From weng@cse.msu.edu  Tue Mar  7 13:58:05 2000
From: Dr John J Weng <weng@cse.msu.edu>
Date: Tue, 7 Mar 2000 13:58:02 -0500 (EST)
To: wdl@cse.msu.edu
Subject: WDL Dis. #25

======== WDL Dis. #25 (Bcc: all WDL participants) ======

Chris,

You wrote: 
> I thought the real mystery was ---
> "experience-expectant" processes.

Let us clearly define what ``experience-dependent''
and ``experience-expectant'' so that we have the real content
instead of just arguing about terminology.

(A) Experience-expectant: the representations that are instantiated
     though development are predetermined by genetics and they are merely 
     ``triggered,'' ``filled-in,'' ``slowed down or speeded up in 
     unfolding'' by the sensory stimuli.

(B) Experience-dependent: what representations are developed 
      depend very much on the nature of sensory stimuli.  For 
      example, even what source of input that a neural zone
      takes is dependent on sensory stimuli. 

I guess that the psychologists who emphasize "innateness" tend to 
like (A) explanation. 

>From what I read, existing studies in neuroscience support (B) 
instead.  

There exist rich studies of brain plasticity in neuroscience, 
from varying extent of sensory input, to redirecting input, to 
transplanting cortex, to lesion studies, to sensitive periods.
I guess that the neuroscientists who will give talks at our WDL
will explain some major results.  To be short and to the point, I just 
mention a single study by Dr. Michael M. Merzenich and his colleagues.  
(He will give a talk at WDL.)

In this study, two bars are consistently pressed againt the fingers of 
*adult* owl monkeys.  One bar was pressed across the distal
ends of three middle fingers.  Another one across the proximal end
of same three fingers.  The following gives a rough illustration: 

               finger 1   finger 2   finger 3
                 |           |          |
     bar 1  =================================
                 |           |          |
                 |           |          |
                 |           |          |
     bar 2  =================================
                 |           |          |
                 |           |          |
                 
Each bar gives different pulses.  Monkeys are trained repeatedly
over multiple days with such stimuli.  Result?  The cortical mapping 
architecture in the somatosensory cortex has changed: the normal 
infra-finger receptive fields are mostly replaced by cross-finger
receptive fields.  That is, normally most neurons take input from
the nearby receptors of the same finger.  Now, most neurons take
inputs from receptors from multiple fingers, because the sensory 
signals from multiple fingers are higher correlated.  Interested? The
following is the reference: 

X. Wang and M. M. Merzenich and K. Sameshima and W. M.  Jenkins, 
"Remodeling of hand representation in adult cortex determined by timing
of tactile stimulation."  Nature, 378 (2), pp. 13-14, 1995. 

It is very important to find out exactly what self-organization 
mechanisms the brain uses to organize its cortical mappings. 
What kind of computational models are behind such mechanisms?
Why do ADULT cortical mappings drastically change based on sensory 
signal? (Does it seem easier for nature to 100% design representation??!!)
Are these self-organization mechanisms also widely used in other cortices? 
What implications these studies suggest to developmental robots/machines? 

John