Some evidence in very recent psychological studies have demonstrated that motor simulation ability is crucial for the correct understanding of social intentions. The present study was conducted first to confirm that the nature of the motor intention leads to early modulations of movement kinematics. Then, we tested whether humans could read an agent’s intention when observing the very first element of a complex action sequence. Results revealed early variations in movement kinematics and further showed that human agents can use these deviants to distinguish above chance level between three different social actions. Similar performance levels were found using an artificial classifier (Neural Network) and this procedure demonstrated furthermore that decisions could be taken on the basis of information contained in the first 500ms of movement kinematics. Taken together these results confirm the importance of motor simulation for adapted social interaction, and suggest how robotic adaptive controllers may use as input low-level motor information (e.g. kinematics) to afford biologically inspired social behaviors.