Projects

Collaboration—co-laboring to achieve a common goal and then sharing the resulting benefits—is ubiquitous in humans but rare in other primates. Collaboration also elicits intuitions that resources should be shared equally. Whereas chimpanzees rarely share, human children from a young age show a strong motivation to divide resources equally when acquired collaboratively. However, the evolutionary and cognitive basis of collaboration remains poorly understood. This research addresses this gap by treating collaboration as a distinct cooperative domain supported by specialized cognitive and motivational systems.

Identifying collaboration

Core cues of linked success and coordinated joint action trigger collaboration-specific responses; regulating equal sharing decisions, social emotions (e.g., gratitude and anger), and partner choice motivations.

Benkley, D., Tooby, J., & Cosmides, L. Cognitive adaptations for collaboration. Manuscript in prep.

Evaluating collaborations

I also study how people evaluate potential collaborations. One line of inquiry explores how specialization of labor stabilizes collaboration. Game theoretic analysis shows that role specialization can incentivize contribution by making roles essential, and deter free-riders by making them easier to detect. Additionally, specialization can boost efficiency by building task-specific skills. My work shows that people intuitively understand these advantages and prefer specialized collaborations.

Benkley, D. & Cosmides, L. Specialization of labor: A game changer for collaboration. Manuscript in prep.

Complementing my empirical work, I also use computational models to explore the learning problems underlying cooperative partner choice.

Reinforcement learning choice models

Partner choice is central to the evolution of cooperation but poses substantial computational challenges, including an exploration-exploitation tradeoff: When should an individual stay with an existing partner versus search for a better one? Using evolutionary agent-based modeling I show that reinforcement learning (RL) provides an efficient solution that outperforms alternative strategies and promotes the evolution of cooperation 

Benkley, D & Conroy-Beam, D. Reinforcement learning models of cooperative partner choice. Manuscript in prep.