HPCT and object recognition

I am leading a discussion this week on a paper by G.C Baylis and J.
Driver (1993). Visual Attention and Objects: Evidence for hierarchical
coding of Location. JEP/HPP, 19, 451-470. It has relevance to
questions/criticisms I had a number of months ago for HPCT concerning
neuroanatomy.

I had said a while ago that HPCT just doesn't fit very well with what we
know of neuroanatomy. For instance, I mentioned that one visual pathway
(ventral from V1 to Inferior Temporal) is concerned with object
recognition--something analogous to the configuration level--while the
dorsal pathway (from V1 thru MT and parietal regions) was concerned with
coding spatial location--something analogous to the relationship level.
These systems are conceptualized as being in parallel.

But the article above reformulates our understanding of these separate
pathways such that the ventral object recognition pathway (also known as
the What pathway) lies hierarchically below the paretial spatial pathway
(the Where pathway). The shift comes in conceptualizing the What pathway
as a system which encodes the spatial relationships between the contours
of the object. Hence, the What pathway encodes relationships between the
oject's contours while the Where pathway encodes relationships between the
objects themselves.

Baylis and Driver didn't think come up with this theory (I think Marr
did), but this paper gives good evidence for it. Anyway, it has obvious
implications for HPCT so I thought I'd mention it to ya all.

Mark Olson
Neuroscience Program
University of Illinois
m-olson@uiuc.edu