Re: COMP/NEURO: Images Extracted from Cat Brain

Robert J. Bradbury (
Mon, 11 Oct 1999 12:10:00 -0700 (PDT)

On Mon, 11 Oct 1999, Bryan Moss wrote:

> I wrote:
> > You could say the same about language -- even isolated
> > deaf/mute people will "invent" a language to communicate
> > with.

> > What form does the 'language' of an isolated deaf/mute person take?

Sorry Brian, use of the word "isolated" was poor choice. I'm refering to deaf people who are not otherwise trained to sign (say in a culture where "signs" are not known). They will invent a "unique" sign language.

> If it's gestures/noises that correspond to objects I would suggest that
> it serves the same purpose as counting fingers or sketching out ideas,

You could go a step lower and look at human ability to read "faces" or wolves or chimps ability to read "postures" as expressions of emotion (or intent). I suspect our ability to communicate evolved from a two-fold selection pressure -- friend or foe determination and communicating complex "learned" software.

I believe there has also been a gene discovered that seems to delay the learning of correct verb conjugations and plural syntax (primarily in males). It turns out to be relatively "common" (something around 1 in 50) and strikes males more than females (which accounts for some of the "slow Johnny's" we have met in our early childhood). I believe that there is still a fair amount of debate regarding whether Chomsky's theory that the semantics of language is "hardwired" is accurate.

But I think I would generally agree with Eliezer, that at least in 2 respects (a) low level emotions/fundamental "drives" and (b) the ability of our senses to select features and/or assemble patterns is pretty hardwired. Those will form a common foundation that the nanobots can clearly latch onto (i.e. gee these neurons always fire when the color "red" is put in the visual field, these neurouns always fire when the word "Mommy" is heard). Once you can draw up the statistical correlation map (and this will not necessarily be easy), you can backtrack to the "core concept" neurons. Building up from there you will have the ability to provide very primitive communications.

Now, when you get up to more abstract concepts like "table" (that is linked in my brain to all the tables I have ever experienced) and must therefore be different from all of the tables you have experienced, you are starting to deal with abstract concepts. Whether there are core neurons involved in abstract concepts or whether they are just a "firing patterns" that gets copied around (as Calvin would argue) isn't clear to me just now. I suspect at the most abstract and theoretical levels all you are dealing with is firing patterns. So it will be very interesting to see *what* I get if my brain can be induced to "play" patterns produced by Steven Hawking's brain.

There you go, I've just convinced myself that there may be a really good reason not to go take all the courses I didn't particularly like to get a degree -- it will be so much simpler to buy the associated knowledge/patterns in the future. Maybe we should suggest Robin H. start an idea futures market on pricing for college course material as "meme-xeroxes".