Abstract
It has been shown that crowding could be alleviated by perceptual training. The underlying mechanism of the phenomenon has yet to be clarified. Here we tackled this issue through investigating the specificity and generalization of perceptual learning of crowded orientation. Experiment 1 consisted of five phases – pre-test, training, mid-test, training and post-test. Subjects were trained to perform an orientation discrimination task with a target grating (radius: 1.5°; spatial frequency: 2 cycles/°; contrast: 1; orientation: 22.5° or -67.5°; eccentricity: 10°). The target was flanked by two other gratings positioned radially, which were identical to the target except that their orientations were randomized. In the tests, subjects’ orientation discrimination thresholds were measured with the target and its orthogonal version, which were presented either alone or with the flankers. After practicing about 1700 trials, subjects’ performance was improved significantly. Surprisingly, the crowding effect could be completely eliminated – subjects performed equally well when the target was present alone or with the flankers. More importantly, the improvement can completely transfer to the crowded orthogonal grating, but the transfer to the isolated gratings was weak. Subjects were further trained for 7200 trials over six days. Their performance improvement was found to be largely specific to the trained orientation. Experiment 2 further showed that the orientation discrimination training in Experiment 1 could even completely break motion crowding. Experiment 3 demonstrated that training had little effect when the trained and test stimuli were in different quadrants. These findings suggest that perceptual learning of crowding orientation comprises two stages. In the early stage, subjects acquired a relatively general ability to separate the target and the flankers. In the late stage, they learn to process the target information more accurately and more specifically.
Meeting abstract presented at VSS 2013