PS202INTERMEDIATEPSYCHOLOGY & PSYCHIATRY

Computational Cognitive Science

How space and language emerge from sensorimotor learning, modelled with neural networks and RL.

Course Information

Course Information
ModeInstructor-led or blended
TermFA2 · Oct–Nov · proposed
Duration8 weeks · about 40 learner hours · 9 modules
Credits1.5 credits
Counts towardStandalone course
PrerequisitesNone. Undergraduate-level mathematics is taught as needed.
Who it is forPsychology, cognitive science, linguistics, education and engineering students.
FacultyNeurovidya faculty
AssessmentLabs and a graded final project.

About the course

A course in cognitive science based on deep networks, recurrent networks and reinforcement learning. It examines the development of spatial and linguistic representations from sensorimotor learning, alongside models of reading, dyslexia and Indian writing systems.

Learning outcomes

  • Build and train networks that model perception, memory and decisions
  • Explain how place and grid-like codes emerge from movement
  • Compare grounded and distributional models of language
  • Model reading and dyslexia across English and akshara scripts
  • Relate model parameters to individual learner profiles

Course modules

9 modules

  • Networks as models of mind

Labs and projects: A training lab for every module in Python.

Books and readings: O’Reilly & Munakata, Computational Cognitive Neuroscience; Sutton & Barto, Reinforcement Learning