Learning joint actions in human-human interactions

Understanding human-human interactions during the performance of joint motor tasks is critical for developing rehabilitation robots that could aid therapists in providing effective treatments for motor problems. However, there is a lack of understanding of strategies (cooperative or competitive) adopted by humans when interacting with other individuals. Previous studies have investigated the cues (auditory, visual and haptic) that support these interactions but understanding how these unconscious interactions happen even without those cues is yet to be explained. To address this issue, in this study, a paradigm that tests the parallel efforts of pairs of individuals (dyads) to complete a jointly performed virtual reaching task, without any auditory or visual information exchange was employed. Motion was tracked with a NDI OptoTrak 3D motion tracking system that captured each subject’s movement kinematics, through which we could measure the level of synchronization between two subjects in space and time. For the spatial analyses, the movement amplitudes and direction errors at peak velocities and at endpoints were analyzed. Significant differences in the movement amplitudes were found for subjects in 4 out of 6 dyads which were expected due to the lack of feedback between the subjects. Interestingly, subjects in this study also planned their movements in different directions in order to counteract the visuomotor rotation offered in the test blocks, which suggests the difference in strategies for the subjects in each dyad. Also, the level of de-adaptation in the control blocks in which no visuomotor rotation was offered to the subjects was measured. To further validate the results obtained through spatial analyses, a temporal analyses was done in which the movement times for the two subjects were compared. With the help of these results, numerous interaction scenarios that are possible in the human joint actions in without feedback were analyzed.