in defence of burge's thesis

In Defence of Burge's ThesisAuthor(s): Sarah SawyerSource: Philosophical Studies: An International Journal for Philosophy in the AnalyticTradition, Vol. 107, No. 2 (Jan., 2002), pp. 109-128Published by: SpringerStable URL: http://www.jstor.org/stable/4321214 .

Accessed: 28/06/2014 08:45

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: AnInternational Journal for Philosophy in the Analytic Tradition.

http://www.jstor.org

This content downloaded from 185.31.195.34 on Sat, 28 Jun 2014 08:45:30 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=springer

http://www.jstor.org/stable/4321214?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


SARAH SAWYER

IN DEFENCE OF BURGE'S THESIS

(Received 31 January 2001; received in revised version 5 April 2001)

ABSTRACT. Burge's thesis is the thesis that certain second-order self-ascriptions are self-verifying in virtue of their self-referential form. The thesis has recently come under attack on the grounds that it does not yield a theory of self-knowledge consistent with semantic externalism, and also on the grounds that it is false. In this paper I defend Burge's thesis against both charges, in particular against the arguments of Bernecker, Gallois and Goldberg. The alleged counterexamples they provide are merely apparent counterexamples, and the thesis is adequate to its proper task. To think otherwise is simply to misunderstand the thesis.

INTRODUCTION

Burge's thesis is the thesis that instances of a certain class of second-order judgement are self-verifying in virtue of their self- referential form. As such, the judgements concerned are immune to error. This thesis was originally presented by Tyler Burge (1988) in partial response to the allegation that semantic externalism is unable to account for the authoritative nature of self-attributions of thoughts. Since then, the thesis has been the subject of attack on three grounds. First, because it deals only with the content of thoughts and says nothing of the attitudinal component of those thoughts. Second, because it deals with the contents of the members of only a small subset of second-order thoughts. Third, it has been claimed outright to be false. In this paper I will defend Burge's thesis against all three charges. In particular I will focus on the criticisms presented by Sven Bernecker (1996), Andre Gallois (1996), and Sanford Goldberg (2000).

The defence falls into two parts. First, I will defend Burge's thesis against putative counterexamples. Second, I will argue that while critics are right to maintain that Burge's thesis cannot provide an adequate account of the authoritative nature of self-ascriptions in general, neither Burge nor so-called "Burgean compatibilists"

La Philosophical Studies 107: 109-128,2002. 0 2002 Kluwer Academic Publishers. Printed in the Netherlands.



110 SARAH SAWYER

maintain that it can. Compatibilists do not typically derive their compatibilism from Burge's thesis; but the thesis is adequate to its proper task.

1. THE PROBLEM OF INCOMPATIBILITY AND BURGE'S THESIS IN BRIEF

The emergence of semantic extemalism, according to which a subject's set of actual and possible thoughts is dependent upon, and restricted by, relations that subject bears to her environment, has been taken to threaten the claim that a subject can be authoritative about which thoughts she has. This can be most clearly illustrated by considering a "travelling case" in which a hypothetical subject is unknowingly switched between different environments. The environments are assumed to differ in imperceptible ways, ways which nevertheless suffice to cause a change in the concepts that subject possesses and hence in the thoughts it is possible for her to entertain. The differences between the environments, then, are semantically relevant. The question is how a subject could be authoritative about the contents of her thoughts given that she would be ignorant of a change in her body of concepts were she to be switched between such. It would seem that a subject is, as Gallois puts it, "subject to facsimile-based error".1 If one's thoughts depend essentially on one's relations to particular environmental conditions, and knowledge of those environmental conditions is available only through empirical investigation, knowledge of one's thoughts can likewise be available only through empirical investigation. So the argument goes.2

In his (1988), Burge draws our attention to instances of what he terns "cogito-like judgements". Even though the contents of such judgements may be determined by relations the subject bears to her environment, such judgements are, Burge maintains, self- referential, and self-verifying in virtue of being so. The existence of such judgements demonstrates that semantic externalism need not threaten authoritative self-knowledge. He offers as examples of cogito-like judgements those such as I am now thinking that writing requires concentration, and I hereby judge that examples need elaboration. Burge writes,



IN DEFENCE OF BURGE'S THESIS 111

The content of the first-order thought is fixed by non-individualistic background conditions. And by its reflexive, self-referential character, the content of the second-order judgement is logically locked (self-referentially) onto the first-order content which it both contains and takes as its subject matter. (Burge, 1988, pp. 659-660)

The first- and second-order thoughts cannot come apart. Accord- ingly Burge claims that "[n]o errors at all are possible in strict cogito judgements."3

An error based on a gap between one's thoughts and the subject matter is simply not possible in these cases. ... If background conditions are different enough so that there is another object of reference in one's self-referential thinking, they are also different enough so that there is another thought. (Burge, 1988, p. 659)

Cogito-like judgements, instances of what Burge calls "basic self-knowledge", are taken to be paradigms of authoritative self- knowledge. Such judgements are not subject to error, and afortiori not subject to facsimile-based error, even if an externalist theory correctly applies to the content of those judgements.

2. PUTATIVE COUNTEREXAMPLES: GALLOIS

Both Goldberg and Gallois maintain that Burge's thesis is subject to counterexample. I begin with a discussion of Gallois's claims. Gallois identifies in Burge's (1988) three distinct reasons for maintaining that second-order thoughts are not subject to facsimile- based error. The first is that the content of a second-order thought will be determined by the same external conditions that determine the content of the corresponding first-order thought. Gallois does not himself dispute this first claim, and the claim is prevalent in the literature. I will return to it in Section 5 below. The second reason he identifies is that cogito-like judgements are infallible. This is the heart of Burge's thesis, as indicated above, and it is this to which Gallois primarily takes exception.

Gallois writes,

the ... thesis that thinking that one thinks that p entails thinking that p ... seems implausible. Surely I can mistakenly ascribe a thought to myself. Convinced by a dogmatic psychoanalyst I think that I think that my mother does not love me even though I have no such first-order thought. (Gallois, 1996, p. 173)



112 SARAH SAWYER

Thus the fact that I can think that I think my mother does not love me even when I do not think any such first-order thought, is taken by Gallois to be a counterexample to Burge's thesis about the self-verifying nature of certain self-ascriptions. However, to take Gallois's example to be a counterexample is to fail to recognise an explicit restriction that Burge places on his thesis.

Burge maintains that not all present-tense self-ascriptions are cogito-like judgements, and hence that not all self-ascriptions of thought-contents will be self-verifying. He takes cogito-like judgements as paradigm instances of self-knowledge, but maintains that "[lthere is a range of cases of self-knowledge which extend out from this paradigm", and remarks that "[d]ealing with the whole range requires subtlety."4 He contrasts such cogito-like judgements with, amongst other forms of self-ascription, self-ascriptions of standing propositional attitudes. This is far the more common case. Examples would be judgements such as I believe that Paris is the capital of France, or I would like to live by the sea, where the judgement is in each case about a standing state, which together with a triggering occurrence helps cause the self-attribution. He maintains that such self-attributions can still be authoritative even though they are not self-referential, and hence not self-verifying: an indication that he does not take his thesis to provide an adequate account of authoritative self-knowledge simpliciter (of which more later). In any case, the restriction here is crucial, and paying heed to it allows us to dispose of the putative counterexample to Burge's thesis Gallois offers.

The problem with Gallois's example is simply that it is not an instance of a cogito-like judgement, and hence falls outside the scope of Burge's thesis. That Gallois fails to acknowledge this restriction in scope, that he fails to distinguish between the cogito- and the non-cogito-like judgements, is further evidenced by his endorsement of the following principle, a principle that he complains Burge has given no reason to think correct:

(1) If any first-order thought corresponds to the second-order thought that I think that p, it must have the content that p.

He illustrates his point by considering the following judgement:

(2) I think that writing requires concentration.




Gallois writes,

Certainly, when I entertain [(2)] I have a thought whose content is, at least in part, that writing requires concentration. Nevertheless, for [(2)] to be true, it must correspond to a thought whose content is solely that writing requires concentration. What precludes [(2)] from corresponding to a thought whose content has nothing to do with writing? What prevents [(2)] from corresponding to the thought that doodling requires concentration?

Answers to this question depend on what is meant in saying that one thought corresponds to another ... suppose we take [(2)] to be about whichever thought causally occasions [it] in the right way. If so, then for all Burge has said, [(2)] could correspond to a thought whose content has nothing in common with that of [(2)]. ... Consequently, [Burge] has failed to show that basic self-knowledge is even self-verifying in the limited sense implied by [(1)]. (Gallois, 1996, pp. 173-174, original emphasis)

This passage is indicative of Gallois's belief that every second- order judgement, if it is to be knowledgeable, must correspond to an independent first-order thought with the appropriate content. But Burge explicitly rules out any such notion of correspondence between cogito-like judgements and their subject matter. He maintains that, "[i]n the case of cogito-like judgements, the object, or subject-matter, of one's thoughts is not contingently related to the thoughts one thinks about."5 In such cases neither the truth nor the warrant for the judgement depends on any causal relation between judgement and subject-matter, since the subject-matter and the judgement about it occur in the same act. Such judgements are either logically or performatively self-verifying. (A judgement such as I am now thinking that writing requires concentration would be an example of the former; one such as I hereby judge that examples need elaboration an example of the latter.) The warrant for such judgements depends on understanding, which is to say that if one understands the judgement in question the fact that it is self- verifying is evident to one, and hence one acquires a warrant for the judgement through understanding, since this involves understanding that it is self-verifying.

These are contrasted with non-cogito-like judgements, including self-attributions of standing states, the truth and warrant for which do depend on a contingent, presumably causal, relation between themselves and the thoughts they are about. Gallois's principle (1), then, applies to non-cogito-like cases only. To think that Burge must



114 SARAH SAWYER

offer us reason to believe (1) for all self-ascriptions is to misunderstand the nature of the cogito-like cases to which Burge draws our attention. It is precisely the absence of contingency, the absence of a causal relation between judgement and subject matter, which ensures that (2) could not correspond to a thought whose content has nothing in common with it.

Let us now turn to the third reason Gallois attributes to Burge for maintaining that second-order thoughts are not subject to error. Gallois cites the following two passages from Burge:

Consider the thought, 'I hereby judge that water is a liquid'. What one needs in order to think this thought knowledgeably is to be able to think the first- order, empirical thought (that water is a liquid) and to ascribe it to oneself, simultaneously. (Burge, 1988, p. 656)

and:

One knows one's thought to be what it is simply by thinking it while exercising second-order self-ascriptive powers ... Getting the 'right' one is simply a matter of thinking the thought in the relevant reflexive way ... We 'individuate' our thoughts, or discriminate them from others, by thinking those and not the others self-ascriptively. (Burge, 1988, p. 656)

Gallois interprets these passages as suggesting that, according to Burge, second-order thoughts are "constructed upon" first-order thoughts, in much the same way that 'I am writing "writing requires concentration",' is constructed by writing 'Writing requires concentration', placing it in quotes, and prefixing it with 'I am writing'. By analogy, then, when a subject entertains (2) above it is constructed from the thought that writing requires concentration by prefixing it with the mental analogue of the expression 'I think that'.6

Gallois admits that if this were the correct interpretation of basic self-knowledge, then such knowledge would be self-verifying. However, he then goes on to say that the interpretation must be mistaken on the grounds that "[ilt does seem possible to misattribute a current thought to oneself on the basis of misleading evidence",7 and we are directed to his earlier example of the thought self- attributed on the basis of evidence provided by the dogmatic psychoanalyst. Accordingly, Gallois claims that "[a]ny theorist who rules out this possibility has to explain why, appearances to the contrary, this is not possible."8




However, as we have seen above, the possibility that one might misattribute a thought to oneself is not ruled out by Burge's thesis. It is ruled out in the case of cogito-like judgements, but these form a (rather small) subclass of self-attributions. Since Burge maintains that error is possible in a whole range of cases outside cases of basic self-knowledge, including, importantly, the case in question, he is not one of the theorists of whom Gallois can legitimately demand an explanation. In short, there is no contradiction in upholding Burge's thesis and allowing the possibility of error in cases such as those to which Gallois refers.

Neither does Burge intend his claims about the self-verifying nature of cogito-like judgements to provide an adequate account of the knowledgable and authoritative status of self-ascriptions in general. Reflection on the cogito-like cases yields, Burge says, "at most an illuminating paradigm for understanding a significant range of phenomena that count as self-knowledge. ... A full discussion of authoritative self-knowledge must explicate our special authority, or epistemic right, even in numerous cases where our judgements are not self-verifying or immune to error."9 In subsequent writings he goes on to offer an account of the epistemic status of self- ascriptions, maintaining that it is the role that self-ascriptions play in critical reasoning that accounts for their special authority. 10 We will not be concerned with his account of authoritative self-knowledge in any detail here, although I will briefly reconsider it in Section 5 below.

For the moment, let us turn to an examination of Goldberg's putative counterexample.

3. PUTATIVE COUNTEREXAMPLES: GOLDBERG

Goldberg asks us to consider the following dialogue.

(tl) Sam: I am thinking that Uncle Harry is awfully ugly. (t2) Nancy: You must be thinking that Uncle Henry is awfully

ugly; everyone knows that Uncle Harry is a dreamboat (and the ugly man you saw yesterday standing by the soda fountain was Uncle Henry)!

(t3) Sam: You're right, I must have been thinking that Uncle Henry was ugly. (Goldberg, 2000, p. 3)



116 SARAH SAWYER

Sam's agreement at t3 is taken, by Goldberg, to establish that Sam's utterance at tI, if taken literally, is false. Hence, Goldberg maintains, if we are not to view this as a counterexample to Burge's thesis, Sam's utterance must be given a plausible non-literal interpretation. Any plausible altemative interpretation must be consistent both with, minimally, the truth of the self-ascription thereby expressed, and with the general externalist position. Goldberg considers two alternative interpretations. According to the first, Sam's utterance expresses the thought that he is thinking that Uncle Henry is ugly. According to the second, Sam's utterance expresses a descriptive thought of the form I am thinking the qo is F. Both interpretations run into difficulties. Hence, Goldberg maintains, we are forced to interpret Sam's utterance at tl literally, in which case we have a counterexample to Burge's thesis.

I do not intend to dwell on Goldberg's discussion of the non- literal interpretations he offers of Sam's original utterance. I agree that each is problematic. 1 Rather, I will argue that Sam's agreement at t3 does not establish that a literal interpretation of his original utterance renders that utterance false. Indeed, viewing Goldberg's example as a counterexample rests, I maintain, on a misconstrual of Burge's thesis.

To begin, let me offer an interpretation of Sam's situation. It would seem that the most natural interpretation of his situation is one according to which he met both Uncles in a somewhat rushed and confused way, and that he simply got their names mixed up. In such a situation, Sam would have a de re thought of Uncle Henry (the ugly man he saw standing at the fountain), to the effect that he is ugly and an additional belief to the effect that he is called "Uncle Harry". Consequently, Sam would agree with Nancy's correction in virtue of accepting that he got the names mixed up, and not in virtue of changing his mind about what he was thinking about the man in question. On this interpretation, however, Goldberg's example is not a counterexample to Burge's thesis. This is because Burge's thesis is not, and was never intended to be, a thesis that concerns the res in de re thoughts.'2 In a particularly apposite footnote Burge writes,

Mistakes about the res in de re judgements are not counterexamples to the claim that basic cogito-like judgements are self-verifying. Suppose I judge: I am thinking that my aunt is charming; and suppose that the person that I am judging




to be charming is not my aunt (I have some particular person in mind). It is true that I am making a mistake about the person thought about [to the effect that she is my aunt]; I have no particular authority about that.... But I am not making a mistake about what I am thinking about that person; there is no mistake about the intentional act and intentional content of the act. Authority concerns those aspects of the thought which have intentional properties. (Burge, 1988, p. 74, fn. 8)

For Burge, the object of a de re thought, the object the thought concerns, is determined by a non-conceptual causal relation between it and the thinker - not by the concepts the thinker employs. The individual thought about does not enter into the content of the thought, and hence the identity of the individual is not something about which the thinker can be authoritative.13

If Goldberg's example is to be relevant to Burge's thesis, then, we have to interpret Sam's judgement as de dicto. On this interpretation, however, Goldberg and Gallois are seen to be guilty of the same mistake - that of failing to take seriously the division within the set of self-attributions between the cogito-like and the non-cogito- like cases. This leads to a conflation which can then give rise to the appearance of a counterexample.

That Goldberg fails to pay heed to this crucial division is evident in the formulation he offers of Burge's thesis. He states,

the Burgean strategy depends on the truth of the following claim:

(*) The class C of present-tense self-ascriptions of thought-contents is such that (i) each of the members of C is self-verifying in virtue of its self-referential form and (ii) each of the members of C warrants the attribution of the corresponding self-knowledge of content to the person making the judgment. (Goldberg, 2000, p. 2)

where (i) here is Goldberg's statement of Burge's thesis. The class of present-tense self-ascriptions is simply not such that each of its members is self-verifying. Nor does Burge maintain that it is. The class of present-tense self-ascriptions will contain both cogito-like and non-cogito-like cases, and according to Burge it is only the former that will be self-verifying.

We turn now to the implications of this illegitimate extension of Burge's thesis to cover the non-cogito-like judgements.



118 SARAH SAWYER

4. A CONFLATION

If Sam's utterance is to be interpreted as expressing a de dicto thought (contrary to the natural interpretation provided above), then viewing it as a counterexample to Burge's thesis depends on a conflation; a conflation between the judgement that I am thinking that p and the judgement that I think that p. This conflation arises directly from the illegitimate extension of Burge's thesis mentioned above. The former is a cogito-like judgement. It is self-reflexive and is self-verifying in virtue of being so. The latter, on the other hand, is, within this context, to all intents and purposes synonymous with the judgement that I believe that p. This judgement is not self-referential. It is a judgement which refers to and attributes to the subject an independent standing belief. As such, it is not self- verifying. It will be true just in case the subject does indeed have an independent first-order belief that p. I am thinking that p does not entail I think that p.

(Note the analogous contrast between the judgement that I am not thinking that p, which is self-defeating, and the quite distinct judgement that I do not think that p, which may or may not be true, depending on the first-order thoughts of the subject in question.)

In the context of Goldberg's example, then, we need to distinguish between what Sam is currently thinking and what Sam thinks, what he believes. It is clear that Sam does not think, at tl or subsequently, that Uncle Harry is ugly, but this does not entail that he was not thinking at tl that Uncle Harry is ugly. Indeed, he was thinking that Uncle Harry is ugly at tl. (That is, he was employing the concept Uncle Harry and the concept is ugly conjoined in the appropriate way.)

Contrast the original conversation, then, with a conversation of the following form (and interpret all the utterances literally).

(t4) Sam: I think Uncle Harry is awfully ugly. (t5) Nancy: You must think Uncle Henry is awfully ugly. (t6) Sam: You're right, it's Uncle Henry whom I think is

awfully ugly.

The judgement expressed by Sam's utterance at t4 is not self- referential, is not self-verifying, and could well be false. However,




since it is not a cogito-like judgement, the fact that it may be false casts no doubt on Burge's thesis.

Looking back at the original conversation, it is no doubt because Sam intended to express at tl the judgement expressed at t4 that he is willing to accept Nancy's correction at t3. Similarly, it is only if we mistake the judgement expressed at tl for the judgement expressed at t4 that we are likely to see his agreement with Nancy at t3 as worrying, and see Goldberg's example as a counterexample to Burge's thesis. The judgement expressed at tl is self-verifying, but it is the judgement expressed at t4 which is the judgement that Sam sees as mistaken and wishes to withdraw.

That this is the mistake that Goldberg makes is indicated by his response to the first non-literal interpretation of Sam's utterance he considers. The interpretation, that Sam was thinking I am thinking Uncle Henry is awfully ugly, was seen to fail because it was thought to be plausible for Sam at the same time to think I do not think Uncle Henry is awfully ugly. Bearing in mind the distinction I am urging, these thoughts do not contradict one another.

Gallois is guilty of the same conflation. In the context of his example, we need to distinguish between the thought I am thinking my mother does not love me and the thought I think my mother does not love me. Once again, the former is self-referential and self- verifying in virtue of being so, whereas the latter is an attribution of a standing state with which it can at best be contingently related. The conflation of the two results in the appearance of a counterexample to Burge's thesis.

5. THE ATTITUDINAL COMPONENT OF SELF-KNOWLEDGE

Bernecker, unlike Gallois and Goldberg, is happy to accept Burge's thesis concerning the self-referential nature of certain self-ascriptions. Nevertheless, it is evident that he too has failed to grasp the distinction between instances of basic self-knowledge, on the one hand, and the more prevalent cases of non-cogito-like judgements on the other, and hence has failed to grasp the significance of the cogito-like cases. Thus, even while he explicitly acknowledges that Burge restricts his discussion to the former kind of judgement, when he comes to give examples the distinction is ignored. He writes,



120 SARAH SAWYER

Consider the first-order thought 'water is wet' and the self-referential thought 'I believe that water is wet'. The content of the that-clause (in the latter thought) is inherited from the first-order thought since the intentional content mentioned in the that-clause is not merely an object of reference or cognition; it is part of the higher-order cognition itself. (Bernecker, 1996, p. 264)

Again, later, he talks as if the belief that I believe that arthritis is painful were a cogito-like judgement, which it clearly is not.14 Maintaining that it is makes the conflation discussed in the previous section inevitable. Bernecker's examples bring to mind Gallois's final interpretation of Burge, according to which second- order thoughts are constructed upon first-order thoughts. In fact, since, according to Burge, the content of a first-order thought is included or contained in the content of the appropriate second- order cogito-like judgement, Bemecker refers to Burge's thesis as "the inclusion theory of self-knowledge."'15 This points to a further confusion. If inclusion, or construction upon, were to the point, there could be no distinction between the cogito-like and the non-cogito-like cases. Take any true self-ascription, including any non-cogito-like case, and the content of the first order thought will be included in the content of the second-order thought. For example, suppose the first-order thought is the belief that p, and the second-order thought is the judgement that I believe that p. Despite the fact that the second-order self-ascription is a non- cogito-like judgement, it contains the content of the first-order thought, namely the content that p. The crucial point here is that although in every case of a true self-attribution there is a partial overlap of content between the first- and second-order thought, there is not always an overlap of thought, or mental act. It is this latter kind of overlap which is of importance for Burge's thesis. To explain a little further, what distinguishes the cogito-like self- attributions from the non-cogito-like self-attributions is that in the latter case there are two distinct mental states - one standing, and the other occurrent. The cogito-like self-attributions, in contrast, are such that the first-order thought and the second-order judgement occur in the same mental act. The inclusion theory fails to acknowledge this crucial point, since it focuses on inclusion of content, rather than on the all-important issue of inclusion of mental act. 16




It is also evident that Bernecker views Burge's thesis as a theory of self-knowledge. He identifies Burge's thesis as "[t]he leading externalist theory of self-knowledge,"17 and claims that "Burgean compatibilism is not only the most promising but also the most widely accepted externalist theory of privileged self-knowledge."18 In this latter he agrees with Goldberg's claim that "the vast majority of philosophers who wish to show that semantic externalism is compatible with authoritative knowledge of [thought] content derive such a conclusion from ... Burge's thesis."19 Unfortunately, Gold- berg does not cite a single member of this so-called "vast majority", and those cited by Bernecker as endorsing it do not obviously fit the bill. For instance, Harold Noonan and Anthony Brueckner are cited, and yet the fomer is an ardent advocate of internalism, and the latter has in several places questioned the compatibility of externalism and privileged access to thought content. Moreover, the particular papers cited (viz. Brueckner (1992) and Noonan (1993)) do not directly concern the issue in hand. Brueckner's paper concerns transcen- dental arguments derived from externalist premises; and Noonan's concerns the explanatory necessity of object-dependent thoughts.

Nevertheless, the crucial mistake here is to suppose that Burge's thesis constitutes a theory of authoritative self-knowledge. As we have seen, Burge does not maintain that all self-ascriptions are self- verifying, and does not derive his compatibilism from his thesis that cogito-like judgements are. The cogito-like cases merely serve to establish that thoughts whose contents are externally individuated need not be subject to facsimile-based error. The warrant for all self- attributions, for Burge, is derived from the role such self-attributions play in critical reasoning. To his credit, Bernecker recognises that Burge does not himself apply his strategy to self-ascriptions of standing states, although given the confusion demonstrated by his examples one has to wonder what Bernecker takes this claim to amount to. Moreover, he goes on to maintain that a number of compatibilists do extend Burge's thesis to cover self-attributions of standing states. Here he cites as examples Kevin Falvey and Joseph Owens (1994), Sydney Shoemaker (1994), and Crispin Wright (1991). And once again the citations are, I believe, inappropriate. As I understand it, these and other compatibilists have taken from Burge only what we can call (following Christopher Peacocke20)



122 SARAH SAWYER

the redeployment thesis. The redeployment thesis states that the concepts used in second-order beliefs will be externally determined and will be fixed by the same external conditions that fix the concepts used at the first-order level. This is the first of the three reasons that Gallois identifies in Burge's writing to favour a compatibilist position. There is no mention of self-reference or self- verification here. The account is widely accepted amongst semantic externalists. Thus, in addition to Peacocke, Wright, Shoemaker, John Heil, and Donald Davidson,21 to mention but a few, appeal to the redeployment thesis. This, then, rather than Burge's thesis, provides the basis for an account of authoritative self-knowledge consistent with semantic externalism.

And even so, the redeployment thesis is typically invoked to show that semantic externalism has no additional problems in accommodating authoritative self-knowledge over and above those that internalism faces. It is invoked to avoid the facsimile-based error about content to which semantic externalism is supposed specifi- cally to give rise. It is not typically thought to constitute by itself an adequate account of authoritative self-knowledge. That is why Burge, Davidson, Gallois, Cynthia MacDonald, and Peacocke, for example, have all offered independent accounts of the warrant possessed by self-attributions of externally individuated thoughts.22 The merits of these will have to await discussion on a future occasion.

It is now time to turn to Bernecker's main charge against Burgean compatibilism. Bernecker claims, quite correctly, that authoritative self-knowledge requires the identification of both the content and the attitudinal component of the first-order thought in question. Thus if I am to know that I believe that p, not only must I know that it is p that I believe, but I must also know that I believe it rather than, for example, desire or fear it. He then points out that while Burgean compatibilism allows one to identify the content of one's thoughts, it says nothing of how one is to identify the attitudinal component of one's thoughts, and hence is an inadequate theory of authoritative self-knowledge. He boldly states that "most compatibilists, since unaware of the issue of attitude-identification, presume that the inclusion theory solves all compatibilist problems".23 In part this charge against Burgean compatibilism is deflected by considerations




laid out in the previous two paragraphs. Burgean compatibilism, if what is meant by that is the appeal to Burge's thesis, is not assumed by compatibilists to be a theory of authoritative self-knowledge. In addition, the fact that one can misidentify an attitude, mistake a state of supposition for a state of belief, or a state of jealousy for a state of envy, for instance, is generally conceded.

However, Bemecker goes on to offer an additional reason to believe that semantic externalism in particular will have difficulty in accommodating the authority with which one can know which kinds of thought one is having. He identifies three sources of possible efror about the attitudinal component of one's thoughts. The first two, those of self-deception and lack of attention, are left to one side, and it is clear that they need pose no specific difficulties for compatibilists. The third source of possible error arises, according to Bemecker, from the assumption that one could possess an "incom- plete understanding of the concepts used to describe one's mental condition."24 If this is indeed a possible source of error, it is one that perhaps arises only on the assumption of semantic extemalism. Accordingly this is the focus of Bemecker's paper.

Bemecker proceeds by arguing that attitudinal concepts such as those of belief and desire are themselves open to extemalist considerations. Thus he writes,

Suppose Bert has a brother, Oscar. While Bert is confused about 'arthritis', Oscar deviates from the language community with respect to 'to believe'. Oscar has an unusually coarse-grained concept of belief. It is in fact so general that he takes 'to believe', 'to suppose', 'to decide' and 'to consider' to be synonymous terms.... Now, despite Oscar's confusion, social externalism seems committed to saying that the meaning of his concept [belief] doesn't differ from what his language community means by it. (Bernecker, 1996, p. 269)

His argument against Burge's thesis, and hence against Burgean compatibilism, is then stated as follows.

One day Oscar entertains the self-referential thought that he expresses by 'I am believing that arthritis is painful'. Given the inclusion theory, Oscar knows what his first-order thought is about, i.e. he knows its propositional content. However, let's suppose that, contrary to Oscar's contention, he doesn't believe but supposes that arthritis is painful.... Hence, Oscar's judgement 'I am believing that arthritis is painful' is (partly) false. (Bemecker, 1996, p. 269, original emphasis)



124 SARAH SAWYER

Let us suppose that Bernecker has identified a further possible source of error concerning the identification of the attitudinal components of one's thoughts. Why should the possibility of this kind of error be a problem either for Burge's thesis, or for Burgean compatibilism? Consider Burge's thesis first. We can agree with Bernecker that the judgement that I am believing that arthritis is painful is not self-verifying. As Bernecker points out, the subject concerned may not be believing any such content. However, it is precisely for this reason that Burge is careful not to offer as examples of cogito-like judgements any which make reference to a specific type of mental attitude, such as that of belief or that of supposition. In order to understand Burge's thesis it should be understood that "thinks" satisfies the condition that any attitude of the form "S p's that p" (e.g. "S desires that p", "S fears that p") entails "S thinks that p". Thus the judgement that I am thinking that arthritis is painful is indeed self-verifying. Indeed, "is thinking" in the context of Burge's thesis is to be understood as "is entertaining the content". This is implicit when, in illustrating the nature of self-reference involved, Burge says, "[i]f a person ... were to ask 'Am I now thinking about water or twater?', the answer is obviously 'both'. Both concepts are used."25 It is only if "thinking" is construed broadly that Burge's claims make sense here. That is, "thinking" is taken to be equivalent to "deploying concepts in thought", where this is neutral on the issue of whether the concepts thus deployed are deployed in beliefs, desires, hopes, fears, and so on. Accordingly, Bernecker has not offered us a reason to doubt Burge's thesis, since he has not offered us an example of a cogito- like judgement. In this his mistake is seen to parallel those of Gallois and Goldberg. All three find fault with Burge's thesis because of an illegitimate extension of that thesis to non-cogito-like cases; or, equivalently, because of a conflation between the two types of case. The difference lies only in the nature of the illegitimate extension. Thus the fact that attitudinal concepts can be grasped only partially does not pose a threat, as Bernecker maintains, to Burge's thesis. Burge's thesis does not concern or imply anything in particular about attitudinal concepts about the extension of which one could be mistaken.




What about Burgean compatibilism? As Bernecker understands it, Burgean compatibilism is the position which extends Burge's thesis to cover all instances of self-attributions, thereby supposedly providing an adequate account of authoritative self-knowledge. However, since no-one has been identified as holding such a position, we can surely have no stake in its defence. Nevertheless, it might be thought that Bemecker has posed a general threat to the compatibilist position. He does, after all, open up the possibility that a subject could be in error about the type of thought she is entertaining; a possibility which, on the face of it, would thwart the very possibility of an account of authoritative self-knowledge consistent with extemalism. But this is not the case. Once one moves outside the realm of cogito-like judgements, one moves outside the realm of infallibility. That is, the extemalist can happily tolerate the possibility of certain kinds of error. Bemecker is himself quite happy with this thought. He does, however, think the kind of error which he has identified is unacceptable. He writes,

... [the subject] cannot even distinguish among different attitudes. When he thinks 'I am believing that arthritis is painful', for all he knows he could suppose, decide, or consider that P. If he actually believes that arthritis is painful, his self-referential judgement comes out true, but it doesn't thereby qualify as knowledge.... When ... [it] is true, it is accidentally so. (Bemecker, 1996, p. 271)

But it is hard to see how the kind of error to which Bemecker points differs in kind from the other kinds of misidentification of attitude to which one can be subject, kinds which are taken by Bemecker himself not to undermine the possibility of authoritative self-knowledge consistent with extemalism. The conflation between the cogito-like and the non-cogito-like cases which was evident in the arguments of Gallois and Goldberg may be playing a role here. In addition, as stated above, compatibilists cannot be classified as Burgean compatibilists along Bernecker's lines. Accordingly, accounts of authoritative self-knowledge consistent with extemalism have been offered. It is worth drawing attention in particular to Peacocke (1996) and (1999), which provide a detailed and plausible account of how the application of the concept of belief tracks the property of belief, even on the assumption of semantic externalism. A similar account could presumably be given for the other propositional attitudes. Suffice it to say that the appearance



126 SARAH SAWYER

of a threat to compatibilism once again rests on a mistaken interpretation of Burge's thesis and the uses to which it is intended to be put.

6. CONCLUSION

We have not been given a true counterexample to Burge's thesis. We have also seen that compatibilist strategies do not, as Bemecker and Goldberg maintain, derive their compatibilism from Burge's thesis. Failure to distinguish cogito-like judgements from other types of self-attribution, and failure to distinguish between Burge's thesis and the redeployment thesis, will clearly lead one astray.26

NOTES

1 See Gallois (1996), Chapter 9. 2 As Peacocke has shown, the model of needing to check on content-fixing relations before the contents of one's own thoughts can be known must in any case be mistaken. For even when the content of a thought is not externally individuated - as with the content of a logical or arithmetical belief - "a thinker equally does not have to check on the internal relations of the conceptual constituents of her beliefs before coming to know what it is that she believes" (1999), pp. 203-204. 3 Ibid., p. 70. 4 Ibid., pp. 73-74. 5 Ibid., p. 70. 6 Gallois (1996), pp. 174-175. 7 Ibid., p. 175, emphasis added. 8 Ibid. 9 Burge (1988), fn. 11, p. 79, emphasis added. 10 See his (1996, 1999). For insightful criticisms of this view see Peacocke (1996). 1 In fact, I think the focus on linguistic utterances rather than the judgements themselves is less than helpful, and contributes to the appearance of a counterexample. Burge never maintains that the statements which express a thinker's judgements regarding her own occurrent thoughts are self-verifying. 12 Burge's thesis will of course apply to the intentional contents of de re thoughts just as much as to those of de dicto thoughts. It is the res that the thought concerns and the relations the subject bears to the res in question, if there is one, over which Burge's thesis does not extend. And, once again, this will only be in the case of cogito-like judgements. 13 See Burge (1977).




14 Bernecker (1996), p. 269. 15 Bemecker (1996), p. 265. 16 With thanks to an anonymous referee for prompting me to clarify my objection to the inclusion theory. 17 Bernecker (1996), p. 262. 8 Ibid.

19 Goldberg (2000), p. 1. 20 See his (1996, 1999), Chapter 5. 21 See Wright (1991), p. 76; Shoemaker (1994), p. 260; Heil (1988), p. 251; and Davidson (1987). 22 See Burge (1996, 1998, 1999); Davidson (1984); Gallois (1996); MacDonald (1995) and (1998); and Peacocke (1996, 1999). For a recent criticism of the accounts offered by Burge and Gallois, see Brueckner (1999). 23 Bernecker (1996), p. 267. 24 Ibid., p. 268. 25 Burge (1988), p. 75. 26 This investigation was supported by the University of Kansas General Research Fund allocation #2301795. I would also like to thank Jack Bricke and Brad Majors, and audiences at the University of Missouri at Columbia and the University of Kansas.

REFERENCES

Bernecker, S. (1996): 'Externalism and the Attitudinal Component of Self- Knowledge', Nous 30, 262-275.

Brueckner, A. (1992): 'What an Anti-Individualist Knows A Priori', Analysis 52, 111-118.

Brueckner, A. (1999): 'Two Recent Approaches to Self-Knowledge', Philosoph- ical Perspectives 13, 251-271.

Burge, T. (1977): 'Belief De Re', The Journal of Philosophy 74, 338-362. Burge, T. (1988): 'Individualism and Self-Knowledge', Journal of Philosophy

85, 649-663. Reprinted in Cassam, Q. (ed.) (1994): Self-Knowledge, Oxford: Oxford University Press. All page references to the reprint.

Burge, T. (1996): 'Our Entitlement to Self-Knowledge', Proceedings of the Aristotelian Society XCVI, 91-116.

Burge, T. (1998): 'Reason and the First Person', in C. Wright, B. Smith and C. MacDonald (eds.), Knowing Our Own Minds (pp. 243-270), Oxford: Oxford University Press.

Burge, T. (1999): 'A Century of Deflation and a Moment about Self-Knowledge', Proceedings and Addresses of the American Philosophical Association 73, 25- 46.

Davidson, D. (1984): 'First Person Authority', Dialectica 38, 101-111. Davidson, D. (1987): 'Knowing One's Own Mind', Proceedings of the American

Philosophical Association 60, 441-458.



128 SARAH SAWYER

Falvey, K. and Owens, J. (1994): 'Externalism, Self-Knowledge, and Skepticism', Philosophical Review 103, 107-137.

Gallois, A. (1996): The World Without, The Mind Within: An Essay on First- Person Authority, Cambridge: Cambridge University Press.

Goldberg, S. (2000): 'Is "I Am Presently Thinking That p" Self-Verifying?' Presented at the Central States Philosophical Association Meeting in Lincoln, Nebraska.

Heil, J. (1988): 'Privileged Access', Mind 97, 238-25 1. MacDonald, C. (1998): 'Externalism and Authoritative Self-Knowledge', in C.

Wright, B. Smith and C. MacDonald (eds.), Knowing Our Own Minds (pp. 123- 154). Oxford: Oxford University Press.

Noonan, H.W. (1993): 'Object-Dependent Thoughts: A Case of Superficial Necessity but Deep Contingency?', in J. Heil and A. Mele (eds.), Mental Causation (pp. 283-308), Oxford: Clarendon Press.

Peacocke, C. (1996): 'Entitlement, Self-Knowledge, and Conceptual Redeploy- ment', Proceedings of the Aristotelian Society XCVI, 117-158.

Peacocke, C. (1999): Being Known, Oxford: Oxford University Press. Shoemaker, S. (1994): 'Self-Knowledge and "Inner Sense"', Philosophy and

Phenomenological Research 54, 249-314. Wright, C. (1991): 'On Putnam's Proof that we are not Brains-in-a-Vat', Proceed-

ings of the Aristotelian Society 91, 67-94.

University of Kansas Lawrence, KS USA E-mail: [email protected]



in defence of burge's thesis

Documents