Idiosyncratic Random Behavior in a Massive Online Sample

What factors govern and predict an individual’s random behavior? Under many views, human random behavior arises from a grab bag of group-level heuristics (e.g., the representativeness heuristic, the law of small numbers, etc.). Contrary to this view, we present evidence that random behavior arises from highly idiosyncratic processes, borne out even with minimal amounts of data. Over 50,000 people generated random sequences of (a) coin flips, (b) dice rolls, and (c) locations in a 3-by-3 grid. Importantly, participants submitted only 8–12 responses in each task. This very small window minimizes many factors that might create individual differences in random behavior on their own, such as differences in working memory or attention. Yet, individuals demonstrated consistent behavior across these three different tasks, as reflected both by their sequence features and by a simple computational model of random behavior. Thus, these results indicate that, while some general heuristics guide random behavior, there also exist unique, idiosyncratic patterns that warrant attention—raising broader questions about how random behavior is realized in the human mind.