Wednesday, November 28, 2007

On inference relations and constituents of representations

Today Pepa Toribio gave a thoughtful and dense talk on nonconceptualism, and the very beginning of it she told us that
"For to contentful mental states to be inferentially related, they ought to have at least one constituent in common"
That puzzled me, because it seems easy to give examples of inferences in which none of the premises share a constituent with the conclusion. Take for example the inference from "b is red" to "There are non-blue things". The inference works because "red things are not blue" is analitically true (though not being logically true, or true in virtue of the sintax alone.) Does anyone else shares my feelings?


Oscar Cabaco said...

A better example is the following: the inference from "Pa" to "Something exists".

In fact, I think that several examples showing what i defend were discussed in the last Barcelona Workshop (BW5). But I don't remember the details of the discussion.

Dan López de Sa said...

Yep, that sounds right to me.

Manolo Martínez said...

Even without countenancing analytic truths:

from p and not-p infer q

from p infer q or not q

Dan López de Sa said...

I was trying to remember: did the original (controverted) contention play a substantive role in Pepa's argumentation?

Pepa Toribio said...

When a sentence in a text is taken out of its context, all kind of funny things happen. The assertion that puzzled Oscar was made while discussing Evans’ Generality Constraint as a necessary condition for thought.

The idea behind it is that (genuine) thinking requires the exercise of certain capacities. In particular, I wanted to capture the idea that genuine thinking must draw on capacities that meet some compositionality principle, i.e., that “if a subject can be credited with the thought that a is F, then he must have the conceptual resources for entertaining the thought that a is C, for every property of being C of which he has a conception’ (Evans, 1982, 104). In other words, a subject can entertain the thought that Lolo (my cat) is white, only if she can also entertain the thought that Lolo has some other property such as e.g. being fat, of which she has a concept. Evans also states the constraint in a slightly different fashion. According to this other version, a subject can be credited with the thought that a particular object —such as Lolo— has a property —such as being white—, only if she can also think about other objects —let’s say this piece of paper on my desk— as having this very same property. The Generality Constraint is usually understood as the union of these two theses and thus reads as follows: the attribution to a subject of contentful states of the form a is F and b is G commits us to the idea that that system should also be able to represent a as G or b as F (Evans, 1982, p. 104, ft. 21).

The constituents in my (maybe infelicitously) sentence were the ‘a’s and ‘b’s and ‘F’s and ‘G’s of the above quote. Its import thus was epistemic. So, to take Oscar’s original example. He says: “Take for example the inference from "b is red" to "There are non-blue things". The inference works because "red things are not blue" is analytically true” and he takes this to be a counterexample to the original claim. However, using this example, my point can be put like this. Nobody could be credited with the thoughts that ‘a is red’ and ‘b is blue’ unless she can also entertain the thought that ‘a is blue’ or ‘b is red’.
I wasn’t talking about what follows from what, or about valid inferences, but about the necessary conditions for crediting someone with a thought in the first place.
Why talk about ‘inferences having constituents in common’? Because the key feature for taking some mental state’s content to figure in a subject’s rational organisation of her behaviour (believing and acting) is that there is some kind of argument concerning the subject’s other mental states in which such content plays a role, an argument whose validity cashes in on the subject’s capacity to keep that content fixed through premises and conclusion, i.e., the subject’s ability to redeploy a particular content in some other thoughts.
Of course, that this is properly understood is essential to my (and any) argument on conceptualism vs. nonconceptualism.