Abstract
Objects differ along a number of different dimensions (e.g., shape, color, size) but only some of these dimensions are relevant for successful categorization. Learning to categorize cars along relevant dimensions can lead to stable changes in the visual system. These changes involve stretching psychological space to make exemplars differing in relevant features more dissimilar than exemplars differing in irrelevant features. This selective stretching of psychological space along relevant dimensions is known as dimensional modulation. Here, we (1) provide further evidence that category learning causes stable dimensional modulation, observable in tasks where learned categories are irrelevant and (2) probe the time course of these effects. Participants learned to categorize morphed cars according to a relevant shape dimension while ignoring an irrelevant shape dimension. After four sessions of categorization training, EEG was recorded during a task requiring the identification of a single target car. Importantly, learned category boundaries were not relevant to this task so that participants had to attend to both relevant and irrelevant shape dimensions. The critical comparison was between two types of non-target cars: those that differed from the target along category-irrelevant dimensions and those that differed along category-relevant dimensions. If category learning caused stable dimensional modulation, then non-targets with irrelevant differences should appear more target-like than non-targets with relevant differences, eliciting more target-like ERPs. As predicted, targets elicited a larger selection negativity, a postero-lateral component sensitive to detection of target features, than non-targets and non-targets with irrelevant differences elicited a larger selection negativity than non-targets with relevant differences. Interestingly, category-relevance of perceptual differences did not modulate the P300, a later decision related component. Overall, the results suggest that category learning causes stimuli with relevant differences to be perceived as less similar than stimuli with irrelevant differences, even when attention is directed to both kinds of features.
Meeting abstract presented at VSS 2016