Title:
Social Norms as Social Habits
Abstract:
Standard philosophical accounts (Bicchieri 2006, 2016; Brennan et al. 2013) treat social norms as rules sustained by conditional preferences to conform given (i) empirical expectations and (ii) normative expectations, backed by sanctions. While explanatorily powerful, these views impose heavy cognitive demands and strong rationality requirements. I develop a habit-based account: norms as habitual patterns of response formed through reinforcement learning (RL)—that is, trial-and-error adjustment to social feedback. On this view, compliance, internalization, and sanctioning arise as agents learn from approval, disapproval, and payoffs. The normative force of habits appears in their stability under repetition and their sensitivity to sanctions and nudging interventions. The framework also supports norm-sensitive interventions (e.g., nudging) with guardrails against backfire. Although general, the account has digital social norms in mind—the unwritten rules that structure online behavior. I will present multi-agent RL simulations and outline how digital reward proxies—likes/upvotes, reports/blocks, and moderation outcomes—let us calibrate and test norm dynamics with platform data.