Abstract: PURPOSE: In typical visual perceptual learning (VPL), only a specific stimulus is practiced and learning is often specific to the trained feature. Previously we demonstrated that multi-stimulus learning (e.g., TPE procedure) has the potential to achieve generalization. However, it is still unclear whether feature variability plays a role in learning generalization in multi-stimulus learning.
METHODS: We adopted a feature detection task, in which an oddly oriented target bar differed by 16° from the background bars. The SOA threshold between the target and the mask was measured with a staircase. Observers were trained with four orientation search stimuli either with a 5° deviation (30°-35°-40°-45°) or with a 45° deviation (30°-75°-120°-165°). The transfer of learning to the swapped target-background orientations was evaluated after 5 sessions of training.
RESULTS: (1) Multi-stimulus training with a 5° deviation resulted in significant learning improvement, but learning failed to transfer to the swapped target-background orientations. (2) In contrast, training with a 45° deviation resulted in a significant learning transfer to swapped orientations. (3) A modified TPE procedure, in which observers were trained with four orientation search stimuli with a 5° deviation and simultaneously passively exposed to orientations with larger orientation variability (45° deviation), resulted in significant orientation learning generalization.
CONCLUSIONS: Training with large feature variability enables generalization to untrained features, by eliminating the “overfitting” of specific stimuli in typical VPL paradigms and endowing the visual system with more robustness toward changes in the stimulus material. Our results could motivate the development of efficient training paradigms in clinical.

Key words: perceptual learning, feature variability, specificity, transfer