Stochastic dynamic models of response time and accuracy: a foundational primer
Journal of Mathematical Psychology
Reaching pareto-optimality in prisoner's dilemma using conditional joint action learning
Autonomous Agents and Multi-Agent Systems
Simple games as dynamic, coupled systems: randomness and other emergent properties
Cognitive Systems Research
Hi-index | 0.00 |
This paper examines strategic adaptation in participants' behavior conditional on the type of their opponent. Participants played a constant-sum game for 100 rounds against each of three pattern-detecting computer algorithms designed to exploit regularities in human behavior such as imperfections in randomizing and the use of simple heuristics. Significant evidence is presented that human participants not only change their marginal probabilities of choosing actions, but also their conditional probabilities dependent on the recent history of play. A cognitive model incorporating pattern recognition is proposed that capture the shifts in strategic behavior of the participants better than the standard non-pattern detecting model employed in the literature, the Experience Weighted Attraction model (and by extension its nested models, reinforcement learning and fictitious play belief learning).