Teleosemantics and Productivity
This paper of mine — of which I had posted previous drafts here and here — has been accepted for publication in Philosophical Psychology. The first link is to the accepted version. The abstract:
There has been much discussion of so-called teleosemantic approaches to the naturalisation of
content. Such discussion, though, has been largely confined to simple, innate mental states with
contents such as There is a fly here. Even assuming we can solve the issues that crop up at this
stage, an account of the content of human mental states will not get too far without an account of
productivity: the ability to entertain indefinitely many thoughts.
The best-known teleosemantic theory, Millikan’s biosemantics, offers an account of productivity in
thought. This paper raises a basic worry about this account: that the use of mapping functions in the
theory is unacceptable from a naturalistic point of view.
It’d be cool to know what you guys think about it!
Naturalizing Stuff 
Very interesting paper (I specially like section 2). I have two questions.
First, I do not see the connection between the first part and the main dsicussion of paper. It seems to me that if you are right and Mapping Functions cannot account for the productiviy of thought, then this is so even if we put concept-like representations first.
In the last paragraph you seem to be saying that the first and second part of the paper are related because there is a tension between top-down content determination (from thoughts to concepts) and bottom-up (from concepts o thoughts using compositionality). I think you are right that there is such a tension, but your argument does not seem to rely on it; there is no problem in concept-like representations deriving their content from thought-like representations. So there might be no problem in thought-like representations deriving their content from concept-like representations.
So I do not see why section 2 is relevant.
My second question concerns the main argument of the paper. The worry is that the problem you are pointing out may be caused by your own formulation of the theory. As L3 shows, you think Millikan is commited to the existence of a system of signs with a certain mapping function onto world affairs that is abstracted from causally grounded relations. But there might be different ways of cashing out the theory that do not involve L3.
Here is another way to put things that may avoid your problem: The Mapping function of a sign S is determined by the etiological function (EF) of the mechanism that produces S. The Etiological Function of the mechanism is something like ‘produce S when X’ (where there is a mapping function between S and X) So the mapping function is determined by the Eti-Function of the mechanism. Now, the Eti-Function is something like the least detailed effect that explains why the system was selected for; the important bit here is “least detailed”; there are many further effects or compicated descriptions that are not included in the function. My suggestion is to use the ‘least detailed’ principle to avoid quus-like rules. In other words, the strategy I am proposing is to rule out PQR and PHR for the same reasons you rule out complex effects as the function of a mechanism. As I told the story, there is no process of abstraction (L3), but just effects, functions and selection processes.
Hi, Marc, thanks for your reply 🙂 Let me go point by point:
>First, I do not see the connection between the first >part and the main dsicussion of paper. It seems to me >that if you are right and Mapping Functions cannot >account for the productiviy of thought, then this is >so even if we put concept-like representations first.
Well, this depends on the content-attributing recipes for these concept-like representations. The idea would be that the content of “There is food in Mars in the distant future” is to be understood as a structured proposition, made of “Mars”, “food”, etc. and the way in which they are composed. Admittedly, I do not say much about how this other story would go; but the paper was already too long. Hopefully I’ll write the sequel soon 🙂
>So there might be no problem in thought-like >representations deriving their content from >concept-like representations.
I certainly hope that there isn’t! The thing is that Millikan does not use this bottom-up route –incorrectly, I argue. And, while you right and she does use top-down determination to account for the content of concepts, these concepts are not subsequently used, in her account to provide (bottom-up) the content of new thoughts.
Regarding your proposal on behalf of Millikan, I don’t know how ‘least detailed’ — if this notion is non intentional, that is — is supposed to prefer plus to quus. Could you flesh out a bit the suggestion?
Cheers
M