An empirical analysis of the methodology of automatic imitation research in a strategic context

National audienceSince the discovery of the mirror neuron system, it has been proposed that the automatic tendency to copy observed actions exists in humans and that this mechanism might be responsible for a range of social behavior. A strong argument for automatic behavior can be made when actions are executed against motivation to do otherwise. Strategic games in which imitation is disadvantageous serve as ideal designs for studying the automatic nature of participants' behavior. Most recently, Belot, Crawford, and Heyes (2013) conducted an explorative study using a modified version of the Rock-Paper-Scissors game, and suggested that in the case of asynchrony in the execution of the gestures, automatic imitation can be observed early on after the opponent's presentation. In our study, we video recorded the games, which allowed us to examine the effect of delay on imitative behavior as well as the sensitivity of the previously employed analyses. The examination of the recorded images revealed that more than 80% of the data were irrelevant to the study of automatic behavior. Additional bias in the paradigm became apparent, as previously presented gestures were found to affect the behavior of the players. After noise filtering, we found no evidence of automatic imitation in either the whole filtered data set or in selected time windows based on delay length. Besides questioning the strength of the results of previous analyses, we propose several experimental and statistical modifications for further research on automatic imitation