Inferotemporal cortex (IT) plays a crucial role in object vision as indicated by the fact that injury to it results in profound recognition deficits. The role played by IT in object recognition has been elucidated by research carried out over several decades in macaque monkeys. This has indicated that neurons in IT are pattern selective. Each neuron responds to certain complex images and not others. To gain an understanding of the nature of neuronal pattern selectivity in IT - and hence to cast light on the nature of the neural machinery that underlies visual pattern recognition - is the aim of experiments described in this proposal. Three series of experiments will be carried out. The approach in each case will be to measure the responses of neurons to displays consisting of discrete parts that can be manipulated independently so as to determine how the response to the whole is built up from the responses to the parts. Series 1 will ask whether IT neurons as a population represent, in a dimensionally reduced code, a family of shapes consisting of joined linear segments and having roughly the complexity of alphanumeric characters. Series 2 will ask whether neurons in IT are selective for the arrangement of elements in an image and, if so, whether selectivity is invariant across changes in size and location. Series 3 will ask how neuronal activity in IT represents displays consisting of items in an array, including words and hierarchical stimuli.
Showing the most recent 10 out of 16 publications
