deciqAI Three-Radius Model — 认知/能力/行动三半径
Overview
Three distinct circles govern performance: what you understand (cognition radius), what you can do (capability radius), and what you actually do (action radius). These are almost never the same size — the gaps between them are where most performance is lost.
The Cognition-Capability gap is a learning and resource problem. The Capability-Action gap is an execution problem driven by fear, inertia, or misaligned incentives. Treating them as the same problem is the default failure mode.
Skill composition: Use AFTER [metacognition] to accurately place each circle. Use BEFORE [okr-goal-setting]. Complements [circle-of-competence]: circle-of-competence sets DOMAIN boundary; three-radius-model maps DEPTH layers within that domain.
When to Use
- Someone says "we know what to do but aren't doing it" — Capability-Action gap
- Someone says "we know the problem but don't know how to fix it" — Cognition-Capability gap
- Effort is not translating to results despite apparent competence
- Pre-planning skill or team development goals; post-mortem gap analysis
When NOT to use: gap is already identified and solution is purely tactical; organizational constraints (not individual gaps) are clearly the sole cause; domain too new for a cognition radius to be established.
Coaching Novices (Adaptive Front Door)
- Engine mode: user has a concrete case → run The Process directly.
- Coach mode: user is unfamiliar or describes being "stuck" without knowing why → 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.
- One-line what-it-is: "What you understand, what you can do, and what you actually do are three different circles — and the gaps between them are where most performance is lost."
- Check fit: "Are you making less progress than you expect, and unsure exactly where the bottleneck is?"
- Elicit their real case: "Tell me about a specific domain where progress feels slower than it should. What do you understand? What can you do? What are you actually doing?"
[WAIT — do not advance until user responds]
- Run The Process one step at a time. Start with cognition only. Do not move to capability or action until cognition is mapped.
[WAIT — do not advance until user responds]
- Close by naming the insight uncovered. Is the bottleneck learning, skill-building, or just starting? Name it explicitly.
[WAIT — do not advance until user responds]
The Process
Output artifact: Three-Radius Diagnostic. Gate rule: Each circle requires specific, observable evidence — not self-report.
Step 1 — Map cognition radius. List 3-5 things you understand well enough to evaluate others' work, explain mechanisms, predict outputs, and name your understanding boundary. Validation test: could you teach each item to a peer and receive useful critique? If not, it is in your awareness radius, not cognition radius. Gate 1: 3-5 items passing the validation test.
Step 2 — Map capability radius. List 3-5 things you can DO with demonstrated, externally evaluated performance. Understanding how something is done ≠ being able to do it. Gate 2: 3-5 items, each with a demonstrated output and external evaluator.
Step 3 — Map action radius. List only what you specifically did in the last 30 days — behaviors, frequency, outputs, decisions. No intentions or "typical" behavior. Gate 3: Specific, dated behaviors only.
Step 4 — Identify gaps. Cognition-Capability gap: understand but can't execute — intervention: deliberate practice, resource acquisition. Capability-Action gap: can do but not doing — intervention: commitment devices, environmental design, incentive restructuring. Adding cognition to a Capability-Action problem is the most common misdiagnosis. Gate 4: Both gaps with ≥2 items each, root causes named.
Step 5 — Design 30-day closure. For each gap, name the ONE highest-impact item, the specific intervention, the minimum viable demonstration within 30 days, and the named validation mechanism. Gate 5: One 30-day action per priority gap with named validation.
Stop-rule: If all three circles appear identical, you have mapped aspiration. Return to Step 3; list only behaviors from the last 30 days.
Three-Radius Diagnostic — [Domain] — [Date]
Cognition Radius: 1.[Item — validated by: X]
Capability Radius: 1.[Item — demonstrated by: output — evaluated by: who]
Action Radius (last 30 days): 1.[Behavior — frequency: X/week]
Cognition-Capability Gaps: Item | Root Cause | 30-Day Action | Validation
Capability-Action Gaps: Item | Root Cause | 30-Day Action | Observable Signal
Priority: Cognition-Capability → [item + action] | Capability-Action → [item + action]
→ Method in Action: Nikola Tesla's Three-Radius Divergence, 1880–1943
Three-Radius Packs
Pack 1 — Founders/Operators: Cognition = validated by 50+ customer conversations. Capability = teachable and replicable. Action = what specifically happened in the last 30 days to expand market position.
Pack 2 — Knowledge Workers: Cognition radius often large but boundary unvalidated. Capability rarely externally evaluated. Action is the bottleneck — classic Capability-Action gap: commitment device + accountability partner.
Pack 3 — Organizations: Strategic intelligence ≠ capability. Documented processes ≠ demonstrated capability. Measure actual decisions made, not activity (meetings, reports).
Applying It Well
- 30-day rule: "What did I do in October?" shrinks the action radius to reality; "what do I usually do?" inflates it.
- External evaluator rule: self-assessed capability is biased; without external evaluation, the item belongs in the cognition radius.
- Do not treat Capability-Action gaps with learning. Change environment, commitment structure, or incentives — not the information landscape.
- Expand circles in sequence: close Capability-Action first (releases existing potential), then Cognition-Capability, then expand cognition.
- Revisit every 90 days. Skills decay; new understanding reveals new ignorance; commitments change.
→ Primary sources: references/sources.md
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake move | Reality |
|---|---|
| [D] "I understand the domain, so I could do this if I tried." | Understanding and executing are different circles. No capability without deliberate practice and demonstrated output. |
| [D] "I'm working on it — just not there yet." | "Working on it" is not the action radius. Specific behaviors with measurable frequency are. |
| [D] "I need to learn more before I can start." | Most often the bottleneck is a Capability-Action gap. More learning delays action without fixing the root cause. |
| [D] "I can't close the capability-action gap without more motivation." | Motivation is unreliable. Commitment devices and environmental design are more reliable. |
| [D] "My team's capability is as large as our best individual's." | Team capability requires coordination and repeatability. The org radius is almost always smaller than the sum of individual radii. |
| → Add [O] entries here after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags
- Stated expertise significantly larger than demonstrated outputs — and gap not actively closing
- Team's action radius excludes their highest-stated-priority domain
- Gap diagnosis done with self-report only; no external validation
- Capability-Action gap addressed by adding training, not changing environment
- Map not revised in 90+ days despite significant domain changes
Verification
- Cognition radius: items passing the validation test only (teachable, peer-reviewable)
- Capability radius: items with external evaluation only, not self-assessed
- Action radius: behaviors from last 30 days only, not "typical"
- Both gaps identified with specific items and named root causes
- Priority gap for each category has a 30-day closure action with named validation
- 90-day revision date is set
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/three-radius-model · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/three-radius-model.json