36Revisiting the Six Stages of Skill Acquisition

B. Scot Rousse and Stuart E. Dreyfus, in Teaching and Learning for Adult Skill Acquisition: Applying the Dreyfus & Dreyfus Model in Different Fields, ed. Mangiante and Peno, Information Age, 2021

Linked to 26 chapters: supports 12 · contradicts 7 · mechanism 18 · same claim 4 · pairing 6

Stuart Dreyfus and his brother Hubert first set this account down in an eighteen-page report dated February 1980, written at Berkeley's Operations Research Center under contract to the Air Force Office of Scientific Research and illustrated from flight instruction, chess and adults learning a second language. That report named five stages: novice, competence, proficiency, expertise and mastery. Mind Over Machine, the book the brothers built on it in 1986, inserted the advanced beginner in second place and let mastery fall away, leaving the five-stage sequence that spread through nursing, medicine, teaching and management. Hubert Dreyfus died in 2017. Here Stuart Dreyfus, writing with the philosopher B. Scot Rousse, restates the model with mastery restored as a sixth stage and with the transitions between stages, rather than the stages themselves, as the subject. What they defend is that skill does not grow by accumulating better rules. Rules, and the analytic reasoning that applies them, are the scaffolding of the first three stages, and they are what has to be let go. What replaces them is a perspective: a way a situation shows up such that some of its elements stand out as relevant and the rest recede and can be ignored. The competent performer has to choose that perspective, and pays for the choice in emotional exposure, because the outcome is now his. The proficient performer finds that it arrives on its own. The expert finds that it arrives with the action already attached. The account is meant to cover walking down a crowded pavement as much as playing the guitar, because it rests on a tendency of experienced people to acquire an intuitive familiarity that orients them without calculation.

From novice to expert

Instruction begins by decomposing the task into context-free features, features a beginner can recognise without any experience of the domain, and supplying rules that combine them. The student driver is given speed off the speedometer and a formula for safe following distance as a function of it. The novice chess player is given a number for each kind of piece, independent of position, and the rule to exchange whenever the value captured exceeds the value lost. The novice cook is given a recipe: preselected ingredients, measured amounts, an oven temperature, an interval. What the novice receives is theoretical knowledge, rules abstracted from any involvement in the domain, and what the novice does with it, applying a general rule to an objective feature, the authors call calculative reasoning. It performs poorly, because the world is a repository of contingencies no rule anticipates: the distracted driver who careens into your lane, the opponent who gives up valuable pieces to gain position, the cooking time that depends on the pan and the state of the stove. The rules also rest on discriminations nobody teaches. No one gives the student driver a rule for understanding that what a mirror shows is behind them, or for telling a truck from a motorcycle, or a knife from a fork; the learner brings a whole background of ordinary competence to the first lesson. The novice, absorbed in getting his own performance out, treats other people as one more feature to be taken into account. Attending to what others are about to do comes later.

Experience then enriches perception in two ways. The first is more of the same: more features and more elaborate rules. The driver comes to know the fuel gauge, the seatbelt light, the mirrors, the sequence for leaving a parking space. Nothing has changed in kind. The second is a change in kind, and it defines the advanced beginner. The learner starts to recognise what the authors call situational aspects, recurrent phenomena an instructor can point at and name but cannot define: the sound of a racing engine, the head-tilt of a driver looking down at a phone, the smell of olive oil beginning to burn, a weakened king's side, a strong pawn structure. Aspects support maxims rather than rules. Slow down and keep back when the other driver's head is tilted. Reduce the heat and add the garlic when the oil starts to burn. Attack a weakened king's side. A maxim is useless to anyone who cannot make the discrimination it names, and that discrimination is not transmitted by explanation. There is no rule for the smell of burning oil; there are only examples.

The advanced beginner pays for the enrichment in congestion. The field of awareness fills with elements demanding attention and with more rules and maxims to juggle; the task seems harder as the situation becomes a morass of details, and even work that will later look routine feels overwhelming. The authors take this seriously as a teaching problem: at this stage and the next, a mentor's job is as much to cultivate an emotional disposition that keeps the learner going as to supply more content. They also note that the stages are not a ladder climbed once. A learner can enter a domain in a composite of the first two, as musicians do who mimic and tinker before they learn any rules, or as someone does who skims the manual for a new program, starts tinkering, and returns to the manual only on hitting a wall.

Competence arrives when the learner learns to choose a perspective. Instead of an aggregation of features competing for attention, the situation is taken up in relation to a goal, and the goal sorts it: some elements become salient, others become ignorable, and the steps of the task fall into a provisional order. A driver who has chosen to reach a destination efficiently finds lanes, lights and traffic density salient, while scenic beauty, driving courtesy and the open sunroof are not, until it starts to rain. A competent cook runs two dishes in parallel under a chosen scheme, both ready when the guests arrive, or the cold salad before the hot pasta, and each scheme foregrounds a different sequence of chopping and a different choice of pans. The trouble is that no rule chooses the perspective. The competent performer looks for one, and often there is none; he must decide for himself what is appropriate here, with no guarantee it will work.

That is where emotion enters the model, and the authors make it load-bearing rather than decorative. Up to this point a learner could stay detached and blame a poor result on inadequate rules. Now the result follows from his own choice, so he feels remorse when it fails and elation when it succeeds. Of a competent chess player they write that "Successful attacks induce euphoria, while mistakes are felt in the pit of the stomach". The feeling is the mechanism by which experience accumulates: the joy of a job well done makes it more likely that the perspective which produced it will occur to the performer unbidden next time. Proficiency, they say, seems to develop only if experience is assimilated in this involved, affect-laden, atheoretical way. Which is also why many performers stop here. Competence is adequate in most domains, and there is comfort in falling back on rules and procedures once the perspective has been chosen, a comfort a rationalistic culture flatters by treating detached calculation as the best thing a mind can do. Going further means taking a different risk: giving up rules and decisions in favour of an intuition that is only beginning to quicken.

Proficiency and expertise

The mark of the fourth stage is that the perspective is no longer chosen; it occurs. Typical situations group themselves, and their elements arrive already polarised into foreground and background. Borrowing James Gibson's term, the authors say the proficient performer begins to pick up the affordances of a situation directly: a gap in the traffic that affords a lane change, a tree that affords escape. There is correspondingly less doubt about whether one has read the situation right. But the reading and the response have come apart, because there are far fewer ways of seeing a situation than of reacting to it, and the proficient performer has not yet lived through the outcomes of enough responses. So he sees intuitively and then decides analytically. The proficient driver feels in the seat of his pants that he is too close to the car ahead on a wet curve, and must then work out whether to brake, lift his foot, or merely ease the pressure. The proficient cook knows at once that the sauce is thin and must consider arrowroot, cornstarch, gelatin or xanthan gum. The proficient chess player sees the salient forces in a position immediately, knows she should attack, and has to calculate how.

Expertise closes that gap. With enough experience of what follows from acting in a given kind of situation, the salience of the situation is experienced as a direct solicitation: attractive and repulsive forces that draw the performer into action without comparison of alternatives. The expert driver's foot comes off the accelerator. The expert cook, shown a pantry, sees a set of dishes it makes possible, the order of preparation, and the first moves, and starts. The grandmaster plays at five to ten seconds a move without noticeable degradation, which is only possible if intuition and not analysis is leading; an expert player is estimated to discriminate something like a hundred thousand types of position. The expert nurse of Benner's studies knows a patient is near cardiac arrest from his colour, his eyes, a tremor, before the monitors show anything, and knows what to adjust. Expertise in this sense is not reserved for elite performance: given enough experience, people become expert at walking down pavements, greeting colleagues and ordering food in a crowded restaurant, and animals become expert at what their survival requires. Elite skills need talent and sustained motivation on top of the experience; everyday ones need only the experience.

The expert's repertoire is a repertoire of perspectives, and a second capacity comes with it. An expert nurse watching a patient through a transition knows, without deliberating, when the evidence no longer fits the perspective she has been working under, and lets a new one take hold. The authors call this reflexive reorientation, and they insist it need not be a decision. An outfielder under a fly ball sees the angle of ascent, the other fielders, perhaps the sun; if the ball turns out to have been hit harder than he read, the situation regestalts, and the wall, the runners on base and the batter's known speed become salient instead, with different actions solicited. Animals incapable of any deliberation switch in the same way from chasing prey to blocking the tree it might climb.

Performing across the stages

Experts, in sport especially, sometimes report periods of peak performance — flow, being in the zone, playing out of one's head — in which everything is easier, time passes unnoticed, and the monitoring the mind normally does falls silent. The authors do not count this as a further stage. They treat it as expertise at maximum focus, a transient state that cannot be chosen, arriving when the situation happens to mesh with what the performer is tuned for, and they suggest that during it the learning from experience which normally accompanies even expert performance briefly stops.

Nor does becoming an expert mean performing as one from then on. The central claim has always been a conditional, and the authors restate it because critics have taken it for an unconditional one: when things are proceeding normally, "experts don't solve problems and don't make decisions; they do what normally works". A performer's level may therefore vacillate across the stages within a single performance. Christensen, Sutton and McIlwain read that vacillation as an inconsistency in the model; the authors reply that it is the model's range, an account of the whole spread of skilled action rather than of its best moments only. Where the published record has been read narrowly, as they say it has been by Gobet, Høffding and Montero, what results is a caricature in which experts are supposed never to think.

Novelty is what forces an expert down the stages, and it has several sources. Experience can be narrow: the musician who is expert in small clubs may find a large hall with stage monitors and a light show disorienting, and have to work with a new maxim about stepping away from the amplifier when the feedback starts. Domains also change underneath their experts. Tennis rackets changed material and head size and altered the dynamics of play; basketball added the three-point shot and the slam dunk to its constitutive rules; software updates change what the program can be asked to do. Local regulations vary: an expert driver or pedestrian in a country where the traffic runs the other way has to keep reminding himself that his intuitions will turn him into it. Even everyday expertise degrades when norms shift, as for newcomers from quiet places who brace for collision on a crowded pavement and have to work out a rule of thumb about turning the shoulders when an oncoming walker is an arm's length away.

Perspectival deliberation

Falling back on rules and maxims is one kind of deliberation, and the authors call it calculative: the performer has no perspective, so the situation presents itself as a collection of facts wanting attention, and general rules have to substitute for a sense of the whole. It is a degradation of expert performance, even when it later yields a new perspective. But there is a second kind, which the 1986 book called deliberative rationality and which this chapter renames perspectival deliberation. Here the perspective is already in place; the deliberation tests its limits, imagines alternatives, and tries to improve on it.

They describe two forms. In the first, more than one intuitive perspective has occurred, and the performer extrapolates: entertains the positions that would follow from acting on each and weighs them for confidence, risk and urgency, and if there is time, asks someone more experienced. This puts detached deliberation at the service of intuition. Deliberately taking up the opponent's perspective belongs here too, as protection against the blunder that comes of not anticipating his move, and against the tunnel vision of staying inside one perspective after the evidence has moved.

In the second, a contingent variation stays marginal under the current perspective but registers as unease, because past experience has taught that marginal things sometimes turn out to matter. Sometimes the tension trips a reflexive reorientation on its own. Sometimes the performer can hold on to it and attend deliberately to the marginal element until the situation regestalts around it and what was peripheral becomes central. The authors' example is the Mann Gulch fire of August 1949, by way of Karl Weick's study of it and Norman Maclean's account. A crew of smokejumpers dropped on what was taken for a manageable fire found it roaring towards them, two hundred yards off, with flames thirty feet high. The foreman, Wagner Dodge, read the situation as one for retreating to the ridge where the fuel is thinner, and ordered the climb. On the way up they were moving through grass two and a half feet tall while the fire gained. Under the perspective of getting uphill fast, the grass was an obstacle to push through, though Dodge, the authors suggest, had registered it as possibly relevant. When the climb stopped looking survivable, he attended to the grass, and a new perspective occurred in which it afforded an escape fire. He told the crew to drop their tools, lit the grass in front of them, and ordered them to lie down in the ground it had burnt. They heard the order as evidence that he had gone mad. Dodge survived in the ashes; thirteen of his men did not.

Mastery

The sixth stage is for performers who are not satisfied with doing what normally works. Where the expert engaged in perspectival deliberation withholds his solicited action and waits, receptive, to see whether another perspective will occur to him, the emerging master actively and experimentally chooses a different one. Because a chosen perspective carries no action with it, the action has to be chosen too, which puts the master deliberately back in the position of the merely proficient, and risks a worse performance. That is why it is done in practice and rehearsal: the shooter who becomes exceptional from the three-point line has done it thousands of times off the court, and has replayed the successes afterwards.

Masters also watch the margins of their domain. Deep foam matting as a landing surface was a marginal development for most high jumpers and their coaches. For Dick Fosbury it afforded a jump that did not have to be landed on the feet, and his gold medal in 1968 changed the event permanently. B. B. King was among the first to play the electric guitar under a perspective in which its capacity to sustain a note, the thing that distinguishes it from an acoustic guitar, was the central fact, so that the instrument could be made to sing. Billy Cobham was among the first to compose for two bass drums with the interplay between them as the centre of the pattern rather than a flourish in passing, and the shuffle that carries his name has been copied across rock, metal and jazz. When withholding pays off in this way, the reward makes the new perspective and its action likely to recur unbidden, and what the authors call enhanced expertise results. Mastery, on their account, is the point at which a performer stops only inhabiting the possibilities of a domain and starts disclosing new ones.

Since 1980

The model travelled fastest through nursing. Patricia Benner applied the five stages to clinical practice in From Novice to Expert in 1984, and the interviews she and her colleagues later published supply this chapter with its nurses who see a crisis coming before the instruments do. From nursing the stages passed into medical education, teaching, social work and software, usually as rungs in a competency ladder to be assessed against checklists, which is a calculative use of a model whose point is that calculation runs out.

The strongest opposition came from Anders Ericsson's work on deliberate practice, which holds that expert performance is not automatic but sustained by continued effortful control, and from philosophers working on the same evidence: Barbara Montero argues from dancers and musicians that thinking during performance does not wreck it, Simon Høffding argues from a string quartet that absorption comes in several varieties, and Christensen, Sutton and McIlwain propose that cognitive and automatic control are meshed rather than successive. Fernand Gobet objects that the model rests on description of experience where a science of expertise needs measurement. The reply offered here is the conditional, and the claim that the vacillation these critics point to was always in the model.

What this restatement adds is the machinery between the stages: the distinction between calculative and perspectival deliberation, reflexive reorientation as a non-deliberative switch of perspective, and mastery brought back from the 1980 report and given a mechanism. The judgment the authors have not softened in forty years is about teaching. A course of instruction can carry a learner to competence and no further, because everything beyond competence is stored as experience rather than stated as rules; the mentor's remaining work is to name aspects by example, to keep the learner from drowning at the advanced-beginner stage, and to point out and celebrate the moments when a perspective arrives unasked, so that the learner begins to trust it. A literature that stops at competence, they say, leaves behind an overly rationalised picture of what skilled people do.