The ABC of Heuristics: Approximate Bayesian Computation as a Framework for Heuristic Inference

Heuristics are often characterized as simple decision rules that reduce cognitive effort in inference and decision making, but this view underplays their flexibility and adaptiveness in complex, real-world environments. This paper proposes a reinterpretation of heuristic inference as realizations and products of Approximate Bayesian Computation (ABC): a likelihood-free method that approximates Bayesian inference by simulating from prior beliefs and comparing observed data using low-dimensional summary statistics, or cues. On this view, many heuristic patterns in inference and choice can emerge from an adaptive inference process that avoids explicit likelihood calculation while remaining sensitive to how informative and reliable available cues are in a given setting. This framework situates heuristics within a probabilistic account of decision making and offers several theoretical advantages: it relaxes strong assumptions about cue-target relationships, accounts for stochasticity in heuristic choices, and invites a more appropriate benchmark for evaluating heuristic performance. We illustrate the practical applicability of the framework with two examples: a classic cue-based decision task and an ecological causal learning task. By unifying heuristics and Bayesian inference through the lens of likelihood-free methods, this work provides a more nuanced understanding of human inference and decision making within an interpretable framework.