Abstract
Goal: Perceptual learning (PL), training-induced perceptual improvement, is usually highly specific, limited to trained locations or features. Previously, we reported that covert attention helps generalize PL to an adjacent, attended location (VSS 2009). Here we examined the effect of attention on PL at attended locations as well as unattended locations in a different quadrant of the visual field. Procedure: Participants performed an orientation discrimination task, indicating whether a briefly-presented Gabor patch of variable contrast was oriented clockwise or counterclockwise from vertical. We derived psychometric functions and calculated 75% contrast thresholds. In an initial testing session, performance was measured at four locations arranged in an imaginary square around fixation. This was followed by five training sessions, in which stimuli only appeared at two locations, placed diagonally from each other. A final testing session, identical to the first, assessed performance and improvement at all four locations. There were two groups of participants, which differed only during training sessions: For the attention group, a briefly-presented transient precue directed attention to the Gabor’s location. For the neutral group, a similar cue was presented at fixation. Testing sessions were identical for the two groups and employed fixation cues. Results: During training, both groups’ performance improved but improvement was greater for the attention group. Comparing the initial and final testing sessions, we found a standard PL pattern for the neutral group: performance improved at trained locations, but was unchanged at untrained locations. For the attention group, however, we found not only a large improvement at trained locations (threshold improvement was double that of the neutral group), but also a large improvement at the distant untrained locations (equivalent to the neutral group’s improvement at trained locations). We conclude that attention enhances PL at trained locations, and also transfers it to unattended, untrained locations in a different quadrant.
Meeting abstract presented at VSS 2012