Data analysis

During the experiment, for each participant we recorded 40 actions in response to the iCub requests (10 gentle requests with happy face, 10 gentle requests with angry face, 10 rude requests with happy face and 10 rude requests with angry face). Each motor response performed by participants was divided in two phases of interest: the reaching phase, during which participants reached the little ball and grasped it, and the passing phase, during which participants moved the little ball towards the monitor and positioned it on the requested target. For both reaching and passing phases, we extracted specific kinematic parameters: peak velocity, peak acceleration, z-coordinate trajectory (representing how much participants raised their right hand), action phase duration and time to peak velocity. For the reaching phase, we also calculated the maximum aperture of the right hand. For each participant and for each kinematic parameter, we averaged the 10 values recorded for each condition. Then we normalized these averaged values with the average of the baseline condition in which participants performed the task without receiving any request (see above). In this way, for each participant and for each kinematic parameter we obtain a single value per condition. Finally, these averaged and normalized values relative to the extracted kinematic parameters were organized to carry out a General Linear Model (GLM). The GLM considered VITALITY (gentle and rude) and FACIAL EXPRESSION (happy and angry) as two factors of interest. This model allowed us to investigate whether and how facial expressions of the iCub robot and its action vitality forms could influence the motor response of participants in terms of kinematics. All kinematic data were normalized to the baseline condition, in which participants performed the task without receiving the iCub request before. In order to investigate whether and how facial expressions and action vitality forms of the iCub robot could influence the motor response of participants, we measured possible differences of kinematic features characterizing their actions. To this aim, for each parameter, data were organized to carry out a General Linear Model (GLM), with VITALITY (gentle and rude) and FACIAL EXPRESSION (happy and angry) as two factors of interest. In addition to kinematic features, we computed the reaction time, i.e. the time elapsing between the end of the iCub request and the starting movement of participants. Also in this case, we analysed possible differences of reaction times among conditions by organizing data in a GLM with VITALITY and FACIAL EXPRESSION as factors of interest.