Non-probabilistic decision making with memory constraints

A. Vostroknutov*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

The single decision maker chooses one of the actions repeatedly. She chooses the action with the highest weighted average of the past payoffs. In the long run either the action with highest expected payoff or the action with highest minimal payoff is chosen depending on how weights evolve. 

Original languageEnglish
Pages (from-to)303-305
Number of pages3
JournalEconomics Letters
Volume117
Issue number1
DOIs
Publication statusPublished - Oct 2012

Keywords

  • Adaptive learning
  • Constrained memory
  • Bandit problems
  • SIMPLE DYNAMIC-MODEL
  • CHOICE

Cite this