When Experts Can Trust Their Gut

A fire commander walks into a house.

The kitchen fire is not behaving normally.

He cannot immediately explain why.

He orders the team out.

Soon after, the floor gives way.

That is the kind of story people tell when they want to prove intuition is powerful.

And it is powerful.

But the wrong lesson is:

Always trust your gut.

The better lesson is:

Ask what trained the gut.

Two researchers who were supposed to disagree

Daniel Kahneman became famous for research showing predictable errors in judgment.

Gary Klein became famous for studying experts making fast, effective decisions in the real world.

One tradition emphasized bias.

The other emphasized expertise.

In 2009, they published a paper together.

The title was revealing:

Conditions for Intuitive Expertise: A Failure to Disagree. [S01]

Their central conclusion was not that intuition is good.

It was not that intuition is bad.

It was that intuition has conditions.

Condition one: the environment must contain regularities

A person can only learn a pattern that exists.

Consider a nurse who sees the same kinds of patients every day.

There may be subtle combinations of:
- skin color;
- breathing;
- movement;
- speech;
- vital signs;
- timing;

that predict deterioration.

The environment contains signal.

Now consider predicting the next lottery number.

No amount of watching prior draws gives the intuition the needed predictive cue.

Experience cannot manufacture a regularity that is absent.

Kahneman and Klein called attention to the validity of the environment. [S01]

High-validity environments contain cues that genuinely predict what comes next.

Low-validity environments do not.

Condition two: the person needs a chance to learn

Signal is not enough.

The learner needs feedback.

A chess player makes a move.

The consequences arrive.

A firefighter chooses a tactic.

The fire responds.

A pilot performs a maneuver.

The instruments and outcome provide information.

Repeated exposure plus feedback can calibrate pattern recognition.

But feedback can fail.

The problem of delayed feedback

Imagine a hiring manager.

They meet a candidate.

Their intuition says:

This person is a star.

The candidate is hired.

Six months later, performance is mixed.

Was the intuition wrong?

Maybe.

Or the onboarding was poor.

Or the role changed.

Or the manager forgets the original prediction.

Or the person leaves before the outcome is clear.

The feedback loop is noisy.

That makes expertise difficult to build.

Selective feedback is even worse

Suppose the manager remembers spectacular successes.

They forget the candidates they loved who failed.

They never see what would have happened with the candidates they rejected.

Now confidence can grow without calibration.

The person has experience.

But the learning environment is broken.

Recognition-primed decisions

Klein's naturalistic decision-making research studied people making consequential decisions under time pressure, including fireground commanders and military personnel. [S02]

The recognition-primed model describes a process unlike textbook option comparison.

The expert may:
1. recognize the situation as familiar;
2. identify a plausible action;
3. mentally simulate whether it will work;
4. modify or act.

They do not necessarily generate ten alternatives.

That speed can be a product of expertise.

Why verbal explanation can lag behind recognition

A skilled person may say:

Something is wrong.

Then, after reflection:

"The smoke was too quiet."

"The heat pattern did not fit."

"The structure sounded wrong."

The cues were available.

They simply were not consciously enumerated before the judgment.

This is one reason forcing an expert to produce immediate analytical justification can sometimes interfere with performance.

But there is a danger.

A biased person can also fail to explain a judgment.

Lack of explanation is not evidence of expertise.

Direct evidence for domain expertise

Dane, Rockmann and Pratt experimentally compared intuitive and analytical decision making in basketball and designer-handbag authentication. [S03]

They found that intuitive decision effectiveness relative to analysis increased with domain expertise.

This supports a crucial rule:

Trust should be domain-specific.

A basketball expert's fast judgment about basketball can be useful.

That does not make their intuition generally superior.

Experience is weaker than people think

Clinical judgment research offers an important warning.

A large meta-reanalysis by Spengler and Pilipis examined 113 studies involving more than 11,000 clinicians. [S16]

The relationship between experience and judgment accuracy was small.

Years in a profession do not automatically create large improvements in judgment.

Why?

Because time spent doing something is not the same as learning from valid feedback.

Expertise needs error correction

Real expertise is not:

I have done this for twenty years.

It is closer to:

I have made thousands of relevant predictions, observed the results, studied the failures and updated the cues I rely on.

That is a different claim.

Some environments fool experts

Financial markets are a classic example.

Outcomes are noisy.

Luck can look like skill.

The environment changes.

Feedback can be delayed.

A good decision can lose money.

A bad decision can make money.

This makes intuition difficult to calibrate.

The same problem appears in:
- forecasting;
- complex hiring;
- long-term relationships;
- political prediction;
- entrepreneurship.

The environment contains some signal.

But it may be weak, changing or difficult to separate from noise.

Why confidence is dangerous

Kahneman and Klein explicitly note that subjective experience is not a reliable indicator of judgment accuracy. [S01]

An expert can feel certain in a domain where certainty is not earned.

Confidence can come from:
- familiarity;
- social status;
- repetition;
- selective memory;
- institutional authority.

That is why expert intuition needs an external audit.

A five-question expertise audit

Before treating a gut judgment as expert intuition, ask:

1. Is this the person's actual domain? Not adjacent.

Actual.

2. Does the environment contain stable predictive cues? Could a learner in principle get better?

3. Has the person seen many relevant examples? Not just dramatic ones.

4. Did they receive timely and accurate feedback? Could they learn which cues were correct?

5. Is performance tracked? Do we know the intuition is calibrated, or do we only know the person is confident?

If those conditions are weak, the word "expert" should receive less weight.

What about medicine?

Expert intuition can be valuable in medicine.

It can also be dangerous.

A radiology review in 2022 emphasized both sides: intuition is pervasive in expert practice, but developing reliable expert intuition requires repeated exposure to valid cues and feedback. [S18]

High-stakes practice works best when intuition and verification cooperate.

The expert notices.

Then the system checks.

What about relationships?

Relationship intuition is hard to calibrate.

You do not get hundreds of identical partners.

Feedback is emotionally loaded.

Outcomes take months or years.

Past relationships change future perception.

That means a "bad feeling" can still be worth attention.

But the correct next step is usually observation.

Not accusation.

The strongest intuition is testable

Paradoxically, the expert who deserves trust is often willing to check.

They do not say:

My gut is never wrong.

They say:

Something here is unusual. Let's verify it.

That is the difference between intuition as a tool and intuition as ego.