CMPUT 654 (Winter 2019)

Modelling Human Strategic Behaviour

Syllabus

Class Times:
Tuesdays and Thursdays 2:00–3:20pm
First class:
January 8, 2019
Location:
C-E4-36
Instructor:
James Wright (james.wright@ualberta.ca)
Office:
ATH 3-57
Office hours:
Available after each class for at least half an hour, and by appointment.
Class forum:
https://piazza.com/ualberta.ca/winter2019/cmput654
Description

This course examines the mathematical and empirical foundations of modelling behaviour by human (and other boundedly rational) agents, in scenarios where the agents have independent goals and priorities. The course will cover the classical game theoretic results for fully rational agents, before surveying the results of modern empirical behavioural research. The course will be capped off by a small research project, in which students will survey an area of existing literature and propose directions for further study.

Topics
  1. Game theory
  2. Bounded rationality and behavioural game theory
  3. Applications of behavioural modelling
Objectives

After taking this course, you should be able to formally represent an arbitrary strategic scenario in the game theoretic paradigm. You should understand the main implications of the full rationality assumption in game theory, and the most common ways in which humans fail to satisfy it.

Research survey

The final third of the class will be driven by a small research project, which will be presented to the class. The project can be a survey of the literature of a specific sub-area that we did not cover in class, ideally with a proposed direction for new research. Novel research results are NOT REQUIRED for full marks on the survey, but may be awarded bonus marks.

Each survey will be presented to the class in a workshop-style talk. The project evaluation will be based partly on the project itself, partly on the presentation of the survey, and partly on the quality of the peer review of other students’ presentations.

Grading

Grade breakdown
Late assignments

Assignments are to be handed in electronically via GradeScope by the start of lecture on the due date. Late assignments will have 20% deducted for each day that the assignment is late, up to a maximum of three days late.

Academic conduct

Submitting the work of another person as your own constitutes plagiarism. The department is very strict about plagiarism and other academic misconduct: ALL forms of cheating are referred to the Dean’s office.

The rules for this course allow consultation collaboration. The specific rules are:

Texts