Computational Cognitive Science

This course is usually offered in the winter semester. It is open to graduate students and upper-year undergraduates, and involves a project component.

Overview

This course is an introduction to computational theories of human cognition. Drawing on formal models from classic and contemporary artificial intelligence, cognitive science and experimental psychology, we will explore how to understand human cognition through the lens of computation. What are the forms that our knowledge representations take? What are the computational principles by which we form hypotheses and learn? How do people make choices given incomplete information about alternatives? Students will learn how to understand experimental data by applying existing computational models, and building models of their own. There are no formal prerequisites for this course, however the course requires mathematical literacy, and some experience with programming and statistics. Any knowledge of neuroanatomy, perception, philosophy of science, experiment design, machine learning and psychiatry will be an asset.

Class sessions will comprise a mixture of lectures and discussion. Readings will include classic and recent research papers from the cognitive science, AI, and neuroscience literature, as well as textbook chapters and tutorials on technical methods. Assignments will consist of several problem sets, and a final modeling project or paper.

Students should expect to choose a project topic half-way through the term, and will be offered an opportunity to develop their project into a conference paper -- although a paper submission is not a requirement of the course.

We will discuss a range of formal modeling approaches and their applications to understanding core areas of cognition. Cognitive science topics will include:

• Theories of Rational Choice
• Learning, Perception and Attention
• Theories of Cultural Transmission
• The Structure and Formation of Intuitive Theories of Biological and Social Systems
• Mental Representations of Time and Space

Formal modeling topics will include:

• Bayesian inference
• Controlling complexity, Minimum Description Length
• Sampling algorithms in probabilistic models
• Program synthesis and program induction
• Methods of fitting models to data

Prerequisites

There are no formal prerequisites for this course, as the goal is to make this course maximally accessible to students from diverse backgrounds. The course is open to Graduate Students from Computer Science and Psychology and Neuroscience. Students from other departments (e.g. Medicine) can enroll with approval of their supervisory committee. Undergraduates can enroll with the instructor’s permission – you will nee to get in touch with the instructor and ask for an override to join this course. Undergraduates in double major programs such as CS and Neuroscience have previously done very well in this course, and are encouraged.
Necessary backgrounds to take this class are mathematical (basic linear algebra, calculus) and programming literacy). A set of math literacy questions that you should be able to answer to enroll in this course can be found here. Students should ideally be comfortable with a programming language of their choice (e.g. Python, R) and should have taken a class in probability or statistics. A background in AI methods will be useful. An effort will be made to accommodate a variety of interdisciplinary backgrounds by providing supplementary resources, but students who feel uncertain about the basic prerequisites are encouraged to contact the instructor.

Sample topics from previous year

• Cognition as Computation
• Computation as Cognition
• Probability perception. Hypothesis testing
• Mental sampling. Learning with a few examples.
• Principles of visual representation
• Objects and Concepts
• Language and language of thought
• Theory of Mind
• Planning
• Neuroeconomics
• Time and space. Cognitive maps.
• Computational psychiatry and phenotyping