76The Reflective Practitioner

Donald A. Schön, The Reflective Practitioner: How Professionals Think in Action, Basic Books, 1983

Linked to 18 chapters: supports 5 · contradicts 6 · mechanism 6 · same claim 5 · pairing 3

Donald Schön finished this book at MIT in 1982, at the end of two decades in which the American professions had gone from being celebrated as the engine of national progress to being blamed for a war, a run of failed social programmes and a series of scandals. His diagnosis is not a moral one. The professions are in trouble, he argues, because the account they give of their own knowledge is wrong, and wrong in a way that hides from them the thing they are best at. That account, which he calls Technical Rationality, holds that professional work is instrumental problem solving made rigorous by the application of scientific theory and technique. It cannot describe what a competent practitioner does in the situations that matter most, which are complex, uncertain, unstable, unique and torn between ends. Schön's alternative begins from the observation that skilled people know more than they can say, and that when action produces a surprise they think about what they are doing while doing it. His evidence is a twenty-minute conversation between an architecture teacher and his student, transcribed and read line by line, out of which he draws a structure: a frame imposed on a situation, a web of moves that tests it, a situation that answers back, and an appreciation that the answer revises.

The crisis of confidence in professional knowledge

In 1963 Daedalus, the journal of the American Academy of Arts and Sciences, gave an issue to the professions and opened it by declaring them triumphant everywhere in American life. The evidence was growth: professionals had risen from four percent of the labour force in 1900 to thirteen percent in 1966, and Daniel Bell expected twenty-five percent by the year 2000. Even the complaints were complaints of success: the overtaxed physician, the proliferating specialties, the burgeoning mass of legal data to be assimilated. Only William Alonso struck another note, writing of cities that had outgrown the tools of the planners who served them.

By 1982, Schön writes, no profession would celebrate itself in those terms. A professionally managed war had been widely seen as a national disaster, and the announced national crises — the deteriorating cities, poverty, pollution, the energy shortage — seemed rooted in the practices called on to relieve them. Urban renewal had become a destroyer of neighbourhoods, and Medicare fraud, Watergate and Three Mile Island did the rest. Professionals, Charles Reich wrote of the Bureau of Reclamation, could be counted on to do their job but not to define it.

When leading professionals diagnosed their own trouble they said one thing in several vocabularies: that professional knowledge is mismatched to the situations in which it has to be used. Erik Erikson described each patient as a universe of one, and a physician told Schön that eighty-five percent of what walks into an office is not in the book. Russell Ackoff announced to his fellow operations researchers that the future of their field was past, because managers are not handed independent problems but dynamic systems of interacting ones; problems, he said, are abstractions extracted from messes by analysis, and "Managers do not solve problems: they manage messes." Complexity, uncertainty, instability, uniqueness and conflict of values are the five terms that recur through the book as the features technique does not reach.

Schön then turns the self-criticism around, and the rest of the book depends on the turn. If practice had really been defeated by those five features the complaint would make sense, but it has not: some engineers and managers do make sense of complexity and reduce uncertainty to manageable risk, and some physicians find the problem of a particular patient in ways that go past the boundaries of diagnosis. The distress is therefore not distress at failure but at being unable to account for a competence that can be seen. Problem setting has no place in a body of knowledge concerned only with problem solving, and an artful handling of the unique case looks anomalous when competence is modelled as the application of technique to recurrent events. What unsettles the professions is an epistemology of practice that leaves them unable even to describe the abilities they value most.

From Technical Rationality to reflection-in-action

The model has a clear shape. Edgar Schein set it out as three components: an underlying discipline or basic science; an applied science yielding diagnostic and problem-solving procedures; and a skills and attitudinal component covering the delivery of service to the client. The order of application is also an order of derivation, and also a ranking, since those who make theory stand above those who apply it. Schooling follows the same order, the science first and the practicum later: since the Flexner Report the first two years of medical school have gone to chemistry, physiology and pathology. Law took the shape too, Christopher Columbus Langdell telling the Harvard Law School in 1886 that law is a science and that all the materials of that science are contained in printed books.

Nathan Glazer had drawn the consequence sharply. Major professions such as medicine and law are disciplined by an unambiguous end — health, success in litigation, profit — and so can ground themselves in systematic knowledge; minor ones such as social work, divinity, education and town planning have shifting ends and unstable settings, and so cannot. Schön objects not to the picture but to the direction of the diagnosis. The origin of the arrangement is Positivism, a nineteenth-century account of the rise of science and a programme for extending its control to human affairs, institutionalised in the American research university just as the professional schools were entering it. Thorstein Veblen fought to keep them out and lost; the professions got their place and paid for it by accepting an epistemology already built into the tissue of the place.

The trouble is that this account of practice begins one step too late. Problems do not present themselves as givens; they have to be constructed out of situations that are puzzling and uncertain, and that construction is not itself a technical operation. When professionals decide what road to build they work in a tangle of geographic, financial, economic and political matter; how to build it, once decided, may be answerable by technique; and when the road destroys a neighbourhood they are back in uncertainty. To set the problem is choosing what to treat as the things of the situation, drawing the boundary of attention, and imposing a coherence that allows one to say what is wrong and in which direction the situation should change. Schön names the two halves: we name the things we will attend to and frame the context in which we will attend to them.

Even a well-constructed problem can escape applied science by being unique or unstable. A physician who recognises measles can map it onto a settled scheme of diagnosis and treatment; a case that is not in the books falls outside the categories, and a situation that changes slips out from under them. Schön puts the resulting choice as a piece of topography. There is a high, hard ground where research-based theory and technique work well, and a swampy lowland where situations are confusing messes incapable of technical solution. The problems of the high ground are often of little importance to clients or to society; the problems of greatest human concern are in the swamp.

Those who stay on the high ground keep it dry by cutting the situation to fit the knowledge: growing inattentive to data outside their categories, blaming a failed recommendation on personality or on politics, filing a client who resists help as a problem tenant. Schein, Glazer and Herbert Simon each saw the gap and each closed it in a way that leaves the model standing. Simon comes closest, proposing a science of design to teach what the schools have abdicated; but his science works on optimisation problems, in which ends have already become constraints and utility functions — exactly what has to be constructed out of a mess before anything can begin.

Knowing-in-action

So Schön stands the question on its head and looks for the epistemology implicit in what competent practitioners do. The first half of it is that knowing is often in the action itself. Gilbert Ryle had argued that intelligent cannot be defined in terms of intellectual, and that a person doing something intelligently is doing one thing and not two. Michael Polanyi, who named tacit knowing, points out that in learning to use a stick we stop attending to its pressure on the hand and attend from that pressure to the thing it touches. Geoffrey Vickers draws the general point: we recognise and describe deviations from a norm much more clearly than we describe the norm.

Knowing-in-action names actions and judgments carried out spontaneously, often without memory of having learned them, and usually beyond the actor's power to describe. Every competent practitioner recognises families of symptoms and irregularities in materials for which he can give no accurate description, and makes judgments of quality for which he can state no criteria.

Reflecting-in-action

The second half is that people also think about what they are doing while doing it, and that surprise is what starts them. Jazz musicians improvising together work inside a metric, melodic and harmonic scheme they all know and a repertoire of figures each has ready; as they hear where the music is going they make new sense of it and adjust, reflecting through a feel for the music rather than in words.

Schön's clearest case is an experiment by Bärbel Inhelder and Annette Karmiloff-Smith, who asked children to balance wooden blocks on a metal bar, some plain and some conspicuously or inconspicuously weighted at one end. Nearly every child of six or seven began by trying every block at its geometric centre, and slightly older ones, asked to add smaller blocks to a block already balanced, piled up to ten of them precariously at the centre rather than distributing them at the ends. The authors attribute to the children a theory-in-action: things always balance in the middle. When the counterweighted blocks would not balance there, some children did the same thing harder and finally declared the block impossible. Others, generally between seven and eight, began to shift the blocks off centre, first the conspicuously weighted ones and then the rest, pausing in the middle of their action sequences; at the end they hefted each block, inferred the probable point of balance and placed it there directly. They had exchanged a theory of the geometric centre for a theory of the centre of gravity, and with it an orientation towards success for an orientation towards the blocks.

Reflection-in-action is not rapid by definition. It is bounded by what Schön calls the action-present, the stretch of time in which action can still make a difference, running from seconds in the give-and-take of a courtroom to months in an antitrust case. A practice is a set of case types met again and again; its benefit is that knowing-in-practice becomes tacit and automatic, and its cost is that a practitioner may overlearn what he knows and be drawn into patterns of error he cannot correct, like the younger children with the blocks.

Reflection is the corrective. In a programme for serving teachers at MIT, a group watched a videotape of two boys failing to reproduce a block pattern across a screen, and described a boy unable to follow directions. Told that the instruction-giver had asked for a green square when every square was orange and only triangles were green, they watched again, saw a chain of reasonable moves following from one impossible instruction, and began to speak of giving the child reason.

Design as a reflective conversation with the situation

The case that carries the book comes from a study of architectural education Schön joined in the late 1970s, funded by the Mellon Foundation and directed by the deans of the MIT and Harvard design schools. Participant observers sat in design studios; one of them, Roger Simmonds, recorded the protocol Schön uses and named its participants Quist and Petra. Quist is the studio master; Petra is one of twenty students given a programme for an elementary school and a description of its site. The review runs twenty minutes, early in a semester.

Petra is stuck, and says so. She has taken the contours of the land seriously and tried to butt the shape of her building into them, and the shape will not fit the slope. She has also found her six classroom units too small in scale to do much with, so she has regrouped them into three larger L-shapes. Quist looks at the drawings, lays a sheet of tracing paper over them, and begins to draw and talk at once. The site, he tells her, is too screwy to give an order to the building. The order has to come the other way: "You should begin with a discipline, even if it is arbitrary" — and the discipline can always be broken open later.

What follows works the reframing out. The geometry of parallels that the reshaped classrooms suggest is carved into the slope. The total drop across the three classrooms is fifteen feet, so the slope can be taken in three levels with five-foot intervals; five feet is maximum height for a child, so the spaces under the roofs can be made into nooks. The gallery, which Petra had drawn as a general pass-through, is invited to become a soft back area to the hard classrooms. When Quist looks at what has accumulated he finds it has become, in a minor way, the major thing — an artifice of the sort Aalto would invent. From there the placement of the administration, which Petra had thought a solution to her access problem, is horrible, because it would ruin the whole idea.

Schön reads three things out of the protocol. The first is the language. Quist's words do not describe what is already on the paper; they run alongside the making of it, so that his here and this and that mean nothing without the moving pencil. He speaks a spatial action language, attributing actions to the elements — a cafeteria that could come down into here to catch summer sun, steps that relate in downward — while imagining the felt path of a person walking through the building. Schön groups the vocabulary into twelve normative domains and finds their frequency telling: organisation of space, scale and form dominate, precedent and structure appear once each, cost and building character never.

The second is implications. When Quist says the kindergarten might go over here, which might indicate the administration over there, he is running a literal logic of design, a pattern of if-then propositions binding earlier moves to later choices. The web branches, and even Quist cannot hold an indefinitely expanding one; at some point a what-if has to become a decision that binds. A move is judged three ways at once: by the desirability of its consequences in the terms of the design domains, by its conformity to the implications of earlier moves, and by the new problems and potentials it has brought into being.

The third is a shift of stance. Quist moves from what can and might happen to what should and must. He urges Petra to step into the problem and impose her own constructs on it, then holds her to what those constructs imply, oscillating between the unit and the total, and between involvement and detachment.

Schön then turns on his own case. Quist's demonstration is so masterful that it hides the process: he detects and corrects no errors of his own, settles at once on schemes that quickly become commitments, and compresses the iterations by which designers actually learn. That is probably because he has developed, as a chess master develops a feeling for a configuration of pieces, a feeling for the kind of conversation this situation will set in motion. But he reflects very little on his own reflection-in-action, and an observer could miss the structure of inquiry underneath the performance.

The structure of reflection-in-action

Beside the design review Schön sets a psychiatric supervision, in which a Supervisor and a Resident work over the case of a patient who keeps herself continually frustrated. The two professions could hardly be less alike, and that is why they are put together: the skeleton is the same. A student has set a problem and cannot solve it as set. The teacher criticises the framing implicitly, leaving the criticism to be inferred from the restructuring, and suggests a direction that requires the student to step into the situation — Petra by imposing her own order on the site, the Resident by treating his relationship with the patient as a microcosm of her life outside. The situation then comes to be understood through the attempt to change it, and changed through the attempt to understand it.

How is such an experiment judged, if not by effectiveness against a fixed objective? Schön extracts five questions from the protocols. Can I solve the problem I have set? Do I like what I get when I solve it? Have I made the situation coherent? Have I made it congruent with my fundamental values and theories? Have I kept inquiry moving? The practitioner cannot know at the moment of reframing whether the new problem is soluble, but he chooses a frame that suits a method he trusts, and affirms it when he values the unintended changes it produces. Quist did not set out to make nooks, and keeps them; he decides for himself that working slightly with the contours is enough.

The second question is how experience can be used in a case treated as unique. A rule drawn from the past would dissolve the uniqueness by making the case an instance of a class, and a description invented out of nothing is not a real possibility. Schön's answer is a repertoire: the examples, images, understandings and actions a practitioner has accumulated. Faced with an unfamiliar situation he sees it as something already in that repertoire — not subsuming it under a category but seeing it as both like and unlike a familiar case, without being able at first to say like in what respect. Thomas Kuhn had described the same thing in science, where the perception of similarity is prior to any criterion by which it might later be justified. Seeing-as brings doing-as with it, and Schön warns against crediting Quist, at the moment he calls the site screwy, with the account he articulates later.

The third question is whether any of this deserves the name experiment. Against the model in which experiment means confirming or refuting a hypothesis, everyday practice looks hopeless: the practitioner cannot shield his experiment from the environment, the situation may change out from under it, and the variables are locked together when he knows what they are. Schön's reply is that hypothesis testing is only one of three kinds. To experiment in the most general sense is to act in order to see what the action leads to, and the fundamental question is: what if? Acting without prediction, simply to see what follows, is exploratory experiment, and it succeeds by turning something up. Acting to produce an intended change is move-testing experiment, affirmed not only when it produces what was intended but when the consequences taken as a whole are liked — a chess player who checkmates by accident does not take the move back. Hypothesis testing is the third. What distinguishes practice is that the same action performs all three functions at once.

It follows that the practitioner's hypothesis testing looks scandalous by laboratory standards. He makes his hypothesis come true, acting as though it were in the imperative mood — let it be the case that X — and shaping the situation until X is true. He does not keep his distance from the object of study, because the unique situation before him is the object of study, and he seeks confirmation rather than refutation. What saves this from being self-fulfilling prophecy is that situations resist. Rigour in on-the-spot experiment therefore means striving to make the situation conform to one's view of it while staying open to the evidence of failing. What ends it is not the exhaustion of alternatives but appreciation: testing stops when the changes produced are found satisfactory, or when something new gives the situation a different meaning.

The full range of outcomes gives a small schema that decides when reflection is called for at all.

The last piece of the structure is that neither Quist nor the Supervisor is doing the real thing. Quist is not moving dirt; the Supervisor is not talking to the patient. Each works in a virtual world, a constructed representation of the world of practice, and the ability to build and use such worlds is what makes rigorous experiment possible at all. Drawing is rapid and spontaneous, but its traces are stable, the pace can be varied, no move is irreversible, and moves that would be costly on the site cost nothing on paper.

The price is that a virtual world works only insofar as its results transfer, and an architect learns by moving back and forth between drawing and building how his drawings will build. He learns what the medium cannot represent — the qualities of materials and surfaces, soil conditions, wind, the costs of labour — and has to hold those absences in mind while reading his results. Underneath all of it is a stance. Technical Rationality asks the practitioner to be a spectator and manipulator, holding a boundary between himself and an objectively knowable world. In a reflective conversation he is an agent and experient: he shapes the situation and makes himself part of it, while it keeps a life of its own that can foil his projects. Schön calls the combination double vision — acting in accordance with the view one has adopted while knowing it can be broken open later, and must be.

Since 1983

Schön spent the rest of his life on the consequences. Educating the Reflective Practitioner, in 1987, took the design studio from this book's third chapter and proposed it as the model for professional education generally: a practicum in which a coach demonstrates, the student does, and both reflect on the doing. The action science he calls for here he went on to build with Chris Argyris, extending a partnership that had already produced Theory in Practice in 1974. He died in 1997.

The uptake was enormous and lopsided. Teacher education and nursing took reflective practice up first and hardest, and from there it passed into medicine, social work and management, usually as a requirement: journals, portfolios, revalidation schemes in which practitioners must document reflection on their cases. That a book written against the reduction of practice to procedure should have been absorbed as a procedure is the standing irony of its reception.

The most careful criticism came from Michael Eraut, who argued in 1994 that the central term is underspecified: reflection-in-action and reflection-on-action are not always kept apart, and the action-present in most professional work is far too short for the deliberation Schön describes, so that what is being praised is either rapid intuition, which is not reflection, or after-the-fact review, which is not in action. A second objection concerns the evidence. The two central cases share a feature Schön does not control for: Quist and the Supervisor are both teaching. What was recorded is a demonstration of practice to a student, slowed down and narrated, and the structure so visible in it may belong to the teaching rather than to the practice.

What has held up is the front half of the argument. Problem setting, and the claim that framing a problem is prior to and different in kind from solving it, arrived at the same time as Horst Rittel and Melvin Webber's wicked problems, and the two together changed how planning and design describe themselves. The claim that what looks like intuition is structured, and can be described without being reduced to rules, was confirmed from directions Schön did not take. The Dreyfus skill model, first published three years before this book, reaches the same conclusion from the side of skill acquisition: rules are the scaffolding of the early stages and what has to be let go, and the expert's perception arrives with the response already attached. Gary Klein's Sources of Power supplied the field studies Schön's critics said he lacked, and found something close to his on-the-spot experiment in the mental simulation by which a fireground commander runs a course of action forward before committing to it.

The judgment Schön did not soften is the one about schooling. The order in which professional schools teach — the science first because it is the real knowledge, the art last or not at all because it is a secondary kind of knowing one hesitates to call knowledge — is not a scheduling convenience. It follows from the epistemology, and it leaves graduates equipped for the high ground and untrained for the swamp, where the work that matters is.