Robert Zajonc's 1968 monograph is a strange document to read now, because the effect it reports is so simple that it seems as though it must be an artifact, and it is not.
Show someone a nonsense word — nansoma, saricik, afworbu — a certain number of times, and then ask how favorable a meaning they think it has. Words shown twenty-five times are rated more positively than words shown ten times, which are rated more positively than words shown twice. Zajonc did it with Turkish-looking pseudowords, with Chinese characters shown to people who did not read Chinese, and with photographs of faces from a college yearbook. The relationship between frequency of exposure and rated liking held in all three, and it has since held for melodies, shapes, ideographs, flavors, and advertisements.
No content is involved. Nothing is learned about the stimulus. It simply becomes more familiar, and familiarity is experienced as mild liking.
The best-supported explanation is perceptual fluency misattribution. Repeated exposure makes a stimulus easier for the perceptual system to process; that ease is registered as a faint positive affective signal; and because the signal arrives without a label, the mind attributes it to the most plausible available cause, which is usually a property of the stimulus itself. The feeling is I like this. The fact is I have processed this before.
Two features make this the quietest mechanism in Part 1.
First, it does not require recognition. Effects have been obtained with exposures too brief or too degraded for participants to report having seen anything, and in some studies the effect is stronger when recognition fails, presumably because a recognized stimulus lets you attribute the fluency correctly. This means the ordinary defense against influence — remembering that you were exposed to something — is unavailable. You cannot discount an exposure you cannot recall.
Second, it produces a preference with no accessible reason. Ask someone why they prefer one of two brands they have never used and they will construct an answer, because people are reliably able to produce reasons for preferences whose actual causes they have no access to. The constructed reason then feels like the real one.
The industries built on this are the ones that spend money on frequency rather than on argument. Advertising reach and frequency planning is an exposure-purchasing business, and the creative content is often less important than the count. Political name recognition behaves the same way, which is why low-information ballot races are won by familiar names and why yard signs, which contain no argument whatsoever, are not irrational. Repetition-based propaganda operates on this and on the illusory truth effect in the next chapter, which is the same fluency mechanism applied to belief rather than to liking.
There is a limit worth knowing: the effect is not monotonic forever. Extremely high exposure produces satiation and then decline, and initially disliked stimuli do not become liked through repetition — exposure amplifies the existing valence rather than reversing it.
The only real defense is structural rather than psychological, because you cannot introspect your way to the source of a fluency signal. When a preference has no reason you can state, treat the absence of a reason as information. Ask what you actually know about this option as against its alternative. If the honest answer is that one of them feels more familiar, you have found the cause, and familiarity is not a property of the thing.
Defense: Vaccinating the Mind: Prebunking and Inoculation Against Manipulation