Perceptual learning with complex visual stimuli is based on location, rather than content, of discriminating features

Exposure to complex checkerboards (comprising a common background, e.g., X, with unique features, e.g., A-D, that are placed in particular locations on the background) improves discrimination between them (perceptual learning). Such stimuli have been used previously to probe human perceptual learning but these studies leave open the question of whether the improvement in discrimination is based on the content or location of the unique stimuli. Experiment 1 suggests that perceptual learning produced by exposure to AX and BX transferred to stimuli that had new unique features (e.g., C, D) in the position that had been occupied by A and B during exposure. However, there was no transfer to stimuli that retained A and B as the unique features but moved them to a different location on the background. Experiment 2 replicated the key features of Experiment 1, that is, no transfer of exposure learning based on content but perfect transfer of exposure learning based on location using a design which allowed for independent tests of location- and content-based performance. In both the experiments reported here, superior discrimination between similar stimuli on the basis of exposure can be explained entirely by learning where to look, with no independent effect of learning about particular stimulus features. These results directly challenge the interpretation of practically all prior experiments using the same type of design and stimuli.