Abstract
[Goal] Microsaccades, incessant “fixational eye movements” (<1°), are an important window into a variety of cognitive functions including attention, expectation and working memory. Yet, its role in visual perceptual learning (VPL)–improvements in visual discrimination due to practice–is practically unexplored. VPL has been extensively studied in peripheral vision while observers maintain fixation. To bridge the gap between visual and oculomotor systems in VPL, we investigated whether and how microsaccades change after training in an acuity task under exogenous (focused) or distributed (neutral) attention. [Methods] Twenty-six observers performed a Landolt acuity task during 5 days, and were assigned to the Neutral or Attention group. Each trial began with a fixation period followed by a neutral or a valid precue. After a brief ISI, two Landolt squares were presented briefly along the top-left/bottom-right or top-right/bottom-left diagonal. Observers reported the gap side of the target stimulus indicated by a response cue. 7 gap sizes were used to assess trial difficulty. Gaze-contingent eye-tracking was employed throughout the experiment. Microsaccades were detected with a standard velocity-based algorithm. [Results] Both groups showed improvements at the trained diagonal and learning transferred to the untrained diagonal. Microsaccades were suppressed during stimulus presentation, and rebounded during the response cue. For both groups, during the response cue microsaccade directionality was biased toward the stimuli diagonal, and this bias increased after training. Additionally, microsaccades were modulated by trial difficulty: they were more reduced for harder trials following the response cue, and such reduction was also found post-training during the response cue. [Conclusion] This study reveals that microsaccade rate and directionality change with VPL in a specific time window. After training, microsaccades are more biased toward the stimulus location(s), and further decrease for harder trials during the response cue, suggesting a functional role of microsaccades during VPL.