My unpublished 1993 paper "A Computational Foundation for the Study of Cognition" is to be the subject of a special issue in the Journal of Cognitive Science. This is the best statement of my views on computation and cognition, and I still agree with most of it. I'm not sure why I never published it. It received an encouraging revise and resubmit from Behavioral and Brain Sciences in 1994, but then I got caught up with finishing The Conscious Mind and then moving to Santa Cruz and never resubmitted it. (The same thing happened with "The Components of Content" at Mind.) Maybe I took the view that web publication was nearly as good as journal publication. That view turns out to be overoptimistic, but the paper has been cited quite a lot over the years and I'm still quite fond of it. So I was pleased when Gualtiero Piccinini, who has himself done important work in this area, suggested this special issue. A number of the leading people in this area have already agreed to comment (there's also a call for submissions), and I'll write a reply.
Other publication news: In addition to this symposium and the Journal of Consciousness Studies symposium on the singularity, there will be symposia on The Character of Consciousness in Analysis Reviews and Philosophical Studies. Also, "Verbal Disputes" is forthcoming in the Philosophical Review, "Revisability and Conceptual Change in 'Two Dogmas of Empiricism'" is forthcoming in the Journal of Philosophy, "Frege's Puzzle and the Objects of Credence" is forthcoming in Mind (I think), and there are previously mentioned forthcoming papers in Nous and Analysis. As someone who's often been bad about journal publication (see above), it's nice to have a few things coming out. In addition, "The Nature of Epistemic Space" will finally appear in the Egan/Weatherson Epistemic Modality volume any day now, and Constructing the World is still on track for publication sometime next year.
You write:
"computations are specified syntactically, not semantically. Although it may very well be the case that any implementations of a given computation share some kind of semantic content, this should be a consequence of an account of computation and implementation, rather than built into the definition."
This would be the crux of any disagreement I have with the paper. To call any changes of state experienced by any finite-state machine, "computation", is to import connotations which just aren't intrinsic to the scenario. In effect, you're saying that any entity with structured states and a dynamical evolution is intrinsically performing a computation. But I would say that as a matter of usage, computation always implies some relationship to representation: either the states are held to already be representing something - then we say they have a semantics - or we say that they have the combinatorial potential for representation - and this is when people talk about "syntax". But syntax is intrinsically a proto-semantic combinatorics. There may not be any intentionality presupposed when you say that something has a syntax, but there's certainly an intentionality-shaped hole being supposed, a hole that is meant to be filled by some interpretation, some semantics.
In the end, I have to take a position with respect to intentionality which is as ontologically radical as your own position with respect to qualia: Intentionality, while not necessarily ontologically primitive, is certainly not reducible to the semantically deracinated elements of standard physical ontology. It's not constituted by causal relations, for example, though it certainly enters into characteristic causal and other relations. I'd further postulate that you cannot ontologically understand intentionality until you understand the self and the unity of consciousness, first at the level of subjectivity (phenomenological ontology), and eventually at the "noumenal" level of objective ultimate ontology. Intrinsic intentionality (as opposed to the "as-if" form) is a structure of consciousness, and its nature is bound up with the global nature of consciousness-as-a-whole.
Posted by: Mitchell Porter | April 04, 2011 at 10:27 PM
can you hint at which parts you don't still agree with?
Posted by: james | April 27, 2011 at 08:37 PM
Mitch: I'm not sure whether we have a nonverbal disagreement here. I agree that there is at least an interesting semantic notion of computation, and you can use the word "computation" for that if you like. I think there's also an interesting nonsemantic notion that I try to characterize in the paper. Maybe you think the latter notion is incoherent, or maybe you think one of the computational sufficiency/explanation theses is false with that notion; I'm not sure. I am not unsympathetic with primitivism about intentionality myself, but I don't think it undermines the theses of the paper.
James: I think that the thesis of computational explanation needs to be qualified a little. See the new footnote 6 at the end of the paper.
Posted by: djc | May 07, 2011 at 11:27 AM