PS201INTERMEDIATEPSYCHOLOGY & PSYCHIATRY

Introduction to Computational Psychiatry

Mental health conditions understood through reinforcement learning and deep networks.

Course Information

Course Information
ModeInstructor-led or blended
TermFA1 · Aug–Sep · proposed
Duration8 teaching weeks; RP301 capstone separately
Credits1.5 credits
PrerequisitesNone. In the certificate pathway, CS201 runs alongside and covers related reinforcement-learning material. Undergraduate-level mathematics is taught as needed.
Who it is forMSc Psychology and similar students, psychiatry residents and clinical psychologists. Labs use Python; CS101 is recommended.
FacultyNeurovidya faculty
AssessmentLab reports, case studies and a graded final assessment.

About the course

Starting from animal behaviourism and ending at deep reinforcement learning, the course treats conditions such as depression, anxiety and addiction as changes in the parameters of a learning agent. A separate RP301 capstone develops an applied research question in computational psychiatry.

Learning outcomes

  • Explain reward, value and prediction error and their links to dopamine
  • Fit learning models to choice data and read the parameters as a profile
  • Describe how deep networks and deep RL model brains
  • Model depression, anxiety, compulsivity and psychosis as altered parameters
  • Discuss how parameter profiles could guide individual care and education

Course modules

8 modules

  • Three traditions and hidden variables

Labs and projects: Python labs in every module, including bandit tasks and agent simulations.

Books and readings: Sutton & Barto, Reinforcement Learning; Redish & Gordon (eds.), Computational Psychiatry