Habitual control of goal selection in humans

Abstract

Human cognition makes widespread use of goal-directed planning. However, exhaustive forward planning for tasks of real-world complexity is prohibitively computationally demanding. Much research aims to find efficient mechanisms for approximate planning. We describe an approach to this problem that exploits the computational efficiency of habit learning to select goal states that are subsequently used in planning. We also provide experimental evidence that humans implement this approach. Our findings illuminate the basis of learning and choice in humans, demonstrate an integration between mechanisms of habitual and planned control, and contribute to the development of computationally tractable planning algorithms.

Document Details

Document Type
Pub Defense Publication
Publication Date
Oct 12, 2015
Source ID
10.1073/pnas.1506367112

Entities

People

  • Adam Morris
  • Fiery Cushman

Organizations

  • Harvard University
  • Office of Naval Research

Tags

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Systems Analysis and Design
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.