Abstract
Perceptual learning training (PL, e.g. Karni & Sagi, 1991; Sigman et al., 2000) typically leads to learning specific to trained stimuli and task. In contrast, perceptual expertise training (PE, e.g. Gauthier, et al., 1997; 1998) leads to greater generalization, both for new objects in the trained category and in new tasks. It is unknown whether the contrasting training effects result from differences in stimuli, training tasks or presentation conditions (e.g., foveal vs. peripheral presentation). We investigated the role of the training task by comparing eight-hour PL and PE learning protocols with identical objects in the same peripheral visual positions. For PL training, participants searched for objects in a specific target orientation among an array of 8 identical objects rotated in 90°, 180° or 270°. PE training involved naming each individual shape. Over trials, the same stimuli, all from the same homogeneous category, were used in both tasks. Both types of training replicated the typical behavioral effects of previous studies. PL training led to behavioral improvement for the visual search task specific to the trained orientation and trained objects, while PE training resulted in improved shape matching performance for trained objects compared to control objects. Both types of training generalized to the untrained task, in contrast to the task specificity in previous PL studies (Fahle, 1997). Thus, observers learned from the irrelevant variability in shape across trials during PL training, and PE training generalized to an orientation task in a more crowded display. In sum, by manipulating training experience with identical training objects, training visual positions and testing tasks, our PL and PE training replicated the typical training effects, and demonstrated divergent behavioral generalization after each type of training. This suggests that the contrasting training effects of PL and PE training can be obtained when stimuli and presentation conditions are matched.