The four nested treatments consist of a replication of the baseline traffic light task without peers (BL) where only intrinsic respect for rules and social expectations can explain rule-following, and three treatments that cumulatively add motives to BL to follow the rule. Treatment EX adds an externality (a foregone donation of US $1 to a charity) triggered by a rule violation. Treatment WP adds a sanction of rule violations to treatment EX with a detection probability of 10% and treatment SP with a detection probability of 90%. Punishment implied the loss of all earnings (except the flat participation fee). Shown are mean rule-following rates in the behavioural experiment ( n BL =501, n EX = 507, n WP = 515, n SP = 484). Dots and whiskers show mean descriptive beliefs (+-1 s.e.m.; n BL = 102, n EX = 98, n WP = 112, n SP = 96). The dark green shaded parts of the bars show the average proportions of participants who conform unconditionally regardless of their normative and descriptive beliefs (averages calculated from the strategy method data eliciting n (.) and d (.), those who always conformed; n BL = 152, n EX = 156, n WP = 154, n SP = 157). The full set of results are shown in Extended Data Figs. 4 and 5 and Supplementary Tables 9 and 10.
