Abstract
We previously showed that actively practising to discriminate the orientation of a line-texture bar increases its saliency and modulates the ERP-component amplitudes reflecting texture segmentation, suggesting that increased saliency results from inhibition of incongruent local orientation (orthogonal to the bar) (Casco et al., 2004).
We now show an experience-dependent increase of saliency even when the texture is an irrelevant background to a central task engaging subject's attention. However, this improved perception in single randomly chosen target trials (Mack, Tang, Tuma, Kahn and Rock, 1992) does not occur in the two groups where, unexpectedly, orientation discrimination of the bar was asked either at the beginning (50% accuracy) or at the end (60% accuracy) of the 300-trials block. Instead, the group for whom the target trial was presented in a second block executed with a delay of several hours and at least one night's sleep showed a significant increase of saliency (72.5 % accuracy) but only when local elements were congruent to the global orientation of the bar. If the elements were incongruent, accuracy dropped to 41%, indicating that increased saliency relies on a larger response to local orientation within the bar. The electrophysiological correlate of this task-irrelevant improvement is a selective ERP modulation on the second day, consisting in a significant increase in the segmentation-component amplitude, but only when elicited by the congruent bar condition.
These results show that perceptual learning may not require attention to be directed on the stimulus. Task-relevant and task-irrelevant learning involve different mechanisms: the former based on inhibition of information irrelevant to the task (Casco et al., 2004) and leading to explicit figure-ground segmentation, the latter based on enhancement of response to local orientation in the texture bar, also resulting in figure-ground segmentation.