Abstract
Goal: Investigations of the time course of Perceptual Learning (PL) have postulated a fast-PL stage (within the first session) and a slow-PL stage (requires several sessions). However, the mechanistic differences between fast–PL and slow-PL are unclear. Recently, using an orientation visual-search task we found that slow–PL is specific to the trained dimension (orientation) but not to the irrelevant dimension (color), and that the specificity reverses for fast–PL. Here, to further explore and characterize fast-PL, we explored whether and how fast–PL relates to the binding of stimulus features by using a conjunction visual search. Method: Observers were trained to detect a target defined by an orientation X color conjunction among 25 line-elements. For example, they trained to detect a red line tilted 80° among 12 red lines tilted 50° and 12 green lines tilted 80°. After 720 training trials observers were tested with a target defined by a new combination of the features, in which the target swapped either its color, its orientation or both. Results: Training significantly improved observers' performance (d'). Learning transferred when target-feature swapped with a distractor its color or orientation; i.e., the performance remained at the level attained with training. However, there was no transfer when both target color and orientation swapped. Conclusions: These results suggest that fast-PL does not reflect an improvement in observers' ability to detect the specific features they were exposed to during training, but rather an improvement of their ability to bind the different features in the conjunction. This finding suggests that fast-PL is mediated by higher cortical areas, where features' representations are bound together, and that attention may play a role in fast-PL.
Meeting abstract presented at VSS 2014