Inference of complex human motion requires internal models of action: behavioral evidence.

Previous behavioral investigation from our laboratory (Pozzo et al. in Behav Brain Res 169:75-82, 2006) suggests that the kinematic features influence the subject's capacity to estimate the final position of simple arm movement in which the last part of the trajectory is hidden. The authors argue the participation of internal information, as the kinematic parameters, to compensate the lack of the visual input. The purpose of this report was to verify if the dependency of visual motion inference to biological displays can be generalized for intransitive and complex human motions. To answer this question, the subjects were asked to estimate the vanishing and final position of the shoulder trajectory of Sit to Stand (STS) or Back to Sit (BTS) motion performed in the sagittal plane, according to a biological or nonbiological kinematics. The last part of the trajectory (i.e., 35%) was occluded. We observed a kinematic effect on the precision of individuals' estimation. The subjects were more precise and less variable to estimate the end trajectory with biological velocity profiles. Moreover, impoverished visual information appeared sufficient to evaluate the final position of an intransitive complex human motion. These results suggest the participation of internal representations to infer the final part of complex motion. We discuss the results in the light of possible neural substrates involved during the inference task.