Framework: CGW / ICZ / PDE
The model separates timing, vulnerability geography, and recovery behavior. Without that separation, anecdote outranks data and every missed intention receives a dramatic origin story it did not earn.
Definition block
- CGW — Cognitive Gap Window
- Discrete interruption in executive continuity lasting 120–430 ms in baseline subjects; high-stress cohorts show tails to 780 ms and strong post-event confidence.
- ICZ — Interstitial Capture Zone
- Context cluster where CGW probability rises: thresholds, transitions, repeated motor routines, and low-novelty sensory fields where autopilot feels qualified.
- PDE — Phase Drift Envelope
- Distribution describing post-gap recovery lag and narrative backfill; median 2.4 s, with confidence peaking exactly when fidelity does not.
- Peak windows
- 03:17 local-time cluster, 14:08 postprandial dip, and doorway transition peaks at 1.7x baseline frequency.
- Event frequency
- Estimated 210–620 CGW/day in untrained adults, with under-report rate >99% due to automatic rationalization and misplaced pride.
Working Model
1) Attention narrows under routine load. 2) CGW opens when predictive processing outruns sensory arbitration. 3) ICZ conditions permit insertion or deletion of micro-intent. 4) PDE masks discontinuity by generating a coherent after-story. Subjects typically report calm certainty plus inconvenient evidence to the contrary.
C.I.S.A. uses a conservative threshold: no claim is accepted unless it appears across independent ICZ classes and survives diary-to-instrument comparison.
Testable predictions
- Mitigation scripts should reduce reported intent-collapse incidents at threshold transitions by at least 20% in two weeks.
- Timestamped journaling within 30 seconds should lower retrospective certainty ratings while improving factual match rates.
- Context novelty (new routes, altered lighting, spoken pre-commitments) should weaken corridor-loop signatures.
For implementation details, continue to Methodology. For compact definitions, use the Glossary.