Breaking it down to build it up: Hierarchical task learning and representation in the anterior cingulate cortex

01 November 2022 → 31 October 2024
Research Foundation - Flanders (FWO)
Research disciplines
  • Social sciences
    • Psychophysiology
    • Cognitive processes
    • Learning and behaviour
Electroencephalography Hierarchical Reinforcement Learning Cognitive Computational Modeling
Project description

The function of the anterior cingulate cortex (ACC) has been well-studied but little understood by cognitive neuroscientists over the last decades. Lesions of the ACC can lead to a wide variety of mental disorders including depression, obsessive compulsive disorder, and schizophrenia. A novel theory of the computational function of the ACC proposes that it is involved in learning and executing hierarchically structured tasks according to the principles of model-based hierarchical reinforcement learning. I aim to test this claim by designing a novel goal-directed navigation task in a hierarchical environment. Participants will be able to simplify the task by representing goals according to their location in one of three ‘wings’ of a museum. Participants will have to learn this hierarchical organization through experience, which requires a specific hierarchically structured surprise signal that I hypothesize is generated in the ACC and can be probed with EEG. Observing this signal would be good evidence for the contribution of the ACC in building hierarchical task models. Additionally, the ACC is proposed to hold hierarchically-organized representations of task progress, which I will try to observe in fMRI. Observing these representations in the novel task will indicate the hierarchical learning mechanism observed in the EEG study is likely to contribute to the ability of the ACC to execute hierarchical action sequences and motivate task engagement or disengagement.