A planned real-world test of a belief. The client states a prediction, designs a safe experiment, performs it, and compares prediction with result. It is often more powerful than discussion because the client collects new experience rather than only new arguments.
Step-by-step guide
- Identify the belief to test.
- State a concrete prediction: what exactly will happen?
- Rate belief in the prediction from 0 to 100%.
- Design a safe and ethical experiment.
- Run the experiment and record what happened.
- Compare prediction and outcome.
- Update the belief and decide the next experiment.
When to use
- Beliefs maintained by avoidance
- Social anxiety and fear of negative evaluation
- Agoraphobia, panic and safety behaviors
- OCD predictions that can be tested without rituals
- When verbal restructuring does not shift belief
Key phrases
Rather than only debating this thought, let's test it. What would be a small experiment that could show whether the prediction is accurate?
Follow-up questions
What exactly do you predict will happen?
How will we know whether the prediction came true?
What did you learn from the result?
Alternative phrasings
Let's treat this as a hypothesis, not as a fact.
What would count as evidence either way?
Warnings
- ⚠️ Experiments must be safe, ethical and collaborative.
- ⚠️ Do not set the client up for humiliation or overwhelm.
- ⚠️ Remove or track safety behaviors; otherwise the experiment may be unclear.
- ⚠️ Plan debriefing before assigning the task.
Source: Bennett-Levy et al. 2004; Beck et al. 1979
Materials are informational and educational and summarize publicly available scientific sources. They are not medical or psychological advice, are not intended for self-diagnosis or self-treatment, and do not replace consultation with a qualified professional.