Item

Abstract

A very important feature of primate vision is the ability to integrate motion signals into a coherent percept. To this end, a two-stage motion integration model has been proposed that selectively integrates local signals over time and space to reconstruct the global motion pattern. It has been suggested that the first stage responsible for local motion detection takes place in lower level visual area(s), while a second stage in higher area(s) integrates the local motion signals in order to extract the global motion direction. Support for this hypothesis stems mainly from recordings in anesthetized non-human primates while evidence in awake-behaving ones is very limited. Furthermore, very little is known about how motion integration is influenced by perception. In this study, we designed a novel pseudo-plaid stimulus that can parametrically modulate coherent or transparent motion perception by changing local feature information. The stimulus consists of two types of apertures over a line plaid display. The first group of apertures allows only single contours to pass through while the second only intersections. Human psychophysics demonstrated that the motion perception changes parametrically with the proportion of the two types of apertures from 100 transparent when only single-contour apertures are present to 100 coherent when only intersection apertures are displayed. Then, we used this stimulus and performed multi-electrode recordings in areas V1 and MT of alert macaques. Analysis of the firing rates during the whole trial (1000 ms) demonstrated that MT neurons were strongly modulated by the proportion of different aperture types reflecting the perception. Specifically, MT neural responses increased when motion perception was more coherent. In contrast V1 neurons did not show any significant changes by using this measure. These data corroborate the hierarchical organization of motion integration and demonstrate the relationship of neural signals with subjective perception.