**Statistics in motion: Does the infant motor system predict actions based on their transitional probability?**
**This study was submitted and in-principle accepted as registered report by Developmental Science.**
Motor theories of action prediction propose that our motor system combines prior knowledge about the most likely outcomes of an action with incoming sensory input to predict other people's behavior. This prior knowledge can be acquired through observational experience, with statistical learning being a candidate mechanism for how humans learn and generate action predictions observing novel and unfamiliar sequences. Recently, it has been shown that the infant motor system uses knowledge from statistical learning to predict deterministic upcoming actions. Here, we build upon these results by asking if during action prediction the activity of infant’s motor system reflects the specific statistical likelihood of upcoming actions to generate action predictions. We will train infants at home with videos of an unfamiliar action sequence featuring different transitional probabilities. At test, motor activity will be measured using EEG and compared during perceptually identical time windows preceding actions linked by four levels of probability (deterministic, high, medium, low). We hypothesize that the activity over the motor area will be parametrically modulated by the transitional probability of action pairs. These results will strengthen previous evidence that the activity of the motor system reflects the specific statistical likelihood of upcoming actions to generate action predictions. Moreover, this study will provide evidence for a functional role of statistical learning for infants developing action understanding.
For a preview of the analysis scripts please visit [the github repository.]