The Similarity Heuristic

Abstract

Decision makers often make snap judgments using fast-and-frugal decision rules called cognitive heuristics. Research into cognitive heuristics has been divided into two camps. One camp has emphasized the limitations and biases produced by the heuristics; another has focused on the accuracy of heuristics and their ecological validity. In this paper we investigate a heuristic proposed by the first camp, using the methods of the second. We investigate a subset of the representativeness heuristic we call the “similarity” heuristic, whereby decision makers who use it judge the likelihood that an instance is a member of one category rather than another by the degree to which it is similar to others in that category. We provide a mathematical model of the heuristic and test it experimentally in a trinomial environment. In this domain, the similarity heuristic turns out to be a reliable and accurate choice rule and both choice and response time data suggest it is also how choices are made. We conclude with a theoretical discussion of how our work fits in the broader ‘fast-and-frugal’ heuristics program, and of the boundary conditions for the similarity heuristic.

Keywords: heuristics and biases, fast-and-frugal heuristics, similarity, representative design, base-rate neglect, Bayesian inference