Monday, 8 November 2010

Clinical judgement, tacit knowledge and recognition in psychiatric diagnosis

Very much a work in progress, obviously, but this is my working up a chapter for Bill Fulford's OUP Handbook of the Philosophy of Psychiatry.

Clinical judgement, tacit knowledge and recognition in psychiatric diagnosis

This chapter contrasts the recent emphasis on operationalism as the route to reliability in psychiatry with arguments for an ineliminable role for tacit knowledge. Although Michael Polanyi is widely credited with arguing for the presence of a tacit dimension, I argue that his account needs to augmented with Wittgenstein’s version of a regress argument for the centrality of know-how over knowledge-that. That argument, however, suggests two possible roles for tacit knowledge. I argue that the more radical version presupposes the very platonic picture that the argument is supposed to oppose. The more modest version presents a tacit element as a kind of limit condition on knowledge.


In this chapter, I will examine the role of clinical judgement in the recognition of psychiatric symptoms via the idea that this involves an ineliminable tacit dimension. I will do this by examining, at breakneck speed, three authors who offer support for the idea.

One reason for doubting, or playing down, a role for tacit knowledge in psychiatric diagnosis is the influence of operationalism in a quest for reliability for the last 50 years or so. There were two main factors which explain this.

Firstly, on its foundation in 1945, the World Health Organisation set about establishing an International Classification of Diseases (ICD). Whilst the chapters of the classification dealing with physical illnesses were well received, the psychiatric section was not widely adopted and so the British psychiatrist Erwin Stengel was asked to propose a basis for a more acceptable classification. Stengel chaired a session at an American Psychological Association conference of 1959 at which the philosopher Carl Hempel spoke. As a result of Hempel’s paper (and an intervention by the UK psychiatrist Sir Aubrey Lewis) Stengel proposed attempts at a classification based on theories of the causes of mental disorder should be given up (because such theories were premature) and to rely instead on what could be directly observed, that is, symptoms.

In fact, Hempel’s paper provided only partial support for the moral that was actually drawn for psychiatry. He argued that:
Broadly speaking, the vocabulary of science has two basic functions: first, to permit an adequate description of the things and events that are the objects of scientific investigation; second, to permit the establishment of general laws or theories by means of which particular events may be explained and predicted and thus scientifically understood; for to understand a phenomenon scientifically is to show that it occurs in accordance with general laws or theoretical principles. [Hempel 1994: 317]

These two requirements – that terms employed in classifications should have clear, public criteria of application and should lend themselves to the formulation of general laws – correspond to the aims of reliability and validity respectively. But it was the former that was adopted by psychiatry as the key aim at the time. With respect to it, Hempel claims that
Science aims at knowledge that is objective in the sense of being intersubjectively certifiable, independently of individual opinion or preference, on the basis of data obtainable by suitable experiments or observations. This requires that the terms used in formulating scientific statements have clearly specified meanings and be understood in the same sense by all those who use them. [ibid: 318]

He commends the use of operational definitions (following Bridgman’s book The Logic of Modern Physics [Bridgman 1927]), although he emphasises that in psychiatry the kind of measurement operations in terms of which concepts would be defined would have to be construed loosely and this view has been influential up to the present WHO psychiatric taxonomy in ICD-10.

The second reason for the emphasis on reliability and hence operationalism was a parallel influence from within American psychiatry that shaped the writing of DSM-III. Whilst DSM-I and DSM-II had drawn heavily on psychoanalytic theoretical terms, the committee charged with drawing up DSM-III drew on the work of a group of psychiatrists from Washington University of St Louis. Responding in part to research that had revealed significant differences in diagnostic practices between different psychiatrists, the ‘St Louis group’, led by John Feighner, published operationalised criteria for psychiatric diagnosis. The DSM-III task force replaced reference to Freudian aetiological theory with more observational criteria. The task force leader, Robert Spitzer, later reported: ‘With its intellectual roots in St. Louis instead of Vienna, and with its intellectual inspiration drawn from Kraepelin, not Freud, the task force was viewed from the outset as unsympathetic to the interests of those whose theory and practice derived from the psychoanalytic tradition.’ [Bayer and Spitzer 1985: 188 quoted in Shorter 1997: 301-2].

This stress on operationalism has had an effect on the way that criteriological diagnosis is codified in DSM and ICD manuals. Syndromes are described and characterised in terms of disjunctions and conjunctions of symptoms. The symptoms are described in ways influenced by operationalism and with as little aetiological theory as possible. (That they are neither strictly operationally defined nor strictly aetiologically theory free is not relevant here.) Thus one can think of such a manual as providing guidance for, or a justification of, a diagnosis offered by saying that a subject is suffering from a specific syndrome. Thus, presented with an individual, the diagnosis of a specific syndrome is justified because he or she has enough of the relevant symptoms which can be, as closely as possible, ‘read off’ the subject. Such an approach to psychiatric diagnosis plays down the role of individual judgement or tacit knowledge amongst clinicians.

Polanyi on tacit knowledge
Whilst the influence of operationalism deployed in the service of reliability aims to remove or reduce the presence of judgement and thus an uncodified tacit element in psychiatric diagnosis, there is a tradition in the history and philosophy of science (dating from about the same time) which stresses an ineliminable role for tacit knowledge in science. In this section, I will examine arguments offered by the chemist turned philosopher of science Michael Polanyi.

But first, what does Polanyi mean by ‘tacit’ knowledge? A clear statement runs thus:
I shall reconsider human knowledge by starting from the fact that we can know more than we can tell. This fact seems obvious enough; but it is not easy to say exactly what it means. Take an example. We know a person’s face, and can recognize it among a thousand, indeed among a million. Yet we usually cannot tell how we recognize a face we know. So most of this knowledge cannot be put into words. [Polanyi 1967: 4]
The suggestion is that tacit knowledge is tacit because it is ‘more than we can tell’. We cannot tell how we know things that we know tacitly. But what argument does he give for this? What are the limits on what can be said still leaving something that can be known?

In Personal Knowledge, Polanyi’s strategy is to examine how what can be said or, more broadly, articulated both leaves room for and depends on something outside what can be articulated. There are two key arguments of relevance to this chapter. One depends on limits on the kind of representation available to summarise explicit knowledge in science, thus indicating a space for tacit knowledge. The other depends on an analysis of what is involved in recognition (an argument which promises to impact on diagnostic judgement), which also connects to Polanyi’s views of how linguistic representation in general is possible. I will suggest that this latter argument is the fundamental argument but start with the former.

To examine the limits of scientific representation, Polanyi considers the understanding that a skilled surgeon has of the spatial configuration and orientation of organs in the body. He argues that this cannot be captured in a representation.
The major difficulty in the understanding, and hence in the teaching of anatomy, arises in respect to the intricate three-dimensional network of organs closely packed inside the body, of which no diagram can give an adequate representation. Even dissection, which lays bare a region and its organs by removing the parts overlaying it, does not demonstrate more than one aspect of that region. It is left to the imagination to reconstruct from such experience the three-dimensional picture of the exposed area as it existed in the unopened body, and to explore mentally its connections with adjoining unexposed areas around it and below it.
The kind of topographic knowledge which an experienced surgeon possess of the regions on which he operates is therefore ineffable knowledge. [Polanyi 1962: 89]

The claim here is that three-dimensional spatial knowledge is ineffable, or tacit, because it cannot be captured in a representation. Polanyi goes on to argue that even if all human bodies were identical and even if there were a map comprising cross sections based on ‘a thousand thin slices’ of the body, that in itself would not articulate the knowledge of a trained surgeon. Someone knowing merely the former ‘would know a set of data which fully determine the spatial arrangement of the organs in the body; yet he would not know that spatial arrangement itself’ [89]. An additional act of interpretation or imagination is needed. But because that act cannot itself be encoded in a representation, according to Polanyi, it remains tacit.

This argument is a little surprising. Polanyi concedes that the set of cross sectional representations, presumably alongside some further information about their inter-relations such as their order and distance apart, ‘fully determine[s] the spatial arrangement of the organs’ and yet denies that this amounts to an articulation of the three-dimensional understanding.

Without the further information about the relations between the set of maps, the maps alone would not be an articulation of the skilled surgeon’s knowledge. But then neither would they fully determine the arrangement of bodily organs. With that addition, however, why would this not count as an articulation of their knowledge and thus imply that it could be explicit knowledge?

A further possible clue to Polanyi’s thinking runs thus:
The difficulty lies here entirely in the subsequent integration of the particulars and the inadequacy of articulation consists altogether in the fact that the latter process is left without formal guidance. The degree of intelligence required from the student to perform the act of insight which ultimately conveys to him the knowledge of the topography, offers here a measure of the limitations of the articulation representing this topography. [ibid: 90]

But there remains something strange about this line of thought. If the integration of the partial representations, such as the set of cross sections, were left without formal guidance then it would be clear why the partial representations could not articulate the surgeon’s knowledge. But neither would they determine the arrangement of organs as Polanyi has previously asserted.

The difficulty with interpreting this argument is that of balancing the claim that spatial configuration is both determined by what can be represented but remains ineffable and thus tacit rather than explicit.
I think that the clue to its interpretation is to realise that whether a symbol logically determines anything always, according to Polanyi, depends on a tacit element. This is supported by a different argument.
I may ride a bicycle and say nothing, or pick out my macintosh among twenty others and say nothing. Though I cannot say clearly how I ride a bicycle nor how I recognise my macintosh (for I don’t know it clearly), yet this will not prevent me from saying that I know how to ride a bicycle and how to recognise my macintosh. For I know that I know how to do such things, though I know the particulars of what I know only in an instrumental manner and am focally quite ignorant of them. [ibid: 88]

Polanyi suggests that the skill involved in the example of recognising a macintosh is akin to the practical skill of cycle riding. In both cases, the ‘knowledge-how’ depends on something which is not explicit: the details of the act of bike riding or raincoat recognition. Whilst one can recognise one’s own macintosh, in the example, one is ignorant of how. Thus how one does this is tacit.

If this argument were successful it would be of general significance because it would carry over to the recognitional skill which underpins classification such as diagnosis in psychiatry but also lingusitic labelling generally. Indeed, Polanyi makes this connection explicilty.
[I]n all applications of a formalism to experience there is an indeterminacy involved, which must be resolved by the observer on the ground of unspecified criteria. Now we may say further that the process of applying language to things is also necessarily unformalized: that it is inarticulate. Denotation, then, is an art, and whatever we say about things assumes our endorsement of our own skill in practising this art. [ibid: 81]

This connection between denotation and tacit recognitional skills appears to be the fundamental argument for the importance of tacit knowledge for explicit scientific accounts. Polanyi summarises the connection thus:
If, as it would seem, the meaning of all our utterances is determined to an important extent by a skilful act of our own – the act of knowing – then the acceptance of any of our own utterances as true involves our approval of our own skill. To affirm anything implies, then, to this extent an appraisal of our own art of knowing, and the establishment of truth becomes decisively dependent on a set of personal criteria of our own which cannot be formally defined.... [E]very where it is the inarticulate which has the last word, unspoken and yet decisive... [ibid: 70-71]

Nevertheless, this argument faces two objections.
First, it involves the claim that recognition of particulars – such as a particular macintosh – depends on features of them of which one is ignorant. The example concerns the recognition of a particular macintosh among a pile of them. Not the recognition that the object is a macintosh, but that that particular macintosh is mine. Now it may be true that the recognition of a particular macintosh (as mine, eg.) turns on features of which I am ignorant (although see my second criticism below) but that is not quite the claim Polanyi needs if he wishes to make a point about the ‘art of denotation’ in general.

For that, he needs the claim that recognition of something as an instance of the type ‘macintosh’ is based on the recognition of subsidiary properties of which one is ignorant. But it is not clear that any such claim is true. Unless at least some properties were directly recognisable, it would lead to a vicious regress. But if some properties were directly recognisable the argument would not establish the generality of a tacit dimension. Some recognitional judgements would depend on tacit elements but not all.

Second, it is not clear that it is right to say that to even in cases where one recognises an F in virtue of its subsidiary properties G, H, I of which one cannot give an independent account, or recognises one particular F (as mine, eg.) on such a bais, that one is ignorant of those properties. It may be, instead, that the awareness one has of G, H, I is manifested in the recognition of something as an F or a particular F. One might say, I recognise that it is my macintosh (or a macintosh rather than any other kind of raincoat) because of how it looks here with the interplay of sleeve, shoulder and colour even if one could not recognise a separated sleeve, shoulder or paint colour sample as of the same type. So it is not clear that Polanyi’s is a convincing account of the tacit properties of recognition even when it does depend on component aspects.

The tacit element in recognition

I argued in the previous section that Polanyi’s argument for the role of a tacit element in science turns on an argument that it is fundamental in recognition, including the recognition which underpins the ‘art of denotation’. He suggests that tacit knowledge is that which falls outside linguistic articulation or representation. But such representation presupposes recognitional know-how rather than the other way round. So the know-how which constitutes recognition is not itself articulated but rather tacit.

I argued, however, that Polanyi’s argument for the role of tacit knowledge in recognition is not successful as a general account because he has to assume that to recognise a feature (F, say) one must a) always recognise it in virtue of something else (subsidiary features G, H and I, eg.) of which b) one is ignorant. But neither point – a) and b) – is compelling. There is, however, another argument which was framed at about the same time as Polanyi’s which turns on an idea that recognition of something might resist being put into words (which Polanyi takes to indicate tacit-status). This is Ryle’s argument that knowledge-how is a concept logically prior to the concept of knowledge-that.

Ryle’s argument takes the form of a regress.
If a deed, to be intelligent, has to be guided by the consideration of a regulative proposition, the gap between that consideration and the practical application of the regulation has to be bridged by some go-between process which cannot by the pre-supposed definition itself be an exercise of intelligence and cannot, by definition, be the resultant deed. This go-between application- process has somehow to marry observance of a contemplated maxim with the enforcement of behaviour. So it has to unite in itself the allegedly incompatible properties of being kith to theory and kin to practice, else it could not be the applying of the one in the other. For, unlike theory, it must be able to influence action, and, unlike impulses, it must be amenable to regulative propositions. Consistency requires, therefore, that this schizophrenic broker must again be subdivided into one bit which contemplates but does not execute, one which executes but does not contemplate and a third which reconciles these irreconcilables. And so on for ever. [Ryle 1945: 2]

There has been a recent flurry of literature on the precise nature of this argument and thus whether it is a successful refutation of intellectualism [eg Stanley and Williamson 2001; Noë 2005]. But it seems to involve something like the following regress:
Suppose all know-how can be articulated (put into words) as a piece of knowledge-that: grasping some proposition that p. Grasping the proposition that p is itself something one can do successfully or unsuccessfully, so it is also a piece of know-how. So, on the theory in question, it will involve grasping another proposition, call this q. But grasping the proposition that q is itself something one can do successfully or unsuccessfully, so it is also a piece of know-how. So, on the theory in question, it will involve grasping another proposition, call this r.... etc

This argument seems to suggest that, if the first step of the argument is designed to ‘articulate’ or represent a piece of otherwise merely tacit knowledge at the heart of recognition, it will lead to a regress. But that may prompt this thought, contra Polanyi. Suppose that one recognises a macintosh as a macintosh and thus denotes it ‘macintosh’. Why is that not what articulating this piece of recognitional knowledge amounts to, thus discharging any tacit element? Why is that not putting all the relevant knowledge in play into words?
One gets a clearer sense of the attraction of Polanyi’s position by seeing both what has to be ruled out in successful recognition and on what limited basis it seems to be known by looking to Wittgenstein’s version of a regress argument (also from about the same time). Wittgenstein considers what is grasped by someone who has grasped a mathematical rule or series which he approaches via the idea of teaching the +2 series.

Now we get the pupil to continue a series (say + 2) beyond 1000–and he writes 1000, 1004, 1008, 1012.
We say to him: “Look what you’ve done!”–He doesn’t understand. We say: “You were meant to add two: look how you began the series!”–He answers: “Yes, isn’t it right? I thought that was how I was meant to do it.”–Or suppose he pointed to the series and said: “But I went on in the same way.”–It would now be no use to say: “But can’t you see....?”–and repeat the old examples and explanations.–In such a case we might say, perhaps: It comes natural to this person to understand our order with our explanations as we should understand the order: “Add 2 up to 1000, 4 up to 2000, 6 up to 3000 and so on.” [Wittgenstein 1953: §185]

In passages such as these, Wittgenstein seems to stress the infinite possibilities of divergence and thus the infinite possibilities of a breakdown of communication. Given his concurrent criticisms, in surrounding passages of the Philosophical Investigations, of physiological mechanisms, mental talismans and platonic structures to explain our grasp of going on in the correct way it can seem the most fragile contingency that communication is possible [cf Lear 1982]. Since teaching rules (whether mathematical or empirical) is possible only either by paraphrase, which merely postpones the problem of explaining the paraphrase, or by finite examples, which seem to underdetermine the rule, grasp of a rule which governs a potentially unlimited number of cases seems to need some further helping hand.

This line of thought provides a role for tacit knowledge. Since everything that can be said still allows for the kind of misunderstanding exemplified by Wittgenstein’s hypothetical deviant pupil, the grasp of a rule that a normal pupil acquires must be based on something unsaid and implicit. It must depend on a tacit element. This seems also to fit Wittgenstein’s own conclusion:
What this shews is that there is a way of grasping a rule which is not an interpretation, but which is exhibited in what we call “obeying the rule” and “going against it” in actual cases. [Wittgenstein 1953: §202]

But whilst I think that there is something right about that line of thought, there is something misleading about it. The risk is that it accepts part of what Wittgenstein criticises: a platonic picture of rules as rails ‘invisibly laid to infinity’ fundamentally distinct from our capacity to articulate them.

That picture is easily prompted by cases of deviant pupils. What they seem to show is both that any finite set of examples underdetermines a correct understanding of the rule but also that such correct understanding must involve grasp of an extra-human or supernatural pattern. Since no actual human enumeration of the pattern seems enough to determine it, it must be extra-human. Hence the metaphor of rails laid to infinity.
With that picture of the way rules determine correct moves in place, there is a substantial role for tacit knowledge to bridge the gap between what can be made explicit in the sublunary realm and the ideal platonic standard. Wittgenstein undermines that gap, however, in passages in which he suggests that there is a close connection between what a teacher can express and what a student can grasp in the examples which manifest the teacher’s meaning.

“But do you really explain to the other person what you yourself understand? Don't you get him to guess the essential thing? You give him examples,--but he has to guess their drift, to guess your intention.” – Every explanation which I can give myself I give to him too. – “He guesses what I intend” would mean: various interpretations of my explanation come to his mind, and he lights on one of them. So in this case he could ask; and I could and should answer him. [ibid: §210]
“But this initial segment of a series obviously admitted of various interpretations (e.g. by means of algebraic expressions) and so you must first have chosen one such interpretation.”–Not at all. A doubt was possible in certain circumstances. But that is not to say that I did doubt, or even could doubt…[ibid: §213]

Whilst Wittgenstein rejects substantive explanations of our grasp of rules by pointing out they could not determine what they are supposed to, he does not promote a kind of sceptical gap between what can be manifested and what must be understood for communication, a gap that has thus to be filled by a tacit element. So the role for tacit knowledge in the recognition is not quite that.

The real role for a tacit element in recognition of sameness and difference is nuanced. It is not that it plugs a gap between what can be expressed and a platonic standard, a gap that would downplay the expressive capacities of explanations by example. Recognising that our understanding can be expressed in examples, for those with eyes to see at least, undermines the gap between the sublunary and the platonic. Nevertheless, the failure of the kind of platonic explanation of what it is to follow a rule suggests an important role for human judgement. It is because we are the kind of subjects that we are, with our shared routes of interest, perceptions of salience, feelings of naturalness etc., that we are able to use finite explanations, as we do, to communicate unending rules [Lear 1982: 386]. Against that background, the rules one grasps can be manifested in explanations. Wittgenstein’s deviant pupil thus illustrates what it would be to lack the right background and thus fail to understand explanations by example as we do.

Now it might be tempting to obviate the need for such a background by trying to articulate or encode how understanding depends on one’s interests, saliences and perceptions of naturalness. But any attempt to articulate that would also have to presuppose the background that underpins understanding of any such explanation.

This suggests a nuanced view of the role for tacit knowledge in rule-governed recognition. There is no gap between explanation and rule to be plugged by tacit knowledge, as would be suggested by partial adoption of a platonic picture. Explanations can manifest one’s understanding of a rule and thus it is not tacit in Polanyi’s sense. But such explanations presuppose a background which cannot be fully articulated. Whatever can be expressed, something remains implicit.

Tacit knowledge, clinical judgement recognition and psychiatric symptoms

I began by outlining the importance of reliability and thus the stress on operationalism for psychiatric taxonomy and diagnosis. For the last half century, syndromes have been defined in terms of conjunctions and disjunctions of symptoms which have themselves been described in terms as free of aetiology as practical. (For some conditions, such as PTSD, that would be impossible.) Thus the diagnosis of a specific syndrome is justified in the presence of a particular subject if he or she has enough of the relevant symptoms.

This approach is aimed at closing the gap between syndrome and subject and thus increasing inter-rater reliability among clinicians. Because of the way both ICD and DSM base syndromes on a combination of conjunction and disjunction of symptoms, it is possible that a syndrome so defined may apply to two individuals with little, or even no, overlap of symptoms. But these differences are codified rather than left to an overall uncodified clinical judgement. Further, the heritage of operationalism suggests that individual symptoms can be tied to subjects through a kind of measuring operation.

There remains, however, a gap between the description or articulation of a symptom and an individual. The concepts of specific symptoms are, despite their specificity, general concepts that can be instantiated in an unlimited number of actual or potential cases. So how can one judge that a general concept applies to a specific individual case or individual person? How can one recognise that they exemplify a type? One can attempt to bridge this gap. Textbooks of psychiatry can describe, rather than merely list, symptoms. But whatever descriptive account they give of symptoms, there will always be a gap between their general descriptions and concepts (which potentially apply to any number of individuals) and any particular individual. Bridging this gap calls for expertise which can be called tacit in the nuanced way set out above. It calls for a skilled recognitional clinical judgement.


Bayer, R. and Spitzer, R.L. (1985) ‘Neurosis, psychodynamics and DSM-III’ Archives of General Psychiatry 42: 187-96
Bridgman, P.W. (1927) The Logic of Modern Physics, New York: Macmillan
Hempel, C.G. (1994) ‘Fundamentals of taxonomy’ in Sadler, J.S. Wiggins, O.P. and Schwartz, M.A. (eds) Philosophical Perspectives on Psychiatric Diagnostic Classification, Baltimore: Johns Hopkins: 315-331
Lear, J. (1982) ‘Leaving the world alone’ Journal of Philosophy 79: 382-403
Noë, A. (2005) ‘Against intellectualism’ Analysis 65: 278-90
Polanyi, M. (1962) Personal Knowledge, Chicago: University of Chicago Press
Polanyi, M. (1967) The Tacit Dimension, Chicago: University of Chicago Press
Ryle, G. (1949) The Concept of Mind, London: Hutchinson.
Ryle, G (1945) ‘Knowing How and Knowing That’ Proceedings of the Aristotelian Society, 46: 1-16
Shorter, E. (1997) A History of Psychiatry, New York: John Wiley and Sons
Stanley, J. and Williamson, T. (2001) ‘Knowing how’ The Journal of Philosophy 97: 411-444.
Wittgenstein, L. (1953) Philosophical Investigations, Oxford: Blackwell.