Shifa Maqsood¹ (), Maryam Vaziri-Pashkam¹; ¹Movement and Visual Perception Lab, Department of Psychological & Brain Sciences, University of Delaware

Humans have the ability to identify and categorise objects in a dynamic visual scene with remarkable accuracy. To achieve this accuracy, our visual system does not only rely on static information of the objects but it can also use dynamic cues. . In static images of objects it has been shown that turning the image upside down hinders recognition. Similar effects have been found in dynamic videos of human action. However, this inversion effect has not been investigated for a broader set of categories when only dynamic cues are available. Here we used object kinematograms to investigate the inversion effect on object recognition using dynamic cues. We used videos of 3 animate and 3 inanimate categories of objects, extracted motion cues from the videos and projected them on randomly moving dots to create object kinematograms that only contain dynamic information. We showed the videos either upright or inverted to participants in an online experiment. For each video, participants were asked to answer three questions: 1) identify if the stimulus was an animal or an object, 2) identify the category to which the object belonged and 3) name the object. Our results showed that the accuracies were higher for animals than objects (p < 0.0001). We observed a significant inversion effect for both animate and inanimate objects (p < 0.001) with no difference between the two (p = 0.1336 ). In addition, the classification accuracy for identifying the 6 objects categories was higher in the upright than the inverted condition (p <0.001 ) and the ability to name the object was significantly hindered when the videos were inverted (p <0.001 ). These results document for the first time strong inversion effects and suggest the role of configural processing in object categorization based on dynamic cues.