Kinematic cues and recognition of self-generated actions

Abstract

In the present study, we addressed the issue of whether healthy individuals can recognize a given gesture as their own, based on kinematic information. To this purpose, we required 36 volunteers to execute a series of hand movements of increasing complexity, while their kinematics was recorded by a motion-capture system. In a later session, we showed them a series of computer animations where a virtual hand, rendered as a simple stick-diagram, was animated by the kinematics recorded from the participants in the previous session. Their task was to recognize their own movements, choosing from three alternatives. To test the contribution of various potential cues to action recognition, the roles of (1) access to motor representation, (2) gesture complexity, and (3) familiarity effects were separately investigated. The results support the hypothesis that kinematic templates rather than single motor parameters contribute to self-recognition in the absence of morphological cues.

Appendix 1

Analyses of kinematic parameters and selection of distractor-participants

Spatial and temporal parameters were analyzed as follows. The time course of the displacement of the markers placed on the fingertips was used to compute the beginning and end of each movement. The single data points were plotted and the waveform of each movement was analyzed by detecting all peaks of the function. Movement start was defined as the sample corresponding to the origin of the rising slope of the first peak, where the difference between two successive points significantly exceeded a constant noise factor (0.001) for the relevant marker (I4 for IT, IF and TL; the earliest finger-tip marker to show a stable displacement for HT and FT). Likewise, the temporal sample corresponding to the end of the falling slope of the last above-noise peak defined the end of the movement. Amplitudes (and angles) were defined on the basis of markers’ displacement compared to the plane of the table for the detected peaks. For all gestures, total movement time (time difference between the start and end of the movement) was computed. For IF, maximal angle of finger flexion was also analyzed. For the three tapping gestures, inter-tap intervals and maximal amplitude of hand/finger lift were computed. For TL, length of the line, peak velocity of the displacement of the marker on the index fingertip and elevation of the little finger on the plane of the table were measured. Means and variability across subjects for the analyzed parameters for the five gestures are reported in Table 1.

Table 1 Means and SD for main spatial and temporal parameters recorded in Experiment 2 (as a measure of overall variability in gesture execution for the group)

Based on this information, distractors for each participant were selected according to four constraints. First, main hand/finger dimensions differed by less than 0.5 cm. Second, for each gesture and for at least a half of the analyzed parameters, mean values recorded from the distractors’ movements were within a tolerance interval equal to the range defined by the observer’s movements. Third, mean values recorded from the distractors’ movements never exceeded this range by more than 25% for the remaining parameters. Finally, this was true for at least four out of five gestures.