Rule and Instance Based Strategies in Expert Judgment

Abstract

Expert judgment has been conceived in contrasting ways. The naturalistic decision making (NDM) paradigm has put forth a largely instance-based account, viewing experts as relying on a storehouse of cases, such as in the recognition-primed decision (RPD) model. Cognitive psychology has instead advanced largely heuristic- or rule-based accounts, such as in the lens model and fast-and-frugal heuristics. To clarify the relationship between these accounts, we performed two experiments in which novices and experts performed a task explicitly designed to reveal signatures in the data of the use of both rule- and instance-based strategies. Modeling revealed that expert judgment benefited from both the use of linear cue-weighting rules and instance memory. Instance memory use was reflected in experts' (but not novices') ability to handle task nonlinearity, and the finding that expert accuracy across instances was positively correlated with the number of times each instance was historically seen in past experiences.