Chapter 22: Expert Intuition: When Can We Trust It?

Type: Chapter 22

Category: Psychology

Difficulty: Intermediate

Reading time: 6 min

Core idea

This chapter is a dialogue between two positions: Kahneman’s skepticism about expert intuition (documented in Chapters 20–21) and Gary Klein’s naturalistic decision-making research, which documents genuine expert intuition among firefighters, military commanders, chess players, and intensive care nurses. Both positions are defensible. The synthesis is about conditions.

Expert intuition is genuine when two conditions are met:

1. A regular environment: the domain has enough structure and stability that patterns repeat and can be learned.
2. Sufficient practice with feedback: the expert has had enough experience with outcomes to calibrate their pattern recognition.

When these conditions hold, System 1 has been trained by experience to recognize patterns and retrieve appropriate responses — the same mechanism that chess grandmasters use to instantly see that a position is dangerous, or that experienced nurses use to sense that an infant is about to crash before the monitors show it.

When these conditions do not hold — when the environment is irregular (financial markets, geopolitics) or feedback is slow, ambiguous, or absent — expert intuition is an illusion of validity, regardless of experience or credentials.

Why it matters

Klein’s research: genuine expert intuition

Gary Klein studied firefighters who made rapid decisions in burning buildings. He expected to find deliberate comparison of options; instead, he found experienced commanders could immediately recognize situations as types and know the appropriate response — without consciously comparing alternatives. The recognition-primed decision model: experts do not compare options; they recognize situations.

This is System 1 expertise: pattern recognition developed from thousands of calibrated repetitions in an environment with reliable feedback. Novices need to compare options; experts recognize and act. The expert’s intuition is not a lucky guess — it is the output of a trained statistical classifier.

The two-conditions test

The question “Can this expert intuition be trusted?” reduces to two diagnostic questions:

Is the environment regular?: Does the domain have enough structure that patterns repeat predictably? Chess is highly regular — the same configurations arise across games, and the same moves succeed against them. Financial markets are not regular — price movements depend on new information by definition, and patterns that once worked stop working when discovered.

Has the expert received adequate feedback?: Has the expert made enough predictions in this domain and received fast, accurate, unambiguous feedback on the outcomes? A cardiologist who reads echos daily and gets immediate confirmation from surgical findings develops valid pattern recognition. A psychotherapist whose patients may improve, plateau, or worsen over years for multiple reasons does not develop the same calibrated feedback.

The misleading feeling

Both genuine and illusory expertise feel the same from the inside. The chess grandmaster and the confident stock-picker both experience their intuitions as immediate, compelling, and grounded in pattern recognition. The feeling of expertise — the sense of “seeing” the situation clearly — does not distinguish between trained pattern recognition and confident narrative confabulation.

This is why the two-conditions test matters: it is an external diagnostic that bypasses the subjective experience of confidence.

Author’s argument: Trust expert intuition when the conditions for valid learning have been met. Do not trust it when they have not — regardless of the expert’s experience, confidence, or reputation.

Key takeaways

Key takeaways

• Expert intuition is genuine when two conditions hold: a regular environment (patterns repeat) and adequate feedback-calibrated practice (the expert has learned those patterns).
• Klein's recognition-primed decision model: experienced experts recognize situations as types and retrieve appropriate responses — they do not compare options, they recognize patterns.
• The two-conditions test: ask (1) Is this domain regular enough for patterns to repeat? (2) Has the expert received sufficient, accurate, fast feedback to calibrate their pattern recognition?
• Both genuine and illusory expertise feel the same: the feeling of 'seeing clearly' does not distinguish trained pattern recognition from confident confabulation.
• Irregular domains (financial markets, geopolitics): no expert intuition is valid, regardless of credentials, because the environment does not provide reliable learning opportunities.
• The synthesis: trust expert intuition in firefighting, chess, surgery, and intensive care nursing; be skeptical in investment management, political forecasting, and clinical prediction from interviews.

Mental model

click to zoom

The two conditions for valid expert intuition — and the domains where each condition does and does not hold.

Read it as: Expert intuition is valid only when both conditions hold: the environment must be regular enough for patterns to exist, and the expert must have received enough calibrated feedback to learn those patterns. When either condition fails, what feels like expertise is an illusion of validity generated by coherence and confidence, not by actual predictive skill.

Practical application

Auditing your intuitions

Before relying on an expert intuition (your own or someone else’s), run the two-conditions test explicitly: (1) Is this domain regular? Do the same patterns reliably recur? (2) Has the expert received fast, accurate, unambiguous feedback from enough similar situations? If either answer is no, treat the intuition as a useful starting point for System 2 scrutiny — not as a reliable guide to action.

Domain classification:

Domain	Regular?	Feedback calibrated?	Trust intuition?
Chess	Yes	Yes	Yes
Surgery	Mostly	Mostly	Often
Firefighting	Mostly	Yes	Often
ICU nursing	Mostly	Yes	Often
Stock picking	No	No	No
Political forecasting	No	No	No
Psychotherapy outcomes	Partly	No	Cautiously
Hiring interviews	Partly	Rarely	Skeptically

Example

A seasoned ICU nurse notices that a premature infant “just doesn’t look right” — though all vital signs are normal. She escalates to the attending physician. Three hours later, the infant develops sepsis. The nurse’s intuition was valid: she had cared for hundreds of premature infants and received fast, unambiguous feedback about which patterns preceded deterioration. Her pattern recognition was genuine expertise.

By contrast, a venture capital partner with twenty years of experience and a strong track record “just knows” that a particular startup will succeed, based on a forty-five-minute meeting. Financial markets are not regular (the successful patterns change), and feedback on venture investments arrives years later, noisily, confounded by market conditions. The confidence is the same as the nurse’s. The validity is not.

Related lessons

Related

Chapter 21: Intuitions vs. Formulas The evidence that formulas beat expert judgment in low-feedback domains — the problem this chapter diagnoses. Chapter 20: The Illusion of Validity How coherence generates confidence independently of accuracy — the psychological basis of illusory expert intuition. Dual-Process Theory System 1 expertise — how fast, automatic pattern recognition develops from calibrated experience in regular environments.