Martin A Nowak | Searcharxiv

arXiv2026

Simple evolution drives direct reciprocity to maximum payoff in social dilemmas

Direct reciprocity is a mechanism for evolution of cooperation based on repeated interactions between the same individuals. Direct reciprocity can help natural selection to favor cooperators over defectors, but whether or not cooperation prevails depends on the details of the evolutionary dynamics. We describe simple processes of evolution that have the astonishing ability of driving direct reciprocity to maximum payoff in all social dilemmas which we study. The basic process is based on mutation and pairwise comparison. Mutation samples strategies near the boundary of the strategy space. Pairwise comparison includes a parameter for intensity of selection. For large population sizes, intermediate to high mutation rates and intermediate to strong intensities of selection, we find that the process leads to communities of strategies that reach maximum payoff in Prisoner's Dilemma, Snowdrift, Stag Hunt and Harmony games. Maximum payoff in all four games is consistently achieved if players have access to memory-2 strategies. Memory-1 strategies have the capacity to resolve all four social dilemmas, but they are usually defeated by a ``Hold-trap'' in Snowdrift games.

q-bio.PE↗

arXiv2026

Strategies of cooperation and defection in five large language models

Large language models (LLMs) are increasingly deployed to support human decision-making. This use of LLMs has concerning implications, especially when their prescriptions affect the welfare of others. To gauge how LLMs make social decisions, we explore whether five leading models produce sensible strategies in the repeated prisoner's dilemma, which is the main metaphor of reciprocal cooperation. First, we measure the propensity of LLMs to cooperate in a neutral setting, without using language reminiscent of how this game is usually presented. We record to what extent LLMs implement Nash equilibria or other well-known strategy classes. Thereafter, we explore how LLMs adapt their strategies to changes in parameter values. We vary the game's continuation probability, the payoff values, and whether the total number of rounds is commonly known. We also study the effect of different framings. In each case, we test whether the adaptations of the LLMs are in line with basic intuition, theoretical predictions of evolutionary game theory, and experimental evidence from human participants. While all LLMs perform well in many of the tasks, none of them exhibit full consistency over all tasks. We also conduct tournaments between the inferred LLM strategies and study direct interaction between LLMs in games over ten rounds with a known or unknown last round. Our experiments shed light on how current LLMs instantiate reciprocal cooperation.

cs.CY↗