In this study, we explored the relation of two different measures used to investigate infants' expectations about goal-directed actions. In previous studies, expectations about action outcomes have been either measured after the action has been terminated, that is post-hoc (e.g., via looking time) or during the action is being performed, that is online (e.g., via predictive gaze). Here, we directly compared both types of measures. Experiment 1 demonstrated a dissociation between looking time and predictive gaze for 9-month-olds. Looking time reflected identity-related expectations whereas predictive gaze did not. If at all, predictive gaze reflected location-related expectations. Experiment 2, including a wider age range, showed that the two measures remain dissociated over the first 3 years of life. It is only after the third birthday that the dissociation turns into an association, with both measures then reflecting identity-related expectations. We discuss these findings in terms of an early dissociation between two mechanisms for action expectation. We speculate that while post-hoc measures primarily tap ventral mechanisms for processing identity-related information (at least at a younger age), online measures primarily tap dorsal mechanisms for processing location-related information.
Keywords: action perception; dorsal-ventral; eye movements; infancy; looking time; predictive gaze.