Abstract
Goal. Perceptual learning is a long lasting improvement in the perception of a stimulus following training. In vision and audition, studies have shown that irrelevant training does not yield learning. However, a recent audition study showed that a given amount of training on one task (frequency discrimination) that is insufficient for learning on its own yields learning when alternated with training on another task (duration discrimination) with a common stimulus (Wright et al. 2010). Here we explored whether a parallel effect is present in vision: Can alternating training between two tasks help learning? Methods. We used orientation (task-A) and spatial frequency (SF, task-B) 2AFC discrimination tasks. On each trial two Gabor patches appeared, a standard and a comparison, and observers compared either their orientation or their SF, depending on the block. During pre- and post-tests, observers were tested on both tasks with a standard Gabor (4cpd oriented 30° or 120°). We trained three groups for three intermediate days with the 30° standard stimulus. Training sessions differed across groups: Group ABAB alternated between the orientation and SF tasks (400 trials/task/day). Groups A–A– and –B–B trained on one task alternating with rest (400 trials/day for each group). Results. We found a significant interaction across groups: Group ABAB had significant learning in both tasks for the trained stimulus, and learning generalized to the untrained orthogonal stimulus in the SF task. In contrast, the other groups (A–A– and –B–B) did not exhibit any learning. Conclusion. As in audition, alternating training between two tasks with the same standard stimulus–even when the same amount of training on each task on its own is ineffective–can facilitate visual learning. These findings suggest a novel visual training paradigm in which task alternation enables learning for both tasks.
Meeting abstract presented at VSS 2013