Course content

The course focuses on processes of statistical inference that can support deciding what happens in the world, motivation in the form of reward and effort sensitivity, to decide whether to act, and reinforcement learning, to decide what action to take.

The processes of statistical inference to be examined are Bayes’ theorem, signal detection theory, and sequential sampling models. Motivation to perform discrete actions will be examined in the form of sensitivity to effort and reward. The marginal value theorem is more suitable when actions and outcomes are continuously variable.

Up to this point, the only decision is whether to act or not, while the action is given. Reinforcement learning deals with which of multiple actions to choose. The course will examine both abstract descriptions of reinforcement learning and concrete examples from the human brain. When reinforcement learning is studied in organisms, it is often assumed that evolution has weeded out maladaptive motivations. When an artificial system is being designed, it is essential to design incentives so that they motivate desirable behaviour. The course will examine incentive design, and examples of learned behaviour that either the individuals who learned it or society may consider undesirable

Learning outcome

Knowledge:

-Students will become familiar with processes of statistical inference that can support decision-making.

-Students can build quantitative models of behaviour.

-Students know how to apply quantitative models to both organisms and artificial systems.

Skills:

Students will be able to assess the strengths and limitations of decision-making algorithms

-Students are able to use decision-making algorithms to explain the behaviour of organisms.

-Students are able to translate equations into graphs and numerical models of behavior.

General competence:

Students will understand the role of quantitative models in the study of behavior.

-Students will be able to evaluate each other’s work and give constructive feedback.

-Students will be able to predict the behavior of simple artificial systems based on learning rules.

Learning methods and activities

Lectures, group work

Compulsory assignments

• Submission of two essays
• Provide feedback on another student's essay.

Specific conditions

Recommended previous knowledge

PSY2102

Required previous knowledge

Spesielle vilkår:

Krever opptak til studieprogram: Psykologi (MPSY) - studieretning i Psykologisk vitenskap og Teknologi