Turning a belief into something you can actually test.
Sometimes a client understands the cognitive argument and nothing changes.
Behavioral experiments create another route: instead of settling the belief
through discussion alone, therapist and client create an experience that can
generate new information.
You’ll practice
Identifying a belief worth testing
Turning a broad fear into a specific prediction
Defining observable outcomes
Identifying safety behaviors that could muddy the result
Processing mixed outcomes without declaring success or failure
Updating a belief proportionately to the evidence
Case orientation
Meet Andre
Andre is 36 and has been invited several times to facilitate a workshop
for colleagues in his professional community.
He has the expertise. He keeps declining.
You are meeting with him after he has received another invitation.
Qualified on paper. Still waiting to feel ready.
Scene 1 · The client wants proof before acting
Andre
“I know this is going to sound ridiculous because technically I’m qualified.”
“But if I actually lead something and I’m awkward or I lose my train of thought,
people are going to realize I’m not as competent as they assumed.”
“I’d rather wait until I feel more confident.”
What is the experiment supposed to test?
Clinical Support
Start with the prediction, not the activity.
A behavioral experiment is not defined by what the client does.
It is defined by what the action is meant to test.
Belief
If I lead while feeling nervous, people will realize I’m not competent.
Prediction
If I hesitate or lose my place, the group will disengage or trust me less.
Observable indicators
People stop participating, leave, or give feedback suggesting lack of competence.
“What are you predicting will happen?”
“How would we know if that prediction came true?”
“If I were watching, what would I actually see?”
“What would count as evidence either way?”
Scene 2 · Too broad to test
Andre
“I think the prediction is basically that it’ll go badly.”
How would you make the prediction more precise?
Experiment update
Belief
If I show uncertainty while facilitating, people will see me as unqualified.
Belief strength
85%
Prediction
If I lose my place or appear nervous, participants will disengage or trust me less.
Scene 3 · What makes this an experiment?
Andre
“Okay. So I just have to do the workshop.”
What would you say?
Clinical Support
Experiment or exposure?
Behavioral experiment
What prediction are we testing?
What data will answer the question?
Exposure
What are we approaching?
What happens when avoidance is reduced?
The same activity can function differently depending on the learning target.
Scene 4 · Safety behaviors enter the room
Andre
“I can prepare every sentence in advance.”
“And I’ll probably keep my notes open the whole time.”
“If there’s silence, I’ll move on immediately so nobody thinks I’ve lost control.”
What do you notice?
Clinical Support
Ask what the safety behavior prevents you from learning.
Fear
“If I hesitate, they’ll lose confidence in me.”
→
Safety behavior
Script every sentence.
→
Ambiguous learning
“It only worked because I scripted everything.”
“What are you hoping that strategy will prevent?”
“If things go well while you use it, what will you conclude?”
“Would that give us a clean test?”
“Which safety behavior could we reduce enough to learn something new?”
Use an outline, not a full script.
Allow natural pauses.
Do not automatically apologize for hesitation.
Track actual participant behavior.
Ask for brief feedback afterward.
Scene 5 · Predict before you collect data
Andre
“I still think at least four of the six people will notice I’m not ready.”
What should be established before the experiment?
Defined before the experiment
Prediction
At least 4 of 6 participants will visibly disengage, communicate loss of confidence,
or give feedback suggesting Andre is not ready to facilitate.
Belief: 85%
Experiment · What actually happens?
The workshop
Andre pauses twiceto find his place.
One participantchecks their phone briefly.
One participantasks for clarification.
Three participantsask follow-up questions.
Feedback is mixedpositive comments plus one constructive suggestion.
Andre
“It still felt really uncomfortable.”
Observation board
What literally happened?
2 noticeable pauses
1 brief phone check
1 clarification request
3 follow-up questions
1 positive comment about pacing
1 constructive suggestion
0 comments questioning competence
0 people left
Scene 6 · The feared outcome partly happened
Andre
“See? Someone checked their phone. And someone needed clarification.”
“So I wasn’t imagining it.”
How do you process the result?
Clinical Support
Do not make “nothing went wrong” the learning criterion.
Reality often produces mixed data. The therapist’s job is not to declare victory.
“What did you predict?”
“What actually happened?”
“Which parts matched?”
“Which parts did not?”
“What does the result allow us to say?”
“What does it not allow us to say?”
Prediction vs outcome
Predicted
At least 4 of 6 participants would disengage or respond as if Andre were not competent.
Observed
1 phone check, 1 clarification request, several follow-up questions,
mixed-positive feedback, and no explicit competence concern.
Andre
“Okay. That is different from what I predicted.”
“It wasn’t perfect. But it also wasn’t the catastrophe I expected.”
Belief updating
Andre
“I still think I could come across badly sometimes.”
“But maybe being visibly imperfect isn’t the same as being unqualified.”
Which revised belief best fits the data?
Re-rating
The prediction changed.
Original belief
“If people see me hesitate, they’ll realize I’m not qualified.”
Belief strength
85% → 55%
New working belief
“People can notice imperfection without making the global judgment I expect.”
The experiment changed confidence in the prediction. It did not prove a permanent truth.
Integration
Andre
“I think I could do another one.”
What would you do next?
Pattern synthesis
Your choices shaped the experiment.
This is not a score. It describes what your choices tended to privilege
in this particular simulation.
Completed experiment
What the experiment established
Belief
“If I hesitate while facilitating, people will realize I’m not competent.”
Prediction
At least 4 of 6 participants will disengage or indicate loss of confidence.
Belief before
85%
Belief after
55%
Experiment
20-minute workshop · 6 peers · normal preparation · outline rather than script ·
natural pauses allowed · no automatic apologizing
The feared outcome occurred far less strongly than predicted.
Revised belief
“People can notice hesitation without concluding that I’m incompetent.”
Debrief
Reflect on your clinical choices.
When did the experiment become specific enough to produce useful information?
Which safety behavior would have made the result hardest to interpret?
What would have happened clinically if Andre had performed flawlessly?
What did the partially negative data contribute?
How did you avoid turning the experiment into something Andre had to “pass”?
What would you test next?
If a behavioral experiment only counts when the client’s feared prediction
is disproved, is it still an experiment?
Continue the learning
See behavioral change unfold inside real clinical conversations.
Behavioral experiments do not have to be dramatic.
They may involve one conversation, one boundary, one attempt at visibility,
or one small action that provides information the client could not obtain
through thinking alone.
For longer-form examples of CBT interventions conducted in live conversations,
continue with Transcripts, Notes and Reflections from The CBT Dive Video Podcast.
