Wednesday, October 23, 1996

Computer models of the brain

> From: Darrell Dodge
> But there are neurological processes that could be hypothesized to play a
> part in situations like Vicki's.
> Enough. This hatchet-job would probably not impress my biopsych prof. And
> that's just one reason why I need to be hitting the books instead of
> continuing here.


It impressed me! It's great that you are able to even begin to think at this level of granularity. It's true that rats can't talk, but it's not completely unfeasible to turn some of these neurological ideas into computer models. Computers CAN speak. The way they have been programmed to do so, so far, has nothing to do with how we produce speech, but it wouldn't be impossible to build a speech producing mechanism based on the kind of control issues we are talking about.

My dream is to produce a speech producing apparatus with a little knob, such that, if I turn the knob a bit, stuttering starts happening. Now, if someone can give me some good arguments why NASA should be funding this type of work...

Some time ago Megan Neilson sent me some very interesting papers on the work she and colleagues were doing at the Univ. of New South Wales. They built a mathematical model (Adaptive Model Theory) according to which a problem in auditory tracking could be responsible for stuttering. The point I want to make is not whether or not their model is valid for this purpose (I've been hoping to get the time to study it carefully...), but that it is in fact possible to start building such models! If Megan is still listening she might give us an update on her work...

Anyhow, Darrell, keep up your model thinking... even though it's sure to be above many of our heads... especially mine.

No comments: