Rationality
In popular culture, we often think of rationality as in tension with intuition. We see this in caricatures of hyper-rational agents like Spock in Star Trek. The probabilistic revolution in cognitive science has changed this notion. It is not rational to treat every task as logical deduction when the world itself is underdetermined and radically uncertain. This session will explore the notion of being rational in an uncertain world.
Primary Readings
Everyone should read these and be prepared to discuss:
| Oaksford, M., & Chater, N. (2001). | The probabilistic approach to human reasoning. Trends in Cognitive Sciences, 5(8), 349-357. |
| Marr, David (1982) | Chapter 1 of Vision, MIT press (on Learn)This book is famous for its presentation of “Marr’s levels”: The three interrelated levels levels of analysis at which one can explain the behavoiur of any computational system (in Figure 1-4). This is an incredibly central idea in cognitive science that has had a major impact on comptuational modelling work ever since and well worth familiarizing yourself with if it isn’t already familiar to you. |
Secondary Readings
The presenter should read and incorporate these:
| Simon, H. A. (1955). | A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99-118.In this research paper, Herbert A. Simon discusses the need to revise the traditional economic model of “economic man” as a rational decision-maker. He proposes a behavioral model of rational choice that takes into account the limited information and computational capacities of individuals. Simon emphasizes the importance of considering the decision-maker’s characteristics and the specific situation they are in when defining rational behavior. He explores variables that can be controlled and optimized for rational adaptation, as well as fixed variables that must be taken into account. The paper also discusses the application of this model to game theory, decision-making, and chess. Simon suggests simplifications and approximations that align with observed human behavior and addresses objections to simple pay-off functions in certain choice situations. He also highlights the importance of information gathering and refining the mapping of possible outcomes in decision-making. The paper concludes by discussing the implications of this behavioral model for understanding human decision-making in various contexts and the potential normative and descriptive value of this model. |
| Lieder, F., & Griffiths, T. L. (2020). | Resource-rational analysis: Understanding human cognition as the optimal use of limited computational resources. Behavioral and brain sciences, 43, e1.This article explores the concept of resource-rational analysis, a cognitive modeling paradigm that integrates rational principles with cognitive constraints to better understand and improve human cognition. The article also discusses the challenges faced by research on human cognition and introduces the concept of bounded optimality as a theory for designing optimal programs for agents with limited resources. It proposes that resource-rational analysis provides a method for understanding how the human mind uses its limited cognitive resources. It suggests that cognitive biases may be a result of limited cognitive resources rather than irrationality, and understanding these limitations can lead to improved AI and a redefinition of cognitive biases. |
Questions under discussion
- What does it mean to be rational?
- What does probability theory have to do with it?
- What does it mean to be boundedly rational or resource rational?
- Broadly speaking, are people any of the above?
