ON GAME-LEARNING THEORY AND SOME DECISION-MAKING EXPERIMENTS

Abstract

The paper reports on games in which a player learns to improve his strategy during the course of a sequence of plays. The fusion model developed by Bush and Mosteller to explain observed behavior of rats in experimental learning situations was used as the basis for both theoretical and experimental investigation of the efficiency of this type of stochastic process in learning to play games. The experiments reported were with human subjects. Their game- learning performance was compared with that of the 'stat-rat', represented by the fusion model with numerical values of the parameters estimated to fit experimental data for rats. The theoretical models accept basic assumptions of von Neumann-Morgenstern game theory and Bush-Mosteller learning theory. The theoretical and experimental results are directly relevant for any situation in which a sequence of decisions is made

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 17, 1952
Accession Number
AD0604157

Entities

People

  • Merrill M. Flood

Organizations

  • RAND Corporation

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Computational Science
  • Decision Theory
  • Distribution Functions
  • Experimental Data
  • Game Theory
  • Learning
  • Markov Processes
  • Mathematical Models
  • Models
  • Numbers
  • Plastic Explosives
  • Probability
  • Probability Distribution Functions
  • Probability Distributions
  • Sequences
  • Stochastic Processes
  • Symmetric Games

Readers

  • Game Theory.
  • Neural Network Machine Learning.
  • Regression Analysis.