This project investigates the cognitive status and role of sense of control (SoC) in complex task situations. To that end, it investigates how embodied cognitive agents create a SoC when receiving complex or conflicting feedback about effects of their actions, how SoC is taken into account when deciding upon actions in an uncertain situation, and whether SoC is task-specific and involved in characterising different tasks in memory, learning, and multi-task action control. In studying these questions, the project will extend its computational cognitive architecture (developed in the first phase) and test it systematically against human behavioural data as well as measures of SoC, to gain a better understanding if and how SoC relates to agency and actual behaviour. Finally, the project will study how SoC arises in situated action control with multiple sensory input modalities. To that end, a predictive processing hierarchy at the sensorimotor layer as well as the cognitive layer of the cognitive architecture will be extended to process visual, acoustic and tactile information about the current action control situation. This will allow for exploring how we can computationally model SoC as it arises from conforming, deviating or conflicting prediction errors and precision in different input modalities.