Running the Experiment and Observing Results

Do the planned activity, observe what actually happens, and record results immediately.

Why it works

The emotional persuasiveness of a behavioral experiment comes from direct experience: not being told by a therapist that the belief is probably wrong, but experiencing the disconfirmation personally. This works through the same mechanism as exposure therapy — prediction violations at the level of experience are more belief-changing than at the level of reasoning. Recording results immediately matters because autobiographical memory is reconstructive and emotionally biased: memories formed hours after an anxious event tend to re-encode the event as more confirming of the anxiety than it actually was.

How to do it

1. Do the planned activity as exactly as designed — do not run a safer version that doesn’t actually test the belief.
2. During the experiment, notice the anxiety level (expected) and observe what is actually happening.
3. Immediately after, write what happened — facts, not interpretations.
4. Compare the actual results to the pre-stated criteria for confirmation and disconfirmation.
5. Rate belief strength in the original prediction after the experiment (0–100%).

Evidence

Behavioral experiments are supported by exposure theory research: violations of threat predictions, when experienced directly rather than discussed, produce stronger and more durable belief change. A study by Salkovskis et al. found behavioral experiments produced greater belief change than verbal challenging alone for panic disorder. (rct)

Evidence is primarily from anxiety disorders; generalization across all CBT applications of behavioral experiments is clinical rather than uniformly RCT-tested.

Sources

• Salkovskis et al. (2007), "Belief disconfirmation versus habituation approaches in exposure therapy", Behaviour Research and Therapy
• Bennett-Levy et al. (2004), "Oxford Guide to Behavioural Experiments in Cognitive Therapy"

Common mistake

Running a diluted version of the experiment that doesn’t truly test the belief — wearing a safety behavior (e.g., speaking quickly in a presentation so people don’t notice nervousness) prevents the prediction from being genuinely tested.