The irreducible epistemic atoms underlying the curriculum. 4,828 atoms across 8 types and 2 molecules
Expect social extinction bursts when you stop behaviors that others have been reinforcing — they will escalate attempts to elicit your old behavior because they are experiencing loss of a reinforcer they relied on.
Track extinction progress using dimensional measurement (frequency, peak intensity, duration) rather than binary measurement to reveal the non-linear decline curve that binary tracking erases.
Explain established systems to newcomers not as onboarding but as defamiliarization practice—articulating what you automate exposes hidden assumptions.
When spontaneous recovery occurs weeks into extinction, compare current intensity to initial burst intensity rather than to immediately preceding quiet days to accurately assess whether progress has been lost.
Interpret a single lapse as a predicted phase of extinction that carries no prognostic significance, and focus solely on preventing the second consecutive occurrence within 24 hours.
Use relapse episodes diagnostically to identify which contexts, reward exposures, or time intervals reactivate the original learning, then strengthen extinction specifically in those conditions.
Pre-commit to the accurate attribution for relapse before it occurs, because once the abstinence violation effect activates, you lose the capacity to generate the reframe in real time.
Re-engage replacement behaviors immediately after setbacks rather than waiting for emotional readiness, because speed of re-engagement predicts long-term success independent of setback severity.
Build lasting behaviors through sequential time-boxed experiments with interim evaluations rather than single permanent commitments, grounding each continuation decision in accumulated evidence.
Create bright-line rules for behaviors maintained by variable-ratio reinforcement, because any engagement re-triggers the complete reinforcement loop.
Calibrate commitment stakes to hurt meaningfully without devastating, because stakes that are too low lack force while stakes that are too high trigger shame spiraling.
Establish accountability relationships with neutral witnessing rather than moral evaluation, because shame transforms accountability into surveillance and reduces honest reporting.
Match redirect behaviors to the specific function the unwanted behavior served, because functional vacuums create pressure for behavioral return.
Ask 'If we were starting from scratch today, would we build it this way?' to simulate the perspective of someone without your accumulated assumptions.
Treat each urge as a wave with a predictable temporal structure (onset, escalation, peak at 8-15 minutes, decline), using observational attention rather than resistance to allow the wave to complete its natural 15-30 minute cycle.
Establish post-extinction monitoring with four defined components—specific observable signal, scheduled observation frequency, explicit threshold for intervention, and predetermined response protocol—to detect behavioral drift before rationalization normalizes deviations.
Taper monitoring frequency as extinction stabilizes (daily → weekly → monthly → quarterly) but increase frequency preemptively during life transitions, because context changes weaken extinction learning while simultaneously reducing monitoring likelihood.
Conduct functional analysis before attempting extinction by observing the behavior without intervention for at least five days using ABC recording, then generate and test a functional hypothesis, because intervention without understanding function produces symptom substitution rather than genuine elimination.
Design replacement behaviors using differential reinforcement strategies (DRA, DRI, DRO) that serve the same function as the unwanted behavior through a more adaptive channel, scoring candidates on functional match, temporal match, and sustainability before implementation.
Establish explicit, falsifiable hypotheses with measurable outcomes and proposed mechanisms before attempting behavior change to enable genuine learning through prediction error.
Measure baseline behavior before implementing interventions to distinguish genuine effects from normal variation and regression to the mean.
Conduct structured evaluations at time-box endpoints using pre-specified criteria and tracked data rather than subjective impressions to make evidence-based continuation decisions.
Isolate variables in behavioral experiments by changing only one factor at a time while holding all other conditions deliberately constant.
Identify and test the behavioral kernel—the irreducible core action without which the behavior does not exist—stripped of duration, intensity, and context complexity.