You're the judge. What bail do you set?

Three psychologists and a one-line review

Jeff Greenberg, Sheldon Solomon and Tom Pyszczynski met in the late 1970s as doctoral students in social psychology at the University of Kansas. Reading Becker, they thought his ideas could be tested in the laboratory. They presented them in 1984, at a meeting of experimental social psychologists, under the name that stuck: terror management theory. The terror in the name is the fear of death.

A paper they sent to American Psychologist came back with a review one sentence long: “I have no doubt that this paper would be of no interest to any psychologist, living or dead.” Another editor told them their ideas would not be taken seriously without evidence. That is how they came to the judges.

From that scheme they drew three predictions, each tested separately.

1. Remind people of death and they will defend their view of the world more fiercely. Hence the bail amounts of 1989.
2. Raise their self-esteem and death will frighten them less. In 1992, students who had received flattering personality feedback reported no more anxiety after graphic footage of death than after a film of landscapes. The praised students also reacted less physiologically when told they would receive electric shocks.
3. Threaten their view of the world and thoughts of death will come to mind more easily. This is measured with word fragments: COFF_ _ can be coffee or coffin, SK_ _L can be skill or skull.

There was one more detail, set out in 1994. The effect appears after a pause, when the thought of death has left attention but is still close. Right after the questions, people defend themselves rationally: I'm healthy, I have time. The defence of values comes later. That is why the exercise has a word grid. That grid is one of the pauses in the materials published by the theory's laboratory at the University of Missouri, with the same neutral words.

And one more thing: in none of the six experiments of 1989 did the questions about death produce measurable anxiety or arousal. The theory is about a fear you are not aware of.

A theory that seems to explain everything

Over the next 30 years the authors counted more than 500 studies, run in more than 25 countries. The theory was tested on the most varied kinds of behaviour. Here are some of the most cited studies.

In 2010 a meta-analysis gathered 277 experiments from 164 articles. 80 per cent had found the predicted effect, and its average size placed it, the theory's authors wrote, among the strongest 20 per cent in social psychology.

A theory that explains voting, hot sauce, tanning and bail is either very deep or very elastic. The only way to find out is to run the experiments again.

The argument that followed

In 2020 Tom Pyszczynski and two colleagues replied that the project had broken its own plan by including laboratories that were too small, and that the effect appears if you keep only the white, US-born participants from the laboratories guided by the authors. In the final version, published in 2022, the organisers applied the inclusion rule written in the plan (at least 60 participants per laboratory) and kept their conclusion. The critics had found that, even under that rule, the effect appeared in their subgroup. But their analysis still included data some laboratories had collected before the plan was registered; without those data, the final version found no significant effect in the same subgroup. An independent reanalysis tried 45 ways of choosing participants and found evidence for the effect in only two of them.

Other large replications, registered in advance, also came out at zero: 1,356 participants in Germany in 2021; 895 adults in Japan in 2023; 784 Americans online in 2022. In Barcelona in 2019, 128 students did not confirm another of the theory's classic experiments either. Five studies published in 2023 that followed the standard procedure did not find the effect either.

One result held. In 2025, 22 laboratories in 11 countries, with 3,447 participants, confirmed that questions about death make death-related words slightly easier to call to mind: about 0.17 more words out of 25 fragments. But the effect was larger without a pause, the opposite of what the theory predicts, and the authors stress that it does not necessarily mean people defend their values more fiercely.

What is still standing

None of this proves Becker wrong. An idea about human nature can be true even when a particular experiment fails to catch it. But terror management theory became famous through its experiments, and it is the experiments that are now shaking.

Other explanations were there from the start. In 2006 Heine, Proulx and Vohs argued that any threat to the meaning of the world, such as uncertainty, rejection or absurdity, produces the same reactions, and death is only one of them. In the 2010 meta-analysis, when the control group received exactly such a threat, the effect of death shrank to less than a third, though on the basis of only seven comparisons. Evolutionary psychologists objected that fear is a useful adaptation, and that evolution would not have produced a mind that has to defend itself against its own fear.

A 2013 reanalysis noticed something else: experiments by teams close to the founders, which made up half of the total, found larger effects than everyone else's. Even the theory's defenders, in their 2020 reply to Many Labs 4, acknowledged that the old studies had too few participants, and that editors often asked authors to drop experiments that had not worked.

What remains is a good question and a powerful intuition, backed by experiments weaker than they looked. And a lesson about ourselves. A theory that explains everything is attractive for exactly the reason Becker described: it puts the world in order.

The enemy who reassures us

Say you read that the big artificial intelligence companies are investigating tens of thousands of security incidents. Two explanations are close at hand. Either they knew all along and hid it, or only now do they have the tools to read and understand what was already in their logs. The first is a conspiracy. The second is more unsettling, because it means nobody was in control.

Becker sensed this choice as early as 1969. We cannot allow ourselves, he wrote, to believe in a disinterested, bureaucratic evil: “Evil must happen because someone cares.” In 2010 Sullivan, Landau and Rothschild, psychologists from the same tradition, put the idea to the test.

Students at the University of Kansas, some of whom had first been shaken in their sense of controlling their lives, read an invented article about Al Qaeda in three versions: a powerful and mysterious organisation, a powerful one well understood by intelligence services, and a weak one (“These guys are a joke”). Among the shaken students, the powerful, well-understood enemy and the weak one raised the risks they saw in their future. The mysterious one did not, and those who had read about it felt more in charge of their own lives. “An enemy known to be incompetent is unlikely to be a satisfactory target…”, the authors had predicted.

The reassurance came from the combination of power and mystery: an enemy big and vague enough to account for all the scattered dangers of the world. The portrait fits the artificial intelligence laboratories of conspiracy theories rather well, all-powerful and opaque.

The same caution applies here. The four studies had between 59 and 104 students each, some results sit at the edge of significance and, as far as I could find, nobody has replicated them directly. The broader theory, that a lack of control pushes us towards conspiracies, has fared badly in tests. A 2020 meta-analysis found a practically null experimental effect, and a replication registered in advance, with 829 participants and co-signed by one of the authors of the original idea, found no effect on belief in a God who controls the world. The link between a lack of control and conspiracy belief shows up mainly in surveys; in experiments that shake people's sense of control, it nearly disappears.

One thread is sturdier. People tend to see intention in ambiguous actions, and harm done on purpose hurts more than the same harm by accident: in one experiment, identical electric shocks felt more painful when a partner had chosen them deliberately. An enemy makes each blow heavier, but makes the world easier to understand.