Milgram Obedience Experiment: An Agent-Based Model

Abstract

An open, interactive agent-based model (ABM) exploring Stanley Milgram’s classic obedience experiments through computational modelling. Built with the Mesa ABM framework and a Streamlit dashboard, it simulates how situational factors (experimenter presence, learner visibility, peer defection, vocal feedback) shift simulated participants’ likelihood of continuing to administer increasing shocks, and compares results against Milgram’s published obedience rates. Intended as a teaching and computational-modelling resource for exploring model assumptions, calibration, and reproducibility - not as a validated psychological model or empirical replication of the original study.