Stimulus recognition may mediate exposure effects.

Moreland and Zajonc presented stimuli with differential numbers of exposures to subjects and obtained measures of affect (e.g., ratings of liking) and ratings of familiarity. Exposure frequency and ratings of familiarity were both significant predictors of affect in a multiple regression equation. Moreland and Zajonc concluded that there are two effects, and thus the exposure effect could not be explained by a stimulus recognition factor alone. However, these results can be explained by the theory that exposure frequency affects a single mediator that is imperfectly correlated with ratings of familiarity and affect. Thus, the null hypothesis that recognition mediates the exposure effect cannot be refuted by the partial correlation and regression analyses of Moreland and Zajonc. Moreland and Zajonc (1977) varied the frequency with which stimuli were presented and obtained several dependent variables including ratings of familiarity and a measure of liking (affect). Since the rating of affect could be predicted from rated familiarity and exposure frequency and since both regression coefficients were significant, it was argued that there are two independent effects. However, the present article shows that the theory that one factor, subjective recognition, mediates the effect of the variable on both dependent variables can predict this outcome.