The spreading activation theory of semantic memory of Collins and Loftus was adapted for the domain of knowledge called "social norms," and the resulting cognitive model of social norms was examined as a means of understanding the transgression-helping phenomena. Briefly, it was proposed that social norms are represented within a knowledge structure in memory and that these representations are associated with each other in varying degrees. With respect to the transgression-helping effect, it was proposed that observation of a transgression activates the representation of the relevant social norm, which in turn facilitates access to the representation of the helping norm to the extent that the two are closely organized within the cognitive structure. Pilot subjects made paired-comparison similarity judgments of a set of social norm statements, and three of these statements were subsequently chosen for the experiment. In the experiment, one of these social norms was made salient within a transgression or no-transgression context, and subjects then had the opportunity to help. The results showed general support for the cognitive model. More helping occurred when a norm closely related to the helping norm was made salient than when a remotely related norm was made salient, compared with no-norm-salient controls. Transgression versus no-transgression context had no effect on helping. The importance of the research and the cognitive model for social norm explanations of helping behavior and of social-norm-based behaviors, in general, was discussed.Research has shown that in many circumstances, persons help more and/or more often when a transgression of some kind occurs prior to the opportunity to help than when no such prior event occurs. Often persons who are in some way responsible for the transgression help both the victim (Carlsmith & Gross, 1969) and a third party (e.g., Katzev, Edelsack, Steinmetz, Walker, & Wright, 1978). In addition, witnessing someone else's transgression has been found to increase observers' helping (e.g., Cialdini, Darby, & Vincent, 1973;Rawlings, 1970).Explanations for this phenomenon have emphasized drive-reduction processes. Hypothetically, a transgression produces in the perpetrator and the witness some type of This research was the first author's PhD dissertation. We would like to thank Allen Dobbs, Brendan Rule, Gary Wells, and Daniel Wegner for their comments on previous drafts of the paper.