Skip to main content

Beyond dopamine: Characterizing the computational functions of midbrain modulatory neurotransmitter systems in human reinforcement learning using model-based pharmacological fMRI

Objective

Understanding how humans and other animals are able to learn from experience and use this information to select future behavioural strategies to obtain the reinforcers necessary for survival, is a fundamental research question in biology. Considerable progress has been made in recent years on the neural computational underpinnings of this process following the observation that the phasic activity of dopamine neurons in the midbrain resembles a prediction error from a formal computational theory known as reinforcement learning (RL). While much is known about the functions of dopamine in RL, much less is known about the computational functions of other modulatory neurotransmitter systems in the midbrain such as the cholinergic, norcpinephrine, and serotonergic systems. The goal of this research proposal to the ERC, is to begin a systematic study of the computational functions of these other neurotransmitter systems (beyond dopamine) in RL. To do this we will combine functional magnetic resonance imaging in human subjects while they perform simple decision making tasks and undergo pharmacological manipulations to modulate systemic levels of these different neurotransmitter systems. We will combine computational model-based analyses with fMRI and behavioural data in order to explore the effects that these pharmacological modulations exert on different parameters and modules within RL. Specifically, we will test the contributions that the cholinergic system makes in setting the learning rate during RL and in mediating computations of expected uncertainty in the distribution of rewards available, we will test for the role of norepinephrine in balancing the rate of exploration and exploitation during decision making, as well as in encoding the level of unexpected uncertainty, and we will explore the possible role of serotonin in setting the rate of temporal discounting for reward, or in encoding prediction errors during aversive as opposed to reward-learning.

Field of science

  • /medical and health sciences/clinical medicine/radiology/medical imaging/magnetic resonance imaging
  • /natural sciences/computer and information sciences/artificial intelligence/machine learning/reinforcement learning

Call for proposal

ERC-2009-StG
See other projects for this call

Funding Scheme

ERC-SG - ERC Starting Grant

Host institution

THE PROVOST, FELLOWS, FOUNDATION SCHOLARS & THE OTHER MEMBERS OF BOARD OF THE COLLEGE OF THE HOLY & UNDIVIDED TRINITY OF QUEEN ELIZABETH NEAR DUBLIN
Address
College Green
2 Dublin
Ireland
Activity type
Higher or Secondary Education Establishments
EU contribution
€ 1 841 404
Principal investigator
John O'doherty (Prof.)
Administrative Contact
Deirdre Savage (Ms.)

Beneficiaries (1)

THE PROVOST, FELLOWS, FOUNDATION SCHOLARS & THE OTHER MEMBERS OF BOARD OF THE COLLEGE OF THE HOLY & UNDIVIDED TRINITY OF QUEEN ELIZABETH NEAR DUBLIN
Ireland
EU contribution
€ 1 841 404
Address
College Green
2 Dublin
Activity type
Higher or Secondary Education Establishments
Principal investigator
John O'doherty (Prof.)
Administrative Contact
Deirdre Savage (Ms.)