Research has shown that we process other people’s actions by simulating them in our own motor system. This means that we represent observed actions as if we executed them ourselves. However, social life requires us to process not only observed actions but also observed interactions. Could interaction representation likewise rely on motor simulation? In my PhD, I showed that the motor system can simultaneously represent multiple observed actions. Yet, representing multiple actions is not the same as representing interactions. Therefore, in this project, I aim to combine fMRI with state-of-the-art analysis techniques to test whether motor processes contribute to understanding observed interactions. First, I will investigate if it is possible to predict from brain activity in the motor system which kind of interaction participants observed (e.g. two persons celebrating or rather two persons having an argument). If confirmed, this would mean that participants represented the actions of both persons in their motor system and that this system hence “knew” what type of interaction was observed. However, it also begs the question how the brain then knows which person performed what action. Therefore, next, I will investigate how the brain keeps track of which agent is doing what. Together, these two studies provide a highly novel, innovative contribution to the question of how we are able to understand social interactions between others simply by looking at them – a key social skill.