PSYCHOLOGICAL SCIENCE VOL. 9, NO. 1, JANUARY 1998 Copyright © 1998 American Psychological Society 45 Abstract—According to a two-step account of the mere-exposure effect, repeated exposure leads to the subjective feeling of perceptual fluency, which in turn influences liking. If so, perceptual fluency manipulated by means other than repetition should influence liking. In three experiments, effects of perceptual fluency on affective judgments were examined. In Experiment 1, higher perceptual fluency was achieved by presenting a matching rather than nonmatching prime before showing a target picture. Participants judged targets as prettier if preceded by a matching rather than nonmatching prime. In Experi- ment 2, perceptual fluency was manipulated by figure-ground contrast. Stimuli were judged as more pretty, and less ugly, the higher the con- trast. In Experiment 3, perceptual fluency was manipulated by presen- tation duration. Stimuli shown for a longer duration were liked more, and disliked less. We conclude (a) that perceptual fluency increases liking and (b) that the experience of fluency is affectively positive, and hence attributed to positive but not to negative features, as reflected in a differential impact on positive and negative judgments. 0 Does processing fluency enhance preference for neutral stimuli? This possibility has been proposed as an explanation of the mere- exposure effect (Bornstein, 1989; Kunst-Wilson & Zajonc, 1980; Sea- mon, Marsh, & Brody, 1984; Zajonc, 1968). According to the two-step attributional theory, the repeated presentation of a stimulus enhances the subjective feeling of processing fluency when the stimulus is encountered again. This enhanced perceptual fluency is then misattrib- uted to liking, resulting in a preference for old over new stimuli (Bornstein & D’Agostino, 1994; Jacoby, Kelley, & Dywan, 1989). However, repetition may affect preference by mechanisms other than perceptual fluency. For example, Zajonc (1971, 1997) proposed that repeated exposure may enhance preference via habituation of the ori- enting response. Thus, to test if processing fluency enhances liking, it is useful to turn to manipulations other than mere exposure. Whittlesea (1993, Experi- ment 5) manipulated conceptual fluency by presenting words in a neu- tral or predictive semantic context (“The evening gown was missing a . . . bead” vs. “The bored student opened her mouth to . . . yawn”). The words that appeared in the predictive context were pronounced faster, indicating enhanced processing fluency, and were judged as more pleas- ant. Although suggestive, Whittlesea’s findings leave unclear whether the predicted words were rated as more pleasant because of high pro- cessing fluency or because they were consistent with the semantic con- text whereas the unpredicted words were somewhat incongruous with the semantic context. Moreover, Whittlesea’s findings leave open if per- ceptual, as opposed to conceptual, manipulations of fluency can enhance preference. The current experiments tested the relation between perceptual fluency and preference with three different manipulations. In Experiment 1, the 0. dsadsadsa processing of a target picture was either facilitated or disrupted by a briefly presented visual prime. We predicted that target pictures preceded by a facilitating prime would be judged more positively. The logic of this experiment is similar to that of Whittlesea’s (1993) in that fluency was manipulated by the preceding context. In Experiment 2, perceptual flu- ency was manipulated by changes in the figure-ground contrast. Pictures with higher figure-ground contrast are clearer, and we hypothesized that they would be judged more positively. The logic of this manipulation was based on Checkosky and Whitlock’s (1973) finding that low clarity of pat- terns increases reaction times in a recognition task, indicating slower encoding for low-clarity stimuli. In Experiment 3, fluency was manipu- lated by presentation time of the target stimulus. Our hypothesis was that people would like the pictures more the longer they were presented. The logic of this manipulation was based on Mackworth’s (1963) finding that enhancing the presentation time of stimuli enhances the accuracy of per- ceptual identification. Note that visual clarity and presentation duration were used in pre- vious research as dependent variables. Specifically, repeated exposure to a stimulus, presumably resulting in higher perceptual fluency, was found to elicit judgments of higher visual clarity (Whittlesea, Jacoby, & Girard, 1990, Experiment 4) and of longer duration of stimulus pre- sentations (Witherspoon & Allan, 1985). Our experiments tested a reversal of this influence: If repeated exposure increases experienced fluency, which in turn results in judgments of higher clarity and longer duration, direct manipulations of the latter variables may also result in experiences of greater perceptual fluency. Increased fluency, in turn, may influence affective judgments, thus paralleling the well-known effects of mere exposure, without changes in exposure frequency. The current experiments addressed two additional issues. In addi- tion to assessing preferences, Experiment 1 measured perceptual flu- ency directly by collecting recognition latencies. This allowed us to examine if gains in liking parallel gains in recognition speed. Second, Experiments 2 and 3 examined if perceptual fluency itself is affec- tively neutral or affectively positive. If perceptual fluency is neutral, it may be misattributed to positive features if participants are asked a question framed in a positive way (e.g., “prettiness,” “liking”), and to negative features if the question is framed in a negative way (e.g., “ugliness,” “disliking”; see Bornstein & D’Agostino, 1994, p. 125). If perceptual fluency is affectively positive, however, it should always lead to more positive judgments, independent of the question’s focus. Some findings indicate that perceptual fluency may be attributed to whatever feature participants focus on, resulting in evaluations of the stimulus as, for example, brighter, darker, and clearer (e.g., Mandler, Nakamura, & Van Zandt, 1987; Whittlesea, 1993). So far, however, effects of perceptual fluency on affective judgments have been obtained only for positively valenced judgments. In the study by Man- dler et al. (1987), participants liked repeatedly shown polygons more than a new polygon, but when participants were asked which polygon they disliked more, their judgments were not significantly different from chance level. To address these issues, Experiments 2 and 3 explored the impact of contrast and presentation time, respectively, on judgments of prettiness (or liking) and ugliness (or disliking) of visual EFFECTS OF PERCEPTUAL FLUENCY ON AFFECTIVE JUDGMENTS Rolf Reber, 1 Piotr Winkielman, 2 and Norbert Schwarz 2 1 Université de Bourgogne, Dijon, France, and 2 University of Michigan Address correspondence to Rolf Reber, University of Berne, Depart- ment of Psychology, Muesmattstrasse 45, CH-3000 Bern 9, Switzerland; e-mail: rreber@psy.unibe.ch. Research Report