This paper argues that first person thought and talk presents a significant problem for the normatively pragmatist conception of semantic and intentional content advocated by Robert Brandom. Section 1 identifies the central theses which define this conception of content. Section 2 then outlines Brandom’s account of the use of the first person singular pronoun, at the same time indicating how Brandom’s normatively pragmatist conception of content has shaped the way he deals with the first person, in particular, how he responds to the issues raised by Perry and others concerning the distinctive contentfulness of first person assertion and belief. Section 3 first argues that Brandom’s account of the use of first person singular expressions is wrong under a literal interpretation of it, then draws upon J. L. Austin in order to reconstruct Brandom’s account so as to reveal the truth it contains. Section 4 then argues that when thus reconstructed this account presupposes a notion of saying something, hence of contentful or meaningful linguistic performance, incompatible with Brandom’s conception of content. Finally, in Section 5, it is argued that Brandom fails to address the issues raised by Perry. He gives no unequivocal account of just what the content of first person assertion and belief is, sometimes creating the impression that the distinctive accomplishment of the first person singular pronoun is located solely at the level of linguistic pragmatics rather than semantics. This equivocalness is traced back to his inferentialism and his account of norms: these work against acknowledging that first person assertion and belief are contentful in a manner different for other kinds of assertion and belief.
Introduction: Two Strands of Argument regarding the First Person
What exactly is asserted or believed when a speaker S asserts or thinks, “I am Φ” (for whatever value of Φ)?1 On the face of it, the answer is easy: what S asserts or believes is that he is Φ. But what exactly is the semantic or intentional content expressed by the anaphorically dependent2 sentence “he is Φ”? This cannot be a proposition, at least not if by proposition one understands something which is (i) expressible by a full sentence, that is, a sentence which, unlike the sentence “he is Φ”, requires no embedding in a larger sentential context for its intelligibility; and (ii) graspable by all subjects, whoever they may be, in particular, by subjects other than S (so that it can, in principle at least, form the content of any subject’s assertion or belief). This can be seen as follows: let t be any arbitrary name or definite description. It is clearly possible that S should believe the proposition expressed by the sentence “t is Φ” without believing that he is Φ; this will be the case whenever S does not believe that he is t. Now the name or description t was arbitrarily chosen. So for any t S’s assertion or belief cannot be an assertion or belief whose content is expressed by the full and unlimitedly accessible sentence “t is Φ.” Consequently, S’s assertion or belief cannot be any assertion or belief that p (in the sense defined by conditions (i) and (ii)).
This argument derives, of course, from John Perry.3 But in his original discussion Perry provides another argument, or strand of argument, for the conclusion that S’s assertion or belief is not an assertion or belief that p (in the sense defined by conditions (i) and (ii)). This further strand of argument concerns the functional or causal role of assertions or beliefs in the first person. Perry points out that the behavioural impact of such an assertion or belief can differ significantly from that of any strictly propositional belief, even when the latter is objectively co-extensive with the former. Thus, the impact of S’s believing the sentence “I have won the lottery” will be differ very greatly from the impact had by his believing the sentence “The person who bought the 1,000th ticket has won the lottery,” at least if in the latter case S does not appreciate that he is the person who bought the 1,000th ticket. Perry takes this difference in impact on behaviour to be explained by (hence as further corroborating) the conclusion reached in his first strand of argument that assertions or beliefs involving the first person have a distinctive contentfulness which cannot be assimilated to the contentfulness of assertions or beliefs whose content is propositional in the sense defined by the two conditions (i) and (ii) above.
Now at the very least these kinds of consideration show that the first person presents special challenges to any attempt to explicate the contentfulness of assertion and belief. In sub-sections 5.2 and 5.3 of chapter eight in his book Making It Explicit (Cambridge, Mass.: Harvard University Press, 1994), Robert Brandom attempts to meet these challenges, that is, to accommodate and explicate the phenomena upon which Perry bases his arguments, within the framework of a normatively pragmatist conception of content. How successful is he? Or does first person thought and talk constitute a great, perhaps indeed insuperable obstacle for the normative pragmatist?
§ 1: Brandom’s Normatively Pragmatist Conception of Content
Before we can understand Brandom’s account of the first person, we must first understand his general project. In Making It Explicit Brandom seeks to give an account of what it is for assertions, beliefs and indeed all other forms of thought and talk to be contentful. Whenever one asserts or believes, there is, of course, always something that one asserts or believes – the semantic content of one’s assertion, or the intentional content of one’s belief. But what is it for an assertion or a belief to have content, the kind of thing one typically expresses by using nominal clauses of the form ‘that p’? According to Brandom, such contentfulness can be explicated in terms of the normative commitments and entitlements which a speaker undertakes by uttering linguistic expressions according to linguistic norms, therein performing a particular act of assertion. What this claim comes to can best be seen by examining, at least in rough outline, the four component theses which yield it.
The first thesis underpinning Brandom’s approach to the question of content locates him in a tradition which includes William Alston and, notwithstanding significant differences, John Searle. Like Alston and Searle, Brandom thinks that speech acting is essentially behaviour subject to, and constituted by, norms (‘rules’) in the following sense:
Thesis 1: To make an assertion, or to acquire a belief, is to undertake a certain set of commitments4—commitments (a) to being able to provide both a reason for one’s utterance or belief; and (b) to being able to provide one’s utterance or belief as a reason for further utterance or belief.5
To assert or to believe is to take position within Sellars’s ‘space of reasons’, a space whose ‘laws’ are norms specifying what it is right or wrong, correct or incorrect, to say or think, given what one has said or thought.6
The second thesis underpinning Brandom’s programme derives specifically from Sellars and constitutes his inferentialism. It may be put as follows:
Thesis 2: The content of assertion or belief, that is, what is asserted by, or believed in it, is fixed by the set of inferential relations which this content has to other such contents.
Like Sellars, Brandom believes that to say that an assertion or belief has as its content, say, the proposition that p, simply is to say that from this assertion or belief one can infer only things which are true if p. This set of inferential consequences just is the content of the assertion or belief.
Brandom’s third thesis constitutes a distinctive elaboration of the inferentialism he inherits from Sellars. For Brandom goes on to maintain the following:
Thesis 3: The inferential relations between contents can themselves be defined in terms of relations of incompatibility between the commitments undertaken in assertion or belief.
For example, to say that the proposition that p entails the proposition that q7 is, roughly speaking, to say that asserting or believing that not q is incompatible with asserting or believing that p – where asserting or believing that not q is a matter of undertaking commitment to all claims incompatible with the claim that q. 8
Clearly, the third thesis maps the dimension of content onto the dimension of linguistic (or ‘mental’) acting, understood in the spirit of the first thesis. This third thesis thus reflects Brandom’s conviction that the contentfulness of thought and talk can and must be understood in terms of the pragmatics of speech acting and thinking. Yet it cannot stand alone since one might object that the notion of assertion or belief qua the undertaking or possession of commitment itself presupposes the character of assertion or belief as contentful, in which case the whole project would be viciously circular. We need an account of what it is to undertake commitment, or to be committed, which shows that this does not in fact presuppose the contentfulness either of the linguistic behaviour through which commitment is acquired, or the psychological condition in which being committed consists. The fourth and final thesis articulates the general form of this account:
Thesis 4: Relations of compatibility and incompability between assertions and beliefs qua undertakings of commitments as outlined in thesis 1. are constituted by norms specifying conditions under which linguistic expressions, and in particular, full sentences, are correctly uttered.
More specifically and precisely, what determines relations of compatibility and incompability between assertions and beliefs are norms specifying (a) what circumstances in the world make it correct to utter the sentences used to make or express these assertions and beliefs; and (b) what further utterances of linguistic expressions one is or is not entitled to, given utterance of the sentences used to make or express these assertions and beliefs. By thus fixing relations of commitment compatibility and incompatibility between assertions and beliefs, the rules or norms of language use (whose formulation appeals merely to normative notions of right or wrong rather than notions of meaning, reference or representation), fix the contents of individual assertions and beliefs, and thereby the very identity of these assertions and beliefs themselves. And with this Brandom can give his normatively pragmatist account of the meaningfulness of linguistic expressions in general and the contentfulness of assertion and belief in particular: the contentfulness of assertion or belief consists in the commitments and entitlements a speaker undertakes by uttering the relevant linguistic expressions in accordance with the norms governing their use.
Now Brandom distinguishes undertaking commitments explicitly from undertaking commitments simpliciter because the content of a given assertion or belief will have all sorts of inferential relations to other contents of which the speaker may well be unaware. Clearly, speakers do and must associate comparatively determinate, distinguishable meanings with the linguistic expressions they utter, and for Brandom this is to say that they only associate a certain limited range of commitments and entitlements with these expressions. Only these they undertake explicitly, or, as Brandom puts it, only these they acknowledge;9 all the other commitments are undertaken merely implicitly, more accurately, are just undertaken. It follows, then, from this normatively pragmatist conception of assertion and its content that all linguistic items just are norm-governed and -constituted means either for explicitly undertaking commitments themselves (in the case of full sentences), or for constructing such means (in the case of sub-sentential expressions). In particular, full sentences just are means of acknowledging commitments; this is what their meaningfulness, their capacity to express, as uttered on a specific occasion, a more or less determinate semantic content. Consequently, to utter a sentence correctly, that is to say, in accordance with its linguistic meaning, must be to perform a speech act, such as an assertion, in which one undertakes a certain set of commitments, understood as an array of possibilities for another’s evaluating one’s performance as entitled or even required, given the previous moves one has made within the ‘space of reasons’.
§ 2: Brandom on the First Person and the First Person Pronoun
How, then, will this general programme deal with first person assertion and belief? Brandom clearly appreciates the two strands to Perry’s argument. He presents the first strand thus:
Following Castañeda, Perry has argued persuasively that there are things that can be said using expressions such as ‘I’, ‘now’, and ‘here’ that cannot be said by using nonindexical vocabulary. I can of course refer to or describe myself in many ways, but no other term can do the … job done by ‘I’. For any other term ┌ t ┐ I might use to pick myself out, there would always be some possible circumstances in which I could believe that t had a property – was about to be eaten by a bear, was standing in the Stanford library, was leaving a trail of sugar on the floor of the supermarket, and so on – without thereby believing that I have that property. (552)
He then notes the second strand, which, as we have seen, concerns how the difference in content argued for in the first strand
… can be manifest in my actions, since a belief I would express using ‘I’ can be relevant to my practical reasoning and action in a way that no belief that can be otherwise expressed can match across the whole range of counterfactual situations. There will always be some of those situations in which I fail to realize that I am t. In those cases, I will fail to form intentions and to be motivated to act on the basis of beliefs about what t should do in the same way I would on the basis of beliefs about what I should do. (552)
Yet having presented Perry’s strands of argument Brandom immediately shifts attention from the issue which concerned Perry, namely, what in the light of these two strands of argument one is to say about the content of first person assertion and belief, to the issue of what use one must attribute to a certain linguistic expression, specifically, the first person singular, in order to account for the phenomena noted by Perry. According to Brandom,
… the key feature of the use of ‘I’ that is not reproduced by other coreferential expressions (even those that are necessarily coreferential, since one could always fail to realize that they were) is its use in expressing the acknowledgment of a commitment. What ‘I’ expresses is a potentially motivating acknowledgment of a commitment. The only expression that cannot counterfactually be separated from this motivational role is ‘I’. (552)
In short, in order to understand and explain the phenomena noted by Perry, we must appreciate “the expressive job done by ‘I’” (552), its distinctive use as a means of expressing acknowledgement of commitment.
Now understanding what Brandom is getting at here is not easy. Nor is this due simply to the brevity of Brandom’s account; we shall shortly see that, when taken literally, Brandom’s official account of the distinctive use of the first person singular is incapable of explaining what it purports to explain, namely, the distinctive motivational force of first person assertion and belief. Extracting a plausible position from this account will in fact require us to engage in significant reconstruction. Before, however, we embark on such reconstruction, let us reflect a little on the significance of the way Brandom shifts the focus of attention away from the issue which concerned Perry, namely, what distinctive content first person assertion and belief must have, to the issue of what distinctive use linguistic expressions used to accomplish or express such assertion or belief must have.
Perry regarded the fact that assertions or beliefs in the first person have a functional role and causal properties not possessed by any assertion or belief that p as something explained by (and only for this reason corroborating) the central thesis that such assertions or beliefs have a content which is not propositional in the sense defined by conditions (i) and (ii) above. Note, however that in what Brandom says about the use of the first person pronoun ‘I’, all reference to this central thesis has been dropped in favour of an alternative explanation of why first person assertions or beliefs have a distinctive functional role and causal properties not possessed by any assertion or belief that p. Whereas Perry explained this distinctive role in terms of the distinctive content of first person assertions or beliefs, Brandom wants to explain it in terms of the distinctive use of first person expressions.
Brandom does not make clear why he shifts away from the issue of content to an attempt to explain the distinctive functional role and causal proclivities of first person assertions and beliefs in terms of language use. Nonetheless, it is not hard to find a reason which is plausible from his perspective. Arguably, Brandom wants to oppose the view that a difference in content explains the difference in functional or, as he puts it, motivational role of first person assertion and belief. Rather, he wants to say that this difference derives from the way first person linguistic expressions are used to accomplish first person assertion, or to express first person belief. This is not to imply that Brandom wants to forget about the issue of content altogether, for he does eventually return to it, namely, in sub-section 5.3 of Ch.8. He does, however, want to argue that the norm-governed use of first person singular linguistic expressions is the basic phenomenon – basic indeed in the sense that it not only explains the distinctive motivational role of first person assertive and belief, but also ultimately explains its distinctive contentfulness (or perhaps rather, as we shall see, why first person assertion and belief can be contentful in a completely standard way and yet still have its distinctive motivational role). In short, the order of explanation to be found de facto in Perry is to be reversed.10
Certainly, if this is right, then the structure and sequence of Brandom’s discussion becomes clearer: in the first two or so pages of sub-section 5.2 in Chapter 8 (552-554), Brandom outlines in quite general terms what he regards as the normatively pragmatic role played by the first person singular. This is, as we have already seen, its role or function as “… expressing the acknowledgment of a commitment” (552) – a claim which, as we have yet to see, requires significant reconstruction before it yields plausible sense. Then, having outlined in general terms “the expressive job done by ‘I’” (552), Brandom proceeds in the remaining pages of this sub-section (554-559) to tell a hypothetically genetic story about how a term capable of playing this role could emerge out of the use of what are essentially proper names (or at least non-first person referring expressions) to make non-first person assertions. Then, in the next sub-section (sub-section 5.3), Brandom concludes his discussion of the first person by returning to the question of first person singular content raised by Perry. Here the point is apparently to show how this question is answered correctly, namely, after a consideration of the use of first person singular linguistic expressions, in particular, on the light of the general account of such use given in the first part of sub-section 5.2. For in the light of this account, or so Brandom appears to imply, the issue of content turns out to contain no particular mysteries at all.
The hypothetically genetic story Brandom tells of how the use of a genuinely first person singular pronoun, and thus the first person perspective itself (self-consciousness), could evolve out of the use of mere proper names and a third person perspective (consciousness)11 consists in appropriating and elaborating G.E.M. Anscombe’s thought-experiment of a so-called ‘A’-language: Brandom adopts the scenario imagined by Anscombe of a linguistic practice in which the letter ‘A’, although serving as “… the word each one uses in speaking of himself …,”12 nonetheless cannot count as a true first person singular pronoun because its use lacks the non-inferentiality and immunity to misidentification displayed by use of a genuine first person singular. Brandom then augments (555-558) Anscombe’s scenario by taking it as the point of departure for a hypothetical process whereby the use of the letter ‘A’ acquires these features. As a final move, Brandom claims (558-559) that if the use of Anscombe’s letter ‘A’ could evolve to the point where it had acquired these features, then it would also have eo ipso evolved to possess the features highlighted by Perry. In other words, it would have acquired precisely what Brandom regards as the ‘key’, that is to say, defining feature of the use of any first person singular pronoun.
Brandom’s story about how Anscombe’s letter ‘A’ could acquire what he regards as the significance and use of a first person singular pronoun is hard to assess because it is sparse, its individual evolutionary steps undermotivated. Fortunately, we do not have to subject it to critical scrutiny: the story presupposes Brandom’s general, non-genetic, strictly analytic account of the key feature and accomplishment of the first person singular and this itself encounters significant difficulties. It therefore seems quite legitimate to leave Brandom’s appropriation of Anscombe to one side in order to concentrate in the first instance on his general account of the use of the first person singular pronoun. In § 3 I first show that this account is not correct under any literal interpretation of it. I then radically reconstruct the account in order to extract from it the germ of truth it does contain. The next section, § 4, argues, however, that if this is the truth contained within Brandom’s official account, then one must admit a notion of saying something, hence of contentful or meaningful linguistic performance, which is precluded by Brandom’s conception of semantic and intentional content. Clearly, if in order to accommodate what is right in Brandom’s talk of “the expressive job done by ‘I’” (552), one must admit a notion of saying incompatible with his overall normatively pragmatist programme, then one would have a ready explanation of why Brandom’s account is very difficult to understand: the first person presents a fundamental, indeed an insuperable obstacle to a conception of content which, notwithstanding its commendably non-reductive attitude towards the normative, betrays a residually behaviourist hostility towards the ‘inner’ perspective of the first person (and thus indeed to any perspective, ‘inner’ or ‘outer’, first or third person, at all).
Of course, if one can only do justice to what Brandom is getting at when he speaks of “the expressive job done by ‘I’” (552) by admitting something fundamentally at odds with his normatively conception of content, then there is no longer any point in arguing that if one adopts this conception of content, one can easily answer all questions concerning the nature of first person content; the whole issue is thrown wide open again. Indeed, if one can only do justice to the first person at the cost of Brandom’s conception of content, then one must suspect that what Brandom says in his subsequent discussion on the content of first person assertion and belief will also encounter difficulty. In the final section, § 5, I show that this is indeed the case: in sub-section 5.3 of Ch. 8 in Making It Explicit we find that Brandom in fact fails to address the question of first person content. For close examination of what he says shows that he misconstrues what addressing this question requires, i.e., fails to accommodate the lesson implicit in Perry, Castañeda, Anscombe and indeed others well before them, e.g., Natorp, Herbart and possibly also Fichte.
These difficulties intimate something of fundamental importance: the conclusion implicit in these actually very traditional and long-standing observations about first person assertion and belief, namely, that the contentfulness of first person linguistic and psychological intentionality is not “subject-object,” that is to say, not propositional in the sense defined by conditions (i) and (ii) above, is incompatible with Brandom’s general approach. Section 5 thus concludes by showing that this is indeed the case and, more importantly, by identifying its sources: on the one hand, Brandom’s inferentialism commits him to maintaining that all semantic and intentional content satisfies condition (i); on the other, his account of what it is for behaviour to be subject to a norm in terms of the habits of normative assessment displayed by others toward such behaviour commits him to maintaining that all content satisfies condition (ii). Brandom’s programme is thus de facto committed to maintaining that at the semantic or intentional level, that is, in the way they are contentful, first person assertion and belief do not really differ from their third person (or at least non-first person) counterparts; what is distinctive about the first person, hence about the linguistic contribution of the first person singular, occurs solely at the pragmatic level of acknowledging commitment.
§ 3: The ‘Expressive’ Job done by ‘I’ – A Reconstruction
At first sight, Brandom’s claim that ‘I’ is used to express acknowledgement of commitment is puzzling. For example, at one point he says of the first person singular pronoun ‘I’ that “(i)ts primary use is as acknowledging a commitment.” (553) But according to Brandom the very meaning of linguistic items can be analysed as their use in explicitly undertaking, i.e., acknowledging, commitments. So all linguistic items are, in one way or another, devices for acknowledging such commitments; all “(a)ssertional performances or avowals are performances that express the deontic attitude of acknowledging doxastic commitments.” (196) What Brandom has just said about the first person singular thus seems to be no explanation at all of just what it specifically is used to do.
At this point, we might seize upon what Brandom says in the very last sentence of the paragraph in which he outlines his account of the use of ‘I’. Having said that the first person singular pronoun ‘I’ is the only expression that cannot counterfactually be separated from its motivational role, Brandom immediately goes on to say that “(a)cknowledgment of a practical commitment is the deontic attitude that corresponds to forming an intention – what is expressed by a sentence of the form “I shall …” (552) This might suggest that for Brandom the ‘key’ use of ‘I’, the use which defines any expression as a first person singular pronoun, is its use in acknowledging a particular kind of commitment. As we already know, to assert or believe something is, according to Brandom, to undertake an array of commitments, in the first instance, to being able to justify one’s utterance or belief and to using it as a reason for other utterances or beliefs. In addition, one also undertakes commitments to believing and entitlements to asserting things entailed by one’s utterance or belief.
Now Brandom calls the kinds of commitment just mentioned doxastic commitments. He claims, however, that there is another, quite distinct kind of commitment. These are practical commitments to performing some (typically non-linguistic) action A. Furthermore, there is, he insists, a class of sentence whose correct use constitutes the undertaking, indeed acknowledging of such a practical commitment. These are sentences of the form “I shall do A.” And so, given that Brandom himself concludes his official account of the use of ‘I’ with a sentence which mentions acknowledgement of practical commitment and the distinctive kind of sentence by which such acknowledgement is ostensibly accomplished, it seems reasonable to suppose that what Brandom sees as distinguishing the first person is its use as a device for acknowledging a practical commitment. All linguistic expressions, including the first person singular pronoun ‘I’, are means of acknowledging doxastic commitments.13 The first person singular pronoun ‘I’, however, is used to do something more, namely, to express, potentially to another, one’s acknowledgement of a practical commitment to performing some further (non-linguistic) action. Its ‘key’ use, that use which defines it as a first person singular pronoun, lies in its teaming up with the modal auxiliary ‘shall’ in a sentence of the form “I shall do A” in order to express one’s acknowledging that, all things considered, doing A is the rational or correct thing to do.
It is clear that in at least several places Brandom wants us to understand things this way – although we should note that he never explicitly enjoins us to do so. How else are we to make sense of the way he ends his official account? How else are we to understand his subsequent claim that “‘I’ finds its home language-game in acknowledgments of commitments to act, and secondarily in the expression of beliefs and desires that are directly relevant, as premises, to bits of practical reasoning that have formations of intention as conclusions”? (553) Or the conclusion that he draws from this, namely, that “(t)he central defining uses of ‘I’ are not its uses in such sentences as “I can run the mile in five minutes” but its uses in “I shall open the door,” as expressing the conclusion of practical deliberation, and therefore as used in the expression of the premises”? (553) Yet if this is how Brandom is understood, then it is not clear what his account comes to. Nor indeed can the account explain the distinctive motivational force of first person assertion and belief.
What, after all, does it mean to say that ‘I’ finds its home-language game in utterances of sentences of the form “I shall do A”? What non-question-begging reason is there for thinking that the use of ‘I’ in such sentences is prior to its more reporting uses in “I can run the mile in five minutes” or “My dog has dingo in him”? Presumably, the idea is that the use of ‘I’ in sentences of the form “I shall do A” is that use which really explains or identifies the character of an expression as a first person singular pronoun. Now it certainly seems an accurate reflection both of linguistic meaning and linguistic phenomenology to say that sentences of the form “I shall do” do not (just) report on some independent process of forming an intention to do A, but rather constitute the very vehicle in and through which intentions are formed.14 In this sense, then, the correct use of sentences of the form “I shall do A” is indeed, to use Brandom’s terminology, to acknowledge practical commitment. But this only reveals the home language game of the whole sentence; it surely says nothing specifically or solely about any home language game of the first person singular pronoun ‘I’, in abstraction from its occurrence together with the modal auxiliary ‘shall’. The mere fact that one can use sentences of this form to “acknowledge practical commitment” merely sets this kind of sentence apart from other sentences, amongst them, other first person sentences of the genuinely ‘reporting’ kind. It tells us nothing about the distinctive accomplishment of the first person pronoun contained in such sentences.
Furthermore, if it is understood in this way, then Brandom’s account cannot do what it claims to do, namely, explain Perry’s point about the peculiar difference first person assertion and belief make to action. One does not address this issue by explaining how first person assertion or belief not of the form “I shall do A”, but rather of the indisputably cognitive, assertoric form exemplified by “I am being threatened by a bear,” is able to generate first person assertion, belief or intention of the kind expressed by sentences of the form “I shall do A.” Nor indeed does one address this issue even by explaining why first person assertion or belief of the kind realised or expressed by sentences of this specific form is able to generate action. For what is sought is an account of that structural feature in virtue of which first person assertion or belief of any form whatsoever is able to generate action in a way in which corresponding (co-referential) non-first person assertion or belief is not. Perhaps indeed first person assertion or belief not of the form “I shall do A”, but rather of the cognitive form exemplified by, e.g., “I am being threatened by a bear,” only takes effect by generating (via the appropriate practical inference) first person assertion, belief and intention of the kind expressed by sentences of the form “I shall do A,” e.g., “I shall run away from the bear.” This is, however, simply to explain how, for this kind of first person assertion or belief, the distinctive motivational force or role possessed by first person assertion or belief takes effect; it in no way explains what this distinctive motivational force or role is, but rather simply presupposes that first person assertion and belief have it. And so the claim that the home language game of ‘I’ is its use in sentences which express resolve to do something does not explain Perry’s point.
Indeed, the claim would seem to presuppose Perry’s point. For the fact that the strictly cognitive, assertoric or reporting uses of the first person only take effect via those uses of ‘I’ (together with ‘shall’) which express resolve to act does not show a one-sided dependency of the former upon the latter so much as a mutual, albeit complementary dependency of each kind of first person assertion or belief upon the other. After all, asserting or believing, “I shall run away” requires for its very existence asserting or believing “I am being threatened by a bear” (or some similar first person assertion or belief which rationalises resolving to run away) just as much as the latter requires the former at least in order to take effect. So ‘I’ is just as much, just as fundamentally at home, in its cognitive or assertoric uses as it is in its ‘volitional’ use in sentences of the form “I shall do A.”
Consequently, it is no more legitimate to appeal to the use of ‘I’ in sentences specifically of the form “I shall do A” in order to explain anything about first person assertion or belief (up to and including its capacity to affect behaviour in a way in which co-referential third person assertion or belief cannot) than it is to appeal to the use of ‘I’ in any other kind of sentence. Let us grant that the use of ‘I’ explains both the distinctive contentfulness and the distinctive motivational role of first person assertion or belief, perhaps because it enables assertion and belief with a distinctively first person contentfulness, which distinctive contentfulness in turn manifests itself, just as Perry points out, in a distinctive behavioural impact. Just this, however, shows that the use of ‘I’ at issue can only be its use in all sentences to make or form all kinds of first person assertion or belief. So one does not indeed address the issue of the first person by pointing out, no doubt correctly, that sentences of the form “I shall do A” are not primarily means for reporting, but rather means for making, one’s decisions.15
Is there then no element of truth in what Brandom says, something which he is expressing poorly because to bring it out more clearly would expose a tension in his overall project? That there is can be seen once one takes the notion of expression just as Brandom, at least in other contexts, wants it to be taken. This notion plays a central and fairly technical role in Brandom’s whole account of language, reason and content. When I look out the window and see that it is raining, I might well infer such things as that the streets outside will be wet. But I do not get from my perceptual observation of rain to the judgement that the streets will be wet by modus ponens. That is, I do not move from the observation to this judgement via belief in the conditional that if it is raining, then the streets will be wet. No, I move directly and immediately, without intervening inferential steps, to my judgement about the streets. For in the course of everyday experience of the world I have acquired the habit of making such moves. Retrospectively, of course, logicians can treat me as if I were in fact performing the operation of modus ponens in my head. That is, they can retrospectively formulate, in other words, make propositionally explicit, my normatively assessable habit of inference by casting it as the conditional “If it is raining, then the streets will be wet.” But I do not really perform in my head the formal inference which logicians can construct on the basis of how I actually reason.16
Bearing this in mind, let us return to Brandom’s initial claim that the “expressive job” performed by ‘I’, its ‘key’ feature, “… is its use in expressing the acknowledgement of a commitment.” (552) Note that in this initial claim there is no reference at all to distinctive practical commitments, nor indeed to uses of ‘I’ in the sentences of the form “I shall do A.” Rather, there is reference to expressing acknowledgement of commitment. And expressing has for Brandom the significance of making propositionally explicit, of saying something that is usually and initially simply done (as know-how). This insinuates that the distinctive use of the first person singular pronoun consists not in its use as a means of acknowledging a distinctive kind of commitment, but rather its use as a means of acknowledging commitment in a distinctively, uniquely expressive or ‘explicitating’, discursive way. All linguistic items are means of acknowledging commitments, in particular, doxastic commitments. But what is distinctive about the first person singular is that it is a means of acknowledging commitment by expressing the fact that one is acknowledging commitment. Clearly, if this is to do any real work in distinguishing use of the first person from the use of other linguistic items, the notion of expression appealed to here must connote something quite specific and comparatively precise. In particular, it must connote expression in linguistic form of the fact that one is acknowledging commitment. The first person must be a means of acknowledging a commitment by expressing linguistically or discursively, in other words, saying, that one is.
To old hands, this will immediately ring a bell. In the first part of his William James lectures, held at Harvard in 1955 and otherwise known as How to do Things with Words, J. L. Austin pointed out that some sentences were used, not to say things, but to do them. For example, when I, as chairman of a meeting, utter the sentence, “I hereby declare this meeting to be closed,” I am not saying, i.e., asserting or reporting, that I am declaring the meeting to be closed; I am actually closing it. Following Austin, one can call sentences usable in this way performative sentences, distinguishing them from sentences typically used to assert or report things, which we can call, once again following Austin, constative sentences.
Now in the second part of his lectures, Austin goes on to point out that the class of performative sentences is wider than those which accomplish the performance of strictly conventional acts such as closing meetings and instituting marriages.17 In particular, there are performative sentences for all speech acts. Thus, the policeman can order the demonstrators to leave the building not simply by uttering the imperative sentence “Leave the building!” but by uttering the indicative sentence “I hereby order you to leave the building!” And in doing so the policeman is not asserting the sentence, “I hereby order you to leave the building!” Of course, there is a sense in which the policeman is still saying something with this sentence, that is, expressing its semantic content as this is determined both by the meaning of the sentence and the context in which it is uttered.18 Crucially, the same applies to assertion: instead of asserting that it is raining by uttering the sentence, “It is raining,” I could assert this same propositional content by uttering the sentence, “I hereby assert that it is raining.” Here, too, I do not assert that I am hereby asserting that it is raining, although once again I do say something in the sense of expressing the content of the sentence “I hereby assert that it is raining.”
Austin’s generalisation of the notion of a performative sentence, as well as the fact that for any speech act there are performative sentences permitting performance of this speech act intimates, I think, the only coherent sense one can give to what Brandom says about the use of the first person singular: the ‘key’ use of ‘I’ is its use in speech acts performed by uttering performative sentences. Naturally, this use is not ‘key’ in the sense of statistically most common, but rather in the sense that that referential behaviour, that contribution to fixing overall semantic content, which distinguishes an expression as a first person singular pronoun are precisely those which an expression must have in order to assume subject position in an explicitly performative sentence.19 Somewhat more crudely put, an expression just is a first person singular pronoun insofar as it has the properties needed for it to play the subject role in the serious and literal utterance of such sentences, the kind of utterance of such sentences in which the speech act done is the speech act said. Consequently, to understand an expression as a first person singular pronoun is to know how to wield it as subject term in the performance of explicitly performative utterances.
That this is indeed what Brandom’s account of the first person singular pronoun ‘I’ comes to is intimated by such claims as that what the first person expresses is “the undertaking by acknowledging of a commitment.” (553) This kind of claim indicates that what Brandom is seeing through a glass but darkly is precisely the fact that to be able to wield an expression as a first person singular pronoun is to know how to perform speech acts explicitly. That is, it is to know how wield an expression as subject term in an explicit performative sentence, the kind of sentence through whose serious and literal utterance one does something linguistic in making discursively explicit one’s doing of it – which discursive explicitness is itself both expressed and presupposed by the (naturally itself always only optional) self-referential prepositional phrase ‘hereby’.20
Now this admittedly highly reconstructive interpretation certainly makes Brandom’s account of the use of ‘I’ look far more plausible. Why, then, have we had to work so hard for it? In short, why could not Brandom have said all this himself? Because the phenomenon of performative speech acts and sentences, hence the first person, understood as something essentially ‘for’ performing such speech acts, presents Brandom with a fundamental and insuperable difficulty.
§ 4: Acknowledging, Attributing and Saying
Acknowledging doxastic commitments to p can never be construed as a matter of attributing these commitments to oneself, for this would be to say that one could assert that p by asserting (or reporting) that one is asserting that p. Or as Brandom puts it, it would be to think that commitment could be attributed in such a way that “attributing a commitment (for instance, an intentional action or a belief) … would have the same practical significance as undertaking that commitment.” (554) Acknowledging the doxastic commitment C and attributing C to someone are thus distinct acts in the sense that to attribute C to oneself is never eo ipso to acknowledge C. At least in part for this reason, Brandom insists that “… acknowledging commitments … cannot be reduced to attributing them, even to oneself.”21 (554) What, however, are we to say when S utters the performative sentence, “I hereby assert that p”, therein asserting that p, that is, acknowledging commitment to p? It is intuitively clear that in this particular case S is saying something, namely, that he, S, is asserting hereby that p. What is this saying, given that it cannot be a matter of S’s attributing to himself doxastic commitments to p22 (since this saying just is S’s acknowledging these commitments)?
Considerations such as these once moved Austin to make another important distinction, this time between what he called locutionary and illocutionary acts: illocutionary acts are Brandom’s kind of speech act, that is to say, acts which, in Brandom’s terminology, consist in acknowledging commitments, whether these be commitments of the kind one undertakes when asserting, or in promising, or in ordering, requesting and so on. Locutionary acts, by contrast, are acts in which the speaker uses language, i.e., adheres to linguistic rules in order to perform acts of saying something – of expressing a proposition or, if this is one’s preferred jargon, of determining a truth condition. Strictly speaking, Austin’s talk of acts is misleading; one should rather speak of the locutionary and illocutionary dimensions of the one speech act, whereby the locutionary dimension is contained within the illocutionary. For Austin’s idea was that an utterance, insofar as it was an act of asserting, requesting, ordering, promising, and the like – an illocutionary act in his terminology, an act of acknowledging commitment in Brandom’s – presupposed that this same utterance was an act of saying something, of uttering expressions according to linguistic rule or norm, therein expressing a proposition (or, if you prefer, therein determining a truth condition).23 In this its locutionary character an utterance was a saying in a sense not identical with, but presupposed by, its character as an asserting (or ordering, promising, etc.).
Now Brandom cannot accept this as an accurate phenomenological description of language use. In particular, he cannot accept that the illocutionary character of an utterance as the acknowledging of commitments should presuppose the locutionary character of the utterance, understood as the use of linguistic expressions according to rule in order to say something, that is, in order to express a proposition or to determine a truth condition. For this is to concede that, contrary to the fourth thesis of Brandom’s programme, the use of linguistic expressions according to linguistic norm or rule is not a matter of acknowledging commitment and that consequently the contentfulness of utterances made in the (correct and literal) use of linguistic expressions cannot be analysed in terms of the commitments a speaker acknowledges in uttering these expressions. Whatever rules or norms assertion, in Brandom’s language, the acknowledging of doxastic commitment, is subject to, these rules or norms are not the rules or norms which constitute the contentfulness of linguistic utterances, hence the distinctively semantic use of linguistic expressions. And all one can then say about the rules or norms of the distinctively semantic use of language is that they are rules and norms for using linguistic expressions in order to say something, understood simply as a matter of expressing a proposition or of fixing a truth condition, indeed, of referring to an entity and predicating something of this entity. In short, to concede that the illocutionary dimension of speech acting presupposes a locutionary dimension is to concede that the normatively pragmatic dimension of speech acting presupposes precisely that dimension of semantic contentfulness – what Brandom calls representation – which he seeks to explicate in terms of commitments undertaken in the utterance of linguistic expressions.
Let us recast this point in Brandom’s own terminology, for by so doing, one can bring out very clearly the way in which Brandom’s conception of content constrains his effort to do justice to the phenomenon of ‘I’-saying and –thinking, hence of self-consciousness: Brandom cannot allow that there can be an act of expressing, of making discursively explicit, acknowledgement of commitment which is not an act of attributing commitment (since this is to concede that locution is prior to, and presupposed by, illocution). But as we have already seen, to attribute commitment can never be to acknowledge this commitment oneself. Consequently, attributions of commitment are always only a matter of asserting or reporting of another that this latter is now acknowledging commitment; or of asserting or reporting of oneself that one was then (or perhaps will in future be) acknowledging commitment. The very idea of attributing commitments to oneself, i.e., to one’s current self, therein acknowledging them, must be an oxymoron on Brandom’s account. True, Brandom does claim at one point that “(t)he attitude of acknowledging a commitment is in effect that of attributing it to oneself.” (196) But Brandom himself admits24 that this claim needs ‘qualification’ in the light of chapter 8, sub-section 5.2. The problem is, however, that the qualification needed would appear to be negation – something that Brandom himself comes close to conceding when he says, as we have already seen, that “… acknowledging commitments … cannot be reduced to attributing them, even to oneself.” (554)
What would it take to admit some sense in which one could “attribute commitment” to oneself, therein acknowledging it? What this requires has already been intimated: one must allow that a speaker can “attribute commitment” in a manner which is not a matter of asserting, or indeed of performing any other illocutionary act at all, but rather of using language to express a content in a manner neutral as to whatever illocutionary character this same use has. Of course, this requirement makes it strange to speak of attributing commitment at all since the notion of attribution ineluctably insinuates that of truth-claiming, i.e., assertion. To attribute something x to something y is surely to assert that y has x. Precisely for this reason, Austin preferred to speak of what a speaker literally says, that is, of the locutionary character of a speech act. Only such a neutral dimension of locution permits one to acknowledge a commitment by “attributing” this commitment to oneself, or rather, by saying that one is acknowledging this commitment hereby. And so only by recognising such a conceptually prior, ‘neutral’ dimension of locution, in other words, a distinctly locutionary sense in which a speaker expresses acknowledgement of commitment, can one hope to accommodate the use of explicitly performative sentences, hence all speech acting insofar as this latter is something which the speaker can always in principle accomplish explicitly by using explicitly performative sentences.
Of course, what is distinctive about speech acts performed by explicitly performative means is their indexical self-referentiality,25 typically expressed by the (always only optional) prepositional phrase ‘hereby’. Now when we say that all speech acts can in principle be performed by explicitly performative means, we are in effect saying that this indexical self-referentiality is not something which only occurs in speech acts actually performed by explicitly performative means; rather, we are saying that it is a feature of all speech acts which performance by explicitly performative means simply makes explicit. Such indexical self-referentiality is an inherent feature of all speech acting and indeed of the first person insofar as this is or entails the ability to accomplish explicitly performative acts of thought or talk.26 Thus, as long as one refuses to admit such a neutral dimension of locution, one finds oneself having to give a distorted picture not just of the use of explicitly performative sentences but, even more fundamentally, of the capacity for ‘I’-saying and –thinking which underlies all speech acting insofar as it is essentially accomplishable in explicitly performative fashion. In effect, one finds oneself having to construe ‘I’-saying and -thinking very much in the manner of Hume, Mach and the early Husserl, who all refuse to take the self-referentiality of self-consciousness seriously, construing it rather as a matter of higher order consciousness (or ‘representation’) of lower order consciousness of something.
Evidently, very important and venerable issues hang on whether one can admit a locutionary dimension to speech acting – issues which cannot, of course, be pursued here. Let us simply note once again that Brandom cannot admit such a dimension because it conflicts fundamentally with his normatively pragmatist conception of content. For to admit such a locutionary dimension is to concede that the activity of acknowledging and attributing commitments, while it involves the utterance of linguistic expressions according to the (content-fixing) rules or norms of language use, cannot be identified with the same. And so these latter norms or rules can indeed only be characterised as specifying just what one may or may not say with linguistic expressions, i.e., what one may and may not refer to, and predicate with, these expressions. This is a result clearly unacceptable to Brandom, one he must reject, even if the cost of so doing is inability to do justice to the phenomenon of explicitly performative speech acting, hence ‘I’-saying and -thinking, in other words, self-consciousness, itself.
But, one might respond, this argument is too hasty. The argument thus far would have us conclude that acknowledgement and attribution of commitment does not proceed simply by uttering linguistic expressions according to linguistic norms (which, if it were true, would mean that these norms specify what commitments one acknowledges or attributes when expressions are correctly used); rather, it proceeds in or through saying something, and saying something is what is accomplished by the truly linguistic, i.e., semantic use of language according to truly linguistic norms (so these genuinely linguistic norms must be understood as specifying merely what things one says when expressions are correctly used). But does this anti-pragmatist conclusion really follow? Perhaps indeed in the case of speech acts performed by uttering explicitly performative sentences, commitment is acknowledged in some sense by ‘saying’ that it is being acknowledged. Yet this sense of saying is completely harmless from Brandom’s point of view because it occurs so to speak in direct rather than indirect speech. In other words, when S utters the sentence, “I hereby assert that p,” the only sense in which S says (as opposed to asserting) anything is captured quite adequately by using direct speech: in uttering this performative sentence, S asserts that p by saying, “I hereby assert that p.”
Clearly, the idea underlying this objection is to account for how S can be asserting that p by uttering this sentence without implying in any way that S accomplishes this in virtue of what this sentence, as uttered by S on this particular occasion, means, that is, what it says. For example, one might say that there is some special convention governing the performative operator “S asserts hereby that …” in which an arbitrary sentence ‘p’ is embedded such that to utter ‘p’ as embedded in this operator is simply to assert ‘p’. If this is so, then one could say that in uttering this sentence, S asserts, and intends to assert, that p simply by saying, “I assert hereby that p,” i.e., by uttering something which, by this special convention, constitutes the utterance as an assertion that p. And a competent interpreter of S’s utterance, should there be one in the vicinity, would appeal directly to this convention, without attempting to determine what S’s sentence says, in order to understand this utterance; the interpreter would relate to (the token of) the sentence uttered by S simply as sign signalling that S is making that assertion which one makes by seriously and literally uttering the embedded sentence p.
It does not take much to see that this suggestion is spurious. It is not enough simply to find some alternative way in which a speaker could assert (and be interpreted as asserting) that p by uttering the explicitly performative sentence, “I hereby assert that p” – some alternative way that sits better with a normatively pragmatist conception of content. For the real problem is that on this conception of content the explicitly performative sentence “I hereby assert that p” cannot say what it, as uttered by S on a certain occasion, manifestly does say. In other words, it is not enough simply to find an alternative conception of how performative sentences work and are understood as working. One has first to show that such sentences simply do not express the content they certainly appear to express; providing an alternative account of how performative sentences work which utilises direct rather than indirect speech is a conceptually secondary exercise appropriate only after this first task has been accomplished.27
It is quite clear, however, that any attempt to show that such sentences do not express the content they appear to express is futile. One cannot seriously doubt that the self-referential prepositional phrase ‘hereby’ is doing some work, that is to say, that it is genuinely referring to something. But if the prepositional phrase “hereby” genuinely refers, then so does the rest of the sentence; the sentence must express a semantic content, indeed precisely the content it appears to express. Given this, no point remains in maintaining that a speaker does not assert that p in virtue of his saying that he is. In fact, it would be absurd to deny that the performative sentence “I hereby assert that p” is means of, or vehicle for, asserting that p in virtue of its linguistic meaning and thus the content which this linguistic meaning imparts to any particular utterance of it. For one cannot deny that the individual parts of this sentence have meaning. Given this, however, and the fact that the meaning of a sentence is in some sense a function of the meaning of its parts, it follows that the performative sentence does have a meaning; it is indeed such that when an arbitrary subject S utters, S says something, namely, that he is asserting (by means of this very utterance) that p. So simply in virtue of its meaning, this sentence must be a means of saying that one is asserting hereby that p and this fact alone suffices to render it a potential means of asserting that p. All one has to do in order to turn one’s saying into an actual doing of what one is saying is to mean what one says.28
There is thus no escaping the conclusion that, for at least one kind of sentence, its character as a means of acknowledging commitments presupposes, hence cannot be used to ‘define’, its character as expressing a proposition, determining truth conditions, etc. Brandom’s programme cannot, therefore, account at least for this kind of sentence and the kind of speech act it is used to accomplish, namely, explicitly performative speech acts. And this means that Brandom’s programme cannot account for the first person singular pronoun ‘I’ since this expression is defined by its key use in the performance of such speech acts – something Brandom himself, at least when reconstructed, himself recognises. With this, Brandom’s programme founders – founders on the phenomenon of ‘I’-saying and -thinking, that is to say, on the phenomenon of self-consciousness. Note, too, how Brandom’s own notion of expression, of making things explicit, has played a crucial role in this foundering. There is no problem as long as expression is merely something done retrospectively, by logicians in acts of thought and talk which are distinct from the acts of thought and talk they make explicit. But as soon as what makes explicit and what is made explicit coincide, making it explicit becomes something Brandom cannot do.
§ 5: Brandom on the Content of First Person Thought and Talk
Note that we have been able to reach this negative assessment of Brandom’s account of the use of the first person without considering in any way how he might respond to the issue central to Perry, namely, the question of just what semantic or intentional content first person assertion or belief has. Brandom had originally passed over this issue in order to outline the ‘expressive’ job done by the first person singular. In the sub-section following his discussion of the use of first person singular expressions, however, he returns, albeit obliquely, to the issue of content. Here, too, substantial reconstruction is needed because here, too, it is very hard to pin Brandom down. Nonetheless, it is possible to determine what Brandom at least ought to say, indeed what at one point he probably does say, by way of answer to the issue of just what distinctive content first person assertion and belief have. And when we do this, we see why Brandom is so hard to pin down: his Sellarsian inferentialism conspires with what one might call his socially constructivist account of norms to paint him into a corner from which he can never deal adequately with the issue of content as this was raised by Perry.
As we know, Brandom wishes to construe the norms and rules of language use as specifying what commitments one acknowledging when one utters linguistic expression according to norm or rule. And as we have also seen, this general normatively pragmatist conception of content contains the four component theses listed in § 2 above. Of particular relevance here is the second of these four theses, which articulates the inferentialism Brandom derives from Sellars: the content of assertion or belief is fixed by the set of inferential relations which this content has to other such contents. Now this inferentialist thesis directly entails that any semantic or intentional content – what is asserted or believed or judged to be true – is by definition something standing in inferential relations to other such contents. So any semantic or intentional content must necessarily be something expressible by full sentences, that is, sentences which can stand alone, in particular, as lines in a proof, without relying on any embedding in a wider sentential context for their intelligibility. Attributions of content which occur in indirect speech using anaphorically dependent sentences, as in “S asserts that he is Φ,” must strictly optional, derivative forms for making such attributions. All in all, Brandom’s inferentialism forces him to insist that all semantic and intentional contents fulfil condition (i) of the two conditions given above on what it is to be a genuinely propositional content.
But there are further constraints to which Brandom should, in all consistency, subject the notion of content. He rightly refuses to take as an unanalysed primitive the notion of norm and norm-governed behaviour to which his account of semantic and intentional content appeals. Thus it is that in Making It Explicit, before turning to expound and elaborate his normatively pragmatist conception of content, Brandom first provides an account of what it is for behaviour to be subject to a rule or norm. In this discussion, he attempts to steer a middle course between so-called regularism and regulism, that is, between a conception of such behaviour as a mere regularity in what actors do (which is manifestly insufficient to constitute behaviour as norm- or rule-governed); and a conception of such behaviour as simply a matter of having some explicit formulation of the rule or norm in one’s head from which one deduces what the rule or norm determines as conforming to and diverging from it (which, according to Brandom, is also insufficient, indeed, when taken as sufficient, leads to the kinds of paradox to which Sellars and Wittgenstein independently draw attention). After some considerable dialectic, Brandom eventually concludes that a regularity in agents’ behaviour counts as norm-governed insofar as it is sustained by the willingness of these agents to regulate and adjust their behaviour in the light of others’ assessments of it as right or wrong, correct or incorrect.
Now this conception of what it is for behaviour to be subject to rule or norm ultimately boils down to the claim that all norms and rules are constituted by the normative or, to use Brandom’s language, the deontic attitudes of subjects assessing the performances of others. In other words, that a certain kind of behaviour is subject to a norm N entails that this kind of behaviour is something done regularly by a sufficient number of subjects,29 whereby adherence to this regularity is sustained by the preparedness of most subjects to adjust their behaviour in the light of assessments by other subjects of their individual performances as correct or incorrect. And when this conception of norm- or rule-governedness is applied to Brandom’s views on assertion, belief and their semantic or intentional content, it entails the following important consequence: what determines that a particular subject’s assertion or belief constitutes the undertaking of one set of commitments and entitlements, hence has one content rather than another, are the views of other subjects as to what set of commitments and entitlements are undertaken when the sentence used to make the assertion, or to describe the content of belief, is correctly used to make an assertion. What set of commitments and entitlements one acquires by asserting or forming a belief, hence what content one’s assertion or belief has, is thus a function of objective congruence between the opinions of several other subjects about how the sentence used to make the assertion is correctly asserted.
From this, however, it immediately follows that the very identity of semantic and intentional contents is fixed intersubjectively. Such contents are necessarily graspable by more than one subject. On Brandom’s account, that another “be able to attribute the very same commitments that [a speaker] undertakes or acknowledges …” (562) by performing speech acts is not just a necessary condition on successful communication, it is constitutive of semantic and intentional contentfulness as such. It would seem, then, that Brandom’s social constructivism about norms forces him to insist that all semantic and intentional contents fulfil condition (ii) of the two conditions on propositional contentfulness given at the beginning of this paper. Consequently, sentences in the first person such as “I am Φ”, “I was Φ,” etc., cannot be the sole or primary, the only truly adequate means of attributing content to the assertion of such sentences since any sentence in the first person can only attribute the right content to S’s assertion if S is the attributor.
In short, Brandom’s inferentialism and his socially contructivist conception of normativity together force him to maintain that what S asserts in asserting the sentence, “I am Φ,” is propositional in the sense defined by conditions (i) and (ii) above. If, however, this is so, then all semantic and intentional contents must be expressible by some full sentence in the third person, something clearly contradicted by Perry’s insights into the content and distinctive motivational force of first person assertion and belief. Perhaps for this reason Brandom at no point acknowledges that he is forced by his programme to insist that all content is subject to condition (ii). Indeed, he writes as if it were unquestionably open to him to deny this. Thus, although he never makes his position completely clear or explicit, he seems quite happy to endorse the view that the content of first person assertion or belief is only truly expressed by sentences in the first person. Since such sentences are themselves full sentences, the content of first person assertion or belief is thus guaranteed to satisfy condition (i). While this permits Brandom to maintain that the content of first person assertion and belief is constituted by its inferential relations, thereby preserving consistency with his inferentialism, it of course immediately commits him to a Fregean doctrine of ‘I’-thoughts accessible only to their asserter or believer.
That Brandom accepts just this commitment comes out in his claim to be able to make precise McDowell’s notion “… of suitably related corresponding thoughts or de re senses.” (566) According to McDowell, the content of a first person assertion or beliefs is inaccessible to individuals other than the subject of this assertion or belief yet this does not make it unintelligible how such assertions and beliefs
… can be understood by and communicated to others. No one else can have the thought that Michele would express by saying, “I am threatened by a bear,” but this does not mean that no-one else can understand what thought she would be expressing by that claim. [McDowell] says: “Frege’s troubles about ‘I’ cannot be blamed simply on the idea of special and primitive senses; they result, rather, from the assumption – which is what denies the special and primitive senses any role in communication – that communication must involve a sharing of thoughts between communicator and audience. That assumption is quite natural, and Frege seems to take it for granted. But there is no obvious reason why he could not have held, instead, that in linguistic interchange of the appropriate kind, mutual understanding – which is what successful communication achieves – requires not shared thoughts but different thoughts which, however, stand and are mutually known to stand in a suitable relation of correspondence.”30 (562)
Brandom then joins McDowell in this latter’s retreat “… from his earlier endorsement of samesaying as the relation between reported and reporting tokenings in (de dicto) ascriptions of such beliefs and their expressions.” (562) All that is required for successful communication is that the content expressed by first person assertion or possessed by a first person belief stand in the right relation to the content of that assertion or belief which reports on the former, i.e., attributes content to it. When two different speakers S and S* each assert, “I am Φ,” they do not assert “the same thing” in the sense of what one reports each as asserting when one uses indirect speech, that is, when one uses anaphorically dependent sentences of the form “he is Φ”. Nonetheless, they both hold the same sentence true and this is all the sameness needed for successful communication since it enables any reporter to map what each speaker asserts onto its own distinct set of further contents.
In the first instance, these further contents will be themselves essentially indexical or, as McDowell would put it, de re: one can report both S and S* as asserting, “You are Φ” (or “He is Φ or “That man there is Φ”), and in each case one will be attributing a different content which can be attributed in no other way. But ultimately, the reporter can discriminate these different, essentially indexical or de re contents by mapping them onto distinct, strictly propositional contents which fulfil condition (ii). S and S* will, after all, be spatio-temporally distinct individuals and so, if S is the speaker at 〈s, t〉 while S* is the speaker at 〈s*, t*〉 (where 〈s, t〉 ≠ 〈s*, t*〉), S’s assertion will be true just in case the proposition that the speaker at 〈s, t〉 is Φ is true while S*’s assertion will be true just in case a different proposition is true, the proposition, namely, that the speaker at 〈s*, t*〉 is Φ. The content or, in McDowell’s terminology, the sense, expressed by ‘I’ in S’s mouth and the content or sense expressed by ‘I’ in S*’s mouth, while accessible only to S and S* respectively, can be mapped onto correspondingly distinct contents or senses accessible to others, indeed ultimately, accessible to all across the board. And this correspondence with distinct sets of such allo- and intersubjectively available contents, which constitutes an equivalence relation stronger than mere co-reference yet weaker than identity, is all that communication requires.
Clearly, McDowell is right in his claim that successful communication requires no stronger equivalence relation than such correspondence. Yet the question of what kind of equivalence relation must exist between what a speaker says and what another reports him as saying in order for communication to be possible is not the primary issue as far as Brandom’s (or indeed any one else’s) account of the first person is concerned. The relevant question in this regard is whether one can admit the kind of content which is to stand in the equivalence relation required for communication, whatever this content, and thus whatever this equivalence relation might be. Brandom’s whole discussion in sub-section 5.3 of chapter (559-567) only addresses the issue of just what this equivalence relation is. Following McDowell, he argues that this need not be strict identity – as if the only obstacle to his embracing (what Perry has called) “… the doctrine of limited accessibility toward strongly de re beliefs …” (562), thereby ensuring that the content of first person assertion and belief satisfies condition (i), were to secure enough equivalence between what is said and what another says is said for communication to be possible. But this is not so at all; the real question is whether the doctrine of limited accessibility is an option for him at all. From the outset, an affirmative answer to this question is simply begged.
Yet above we have already seen that the doctrine of limited accessibility is not really an option for him: Brandom’s socially constructivist conception of the normative entails that contents are so thoroughly public, intersubjective affairs that no room is left for (what one might call) ‘private’ ‘I’-thoughts. Or, to put the point in another, less misleading way, no room is left for the idea that when two different speakers S and S* each assert the indexical sentence “I am Φ,” what each asserts is at once “the same” (in that each holds the same sentence true) yet “different” (in that the truth conditions of each speech act differ in virtue of the different ‘perspectives’ of the speakers).31 What S and S* say simply is what others would say they say: the perspective from which each says, “I am Φ,” is ceaselessly deferred to that from which others understand what they say. So this sentence, and, mutatis mutandis, all other sentences, whether indexical or non-indexical, are in truth said from nowhere – this notwithstanding Brandom’s insistence that his account is well able to capture how “thought and talk give us a perspectival grip on a nonperspectival world.” (594)
Importantly, at one point Brandom appears de facto to concede this, for here he makes claims which, under any plausible interpretation of them, directly entail that the semantic or intentional content of first person assertion or belief must be strictly propositional in the sense defined by conditions (i) and (ii) above, hence must be expressible by some full sentence in the third person. In sub-section 5.3 of Chapter 8, we find him saying that “Michele, who believes “I am threatened by a bear,” and Nicole, who believes “Michele is threatened by a bear,”” although in what Perry calls different belief states,32 nonetheless “share a content.” (561) And, on this same page, he also says, “Two different people who each believe something they could express by asserting the sentence “I am threatened by a bear” share a belief state, though the contents of their beliefs (who it is they take to be threatened) differ.” (561)
Now this latter claim might at first suggest that the notion of content which Brandom has in mind here is so to speak de re: two beliefs have the same content if and only if they predicate the same propositional function of what, objectively speaking, are the very same things. But this cannot be right because this notion of content would be too weak to do any real work: in this sense of content beliefs are contentless if and when their reference fails, and this shows that it is not the kind of contentfulness we are interested in, viz., that kind of contentfulness which underpins a belief’s motivational force and indeed its interpretability and communicability. It seems we must take seriously Brandom’s parenthetic gloss about who it is a speaker takes to be threatened: the notion of content he is working with here must be such that two beliefs have the same content if and only if, in addition to predicating the same propositional function, they also refer in the same way, i.e., have the same referential properties.
Consequently, if the belief that Michele has in believing, “I am threatened by a bear,” has the same content as the belief that Nicole has in believing, “Michele is threatened by a bear,” then both beliefs must refer to Michele in the same way. But Nicole’s belief refers to Michele in the manner determined by the proper name ‘Michele’.33 So the sentence which picks out the content of Nicole’s belief picks out the content of Michele’s equally well. And this sentence is none other than the completely third person sentence “Michele is threatened by a bear.” This is in effect to say that both beliefs have the very same propositional content. Or, to put the point another way, the biconditional which Nicole’s belief inherently34 determines as specifying the conditions under which it is true, namely, that it is true if and only if Michele is threatened by a bear, also does the same job for Michele’s first person belief. In all consistency, Brandom should say that the biconditional which specifies the truth condition that a first person assertion or belief inherently determines for itself is one which does the same job for a third person assertion or belief in which the same propositional function is predicated of the same entities.
This position is untenable. For one thing, it immediately entails that a first person assertion or belief would inherently determine any number of biconditionals as fixing truth conditions for it; for another and more important thing, it contradicts Perry’s insights into the distinctive content and motivational force of first person assertion or belief. For Perry’s original insight was that no full sentence whose intelligibility does not rely on some larger sentential embedding suffices to express the content of first person assertion or belief – as is corroborated by the fact that the motivational role or force of first person assertion or belief differs from that of any assertion or belief whose content is expressed by such a sentence. That is, no full sentence whose intelligibility does not require larger sentential embedding can form the right hand side of the biconditional which an assertion or belief in the first person intrinsically determines as specifying its truth conditions. Yet commitment to this untenable position is implicit both in claims Brandom makes as a matter of brute fact and, more deeply, in his simultaneous allegiance to inferentialism and a socially constructivist conception of normativity.
At this point, we surely find the explanation for why Brandom does not address the real questions raised by first person assertion and belief, and why sub-sections 5.2 and 5.3 of Chapter 8 are so hard to understand. Brandom’s views, both on the use of the first person singular pronoun ‘I’ and on the content of first person assertion and belief, are so hard to pin down because the first person constitutes a fundamental difficulty for his whole approach to semantic and intentional content. Dealing adequately with the first person would require him either to renounce that inferentialism which embodies his Sellars-derived rationalism; or to give up that socially constructivist, anti-subjectivist account of normativity which naturalises this rationalism, thereby immunising it against the objective idealism which, historically speaking, has always been a major source for the pragmatist tradition.
Alston, William P. 1964. “Linguistic Acts,” in American Philosophical Quarterly, Vol 1., No. 2, pp.138-146.
Anscombe, Gertrude E. M. 1974. “The First Person,” in Mind and Language, edited by Samuel D. Guttenplan, Oxford: Oxford University Press, pp.45-65.
Austin, John L. 1975. How to do Things with Words, Oxford: Clarendon Press:
Brandom, Robert 1994. Making It Explicit, Cambridge, Mass.: Harvard University Press.
Castañeda, Hector-Neri 1966. “‘He’: A Study in the Logic of Self-consciousness”, in Ratio, Vol. 8, pp.130-157.
Castañeda, Hector-Neri 1967. “On the Logic of Self-Knowledge,” in Noûs, Vol. 1, pp.9-21.
McDowell, John 1984. “De Re Senses,” in Philosophical Quarterly, Vol. 34, No. 136, pp.283-294.
Perry, John 1979. “The Problem of the Essential Indexical,” in Noûs, Vol. 13, pp.3-21.
Perry, John 1993. The Problem of the Essential Indexical and Other Essays, Oxford and New York: Oxford University Press.
Rorty, Richard 1979. Philosophy and the Mirror of Nature, Princeton, NJ: Princeton University Press.
Searle, John R. 1969. Speech Acts—An Essay in the Philosophy of Language, Cambridge: Cambridge University Press.
Sellars, Wilfrid 1963. “Empiricism and the Philosophy of Mind,” in Wilfrid Sellars, Science, Perception and Reality, London: Routledge & Kegan Paul, pp.127-196.
Or, for that matter, any other sentence containing first person pronouns, whether in the nominative or not, e.g., ‘I’, ‘me’ and ‘my’, perhaps indeed whether in the singular or not, e.g., the corresponding plural forms ‘we’, ‘us’ and ‘our’. ↩
The sentence “he is Φ” contains the pronoun ‘he’, which is used anaphorically, that is, as a singular term which derives its reference from some other singular term contained within the larger sentential context to which the dependent sentence “he is Φ” belongs. What singular term the pronoun derives its reference from is determined purely by considerations of what makes sense in the context of speech. In this case, it is the singular term ‘S’ contained within the larger sentential context “S believes that …”. One can only understand the sentence “he is Φ” by appeal to this larger sentential context. ↩
See Perry 1979 and 1993. ↩
Strictly speaking, one should add: “and to acquire a certain set of entitlements.” ↩
And also, at the practical rather than the cognitive or, as Brandom calls it, doxastic level, to acting consistently with one’s utterance or belief. ↩
I doubt that coherent sense can be made of the notion that assertion as such is a matter of undertaking of commitments, whether to the truth, sincerity or justifiability of what one says, or of acquiring and losing entitlement to further moves in the game of giving and taking reasons within the space of reasons. Brandom and the many others who also endorse this view of thought and talk can only do so because of continual equivocation between a notion of the normative which, while true of all acts of thought and talk, is too weak to do any serious work; and a notion of the normative which, while reflecting what one ordinarily understands by the ‘normative’, does not apply to all assertion and belief across the board. Unfortunately, the normative is an idea whose time has come: it is seen as providing the key to overcoming a genuinely metaphysical dualism in favour of an allegedly non-metaphysical dualism of the natural and the normative which avoids the reductionism or eliminativism of traditional naturalism. Strictly speaking, of course, one should describe the normative as an idea which time has come again, since current neo-pragmatist celebrations of it (Sellars, Rorty and Brandom) are in fact reviving an idea which goes back to Hermann Lotze (1818-1883) and his distinction between efficacious Sein (Being) and inefficacious Geltung (validity). In various forms, this kind of distinction permeated much neo-Kantian philosophising in mid- to late 19th century Germany; indeed, it manifests its influence in Frege’s distinction between the Wirklichkeit of physical and psychological reality and the Objektivität of thoughts. Since my current concern is not with this fundamental assumption of Brandom’s normative pragmatism, I will just acquiesce in it here. ↩
Or, in other words, that the conditional proposition “If p, then q” is logically true. ↩
See 115f. ↩
“The fact that one thereby undertakes consequential commitments that may reach beyond what one acknowledges just shows that the generic attitude of undertaking a commitment is not to be identified with its species attributing a commitment to oneself, which is acknowledging it.” (196) ↩
There is perhaps a variation on a theme of Anscombe here: Anscombe writes, “Getting hold of the wrong object is excluded, and that makes us think that getting hold of the right object is guaranteed. But the reason is that there is no getting hold of an object at all. With names, or denoting expressions (in Russell’s sense) there are two things to grasp: the kind of use and what to apply them to from time to time. With “I” there is only the use.” (Anscombe 1974, p.59) Admittedly, Brandom does not deny that ‘I’ is a referring expression in any sense at all, hence does not maintain that there is only what Anscome means by use. Even so, he apparently believes that what distinguishes ‘I’ from other referring expressions, in particular, denoting expressions in Russell’s sense, is a distinctive use in the sense of a distinctive contribution at the pragmatic level of speaker-audience interaction rather than at the semantic level of the kind of truth condition determined. If this (somewhat speculative) interpretation is right, then in all consistency Brandom should say that ‘I’, viewed simply in its capacity as a referring expression, imparts (in the case of first person assertion) or expresses (in the case of first person belief) a contentfulness which is nothing out of the ordinary, so that in respect of this its purely semantic function there is nothing special about ‘I’. ↩
Brandom himself describes his story in terms very similar to these – see 559. ↩
Anscombe 1974, p.47. ↩
Or, as Brandom at one point also says, of attributing such commitments to oneself: “The attitude of acknowledging a commitment is in effect that of attributing it to oneself.49” (196) In the note to this passage (note 49, p.675), Brandom acknowledges that “(t)here are subtleties that require qualifying this formula, some of which are discussed below in 8.5.2 …,” that is to say precisely in his discussion of the first person. ↩
This way of thinking about what it is to form an intention explains the close connection one intuitively feels to exist between forming an intention to do something and self-consciously deciding to do something. Of course, if this is what it is to form an intention, then non-linguistic beings, who do not dispose over sentences of the form “I shall do A” (or its equivalent in some other language), cannot form intentions. This is not a particularly worrying conclusion, however. It does not necessarily entail that such beings do not have any intentional states or experiences at all, but merely that their beliefs and desires do not interact causally to produce intentional behaviour via conscious decision – surely a perfectly acceptable and plausible position. Of course, the intentional behaviour of self-conscious, linguistic beings is often similarly not the result of conscious decision, hence of an actual intention to engage in this behaviour. Note, incidentally, that this way of interpreting the linguistic function of the sentence “I shall do A” implies that such sentences have a quasi-performative character: to assert or think to oneself, “I shall do A” is in effect to assert or think to oneself, “I hereby resolve to do A.” ↩
That the whole issue is not addressed becomes clear once one unpacks the difficult paragraph on 552-553. ↩
It immediately follows from this conception of what it is to formalise processes of reasoning that the various logical truths which one can associate with the formally valid rules of inference formulated by the logician do not explain, but simply reflect or express, the validity of how we do as a matter of fact reason. ↩
As when the minister says, “I hereby declare you husband and wife.” ↩
For of course the content of any indexical sentence varies from context to context, speaker to speaker. ↩
What this referential behaviour or semantic contribution is forms the topic of another paper, currently in preparation, examining Anscombe on the first person. ↩
Naturally, one can generate an infinite regress by substituting for the prepositional phrase ‘hereby’ the more complex and more descriptive prepositional phrase “by expressing that one is performing such and such a speech act hereby,” since this more complex phrase must itself contained what it replaces. But this regress remains harmless as long as one does not regard (or is not forced by external philosophical commitments to regard) one’s substitution of the more complex phrase for the phrase ‘hereby’ as explaining or analysing what the latter accomplishes. The phrase ‘hereby’, however disconcerting its self-referentiality might make it in the eyes of some, is the original phenomenon. ↩
Actually, Brandom says here, “But acknowledging commitments is the basic way of undertaking them, and undertaking commitments cannot be reduced to attributing them, even to oneself.” Clearly, if undertaking commitments cannot be reduced to attributing them, even to oneself, then neither can acknowledging commitments (which, according to Brandom, is simply a matter of undertaking commitments explicitly). ↩
I.e., cannot be a matter of S’s asserting or reporting that he is asserting that p. ↩
Austin may conceivably think that at least in some cases, for example, speech acts made by actors playing roles on the stage, an utterance can be a locutionary act without being an illocutionary act. It is never clear, however, whether he thinks a speaker can make an utterance which is a locutionary act and nothing more. But these issues are not relevant here. ↩
See the note accompanying this observation (note 49 to p.196 (675)) ↩
Such indexical self-referentiality is clearly a species of what was once called token-reflexivity. ↩
This suggests that the ability to think to oneself, “I am Φ,” trades on, hence presupposes the ability to perform speech acts, in particular, explicitly performative speech acts. This is surely plausible since it suggests an explanation of the phenomenological observation that self-conscious thinking to oneself is a matter of hearing oneself speak in one’s mind – precisely inner speech, a dialogue of the soul with itself. To think to oneself, “I am Φ,” is indeed to imagine (not in the sense of discursively conceiving but of literally ‘image-ing’) oneself actually asserting the sentence, “I am Φ.” ↩
Assuming, of course, that there is no other reason to propose this alternative account than the need to save one’s preferred normatively pragmatist conception of content. ↩
It is also worth pointing out that if one were to deny that the prepositional phrase ‘hereby’ refers, that indeed the whole sentence really does say something, one would find it impossible to explain why this phrase should ever have found its way into this performative sentence, why indeed this performative sentence (as opposed to, say, a specially appointed grunt or groan accompanying the sentence ‘p’) should be use to perform an explicitly performative assertion that p. ↩
To say that a certain kind of behaviour B is done regularly is to say that across a number of subjects and across a sufficient length of time, B is done if and only if A, where A is some (possibly disjunctive) set of conditions concerning the context of behaviour. ↩
The quote from McDowell is from McDowell 1984. The page number reference Brandom gives, namely, to p.105, appears to be incorrect. ↩
To put the point using Brandom’s own language, his problem is that he does not have conceptual room to make Perry’s distinction between “… a belief state exhibited by the believer and a belief content exhibited by that state …,” a distinction which corresponds to “… Kaplan’s notions of character and content.” (561) According to Perry, state (or character) and content are individually shareable; what is not shareable is their conjunction. For Perry, then, “(w)hat needs to be worked out is an account of the sense in which being able to share these individually suffices to make intelligible (hence communicable) their conjunction, which is not shareable.” (561) According to McDowell, only the belief state is shareable so he needs to work out an account of communication which shows it not to require strict shareability, i.e., identity of content. ↩
“Two different people who each believe something they could express by asserting the sentence “I am threatened by a bear” share a belief state … .” (561) This clearly permits us to gloss two individuals’ being in the same belief state as a matter of each holding, or being prepared to hold, the same sentence to be true. ↩
The fact that one could equally argue that Nicole’s belief refers to Michele in the manner determined by the first person pronoun ‘I’ as uttered by Michele only brings out more starkly the difficulty implicit in Brandom’s account. ↩
To say that a belief or assertion has the proposition that p as its content is, after all, simply to say that the assertion or belief inherently satisfies the complex biconditional propositional function of being true if and only if p – where ‘inherently’ here means “simply and solely in virtue of its being the belief or assertion that it is.” (Any assertion of the sentence “Snow is white” satisfies the complex property of being true if and only if grass is green, but not solely in virtue of this assertion’s being the assertion that it is; one must take the assertion or belief in conjunction with the further, quite contingent fact that snow is white if and only if grass is green.) ↩