| |
Abstract:
Area TE of the monkey inferotemporal cortex represents the
final pure visual stage of the occipitotemporal pathway, which is
essential for visual object recognition. We had previously found
that cells in TE respond to moderately complex visual features,
that cells responding to similar features cluster in a columnar
region, and that columns representing different but related
features are located at neighboring positions with partial
overlapping. To investigate the roles of these functional
structures in object recognition, we have recently conducted two
projects with behaving monkeys.
In one project, single-cell activities were recorded from TE of
one monkey performing a tacto-visual matching task. The monkey
grasped one of 18 differently shaped objects under an opaque plate,
and then matched visual objects presented on a display with the
grasped object. About 1/5 cells recorded from the dorsolateral part
of TE showed stimulus-selective discharges to the grasped objects
before the start of the visual image presentation. Tacto-visual
matching may occur either 1) in the polymodal areas that receive
converging inputs from the visual and tactile pathways, or 2) in
the interaction between the modality-specific pathways. The results
suggest that the latter is at least partially working.
In the second project, a visually cued GO/NOGO task with
reversals was trained on monkeys, and single-cell recordings were
made from the ventral regions of the prefrontal cortex, to which TE
projects. One of 2 visual patterns was presented as a cue, and
after a delay the monkey had to perform either GO or NO-GO motor
response depending on the cue. Rewards were asymmetrically provided
after either correct GO or correct NOGO responses. The monkeys were
trained to quickly learn the stimulus-motor correspondences with
daily introduced new visual patterns. They also had to switch or
maintain the stimulus-motor correspondences when the motor
requirement or the reward condition was reversed. For cells that
showed differential responses to the visual patterns, we determined
which of the three aspects of stimuli, i.e., 1) the visual
features, 2) associated motor responses, or 3) associated rewards,
was essential for the responses by testing their activities across
both motor and reward reversals. We found that most cells in the
orbitofrontal regions represent associated rewards, but not the
other two aspects. Although we have not localized the site of
stimulus-motor association, the results suggest that the
stimulus-motor and stimulus-reward associations separately occur at
different brain sites.
In addition to these animal experiments, we are now setting a
fMRI system at 4T to realize a non-invasive human imaging with
columnar level resolution. We have tried imaging the ocular
dominance columns in the human primary visual cortex. By using
surface coil and 32-segmented EPI, we have obtained activation
patterns consistent with the known properties of the ocular
dominance columns.
|
