Item

Abstract

The inferotemporal cortex of the monkey (IT) is generally thought of as a visual area critical for object recognition. Physiological support for this claim comes from numerous reports of inferotemporal neurons that respond most robustly to the presentation of complex visual patterns, most notably faces. We have recently investigated the physiological properties of hundreds of single IT neurons while rhesus monkeys viewed and identified images of faces, animals, and other objects. Among these neurons, we found a small number of cells whose activation patterns were both highly selective for particular stimuli and limited to periods during which the monkey's behavior indicated that the stimulus was perceived. Indeed, these cells appear to play a role in the process of converting sensory fragments into recognized wholes. Here, we consider the properties and possible function of the large number of remaining cells. For all recordings, analog neural signals were stored digitally, allowing a careful analysis of not only spike timing, but also of extracellular waveforms. In addition to the highly selective visual neurons, we found at least three types of other cells. First, we found cells whose spontaneous rates were extremely low and whose extracellular waveform was indistinguishable from other highly selective neurons. Unlike the selective cells, these units rarely fired in bursts and they were not obviously visual. Second, we found neurons that looked and sounded like other visually selective neurons, but whose onset latencies were significantly longer than other visually responsive cells, indicating that their activation may have been driven by feedback. Finally, a significant number of neurons with thin spikes were found, whose background activity was higher than other neighboring cells. These cells were not highly selective, but were clearly visually modulated. Similar cells have been described in sensorimotor, frontal, and primary visual areas, and they are likely small neurons participating in local microcircuits within IT. By considering all of these neurons as a group, we can appreciate both the commonalities of IT cortex with other cortical areas, and the richness of the neural components within IT that is often overlooked or indiscriminately lumped together.