We report on an experiment designed to explore whether the effects of expressing one’s emotions spill over into future interactions, thereby curtailing subsequent selfish decisions. In between two identical public goods games, participants play a binary-choice dictator game which, depending on the treatment, either gives or does not give the recipient the opportunity to text the dictator. The recipients of an unfair offer—in contrast to the recipients of a fair offer—contribute significantly less in the second public goods game. Yet, their contribution reductions are significantly smaller in the treatment allowing for recipient communication. To control for a belief-based explanation of these findings, we run treatments where we elicit beliefs about the other’s contribution. We find that belief elicitation affects the efficacy of communication.