Abstract
Perceptual learning is generally known to be highly location specific in that sensitivity-improvement are greatest at the location of a learned task. Our recent research has found that task-irrelevant visual features can be learned when those features are presented during an attentionally demanding task (Watanabe et al. 2001; Watanabe et al. 2002). This task-irrelevant learning is observed only when the presentation of the learned feature is temporally paired with the targets of training task (Seitz and Watanabe 2003). However, spatial and temporal extent of task-irrelevant learning signals has yet to be clarified. In this study, we presented stationary Gabor patches on a noisy background while subjects performed a rapid serial visual presentation (RSVP) task. We measured the performance of orientation detection for the Gabor patches at two different spatial locations and in four different timing relative to task targets. By comparing the performance before and after seven-day learning, we found significant improvement only for Gabor stimuli that had been presented spatially close to the attended location. The results show that while task-irrelevant perceptual learning is not specific to a task location, the spatial extent is limited. Preliminary results indicate a broad temporal window for learning. These suggest an underlying mechanism for task-irrelevant perceptual learning that has rough spatial-temporal tuning.
This research was supported in part by the National Institute of Information and Communications Technology (NICT), NSF (BCS-9905914), NIH (R01EY015980-01), Human Frontier Research Fiundation (RGP18/2004) and NSF (CELEST).