Abstract
In natural vision, the visual system must dynamically integrate form and motion. Here we explored the influence of an object's form on the perception of its motion. In a 2IFC paradigm, observers judged whether or not the motion of a target object was more natural in comparison to a reference object, a rectangle. The target motion had seven different levels of biological-ness, obtained by parametrically varying the amount of minimum jerk (MJ) and constant velocity (CV) in the trajectory. The rectangle always moved with 50% MJ and 50% CV. In Experiment 1, the target objects were a robot hand and a human hand. While these objects had similar shape, the former was noticeably metallic and artificial. Comparing the response curves revealed the robot hand needed significantly less biological motion (% MJ) than the human hand for its motion to be judged at the same level of biological-ness. We hypothesized that this effect might be driven by predictive influences; e.g., the robot hand might evoke prediction for clunky, mechanical movement, leading to its actual motion appearing more natural than it is. In Experiment 2, we included two more conditions: car (an inanimate object that moves) and house (an inanimate object that is typically stationary). Interestingly, responses for the car and house conditions did not differ from the human. For the robot hand, once again, significantly less biological motion was needed for naturalness judgments. Thus, while the form of an object can influence biological-ness judgments of its movements, there are constraints on this effect. We suggest that an object may need to be sufficiently similar to a biological object for biological motion prediction mechanisms to be able to influence its perception. Data thus far indicates similarity of shape (and not color or texture) might constrain interactions between form and biological motion processing.
Meeting abstract presented at VSS 2014