Neuronal mechanisms of motion perception.

An enduring problem for sensory neurophysiology is to understand how neural circuits in the cerebral cortex mediate our perception of the visual world. In part, the problem endures because it is difficult; the circuits in visual cortex are formidable both in their number and in their complexity. Of equal importance, however, is that investigation of the visual system has yielded a stream of fascinating insights into the nature of cortical information processing. Perhaps foremost among these insights is that individual cortical neurons, in contrast to retinal photoreceptors, respond selectively to perceptually salient features of the visual scene. For example, neurons in striate cortex (or V1) respond selectively to the orientation of local contours, to the direction of motion of a visual stimulus, or to visual contours that fall on disparate locations in the two retinae (for review, see Hubel 1988). Selective neurons of this nature are often thought to be related to specific aspects of visual perception. For example, orientation-selective neurons could provide the basic information from which we perceive shape and form, direction-selective neurons might play a prominent role in seeing motion, and disparity-selective neurons could mediate the sensation of stereoscopic depth. Although straightforward links between neuronal physiology and visual perception are intuitively appealing, the evidence for such links is generally indirect (see, e.g., Teller 1984). The goal of our research is to explore--in as direct a manner as possible--the relationship between the physiological properties of direction-selective cortical neurons and the perception of visual motion. All of the physiological experiments were conducted in the middle temporal area (MT, or V5) of rhesus monkeys, a higher-order visual area that lies near the junction of the occipital, parietal, and temporal lobes as illustrated in Figure 1. We chose MT for these experiments because it contains a conveniently organized population of direction-selective neurons. More than 90% of the neurons in MT are direction-selective (Zeki 1974; Maunsell and Van Essen 1983), and they reside in a series of "direction columns" that systematically represents direction of motion at each point in the visual field (Albright et al. 1984). MT is thus a logical site to investigate the role of direction-selective neurons in motion perception. Our general strategy is to conduct physiological experiments in rhesus monkeys that are trained to discriminate the direction of motion in a random-dot motion display. In such experiments, we can simultaneously monitor physiological events and perceptual performance. The psychophysicat task is designed so that good performance depends on signals of the kind carried by direction-selective cortical neurons. We asked three basic questions during the course of the investigation: (1) Is performance on the direction discrimination task impaired following chemical lesions of MT? (2) Are cortical neurons sufficiently sensitive to the motion signal in the random-dot display to account for psychophysical performance? (3) Can we influence perceptual judgments of motion by manipulating the discharge of directionally selective neurons with electrical microstimulation? The answer to each of these questions is "yes" (Newsome and Par6 1988; Newsome et al. 1989a,b; Salzman et al. 1990); we therefore conclude that under the conditions of our experiments, perceptual judgments of motion direction rely heavily on information carried by direction-selective neurons in MT.