Abstract
Human eyes continuously and dynamically feed the retina with distinct visual inputs. Even during gaze stability, the eyes are performing small relocations to prevent for visual fading. Among these fixational eye movements, the fastest and the largest are microsaccades. Microsaccades were asserted as being functionally similar to saccades, scanning visual inputs to gather relevant information for the task at hand. This functional hypothesis remains controversial and has never been yet investigated for the biologically relevant class of faces. This is even more interesting, since Westerners preferentially fixate the eyes and mouth during face identification. Strikingly, Easterners focus more on the central facial region and weather this cultural perceptual bias is coded by microsaccades is unknown. To this aim, we examined the occurrence of microsaccades while Western and Eastern observers were performing a face identification task with their fixation maintained on the center of the screen. Observers first learnt 8 face identities. To investigate microsaccade orientation, we defined 9 equidistant Viewing Positions (VPs) covering the internal facial features. Faces stimuli were presented on a random VP centered with the fixation cross. Western and Eastern observers showed a similar face identification performance and comparable microsaccades patterns. Microsaccade rate followed a typical time course, an early inhibition from 100ms with a burst occurring between 150 to 300ms. The eyes elicited the highest rate of microsaccades in this later time-window. In addition, all the microsaccades occurring on the outward VPs were strongly oriented to horizontally center the face stimulus, with no specific orientation preference in the midline VPs. Crucially, observers identified faces significantly faster if they performed a microsaccade in this time-window. Our data show that microsaccades are boosting face identification performance by centering facial information and speeding response times. Microsaccades are universally tuning visual inputs to optimize face processing and identification.
Meeting abstract presented at VSS 2013