Abstract
Motion perceptual learning had been found with task-irrelevant motion stimuli. We further investigated this phenomenon using two motion tasks. In coherence detection, the precise motion direction of coherent dots was not critical. In direction discrimination, however, coherence detection was necessary to discriminate subtle motion directions. By using these highly similar, supra-threshold stimuli and closely related tasks, we hypothesized that task-irrelevant, direction-discrimination learning had the best chance in the detection task. The discrimination task served as a control because coherence-detection learning was expected in this task.
Method: two random-dot motion stimuli were presented, one without coherent dots, the other with a certain proportion of coherent dots moving along one of two possible directions. Pointed along the two's average direction was an uneven cross, serving as the fixation. In detection, participants identified the coherent stimulus. In discrimination, they determined whether the directional difference between the coherent motion and fixation was clockwise. The detection psychometric curve was measured as a function of coherence, with a constant directional difference of ±8°. The discrimination psychometric curve was measured as a function of directional difference, with detection coherence at 95% correct. Participants were trained with either task. For detection training, the coherence was maintained at 80% correct by staircase while the directional difference was at the pre-training, 80% correct threshold; and vice versa for discrimination training. After eight days' training, psychometric curves were re-measured.
Participants trained with discrimination improved on both discrimination and detection psychometric measurements; whereas those trained with detection improved only on detection, but not discrimination.