I recently heard a talk by Elissa Newport, which reviewed
some current thinking on brains and language. It led me to a question. Let me
back into it.
Brains it seems are quite plastic. What this means is that
if a part of the brain usually tasked with some function is incapable of so
performing (e.g. it has been insulted in some way (or even removed (quite an
insult!)) other parts of the brain can pitch in to substitute. If the insult is
early enough (e.g. in very young kids) the brain substitutes for the lost brain
parts are so good that nary a behavioral deficit is visible, at least to the
untrained eye (and maybe even to the experts). So, brains are plastic. The
limiting case of plasticity is the equipotentiality thesis
(ET); the idea that any part of the brain can substitute functionally for any
other part.
Now one can imagine how were ET true, one might be led to
conclude that minds are also equipotential in the sense that the computations
across cognitive domains must be the same. In other words, it strikes me that
ET as a brain thesis leads to General Learning theory as a psychological
thesis. Why? Well, on the assumption that minds are brains when viewed
computationally, then there would be little surprise were any part of the brain
able to compute what any other part could if in fact brains only did one kind
of computation.
Conversely, if brains are not fully labile (e.g. if they had a more modular design) then this
would suggest that the brain carries out many distinct kinds of computation and
that some parts cannot do what other parts do. In other words, the reason that
they are not fully labile is that they differ computationally. I mention this
because it appears that current wisdom is that brains are not completely equipotential. So, though it is true that brains are
not entirely fixed in function, it also seems to be the case that there are
limits to this. Elissa reported, for example, that language function, which is
generally lateralized in the left, migrates to the analogous place in the right
hemisphere if the left side is non-functioning. In other words it moves to the
same place but on the other side. It cannot, it seems, move just anywhere (say
to the hippocampus (by far my favorite word in neuroscience! I makes me think
of a university version of this).
That the language slack left by the left hemisphere is taken up by the same
place in the right hemisphere makes sense if these areas are specialized and
this makes sense if they execute their own unique brand of computations.
Or at least that’s the way it seems to me at first blush. So
here’s my question: is that the way it seems to you too? Can we argue from
brain “modularity” against general learning? Or put another way; does a mind
that uses general learning imply a brain that is equipotential and one that is
not fully plastic argue in favor of mental modules? I feel that these notions are linked, but I
have not been able to convince myself that the link is tight.
Let me add one more point: even modularists could concede
that different parts of the mind use partially overlapping cognitive operations.
What is denied is that it is all the same all the way down. But let’s forget these niceties for now. What
I want to know concerns the pure case: does brain modularity imply mental
modularity and vice versa? Or are the two conceptions entirely unrelated?
Here is mark Johnson via me. For some reason he is still having trouble replying. Any of you with compassion for our technologically challenged CS friends, feel free to advice. BTW, Mark insists that this is not competence but prioritizing. As my daughter would once have said: whatever:
ReplyDeleteSounds reasonable to me, but I'd like to hear what real neuroscientists (e.g., Hickock and Poeppel) have to say.
My neuroscience is a bit old, but I believe that roughly speaking brain development proceeds by laying down an extremely rich set of neuronal connections along predefined pathways and experience determines which of these die off, which is certainly consistent with a nativist position.
However, there are experiments (with chinchillas, I think) where the auditory and visual nerves are swapped at birth (I think of this as "plugging" the audio cable into the video input, and vice versa) and apparently the animal develops sort of normally. I amazed this can be done at all (think how hard it is to get your computer to talk to a video projector, even when the plugs are in the right sockets). The fact that the animal can apparently (sort of) see and hear is an amazing demonstration of plasticity, since I would have expected that the visual system, being evolutionarily rather ancient, would be somewhere where you'd find all kinds of specialised "hardwired" circuitry. This is consistent with the hypothesis that the genetic code specifies fairly generic domain-independent circuitry, which experience (dare I say "learning" on Norbert's blog?) specialises to the domain. But I have no idea what kinds of circuits might somehow develop both into (chinchilla) sound or vision, and what kind of "learning" process might enable them to do this.
It would be interesting to hear about the "rewiring experiments" that don't work. What happens if you plug a sensory nerve into a motor area, for example? This would give us a hint about the limits of this plasticity.
So there are deep mysteries here!