Selective attention in the primate visual system.

Abstract

A series of single unit PET and behavioural studies is described addressing the functions and neurophysiology of visual attention. Beyond striate cortex, visual information is processed in a network of separate cortical areas, specialized in part for analysis of different visual attributes. Issues arising in such a modular system include the nature of the attentional state in extrastriate cortex, its flexible control by the requirements of current behaviour, and the coordination between areas implied by attention to whole objects.The single unit studies (Chelazzi, Miller, Duncan, & Desimone, 1993) dealt with control of attention by complex object features. Macaque monkeys were trained in a simple form of visual search. The task was to make an immediate saccade to whichever of two stimuli (digitized pictures) in a visual test display matched a previously - presented cue. Recordings were made from single units of inferotemporal (IT) cortex, an area specialized for object recognition. In the delay period between cue and test, there was enhanced spontaneous discharge in cells selectively responsive to the target. Such priming of target - selective cells has the properties required of an attentional control signal. When the test display was then presented, cells responsive to both target and nontarget stimuli showed an initial discharge. Beginning 90 - 120 msec before the saccade, however, nontarget responses were suppressed. Suppression of responses to ignored stimuli has previously been observed in this area of cortex in spatial selection tasks. When selection is by object features, it is possible that suppression is a direct consequence of pre - existing priming of target - selective cells.In behavioural studies, different objects compete for visual attention, while different attributes of the same object do not. Thus concurrent discriminations concerning the same object can be made without mutual interference, while discriminations concerning different objects cannot. This comparison was pursued in the current PET study (Orban, Duncan, Dupont, Ward, Bormans, De Roo, & Mortelmans, 1993), measuring regional activity in the human brain during concurrent discriminations of orientation and spatial displacement. In two conditions, discriminations concerned the same object (to left or right of fixation), while in two more, they concerned different objects (one to each side). Visual regions activated by these tasks were occipital area 19, especially in the right hemisphere, and parietal area 7 bilaterally. The area 19 activity was close to that previously observed during orientation discriminations, while parietal activity may be related to spatial analysis. …