An exploration of central topics in the foundations of cognitive science through the lens of reasoning and decision-making with limited cognitive resources.
The approach to rational decision-making known as resource rationality has recently seen rapid development in the cognitive sciences. In Resource Rationality, Thomas Icard explores central topics in the foundations of cognitive science through the lens of reasoning and decision-making with cognitive resource constraints. He utilizes resource rationality both as a framework for normative assessment and as a source of scientific hypotheses about how mental processes work. He also discusses resource rationality in the context of large language models and contemporary topics in deep learning and neural network interpretability (a method for explaining the behavior of AI neural networks).
Icard begins with background on traditional decision theory, basic learning theory, information theory, and elementary theory of computation, then presents empirical studies of human perception, reasoning, and decision-making. Finally, he turns to core theoretical questions in cognitive science on the nature of cognitive costs, human "rational faculties," and algorithms and algorithmic implementation. He offers a novel account of cognitive costs: what they are and how they can be measured. Throughout, he grounds his discussion in the productive tension between the normative and the descriptive in the computational, philosophical, psychological, economic, and mathematical foundations of resource rationality.
