Self-prioritization depends on assumed task-relevance of self-association.

We conducted three experiments to test the effect of assumed task-relevance of self-association on the self-prioritization effect (SPE). Participants were first performing the standard matching task, and then a pseudo-word matching task, in which familiar labels from the standard task were replaced with pseudo-words. In the pseudo-words task, the association between stimuli and the self was, in fact, task-irrelevant. Learning instructions were varied to make participants believe that the self-association was task-relevant (Experiments 1 and 3) or task-irrelevant (Experiment 2). In Experiment 1, we told participants that geometrical shapes represent specific identities (self, friend, none) and pseudo-words' meanings reflect each of these identities to make participants believe that semantic associations of pseudo-words were task-relevant. In this experiment, a SPE was present in the pseudo-words task. Experiment 2 was identical to Experiment 1, but participants were told to form association pairs between pseudo-words and shapes, and not pseudo-words and identities. Thus, because the matching task presented only shapes and pseudo-words, participants were led to correctly believe that associations were task-irrelevant. Under these conditions, we did not observe a SPE in the pseudo-words task. Finally, in Experiment 3, participants were told that pseudo-words are "paired with" identities, making their relations with identities weaker than in Experiment 1, but still task-relevant. The SPE in the pseudo-word task reappeared. Together, the results suggest that a SPE can be observed in the absence of any stimuli with established self-associations, but only if self-associations are represented as task-relevant.