The contribution of qualitative and quantitative shape features to object recognition across changes of view

Abstract

Two experiments investigated the influence of qualitative and quantitative shape features on recognition of novel, four-component objects. Quantitatively different objects had different connection angles between the components. Qualitatively different objects had different connection angles and differently shaped components in some of the four positions. Old-new recognition declined less with changes of view for qualitatively different objects (Experiment 1). However, recognition of these objects was made to decline sharply with changes of view if subjects were biased to attend to the connection angles rather than the component shapes (Experiment 2), suggesting that the influence of different features depends on visual experience with those features. These results favor a feature-based model of shape representation that utilizes multiple feature types and that can rely on different features depending on particulars of the objects and the task.