Predicting others’ actions is crucial for acting in a social world. How we come to these predictions is hotly debated. The simulation account states that the motor commands used to predict the sensory consequences of our own actions are also used to predict others’ actions. In contrast, rationality theory suggests that predictions of others’ actions are based on inborn, abstract rules and thus rely on a different mechanism than predictions of our own actions. To address this controversy, I suggest to study how predictions of own and others’ actions develop during infancy. In Study 1, infants of 6- and 12-months of age will reach and grasp for objects, and observe someone else reaching and grasping for objects, while their hand and eye movements are tracked. Hand and eye movement tracking can reveal whether and when infants predict the goal of the action. Results will reveal whether the two types of predictions develop in synchrony. Study 2 tests the congruence in brain activation between reaching performance and observation while infants’ brain responses are measured using fNIRS. Study 3 will verify whether the basic mechanism found for predicting others’ simple reaching actions also applies to a real-world situation involving multiple objects and a sequence of reach and grasp actions, such as sandwich making. To this end, 3-year-old children will be asked to make a sandwich and to observe someone else making a sandwich while their eye and hand movements will be tracked. Taken together, the three proposed studies will show whether the same or a different neurocognitive mechanism serves to predict own and others’ actions and whether this mechanism extends to real-world applications. If so, this mechanism can enable people to understand each others’ behaviour. The findings will be informative for psychology at large as the mechanism for understanding others plays a fundamental role in social interaction.