GUIDE · CBT

Socratic questioning: how to check a thought before you believe it

Your manager emails “Got a minute tomorrow at nine?” and by dinner you’re sure you’re about to be let go. A friend says you’re overthinking it, and the certainty doesn’t budge. Socratic questioning, a core skill in cognitive behavioral therapy (CBT) that takes its name from Socrates and his question-and-answer method, comes at the thought from the other side: you put a few plain questions to it and see what your own answers say.

What Socratic questioning means in CBT

In CBT, the thought you question is usually an automatic thought, the split-second sentence your mind supplies when something happens, like “I’m getting fired” or “I ruin everything.” A therapist using Socratic questioning doesn’t argue that the thought is wrong or hand you a better one. They ask questions that bring in what the thought leaves out, and you draw the conclusion yourself.

The cognitive therapist Christine Padesky drew a useful line in a 1993 keynote address. One kind of questioning sets out to change your mind: the questioner already has the answer and steers you to it with leading questions. The other kind, which she called guided discovery, starts from curiosity and leaves the conclusion to you. She favored the second, partly because in her experience few clients changed for good just because a therapist had shown their thinking to be illogical.

In her definition, good questions are ones you can actually answer, and they point to facts that matter but sit outside your attention right now. They usually start with a concrete moment before moving to what it means. Her reasoning was that mood steers memory, and a well-aimed question can bring back what a dark mood screens out. Padesky also handed her list of questions to clients and found it helped them write balanced responses on their thought records, the worksheet where these questions usually live.

Six kinds of Socratic questions

Judith Beck’s thought-record worksheets for the Beck Institute group the questions much as they appear below, and the control questions echo a Beck Institute post on living with uncertainty. You don’t need all six every time. Pick the two or three that fit the thought and answer those properly.

• Evidence for and against. What makes me think this is true? What makes me think it isn’t, or isn’t completely?
• Other explanations. What else could explain what happened? How would someone who likes me read it?
• What you’d tell a friend. If a friend had this thought, what would I say to them? Have I been somewhere like this before, and how did it go?
• Worst case, best case, most likely. If the worst happened, how would I cope? What’s the best that could happen? What will probably happen? The decatastrophizing exercise works through this family in detail.
• The effect of believing it. What happens when I keep telling myself this? What might change if I saw it differently?
• What’s in your control. Which part of this can I do something about? What do I have to accept I can’t control?

Socratic questioning examples: anxiety, self-criticism, a relationship worry

The three people below are made up to show how the questions work, and their numbers aren’t data.

Anxiety: Maya’s manager emails late in the day asking her to stop by at nine tomorrow, with no reason given. Maya’s thought is “I’m getting fired,” and she believes it 80 out of 100.

• What makes me think it’s true? The team lost a big client last month, and the email didn’t say why she wants to meet.
• What makes me think it isn’t? My review in the spring was good, and last week she gave me a new account.
• What else could explain it? She might want to talk about the new account or a deadline.
• What would I tell a friend? “You don’t know what it’s about yet. Get some sleep and find out at nine.”

Asked how “I’m getting fired” looks now, she writes: “Something’s up, and I don’t know what. Being let go is possible, but it isn’t the likeliest reading.” Her belief drops to 30. She still sleeps badly, and she walks in the next morning less braced for bad news.

Self-criticism: Jordan sends a report to the whole department, and a coworker replies that a total in the first table is wrong. His thought: “I’m an idiot, and I always mess things up.” He believes it 90 out of 100.

• What actually happened? One number was wrong. The rest of the report was fine, and I sent a correction within the hour.
• Is “always” true? Last quarter’s report went out clean, and so did the budget sheet in May.
• What would I tell my sister if she did this? “You made a typo and fixed it the same day. Nobody thinks you’re an idiot.”
• What does believing it do? I’ll replay it all evening and stay quiet at Monday’s meeting.

His new version: “I made a mistake and handled it, and next time I’ll check the totals before I hit send.” His belief drops to 25. “Idiot” and “always” are two of the cognitive distortions CBT has names for: labeling and overgeneralization.

A relationship worry: Sam’s partner, Alex, was quiet all evening and went to bed early. Sam’s thought is “He’s losing interest in me,” and she believes it 75 out of 100. She wants to ask him again whether everything’s okay, though she already asked twice yesterday.

• What do I actually know? He was quiet tonight. On Sunday he told me his mom’s test results come back this week.
• Am I reading his mind? Yes. He hasn’t said anything about losing interest, so that part is my guess.
• What else could explain it? He’s worried about his mom, or he’s worn out after a long week.
• What’s in my control? Asking about his mom tomorrow, and not asking a third time whether we’re okay.

Her new version: “He’s had a lot on his mind this week, and I don’t know that it’s about me.” Her belief drops to 40, and the urge to ask again eases without going away. The guide to mind reading covers this trap in more depth.

Question one thought on paper

Pick one thought that bothered you recently and write your answers down, on paper or in a notes app, so you can see them side by side.

1. Pick one moment and one thought. Choose a specific time the feeling hit, like Tuesday’s meeting rather than “work in general.” Write the thought as one sentence and rate how much you believe it from 0 to 100.
2. Write down what happened. Note where you were and who said or did what. Padesky’s advice is to get concrete before asking what it all means.
3. Choose two or three questions. A prediction calls for evidence and the most likely outcome. A harsh label calls for the friend question, and a guess about what someone else thinks calls for other explanations.
4. Answer with what you know. Stick to what you know rather than what you fear. If you can’t answer a question, write “I don’t know” and move on.
5. Circle the answer that surprises you. Padesky said that if she wasn’t regularly surprised by her clients’ answers, she suspected she was asking dull questions or not listening. An answer that doesn’t fit the thought is the one to notice.
6. Sum it up. Read your answers back and write a two- or three-line summary. Padesky noted that seeing the answers together sometimes hits harder than any one of them.
7. Ask the tying-together question. Given everything you wrote, how does the original thought look now? Padesky saw therapists, herself included, skip this question after dozens of good ones. Write a new version in your own words and rate your belief in the old thought again.
8. Pick one thing to do, then stop. If an answer pointed to something in your control, put it on your calendar. If the thought is a prediction you can test, try a behavioral experiment. If you catch yourself starting over to feel sure, close the notebook, since that’s a sign the questions have turned into reassurance.

If the same thought comes back next week, reread what you wrote instead of starting from scratch. The site’s free thought record worksheet has a field for the balanced thought you land on.

Asking yourself without arguing or chasing reassurance

Padesky’s advice to therapists works when you’re asking yourself, too: if you aren’t curious about the answer, don’t ask the question. Ask “What else could explain this?” and wait for something to come, instead of firing back a rebuttal you had ready.

Arguing with a thought sounds different. You tell it “That’s irrational,” it answers “But what if,” and the exchange goes in circles. Sally Winston and Martin Seif, two anxiety clinicians writing for the Anxiety and Depression Association of America, describe this looping internal debate, where what-ifs and rational responses take turns and never stop. In their account, arguing with a doubt or reassuring yourself about it strengthens the doubt, and the brief relief rewards the worry that came before.

Reassurance seeking can also hide inside the questions. In a 2013 questionnaire study, Osamu Kobori and Paul Salkovskis found that people with obsessive-compulsive disorder (OCD) or panic disorder sought reassurance more often than people without either condition, and that people with OCD leaned more on self-reassurance. Signs your questioning has crossed over:

• You’re asking a question you already answered yesterday.
• You need the answer to feel exactly right before you can stop.
• The relief lasts a few minutes, and then the doubt is back.
• You’re asking other people, or searching online, for the same answer.

When that happens, ask fewer questions. Answer the thought once in writing. When the doubt comes back, notice it and return to what you were doing. Winston and Seif point out that absolute certainty can’t be had and isn’t needed to make a decision.

When questioning isn’t the right tool

These questions need a reasonably calm mind and a thought that’s actually open to question. In four situations, reach for something else.

In the middle of a panic attack, a list of questions is hard to hold onto when your heart is pounding and you can’t catch your breath. Settle your body first with slow breathing or 5-4-3-2-1 grounding, and save the questions for afterward, when you can look back at what the panic told you. The panic attack guide covers the attack itself.

With intrusive thoughts, arguing can feed the loop. These are unwanted thoughts or images that pop in, like a flash of harming someone you love. OCD researchers call acts such as checking, praying, or thinking an opposite thought neutralizing, and arguing a thought down can work the same way. In a 2003 experiment led by Paul Salkovskis, 29 people with OCD heard recordings of their own intrusive thoughts and either neutralized them or distracted themselves. Discomfort fell faster at first for those who neutralized. Later they felt more discomfort and stronger urges to neutralize than the people who had distracted themselves. The intrusive thoughts guide and cognitive defusion skills fit these thoughts better than a debate.

With grief, the pain fits the loss. A thought like “I should have visited more” may be plain regret, and picking it apart in the first raw weeks can feel like arguing with your own sadness. If the guilt hardens months later into a verdict about yourself, the questions can help then, ideally with a therapist who works with grief.

When the danger is real, the useful question is what to do. If a symptom won’t go away or your company has announced layoffs, the evidence may well support the worry, and no question should talk you out of calling a doctor or someone you trust. If you’re in danger right now, get help before anything else.

What the research does and doesn’t show

Experts call Socratic questioning a key part of cognitive therapy, yet for a long time it was barely measured. A 2015 review by Gavin Clark and Sarah Egan found no direct evidence that it helps in CBT. The same year, Justin Braun and colleagues at Ohio State knew of only one earlier study of how therapists use it.

Braun’s team followed 55 adults through 16 weeks of cognitive therapy for depression, with observers rating the therapists’ questioning in the first three sessions. After sessions with more questioning than was typical for that patient, depression scores fell more by the next session, even after accounting for the therapy relationship. A 2022 study from the same lab, with 123 clients, found results consistent with the questions working through a change in how clients think, more so for clients who started with weaker CBT skills. Viewers like the style too: in a 2018 study, 144 adults rated the therapist’s questions in a mock Socratic session as more helpful than a session where the therapist explained things directly, though they were only watching.

These findings have real limits. They’re correlations from therapy for depression, so they can’t prove the questions caused the improvement, and Braun’s team says so. A 2023 review by James Overholser and Eleanor Beale called the research scant and found none on how people are doing at the end of therapy. We found no study that measured Socratic questioning this way in anxiety, and none that settled whether guided discovery beats questioning aimed at changing minds, which Padesky flagged as untested in 1993.

Self-guided questioning has barely been tested. In a 2023 experiment from the Ohio State group, 230 people filled out one worksheet on either cognitive restructuring or positive thinking, and negative mood dropped after both with no significant difference. A 2025 feasibility study of Socrates 2.0, an AI tool that walks users through Socratic questions, found small to moderate symptom drops over four weeks in 61 adults. It had no comparison group, and some users found the questions repetitive. That tool isn’t Helpy, and neither study shows what asking yourself these questions adds. Treat them as a skill worth trying, and judge them by what happens to your own thoughts.

Question a thought with Helpy

The chat below runs with Helpy, the site’s AI assistant. Tell it the thought that’s bothering you, and it can ask you a few of these questions and help you sum up what your answers say.

Evidence and sources

• Padesky (1993), “Socratic Questioning: Changing Minds or Guiding Discovery?”, keynote, European Congress of Behavioural and Cognitive Therapies: Contrasts questions that steer a client to the therapist’s answer with guided discovery, and describes four stages ending in a synthesizing question. A clinician’s talk, not a study.
• Beck Institute, CBT Worksheet Packet (Judith Beck, 2020): Its thought-record pages list the question families on this page, from evidence to what would be good to do. A worksheet set, not a trial.
• Beck Institute Cares, “A CBT Approach to Living with Uncertainty”: Advises focusing on what you can control and accepting that complete control isn’t possible. A blog post from the COVID-19 pandemic.
• Nails and Monoson, “Socrates,” Stanford Encyclopedia of Philosophy (revised 2022): Describes Socrates’s question-and-answer way of probing serious matters. Background on the name only.
• Clark and Egan (2015), Cognitive Therapy and Research: A narrative review: the Socratic method is called an important part of CBT, but no direct evidence showed that it helps. A review, with no new data.
• Braun, Strunk, Sasso and Cooper (2015), Behaviour Research and Therapy: In 55 adults in cognitive therapy for depression, more questioning than usual for a given patient predicted a bigger symptom drop by the next session (1.51 BDI-II points per standard deviation), independent of the alliance. Correlational, three early sessions, depression only.
• Vittorio, Murphy, Braun and Strunk (2022), Behaviour Research and Therapy: In 123 clients in CBT for depression, results fit cognitive change carrying the effect of Socratic questioning on early symptom change, more so for clients with lower starting CBT skills. Therapist-led and depression only; we read the abstract.
• Heiniger, Clark and Egan (2018), Journal of Behavior Therapy and Experimental Psychiatry: 144 lay observers rated a Socratic therapy video as more helpful and more supportive of autonomy than a didactic one. Observers, not clients in treatment.
• Overholser and Beale (2023), Psychotherapy Research: A review: the scant evidence points to session-to-session drops in depression, with no research on end-of-therapy outcomes. Blends research with clinical experience.
• Murphy, Vittorio and Strunk (2023), Psychotherapy Research: 230 people were randomized to one worksheet on cognitive restructuring or on positive thinking; negative mood fell after both, with no significant difference. Immediate mood only.
• Held and colleagues (2025), JMIR Mental Health: 61 adults used Socrates 2.0, an AI tool for Socratic dialogue, for four weeks; symptom drops were small to moderate, and some users found it repetitive. No comparison group, and a different tool from Helpy.
• Salkovskis and colleagues (2003), Journal of Abnormal Psychology: 29 people with OCD neutralized or distracted themselves while hearing recordings of their intrusive thoughts; neutralizing eased discomfort at first, then brought more discomfort and stronger urges. A small lab study, not a test of Socratic questioning.
• Pascual-Vera and colleagues (2025), European Journal of Investigation in Health, Psychology and Education: Its questionnaire gives checking and thinking an opposite thought as examples of neutralizing. Cited only for those examples.
• Kobori and Salkovskis (2013), Behavioural and Cognitive Psychotherapy: People with OCD (153) or panic disorder (50) sought reassurance more often than 52 healthy controls, and the OCD group used more self-reassurance. A questionnaire study.
• Winston and Seif, ADAA, “When Reassurance Seeking Becomes Compulsive”: On looping internal debates, and on how brief relief from reassurance strengthens worry. Expert commentary, not data.

Read next

How to Fill Out a Thought RecordWhere the questions get usedDecatastrophizing: give the worst case a size, then a planQuestions for the worst caseWhich Cognitive Distortions Do You Have?Name the trap you're questioningCognitive restructuring: how to rewrite a thought so it fits the factsThe core CBT method, step by step

← All CBT guides