Why I am not a modularist

by John

Will is taking to his hobby like a Leiter-approved professional. But I still have to dissent from the thrust of his analysis.

I agree that the differences in intrinsic neural structure, not to mention the striking variety of ways that neurons come together to form the brain's proper parts, shows that the brain is not "an undifferentiated blob". But I don't see why anything as strong as what Will is calling "hard modularity" - that is, the idea that specific cognitive processes can be localized to neural "modules" whose proper functionings suffice for the realization of the processes in question - falls out of this observation, nor (and perhaps more importantly) do I see why the thesis of hard modularity would need to be true of the brain in order for its inner workings to be "intelligible". Connectionist networks, for example, are very often notoriously non-modular, but that does not stop cognitive scientists from providing illuminating analyses of how they work in both computational- and algorithmic/implementation-level (read software- and hardware-level) terms. It is the existence of what we might call dedicated neural substructures - that is, regions of the brain whose operations are specialized for those very cognitive functions that are turned up by neuropsychology - that is at issue here, and just as you (or Will, or perhaps my techie father, anyway) can understand how the hunk of silicon that is my computer manages to do addition even though there's no addition-dedicated part of it, you can analyze my brain's inner workings and account for its roles in any of the various components of my mental life without taking each such component to correspond to a specific part of the brain with respect to which the accounting must be done.

I fear that I have missed something. But I think that I have not. In each case it seems likely to be the entirety, or very nearly so, of the gray lump in my head that is responsible.

Comments

[imponderabilion]

There is a very nice discussion of functional brain architecture here:
http://www.agcognition.org/papers/anderson_bbs_2010.pdf
you may find it at least interesting. It is clear and understandable for non-specialists. There you will find some arguments against modularity.

[imponderabilion]

If by modularity one means "functional segregation" then it's trivial, although it remains to be aswered what one means by function ("problem solving" for example would be very distributed and thus non-modular). On the other hand, modularity as proposed by Fodor is not likely (at least no likely as a general organizational principle). There are options in between, sure, but still controversial.
Look though the article anyway, you could find it helpful.