Abstract
A moving grating contains two types of motion signals: contour signals found along the stripes of the grating, and terminator signals at the edges of the aperture. The terminator signals are especially salient perceptually, so that a grating displayed in a rectangular aperture appears to slide along the long axis of the aperture. We used this “barber-pole illusion” to probe the integrative properties of MT neurons in alert macaque monkeys. The vast majority of MT neurons responded preferentially to motion along the long axis of the aperture, in accordance with the perceptual barber-pole illusion. The extent of the bias introduced by the aperture elongation was dependent on the aspect ratio, but not on the size of the stimulus, or its position within the MT receptive fields. This effect was reduced, but not abolished, when the terminator signals at the grating edges were rendered “extrinsic” by a frame that simulated occlusion. We studied the effect of monocular occlusion cues further, using a square grating patch with occluders at either the horizontal or vertical edges of the patch. In this case, the neurons were biased in the direction of the “intrinsic” terminators. However, MT neurons were insensitive to other stimulus manipulations that changed the global surface layout without altering local occlusion cues. We used the same stimuli to study neurons at an earlier stage of motion processing in striate cortex. We found that V1 neurons responded only to local motion cues (intrinsic terminators and contours) placed within their receptive fields, but not to the global motion of the barber-pole stimuli. We conclude that MT neurons accord more weight to terminators than to contours, and can use monocular cues to distinguish between intrinsic and extrinsic terminators. The properties of end-stopped, direction-selective cells in V1 suggest that they are well-suited to making both of these distinctions based on local cues.
Supported by NIH EY11379.