Assembling a Pattern Language for Naturalistic Decision Making

Assembling a "pattern language" (inspired by architect Christopher Alexander) under the framework of Naturalistic Decision Making (NDM) (pioneered by Gary Klein) bridges the gap between tacit expert intuition and actionable, structured training.

GoogleGemini says:

In NDM, experts do not mentally calculate odds or compare extensive matrices; they use Recognition-Primed Decision (RPD) models. They match the current situation to a library of mental prototypes (patterns) built from years of experience. By externalizing these prototypes into a formalized pattern language, you can accelerate training, improve team coordination, and preserve institutional knowledge. [1]

Here is a blueprint for structuring and building this language.

1. The Anatomy of an NDM Pattern

Traditional software or architectural design patterns focus on structural solutions. An NDM pattern must focus on cognitive framing and situational awareness. Every pattern in your language should include these five elements:

• The Cue/Trigger (The Context): What environmental signals, data points, or subtle anomalies should an operator notice?
• The Diagnostic Match (The Problem): What mental model does this situation map to? What is the real underlying problem or threat?
• Expectancies: What should happen next if this diagnosis is correct? (If these expectancies fail, the operator must immediately re-evaluate).
• The Action Script (The Resolution): The default course of action that an expert instinctively knows will work, without needing to compare alternatives.
• The Umwelt / Contextual Fit: The specific constraints, high-stress variables, or emotional states of the team that alter how this pattern plays out. [2]

2. Framework for the Pattern Language

To make the language scannable and functional, group your patterns into a hierarchical taxonomy progressing from micro-cues to macro-strategies:

🟢 Level 1: Cue Recognition & Sensemaking Patterns

These patterns help operators translate raw, chaotic data into situational awareness.

• Example Pattern: "The Missing Sound"

• Context: Routine operational environment.
• Cue: The absence of a standard ambient noise (e.g., a specific hum in a plant, a predictable cadence in team radio chatter).
• Expectancy: Teams usually miss this because of automation bias.
• Action: Explicitly halt or trigger a diagnostic check to find out why the baseline changed.

**🟡 Level 2: Expectancy-Violation Patterns

These patterns dictate what to do when a chosen course of action deviates from reality.

• Example Pattern: "The Sticky Schema"

• Context: A crisis where the initial diagnosis was made, but incoming data is beginning to contradict it.
• Problem: Confirmation bias causes the commander to ignore new anomalies.
• Action: Appoint a "Devil's Advocate" or a dedicated red-team role whose sole job is to validate the contradictions against expectancies.

🔴 Level 3: Action Scripting under Time-Pressure

These patterns map specific crisis archetypes directly to singular, robust solutions.

• Example Pattern: "Sacrifice Play (Stabilization First)"

• Context: Multi-system failure where saving everything is impossible.
• Problem: Choice paralysis caused by trying to optimize all variables.
• Action: Sever secondary priorities immediately (e.g., drop a specific server cluster, abandon a room in a burning building) to consolidate resources on the core asset.

3. How to Extract and Assemble the Language

Because expert knowledge is largely tacit and contextual, you cannot gather these patterns via standard interviews or surveys. You must use Cognitive Task Analysis (CTA) methods: [1]

1. Critical Incident Technique (CIT): Interview your most seasoned experts. Do not ask them how they usually do their job. Ask them to recount the worst, weirdest, or most complex incidents they ever handled.
2. Probe for the Pivot Points: During the interview, relentlessly drill down on the micro-moments. Ask: "What exactly did you see or hear that made you realize the situation had changed?" or "What did you expect to happen next?"
3. Identify "Pattern Smells": Document where less-experienced operators misapply these patterns or execute them mechanically without understanding the underlying environmental context. [2]
4. Format for Vividness: Keep the final pattern descriptions concise, clear, and highly expressive. Use sketches, timelines, or brief anonymized war-stories so the pattern immediately resonates with a reader's own experience. [2]

[1] https://multiplenatures.com
[2] https://wirfs-brock.com