Social preferences depend on emotional states like compassion and anger. Since emotions are fleeting and subject to manipulation, they may generate demand for commitment. We investigate the use of commitment strategies in an online experiment (n = 1,400), where subjects decide to watch or avoid videos before engaging in a charitable giving task. We find that a video with emotional content increases giving, but is also avoided more than non-emotional videos. We estimate a structural model of state-dependent social preferences, and show evidence for sophisticated commitment to selfishness and altruism. We argue that giving can be fruitfully analyzed as a self-control problem.

Previous literature has shown that people are often reluctant to learn whether individually profitable actions have negative consequences for others. In an experimental allocation decision, we vary the ‘inconvenience’ of becoming informed about the payoffs of another player by changing the costs and benefits of choosing the fair outcome. Lowering the cost of fairness turns out to have a multiplier effect, raising both altruistic choices of informed subjects and the fraction of subjects that chooses to become informed. Thus, in situations of uncertainty, subsidizing altruistic choices to decision makers is an effective tool for raising social welfare. By contrast, variations in the size of recipients’ potential benefit have a smaller and insignificant effect on ignorance and fair choices.

We examine how narratives about a past conflict shape present-day political preferences and how these effects interact with family memories. We focus on Spain and the Spanish Civil War and conduct an online survey experiment (n=2,593). Individuals are exposed to either an Extreme Right-Wing, a Balanced or an Unrelated narrative about the war and its aftermath, with or without prior activation of family memories, and then allocate a donation between left- and right-wing organizations. Exposure to the Extreme Right-wing narrative leads to a 12-19% increase in right wing donations (vs Unrelated and Balanced, respectively). This effect mostly driven by those on the right. Memory activation seems to buffer this radicalizing influence of the Extreme Right-Wing Narrative. Analysis of participants’ memories shows that memories are predominantly emotionally negative and left-slanted, helping explain this buffering. Evidence suggests that both narratives and memories operate through ingroup–outgroup emotional mechanisms, heightening affective polarization rather than alternative channels. Hence, the ways in which past conflicts are remembered and narrated can shape present-day political alignment, and they appear to do so through an identity channel.