Mere-Exposure Effect: Why Familiar Things Start to Feel Better

Something you have met before is easier to process the next time. The eye finds the shape faster, the ear parses the sound more smoothly, the word comes apart into syllables without effort. One account of the effect, the fluency account, says that this ease is a real sensation which arrives without a label saying where it came from, so the mind reads it as a property of the thing: this one is pleasant, that one is odd. Rolf Reber, Piotr Winkielman and Norbert Schwarz showed in 1998 that making a picture easier to process in other ways, with a matching prime, a longer exposure or a higher contrast, also made people like it more, which is what that account predicts. Zajonc himself rejected it. In 2001 he argued that the effect cannot be explained by fluency or by recognition, since it is strongest when recognition is at chance, and proposed that repeated exposure with nothing bad following works like a mild form of conditioning: the thing becomes a signal of safety. The two accounts agree that repetition is converted into liking by a route that never passes through a judgment about quality; they disagree about the route.

Robert Zajonc's claim, which was controversial when he made it, was that this does not require recognizing the thing. Preference can run ahead of memory. The strongest evidence is the 1980 experiment with shapes flashed for a millisecond: people could not say which shapes they had seen, yet they preferred them. Robert Bornstein's 1989 meta-analysis of 208 exposure experiments put the average effect at a correlation of 0.26, found it across stimulus types, and found it larger when exposures were too brief to be recognized than when they were plainly visible, which is consistent with Zajonc's claim.

The effect has limits, and the limits are as useful as the effect. It is most reliable for stimuli that start out neutral or mildly positive; for something already disliked, repetition usually does not help and can harden the dislike. It grows with the number of exposures and then flattens or reverses, because once a thing is thoroughly familiar, further repetition adds boredom rather than fluency. Complex stimuli tolerate more exposures than simple ones before that turn. And the gain is in liking, not in quality: the same mechanism makes a good product and a bad product both feel more comfortable after the twentieth advertisement.

One consequence is awkward and worth stating plainly. Preferences that feel like taste can be records of what you happened to be exposed to, and the exposure may have been bought. That does not make the preference false; a song you love because you heard it 50 times is still a song you love. It does mean that "I just like it better" is not evidence that the thing is better.

Robert Zajonc set out the effect in a monograph called "Attitudinal effects of mere exposure," published as a supplement to the Journal of Personality and Social Psychology in 1968. The best-known of its experiments used nonsense words. Participants were shown seven-letter strings presented as Turkish words, including the five in the warm-up above, and asked to pronounce each one as it appeared. Different words were shown different numbers of times: none, once, twice, five, ten or twenty-five times. Afterwards participants rated whether each word meant something good or something bad. The ratings rose with the number of showings, and the same pattern appeared in companion experiments using Chinese characters and photographs of faces. This page reports the direction of that result rather than its means, because the monograph's tables were not available to check.

The experiment that made the strong claim came 12 years later. William Kunst-Wilson and Robert Zajonc, publishing in Science in 1980, built 20 irregular octagons by a random process and split them into two sets of ten. Half of the participants were shown one set and half the other, five exposures per shape, at a single millisecond each, with the light dimmed by filters until pilot testing showed that recognition had fallen to chance. Participants were told the flashes would be too brief to see anything and were asked to acknowledge each one.

In the second phase, two groups of 12 participants saw ten pairs of shapes under normal viewing conditions, each pair containing one old and one new octagon, and said which they liked better and which they had seen before. Recognition accuracy was 48 percent, which is chance. Liking picked out the old shape 60 percent of the time. Sixteen of the 24 participants preferred old shapes overall, while only 5 recognized them at better than chance.

The demo on this page uses the visible version of the procedure rather than the subliminal one: a browser cannot control exposure at one millisecond, and nothing here is flashed below the threshold of awareness. Three readings of a five-word list is the 1968 design in miniature.

Mere exposure is the quiet part of most persuasion, because it needs no argument. The rows below are illustrations, not findings from the studies cited on this page.

You cannot undo an exposure, so the counter-measures are about keeping familiarity and quality in separate columns.

• Ask what you can say about it other than that you have seen it before. If the only answer is a feeling of ease, you have a familiarity reading rather than an evaluation.
• Compare blind. Cover the labels and test the product, the typeface, the wine or the headline against alternatives. This is the one move that cancels the advantage of the familiar option.
• Count the exposures. A brand you have seen a thousand times and one you have seen twice are not on the same footing before you start.
• Give a new design a second hearing. Rejection on first sight is predicted by the same mechanism; ask again after two weeks, and ask what specifically does not work.
• When the decision matters, write the criteria first. Fixed criteria make a familiar name compete on the same terms as an unfamiliar one.
• Expect the turn. Repetition stops adding liking at some point and starts subtracting it, so a campaign or a jingle that was working can go sour without anything else changing.

Repetition drives several different effects, and they differ in what it changes.

• It is not the illusory truth effect, where repeating a statement makes it feel more likely to be true. That one acts on belief; mere exposure acts on liking, and a shape or a nonsense word has no truth value to be raised.
• It is not the availability heuristic, where things that come to mind easily are judged more common. Availability turns fluency into a frequency estimate; mere exposure turns it into a preference.
• It is not the bandwagon effect, which is liking something because other people do. Mere exposure needs no other people and no information at all, only repetition.