Abstract
People can rapidly and efficiently categorize the animacy of natural objects. Does this imply that the visual system has an unlimited capacity for processing animacy across the entire visual field? To test this, we generated a set of morphed “animacy continua” between animate and inanimate silhouettes. Observers were shown different ratios of these items (2:1, 10:3) and judged whether a set was more animate/inanimate. The distribution of items was either “segmentable” (including only 100% animate/inanimate items) or “non-segmentable” (50% ambiguous morphs were also presented). In Experiments 1 and 2 (N=61), observers failed to integrate animacy from multiple items, as they showed very poor performance and were not sensitive to the distribution type, despite an almost perfect ability to categorize individual objects. The same manipulation with color as a category-defining feature elicited both good individual and ensemble categorization performances and the segmentability effect. In Experiment 3, we parametrically manipulated feature discriminability in both color and animacy to ensure our results were not based on different discriminability. We always presented “segmentable” distributions with a 2:1 numeric ratio and used two levels of color discriminability (easy and hard – 30˚ and 15˚ apart along a CIELab color wheel) and three levels of animacy discriminability (hard – ambiguous morphs of objects, easy – morphed but not ambiguous objects, non-morphed – original silhouettes of real objects). The results again showed that observers were bad at ensemble animacy categorization even at high discriminability and rather good at the color ensemble task even at low discriminability (although they were almost perfectly accurate at individual categorization for both features). Thus, discriminability alone cannot explain the differences in performance for two dimensions. Overall, we conclude that good individual categorization does not always allow people to build strong ensemble animacy representations, which suggests the limited capacity of animacy perception.