Boghossian: Naturalising content

Naturalising Content

Fodor: intentionality is not an irreducible property. If the semantic and intentional are real properties of things, it must be because of identity with/supervenience on properties which are themselves neither intentional, nor semantic.

'If aboutness is real, it must really be about something else'.

.Fodor's commitment to supervenience requires a naturalised theory of meaning. That is, he must explain how meaning is something both physical, and sufficient for the instantiation of an intentional property.

Informational theories and Type 1 Situations

Language of thought: the meaning of a mental symbol is determined by the information it carries. 

This account faces the disjunction problem, however.

Framing the problem in terms of language of thought: every property that can cause a symbol token is by definition in the extension of that symbol.

A situation in which nothing but the referent of a symbol can cause it's tokening is a type 1 situation. Theories which specify meaning in terms of these situations are 'Type 1 theories'.

Teleological Type 1 Theories

There is a normal set of conditions specified by evolutionary biology, under which cognitive mechanisms function as supposed to. Type 1 situations = normal conditions.

Fodor's objections:

1. Rests on an incredible conception of evolutionary selection: assumes that cognitive mechanisms are always selected for the purpose of tracking the truth.

2. Root idea on which it rests ('when things go right' S will be tokened only in application to its referent) seems false. Fodor's counterexample: phenomenon of one thought causing another. In thinking about horses, I might end up thinking about cows. A horse thought causing a cow thought. But, nothing has gone wrong teleologically, so hard to define type 1 situations in terms of 'normal conditions'.

Fodor's Theory of Asymmetric Dependence

What is wrong with teleological theories is that tokens of a symbol can only have one kind of cause: the kind that fixes meaning. 

However, this underestimates the robustness of meaning.

'Cow' tokens can be caused in all kinds of ways, but still mean cow. What is needed is to pick out semantically relevant causal relations, without idealising from contexts of causal homgeneity.

His theory: 

(i) cows cause “cow” tokens

(ii) For all X other than cows that cause “cow” tokens: If cows didn’t cause “cow” tokens, then Xs wouldn’t either. But even if Xs didn’t cause “cow” tokens, cows would still cause “cow” tokens.

That is: break cows ⇒ “cow” tokens and you break Xs ⇒ “cow” tokens, but not vice versa.

That is: the Xs ⇒ “cow” tokens link is asymmetrically dependent on the cows ⇒ “cow” tokens link

So, S means P if: if P ceases being able to cause S tokens, then every other cause of S is also rendered impotent. But any non P could cease, without affecting P's ability to do so.

Boghossian's Criticism

Boghossian argues that this theory still commits itself to the existence of type 1 theories.

Favourable reading of Fodor: if symbol S possesses asymmetric dependence base P, then there exists a…