Talk abstract: In the study of animal cognition and comparative psychology, there is an implicit understanding that experimental approaches alone can tell us how animal minds operate. This has led to a small number of paradigms for a small number of cognitive skills having an outsized impact on our understanding of animal cognition. Vast observational datasets exist, and this talk will provide an example of how we can use them to study cognition in non-human primates using non-experimental approaches. Importantly, decision-making in animals is sequential: every action depends on what happened before, and determines what can happen after, and complexity arises where these cascades are not fixed, but allow for flexible decisions by individuals. Never is this clearer than when animals lose themselves in play, combining hundreds of actions freely and in coordination with their playmates: play is only fun when it is unpredictable, but only as long as some rules are followed. Using the mix of chaos and rules that underlies play, this talk will discuss how detailed observational studies of action sequences can broaden our understanding of other primates’ abilities to cooperate, communicate, and learn from each other.