[from Randy Beer 920302 via Gary Cziko to CSGnet]
Dear Bill,
I apologize for my delay in responding to your last message but, as I
mentioned in my last message, it is difficult to find the time to
actively participate in discussions such as these.
As near as I can tell from your description, it sounds like your
proposed feeding system should work. One clarification. At several
points in your message, you state that one advantage of your proposed
control system over the current appetitive and consummatory circuits
is that it would allow the insect to move as food is eaten or moved
out from under it. The implication seems to be that the present
circuits do not have this capability. But, in fact, they do. Indeed,
in my original simulation, food patches shrunk as they were consumed,
so this capability is crucial. If the insect loses contact with the
food while it is eating, the consummatory controller immediately
relinquishes control to the the appetitive controller, which will
reorient the insect to the patch. This can be seen in Figure 8.7 of
my book. If the insect loses contact with the food, the consummatory
command neuron will stop firing and its inhibition of the search
command neuron will disappear until contact is reestablished.
I seem to detect a strong aversion to having "superordinate" control
systems turning off subordinate control systems in your comments.
Perhaps I would understand this aversion better if I was familiar with
HCT. However, I should point out that, to take just one example, the
consummatory controller doesn't literally turn off the appetitive
controller when it is activated, it simply denies it access to the
motor system for turning. The relative odor strength on each side of
the insect is continuously computed. However, this odor strength can
only affect the insect's turning when the search command neuron is
enabled.
In fact, I don't really think of the appetitive CIRCUIT as being
subordinate to the consummatory CIRCUIT at all, since portions of both
are always active. Rather, the appetitive BEHAVIOR is subordinate to
the consummatory BEHAVIOR. I tend to think of the artificial insect's
nervous system as consisting of a number of partially overlapping
circuits "fighting" for control of the periphery. Which circuit wins
at any particular time depends upon both the external stimuli and the
internal state of the insect (consider, for example, the interaction
between edge-following and the appetitive phase of feeding). There is
some (rather tenuous) neuroethological evidence for this way of
looking at things, but as you probably know, work on the neural basis
of behavioral choice is still at a rather primitive stage.
More generally, I'm afraid that, due to my inexperience with this
group, I don't have a very clear idea of the purpose of this
discussion, though I sometimes have the vague sense that you are
arguing for or against something. Are you interested in how nervous
systems control behavior? Are you interested in designing artificial
autonomous agents? Are you interested in how the artificial insect
works? Are you interested in a rational reconstruction of the
artificial insect using HCT? Each of these topics lead to a somewhat
different perspective on the issues we have been discussing. If I
better understood your motivation, then perhaps I could tailor my
comments accordingly.
Best regards,
Randy Beer
ยทยทยท
====================================================
P.S. from Gary Cziko
Randy Beer is not on CSGnet. Responses to him should therefore be sent to
his personal address <beer@cthulhu.ces.cwru.edu> with a copy to CSGnet.
My own preference right now would be for Beer's involvement with CSGnet to
be limited to interactions with Powers, lest we lose Beer completely as the
result of heavy salvos from CSGnetters.
But who am I to tell anybody on CSGnet what to do (other than the fact that
as "listowner" I can cut anybody off from the net whenever I wish and there
just ain't nothing nobody can do about it, so there!)?--Gary