In the 1940s, psychologist Carl Hovland was studying wartime propaganda for the U.S. Army when he noticed something unexpected: messages from credible sources were more persuasive immediately after delivery, as you would expect. But something strange happened over time. The effect of source credibility faded. And in some cases, messages from low-credibility sources became more persuasive over time — not less.
Hovland and Walter Weiss published their findings in 1951, calling this the sleeper effect. It is one of the most counterintuitive findings in the psychology of persuasion, and it tells us something important about how memory, belief, and communication actually work.
The Classic Experiment
In Hovland and Weiss's original study, participants read persuasive messages on a set of topics. Some messages were attributed to high-credibility sources (e.g., a respected scientific journal) and others to low-credibility sources (e.g., a tabloid). Immediately after reading, participants showed the expected pattern: high-credibility messages were more persuasive.
But when Hovland and Weiss tested participants again four weeks later, something had shifted. Attitude change from high-credibility sources had decreased. Attitude change from low-credibility sources had increased. The gap had essentially closed.
The sleeper effect refers to this dissociation: delayed persuasion from a discounted source, or the reversal of initial attitude change over time.
The "Discounting Cue" Hypothesis
The dominant explanation comes from what psychologists call the discounting cue hypothesis, developed in the 1970s by Gruder, Cook, Hennigan, and colleagues, and elaborated by Anthony Pratkanis and colleagues at UC Santa Cruz.
The idea is this: when we encounter a persuasive message from a low-credibility source, we discount the message — we mentally tag it as "this came from an unreliable source, so I shouldn't change my mind." The message content and the discounting cue are both stored in memory, but they become dissociated over time.
Memory is imperfect and unequal. We tend to retain the gist of a message — its central claims — longer than we retain its contextual tags, including where it came from and why we should doubt it. As the discounting cue fades, what remains is the underlying argument, stripped of the reason we initially dismissed it.
The result: attitude change that was initially suppressed gradually surfaces as the inhibiting cue becomes less accessible.
Why the Effect Is Difficult to Replicate
The sleeper effect has a complicated history in psychology. Early attempts to replicate it often failed, leading some researchers to question whether it was real at all. Later work by Pratkanis, Greenwald, Leippe, and Baumgardner (1988) identified the specific conditions under which it reliably occurs:
- The message must be strong enough to produce attitude change on its own — a weak argument will not produce delayed persuasion even without discounting.
- The discounting cue must come after the message, not before. If you are told the source is unreliable before you read the message, you process it skeptically from the start. If you learn the source is unreliable after, you have already engaged with the content.
- Sufficient time must pass — typically at least three to four weeks — for dissociation to occur.
When these conditions are met, the sleeper effect is robust. When they are not, it does not appear. This explains why initial replications failed: many studies presented the source information before the message.
What the Sleeper Effect Reveals About Memory and Persuasion
The sleeper effect matters beyond its historical curiosity because it exposes something true about how persuasion and memory actually work.
We are less protected from persuasion over time than we think. When we encounter an argument and dismiss it — whether because the source is dubious, the timing is bad, or we are in a skeptical mood — we often believe we have neutralized it. We have not necessarily done so. The argument may remain in memory, and our reasons for dismissing it may fade faster than the argument itself.
This has implications for misinformation. Research on corrective information suggests that debunking false claims can paradoxically reinforce them under some conditions: repeating the false claim to deny it ensures the claim stays in memory, even as the denial fades. The myth can outlast the correction.
Familiarity is not the same as credibility. One of the most reliable findings in attitude research is that mere exposure to an idea increases its perceived truth — the so-called illusory truth effect. The sleeper effect intersects with this: repeated exposure to a claim, even a discredited one, can increase its persuasive power over time simply by making it feel more familiar and therefore more plausible.
Practical Implications
For anyone who communicates — educators, journalists, public health workers, policymakers — the sleeper effect suggests a few things worth taking seriously:
Present the discounting information before the message, not after. If you are warning people about a misleading claim, lead with the warning. Letting people engage with the content first and then revealing the source is unreliable may not work as intended.
Repetition carries risk. Repeating a false claim to debunk it keeps the claim alive in memory. The inoculation research tradition (Lewandowsky, Cook, and colleagues) suggests that explaining the technique of manipulation, rather than repeating the specific false claim, is a more effective approach.
Do not assume that initial rejection is final. If a message was compelling on its own merits, the discounting that prevented initial attitude change may erode over time. In persuasion contexts, what someone says after reading is not always what they believe three weeks later.
A Note on What Has Changed
It is worth acknowledging that the sleeper effect was identified in a media environment very different from our own. Hovland studied propaganda through pamphlets and newspaper attribution in the 1940s. Today, source information is fragmented, viral, often stripped away by the time a message is shared, and rarely associated with a single credible or non-credible outlet.
The discounting cue that inhibits persuasion depends on source awareness. In an environment where sources are routinely invisible, the inhibiting tag may never form in the first place — meaning the sleeper dynamic might be less relevant, and direct, undiscounted persuasion more pervasive, than the original studies suggest.
What the sleeper effect reveals remains: the persuasiveness of a message and our initial assessment of it are not the same thing, and memory slowly separates them.
