Partner Representation in Competitive Interaction: Implications for Neurorehabilitation

During rehabilitation patients are required to perform movements that are difficult for them to do alone. By effectively promoting player engagement and motivation, competitive settings are potentially useful in rehabilitation. However, in these scenarios the mechanisms underlying adapting to each partner are poorly understood. Here we address competitive interaction by using dyadic haptic interfaces, where pairs of participants repeatedly play a ball game (penalty kick). We manipulated the amount of information each participant had about their partner location. We found that in different experimental conditions the participants changed their behavior. We compared these results with simulations based on a computational model that assumes optimal action estimation and selection. Our findings are consistent with the model predictions: the players seem to develop a model of the partner during the interaction and adapt the model to different conditions.