Abstract

Punishment is fundamental to the evolution of cooperative norms in teams, organizations, and societies. Based on findings that people are faster when punishing others (relative to when withholding punishment), dual-process theories of punishment assert that humans have an intuitive tendency to punish, which requires effortful deliberation to overcome. Here, we propose an alternative single-process theory that models punishment decisions as a sequential sampling process. We provide supporting evidence for this theory using a public goods game experiment that experimentally manipulates the cost–benefit tradeoff across the game. We show that people are not systematically faster when punishing (versus withholding) across tradeoffs. We also find an inverted-U-shaped relationship between response times and the strength of preferences for punishing, and a negative association between punishment rates and the relative speed of punishment across individuals. Further computational analysis using the drift–diffusion model (DDM) reveals that, on average, people exhibit a pre-disposition to withhold punishment. Our study provides a unified single-process framework for studying the micro-foundations of punishment and integrating process measures to better describe and predict behavior.