What are the origins and key ingredients of the cultural abilities that distinguish humans from other animals? Are some or all present in our closest living primate relatives? Can this knowledge help us design systems capable of interacting with humans cooperatively?

Humans are not alone in displaying cooperative behaviors. Ant colonies build complex nests, groups of songbirds mob predators, and wolf packs capture large prey. Yet even children cooperate with a level of sophistication unmatched by most other animals. They help others complete tasks, divide labor, and communicate plans to solve social dilemmas. Such behaviors are the seeds of adults’ unprecedented levels of cooperation.

The difference between human and animal cooperation may lie in the psychological mechanisms underpinning joint action. Although many species engage in joint action, evidence suggests only humans have ​shared intentions that ​flexibly support it. Do shared “we-intentions” drive human sociality and culture? Are these abilities present but undiscovered in other animals?

Traditional experimental approaches cannot reproduce real-world ecologies that would reveal such abilities. Using virtual reality (VR), this project addresses these important challenges. It combines rich virtual mini-worlds with state-of-the-art behavioral analysis based on novel computational models of learning and cooperation.

Harnessing the generative nature of VR, a team led by Francine Dolins will study social problem solving in apes, humans (including children), and artificial agents across episodes and environments varying in both visual and ecological realism. The project has three goals:

1. to examine we-intentions and their role in the distinctiveness of human cooperation;
2. to develop artificial agents with diverse learning and planning capacities, enabling a model-based fine-grained analysis of ape and human social behavior; and
3. to test hypotheses derived from primate fieldwork.

The team’s unique interdisciplinary makeup enables an ecologically-grounded investigation. The tools they create will gather rich empirical data, generating important theoretical advances in social intelligences. This will, in turn, drive new experiments to further refine theoretical models of cooperation.