I use methods from experimental economics to explore questions of interest in social psychology, organizational behavior, and complex systems. I collect groups of subjects to play simple games for money. Between experiment and individual-level cognitive modeling, I try to understand collective outcomes in terms of individual behavior, with a special focus on the emergence of group structure (how groups adapt to the problem they are solving) and dynamics (describing the time course of individual and group behavior, and its evolution). Previous work in multi-player economic games has shown that large groups (established exogenously by the experimenter) can work together better if they grew incrementally from small groups. My goal was to examine this finding in endogenous groups---subgroups that emerge as a result of individual interactions in the task. The results suggest that subjects transferred their experience in a very different game to their new environment, and groups with this theoretically irrelevant experience actually performed better than groups who had only experienced the test environment. Learning transfer in groups is a new field with implications for interaction design. This work can inform organizational design. When should there be a boss? When should a problem be split between groups?