A Neural Theory of Mind

Paul Bush paul at phy.ucsf.edu
Thu Apr 25 15:57:45 EST 1996


I think you bring up a very important concept, even though I don't agree with the
first bit:

In article <191790551wnr at grithill.demon.co.uk>, robin walker <rwalker at grithill.demon.co.uk> writes:
|> 
|> I fully agree with your idea of a complexity pyramid, but respectfully suggest that in this last section you are missing the 
|> point. There may well be a single concept at the top of W but this is not the same as the single concept at the top of sW.
|> At the top of sW is the concept of "self" which is the nucleus around which all other concepts, with complexity decreasing 
|> inversly with radius, orbit. 

I think the concept of self is obviously a large part of each sW, but as we
approach the most complex concepts, the final knowledge singularity, the
objective structure of W becomes such a massive constraint that the top concept
is almost independent of data and therefore subjectivity. There can only be one
top concept, and it is objective. I don't know. This stuff is weird.

|> Putting "self" at the top solves some interesting problems, not least the puzzle of multiple personalities where one could 
|> speculate that some trauma has resulted in the pyramid having two or more peaks of the same complexity level, all using 
|> sub-sets of the same lower order concepts but only one having attention (i.e. being in control) at one time. (Probably only 
|> one concept at a given level can ever have attention at one time.) Of course a good method actor has something similar 
|> except his acting persona are represented by smaller peaks on the flanks of the complexity pyramid and are therefore 
|> subordinate to the central self and can be properly integrated. Control can be temporarily passed to the role without 
|> attention ever being totally lost by the "self".

Yes, I think something exactly like this goes on. Not just for personality (in
orbitofrontal cortex?), but for every brain process. A pyramid (or cone) of
neurons for every behavior you exhibit, based in the entire feature space of the
lowest sensory cortex, peaking to a few (one?) neurons in higher association 
cortex. Other people have had the same idea before us. The higher the cone, the 
more attention must be devoted to it to build it further. New behavior patterns 
split off old ones, as you said. I will say more about this in my next post. 

Paul





More information about the Neur-sci mailing list