The fact that people frequently cooperate in the single-trial Prisoner's Dilemma (PD) game indicates that their decision making in conflicts is not always guided by game-theoretic analyses of expected outcomes. Whereas most theorists have accounted for cooperation in terms of an ethically rooted concern for matching another's “good faith” cooperation (Hofstadter, 1985), others have argued that cooperation reflects several distinct social norms or heuristics (Elster, 1989). In the latter spirit, the current paper argues that whereas one source of cooperation is an effort to match cooperation of others, another source of cooperation is an illusory effort to exert control over others. To distinguish these social decision heuristics and investigate their boundary conditions, we test hypotheses about how their occurrence depends on the timing of decisions in a PD game in which the two players move on different days. Two experiments consistently supported the hypothesis that the control heuristic is more frequent when participants believe that their counterpart's move lies in the future rather than in the past. Support was also obtained for a second hypothesis that the matching heuristic, by contrast, is less frequent when participants believe that the counterpart's move lies in the future rather than in the past.