Abstract
After prolonged exposure to a female face, faces that had previously seemed androgynous are more likely to be judged as male. Research into such high-level categorical aftereffects has predominantly focussed on human faces, but more recent research suggests that such effects are generic, impacting judgments involving multiple object classes. High-level categorical aftereffects are thought to be caused by adaptation within a norm-based opponent code, akin to low-level analyses of colour. While a good deal of evidence is consistent with this, some recent data is contradictory, motivating a more rigorous test. In behaviourally matched tasks we compared the characteristics of aftereffects generated by adapting to colour, to expanded or contracted faces, to facial gender, and to different species of animal. In our experiments opponent coding predicted that the appearance of the adapting image should change and that adaptation should induce symmetrical categorical boundary shifts. This combination of predictions was firmly supported for colour adaptation, but not for any other condition. Interestingly the two face aftereffects tested were caused by distinctly different patterns of response shift relative to colour adaptation, and relative to each other. Our data suggest that high-level categorical aftereffects are not caused by adaptation within an opponent code. Instead, superficially similar aftereffects seem to ensue from several different combinations of visual adaptation within a population coding scheme, contrast effects, and/or changes in decision-making criteria.
Meeting abstract presented at VSS 2012