Understanding as Inferential Ability
3 Normative Inferentialism
The basic claim of inferentialism is that meaning (i.e. conceptual content) should not be analysed in terms of reference but in terms of inference. The fact that what we say and do, that a statement means something is a fact that we should try to understand not in terms of the relation between a statement and a state of affairs, but in terms of inferential relations between a statement and other statements. However, although inferentialism is first and foremost a semantic theory, concerned with how what we say has meaning and content, Brandom proposes to ground semantics in pragmatics. Pragmatics is concerned with what one is doing when using linguistic expressions, which means Brandom analyses conceptual content and understanding in terms of what one is doing when using particular terms and expressions. This order of explanation thus builds upon the later Wittgenstein, in which meaning is analysed in terms of use. To clarify Brandom’s particular version of inferentialism, we need to contrast it with two other approaches to semantic content: (i) semantic representationalism, and (ii) causal inferentialism.
First, Brandom opposes what he takes to be the orthodox representationalist approach to semantics. Representationalism attempts to account for the semantic content of linguistic expressions by invoking various forms of word-object or word-world relations. As a very basic characterisation, the representationalist picture holds that words are contentful/meaningful insofar as they represent, or stand for, something (on the basis of suitably worked out truth- and/or reference-conditions). Whole sentences are then explained as being composed out of the meanings of the words they comprise, which then can be used to explain which inferences are valid. Furthermore, a representationalist semantics is often, but not necessarily, coupled with the idea that ‘the mind’s representational capacities precede, cause and explain linguistic meaning and what one is doing in using language’ (Maher 2012:
78).
Semantic representationalism can be thought of as the conjunction of two assumptions (Wanderer 2008: 96). The first claims the explanatory priority of semantics over pragmatics in providing a philosophical account of language. The assumption here is that we first need to
secure our semantic theory before moving on to understand our use of expressions. The second assumption holds that semantic notions like truth and reference are explanatory primitives.
The representationalist approach thus adheres to a particular order of explanation with regard to language: we (1) begin with an account of truth and reference as the basic semantic concepts; (2) proceed from there to explain meaning and inference in terms of truth and reference; before finally moving to (3) the pragmatic level, an account of what we are capable of doing with our semantics in place. This implies a division of labour, such that semantics and pragmatics can be treated as conceptually independent from one another. While philosophers might agree in principle that a full-blown theory of language will eventually need both, in practice, they are approached with relative autonomy.
The basic move of Brandom’s inferentialism is to reverse this order of explanation. Like the representationalist approach, it can be presented as proceeding in three stages. The first stage of Brandom’s alternative begins in an account of our linguistic practice: an account of what we are doing when using expressions. The specific kind of use or activity that Brandom primarily focuses on is the activity of asserting which I shall discuss in greater detail below.5 With his account of linguistic practice in place, the next stage develops our understanding of semantic content in terms of its inferential role. The aim of the third stage is to offer an account of the representational dimension of our language in terms of the previous two stages. Like the representationalist order of explanation, each stage thus builds on the previous one.
Wanderer (2008: 97) schematises Brandom’s general strategy as the following:
B1 An account of linguistic practice, incorporating the speech act of asserting
B2 An account of the semantic content of linguistic items in terms of the inferential role of such linguistic items. The concept of inferential role is explained in terms of such linguistic items being caught up in the linguistic practice outlined in B1.
B3 An account of the representational dimension of semantic content, including the use of terms such as “true” and “refers”, in terms of the notion of inferential role outlined in B2.
Thus, Brandom’s inferentialism challenges and rejects both assumptions that motivate semantic representationalism: (1) the idea that semantics precedes pragmatics in the order of
5 Brandom writes, ‘The explanatory strategy pursued here is to begin with an account of social practices, identify the particular structure they must exhibit in order to qualify as specifically linguistic practice, and then consider what different sorts of semantic content those practices can confer on states, performances and expressions caught up in them in suitable ways’ (MIE: xiii).
explanation, and (2) that truth and reference are semantic primitives. Both rejections are undoubtedly controversial and Brandom is by no means the first to argue for the priority of pragmatics over semantics. However, it is relevant for the following discussion that we recognise the constraints Brandom’s commitment to this order of explanation place on his account. The key point is that for each stage to do explanatory work - that is, to explain the notions and concepts introduced in later stages – Brandom cannot appeal to concepts that occur in later stages on pain of begging the question. For example, as we shall see, Brandom cannot and does not appeal to notions of meaning and content to explain linguistic practice.
Similarly, he cannot and does not appeal to notions of truth to explain valid inference. In the next section, I shall only sketch out the basic contours of Brandom’s first two stages, B1 and B2, which provide the main ingredients for my approach to the concept of understanding. This does not mean that I take the third stage B3 to be irrelevant, but that it would need more space to work out adequately.
Setting out these starting points for Brandom’s account helps to clarify the inferentialist approach to conceptual understanding. If Brandom’s order of explanation for conceptual content can be treated as an analogue for an account of understanding, then it would suggest that we treat understanding first in terms of what we do in our practices, rather than making sense of understanding primarily in representationalist terms. Another implication is that the representational capacity of understanding should be treated as dependent on this more basic pragmatic understanding, a point that we saw in Heidegger’s conception of understanding. To reiterate the point, the basic pragmatist commitment underlying Brandom’s inferentialism is to explain knowing that in terms of knowing how (see MIE: 135-6).
The second contrast between Brandom’s version of inferentialism and others concerns the notion of inference itself. An inferentialist semantics holds that semantic content is a function of inferential role: content is conferred on our expressions in virtue of its role in inference and reasoning, rather than in terms of its origin in experience (as an empiricist might hold). Stated as such, Brandom’s inferentialism bears resemblance to what is known as inferential role semantics or more generally, conceptual role semantics,6 which claims, in the words of Ned Block, that the ‘meaning of a representation is the role of that representation in the cognitive life of an agent, e.g., in perception, thought and decision-making’ (Block 1998).
It is necessary, however, to distinguish Brandom’s approach from these other varieties of inferentialism to avoid confusing them. Although both stress the idea that meaning and content be considered as a function of inferential role, the crucial difference between them lies in how they make sense of this role. Consider Boghossian’s characterisation of inferential role semantics in the following:
6 For example, as discussed in Fodor and Lepore (1992) and Boghossian (1993).
Let’s suppose that we think in a language of thought and that there are causal facts of the following form: the appearance in O’s belief box of a sentence S1 has a tendency to cause the appearance therein of a sentence S2 but not S3. Ignoring many complications, we may describe this sort of fact as consisting in O’s disposition to infer from S1 to S2, but not to S3. Let’s call the totality of the inferences to which a sentence is capable of contributing, its total inferential role. A subsentential constituent’s total inferential role can then be defined accordingly, as consisting in the contribution it makes to the total inferential role of the sentences in which it appears. (Boghossian 1993: 73)
On Boghossian’s construal, the inferential role that confers content on an expression is characterised first and foremost in causal or dispositional terms. The focus is accordingly on what inferences the subject actually makes or is disposed to make. By contrast, Brandom’s inferentialism primarily concentrates on the inferential rules and norms that govern the inferences made by the subject, rather than the actual inferences a subject makes. That is, Brandom’s approach shifts attention to the normative proprieties that govern our inferences, rather than being concerned with the actual inferences themselves. As such, the relevant distinction here is between causal inferentialism and Brandom’s normative inferentialism (Peregrin 2014: 8).
This marks a significant contrast with the kind of account of inferential ability proposed by Wilkenfeld or Newman, discussed in the previous section. While the ability to infer is something that one does on the basis of certain cognitive abilities, inferential rules are not themselves states or events in the causal order. Brandom’s normative variety of inferentialism is committed to an account of the content of our thoughts and expressions in terms of what we ought to do if what we say and think is to have determinate semantic content. Normative inferentialism holds that the ability to grasp conceptual content, and thus to count as understanding, is a matter of the proprieties and norms of inference, rather than in terms of the dispositions of subjects to infer one thing rather than another. For Brandom, ‘the content of any concept is the conceptual content that it is only in virtue of playing a specific normative role or having a certain normative status within the logical space of reasons’ (O’Shea 2010:
462). In the same sense that something only counts as a chess piece in virtue of being subject to the rules of chess, for the normative inferentialist, meaning and content are only understood in relation to such inferential rules. Meaning, content and understanding are thus dependent on being situated within a rule-governed, inferentially articulated space. Thus, Brandom’s account of inferential content should be understood in the light of Saul Kripke’s interpretation of Wittgenstein’s rule-following problem, which concludes that an adequate account of meaning and content must not simply describe actual behaviour or dispositions to behave.
Rather, ‘the relation of meaning and intention to future action is normative and not descriptive’
(Kripke 1981: 37).
To sum up: in this section I have introduced two distinctions to clarify the nature of Brandom’s approach to conceptual understanding. The first distinction was between representationalism, which understands conceptual content in terms of word-world relations, and inferentialism, which argues that conceptual content acquires meaning through its use in reasoning. The second distinction was between causal inferentialism, which explicates inferential ability in terms of the inferences the subject actually makes or is disposed to make, and normative inferentialism, which focuses on the rules and norms that govern the appropriateness and legitimacy of those inferences. For Brandom, conceptual understanding is less about the particular cognitive abilities one may or may not possess, and more about the way in which we are responsive to norms. Conceptual understanding, in this respect, is more about being able to appreciate the distinction between correct or incorrect ways in which we take the world to be and to be responsive to the norms governing our inferential practices.