This paper describes interactions between agents who sometimes choose as individuals and sometimes as members of teams. Choosing as a member of a team entails not only being motivated by the team's objective, but also a distinctive pattern of reasoning: an agent who “team reasons” computes, and chooses her component in, aprofile evaluated using the team's objective function. It is not assumed that a given agent team reasons for a particular team; there may be more than one team, and which she reasons for is here treated probabilistically. Ordinary reasoning is a special case in which the team is a singleton. The framework therefore encompasses interactions paradigmatic in the theory of co-operative behaviour in which agents may choose either for the group of for themselves. The hypothesis that people may team reason if and when they are group motivated can, it is shown, explain some puzzling aspects of co-operative behaviour in a new way.