From newshub.ccs.yorku.ca!torn!utcsri!rpi!usc!cs.utexas.edu!uunet!trwacs!erwin Tue Jul 28 09:41:26 EDT 1992
Article 6456 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:6456 comp.ai.neural-nets:3580 sci.cognitive:195
Path: newshub.ccs.yorku.ca!torn!utcsri!rpi!usc!cs.utexas.edu!uunet!trwacs!erwin
>From: erwin@trwacs.fp.trw.com (Harry Erwin)
Newsgroups: comp.ai.philosophy,comp.ai.neural-nets,sci.cognitive
Subject: Visual Keys
Keywords: vision
Message-ID: <663@trwacs.fp.trw.com>
Date: 15 Jul 92 17:30:57 GMT
Followup-To: sci.cognitive
Organization: TRW Systems Division, Fairfax VA
Lines: 37

In "Objective analysis of the topological organization of the primate
cortical visual system" Malcolm P. Young (Nature, Vol 358:152-155, 9 July
1992) presents evidence that the visual processing system is divided into
two streams with limited cross-talk. The dorsal stream (V2-MST-FEF-
STPp-Area 46/STPa) apparently processes spatial/movement data while the
ventral stream (V2-V4/VOT-IT-Area 46/STPa) is involved in pattern
recognition. Young identifies the prestriate group to MST/posterior
parietal complex connection as being highly redundant, although I suspect
the redundancy may parallel the redundancy seen in the pyramidial tracts,
where it may involve inhibitory processing of a generalized motor stream.
Young notes that the reconvergence seen in Area 46 and the anterior
superior temporal polysensory area (STPa) is an argument against the
recent models providing coherence of visual perception by the
synchronization of oscillatory activity in neurons distributed widely in
the brain.

Although this analysis provides evidence for reconvergence, ghosts can be
expected in the resulting visual images unless some sort of key is
transmitted with the pattern and location/movement data. Spence et al.
(Neuronal maps for sensory-motor control in the barn owl, NIPS-1, 366-374)
identify narrow band spectrum as the corresponding key for aural
processing in the barn owl. What is the key for visual processing? I have
conjectured that it is texture, but I have no evidence. Interestingly,
Marr identifies a bank of filters in the ganglion cells of the retina.
This is similar to the bank of filters in the cochlea that serve to
construct the spectrum for aural objects. Marr's filter bank might be able
to serve in that capacity, particularly as it is apparently not involved
in identifying zero crossings as Marr conjectured.

Note that the synchronization of oscillatory activity in the distributed
models of vision serves as a similar "key," although one I feel is much
less robust in its operation.

-- 
Harry Erwin
Internet: erwin@trwacs.fp.trw.com



