges to be the case, as long as that self-ascribed belief is, in fact, a mental disposition.27 Intentional mental states, like beliefs and desires, as dispositions are typically manifested in a subject’s behaviour and need not be self-known.28 More importantly, they are not intrinsically normative mental states. They are not governed by intrinsic “oughts” a subject should be in a position to appreciate in order to have the relevant mental state. They merely mediate between certain inputs and behavioural outputs (together with other mental states) and do not depend on a subject’s (mental) agency. That is to say, on her ability, in the case of belief, to gather and assess evidence in favour of [p] and to use [p] as a 25

As will become evident in the following, Jane’s irrationality is not the irrationality of knowingly holding a

contradiction or of undertaking mutually inconsistent commitments (if this is a genuine possibility at all). Rather it is the irrationality of not aligning one’s dispositions with one’s commitments. 26

Functionalism treats beliefs as dispositions, and therefore as lacking intrinsic normativity. Hence, as Heal

notices (1994, pp. 12-20), it is bound to dissolve Moore’s paradox. A similar point is made by Baldwin (2007, p. 84). 27

Bilgrami (2006, Chs. 4-5) distinguishes between intentional mental states as “dispositions” and as

“commitments”. He point out that the first philosopher who systematically drew such a distinction is Isaac Levi (1983). Scanlon (1998, Ch. 1) and Moran (2001) distinguish between “brute” or “non-judgment sensitive” and “judgmentsensitive” mental states. Millar (1994) considers beliefs as intrinsically normative mental states. For further discussion, see @2012 and forthcoming. 28

When they are, they are usually known in a third-personal way. That is to say, by observation and inference

to the best explanation.

14

premise in theoretical and practical inferences. Hence, one can find oneself saddled with beliefs as dispositions, while also reflectively criticising and distancing oneself from them. Freudian unconscious beliefs like Jane’s with respect to her husband’s infidelity are a case in point.29 Indeed, in that case, the self-ascription could be a first step towards trying to put one’s beliefs in line with each other. For Jane could reason as follows: [I do have the belief (as a disposition) that my husband is unfaithful. Yet my husband is faithful. Therefore, I should stop having the belief (as a disposition) that he is unfaithful]. Accordingly, she could then undertake psychoanalytic therapy, which may eventually help her to overcome her belief as a disposition that her husband is unfaithful to her. Hence, the suggestion is that the perceived absurdity of holding (1) and (2) may come from: (III)

Taking the self-ascription of belief, which constitutes the doxastic conjunct of either (1) or (2), as the self-attribution of a commitment.

We have just seen that the self-ascription of a belief as a disposition is not enough to give rise to Moore’s paradox. However, not only do we have a notion of belief as a disposition, but we also have the notion of a belief as a commitment. There are different views about what beliefs (and other intentional mental states like desires and intentions) as commitments amount to, in the literature.30 The largely agreed upon feature of these accounts is that beliefs as commitments are attitudes of acceptance of a given propositional content that are intrinsically normative. To use Bilgrami’s helpful explanation: In this latter normative usage, to desire something, to believe something, is to think that one ought to do or think various things, those things that are entailed by those desires and beliefs by the light of certain normative principles of inference (those codifying deductive rationality, decision-theoretic

29

Freudian unconscious mental states do not exhaust the category of dispositional beliefs and desires. There

may be also beliefs and desires that shape one’s character traits. Moreover, insofar as one is willing to grant beliefs and desires to animals and infants, their intentional mental states would be dispositions. 30

See in particular Moran 2001, Millar 2004, Bilgrami 2006, @2012 and forthcoming.

15

rationality, perhaps inductive rationality and also perhaps some broader forms of material inference having to do with the meanings of words as well). (2006: 213) Typically, we form beliefs as commitments by considering and assessing the evidence at our disposal in favour of a given propositional content. Believing as a commitment is therefore the result of mental agency. It is not merely something that happens to one, perhaps unconsciously, so that, when brought to light, one will find oneself saddled with it. This is not to say that we believe at will. It merely entails that we have to be mental agents in order to form beliefs as commitments, by being able to consider and assess the evidence that bears on the truth of [p] (and of course, we can go astray in doing that). Moreover, we do have an (often implicit) understanding of the principles governing various forms of rationality that normatively constrain our beliefs as commitments. These principles concern their content (e.g. if I believe that it is raining, I will believe that there are clouds). Yet, they also concern what having that kind of propositional attitude involves (e.g. if I believe that it is raining and I don’t want to get wet, I will, ceteris paribus, go out with an umbrella, whereas if I merely so wished, I wouldn’t). All that entails that a subject capable of a belief as a commitment ought to form it on the basis of adequate (still defeasible) evidence; ought to use the proposition believed as a premise in one’s practical and theoretical reasoning; ought to provide (defeasible) reasons for it, if challenged; and ought to revise it, were sufficient contrary evidence to come in. Subjects are therefore rationally responsible for their beliefs as commitments. Should they fail to comply with these ‘oughts’ they ought to be self-critical or accept to incur criticism. The contrast between beliefs as commitments and dispositions, then, could be illustrated by Jane’s situation. She finds herself with a belief as a disposition that her husband is unfaithful to her. That disposition shapes much of her behaviour and can have various causes. However, she also has and avows her belief as a commitment, held on the basis of evidence that she herself has assessed, that he is not. The letter belief exerts normative force on her and, consequently, she ought to try to get rid of her recognizably irrational disposition. In the end, she might not be able to overcome it (completely). Still, if rational, she

16

ought to recognise that that is what is required of her, given her belief as a commitment that her husband is a faithful man. What Bilgrami’s quote leaves unspecified and is not clarified, to the best of my knowledge, in the literature on commitments, is whether one can knowingly and willingly hold inconsistent commitments and therefore bind oneself, no matter how irrationally that would turn out to be, to opposite courses of actions (and, consequently, to accepting criticism, were one not to behave accordingly). We will presently see the bearing of these different takes on the notion of commitment onto the analysis of Moore’s paradox. Yet, before moving on to that, let us consider the relationship between beliefs as commitments and dispositions a bit more. As Bilgrami’s quote made apparent, having a belief as a commitment consists in knowingly and willingly binding oneself to those courses of action that “are entailed by those desires and beliefs by the light of certain normative principles of inference”. If one does not comply with them, one will be held responsible for not doing so and will have to be self-critical or accept criticism from others for it. Thus, to have a belief as a commitment entails seeing oneself as having to implement a certain behaviour (and accepting criticism for not “living up to one’s commitments” should one fail to behave accordingly). Therefore, there is an internal link between the content of one’s belief as a commitment and the kind of actions that one ought to perform given that belief. However, theorists disagree on whether, in order to count as really having the commitment in question, one also has actually to comply with the course of action it mandates, at least on occasion, or whether no such behaviour is required in order to count as having the commitment.31 Those who favour the former option will appeal to the idea that of a subject who, for example, persistently professed “I ought to help the poor” and never did anything to that effect, we would presumably be inclined to say that she thinks she has the commitment to help the poor, but actually she does not. In contrast, those who favour the latter option would argue that insofar as such a subject recognized what would be required of her and were self-critical or accepted criticism for not complying with her

31

Millar (2004) maintains the former option, while Bilgrami (2006) and (2012, p. 265) the latter.

17

commitment, she would count as having it nonetheless. Clearly, the two readings impose more or less stringent requirements with respect to the conditions that have to be met in order to have beliefs as commitments and propose a different configuration of the interplay between commitments and dispositions. This issue is of great interest and significance and it would merit a separate treatment.32 It is not, however, of immediate relevance for the present discussion and I will show why in due course. What is relevant, at this stage, is simply that the internal link between the content of one’s belief as a commitment and the course of action it mandates be recognized (yet not necessarily implemented, on at least one possible understanding of the notion of a belief as a commitment) by the subject who could not otherwise have that belief as a commitment. Let us now turn to the kind of analysis of Moore’s paradox that can be offered if one held a “liberal” notion of commitment, which allows for the possibility that a subject could knowingly and willingly undertake incompatible commitments. Accordingly, the analysis of Moore’s paradox would be as follows.33 Being committed to [p], one ought to use it as a premise of one’s practical and theoretical reasoning. Hence, if Jane’s judgement (or assertion) contained the self-ascription of a commitment – as opposed to the selfascription of a disposition – to holding that her husband is unfaithful to her, this would give rise to a form of

32

On the former option, for instance, beliefs as commitments would involve first-order dispositions (to behave

as believing that p requires), while on the latter they would involve only second-order dispositions (to be self-critical or accept criticism if one did not behave as required by one’s beliefs as commitments). Moreover, on the former option one could fail to be authoritative with respect to intentional mental states as commitments. Thus, one could self-ascribe a mental state as a commitment and not have it. On the latter, instead, such a possibility would seem to be ruled out, so long as one’s self-ascription were sincerely made and were underwritten by the relevant conceptual capacities. They may also have different bearings on whether intentional mental states are ultimately amenable to a naturalist understanding. 33

I surmise that Baldwin (2007) puts forward something along these lines, for he allows for the possibility of

finding “ourselves from time to time with inconsistent commitments” (p. 86). What would be absurd, on his analysis, would be “to make commitments whose inconsistency is obvious in the very judgment itself” (ibid.).

18

irrationality. For if the doxastic conjunct self-ascribed such a commitment, then Jane ought to use [My husband is unfaithful to me] as a premise of her reasoning.34 By also assenting to its negation, however, she would commit herself to using (knowingly and willingly) that content as a premise of her reasoning. Thus, she would commit herself (knowingly and willingly) to reason from contradictory premises. And this would be irrational. Take now an instance of (2) – [I do not believe that p, but it is the case that p]. One may take its first conjunct as a self-ascription of one’s open-mindedness as to whether [p] is the case. Accordingly, it would describe one’s commitment to using neither [p] nor [not-p] as a premise of one’s reasoning. However, by holding the second conjunct, one would also commit oneself (knowingly and willingly) to using [p] as such a premise. And this, once again, would be irrational. On the face of it, then, the notion of a belief as a commitment seems to allow us to save the paradox because it helps us show that while (1) and (2) do not amount to contradictions, holding them in thought would be a manifestation of irrationality. The conclusion one should draw would thus be this: any creature capable of beliefs as commitments ought not to judge either (1) or (2) – when construed as satisfying conditions (I)-(III) – on pain of irrationality.

5. What Moore’s paradox is about—second pass However, this is a somewhat weak result: people are often irrational and so, if we allow that they can have inconsistent beliefs as commitments, it seems that one could really judge (1) or (2). That is to say, it seems that one could really be in the mental state of committing oneself to [p], say, while also knowingly and willingly denying that [p] is the case (and, mutatis mutandis, given the omissive form of Moore’s paradox, one could then be open-minded with respect to [p], while also endorsing it). In particular, it does not seem impossible to reason from contradictory premises, like in the case of (1), or to reason from neither [p] nor [not-p] and [p], in the case of (2). It is just that these mental states would – unluckily – be irrational ones.

34

Obviously she would also commit herself to using [I believe that p] as a premise of her reasoning.

19

Yet, if it were possible for a subject to endorse incompatible commitments, what would then prevent one from judging and even from asserting that one did? Of course such a situation would be comparatively rare – or so one would hope. However, if it is a possibility, the corresponding judgement and assertion should be too. The absurdity perceived in hearing either (1) or (2) would be just due to the fact that we are inductively unfamiliar with open admissions of irrationality, which, however, on the proposed conception of commitment, would be entirely possible. Yet, if this were the explanation of why we find the assertion of (1) or (2) absurd, then to label Moorean contents “paradoxical” would be an overstatement. There are plenty of assertions that strike us as weird because – supposing they are true and sincere – they are at odds with our experience, but this would not justify calling them “paradoxical”. Furthermore, it was a datum of the problem that when we hear Moorean sentences we perceive some kind of real impossibility, despite the lack of a contradiction. Whereas, so far, our analysis has revealed no impossibility, just the fact that if those judgements were actually made, they would correspond to an irrational state of mind. In order to justify the idea that there is a real impossibility here and hence a genuine paradox, we should understand the notion of a belief as a commitment in a more demanding way, and, in particular, as entailing a negative answer to the question whether it is possible knowingly and willingly to hold incompatible commitments. Accordingly, one could not (logically) have the belief as a commitment that p is the case and knowingly and willingly assent to its negation, for this would actually undo one’s previous commitment. To see why this more demanding notion could have traction, consider the internal link between the content of one’s beliefs as commitments and the courses of action, mandated by them, which ought to be recognized by a subject in order to have those very beliefs as commitments. Let me stress a point made in the previous section. Namely, that having a belief as a commitment entails seeing oneself as bound to the kind of action mandated by that commitment, although, on certain readings of commitments, one may not be able to implement it. It is a further theoretical issue, which we left open for the purposes of this paper, whether one actually ought to have the relevant first-order dispositions, in order to count as

20

having the relevant belief as a commitment. For our purposes, it is enough to recognise that having a belief as a commitment entails (at least) the relevant second-order disposition of being self-critical or of accepting criticism from others if one did not behave as mandated by one’s belief as a commitment. Take a case in which the belief is part of practical reasoning. Suppose, therefore, that a subject desires not to get wet, if (and only if) it rains. Suppose, further, that the only possible action, in order not to get wet, is to open the umbrella that the subject happens to carry with her and that there are no countervailing considerations. Now, if she has the belief as a commitment that it is raining, she ought to see herself as bound to opening the umbrella. If she has the belief as a commitment that it is not raining, she ought to see herself as bound to not opening the umbrella. Finally, if she is open-minded with respect to whether it is raining, she ought to see herself as neither bound to opening the umbrella, nor to keeping it close. The three kinds of action, which are respectively internally linked to each kind of belief as a commitment a subject might have, however, are mutually exclusive. Since, in order to have a belief as a commitment, a subject will have to recognize the courses of action it internally mandates, given they are mutually exclusive, she will not be in a position (logically) to bound herself to any two of them at once. Thus, no matter what she could think of herself as doing, she could not actually have incompatible beliefs as commitments.35 To stress, this is a point about logic and the internal link between a belief as a commitment and the kind of action it mandates and one ought to see oneself as bound to, in order to count as having the commitment. Of course, since we are open to the possibility that a subject may have a commitment while behaving contrary to what it 35

A similar conclusion can be reached by considering the ‘oughts’ as input that should be appreciated by a

subject in order to have the relevant beliefs as commitments. Namely, that one ought to believe that p only if there is sufficient evidence in its favour. So one could self-ascribe the belief as a commitment that p only if one had sufficient evidence in favour of [p]. Yet, one cannot have sufficient evidence for [p] and also and at the same time for its negation. Hence, one cannot possibly have the inconsistent commitments entailed by (1). Similarly, one can be committed to open-mindedness with respect to [p] only if one had neither sufficient evidence in favour of [p] nor for its negation. Yet, one cannot have, at once, no sufficient evidence for [p] and [not-p] and sufficient evidence for of [p]. Hence, one cannot possibly have the inconsistent commitments entailed by (2).

21

mandates, we have to be open to the possibility that she might act contrary to what the belief (as a commitment) that it is raining mandates, in the kind of scenario just envisaged. However, in such an event, so long as she did recognize that she hasn’t “lived up” to her commitment, she would count as having it in the first instance, nonetheless. Notice, once again, that even if she behaved contrary to her commitment, and therefore didn’t have the relevant first-order disposition, she should have a related second-order one, which depends precisely, as required, on seeing herself as bound to act in certain ways. Hence, if she didn’t act accordingly, she should be self-critical or accept criticism from others for not doing so. What couldn’t be done, therefore, on such a less demanding notion of a commitment, is not to have a belief as a commitment and act in ways that run contrary to it. Rather, it is to have the former and think it may be all right to act in ways that run contrary to it. Yet, to stress, the constitutive connection that is necessary in order to come to grips with Moore’s paradox, viz. the one concerning the content of one’s commitments and the kind of actions mandated by them one should see oneself as bound to, would be respected nonetheless. That is to say, even if one might fail to bring these actions about. Since the courses of action mandated by inconsistent commitments are mutually exclusive, one can’t have the belief as a commitment that p while also knowingly and willingly assenting to its negation (or be open-minded with respect to [p], while knowingly and willingly assenting to it or to its negation), even if one actually behaved in ways that run contrary to [p]. By contrast, as we have seen with Jane’s case, one can perfectly well have the belief as a disposition that p while also knowingly and willingly assenting to its negation. Hence, one might be saddled with beliefs (as dispositions) that one does not endorse reflectively, act on them and even overtly recognise the kind of situation one is in. Yet, it seems that, on such a more demanding notion of a belief as a commitment, one cannot even have the belief as a commitment that p if one also assented – and thereby committed oneself – to its negation. The assent to the negation would thus undo one’s previous commitment to [p]. Therefore, considering (1), if the doxastic conjunct were the self-ascription of a commitment, understood in the way proposed, that commitment could so much as exist only as long as no known and willing assent to

22

the negation of its content were in place. Hence, if the doxastic conjunct is a self-ascription of a genuine commitment no assent to its negation is possible. Conversely, if such an assent is in place, then there cannot be any commitment to [p] in the first place. Take now an instance of (2) – [I do not believe that p, but it is the case that p] – where the first conjunct is taken as a self-ascription of a commitment, understood in the way proposed. By judging that p is the case one would undo one’s previous commitment to open-mindedness with respect to [p]—i.e. one’s commitment to using neither [p], nor [not-p] as a premise of one’s reasoning. Hence, again, if the doxastic conjunct is an ascription of a genuine commitment to open-mindedness, no known and willing assent to its embedded content is possible. Yet, if such an assent is in place, there cannot be any genuine commitment to open-mindedness in the first place. Therefore, the moral to draw is that any creature capable of beliefs as commitments could not possibly judge either (1) or (2), despite the fact that their conjuncts have independent truth-conditions, when (1) and (2) are construed as satisfying conditions (I)-(III). For to do so would actually be self-defeating. Finally, the absurdity of asserting either (1) and (2) can be explained as follows. Usually, although not invariably, we assert the contents of our beliefs as commitments.36 As we have seen, however, if (1) and (2) are asserted as involving beliefs as commitments, they wouldn’t simply be a manifestation of irrationality, but would actually be self-defeating, in the sense of expressing an impossible cognitive situation – i.e. the situation of endorsing incompatible commitments.

6. An objection The previous account of Moore’s paradox gives rise to an immediate objection, for it seems that sometimes we find ourselves with inconsistent commitments. Consider, for instance, the case of a man with two

36

This does not require taking issue with any of the various proposals about the norm of the assertion available

in the growing literature on that topic. For the claim is that we usually assert the content of our commitments, not that this is the norm of correct assertions.

23

families who, at lunch-time says to his first family “I will go on holiday with you[family 1]”; and at dinner-time says, to his second family, “I will go on holiday with you[family 2]”, and thereby undertakes inconsistent commitments.37 Let us analyze this case in more detail. There seem to be only four options. First, he means to compatibilize by dividing his holidays between the two families (or by taking them all to the same place at the same time). Second, he has changed his mind between lunch and dinner and will go on holiday with the second family only. Third, he has forgotten the promise made over lunch and is therefore not knowingly and willingly undertaking two incompatible commitments.38 Finally, by willingly trying to bind himself to knowingly incompatible courses of action he keeps alternating between the two, up to the point of actually losing the capacity of committing himself to either. Yet, there is no single overall situation in which he is really knowingly and willingly committed to both at once. This point may escape notice because we have what we might call a “social” or “third-personal” notion of commitment, such that people who have heard his pronouncements will actually take him to have bound himself on both occasions and then entered a state of holding incompatible commitments. However, he

37

Another possibility is to have beliefs as commitments that, though not incompatible as such, in given

circumstances, may impose incompatible courses of action. Consider the belief as a commitment that a supervisor should help her students and the other belief as a commitment that one should give an academic position to the best candidate. Obviously, there is no incompatibility between the two commitments. Hence, one can knowingly and willingly assent to both. If, however, the situation arises when there is a conflict, one will have to prioritise one commitment over the other. 38

Cases like Frege’s puzzle where one can assent to ‘Hesperus is a planet’ while denying or not assenting to

‘Phosphorous is a planet’ are cases in which the beliefs are incompatible but this is not known to the subject, who precisely ignores that Hesperus is identical to Phosphorus. Hence, it is important to qualify the claim here: it is only for knowingly incompatible beliefs as commitments that one cannot (knowingly and willingly) have both of them at once. Another way to see the same point is that to judge “Hesperus is a planet, but Phosphorus it is not” (for a subject who didn’t know that Hesperus is Phosphorus) wouldn’t be Moorean-paradoxical.

24

has not. For either he holds compatible commitments, or he has broken or forgotten the promise made to one of the two families – in which case he may be accused of not being trust-worthy or of being absentminded. Yet, he cannot be accused of having personal (or first-personal) incompatible commitments. In the first case, they are compatible, in the second he has only one commitment or neither, and he will have to deliberate whether to confirm the forgotten promise and break the other one or vice versa. Note that I do not wish to oppose such a “social” or “third-personal” notion of a commitment. The point of the last two sections of the paper, however, has been to press the need for the recognition of what we may call a “personal” or “first-personal” notion of a commitment. That is to say, the notion of a particular kind of belief (with a characteristic course of action one would have to see oneself as bound to, even if one might not live up to it), which one can enter only be willingly and knowingly assenting to a given content, and which is incompatible with (at least) believing the opposite. Hence, first, we cannot just find ourselves with first-personal commitments, incurred out of habit or out of compliance with social expectations. Rather, first-personal commitments are always the result of a conscious deliberation and therefore of agency on a subject’s part. Even the humdrum case of believing that it is raining because one sees the rain fall outside one’s window will be the result of consciously considering that perceptual evidence and of deeming it sufficient to warrant one’s belief. Second, for all we have seen in the previous two sections, a person may still have a given belief as a commitment, while not behaving in keeping with it, so long as she were prepared to be self-critical or to accept criticism for her dissonant behavior. Hence, in the case of the man with two families, he could still have the commitment to going on holiday with the first family, while actually ending up going on holiday with the other one (or simply saying so to the second family), so long as he were prepared to be self-critical or to accept criticism for so doing. Notice, in this case, that his first-order dispositions, which would run contrary to his commitment, would give rise to courses of action, which would be carried out somehow “against one’s own will and judgment” as it were, perhaps out of fear or habit. If not, i.e. if the man had in fact willingly taken the second family on holiday (or actually committed himself to doing so), we would then

25

be back to the second possibility envisaged above, i.e. the one in which he would have changed his mind and relinquished the promise made over lunch to his first family. If this account is on the right track, we can actually see why Moore’s paradox is indeed so perplexing: not because it describes a situation, which, as irrational as it may be, can actually happen. For, in that case, one should just acknowledge that there will be occasions that would license the judgment (and the assertion) of either (1) or (2), as weird as they might be. Down that route, however, one would no longer have the means to maintain that Moorean contents are paradoxical. They would just seem weird because they would depict comparatively infrequent and unusual states of affairs. Rather, Moore’s paradox is indeed a genuine and deep paradox because, despite not containing any contradictory content, judging or asserting (1) or (2) would express an impossible cognitive situation. That is to say, the situation of endorsing incompatible (personal) commitments. If this is right, then, we will have the means to understand why, as Moran remarked, Moore’s paradox is indeed “an emblem for peculiarities in the first person point of view”.

References @ 2005. @ 2012. @ forthcoming. Almeida, C. de 2001 “What Moore’s paradox is about”, Philosophy and Phenomenological Research 62/1, pp. 33-58. Ashwell, L. 2013 “Review of Transparent Minds. A Study of Self-Knowledge”, Notre Dame Philosophical Reviews, http://ndpr.nd.edu/news/42227-transparent-minds-a-study-of-self-knowledge/ Baldwin, T. 1990 G. E. Moore, London and New York, Routledge. Baldwin, T. 2007 “The normative character of belief”, in M. Green and J. N. Williams 2007 (eds.), pp. 76-89. Bilgrami, A. 2006 Self-Knowledge and Resentment, Cambridge (Mass.), Harvard University Press.

26

Bilgrami, A. 2012 “The unique status of self-knowledge”, in A. Coliva (ed.) The Self and Self-Knowledge, Oxford, Oxford University Press, pp. 263-278. Brueckner, A. 2006 “Justification and Moore’s paradox”, Analysis 66/3, pp. 264-266. Brueckner, A. 2009 “More on justification and Moore’s paradox”, Analysis 69/3, pp. 497-499. Chan, T. 2008 “Belief, assertion and Moore’s paradox”, Philosophical Studies 139/3, pp. 395-414. Chan, T. 2010 “Moore’s paradox is not just another pragmatic paradox”, Synthese 173/3, pp. 211-229. Collins, A. W. 1996 “Moore’s paradox and epistemic risk”, Philosophical Quarterly 46/184, pp. 308-319. Gallois, A. 1996 The World Without, the Mind Within, Cambridge, Cambridge University Press. Gallois, A. 2007 “Consciousness, reasons and Moore’s paradox”, in M. Green and J. N. Williams 2007 (eds.), pp. 165-188. Gendler, T. 2007 “Self-deception as pretense”, Philosophical Perspectives 21, pp. 231-258. Gendler, T. 2008 “Alief and belief”, Journal of Philosophy 105/10, pp. 634-663. Gertler, B. 2011a Self-Knowledge, London – New York, Routledge. Gertler, B. 2011b “Self-knowledge and the transparency of belief”, in A. Hatzimoysis (ed.) Self-Knowledge, Oxford University Press, pp. 125-45. Green, M., Williams, J. N. 2007 (eds.) Moore’s Paradox. New Essays on Belief, Rationality and the First Person, Oxford, Clarendon Press. Heal, J. 1994 “Moore’s paradox: a Wittgensteinian approach”, Mind 103/409, pp. 5-24. Lee, B. D. 2001 “Moore’s paradox and self-ascribed belief”, Erkenntnis 55/3, pp. 359-370. Levi, I. 1983 Enterprise of Knowledge, Cambridge (Mass.), MIT Press. Linville, K. and Ring, M. 1991 “Moore’s paradox revisited”, Synthese 87/2, pp. 295-309. Malcolm, N. 1995 “Disintangling Moore’s paradox”, in R. Egidi (ed.) Wittgenstein: Mind and Language, Dordrecht, Kluwer, pp. 195-205. Martin, M. 1998 “An eye directed outward”, in C. Wright, B. C. Smith and C. Macdonald (eds.) Knowing Our Own Minds, Oxford, Clarendon Press, pp. 99-121.

27

Millar, A. 2004 Understanding People. Normativity and Rationalizing Explanation, Oxford, Oxford University Press. Moore, G. E. 1942 “Reply to my critics”, in P. Schilpp (ed.) The Philosophy of G. E. Moore, La Salle, Illinois, Open Court, pp. 535-677. Moore, G. E. 1944 “Russell’s theory of descriptions”, in P. Schilpp ( ed.) The Philosophy of Bertrand. Russell, La Salle, Illinois, Open Court, pp. 175-225. Moran, R. 1997 “Discovery, resolution and undoing”, European Journal of Philosophy 5/2, pp. 141-161. Scanlon, T. 1998 What We Owe to Each Other, Harvard, Harvard University Press. Shoemaker, S. 1995 “Moore’s paradox and self-knowledge”, Philosophical Studies 77/2-3, pp. 211-228. Reprinted (with small variations) in Shoemaker, S. 1996 The First Person Perspective and Other Essays (Cambridge, Cambridge University Press). Sorensen, R. 1988 Blindspots, Oxford, Oxford University Press. Turri, J. 2010 “Refutation by elimination”, Analysis 70/1, pp. 35-39. Vahid, H. 2005 “Moore’s paradox and Evans’ principle: a reply to Williams”, Analysis 65/4, pp. 337-341. Williams, J. N. 2004 “”Moore’s paradoxes, Evans’s principle and self-knowledge”, Analysis 64/4, pp. 348353. Williams, J. N. 2006 “”Moore’s paradoxes and conscious belief”, Philosophical Studies 127/3, pp. 383-414. Williams, J. N. 2007 “Moore’s paradox, Evans’ principle, and iterated beliefs”, in M. Green and J. N. Williams 2007 (eds.), pp. 90-113. Williams, J. N. 2009 “Justifying circumstances and Moore-paradoxical beliefs: a response to Brueckner”, Analysis 69/3, pp. 490-496. Wittgenstein, L. 1953 Philosophical Investigations, Oxford, Basil Blackwell. Wittgenstein, L. 1980 Remarks on the Philosophy of Psychology, Vol. 1, Oxford, Basil Blackwell.

28