A Contrast/Surprise Model for Updating Beliefs

Abstract

Although the updating of beliefs is a central concern in many fields, empirical research has produced complex and conflicting results. We first present a psychological model of updating that assumes equal attention is given to all information. The model is based on an anchoring-and-adjustment process that incorporates a contrast or surprise effect; in particular, the larger the current opinion, the more it is discounted by negative evidence and the less it is increased by positive evidence. The model predicts strong recency effects for conflicting evidence and, no order effects for consistent evidence. These predictions are contrasted with those of alternative models and tested in a series of six experiments involving the evaluation of written scenarios containing varying amounts and types of information. Thereafter, we generalize the model to include the effects of differential attention and show the conditions under which attention decrement can lead to primacy rather than recency. Specifically, under attention decrement, people with strong prior beliefs are more prone to primacy than those with weak priors. We then discuss our theoretical framework and results with respect to procedural variables that can affect judgment, the optimal inattention problem, comparisons with alternative models of updating (e.g., Bayesian models), and limitations and extensions of the present approach. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1985
Accession Number
ADA153937

Entities

People

  • H. J. Einhorn
  • R. M. Hogarth

Organizations

  • University of Chicago

Tags

Communities of Interest

  • Biomedical
  • C4I
  • Human Systems
  • Weapons Technologies

DTIC Thesaurus Topics

  • Air Force
  • Applied Psychology
  • Bayesian Networks
  • Behavioral Sciences
  • Biomedical Research
  • Cognition
  • Combinatorial Analysis
  • Computational Science
  • Human Factors Engineering
  • Information Processing
  • Information Science
  • Military Research
  • Models
  • Navy
  • Operations Research
  • Probability
  • Psychology

Fields of Study

  • Psychology

Readers

  • Artificial Intelligence
  • Computational Modeling and Simulation
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference