--- license: Apache-2.0 name: naturalistic-insight description: Research on how insights and creative solutions emerge naturally during problem-solving and expertise category: Cognitive Science & Decision Making tags: - insight - problem-solving - creativity - ndm - cognition --- # SKILL.md: Naturalistic Insight (Klein & Jarosz) ## When to Use This Skill Load this skill when the user faces challenges where **reframing** rather than **optimizing** is needed: - Contradiction signals: Data doesn't fit the current model - Impasse situations: Standard approaches aren't working - Organizational insight barriers: Teams excel at error prevention but fail to generate new understanding - Agent/AI system design: Building systems that need to discover novel solutions ## Decision Points **Primary Decision Matrix: Anomaly Type × Context → Pathway Selection** ``` IF anomalous data persists despite explanation attempts: ├─ IF in research/investigation context → Contradiction Pathway │ └─ Action: Treat anomaly as most reliable data point │ └─ Question: "What frame makes this the normal case?" ├─ IF in operational context with time pressure → Knowledge Shield Check │ └─ Action: Document anomaly for later analysis │ └─ Escalate to subject matter experts └─ IF in learning/exploration context → Pattern Accumulation └─ Action: Log as gradual insight opportunity └─ Set up tracking infrastructure IF experiencing genuine impasse (multiple solution attempts failed): ├─ IF problem is well-defined domain → Execution Problem │ └─ Action: Standard decomposition and optimization ├─ IF problem definition feels wrong → Desperation Pathway │ └─ Action: Question foundational assumptions │ └─ Replace strongest anchors systematically └─ IF expertise feels like obstacle → Expertise Paradox └─ Action: Leverage knowledge for pattern detection └─ Remain open to frame revision IF noticing unexpected connections across cases: ├─ IF connection is metaphorical → Connection Pathway │ └─ Action: Add bridging anchor to mental model ├─ IF connection reveals causal relationship → Analytical Extension │ └─ Action: Standard hypothesis testing └─ IF connection spans domains → Cross-Domain Transfer └─ Action: Map structural similarities IF designing for insight generation: ├─ IF users are domain experts → Expertise-Enabled Design │ └─ Provide pattern libraries and anomaly detection ├─ IF organizational context → Assess Error vs. Insight Culture │ └─ IF error-focused: Create protected spaces │ └─ IF insight-friendly: Integrate with existing processes └─ IF AI/agent system → Multi-Pathway Architecture └─ Build contradiction, connection, and desperation modules ``` ## Failure Modes **1. Rubber Stamp Anomaly Dismissal** - *Detection*: You find yourself explaining away contradictions with increasingly sophisticated technical reasons - *Symptoms*: Multiple "special cases," forced explanations, mounting cognitive dissonance - *Fix*: Deliberately treat the weakest piece of evidence as most reliable; explore frames where anomaly is central **2. Sudden Insight Fixation** - *Detection*: System/process designed only for "aha!" moments and breakthrough experiences - *Symptoms*: Missing 44% of gradual insights, no infrastructure for pattern accumulation - *Fix*: Build tracking systems for incremental evidence gathering across cases over time **3. Expertise Contamination Fear** - *Detection*: Treating domain knowledge as obstacle to fresh thinking - *Symptoms*: Deliberately excluding experts, seeking "naive" perspectives only - *Fix*: Recognize expertise enables 2/3 of real insights; leverage knowledge while staying open to revision **4. Execution Task Decomposition** - *Detection*: Breaking insight problems into sub-tasks and parallel processing - *Symptoms*: Progress stalls, sub-solutions don't integrate, frame remains unchanged - *Fix*: Recognize holistic frame revision needed; don't decompose prematurely **5. Innovation Theater Deployment** - *Detection*: Adding insight processes to error-prevention organizations without addressing conflicts - *Symptoms*: Brainstorming sessions ignored, suggestion boxes empty, innovation initiatives fail - *Fix*: Create protected spaces with different norms or accept insights will remain accidental ## Worked Examples ### Example 1: AIDS Blood Supply Crisis (Contradiction Pathway) **Context**: 1981, CDC epidemiologist reviewing unusual pneumonia cases in gay men. **Initial Frame**: Isolated outbreak, likely lifestyle-related immune suppression. **Anomaly**: Hemophiliac patients presenting identical symptoms despite completely different demographics. **Novice Response**: "Hemophiliacs have compromised health anyway—probably coincidence." **Expert Application of Skill**: 1. **Anomaly Detection**: Two populations with no overlap showing identical rare syndrome 2. **Knowledge Shield Recognition**: Initial tendency to explain away as "different causes, similar symptoms" 3. **Weak Anchor Examination**: What if the connection IS the explanation? 4. **Frame Shift**: Single transmissible agent affecting both populations 5. **Implication Following**: Blood supply contamination hypothesis 6. **Action**: Test blood products, not just lifestyle factors **Outcome**: Discovery of bloodborne transmission route, saving thousands of lives. **Decision Points Navigated**: - Chose contradiction pathway over connection pathway - Treated demographic difference as data, not dismissal reason - Leveraged epidemiological expertise to recognize impossible coincidence ### Example 2: Fire Commander's Escape Decision (Desperation Pathway) **Context**: Fire commander Klein interviewed, routine house fire response. **Initial Frame**: Standard interior attack, crew positioned normally. **Impasse Signals**: Fire not responding to water, heat increasing unexpectedly, no visible progress. **Expert Application of Skill**: 1. **Impasse Recognition**: Standard procedures failing despite correct execution 2. **Strong Anchor Identification**: "Interior attack is always correct for this type of fire" 3. **Anchor Replacement**: "This fire is not what it appears to be" 4. **Frame Revision**: Basement fire burning undetected, floor about to collapse 5. **Immediate Action**: Order evacuation **Outcome**: Crew evacuated 30 seconds before floor collapsed. **Decision Points Navigated**: - Recognized desperation pathway signals (expertise feels like obstacle) - Questioned foundational assumptions rather than execution details - Trusted pattern recognition over procedural compliance ### Example 3: Darwin's Gradual Coral Atoll Insight (Connection Pathway) **Context**: Darwin studying coral formations during Beagle voyage. **Pattern Accumulation**: Observing multiple atolls across different locations over months. **Connection Emergence**: Noticing structural similarity between atolls and barrier reefs around volcanic islands. **Expert Application of Skill**: 1. **Pattern Tracking**: Systematic documentation of coral structures 2. **Cross-Case Analysis**: Comparing formations across geographic regions 3. **Geological Knowledge**: Understanding volcanic island subsidence 4. **Bridging Anchor**: "Atolls are sunken barrier reefs" 5. **Prediction Testing**: Drilling should find ancient reef limestone beneath modern coral **Outcome**: Revolutionary theory of coral formation, confirmed by drilling decades later. **Decision Points Navigated**: - Recognized gradual insight pattern (no sudden "aha!" moment) - Used expertise to detect structural similarities across cases - Built infrastructure for pattern tracking over extended time periods ## Quality Gates **Frame Coherence Check**: [ ] New frame explains previously unexplained anomalies without special pleading [ ] Frame generates testable predictions different from old model [ ] Revised understanding integrates smoothly with established domain knowledge [ ] Frame addresses root cause, not just symptoms **Prediction Power Validation**: [ ] New frame successfully predicts behavior in untested cases [ ] Predictions are specific enough to risk falsification [ ] Frame explains both positive and negative cases in dataset [ ] Model performs better than previous best explanation **Expert Acceptance Assessment**: [ ] Domain experts can follow the reasoning path [ ] Insight leverages rather than dismisses existing expertise [ ] Other experts can replicate the insight given same data [ ] Resistance based on evidence, not just unfamiliarity **Anomaly Resolution Verification**: [ ] Original contradiction/impasse is genuinely resolved, not just deferred [ ] Resolution doesn't create larger contradictions elsewhere [ ] Previously dismissed data now makes sense within new frame [ ] Weak anchors that triggered insight are now integrated **Implementation Readiness**: [ ] Clear next actions follow from the insight [ ] Resource and timeline implications are understood [ ] Organizational barriers to implementation are identified [ ] Success metrics for new approach are defined ## NOT-FOR Boundaries **This skill is NOT for**: - Routine optimization problems with clear metrics → Use `systematic-optimization` instead - Error correction in well-understood systems → Use `debugging-protocols` instead - Incremental improvements to existing processes → Use `continuous-improvement` instead - Problems with obvious decomposition into subtasks → Use `task-decomposition` instead - Situations requiring immediate action without reflection → Use `crisis-response` instead **Delegate to other skills when**: - User needs brainstorming techniques → Use `creative-ideation` instead - User wants systematic innovation methods → Use `systematic-inventive-thinking` instead - User faces behavioral change challenges → Use `behavior-change-design` instead - User needs decision-making under uncertainty → Use `decision-analysis` instead - User requires team creativity facilitation → Use `group-problem-solving` instead **Critical boundary**: This skill focuses on naturalistic insight (how insights actually emerge in real contexts) not laboratory creativity (puzzle-solving, divergent thinking exercises). If the user wants "innovation training" or "creative thinking workshops," redirect to appropriate facilitation skills.