Convert 'I am X' identity claims into 'In situation Y, I will do Z' — make identity testable
Rewrite personal identity schemas as behavioral predictions with specified conditions and thresholds to convert unfalsifiable identity claims into testable hypotheses.
Why This Is a Rule
"I am a good listener" is an identity claim — unfalsifiable because you can always redefine "good listener" after any failure. "In one-on-ones, I will ask at least two follow-up questions before shifting to my perspective" is a behavioral prediction — testable because the conditions (one-on-ones), the behavior (follow-up questions), and the threshold (at least two) are specified.
Identity schemas resist disconfirmation because they're self-referential: the evidence you accept is filtered by the identity claim itself. "I'm a strategic thinker" makes you notice your strategic moments and dismiss your tactical ones. The identity is unfalsifiable not because it's true but because it controls what counts as evidence.
Converting to behavioral predictions breaks this self-referential protection by specifying external, observable criteria that the identity must meet. The prediction can fail — and when it does, you have specific, actionable information about where the identity claim diverges from reality.
When This Fires
- When self-describing with identity claims: "I am [trait]"
- During self-reflection when evaluating personal strengths or weaknesses
- When identity-level beliefs are driving decisions without behavioral evidence
- After receiving feedback that contradicts a held identity claim
Common Failure Mode
Writing behavioral predictions that are too easy to pass: "I will sometimes listen to others." This is unfalsifiable in the opposite direction — the bar is so low it can never fail. Predictions need specific conditions, specific behaviors, and specific thresholds that create genuine pass/fail criteria.
The Protocol
For each identity schema: (1) State the identity claim: "I am [trait]." (2) Rewrite as behavioral prediction: "In [specific situation], I will [specific observable behavior] at least [threshold frequency/quality]." (3) Test: does the prediction have genuine fail conditions? Could you observe yourself not meeting the threshold? (4) Track: over the next 2 weeks, does your actual behavior match the prediction? Where it matches → the identity claim is behaviorally supported. Where it doesn't → the identity claim exceeds what your behavior actually delivers.