chicks with prisms; reinforcement model

[Martin Taylor 951207 11:50]

Wayne A. Hershberger Thu, 7 Dec 1995 10:30:57

       I believe that it has been found that a chick's pecking will adapt
to prismatic displacement if the displacement comes on gradually so that
the chick senses the slight displacement of the kernel with its beak.

Now that's fascinating. I take it by implication that the converse is
true--the chick will not adapt unless the displacement can be sensed
by the beak? So the optical-kinaesthetic map can be changed, all right,
but it has to be done through some kind of touch-related perception. As
if the chick could not determine from vision the direction of error?
I suppose these chicks would not have been allowed to starve to death,
so they wouldn't have had much intrinsic error that might have led to
random changes in pecking direction until an effective remapping was
found. I wonder what is going on in those chicks?

···

===============

Bill Powers (951206.0655 MST)

I've been thinking about it. Your suggestion is that there are two
control loops. One performs the action that the experimenter wants to
see. As a side-effect (via what the experimenter calls the schedule of
reinforcement), it also makes it possible for a second loop to control
the variable the animal wants to control.

...

How about trying a model and seeing what it does?

Wow! I'll have to look closely at your logic and your diagram, but I'm
definitely impressed at first glance. Yes, it sure looks modellable.
Would Rick's spreadsheet approach be suitable, do you think? I've
never tried such a thing, but why not, indeed?

Martin

Martin:
        The chicks' adaptation implies a strong proprioceptive component.
Humans' too. Adapt to prismatic displacement with your left hand and it
doesn't transfer to the right hand.
Regards, Wayne

Wayne A. Hershberger Email: wah@niu.edu
Department of Psychology Fax: (815) 753-8088
Northern Illinois University Phone: (815) 753-7097
DeKalb, IL 60115