Simon Steinkamp
Does cognition use a common reward map? An fMRI study of reward across multiple cognitive tasks
Simon Steinkamp, Danish Research Centre for Magnetic Resonance at Amger – Hvidovre Hospital, has received a Lundbeck Foundation Postdoc grant worth DKK 2,394,385.
About the project
In the 1990s, the discovery that dopamine neurons in the midbrain encode expectations of future rewards, changed our view on decision-making and learning. In a remarkable convergence with artificial intelligence, it was observed that dopamine activity aligns with the learning signal used by reinforcement learning algorithms. A recent study in mice shows that dopamine neurons are diverse in their reward expectation, with some being more optimistic than others.
How this ’distributional’ code is implemented in the human brain is currently unknown. In this project, Simon Steinkamp will collect functional magnetic resonance imaging data from the same participants performing a diversity of cognitive tasks, including sensory, navigational, motor, and reward paradigms. All these tasks rely on the reward system for learning. With this data, he will investigate whether a distributional reward code is commonly deployed across all these cognitive domains. He will use computational parametric mapping, a technique Simon Steinkamp himself has helped pioneer and which will be key to visualize whether and how the distributional code is used.
This project offers fundamental insights into the neural processes underlying learning and decision-making, and builds a novel framework that can be used for modelling and mapping disordered reward processing in neurological and psychiatric disorders.
