Motion matters: learning dynamic objects

Previous research has typically focused on static properties of objects. Recently there has been a growing interest in the role that dynamic information might play in the perception and representation of objects. In this talk we approach this issue by describing how the visual system utilises dynamic information in learning two different classes of visual objects: i) novel deforming stimuli, ii) faces. Object-learning experiments with novel objects show that human observers are sensitive to the motion characteristics. In addition, preliminary results also suggest that learned motion characteristics can reduce the detrimental effects of changing the studied viewpoint. Using faces, we explored how encoding of identity is affected by two different types of facial movements: non-rigid facial motion, and looming facial motion. Using a delayed visual search paradigm we could show that faces learned in motion were found more quickly and more accurately than faces learned from static snapshots. In summary, results from our lab suggest that the visual system uses dynamic information to encode and subsequently recognize new object/face identities