Abstract
[Goal] Visual perceptual learning (VPL) refers to improvements in sensory discrimination due to repetitive practice. VPL has been extensively studied at peripheral locations, while observers fixate at the center. Microsaccades are fixational eye movements with amplitudes ≤1 degree. The role of microsaccades has been established in a variety of perceptual tasks, but it is unknown whether and how microsaccades change along with human perceptual learning. Here we investigated whether and how microsaccade rates vary during and after VPL when observers trained under feature-based attention (FBA) or a neutral condition. [Methods] Observers performed an orientation discrimination task. Each trial began with a fixation period followed by a cue. After an ISI, a Gabor stimulus was presented for 200ms, followed by a response window. Five offsets were used to assess different difficulty levels. Twenty observers participated in a six-day VPL study: 10 in a neutral condition and 10 in a FBA condition. Eye fixation was monitored by an eye-tracker. Microsaccades were extracted with a standard velocity-based detection algorithm. VPL was re-assessed after ~3.5 and ~12 months. [Results] Performance improved at the trained location with the trained feature for both groups, and learning transferred to an untrained location for the attention group. Both groups showed a typical pattern: microsaccades were largely suppressed during the stimulus presentation, followed by a rebound during the response window, regardless of task difficulty. The rebound of microsaccade rates was greatly reduced after VPL, and this reduction gradually emerged during training. This pattern remained after ~3.5 months for both groups and remained after 1 year only for the Attention group. [Conclusion] This study reveals that the rates of microsaccades change after perceptual learning, and that such a change is long lasting, and more so when observers trained with FBA.