Abstract: PURPOSE: According to the Kinematic Specification of Dynamics (KSD) hypothesis, dynamics causes movements, movements exhibit kinematics in the world which projects to the optics forming information known as the optic flow, which underlies visual perception. In addition to dynamics, optical input during movement performance also affects kinematics. For example, when we walk with a heavy backpack, we may look different from without; when we walk with the eyes closed, we may look different from when the eyes being open. With this addition, we propose a revised KS(D+O) hypothesis, which emphasizes on the influences of both the dynamical and optical input during action production on observable movement patterns and hence the optical information and visual event perception. To test this new hypothesis, we designed a visual search task where observers looked at point-light displays of actors, who performed actions with the eyes open or closed, and reported the presence /absence of an open-eyed/closed-eyes actor.
METHODS: A total of 768 point-light display videos of 4 actions (walking, running, jumping and adapted-Y-balance) were recorded from the frontal and side views. While performing these actions, actors’ eyes were either open or closed. We put these PLDs into a visual search task with set sizes equal to 2, 3, 4, or 5. Twenty participants watched the stimuli and reported whether an open-eyed PLD actor was present/absent among close-eyed PLD actors or vice versa. Their response accuracy and time were recorded.
RESULTS: Participants were able to differentiate whether optical feedback was available during action performance (mean correct = 54.8%, SD = 0.025, chance level = 50%). First, views of presentation (frontal vs. side) did not affect search performance (P < 0.001). Moreover, a repeated measures ANOVA showed that set size, action type and target present/absent all affected accuracy, and accuracy decreased as set size increased and was higher for target present trials. The ANOVA also revealed some significant interactions, including target present/absent * set size, action type * set size, and target present/absent * search for eye-open/closed target. Finally, a repeated measures ANOVA showed that the response time was affected by set size, action type, target present/absent and longer RT was related with larger set size, and target absent trials. The significant interactive effect on RT included target present/absent * set size.
CONCLUSIONS: From the movement alone, observers were able to find one (among many) actor with their eyes open or closed. This supported our hypothesis that optical input during action performance affected kinematics and the distinction (with or without optical input) could be picked up by observers. Thus, the KSD hypothesis can be extended to include the optical component and, in general, what affects action is perceptible through the kinematics.

Key words: KSD hypothesis, optical input, movement kinematics