The Dreyfus model of skill acquisition

Stuart Dreyfus, a mathematician, and his brother Hubert Dreyfus, a philosopher, developed the model at the University of California, Berkeley, in a 1980 report for the US Air Force Office of Scientific Research. Their target was artificial intelligence: if expertise were really a very large set of rules, a machine could hold them all. Their counter-claim was that rules are what beginners lean on and what experts have left behind. The model reached a wider audience through their 1986 book Mind Over Machine, and through Patricia Benner, who applied it to clinical nursing in From Novice to Expert in 1984.

The five stages at a glance

What actually changes between the stages

These are not levels of knowledge, and nobody moves up by being given more information. Three things shift, and they shift together:

• From rules to experience. A beginner needs rules that hold without context, because they have no context yet. Experience slowly replaces them with remembered situations.
• From parts to whole. Early on a situation arrives as a flat list of elements, all apparently equal. Later it arrives pre-sorted, with the important thing already standing out.
• From detached to involved. A beginner can stay analytical and unbothered by the outcome. Past competence, the performer has something at stake. The Dreyfus brothers argued this involvement is not a side effect of progress but the engine of it: people who stay detached from results tend to stop developing.

Novice and advanced beginner: let the rules do the thinking

A novice has no experience to draw on, so they need rules stated in terms they can recognise without judgement. "Ask three questions before you propose anything" is a good instruction for a novice. "Use your judgement" is not, because there is nothing yet to judge with. Telling a beginner to read the room produces either paralysis or a guess.

The advanced beginner has been through enough real situations to start recognising patterns that cannot be written down cleanly, and works from maxims rather than rules. But everything still feels equally urgent, so they attend to the wrong thing at the wrong moment. Most workplace training stops precisely here, issues a certificate, and calls it capability.

What to do

• Give explicit rules and a short checklist, and say plainly that these are training wheels you will remove later.
• Grow the number and variety of real situations faster than you grow the theory. Reps beat frameworks at this stage.
• Name the aspects they could not have seen. "You missed that she went quiet after the price came up." That is how situational recognition gets built.

Competent: the stage where most careers park

Competence arrives when a person can see more relevant factors than they can possibly hold at once. The way out of that overload is to choose: pick a plan, a goal, a perspective, and let it decide what counts as important today. That choice is what makes competent performers effective, and also what makes them anxious, because the outcome is now genuinely theirs. They plan, prioritise, cope and deliver, which is why organisations promote them and why so many organisations are staffed entirely by them.

Competence is the first stage that feels stressful, because it is the first stage where the result belongs to you.

The trap is that the easiest way to manage that stress is to retreat into process. Follow the method, and if it fails, the method was at fault. That protects the person and freezes the skill. Movement past competence runs the other way: more variety, more consequence, and honest feedback on the choices made rather than on the results achieved.

What to do

• Make them commit to a plan out loud before acting, then review the plan, not the result. Praising a good outcome that came from a bad read is the most expensive habit a manager has.
• Vary the conditions deliberately. Someone who only ever meets the same situation becomes fast at one thing and brittle at everything else.
• Let them own something end to end, including the part where it goes wrong in public. Removing consequence removes the pressure that builds judgement.
• Stop supplying the answer in the moment. Every rescue keeps them at the stage where rescue is needed.

Proficient and expert: recognition replaces deliberation

The proficient performer stops assembling the situation and starts perceiving it. They walk into a meeting and know it is going badly before anything unusual has been said. What they still do is deliberate about the response, running options against that read. The expert closes the last gap: seeing the situation and knowing the response become a single act, and deliberation returns only when something genuinely novel appears.

Two cautions. Intuition of this kind becomes trustworthy only where the environment is regular enough to learn from and where feedback arrives quickly and clearly, which is the substance of the long argument between researchers on expert intuition and on bias. Negotiating, hiring and coaching give slow, noisy feedback, so confidence there deserves more scrutiny than confidence in a cockpit. Second, experts often teach beginners badly, because they no longer have access to the rules they once used.

• Give them problems that are new, not problems that are merely bigger. Novelty develops an expert; volume does not.
• Ask them to think aloud while working. It is the only reliable way to get what they know back out for everyone else.
• Check their intuitions in slow-feedback domains against something external. Confidence and accuracy come apart quietly.

How to tell which stage someone is in

Do not ask how many years they have done the job. Ask them to talk you through a recent decision, then listen to what they reach for first:

1. A novice cites the rule. "We're supposed to escalate anything over that value."
2. An advanced beginner cites a similar past case, usually the most recent one, and applies it whether or not it fits.
3. A competent performer describes the plan and the priorities, and can tell you what they chose to ignore.
4. A proficient performer describes the situation first. "It was obvious the sponsor had already decided," then explains the options weighed after that.
5. An expert struggles to explain at all, and reconstructs a rationale afterwards because you asked for one.

One caution: fluent explanation is not proof of skill. Plenty of people can narrate expertise they do not have. Watch what they do when the situation changes shape halfway through.

Where the model breaks down

It is taught almost everywhere, and it is worth being honest about what it is not.

1. It is a description, not a measurement. It came from a small set of domains, including chess players, pilots, drivers and adult language learners, and was argued philosophically rather than validated as a rating scale. Scoring people on it in an appraisal cycle turns it into rating theatre.
2. Stage belongs to a skill in a context, not to a person. Someone can be proficient at handling a furious client and a novice at chairing a meeting. There is no such thing as "a proficient", and tenure tells you very little.
3. Experience alone does not carry anyone up the stages. Repetition in a forgiving environment produces confident habits, not expertise. The research tradition around deliberate practice pushes back hard on the idea that experts simply stop being analytical.
4. Progress is not one-way. A new market, a new tool, a new team, and a proficient performer is briefly an advanced beginner again. That regression is normal and short, provided nobody treats it as failure. The same happens to teams, which is the point of Tuckman's stages.

Used properly it does one job very well: it stops you teaching everybody the same way. Most training failures inside organisations are stage mismatches rather than content failures. Four are worth watching for:

• Telling a novice to use their judgement, when they have nothing yet to judge with.
• Handing an expert a checklist as a general management style. Fine for safety-critical steps, corrosive everywhere else.
• Running one workshop for a room holding both novices and proficient performers, then wondering why half were lost and half were bored.
• Promoting people at competence and funding nothing that would move them past it. The most common of the four, and the most expensive.

This is where classroom training and real skill part company. You can teach a novice's rules from a slide deck. You cannot teach the reading of a situation that way, because that reading only develops when a person is inside a situation, has to commit, and then has to live with what follows. That is the whole argument for experiential learning: build a compressed version of the real thing, make people choose under pressure, then debrief the choice rather than the result. Watch a group work through one for twenty minutes and the stages are visible in the room. There is the person still hunting for the rule, the one defending a plan they picked far too early, and the one who saw it coming and cannot tell you how.

Tour De Force runs this as live, experiential training for teams worldwide — online and in person. Talk to us, or play Gamified learning appsThe Weekly Challenge to see the method in ten minutes.