Computational Neural Models of Risk

Abstract

The central aim of the project was to develop computational models of how individual decision-makers learn in real time to anticipate and take into account the risks and potential consequences of their actions. The main focus was on the medial prefrontal cortex (mPFC), an area of the brain known to signal mistakes as well as the level of difficulty or conflict facing the decision-maker. The research effort involved iteratively developing computational models and testing their predictions with fMRI, leading to further refinements of the model. The original goal of developing a model of risk prediction was achieved. Further effort yielded a more general model of how both good and bad potential consequences are learned and anticipated. The model predictions were validated by numerous behavioral and fMRI studies, and the effort also yielded an exact recursive model of hyperbolic temporal discounting. The results overall provide a new and relatively simple computational model of consequence prediction that accounts for and predicts a wide array of empirical data and is well-grounded in the known neurobiologically.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 24, 2010
Accession Number
ADA515423

Entities

People

  • Joshua W. Brown

Organizations

  • Indiana University

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Brain
  • Cognition
  • Cognitive Neuroscience
  • Cognitive Science
  • Computational Neuroscience
  • Executives
  • Human Behavior
  • Learning
  • Mental Processes
  • Monitoring
  • Neural Networks
  • Neurosciences
  • Psychology
  • Reinforcement Learning
  • Students

Fields of Study

  • Biology

Readers

  • Computational Fluid Dynamics (CFD)
  • Neuroscience
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • AI & ML - Neural Networks