Abstract Self-deception is an important construct in social, personality, and clinical literatures. Although historical and clinical views of self-deception have regarded it as defensive in nature and operation, modern views of this individual difference variable instead highlight its apparent benefits to subjective mental health. The present four studies reinforce the latter view by showing that self-deception predicts positive priming effects, but not negative priming effects, in reaction time tasks sensitive to individual differences in affective priming. In all studies, individuals higher in self-deception displayed stronger positive priming effects, defined in terms of facilitation with two positive stimuli in a consecutive sequence, but self-deception did not predict negative priming effects in the same tasks. Importantly, these effects occurred both in tasks that called for the retrieval of self-knowledge (Study 1) and those that did not (Studies 2–4). This broad pattern supports substantive views of self-deception rather than views narrowly focused on self-presentation processes. Implications for understanding self-deception are discussed. Acknowledgement The authors acknowledge support from NIMH (MH 068241). Notes 1We also performed parallel analyses on the rating means from all studies. Study 1 involved self-judgements of emotion. We therefore thought it likely that self-deception would predict ratings in the task. Indeed, there were main effects of Self-Deception on emotion ratings, both for positive targets, F(1, 18) = 6.16, p<.05, and for negative targets, F(1, 18) = 6.09, p<.05. That is, individuals high in self-deception reported more intense positive emotions (Ms = 3.23 vs. 3.99) and less intense negative emotions (Ms = 2.70 vs. 2.01), relative to individuals low in self-deception. However, Studies 2–4 used affective rating tasks of low self-relevance, and thus we thought it less likely that self-deception would predict ratings in these tasks. Indeed, there were no relations along these lines, ps>.10. Priming effects on ratings, and possible interactions by self-deception, were weak and inconsistent across studies. 2Contact the first author for a list of stimuli. 3Note that these are relatively long judgement times. What is consistent with Studies 1–3 is that a differentiated rating task was used, which would somewhat necessarily slow processing speed (Fazio, 1990). However, this is viewed as beneficial for present purposes as longer processing times have produced stronger and more robust priming effects (e.g., Hines et al., 1996; Joordens & Becker, 1997). Judgement times were slower in Study 4 than in Studies 1–3 and this is attributed to a more differentiated scale (1–8) as well as elimination of the two-response aspect of the procedures used in the earlier studies. The evaluations assessed here clearly involved time and deliberation. However, priming of the responses, we believe, is reliant on the sorts of spreading activation processes shown to be involved in the continuous priming task (de Mornay Davies, 1998; McNamara & Altarriba, 1988; Shelton & Martin, 1992).