In this paper, we test the type indeterminacy hypothesis by analyzing an experiment that examines the stability of preferences in a Prisoner Dilemma with respect to decisions made in a context that is both payoff and informationally unrelated to that Prisoner Dilemma. More precisely we carried out an experiment in which participants were permitted to make promises to cooperate to agents they saw, followed by playing a Prisoner's Dilemma game with another, independent agent. It was found that, after making a promise to the first agent, participants exhibited higher rates of cooperation with other agents. We show that a classical model does not account for this effect, while a type indeterminacy model which uses elements of the formalism of quantum mechanics is able to capture the observed effects reasonably well.