Don't Think About a White Bear
In 1987, psychologist Daniel Wegner ran an experiment that would reshape clinical thinking for decades. He asked college students to spend five minutes not thinking about a white bear, and they failed spectacularly, reporting the forbidden thought more than once per minute despite explicit instructions to avoid it. Then came the twist. When subsequently told to think about the bear freely, these same participants thought about it far more than a control group that had never been told to suppress, as if the act of suppression had loaded a spring that now released with amplified force.
Wegner called this the "ironic rebound effect," and it became foundational. His ironic process theory proposed that our minds run a background monitor checking whether the forbidden thought has surfaced, and that very monitoring keeps it alive. The finding penetrated clinical practice with remarkable speed: national treatment guidelines for depression, anxiety, and PTSD now classify thought avoidance as maladaptive, CBT teaches patients to face intrusive thoughts rather than push them away, and therapists routinely warn that suppression makes things worse.
Then two Cambridge researchers decided to test what everyone already knew.
Three Days of Pushing Thoughts Away
Zulkayda Mamat was a doctoral student in Michael Anderson's lab at Cambridge's MRC Cognition and Brain Sciences Unit when the pandemic began driving anxiety and depression to record levels. Anderson had spent years studying inhibitory control over memory, accumulating neuroimaging evidence that the prefrontal cortex can suppress unwanted memories without the predicted rebound. Clinical guidelines said otherwise, so they ran the trial.
They recruited 120 adults from 16 countries, each listing 20 personally feared future events, 20 positive hopes, and 36 neutral scenarios. For each fear, they provided a cue word and a single word capturing a central detail. Participants were randomly assigned: 61 would train to suppress fears (Suppress-Negative), 59 would suppress neutral thoughts instead (Suppress-Neutral), controlling for training effects, social interaction, and placebo.
Training lasted three days via individualized videoconferencing. Red cues meant suppress: recognize the event, then block all imagery; green cues meant imagine vividly. Each fear was suppressed 36 times total across the three days. Mental health was measured at three time points using validated clinical scales for depression, anxiety, worry, affect, and well-being.
Rebound That Never Came
Four decades of ironic-process research predicted the outcome: participants should suppress temporarily, then experience a surge in fear accessibility and emotional intensity, with the most anxious suffering the worst rebound. None of that happened.
After training, participants recalled suppressed fears less often than baseline events (ηp² = 0.066), rated them as less vivid (ηp² = 0.053), and only 1 of 120 showed higher recall for suppressed items. The feared thoughts also lost emotional power. Affective responses dropped far more for suppressed events than for baseline events that were encoded but never suppressed, with a large effect size (ηp² = 0.239) and a relative risk of anxiety rebound of exactly 1.0, meaning suppression caused no more emotional worsening than doing nothing at all.
The mental health results inverted the clinical prediction entirely. Participants who suppressed fears showed reduced depression, worry, and negative affect alongside improved well-being, and a principal components analysis across all six mental health measures found significantly greater overall improvement in the fear-suppression group than in the control group (t = −2.01, P = 0.045). Depression worsening risk dropped 57.4% (RR 0.43, 95% CI: 0.23 to 0.78). Well-being decline risk fell 44.2%.
But the most striking result is this: the sickest got the most help. Higher trait anxiety before training predicted dramatically greater improvement on the composite mental health measure (r = −0.52, P < 0.00001), and among participants with clinically concerning anxiety scores, worry dropped 44%, while 11 participants meeting criteria for probable PTSD showed large improvements in depression (ηp² = 0.605), negative affect (ηp² = 0.439), anxiety (ηp² = 0.427), and well-being (ηp² = 0.425) that persisted at three months. Positive thinking did nothing. Imagining hopeful future events provided no unique benefit over imagining neutral ones.
The Strongest Counterargument
The most forceful objection is that Mamat and Anderson's Imagine/No-Imagine task is not the same kind of suppression Wegner studied. Wegner's participants were told to suppress during free-form stream-of-consciousness reporting, while the Cambridge protocol used structured, cue-directed suppression with explicit instructions to push imagery from mind without substituting a distracting thought. The two may recruit different cognitive mechanisms, and critics could argue the Cambridge results simply don't address the original phenomenon. If distraction-based avoidance (thinking of something else to dodge the feared thought) is the clinically relevant form of suppression, then the finding that directed retrieval stopping works does not invalidate the guidelines warning against avoidance. Anderson and Mamat acknowledge this distinction but argue that clinical guidelines don't make it either, treating all thought suppression as uniformly maladaptive when the evidence against it comes from a narrow experimental paradigm.
What We Didn't Prove
This is one trial of 120 participants with no clinical diagnosis required for enrollment. Participants reported symptoms on validated scales, and the most anxious subgroup showed the largest benefits, but nobody was undergoing active treatment for a diagnosed anxiety disorder or PTSD, making it impossible to know whether this approach would work as well in people currently receiving clinical care for these conditions. The three-month follow-up showed sustained depression reduction in the fear-suppression group, but the aggregate advantage over the control group did not persist at three months for the full sample; durable gains appeared only among the most symptomatic participants, raising questions about whether the technique works broadly or only in people with enough symptoms to improve. No independent replication exists. The online format introduces training-fidelity variability across 16 countries that a clinic-based trial would eliminate.
The Bottom Line
For nearly 40 years, a single 1987 experiment with college students and white bears has shaped how clinicians worldwide treat intrusive thoughts. A Cambridge trial designed to directly test the clinical assumption found the opposite: training people to suppress distressing thoughts reduced their vividness and emotional power, improved mental health on every measured dimension, and produced the largest benefits in the most vulnerable participants. Positive thinking, by contrast, provided no measurable advantage. The study is small, unreplicated, and used a specific suppression technique that may not match what patients do spontaneously. But it asks a question the field had stopped asking: what if the century-old prohibition against thought suppression was wrong?
What You Can Do
When a distressing thought intrudes, try what the Cambridge study used: acknowledge what the thought is about, then deliberately clear your mind of all imagery related to it, aiming not for a replacement thought but for blankness itself. Don't argue with it or distract from it. Push it out and leave the space empty. Practice matters; the study found effects after three days of repeated sessions, not a single attempt. This is not a substitute for professional treatment if you have a diagnosed anxiety disorder, depression, or PTSD. It is evidence that the reflexive clinical advice to never suppress a negative thought may be less universal than we assumed, and that a structured version of "don't think about it" can be practiced as a skill rather than feared as a failure of coping.