The final central thesis of Brandom’s Between Saying and Doing concerns the nature of intentionality – the world-directed character of language and thought – that we call the pragmatically mediated conception of intentionality. In our view, this proposal is of profound importance to the overarching project of Analytic Pragmatism. Yet Brandom’s discussion is underdeveloped and, at times, frustratingly obscure. With this in mind, we propose to start with a careful and charitable reconstruction of the proposal. Having done so, we show that the arguments of the foregoing sections provide reason to reject the proposal.
Motivation: A Persistent Criticism of Semantic Pragmatism
One way to appreciate the central motivation for – and significance of – Brandom’s pragmatically mediated conception of intentionality is to start by focusing on a persistent criticism of the semantic assumptions of Analytic Pragmatism. More or less by definition, Analytic Pragmatism is committed to Semantic Pragmatism: the thesis that linguistic meaning is determined by use. Moreover, the sorts of analyses that Brandom seeks to provide consist in specifying the linguistic practices on which our mastery of various vocabularies are thought to depend. But as Brandom himself readily acknowledges, there is a worry that such analyses are, in fact, not semantic at all; and not for any reason that has anything specific to do with the particular analyses that he advocates. Rather, the worry is quite general. Such analyses fail to be semantic – and hence to do the work required of them by Analytic Pragmatism – because they conform to the general precepts of Semantic Pragmatism, and Semantic Pragmatism is unsatisfactory as an account of semantic properties. On such a view, meaning is ultimately to be understood in terms of the linguistic practices in which we engage, not the relations that terms bear to aspects of the world. But according to many, semantics, if it is anything, is concerned with precisely such word-world relations – e.g., with representation, truth, and reference.Footnote 23 In which case, it may seem that to the extent that Analytic Pragmatism seeks to provide semantic analyses, it is doomed to fail. Thus whatever else the value of pragmatic analysis, it cannot be a satisfactory extension of the project initiated by Frege, Russell and Moore.
The Pragmatist Responds
One obvious response to the above concern is simply to deny that semantics is centrally concerned with word-world relations. Yet clearly this is not Brandom’s response. Rather, he thinks that there is something importantly right about the idea that semantic properties involve word-world relations. The challenge, then, is to explain why this is so in a manner consistent with Semantic Pragmatism. This is precisely the role that his account of intentionality is supposed to fill.
Step 1: The Nature of Practices
Brandom’s response can be viewed as having two main steps. The first is driven by what he takes to be a very general fact about the nature of practices, which he views as a central claim of the pragmatist tradition. Of course language use qua practice involves relations between subjects and the world because in general practices involve aspects of the world: they are thick.Footnote 24 To take some simple examples, the practice of cooking cannot be adequately characterized absent such things as food items and kitchen utensils; and the practice of drinking tea cannot be adequately characterized without reference to such things as cups, saucers, and, of course, tea leaves. But such components of the practices are also aspects of the world, and thus the existence of these practices carry a commitment to subject-world relations. And what goes for cooking and tea drinking is also true of language use. Qua practice, it too ineliminably involves subject-world relations.
So there is a sense in which, according to Brandom, the worry with which we started is no objection at all. That is, if the concern is that Semantic Pragmatism severs the connection between word and world, then the concern is misplaced. At best it is an artifact of assuming that the relevant practices ought to be characterized “thinly”, in such a way as to avoid reference to aspects of the world. But as Brandom clearly recognizes, this does not settle the matter. The real problem for the semantic pragmatist is not that there are no relations between language-using subjects and the world. On the contrary, there are a great many such relations. Such relations are cheap. The real problem for semantic pragmatists is specifying the right relation(s) – those relevant for the purposes of characterizing the intentional character of language use; and this, according to Brandom, requires a more detailed, pragmatist account of intentionality.
Step 2: Combining the Accounts of Modal and Normative Vocabulary
To appreciate Brandom’s account of intentionality one needs first to make three general points clear. The first, already hinted at above, is that he views intentionality of the sort manifested by the practice of language use as a specific instance of a broader phenomenon, manifested by all human practices, which he calls practical intentionality. Practices quite generally, you may recall, involve relations between subjects and aspects of the world. Further, as Brandom notes, practices invariably involve what he calls practical intentionality: feedback-governed relations in which the results of earlier actions are used as guides to future, goal-directed behavior. This is true, for example, of cooking and tea drinking, but also football kicking and everything else for that matter, including language use. According to Brandom, then, the semantic intentionality characteristic of language use is best thought of as a specific instance of practical intentionality.Footnote 25
The second point concerns the order of explanation one ought to adopt in explaining semantic intentionality. According to Brandom, some traditionsFootnote 26 purport to start with independent conceptions of the states of the subject that represent and of the entities so represented, and then seek to characterize the nature of the intentional relation that holds between them —to somehow “bolt” them together.Footnote 27
Brandom maintains that such an approach is wrong-headed in that it gets the proper order of explanation exactly backwards, and thereby ignores precisely that which allows for intentionality in the first place: “the thick, essentially world-involving practices” in which we engage.Footnote 28 In its place, Brandom proposes that should start, as traditional pragmatists have advocated, with the world-involving practices themselves; and only characterize intentionality and its relata as a kind of abstraction from these practices.Footnote 29 Thus it is only by taking our rich, thickly characterized linguistic practices as primary and subjecting them to careful enquiry that, according to Brandom, we have any hope of understanding the general phenomenon of semantic intentionality.Footnote 30
The final general point concerns the specific practices most central to explaining intentionality. According to Brandom, the most central are those concerned with our use of normative and modal vocabularies. Indeed his strategy is, in effect, to explain intentionality by combining the two Kant-Sellars Theses discussed in Section 5, and seeking to draw out certain consequences of their joint endorsement. In Section 6.3 we argue that this strategy does not work. But first some further unpacking is in order.
Recall: for Brandom, though an explanation of semantic intentionality requires an account of the relevant subject-world relations, such an account is to be provided by abstraction from the relevant linguistic practices. The primary task, then, is to supply an adequate account of the relevant practices; and for Brandom the two central practices are those concerning modal vocabulary and normative vocabulary, since they inform us about the relata of semantic intentional relations: the world and the rational subject, respectively. The practices made explicit by modal vocabulary (CRI) impose constraints on the modal structure of reality. Roughly, they reflect constraints, implicit in our thought and language, on ways that the world could be and must be. In contrast, the practices made explicit by normative vocabulary (GAR) impose constraints on rational agency. Roughly, they tell us what is it to be a rational subject – a producer and consumer of reasons. Thus for Brandom these linguistic practices, though not themselves the poles of intentional relations, are crucial to explicating the nature of intentionality, because it is by providing appropriately perspicuous characterizations of them that we are able, by abstraction, to reconstruct the nature of the relata.
So far so good. But an account of intentionality must do more than merely explicate those practices that concern the relata. Rather, one needs to explain how each of these practices are related to each other so as to yield the larger practice on which semantic intentionality depends. In effect, Brandom must address an analog of the problem that confronts more traditional approaches to intentionality of the sort that he rejects. For as we have already noted, on some approaches to the explanation of intentionality, the challenge is to specify the relevant relation between the relata: i.e., subject and world. For Brandom, the challenge is to specify the relationship between two practices: the world-oriented practices made explicit by modal vocabulary and the subject-oriented practices made explicit by normative vocabulary.
How is this to be done? Brandom’s answer is that the two practices intersect in what he calls rational rectification, which is the practice by which new commitments are integrated into one’s cognitive economy – one’s system of beliefs, desires, goals, and other intentional states. Such a practice involves the modification of one’s beliefs – often in the light of new information – so as to eliminate inconsistencies among one’s current commitments. But the capacity to do so presupposes a capacity to recognize inconsistencies, which in turn presupposes a capacity to engage in inferences that draw out the consequences of one’s current beliefs. According to Brandom, all of these cognitive capacities are aspects of the practice made explicit by use of modal vocabulary; and thus are implicit in the practice of rational rectification.
But there is more to rational rectification than this. Rational rectification requires more of us than the mere capacity to recognize inconsistency – and whatever inferential capacities this, in turn, presupposes. Among other things, this alone would be compatible with continuing to endorse inconsistent commitments. In addition to this, rational rectification requires that one take oneself to have an obligation to eliminate such inconsistencies – to structure one’s commitments in such a way as not to violate facts about the modal structure of the world. And according to Brandom, the practice of having one’s commitments conform to such constraints are amongst those made explicit by our use of normative vocabulary.
It is time to bring all the pieces together. What we have said so far on the topic intentionality is quite impressionistic; and indeed most of Brandom’s own discussion is similarly gestural. Nevertheless, in accord with his practice elsewhere in the book, when Brandom seeks to spell out his views precisely, he presents a meaning-use diagram. In the case of intentionality, the diagram that he uses to distill his views is reproduced as Fig. 5.
Although there is a lot going on here, we restrict ourselves to the following observations. First, the diagram aims to depict a kind of practice – indeed any practice of using an autonomous vocabulary – as an object for which intentional vocabulary is VP-sufficient – i.e., either made explicit by or describable by intentional vocabulary. Second, the diagram represents any autonomous discursive practice as incorporating two essential components: a practice made explicit by modal vocabulary and a practice made explicit by normative vocabulary.Footnote 31 Thus the diagram represents mastery of such practices as necessary for being a language-user at all.
Finally, the diagram represents anyone who can engage in these two practices as being capable, in principle, of deploying the vocabularies that make these practices explicit – i.e., modal and normative vocabulary. In other words, the diagram represents the fact that Brandom’s account of intentionality – in conjunction with the earlier discussed claims regarding modal and normative vocabulary – entails the modal and normative Kant-Sellars theses. This is important for present purposes, because by tying his account of intentionality to his previous claims about modality and normativity, Brandom provides a partial explication of the nature of the practices on which intentionality depends. As we noted earlier, the real challenge for a pragmatic account of intentionality of the sort advocated by Brandom is to explicate the word-world relations that is relevant to intentionality; and with respect to this task, the primary significance of the Kant-Sellars theses is that they permit Brandom is specify central aspects of the relevant practices. Brandom is thus relying on his earlier discussion of modality and normativity in order to flesh out his account of intentionality.
In view of our earlier discussion, one problem with Brandom’s proposed account of intentionality should be clear. As we have already seen, modal vocabulary is not universal LX. So, there might be a language user that cannot already do everything needed to be able to use modal vocabulary; and if that is right, then the account of semantic intentionality, as represented in Fig. 5, obviously cannot be right.
How deep does this objection go? To see the significance of the objection, we need to rehearse some aspects of our earlier discussion, and introduce some features of Brandom’s theory of content. First, recall that we showed earlier that the practice of counterfactually robust inference is not PP-sufficient for the practice of using modal vocabulary. But this is just to deny that arrow 1 in Fig. 5 holds. In other words, the practice labeled Pobj-modal in Fig. 5 is not PP-sufficient for the practice of using modal vocabulary; and this means that Brandom has failed to characterize adequately the nature of the practice that, on his view, mediates the intentionality relation. In particular, if our arguments from Section 5 are correct, then the aspect of the practice that is supposed to explicate the “world pole” of the intentionality relation has not have been accurately characterized.
Of course, this is not the end of the matter. Following the dialectic set out in Section 5, Brandom might seek to preserve the connection between Pobj-modal and the practices sufficient for the use of modal vocabulary by enriching Pobj-modal. Specifically, he might add the assumption that Pobj-modal is constituted, in part, by the ability to make assertions. If such a maneuver is pursued, then our earlier objection to arrow 1 fails. But now – just as before – the problem crops up elsewhere. Though arrow 2 holds more or less by definition, arrow 3 fails. That is, as argued earlier, we no longer have any reason to suppose that modal vocabulary is VP-sufficient for Pobj-modal. In which case, modal vocabulary still fails to be universal LX, and the modal Kant-Sellars Thesis is false. As a consequence, even if one pursues this escape route, Brandom’s pragmatic theory of intentionality, at least as represented by Fig. 5, is false.
One might, however, accept our criticism and yet insist that the objection is easily met so as to preserve the spirit, though not the letter, of Brandom’s account of intentionality. Specifically, it seems eminently reasonable to ask why the account of intentionality as such should require that modal vocabulary be VP-sufficient for Pobj-modal. Clearly, on anything like Brandom’s view, Pobj-modal should be PP-sufficient for Pmodal. Absent this connection Brandom would, by his own lights, have failed to explicate the “world pole” of the intentionality relation. Further, on anything like Brandom’s account, a relation of PV-sufficiency must hold between Pmodal and Vmodal. After all, on his view, the existence of such a connection obtains more or less by definition. In contrast, it is utterly unclear why the account of intentionality requires that Vmodal be VP-sufficient for Pobj-modal and, hence, that modal vocabulary be universal LX. To the extent that we concerned only with the account of intentionality, then, such a commitment seems entirely optional.
There is a sense in which we think that this response is entirely correct. Brandom’s approach to intentionality, need not – and indeed should not – demand that Vmodal be VP-sufficient for Pobj-modal. But we think that the assumption that it should is deeply engrained in Brandom’s views about semantics and, hence, that preserving the Brandonian approach to intentionality requires that one give up a deep-seated commitment of Brandom’s philosophical worldview, sometimes called expressive equilibrium. In our experience, the issue is not easily appreciated; and so we propose to creep up on the relevant assumption by first considering another assumption – expressive completeness – also deep-seated in Brandom’s philosophy, that does not suffice to mandate the commitment to arrow 3, but that does motivate the endorsement of expressive equilibrium
Brandom is, of course, a semantic pragmatist, and so his theory of content takes the form of a theory of those discursive practices that are adequate to confer semantic content on the expressions involved. Expressive completeness is invoked as a condition of adequacy on such a theory. To a first approximation, expressive completeness requires that a theory of discursive practices should, in principle, be available to its practitioners. In other words, an adequate semantic theory should explain how linguistic expressions acquire their content in such a way that practitioners are, in principle, able to use the theory in order to explain the practices in which they participate. A corollary of this – one that is central to appreciating the significance of this constraint – is that such a theory will apply to itself. That is, an expressively complete semantic theory will correctly explain how the very linguistic expressions that comprise the theory acquire their meanings.
Why endorse such a commitment? Though we will not explore the issue in detail here, the central motivation is to avoid the sorts of self-refutation worries that notoriously confront various historically influential theories of meaning, most notably early versions of verificationism, but also Wittgenstein’s Tractarian account of meaning.Footnote 32 Clearly, an expressively complete theory will not confront this problem since such theories are self-applicable – that is, they provide a meaning theory for the expressions in the theory itself.
Be that as it may, the assumption of expressive completeness alone fails to explain Brandom’s commitment to the VP-sufficiency of Vmodal for Pobj-modal. This is because the mere commitment to expressive completeness very obviously does not demand that modal expressions make explicit the practices on which they depend. On the face of it, all that is required is that the theory not be self-defeating – that the theory possess the resources to explain how the expressions of the theory themselves acquire their meanings. What does require VP-sufficiency, however, is the strategy that Brandom adopts in order to ensure the expressive completeness of his preferred approach to semantic theorizing – what he calls expressive equilibrium.
To a first approximation, the requirement for expressive equilibrium is the requirement that the broad category of logical vocabulary – which, for Brandom, includes traditional logical connectives and quantifiers, but also normative and modal notions – makes explicit not only those norms that confer content on non-logical vocabulary, but those norms that confer content on logical vocabulary as well. To see the significance of this notion for Brandom’s semantic project we need to rehearse briefly the overall structure of the approach to semantic properties that Brandom adopts. Roughly put, he advocates a two-part approach:
The basic theory: This describes what might be called basic discursive practices – i.e. those involving no logical vocabulary. This theory will, of course, use logical vocabulary, but it will not explain how such expressions get their meanings.
The extended theory: This describes the extended discursive practices in which language-users engage – i.e., those involving logical vocabulary. More specifically, it does so by showing how basic discursive practices can be extended by the introduction of logical vocabulary.
Now as we have seen, the overall theory – comprised of both the basic and extended theories – needs to be expressively complete. Failing that, one risks self-defeat problems of sort that stymied verificationism. Yet clearly, participants in the basic practice lack the resources to state the basic theory because they posses no logical vocabulary at all. In which case, the basic theory is self-defeating. That is where the extended theory enters the picture. This theory explains how to introduce logical vocabulary – including modal and normative vocabulary – into the basic practice. So, participants in the extended practice – i.e., the practice described by the extended theory – possess the vocabulary required to state the basic theory. But now the problem recurs. In order to ensure expressive completeness of the extended theory, such practitioners must not merely have access to the basic theory, but to the extended theory as well. Otherwise, we (once more) risk regress.
Brandom’s proposed solution is that logical vocabulary is in expressive equilibrium —that it makes explicit not only those norms that confer content on non-logical vocabulary, but those norms that confer content on logical vocabulary as well. And in the parlance of Between Saying and Doing, this is just the claim that logical vocabulary, broadly construed to include normative and modal vocabulary, makes explicit the very practices that one needs to master in order to possess logical vocabulary itself. In short: the notion of universal LX vocabulary is a way of making more precise the demand for expressive equilibrium: a demand that Brandom thinks is central to an account of meaning that avoids self-defeat.
We are now in a position to summarize our major concerns about Brandom’s view of intentionality. First, as noted earlier, our objections to the modal Kant-Sellars thesis in Section 5 show that the view as stated is false. Second, there is no reason internal to the theory of intentionality to assume that modal vocabulary is universal LX. At most the theory of intentionality per se requires arrows 1 and 2 – not arrow 3 – because it is the first two connections that are involved in specifying the discursive practices relevant to explicating the “world pole” of the intentionality relation.
Third, arrow 3 is required if one seeks to endorse both the account of intentionality and the strategy for ensuring expressive completeness. But this is just because it is required by the theory of expressive completeness. In other words, arrow 3 comes not from the theory of intentionality but from Brandom’s strategy for achieving expressive completeness for his pragmatist theory of meaning.
Finally, once one sees the problem, it should be clear that arrow 3 is demanded by expressive equilibrium, irrespective of what account of intentionality Brandom endorses. That is, once one adopts expressive equilibrium as a way of attaining expressive completeness, it needs to be the case that modal vocabulary is VP-sufficient for Pobj-modal; and more generally, that logical vocabulary, broadly construed, is VP-sufficient for those practices needed to master exactly that logical vocabulary. What our arguments in the earlier sections of this paper show is that this strategy cannot succeed.