No, this isn’t (again) about the Stanford Prison Experiment or Milgram (although there have been successful replications there)… Today I want to briefly focus on cognitive dissonance. This is one of those psychological concepts that almost everyone has heard of at some point. Even those who have never taken a psychology course usually know the basic idea.

It boils down to the fact that people don’t like it when their behaviour and beliefs clash. So, if they do something that doesn’t actually align with what they believe, they sometimes adjust their beliefs so that everything fits together nicely again.

A few examples to illustrate this:

• Smoking is unhealthy, but I smoke anyway. Then I might start thinking that the health risks are exaggerated.
• I bought a much too expensive car. Then I might start to believe that the purchase was actually perfectly rational. (Note for my wife: this is purely coincidental as an example).

The idea sounds quite intuitively correct and human. And it became one of the most influential theories in social psychology. But in recent years, something interesting happened. Not primarily with the theory itself, but with one of its most famous experiments. And in that way, perhaps ultimately with the theory after all.

In 2024, one of the largest replication studies ever on cognitive dissonance was published. Not a small experiment with a few dozen students, but a preregistered international collaboration between 39 research groups from 19 countries, involving nearly 5,000 participants. In this study, the researchers wanted to retest a classic paradigm: the so-called induced compliance experiment.

In this experiment, researchers ask participants to write arguments in favour of something they actually oppose. Imagine defending higher university tuition fees even though, as a student, you disagree with them. According to the theory, this creates tension between what people believe and what they publicly advocate. That tension, in turn, should push them to adjust their attitude slightly in line with their behaviour.

The crucial prediction does not simply concern writing the essay. The theory places particular importance on choice. To understand why, we need to look a little closer. People should change their attitude mainly when they voluntarily choose to defend a position they do not support. If they feel someone forced them into it, the effect should be much weaker.

And that is exactly where the replication ran into trouble. The researchers did find that participants became slightly more positive about the position they had defended after writing a counterattitudinal essay. However, they failed to find the classic difference between high-choice and low-choice conditions. In other words, writing the essay appeared to influence attitudes, but the famous choice manipulation on which decades of research relied did not produce the expected effect.

That does not mean cognitive dissonance has suddenly vanished. Fortunately, science does not work that way. No single experiment carries an entire theory, and cognitive dissonance rests on far more evidence than this one paradigm alone. Researchers have developed other experimental approaches, reported other findings, and proposed numerous theoretical extensions over the years.

What this study did do was make it harder to treat this particular experiment as definitive proof. The reactions that followed reflected exactly that. Some researchers argued that the replication may not have created the right conditions to trigger the effect. Perhaps the essays were too long. Perhaps participants experienced the choice manipulation differently than in earlier studies. And perhaps the COVID context influenced the results. Researchers have suggested many possible explanations.

Others reanalysed the data and found that the subjective experience of choice may still matter. Participants who genuinely felt free to choose seemed to show more attitude change than those who felt they had little choice.

So what does a failed replication actually contribute to science?

For years, the standard story sounded something like this:

Cognitive dissonance works. Just look at this famous experiment.

Today, the story sounds more like this:

Cognitive dissonance is probably real, but this famous experiment may not demonstrate it as clearly as we once believed.

Again, that does not mean cognitive dissonance has disappeared. But when one of the best-known demonstrations of a theory turns out to be less robust than expected, researchers inevitably reassess how much weight they should give that evidence. That may make science less exciting. You might think, Come back when you know for sure. Yet this is exactly how science should operate.

A small experiment from the last century became famous. Textbooks adopted it. Thousands of students learned about it. Then an international team preregisters a replication and tests it with thousands of participants. The results prove less convincing than expected. Researchers debate the findings, publish new analyses, and refine their theories.

This week, I saw students present research that could challenge part of my own thinking. Rather than feeling threatened, I became curious. They may have made a mistake. Equally, I – and others – may eventually need to revise our views. That is not a reason to stop. That is how science moves forward.