Apparent neural feedback loops

[From Bill Powers (930928.1530 MDT)]

Mark Olsen (copy direct)

Not all neural "loops" that can be found anatomically have to be
functional negative feedback loops. Just consider the general
relationship between two systems at two levels in HPCT:

                        > > \
                       / \ "LOOP" r| \ r other
                     / \ V --->- lower
                   / \ --->IpCeO-->- systems
             other lower | |
               systems from lower to lower
                           systems systems

[IpCeO = input funct-perception-comparator-error-output funct]

Note how the higher-level input function receives information
from lower systems: copies of the perceptual signals reach the
input function of the higher system where they are combined to
create the higher-level perceptual signal. Also, the output of
the higher-level system sends signals that become reference
signals for lower systems, including the one shown.

The part labeled "LOOP" above looks at first like a feedback
loop, but if you trace it out, you'll find that at one point,
where the arrow goes into the input of the lower system that's
completely shown, you've travelling against the input arrow.
There are signals going both ways between the levels, but they do
NOT make a feedback loop.

It's also possible that there are local feedback loops inside
various functions. I suggest looking into the literature of
analog computing to see how short negative feedback loops can be
used to create special computing functions. By "short" I mean
very short -- involving only a few adjacent neurons, as you will
find in the cerebellar cortex where the outputs of Purkinje cells
feed back through basket and stellate cells to modify the same
Purkinje cell's operation. When such loops are present, you can't
say what the input-output relationship is without figuring the
feedback into the equation.

In PCT, which is concerned primarily with overt behavior, we do
not consider any feedback loops that are exclusively internal to
the brain (with the exceptions of the imagination loop, and of
certain possible reorganization loops that act locally, still
under consideration). All the loops in the hierarchy of control
pass through the environment.

Of course your suggestion that the imagination connection might
be involved is still valid. I'm just trying to point out that you
have to trace the circuitry in considerable detail, both
anatomically and functionally, before you can say what the
various conformations mean. The mere fact that you can find
signals travelling both ways between different volumes of the
brain by no means indicates that any kind of feedback is



Bill P.