An interesting paper on the binding problem approached from experiments
using slice preparations of rat hippocampus showed up on page 612 of the
16 February 1995 Nature (Synchronized oscillations in interneuron networks
driven by metabotropic glutamate receptor activation, Whittington, Traub,
and Jeffreys). Douglas and Martin discuss it on page 563. (My interest
involves the fact that I'm simulating a near relative of this system. The
probable loci for the metabotropic glutamate receptors are the inhibitory
interneurons that modulate the excitatory signal received at the apical
dendrite of pyramidal neurons. My experience is that this system doesn't
work unless it is adaptively tuned.) Douglas and Martin point out that the
major problem with associative memories is decomposition of the afferent
scene into component features--the binding problem. The 40-Hz signal
generated by these inhibitory interneurons appears to be a piece of the
solution, since each cycle reports on a different feature. Buonomano and
Merzenich's recent paper in Science (Temporal information transformed into
a spatial code by a neural network with realistic properties, 17 February,
1028) describes how the 40 Hz signal can then be translated into a spatial
coding that can then be recognized. (My simulation work has been
investigating how the 40-Hz signal interacts with an upstream spatial
coding to generate the temporal information and how the cortical-cortico
connections involved in that process can be created, maintained, and
forgotten.)
It appears this area is starting to wake up, and we may see a solution to
the binding problem fairly soon. My suspicion is that we will discover
that the aware brain contains patterns of activation that make up a
transformed representation of the outside world, and that it interacts
indirectly with the outside world by interacting directly with that
transformed representation. Hence the brain operates semantically, not
syntactically. The binding problem (specifically the problem of scene
decomposition into meaningful elements) appears to be the major obstacle
to getting to a generally-accepted definition of that transformation.
--
Harry Erwin
Internet: herwin at gmu.edu
