deciqAI Neurological Levels
Overview
Robert Dilts' Neurological Levels maps how change propagates through a 6-level hierarchy (grounded in Bateson 1972): change at a HIGHER level cascades down automatically; change at a LOWER level reverts under pressure.
L1 Environment (where/when) → L2 Behavior (what you do) → L3 Capability (skills/strategy) → L4 Beliefs/Values (what you believe matters) → L5 Identity (who you are) → L6 Purpose (why beyond yourself).
The diagnostic question: "At which level does this problem actually live?" Most interventions fail because they target L2 for problems that live at L4–L5.
- Use BEFORE [okr-goal-setting] — verify identity/belief alignment before setting L2–3 OKRs.
- Use WITH [intrinsic-drive] — intrinsic drive lives at L4–6; this skill locates where to rebuild it.
When to Use
- A behavior change has been attempted and reverted more than once
- Someone knows what they should do but consistently does not do it
- Strategy understood but execution fails repeatedly in the same ways
- "Playing not to lose" rather than "playing to win" — a L5 identity signal
- Someone adopting a new professional identity (IC to leader, employee to founder)
When NOT to use: Problem is genuinely L1 (environment is the constraint); skills are the bottleneck and person agrees to L3 training; acute crisis (stabilize first); applying it as external label without engagement.
Coaching Novices (Adaptive Front Door)
- Engine mode: concrete case → run The Process directly.
- Coach mode: unfamiliar or no case → guide step by step.
In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.
- Frame: "Persistent failures usually mean intervening at the wrong level. Higher-level changes cascade down automatically."
- Check fit: "What have you tried? At what level? What happened under pressure?"
- Elicit case: "Describe a change that didn't stick. Was the reversion L1 circumstances / L2 actions / L3 know-how / L4 disbelief / L5 identity?"
[WAIT — do not advance until user responds]
- Process: "Name the level where resistance lives. What intervention at THAT level looks like — not fixing the symptom."
[WAIT — do not advance until user responds]
- Close: "What insight did you just uncover about where to intervene?"
[WAIT — do not advance until user responds]
The Process
Gate: Multiple L2 interventions attempted and reverted. First attempt → try L2–3 first.
- State precisely: observable symptom, changes tried, number of reversions.
- Level diagnostic: L1 — environment the constraint? L2 — knows what to do but still doesn't? L3 — lacks skill/strategy? L4 — believes change is right/possible? L5 — change conflicts with self-concept? L6 — aligns with larger purpose?
- Identify intervention level: lowest level with persistent resistance despite lower-level work.
- Design at that level: L4 → surface belief, find contradicting evidence, reframe values. L5 → identity narrative, role models, graduated exposure.
- Check cascade: lower-level symptoms resolve without direct attention → diagnosis correct.
Stop-rule: 3+ L2 interventions failed → diagnose at L4–5 before any further L2 work.
Output: Neurological Level Diagnostic
Problem + change history | Interventions by level | # reversions
L1–L6: [constraint evidence at each level]
Diagnosis: Level ___ [evidence] | Intervention: [at diagnosed level]
Expected cascade: [lower-level changes if correct] | Success indicator: [...]
→ Method in Action: Lincoln's Emancipation Strategy (1858–1865)
Packs · Applying It Well
Packs: Org culture — L5 intervention: "Who are we as a company?" · Executive IC-to-leader — L5: "What does being a leader mean about who you are?" · Product teams — reframe "feature factory" L5: "We create outcomes." · Contribution: domain cases with quantified cascade outcomes especially valuable.
Applying It Well: Diagnose before intervening · Cascade test = proof (lower-level symptoms improve without attention) · Language reveals level: "I should" = L3–4 · "I can't" = L3 · "Not who I am" = L5 · "What's the point?" = L6 · Never skip levels · L5 requires narrative not instruction (stories, role models, lived experience) · Higher levels need more time and safety.
→ Primary sources: references/sources.md
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake Move | Why It's a Trap |
|---|---|
| [D] "Just train people harder on the behavior." | Training = L3. Behavior without L4 belief or L5 identity alignment reverts. |
| [D] "Set clear goals and hold people accountable." | Accountability = L2–3. L5 identity conflicts are experienced as threats, not incentives. |
| [D] "Change the structure and behavior will follow." | Sometimes L1 is the constraint — but most persistent failures in resourced orgs are not L1 problems. |
| [D] "This person just doesn't want to change." | "Not wanting" = L4–5 signal, not a character verdict. Diagnosable and workable. |
| [D] "We have a stated values document." | Documenting values is L4 at best. Embedding requires L2 consequences + L3 capability + L5 storytelling. |
| [D] "Identity work is therapy, not business." | L5 work is the highest ROI intervention. Every "inexplicable" execution failure is usually a L5 problem. |
| [D] "We tried culture change and it didn't work." | A L4 belief preventing L5 intervention. Previous attempt operated at the wrong level. |
| → Add [O] entries after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags · Verification
Red Flags: Same behavior reverted 2+ times after L2 intervention · "Why doesn't it stick?" answered with motivation/culture ("how people are") · "That's not who I am" (explicit L5) · Org calls competitors' capabilities "not our culture" (L5 blocking L3) · Change requires constant management attention · "I know I should" (L3/4 without L5 integration).
Verification checklist:
- Level named with behavioral/verbal evidence — not a guess.
- Lower-level interventions enumerated and their failure documented.
- Intervention design targets the diagnosed level — not a re-application of L2.
- Cascade test defined: what lower-level symptoms should resolve?
- Diagnostic validated through conversation — not applied as external label.
- Timeline set: L5 identity work is measured in months, not days.
- Diagnostic artifact reviewed and agreed upon by person or change team.
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/neurological-levels · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/neurological-levels.json