| |
Abstract:
Neurons in inferotemporal cortex (IT) are well known to
respond selectively to particular complex visual objects. The
mechanisms underlying this selectivity, however, are not well
understood. Here, we asked whether selectivity can be based on the
presence of specific combinations of object parts. Two monkeys were
trained to perform a visual object discrimination task in which
objects were defined by a specific conjunction of parts chosen from
a small set. No individual parts alone were sufficient for correct
performance of the task. After training, we recorded the activity
of single neurons in IT elicited during the presentation of objects
from the training set and from an equivalent control set. We found
that objects from the training set elicited selective neuronal
responses more often than objects from the control set. Further,
restricting consideration to cases in which selectivity was
present, we found that neurons were sensitive to the convergent
influence of multiple parts more often in the case of the trained
than control objects. We conclude that training can induce in IT
neurons a form of visual object selectivity that is based on the
combined presence of multiple parts.
|
