Curtis Bell has recently published a series of papers about the
generation of sensory expectations within cerebellum-like structures.
(BBE 1997;50(suppl 1):17-31), (Nature 387:278-281), (JNeuro,
17(16):6409-6423). The results described in these papers seem supportive
of Pribram's holonomic memory model. It's interesting that Bell
identifies dendritic circuitry in the molecular layer that seems
necessary to produce a negative image that is then sampled by the
principal cells of the system.
Bell suggests that the dorsal cochlear nucleus (DCN) is one of the
cerebellum-like structures that matches expectation against afference.
He also indicates the optic tectum of teleost fish has a similar
structure. I suspect those matching processes are actually multi-staged.
Bats are good at memorizing extensive areas (hundreds of square
kilometers) based on auditory data, but I suspect the DCN is a bit small
(~20,000 primary cells) to store that much information. The colliculi or
an area of the auditory cortex may be the regions where that auditory
map is stored. I have to note, however, that I suspect expectations can
be initialized by higher levels and are not just dependent on current
afference. A fast-flying bat can cross the space it monitors in 200-400
msec, and the process Bell discusses runs more slowly than that.
herwin at gmu.edu