16. Inferentialism

Martín Abreu Zavaleta

June 19, 2014

1 A crash course in conceptual-role semantics

Perhaps the most widespread view about the content of mental states is that there is something, say, a proposition, that they have as their content, and this content represents something or other. We have already seen some attempts to reduce that content to biological functions, but this is not the only kind of reduction that has been attempted, and indeed the reduction need not be a naturalistic one.

Some people think that what determines the content of a state or expression is the conceptual role it plays. This kind of view is usually called conceptual role semantics. It may also be properly called inferentialism, since the conceptual role of a given item is supposed to be given by the kind of role it plays in certain set of inferences.

The easiest way of presenting this view is by explaining how this way of determining content applies to something as simple as logical constants. Logical constants are expressions of logic like ‘and’, ‘or’ or ‘not’. In these cases, it seems easy to define them in terms of the roles they play. For instance, there are two rules of inference that fully characterize the inferential behavior of ‘and’. These are the rules for introducing it and the rules for eliminating it:

From premises p, q, infer p and q.
From premise p and q, infer p; from premise p and q, infer q.

The idea is that the meaning (or semantic content) of ‘and’ is whatever satisfies these two rules of inference. Similar definitions can be given for the rest of the logical connectives.

Inferentialism as we will understand it concerns the content of mental states. In general, for an attitude to have as its content a particular proposition just is for it to play a particular role in cognition, and to grasp that conceptual content is to be prepared to make certain inferential transitions. For instance, for someone to grasp the proposition that something is a vixen is for that person to be disposed to infer, whenever she thinks that x is a vixen, that x is a mammal, or that x is a female fox, and so on. This might remind you of functionalism, but our discussion now is purely about content, and not about mental states in general.

Not all inferentialists are naturalists. For instance, Brandom thinks that the roles that determine the content of beliefs are normative roles; in particular, the roles that determine content are something like rules that one must obey if he is to be able to communicate with other people.

2 Brandom’s assumptions

Brandom wants to distinguish between two kinds of mental qualities. The first, which he thinks we share with animals, is sentience: a kind of awareness in the sense of being awake. Sentience doesn’t amount to the reliable responses we have when we encounter certain objects, since such dispositions could be had by plants or thermostats, but we wouldn’t attribute mentality to the latter.

The second mental quality is sapience, which concerns understanding or intelligence:

One is treating something as sapient insofar as one explains its behavior by attributing to it intentional states such as belief and desire and constituting reasons for that behavior. Sapients act as though reasons matter to them. They are rational agents in the sense that their behavior can be made intelligible, at least sometimes, by attributing to them the capacity to make practical inferences concerning how to get what they want, and theoretical inferences concerning what follows from what. (p. 509)

In addition to this feature, things with sapience are the kinds of things that can believe, act, desire, and have other mental states whose contents can be true or false.

Like most people, Brandom believes that the content of beliefs, desires and the like is a proposition. Propositions are also the things we denote by means of ‘that’ clauses. However, unlike most philosophers, Brandom things that there are two independent things a proposition does: first, say something, and second be about something. It is only the latter that he takes to be a properly representational dimension of propositions and thoughts.

Given that he thinks that saying something, and that thing being about something, are different features, he must explain how it is that these two sides of content interact with each other. In particular, he wants to explain why it is that any state or utterance with propositional content also has representational content. His answers will be along the following lines: a state or utterance manage to represent because of the role it plays in the assessment of how these states or utterances can serve as reasons in a social context.

Unlike other philosophers, Brandom doesn’t take the notion of truth as the main notion in terms of which he will explain representation. Instead, he takes the notion of an inference, and define the things that have propositional contents as the things that can serve as a premise or a conclusion in an inference. He will define truth as whatever is preserved by good inferences.

Roughly, Brandom’s idea is that beliefs have the propositional content that they actually have because of the role they play in inferences and, in turn in the explanation of our behavior as rational agents. In the case of assertions, they owe their propositional content to the role they play in “the game of giving and receiving reasons”.

Brandom thinks that there is some sort of normativity to representationality. He thinks that in having a belief or making an assertion, one incurs a certain commitment, and one is entitled to making such commitment (and thus to make the assertion) to the extent that he or she is capable of justifying them. This need not worry us too much, so it’s ok if it’s not very clear.

To understand the content of a sentence or a thought is to know which kind of commitments one would incur by having the thought, and what would entitle one to have those commitments. Brandom uses his metaphor of scorekeeping in a game to make this point.

3 Carrying out the inferentialist program

Brandom only says this in passing, but he doesn’t really think that there are beliefs. He only thinks that there are belief ascriptions. He also doesn’t think that the notion of representation carves very deep. So what he does is to explain the function of the locutions that we would ordinarily take to indicate what a particular proposition represents.

In particular, he thinks that de re ascriptions of propositional attitudes are the prime explicitly representational locution of natural language. They explicitly state what a proposition is about. Let’s start by distinguishing de re from de dicto ascriptions of attitudes.

Consider the following two sentences:

Arya believes that the inventor of the zipper did not invent the zipper.
Arya believes of the inventor of the zipper that he did not invent the zipper.

If (1) is true, it would seem that Arya is an irrational person, but if (2) is true, she might just have the mistaken belief that the particular person who in fact invented the zipper, didn’t invent the zipper. We say that (1) is a de dicto (from what is said) attitude ascription, and that (2) is de re (about the thing).

In general, we can distinguish de re from de dicto readings of an attitude ascription by the fact that, in de re ascriptions, we can substitute coreferential terms salva veritate. We have seen that people like Chisholm and Brentano take intensional inexistence to be the distinguishing mark of the mental, so in some sense de existence of de re attitudes seems to be at odds with their claim. Still, it’s not clear whether or not the existence of these attitudes falsifies Brentano’s thesis.

Brandom stresses the importance of accounting for de re ascriptions:

What are we doing when we make claims about what someone is talking or thinking about? How must vocabulary be used in order for it to deserve to count as expressing such de re ascriptions? Answering that question in a way that does not itself employ representational vocabulary in specifying that use is then a way of coming to understand representational relations in nonrepresentational terms.

Keep in mind that we should understand discursive practice in terms of the adoption of practical attitudes by which interlocutors keep score on each other’s commitments.

Brandom distinguishes between two different ways of undertaking a commitment—undertaking a commitment, by the way, allows others to attribute the commitment. One may acknowledge a commitment by being disposed to avow it by an overt assertion. Another way to acknowledge a commitment is by employing it as a premise in one’s reasoning.

The second way of undertaking a commitment is consequential, as Brandom puts it. One undertakes a commitment consequentially if the commitment is a conclusion one is committed to as a consequence entailed by the commitments one acknowledges.

Brandom thinks that the function of expressions like ‘of’, or what we would ordinarily take to be descriptions of what something represents, is to aid in communication. He explains this by giving what he takes to be the conceptual role of these expressions in belief ascriptions.

He starts by distinguishing two roles that a given expression can play in a belief ascription: it can be the explicit attribution of a commitment to the subject of the ascription, or it can express a commitment on the side of the person who is attributing the belief. Brandom illustrates with the following case. Consider the following conversation between a prosecutor and his opponent:

The defense attorney believes a pathological liar is a trustworthy witness
Not so; what I believe is that the man who just testified is a trustworthy witness.
Exactly, and I have presented evidence that ought to convince anyone that the man who just testified is a pathological liar.

The disagreement is clearly whether a particular person is a liar, not whether a liar is a trustworthy witness. Compare (3) with

The defense attorney claims of a pathological liar that he is a trustworthy witness.

Brandom claims that

the expressive function of de re ascriptions of propositional attitude is to make explicit which aspects of what is said express commitments that are being attributed and which express commitments that are undertaken.

So how should we characterize the functional role of de re ascriptions?

Suppose that according to A’s scorekeeping commitments, B acknowledges commitment to the claim Φ(t). Then A can make this attribution of commitment explicit in the form of a claim by saying ‘B claims that Φ(t). If in addition A acknowledges commitment to the identity t=t’, then whether or not A takes it that B would acknowledge that commitment, A can also characterize the content of the commitment ascribed to B by saying ‘B claims of t’ that Φ (it). Again, the question is whose substitutional commitments one is permitted to appeal to in specifying the consequence someone is committed to by acknowledging a particular doxastic commitment. (p. 517)

What is expressed by de re specifications of the contents of beliefs is crucial for communication. For it is only if I am ready to infer something from your characterization of a given thing that I can be said to understand it. “Extracting information from the remarks of others requires grasping what is expressed when one offers de re characterizations of the contents of their beliefs... It is to grasp the representational content of their claims. The point I have been making is that doing this is just mastering the social dimension of their inferential articulation." (p. 518)