A Conceptual Analysis of Cooperation in Social Dilemmas: Triangulating Text Mining, Meta-Analyses, and Narrative Reviews

For nearly 75 years, scholars across disciplines have studied cooperation in social dilemmas, but the burgeoning literature makes it difficult to see the complete picture and identify key phenomena associated with cooperation. To address this gap, we first consolidated existing theories, narrative reviews, and meta-analyses on social dilemmas into a conceptually-driven meta-review. We subsequently applied the novel Text Mining Systematic Review (TMSR) method to qualitatively summarize the comprehensive empirical literature, comprised of 1,880 papers represented in the Cooperation Databank (CoDa). We implemented methodological innovations that further automate the TMSR process and used context-aware word embeddings to better capture meaning. Drawing on TMSR, we mapped the frequency and co-occurrence of phenomena within papers, confirming well-established correlates of cooperation and revealing understudied yet promising areas. The meta-review contextualizes our interpretation of these data-driven insights, and CoDa's comprehensive effect size coding allows us to validate our qualitative findings with meta-analyses. From the meta-review and TMSR, we derive 14 essential constructs contributing to cooperation in social dilemmas including the incentive structure, framing, prosocial goals/efficiency, equality/fairness, expectations of cooperation, trust, group identification, punishments, rewards, communication, leadership, direct and indirect reciprocity, and intergroup competition – constructs spanning aspects of the game structure, player characteristics, group processes, and evolutionary mechanisms. The present findings advance a unified understanding of what drives cooperation in social dilemmas and provide a framework for future research and theory development.