| |
Abstract:
It is unclear whether perirhinal cortex (PRh) is best
characterized as being specialized for mnemonic function or whether
PRh is involved in visual information processing leading to "object
identification". One contributing factor has been a puzzling
pattern of effects observed following lesions of PRh. For example,
PRh lesions affect discrimination only when a large set of objects
is used (e.g., Buckley & Gaffan, 1997). We developed a neural
network model to investigate the idea that PRh has similar visual
information processing properties to those of the rest of
inferotemporal cortex (IT). The model is based on two general
assumptions. First, networks of neurons in IT set up and maintain
representations of visual stimuli, and these representations can
change as a result of experience (perceptual learning). Second,
visual representations in IT are organized hierarchically, such
that simple representations in early processing regions combine to
form more complex representations downstream, culminating in
maximum complexity in PRh. We present simulations of several
recently reported effects of lesions in PRh in monkeys that have
provided key insights into the functions of this region. Simulation
results suggest that effects of lesions in PRh may be due not to
the impairment of a specific type of learning or memory, but to
compromising the representations of visual stimuli.
|
