The Threat Simulation Theory: Dreams as Rehearsal for Danger
Why are dreams so often unpleasant? In 2000, Finnish philosopher and neuroscientist Antti Revonsuo proposed a striking answer: dreaming evolved as a nightly rehearsal of danger, a safe place to practice spotting and escaping threats. Two decades of testing later, the theory has real support, real critics, and no verdict.
The threat simulation theory, proposed by Antti Revonsuo in 2000, holds that dreaming evolved to simulate threatening events so we could rehearse detecting and avoiding danger in safety. It has partial empirical support but remains an unproven hypothesis: why we dream has no scientific consensus.
At a glance
- Proposed by Antti Revonsuo in 2000 in Behavioral and Brain Sciences: dreams are an evolved virtual rehearsal of spotting and escaping threats.
- Support is real but partial: traumatized children dream more threats and most people report having had chase dreams, yet realistic survival scenarios are rare in most dream reports.
- It remains an open hypothesis. No study has shown that dream rehearsal improves waking threat avoidance, and why we dream has no scientific consensus.
Where the science stands
What is established here is the record itself: the theory's claims and the published findings around it are accurately reported, and the pattern they form is real. Traumatized children do report more threatening dreams, most people have had chase dreams, and negative content is common in ordinary dream reports. What is open is everything those findings are supposed to prove. Why we dream, and whether dreaming has any function at all, has no scientific consensus. Threat simulation is one proposal inside that open question, not the answer to it.
Even the basic question of how often dreams contain threats is contested. Scored with Valli and Revonsuo's Dream Threat Scale, threatening events appear in a majority of everyday dream reports, at rates above matched waking life. Scored against stricter criteria for realistic life-threats, as Malcolm-Smith's group did, they are rare. That gap is a methods dispute over the definition of 'threat', not a resolved fact, so no single percentage on this page should be read as the number.
One caveat sits outside the academic argument. Whatever nightmares turn out to be for, a nightmare that recurs, distresses you, or follows trauma is a reason to talk to a professional, not something to reframe as useful practice. The research above treats trauma nightmares as data. Your own are not obliged to serve a theory.
What the threat simulation theory says
The threat simulation theory holds that dreaming is an evolved simulation of threatening events. Its biological function, on this account, is the repeated rehearsal of threat perception and threat-avoidance skills, run nightly in a setting where mistakes cost nothing. Antti Revonsuo proposed it in 2000 in the journal Behavioral and Brain Sciences, in a paper titled 'The reinterpretation of dreams: An evolutionary hypothesis of the function of dreaming'. In the ancestral environment, he argued, this rehearsal raised the odds of surviving real dangers.
The theory reframes the unpleasantness of dreams as the point rather than a defect. Chase dreams, attacks, falling, nightmares: negative content is the system doing its job, selecting threatening material because that is what was worth practicing. One thing should be said plainly before any of it is taken as settled: why we dream has no scientific consensus, and threat simulation is one of several competing hypotheses, none of them confirmed.
The evidence Revonsuo built the case on
The 2000 paper assembled several converging observations. Children's dreams and nightmares feature frequent threats and animal aggressors. Recurrent dreams and nightmares preserve old, primitive threat content. Post-traumatic nightmares replay threats long after the danger has passed, and Revonsuo cited reports that dreams in hunter-gatherer societies contain more animal and threat content than modern samples.
The everyday statistics fit the pattern too. Calvin Hall's quantitative content analysis, developed with Robert Van de Castle, had shown by 1966 that aggression and misfortune are common in ordinary dreams. And being chased is among the most widely shared dream themes on record: in the 2003 Typical Dreams Questionnaire study by Nielsen, Zadra and colleagues, about 82 percent of Canadian university students said they had experienced a chase dream at some point. That figure is lifetime prevalence in one student sample, not the share of dreams that are chase dreams.
What happened when researchers tested it
The most direct test came from Katja Valli, Revonsuo and colleagues in 2005. They compared severely traumatized Kurdish children from northern Iraq with less threat-exposed Kurdish children and with non-traumatized Finnish children: the traumatized children reported more dreams, more threatening dream events, and more severe threats. That matches the theory's prediction that real threat exposure fully activates the simulation system.
A 2006 test on 212 recurrent dreams by Zadra, Desjardins and Marcotte returned a mixed verdict. Six of eight hypotheses derived from the theory were supported, and 66 percent of the recurrent dream reports contained at least one threat. But under 15 percent depicted realistic, probable situations critical to physical survival or reproduction: partial support, with a realism problem attached.
Reviewing the accumulated tests in 2009, Valli and Revonsuo concluded that the evidence gave the theory considerable but partial support. Their review used the Dream Threat Scale, a content-analysis system built to score threats in dream reports, and that scoring choice matters more than it might seem, as the next section shows.
The case against it
The strongest published counter-evidence came from Susan Malcolm-Smith and Mark Solms at the University of Cape Town. In 2004 they reported that realistic life-threatening events were rare in a South African student sample, and successful escape from dream threats rarer still. A 2008 follow-up with Turnbull and Tredoux found low rates of threat avoidance, and especially of successful threat avoidance, even among people exposed to real threat. Revonsuo and Valli replied that these tests operationalized the theory wrongly. Much of the disagreement comes down to how strictly 'threat' is defined and scored.
G. William Domhoff has offered a different account altogether. In his reading of the quantitative dream-content record, dreams mostly reflect waking concerns, a pattern called continuity, and dreaming is best understood as an intensified form of mind-wandering with no adaptive function of its own. He does not deny that threat content exists, and he links recurrent dreams and nightmares to the brain's fear systems. What he denies is that threat simulation is what dreaming is for.
Two further problems are worth naming. Laboratory studies of young children's dreams, notably by Foulkes, found them often bland and static, which complicates the argument from children's nightmares. And the theory has a built-in reply to the observation that most modern dreams are not threatening: the system fully activates only under real threat exposure. Critics accept that this is consistent but consider it hard to falsify, and some researchers go further and hold that dreaming is a byproduct of the sleeping brain with no function at all.
Where the theory stands, and how we use it
Threat simulation is a live evolutionary hypothesis with partial empirical support: neither falsified nor confirmed. The central gap is causal. No study has yet shown directly that rehearsing threats in dreams improves waking threat detection or avoidance, and until something like that exists, the adaptive story remains a proposal.
Revonsuo himself has broadened the idea. In 2015, with Tuominen and Valli, he proposed the social simulation theory, on which dreams also rehearse social perception, bonds and interactions. It is likewise a hypothesis under test, and whether it extends, competes with, or absorbs threat simulation is an open question.
On Somniary, threat simulation is one of the lenses our interpreter reads a dream through, alongside emotional-regulation, continuity, Jungian and cross-cultural perspectives. It earns its place because it takes the darkness of dreams seriously without moralizing it. It stays a lens, never a verdict: your own life supplies the meaning, and the theory supplies one way to look.
Frequently Asked Questions
What is the threat simulation theory of dreams?
The threat simulation theory, proposed by Antti Revonsuo in 2000, holds that dreaming evolved to simulate threatening events so that sleepers could rehearse detecting and avoiding danger without real risk. On this view, nightmares and chase dreams are the system working as designed, not malfunctioning. It is a hypothesis with partial empirical support, not an established fact.
Is the threat simulation theory proven?
No. A 2009 review by Valli and Revonsuo found considerable but partial support, while studies by Malcolm-Smith, Solms and colleagues found realistic threats and successful escapes rare in dreams, and Domhoff's dream-content research points to dreams tracking waking concerns instead. No study has shown that dream threat rehearsal improves waking performance, and why we dream has no scientific consensus.
Why do we dream about being chased?
Being chased is among the most commonly reported dream themes: about 82 percent of Canadian university students in a 2003 study said they had experienced a chase dream at some point in their lives. Threat simulation theory reads chase dreams as rehearsal of an ancient pursuit scenario. Continuity-based accounts read them as imagery of waking pressures. Both are interpretations, and neither is confirmed.
Do nightmares serve a purpose?
It is not known. Threat simulation theory proposes that nightmares rehearse danger, and post-traumatic nightmares do fit its prediction that real threat exposure activates threatening dream content. But trauma nightmares disturb sleep and impair well-being, which sits badly with a straightforwardly useful reading, and some researchers hold that dreaming has no function at all. If nightmares recur or distress you, especially after trauma, talking to a professional matters more than any theory.
Go Deeper
The original 2000 paper: the full argument, the evidence base, and peer commentaries with Revonsuo's replies.
View in Sources ↗The 2009 stock-taking by the theory's own camp, honest about where the support is only partial.
View in Sources ↗The main counterposition: continuity with waking concerns, quantitative dream content, and the case that dreaming has no adaptive function.
View in Sources ↗The 1966 normative baseline showing aggression and misfortune are common in ordinary dreams, background to the whole debate.
View in Sources ↗One dream is a night.
Kept together, they begin to show a thread. The Dream Journal letter arrives on the 1st and the 15th: one symbol read closely, one honest idea from the science of sleep.
Free · Twice a month max · One-click unsubscribe
And if you want to keep the dreams themselves: the Companion plan keeps what you cannot see alone.