Internal models of the dynamic world

To interact with other people and with the dynamic world in general, our brain needs to continuously update its representation of external information and generate predictions of future states. Without such predictions, there would be a substantial time lag between states in the real world and the perception of, and reaction to, these states. A fundamental assumption is that the brain constantly generates and updates internal models of the world. However, the representational nature of internal models at different processing levels, and how the dynamics of internal models temporally relate to (e.g. follow or predict) actual events in the real world, remains unknown. 

To investigate the representational dynamics in the brain in response to dynamic events, we developed dynamic representation similarity analysis (dRSA) - which extends 'classic' RSA by using temporally variable models. This approach allows testing for each time point of a dynamic event how different aspects - from lower-level visual and kinematic features to higher-level semantic features - are represented in the brain relative to the modeled event (i.e., in a lagged or predictive manner). Thereby, we aim at understanding how the brain constructs internal predictive models of the dynamic world.