How to Actually Quiet a Thought You Don't Want

Telling yourself not to think something backfires because the instruction contains a contradiction. To confirm that you are not thinking about the thing, some part of your mind has to keep checking for it, and checking means holding it available. Daniel Wegner called the checking process the ironic monitor, and its existence is the best available explanation for why "stop being negative" is such a poor instruction to give a person. The honest version of this story is narrower than the version you will hear repeated. The evidence for a rebound after you stop suppressing is decent. The evidence that suppression makes the thought more frequent while you are actively suppressing is considerably weaker, and the effect sizes throughout are small to moderate rather than dramatic.

The white bear experiment

Daniel Wegner, David Schneider, Samuel Carter and Teri White published "Paradoxical effects of thought suppression" in the Journal of Personality and Social Psychology in 1987. The design is unusually simple, which is part of why it travelled so far.

Participants spoke their stream of consciousness aloud into a tape recorder for five minutes. One group was told to try not to think of a white bear, and to ring a bell every time the bear came to mind anyway. They rang it. Then came a second five-minute period in which the same participants were told they could now think about the white bear. They produced more bear thoughts in that period than a comparison group who had been told to think about the bear from the beginning.

That second finding is the one that matters. Suppression did not remove the thought. It deferred it, and the deferral came with interest. The white bear was chosen precisely because it is arbitrary, unemotional and not something anyone has a stake in avoiding. If the effect appears with a white bear, the reasoning went, the emotionally loaded thoughts people actually try to suppress are unlikely to be easier.

The theory Wegner built to explain it

Seven years later Wegner published "Ironic processes of mental control" in Psychological Review, generalizing the finding into an account of mental control as a whole.

The proposal is that any attempt to control your own mental state runs two processes at once. An intentional operating process searches for content consistent with the goal: distractions, replacement thoughts, anything other than the bear. A separate ironic monitoring process searches for evidence that the goal is failing, which means it searches for the bear. The operating process is effortful and consumes attention. The monitor is comparatively automatic and cheap.

Under normal conditions the operating process stays ahead and control succeeds, imperfectly. Add mental load, time pressure, stress or divided attention and the operating process degrades faster than the monitor does. What remains is a process whose entire job is to detect the unwanted thought, running with nothing to counterbalance it. That is the ironic part: the machinery that produces the failure is the same machinery that was installed to prevent it.

Worth stating plainly, because it usually is not: the monitor is a theoretical construct. Nobody has measured it directly. It is an inference drawn from the pattern of results, and it fits that pattern well, but "the monitor" is a name for a hypothesis rather than an observed structure.

What the meta-analysis actually found

Jonathan Abramowitz, David Tolin and Gordon Street published "Paradoxical effects of thought suppression: a meta-analysis of controlled studies" in Clinical Psychology Review in 2001, pooling the controlled experiments available at the time.

Their conclusion was a small to moderate rebound effect whose magnitude varied with the nature of the target thought and with the method used to count thought frequency. That is a real result, and it is also a considerably more modest one than the popular telling.

Two things deserve emphasis. First, the rebound effect and the immediate enhancement effect are separate claims, and they are not equally supported. Rebound is the finding that the thought returns more often after suppression ends. Immediate enhancement is the claim that suppression increases the thought's frequency during the suppression period itself. The rebound is the more reliably demonstrated of the two. Second, effect sizes in the small to moderate range mean that individual studies with modest samples will often fail to detect the effect at all, and the earlier literature was full of exactly such mixed results. The meta-analysis existed because the primary studies disagreed.

Eric Rassin, Harald Merckelbach and Peter Muris made a related point in a critical review in Clinical Psychology Review in 2000. Laboratory findings on paradoxical effects have been reasonably consistent, they argued, but there is little basis for concluding that thought suppression causes obsessions, phobias or other conditions. A laboratory phenomenon is not automatically a clinical mechanism.

Where the theory has been oversold

Joshua Magee, Kathryn Paige Harden and Bethany Teachman published a quantitative review in Clinical Psychology Review in 2012, testing an assumption that had become common: that people with psychopathology suffer larger suppression failures than everyone else.

They found no overall difference in thought recurrence between groups with and without psychopathology. There was variation by condition, and samples with obsessive-compulsive symptoms showed relatively less recurrence during suppression attempts than non-clinical samples, which runs opposite to the prediction. The differences were typically small and offered only mixed support for the theories that generated them.

You will also see the rebound explained as willpower running out. That is the ego depletion model, and it is in serious trouble. A preregistered replication across 23 laboratories with 2,141 participants, led by Martin Hagger and published in Perspectives on Psychological Science in 2016, found essentially no depletion effect. Ironic process theory does not require a depleting fuel tank. It requires only that the effortful process degrade under load faster than the automatic one, which is a different and less exotic claim.

Why this breaks "stop being negative" as advice

1. The instruction names the target. To follow "stop thinking about how badly that meeting went," you have to identify the thought about the meeting in order to not have it. The instruction supplies the search term.
2. It converts a passing thought into a monitored one. An unpleasant thought you ignore is one event. An unpleasant thought you have committed to eliminating becomes a standing background process.
3. It fails hardest under exactly the conditions that produce it. Stress, fatigue and cognitive load are when people reach for suppression and are also when the operating process degrades. The tool breaks under the load it was picked up to handle.
4. It adds a second failure. The thought returns, and now you have also failed at the instruction, which is its own unpleasant thought to manage.
5. It has no replacement. "Stop" specifies what not to do and leaves the vacancy unfilled, which is the structural problem shared by most negative instructions.

What the evidence supports instead

The workable moves in this literature are substitution and acceptance rather than removal. Giving the operating process something specific to hold beats asking it to hold nothing. Reinterpreting a situation before the emotional response consolidates has better evidence than fighting the response afterward, which is the case laid out in the piece on cognitive reappraisal. And declining to add a second reaction on top of the first, rather than trying to delete the first, is the pattern the acceptance research keeps finding.

This is one of three documented failure modes behind the slogan people repeat most often, which is why "think positive" is usually bad advice. It also explains why a self-talk sentence works better as something to say than as something to stop saying. A statement you have chosen occupies the channel. A statement you have banned patrols it. The broader question of which chosen statements survive contact with a skeptical reader is worked through in why believable self-talk beats grandiose affirmations.