We examine whether spillovers of pro-social behavior depend on how behavioral changes are induced. We conduct a large experiment using economic games, with a Dictator Game (DG) followed by either an identical game or a Prisoner’s Dilemma (PD). We influence initial behavior through widely used policy instruments, either behaviorally informed (default, social norms) or with an economic/regulatory rationale (incentives, regulation). Our results provide evidence of positive spillovers to subsequent economic games (which are not treated), but only for the traditional economic/regulatory interventions and within the same game type. Specifically, inducing higher giving in the first stage leads to subsequent higher altruism in the DG, but not to more cooperation in the PD. The carry over of pro-social behavior appears to be driven by an anchoring on the initial donation. We also measure observers’ beliefs and we find that these results are not correctly anticipated by third parties, who systematically overestimate both the direct effect of behaviorally informed interventions on initial donations and their spillover to subsequent donations.