I have many doubts with the concept of "modularity". Is there good evidence for some modules? Is it only a post hoc speculation? Do we know exactly the concrete details (e.g. neural mechanisms, genetic determinations) of some modules? Do we know which neural subsystems and mechanisms are domain-specific and which ones are not? What's the evidence for the rival proposals (massive modularity, non-massive modularity, general purpose mechanisms)? Could you recommend me some references that clarify the conceptual and empirical issues about modularity?

palindnilap wrote:- Computer systems are much more adaptive when they are made modular. So modularity is an excellent candidate for an adaptation.
- Some functions of the brain, like face recognition for instance, show close to nil continuous individual differences but some spectacular discontinuous difference in people where no other function is impaired. This hints for a good encapsulation of those particular functions.
Not strong evidence at all I reckon. Probably there are many other arguments, but I would be surprised if they were of the smashing kind. By the way, I do believe that Fodor has overblown modularity by a long way.
No other evidence proposed ? Is there a cognitivist in the house ?

I'm not a cognitivist, but I think there's more evidence. Neuropsychological data? Double dissociations? Cross-cultural data?
Sperber defined a cognitive module as "a genetically specified computational device in the mind/brain that works pretty much on its own on inputs pertaining to some specific cognitive domain and provided by other parts of the nervous systems (e.g. sensory receptors or other modules)." He claims that given this notion, the view that perceptual processes might be modular is highly plausible, and the view that conceptual processes might be modular is also defensible.
The research programs that appeal to modules seem to be rather heterogeneous: Chomsky's innate grammar, Karmiloff-Smith's gradual modularization, Sperber's massive modularity, Fodor's non-massive modularity, Marr's modular approach to vision, Liberman's modular approach to auditory processing, studies about unlearned constraints on induction in word learning (e.g. Gelman, Keil, Markman, Newport), studies about folk theories (e.g. Gopnik, Atran, Keil), studies about expert skills (e.g. Ericsson, Simon), comparative research on species-specific unlearned constraints, cross-cultural research (e.g. Berlin studies of some universal regularities related to colour words and biological categories). Is there any book that tries to make a careful assessment of those different proposals of "modularity", and their respective evidence?
Also, is there a conflict between modularity hypotheses and general learning hypotheses? Can those hypotheses be integrated?

I've found an entry about this issue in the Stanford Encyclopedia of Philosophy, that distinguishes different claims (Fodorian, massive, and modest modularity), and summarizes some pros and cons for each one:
http://plato.stanford.edu/entries/modularity-mind/
What do you think?

seeker wrote:I've found an entry about this issue in the Stanford Encyclopedia of Philosophy, that distinguishes different claims (Fodorian, massive, and modest modularity), and summarizes some pros and cons for each one:
http://plato.stanford.edu/entries/modularity-mind/
What do you think?

Yeah I read that article a while ago and I thought it was a fairly decent overview of the issues involved with modularity. I was under the impression that modularity is a problematic concept though because it relied on the computational theory of mind, which isn't a well-accepted as a valid model of the mind/brain processes.

Mr.Samsa wrote:Yeah I read that article a while ago and I thought it was a fairly decent overview of the issues involved with modularity. I was under the impression that modularity is a problematic concept though because it relied on the computational theory of mind, which isn't a well-accepted as a valid model of the mind/brain processes.

Why do you say so? I don't think it's dependent on a computational theory, unless we consider that any neural process qualifies as a kind of "computation", in which case the "computational theory" becomes an anodyne notion. Here's another interesting article:
http://sas-space.sas.ac.uk/944/1/R_Samuels_Mind.pdf
The author argues against "massive modularity", but defends "peripheral modularity":

The case (for modularity) is most plausible for low-level (or “early”) visual processes – for example, those involved in the perception of color, shape from motion, and depth from disparity. In such cases one finds illusions strongly suggestive of mechanisms encapsulated from belief (Fodor, 1983; Pylyshyn, 1999). Moreover, in contrast to central processes, we possess quite detailed and plausible computational models of these perceptual processes: models which suggest the kinds of computations involved are highly specialized for very specific sorts of input and draw on very restricted kinds of information (Palmer, 1999). So, for example, the computations involved in computing depth from disparity are quite different from those involved in computing lightness; and the sorts of information relevant to one are irrelevant to the other. Finally, and again in contrast to central processes, a fair amount is known about the neural basis of such processes; and again what we know suggests that the modular picture is very plausible (Zeki and Bartels, 1998).