Abstract
Visual perception across eye-movements is not veridical. As an example, saccadic eye-movements modulate the perceived location of a briefly flashed stimulus leading to a perceptual compression of space. During everyday-life eye movements occur also during navigation through an environment, thereby challenging the perception of self-motion direction (heading). Here we asked if saccades influence also the perceived heading in humans. We found a perisaccadic compression of perceived heading and aimed to identify a neural correlate of this new perceptual phenomenon in the animal model (macaque monkey). Human subjects were presented brief (40ms) visual sequences simulating self-motion across a ground-plane (random dots) in one of five different directions during fixation or perisaccadically. After each trial the subjects had to indicate their perceived heading. Eye-movements were monitored by an infrared eye-tracking system. During fixation perceived heading was not perfectly veridical but shifted centripetally. In saccade trials performance was very similar to fixation trials for motion onsets long (>100ms) before or after a saccade. Around the time of the saccade, however, perceived heading was strongly compressed towards the straight-ahead direction, being equivalent to a compression of heading space. Precision of behavioral judgments was not modulated perisaccadically. In search for a neural correlate of the perceptual effect, recordings were performed in two dorsal stream areas of two macaque monkeys (areas MST and VIP, respectively). In a first step, we aimed to decode self-motion direction from population discharges of both areas. Heading could be decoded veridically during steady fixation and during tracking eye-movements. During saccades, however, decoded heading was compressed towards straight-ahead. We conclude that saccades compress not only perceived space, but also perceived heading. This newly described perceptual phenomenon could be based on the visual processing in cortical areas being responsive to self-motion information. Functional equivalents of both areas have been previously identified in humans.
Meeting abstract presented at VSS 2015