Cognitive distortions are recurring shapes that unhelpful thinking takes: assuming you know what somebody thinks of you, treating one bad afternoon as proof of a permanent rule, hearing nine compliments and one criticism and remembering the criticism. The term came out of clinical work with depressed patients in the early 1960s, where the content of what people said about themselves varied enormously and the shapes underneath it kept repeating. [beck-1963-thinking]
The list of them is genuinely useful and you can learn it in an afternoon. What almost nobody tells you is that learning it does very little on its own, and the reason is that identification was never the technique.
The distortions worth knowing, and the question that tests each one
Below are eight of the patterns named most often, alongside what each sounds like in ordinary words and a question that puts it to an actual test. The third column is the part that matters, because a thought is only shown to be distorted by evidence and never by the label.
| Distortion | What the thought sounds like | The question that tests it |
|---|---|---|
| All-or-nothing thinking | ”If it is not perfect it is a failure.” | Where between the two did it actually land, and who else would say so? |
| Overgeneralising | ”This always happens to me.” | How many times, exactly, in the last year? Write the number. |
| Mind reading | ”She thinks I am useless.” | What did she actually say or do, and what else could explain it? |
| Catastrophising | ”If I get this wrong I will lose my job.” | What happened the last five times somebody here got that wrong? |
| Emotional reasoning | ”I feel like a fraud, so I must be one.” | What would I conclude about somebody else from this same evidence? |
| Should statements | ”I should be further along by now.” | Whose rule is this, and what actually happens if I do not follow it? |
| Labelling | ”I am an idiot.” | What did I do, described as a behaviour rather than as an identity? |
| Discounting the positive | ”They only said that to be kind.” | What would have to be true for the compliment to have been genuine? |
Most versions of this list run to ten, some to fifteen, and the differences between versions matter far less than they look. There is a section further down on why that is, and it is more useful than memorising any particular version.
Naming the distortion is the easy half
Spotting the pattern is the part that takes an afternoon, and on its own it moves belief hardly at all. The experience that follows, and the one that makes people abandon the whole approach, is thinking “that is catastrophising” and continuing to believe the thought anyway.
This is not a failure of effort or of insight. A label is a description of a thought, and a description is not an argument against it. Knowing the name of an optical illusion does not stop you seeing the illusion, and knowing that a thought is overgeneralising does not supply you with the counter-evidence that would make it feel wrong.
There is a second way it goes badly, with the same root. People who have just learned the list start narrating their own thinking during a bad evening, which adds a commentary track to a loop that was already running. Our guide to what rumination actually is covers why that loop keeps going and why thinking harder about the thinking makes it worse rather than better.
What actually moves a belief is testing it
Three steps, and the change happens in the third. Catch the thought in words, generate at least one alternative account of the same situation, then look for evidence on both sides that somebody other than you could check.
Catch it as a sentence, not as a mood. Write down the exact words that arrived: “I am going to be sacked over this”, not “I felt anxious about work”. A sentence can be tested and a mood cannot, and most of the value of writing it down is that the sentence usually looks weaker in handwriting than it sounded in your head.
Generate an alternative that fits the same facts. This is the step people do badly, because they reach for a nicer thought rather than a plausible one. “Everything will be fine” is not an alternative account, it is a wish, and you will not believe it. “My manager is short with everybody in the week before a deadline” is an alternative account, because it explains the same behaviour differently.
Then look for evidence both ways. Evidence means things a third party could verify: what was actually said, what happened the last five times, what the numbers were. Feelings are not evidence for the thought. They are the thing the thought produced, and treating them as evidence is precisely what emotional reasoning is.
Something worth knowing about the written version of this. In a study of cognitive therapy trainees who practised both techniques on themselves, written thought records and behavioural experiments, which means going out and testing a prediction in the world, were experienced quite differently, with the experiments rated as more powerful and more convincing. [bennett-levy-2003-mechanisms] That study looked at therapists in training rather than patients, and it measured what they reported rather than whether their symptoms changed, so it does not establish that experiments treat anything better. It does match the common complaint that writing “they probably were not annoyed with me” is much weaker than asking them.
Test one thought
Pick one sentence that has been going round. Set the timer, write the thought exactly as it arrived, then write an alternative account that fits the same facts, then two columns of evidence for and against each. Ten minutes is enough for one thought and too short for three.
10:00
If you finish early with nothing in the evidence columns, that is a finding rather than a failure: it usually means the thought is a prediction about the future, in which case the test is something you have to go and do rather than something you can write.
Done. The question now is what you would have to see to change your mind, and whether you could go and look.
If the thinking will not hold still long enough to write anything down, that is a different problem and this is the wrong tool for it. Our guide to what interrupts overthinking covers the approaches that work on the process rather than the content, and it is the more useful place to start.
When the thought is not distorted
Some situations are exactly as bad as they feel, and calling that thinking distorted is not therapy. A thought is distorted when it does not match the evidence, and the evidence is sometimes grim.
This is the failure that does real harm on a page like this one. Somebody who is genuinely being treated badly at work, or whose relationship really is ending, or who really has had a frightening result back, does not need a thought record. They need a plan, or practical help, or somebody to sit with them. Turning the vocabulary on yourself in that situation is a way of dismissing your own read on a real problem, and it works well enough to keep you in the situation longer.
The version aimed at other people is worse. Telling somebody that their reaction is catastrophising, or that they are mind reading, converts a shared vocabulary into a way of closing a conversation, and it is unanswerable in the moment because whoever holds the label decides what counts as distorted. If you find yourself reaching for it during a disagreement, it is being used as a weapon rather than as a tool, and the tell is that you are applying it to their thinking rather than to your own.
The rule that keeps this honest is short. A thought is not distorted because it is uncomfortable, and it is not accurate because it is calm.
The list of ten is a teaching tool, not a diagnostic system
The standard list was assembled to teach the idea rather than derived from data, and the categories overlap so heavily that sorting a thought into the right one is not a meaningful task. The evidence for that is unusually direct: when researchers built a questionnaire specifically to measure ten distortions, the analysis did not find ten separate things.
The Cognitive Distortions Scale asks about ten named errors across two domains, interpersonal situations and personal achievement. [covin-2011-cds] When it was examined in clinical and non-clinical samples, factor analyses supported a one-factor model rather than the ten the instrument was built around. [ozdel-2014-onefactor] That is one instrument in two samples from a single country, so it does not settle the question for every version of the list. It does mean nobody should treat the ten categories as ten distinct mechanisms.
The practical consequence is worth stating plainly. “I am an idiot for not seeing this coming, this always happens to me” is labelling, overgeneralising and hindsight all at once, and the test for all three is the same test. Readers who treat the list as a classification system spend their evenings sorting thoughts into boxes, which feels like progress, produces no change, and is a considerably more comfortable activity than looking for evidence that might contradict something you believe about yourself.
Where this comes from and how well it works
This is cognitive behavioural therapy, usually shortened to CBT, and the whole approach grew out of the observation that the thinking of depressed patients had recurring shapes worth working on directly. [beck-1963-thinking] A review that identified 269 meta-analyses and examined 106 of them described the evidence base as strong, with the strongest support for anxiety disorders alongside several other conditions. [hofmann-2012-efficacy] That is a review of reviews rather than of individual trials, so it inherits whatever weaknesses sit inside them, and it is a statement about averages across many people rather than a prediction about you.
Doing it from a worksheet on your own is a weaker version of the same thing rather than a different thing, and the size of the gap has been measured. In an analysis pooling individual data from thousands of participants across dozens of trials of internet-delivered CBT for depression, guided programmes did better than unguided ones by about 0.8 points on the PHQ-9 at the end of treatment, with no detectable difference at six or twelve months, and the advantage concentrated in people who started with more than mild symptoms. [karyotaki-2021-guided] Read that in both directions. The unguided version worked; it worked somewhat less well, and it worked least well for the people who were struggling most.
No screener on this site measures cognitive distortions, and no validated one exists that would. The free self-assessment hub has screeners for depression and anxiety, which use PHQ-9 and GAD-7, the questionnaires most clinicians use. Those measure symptoms of the conditions this technique is used to treat rather than the distortions themselves, and a number from either is more useful to bring to an appointment than a description of how you have been thinking. Our guides to overthinking and to therapy for anxiety cover the wider picture and what to ask for by name.
When to seek help
Speak to a doctor or a therapist if low mood, anxiety or a change in how you are functioning has lasted more than a few weeks, or if you have been working through thought records on your own for a month or two and the belief has not moved. Say exactly that, because it tells a clinician you have already tried the self-directed version and it shortens the conversation considerably. Ask for cognitive behavioural therapy by name if that is what you want.
Go sooner if the thoughts concern something frightening that actually happened to you, since that points towards trauma-focused treatment rather than this, or if you have stopped doing things you used to manage, or if somebody close to you has said they are worried.
Contact your local emergency services or a crisis helpline if you feel unsafe or have thoughts of harming yourself.
How MyFreud can help
MyFreud is a mobile app that helps you find solutions to problems that have affected your mind and productivity. The alternative account is the step that is hardest to do alone, because the one you generate for yourself is usually the one you already believed, so live coaching sessions give you somebody to produce it with, and each session ends with an actionable plan rather than encouragement. Daily tracking shows which days the thinking bites and what preceded them, which is the evidence a thought record cannot supply from memory, and the notepad is where the exact sentence goes at the moment it arrives rather than four hours later.
Download MyFreud and start today: App Store or Google Play.