Evaluating case-based decision theory: Predicting empirical patterns of human classification learning

• Published in: Games and economic behavior 2013-11, Vol.82, p.52-65
• Main Authors: Pape, Andreas Duus, Kurtz, Kenneth J.
• Format: Article
• Language: English
• Subjects: Agent-based computational economics; Behavior; Case-based decision theory; Cognitive science; Cognitive style; Decision making; Decision theory; Economic behaviour; Empirical research; Games; Human cognition; Learning; Management science; Psychology; Simulation; Studies
• ISSN: 0899-8256; 1090-2473
• DOI: 10.1016/j.geb.2013.06.010

We introduce a computer program which calculates an agentʼs optimal behavior according to case-based decision theory (Gilboa and Schmeidler, 1995) and use it to test CBDT against a benchmark set of problems from the psychological literature on human classification learning (Shepard et al., 1961). This allows us to evaluate the efficacy of CBDT as an account of human decision-making on this set of problems. We find: (1) The choice behavior of this program (and therefore case-based decision theory) correctly predicts the empirically observed relative difficulty of problems and speed of learning in human data. (2) ‘Similarity’ (how CBDT decision makers extrapolate from memory) is decreasing in vector distance, consistent with evidence in psychology (Shepard, 1987). (3) The best-fitting parameters suggest humans aspire to an 80–85% success rate, and humans may increase their aspiration level during the experiment. (4) Average similarity is rejected in favor of additive similarity. •We introduce a case-based decision theory software agent.•We test CBDT against psychological data on human classification learning.•We find CBDT correctly predicts the relative difficulty of problems for humans.•We find similarity functions which match humans are decreasing in vector distance.•We find CBDT with appropriately imperfect memory solves these problems as fast as human problem solvers.