Recognising novel deforming objects

Abstract

Visual object perception is dynamic as the result of an active observer or movement in the environment. Nonetheless, contemporary theories of object recognition commonly focus on how objects are represented in terms of their static properties, e.g., shape. When the contribution of motion has been considered, it has typically been with regards to how rigid rotation in depth could facilitate the reconstruction of an object’s static 3-D shape (e.g., Ullman, 1979). Here, we propose that human observers represent an object’s dynamic properties as a cue to its identity, independent of its contribution to shape perception. This information may facilitate recognition when shape information is less reliable, e.g., viewpoint variations.