The Mere Exposure Effect and Classical Conditioning

This study investigated Zajonc’s hypothesized link between the mere exposure effect and classical conditioning. In the first part of the experiment, participants were presented a photograph of a person. Each photograph was followed by a presentation of an emotionally positive image, a neutral image, a negative image, or a blank screen. Then, participants were asked to rate the faces presented the first part of the experiment, as well as some novel portraits, on a 6-point Likert scale. Faces that had been presented previously differed significantly from faces not previously seen. However, only faces in the 20 Repetitions + Positive Image condition were rated significantly higher than faces in the new condition. These results suggest some support for Zajonc’s hypothesis because only the strongest positive associations led to a mere exposure effect, but this conclusion must be considered carefully given the mixed statistical support. The Mere Exposure Effect and Classical Conditioning 3 Introduction The Mere Exposure Effect In 1968, Robert Zajonc conducted a study which found that repeated exposures to neutral stimuli, such as nonsense words and Chinese characters, led to those stimuli being rated more highly on Likert scales than stimuli that had never been presented. This effect was later named the “mere exposure effect.” Zajonc’s experiments have been repeated with other stimuli, such as photographs of college students and polygons, and the mere exposure effect endures as a fairly replicable psychological phenomenon (Bornstein, 1989). In fact, the mere exposure effect has been witnessed across cultures and even across species (Monahan et al., 2000). One of the first influential explanations for the mere exposure effect was the perceptual fluency/attributional model (Bornstein and D’Agostino, 1994). The theory of perceptual fluency states that after a greater number of presentations, people will process a given stimulus faster. After people are shown a stimulus multiple times, they are usually not aware that their ability to process that stimulus has been enhanced. The ease of processing the stimuli is misinterpreted by the brain as an increase in liking. Thus, the mere exposure effect occurs. Most of the support for the perceptual fluency/attributional model comes from various studies conducted by Bornstein. For example, in a 1992 study by Bornstein and D’Agostino, subjects were presented with photographs of college women at suboptimal (5 millisecond) levels or optimal (500 millisecond) levels. At suboptimal levels (usually at around 4-8 milliseconds), stimuli are presented so rapidly that it would appear as though the subjects had seen nothing at all. After the presentations, the subjects were asked to choose which stimulus they liked more and which stimulus they recognized when presented with a choice of two stimuli. One of the stimuli was present in the previous suboptimal presentations and another was one which had never been presented. At suboptimal levels, subjects could not recognize previously presented stimuli from the new The Mere Exposure Effect and Classical Conditioning 4 stimuli, but liking ratings for previously presented suboptimal stimuli were significantly higher than new stimuli. However, at optimal levels, subjects could readily identify which stimulus they recognized, but the liking ratings between previously presented stimuli and new stimuli did not differ. Bornstein and D’Agostino hypothesized that, at the optimal levels, subjects could readily attribute their perceptual fluency to the greater number of presentations. Thus, the subjects’ minds automatically corrected their liking ratings and the mere exposure effect was not established. However, at the suboptimal levels, subjects could not readily identify the source of their perceptual fluency. Thus, they misattributed the perceptual fluency as greater liking for that stimulus. However, studies by Monahan et al. (2000) questioned the validity of the perceptual fluency/attributional model. In their first study, Monahan et al. put subjects in two groups. In the single exposure condition, participants saw 25 different subliminal stimuli once while subjects in the repeated exposure condition saw 5 different subliminal stimuli five times. When asked about their mood in general, with no reference to the stimuli, people in the repeated exposure condition reported having more positive moods than those in the single exposure condition. Thus, it was hypothesized that the mere repetition of stimuli was enough to create an increase in mood. This “diffuse happiness” would eventually end up being projected on the stimuli, causing the mere exposure effect. In order to test this hypothesis, Monahan et al. ran another experiment with the same singleand repeated-exposure conditions. However, after the exposures, subjects were asked to rate different types of stimuli: stimuli they had previously seen (old), stimuli that resembled the stimuli they had previously seen but were actually new (novel similar), and stimuli that were new and looked very dissimilar from those previously seen (novel different). In both conditions, the old stimuli had the highest liking ratings, followed by the novel similar stimuli, The Mere Exposure Effect and Classical Conditioning 5 and then the novel different stimuli. These results can readily be explained by the perceptual fluency/attributional model. However, the perceptual fluency/attributional model could not account for the fact that all types of stimuli had higher liking ratings in the repeated-exposure conditions compared to the single-exposure conditions. Thus, these results demonstrated that the diffuse happiness created through repeated exposures could be expressed through greater liking for all stimuli, but the effect was amplified for the repeated stimuli or stimuli that resembled the repeated stimuli. Thus, this diffuse happiness could also be a contributor to the mere exposure effect. Classical Conditioning Classical conditioning was discovered through happenstance by the Russian physiologist, Ivan Pavlov (Clark, 2004). In Pavlov’s experiment, dogs learned to associate stimuli with important events. For example, a bell was rung (conditioned stimulus) and the bell ringing was consistently followed by the presentation of food (unconditioned stimulus). Naturally, the presentation of food would cause the dogs to salivate (unconditioned response). With many presentations of the bell ringing followed by the food presentation, the dogs would eventually associate the ringing with food. This would lead the dogs to salivate simply when the bell was rung and the ringing was not immediately followed by the presentation of food (conditioned response). Because ringing a bell would not normally cause a dog to salivate, the emergence of the conditioned response is considered to be an example of classical conditioning. Many experiments involving classical conditioning often involve animal subjects. In fact, some argue that conditioning cannot influence adult humans. Brewer (1974) stated that classical conditioning in adult humans occurs through subjects reacting to the demand characteristics of the experiment, and not through conditioning. Many conditioning experiments have assumed that The Mere Exposure Effect and Classical Conditioning 6 subjects will reflexively react to a conditioned stimulus in a certain way once it has been associated with an unconditioned stimulus enough times. For example, if a subject learns there is a relationship between viewing a stimulus and receiving an electrical shock, the subject will prepare for the shock. However, when the experimenter tells the subject that the stimulus would no longer be followed by shock, the “conditioning” did not persist. However, a 2001 study by Olson and Fazio found that classical conditioning can affect the perception of neutral stimuli under appropriate circumstances. In this experiment, participants saw various pairings of Pokémon cartoon characters and words with either positive, neutral, or negative meanings. These pairings were interspersed with presentations of other stimuli, including other Pokémon, images, and words, which were presented independently or in pairs. When analyzing two target Pokémon (one of which was consistently paired with a positive word and another with a negative word), Olson and Fazio found that the Pokémon paired with a positive word were rated more positively than the Pokémon paired with the negative word. In addition, subjects were asked to rate how confident they felt about seeing certain pairings, including the target Pokémon-word pairs. The confidence results revealed that the subjects were not explicitly aware of pairings presented because participants could not reliably identify if a Pokémon had been paired with a word with a positive valence or a negative valence. Thus, Olson and Fazio’s findings could not be linked to demand characteristics because participants would be unsure whether to respond more positively or negatively if they could not identify the type of word the Pokémon was paired with. The Mere Exposure Effect Meets Classical Conditioning In order to address why repeated exposures would lead to an increase in diffuse happiness, Zajonc (2001) offered another explanation for the mere exposure effect: the mere exposure effect could be a form of classical conditioning. In this theory, the presented image is a The Mere Exposure Effect and Classical Conditioning 7 conditioned stimulus. Because no negative consequences occur after viewing the image, the lack of punishment can be viewed as the unconditioned response that is paired with the stimulus. The absence of negative consequences could be interpreted as a form of negative reinforcement that communicates that the stimulus is safe to view. Thus, after repeated exposures of the stimulus (which reinforce that the stimuli is safe to view), the conditioned response manifests in the form of increased liking for the stimulus. Experimental Design Considerations In most mere exposure experiments, multiple presentations of stimuli are used to obtain the mere exposure effect. Oftentimes, these experiments will utilize subliminal exposures in order to prevent participants from knowing that they have seen the stimuli in the experiment. These subliminal exposures were supported by Bornstein’s (1989) meta-analysis, which found that the mere exposure effect was stronger in experiments with subliminal exposures versus those with supraliminal exposures. Bornstein and D’Agostino (1994) argued that the superiority of subliminal exposures was due to the fact that supraliminal exposures lead to increased recognition, which might inhibit the mere exposure effect. However, these findings have been contested by other researchers. Stafford and Grimes (2012) noted that Bornstein’s meta-analysis included 9 studies which found a stronger mere exposure effect using subliminal exposures and compared them to over 200 studies which had not. Furthermore, there have been a growing number of experiments that have demonstrated that recognition does not inhibit the mere exposure effect, but actually is necessary for it to occur. For example, three experiments by Newell and Shanks (2007) compared recognition and liking across four conditions: 40 ms presentations with 3 exposures, 40 ms presentations with 9 exposures, 400 ms presentations with 3 exposures, and 400 ms presentations with 9 exposures. When stimuli were presented for 400 The Mere Exposure Effect and Classical Conditioning 8 ms presentations and had 9 exposures, both recognition and liking ratings increased. However, the mere exposure effect was not present for any of the other conditions. Similar results were obtained by de Zilva et al. (2013), who generated a mere exposure effect through continuous flash suppression (CFS). In CFS, two different images are flashed to each eye. In the suppressed condition, the target stimulus is flashed in one eye while a Mondrian pattern is flashed in the other. Because of the perceptual dominance of the Mondrian patterns, the suppressed condition leads to the perception of the pattern, but not the target stimulus. The unsuppressed condition is similar to the suppressed condition, but the target stimulus is superimposed on the Mondrian pattern. Thus, the stimulus is perceived. In this study, the mere exposure effect only occurred in the unsuppressed condition. Other researchers have found ways to induce a mere exposure without the use of subliminal exposures. One of these techniques is utilizing a divided attention task. One of the first instances of this was found in Ye and Van Raaij’s (1997) experiment, where they presented Chinese students studying at a Chinese university with numerous trials consisting of pairs of commonly-used Chinese characters. In each pair, one of the characters was marked as the “attended” stimuli with an asterisk. When subjects were asked to evaluate how much they liked the characters, they were put into two groups; one group evaluated the attended stimuli, while the other group rated the unattended stimuli. The attended stimuli group’s liking ratings did not differ from the control group, who were not presented any stimuli and were simply asked to rate the characters. However, the mere exposure effect did occur in the unattended stimuli group. Similar results were obtained by Sutherland et al. (1999). In this study, participants viewed slides with three advertisements arranged horizontally, and participants were asked to focus on the center advertisement. Then, three advertisements were presented in a similar manner and the The Mere Exposure Effect and Classical Conditioning 9 participants were asked to choose which advertisement they liked the most. They found that advertisements that were presented to the sides of the central advertisement (divided-attention condition) were more significantly preferred compared novel advertisements. Because of the controversy surrounding the use of subliminal stimuli, supraliminal exposures were used in the present experiment since classical conditioning experiments with subliminal exposures have only shown slight attitudinal changes (Olson & Fazio, 2001). The amount of exposure time to the stimuli (2000 ms) was chosen because this amount of time was used in Zajonc’s (1968) original experiment. The number of repetitions in each condition (1, 2, 5, 10, and 20) were chosen from a variety of mere exposure experiments in order to assess the effect of repetition in a number of various conditions. In the experiment, a face and then an image were shown on the screen for 2000 ms each, with the face presented first, followed by the image. After seeing all face-image pairs, participants were asked to rate how much they liked the previously presented faces as well as new faces on a 6-point Likert scale. The mere exposure effect would be indicated by higher liking ratings for previously presented than new faces. It was hypothesized that, if the mere exposure effect depends on classical condition, then the mere exposure effect would be the smallest with the faces paired with negative images because this condition would counteract the lack of negative consequence after viewing a neutral image. Then, the scores of faces paired with neutral images and the scores of faces paired with blank screen would be greater than the faces paired with negative images. Both the neutral condition and the blank condition would only be affected by the mere exposure effect, but the blank condition was included in order to replicate the conditions found in other mere exposure effect experiments (where the neutral stimulus is not followed by an image). The faces paired with positive images were expected to benefit from both the mere exposure effect and positive The Mere Exposure Effect and Classical Conditioning 10 reinforcement, and therefore, it was predicted that they would have the largest proportion of preferred faces. Methods