William T. Scott
Tacit Knowing and the Concept of Mind
Philosophical Quarterly, 21 (82)
Jan. 1971, 22-35.
I wish to show that Gilbert Ryle’s account of mental processes as given in The Concept of Mind  can be significantly extended by considering certain features of the philosophical position that Michael Polanyi has developed around the concept of tacit knowing.  The Polanyian themes I wish to employ are outlined in Section 1 and include the integrative and unitary perception of comprehensive entities, the distinction between subsidiary and focal awareness and the consequent “from-to” relation, the hierarchy of levels in the experienced world, the notion of indwelling and the variability of the boundary between self and world that follows from this notion, and the indefinite nature and function of anticipatory imagination.
In Sections 2 to 6 I shall apply Polanyian conceptions to a number of problems arising from Ryle’s book, including the coherence of the indefinite variety of instances of a particular disposition, the relation between sensing and observation, the relation between mind and observed activities to which mental predicates can be applied, the nature of our ability to recognize a candidate-idea as a solution to a problem, and the confidence with which we can describe mental life in the Rylean manner.
I shall begin with a brief account of the tacit dimension of knowledge in terms of my own understanding of Polanyi’s work. Polanyi uses the term ‘tacit knowledge’ to refer to the kind of things which cannot be made explicit in speech - things we know and know that we know, but cannot tell. For example, we know how to walk or ride a bicycle but we cannot tell the particulars of co-ordination and control by which we carry out these muscular activities.  A wide variety of examples of this tacit property of “knowing how” can be given; Ryle himself has several, such as that of the humorist, who knows how to make good jokes but cannot give any recipes for them.  Most, if not all, cases of “knowing how” involve tacit knowledge.
Polanyi shows that tacit elements are also included in “knowings that” or at least underlie them, in spite of the fact that “knowing that” is commonly taken to refer only to propositional knowledge that can by definition be articulated. There are, in the first place, propositions held by persons
1. London, 1949; hereafter referred to as CM.
2. M. Polanyi, The Tacit Dimension (New York, 1963).
3. M. Polanyi, Personal Knowledge: Towards a Post-Critical Philosophy (London, 1958; reprinted New York, 1964), pp. 49-52.
4. CM, p. 30.
who have not themselves articulated these propositions. The success of philosophers in clarifying thought in a wide range of disciplines is a measure of the great extent of propositional knowledge which is held partially or wholly in tacit form. But there are also non-propositional elements that cannot be fully expressed in much that we know to be the case. An example from Ryle’s book is that of an actor knowing the moods of another and being able to act them out, without being able to tell, either at the time or later, what it is that he knows to be the case about the other person.  Knowing how to recognize your face entails knowing that a certain face is yours and yet no rule can be given for this recognition or for any other such case of knowing by acquaintance. Finally, as Ryle himself says, the use of language involves recognizing words on saying and hearing them,  a type of recognition that is itself tacit and not propositional.
Polanyi’s conception of tacit knowledge goes considerably beyond the mere recognition that there are things we know and cannot tell, by providing a structural account of this feature of cognition, which utilizes seeing as a paradigm for knowing, in spite of current philosophical argument against such use. A basic element of Polanyi’s account is the distinction between focal and subsidiary awareness.  Focal awareness is the ordinary kind of fully conscious awareness we have in focusing attention on a specifiable object. Subsidiary awareness, in contrast, refers to the periphera noticing of features of an object that are not attended to in themselves but are seen as pointers or clues to the object of focal attention. Polanyi say that we attend from subsidiary particulars to an entity under scrutiny, and calls the relations between the particulars and the whole entity the “from to relation”. 
In terms of J. J. Gibson’s recent work, we can describe the focal awareness of a pattern or coherence as the recognition of an invariance in the stimulus information received from the world about us.  Gibson also describes the gradients in the stimulus field and the variants occasioned by our moving about, both of which assist us in perceiving the objects and properties of the real world. Such variations in the stimulus field are among the elements that Polanyi classifies as the content of subsidiary awareness. Because the chief character of subsidiary particulars is that we rely on them for attending to something other than themselves (that is, the thing we are looking at or listening to), Polanyi is able to extend the notion of subsidiary awareness to include processes of both objective and subjective character
5. CM, pp. 262-3.
6. CM, p. 234.
7. Personal Knowledge, pp. 55-7; M. Polanyi, “The Logic of Tacit Inference” in Knowing and Being, ed. M. Grene (London, 1969), pp. 140-4 (reprinted from Philosophy Vol. 41 (1966), pp. 1-18).
8. The Logic of Tacit Inference “, p. 146; M. Polanyi, “The Creative Imagination” Psychological Issues VI, No. 2, monograph 22 (1969), discussion, pp. 7 1-3.
9. J. J. Gibson, The Senses Considered as Perceptual Systems (London, 1968). See also “The Creative Imagination”, p. 56.
involving all degrees of observability from the fully clear to the completely unconscious. Besides features of an object which we can easily notice, there are subtler clues whose function for us is obscure, such as details of background and context, sensations of colour, pitch, intensity and the like, motions of the eyes, the focusing muscles and the head that accompany our efforts to perceive, physiological processes that are completely subliminal, and elements of memory and expectation derived from previous experience.
While it may seem to be stretching the term ‘awareness’ rather far to use it for such a variety of more or less subliminal processes, this use has a close parallel in Ryle’s extension of the word ‘knowing’ from the propositional kind of “knowing that” into the unspecifiable kind of “knowing how”. To the argument that the term ‘awareness’ does not properly belong to events of which we are not conscious, the Polanyian reply is that it is neither simple nor straightforward to speak of consciousness of that on which we rely for attending to something else, so that the question of consciousness need not be raised in devising a special term for such occurrences.  Once the term has been accepted, we can of course go on to discuss the qualities and properties of the various types of subsidiary awareness, including their degrees of subliminality.
The functional character of reliance by which we attend from subsidiary particulars to the whole not only unites in one category a variety of different processes, but plays a basic role in determining the aspects of things thus sensed in the subsidiary mode. Things look different and sensations feel different when seen or felt as clues to something else, than when they are attended to directly. The kinds of differences that occur depend, of course, on the type of subsidiary element being considered, but in all cases we can say that the knowledge gained by subsidiary awareness is tacit. To speak in the indicative mode about such elements would require focusing attention on them and then they would no longer be the relied-on subsidiary elements in question. Of course, some of the elements cannot be focused on at all, in which case only the subsidiary mode of attention is possible. I shall come back to this point below in considering the problem of “sense.data”.
Polanyi uses the term ‘integration’ to describe the process of recognizing a coherence from subsidiary particulars. While Gibson describes the perception of information largely in passive terms of immediate recognition, another cognitive psychologist, Ulric Neisser, insists that construction by the perceiver is always involved.  The terms ‘integration’ and ‘construction’ should not be taken to refer to acts of imposing structure on unformed sense data, or of a mechanical or mathematical summation of parts, but rather to mean that the perceiver is active in forming a perception of what it is that he sees or hears, while attending to the object from its particulars.
10. “The Creative Imagination”, pp. 56 and 72.
11. U. Neisser, Cognitive Psychology (New York, 1967), esp. chs. 4, 5 and 6.
This constructive activity is more evident in cases where perception is not so clear and immediate – “moonlit” in Ryle’s phrase  - rather than in the immediate, “sunlit” type of seeing which forms the main burden of Gibson’s work.
The integrative activity involved in perceiving coherences is generally rapid and automatic. The mechanism that may account for its operation does not yet seem to have been elucidated, but whatever it is (and whether or not ‘mechanism’ is the right word for it), we know that we have such an ability. Because of its unobservable and uncontrolled nature, Polanyi refers to it as “intuition”. 
Imagination is a feature which is closely related to perceptive intuition.  According to Polanyi, we construct images of what we may see or expect to see. Imagination guides our powers of integration, for instance as we “home in” on a vague scene in the moonlight. This is not the imagination of a clear, sunlit scene, about which Ryle has a good deal to say,  but rather a vague anticipation directing the intuition in trying out various possibilities and providing the basis for judging the successes or failures of the integrative perceptual system in finding out clearly what had been only dimly anticipated.  Ryle hints at this kind of anticipatory imagination when he speaks of a person catching sight of a thimble and “having a visual sensation in a thimble-seeing frame of mind”. 
An important philosophical aspect of the from-to character of perception lies in its bearing on the conception of the world as composed of a hierarchical set of levels of complexity and organization. Even without the from-to relation, we know that perceptual wholes are not reducible to their parts, and conceptual wholes are not explainable in terms of laws that apply to their constituents. Ryle has a persuasive account of the latter point in his attack on the “Bogey of Mechanism” in which he shows that the laws of physics leave quite open the opportunity for an independent set of laws for mental behaviour. 
12. Ryle quite correctly criticizes the use in epistemology of analogies drawn only from the perception of the “familiar, expected, and sunlit” type rather than from the “belated and hesitant recognition, or misrecognition, of what is strange, unexpected or moonlit” (CM, p. 303). The former type is a marginal or limiting case of the latter the effortful species of perception clearly makes a better paradigm for the development of epistemological models than the variety in which the components of effort involved have been lost to view as a result of experience and practice.
13. M. Polanyi, “Sense-Giving and Sense-Reading”, in Knowing and Being, p. 201 (reprinted from Philosophy Vol. 42 (1967), pp. 301-25).
14. Cognitive Psychology, Chapter 6.
15. CM, ch. VIII.
16. “Sense-Giving and Sense-Reading”, pp. 199-205; “The Creative Imagination
17. CM, p. 230.
18. CM, pp. 76-82; also G. Ryle, Dilemmas (Cambridge, 1954), chapter V. This point is amplified by Polanyi’s recent account of the principle of marginal control, in which the laws of a lower level, say physics, leave open the determination of initial and boundary conditions, which are just those matters that become subject to the laws of a higher level - say chemistry or biology. M. Polanyi, “Life’s Irreducible Structure “, Science 160 (1968), pp. 1308-12; reprinted in Knowing and Being, pp. 225-39.
Since for Polanyi, as it was for Gestaltists, the perceived character of particulars is determined by the whole to which they belong, this whole has an ontological status as a real entity, and not just an explanatory or descriptive status. Ordinary objects, including mechanisms and living organisms, are real things, whose meaning is contained in their organizing principles. What about the principles themselves? I do not wish to enter into the question of whether organizing principles, or patterns, or coherences in general are real existents in some Platonic sense, or are potencies according to Aristotle or to Popper, or have some other status. What is pertinent here is that such entities belong in the same category as propositions, and constitute an extension of this category in terms of representing more complex relations than do propositions, and of being knowable with a more far-reaching component of tacit comprehension. For instance, the concept ‘human being’ expresses an organizing principle of what it is to be human that defies description and yet is clearly comprehensible not only by human beings but also by a number of species of other animals. Thus while avoiding the question of whether the abstract concept exists, we can assert that the class defined by this tacitly-known concept certainly does exist.
Now if we consider the relation of entities to each other, and the status of parts of entities, we recognize that we can shift attention from a whole to one of its parts, or from a generality to one of its instances, and conversely from a whole to a larger whole, as well as from an organizing principle to a wider one, of which the first is a particular for the second. Each real object or entity of the world is in this respect what Arthur Koestler calls a “holon “, a whole but also part of a larger whole.  Since a holon is seen differently when looked at focally and when seen as part of a larger holon, the succession of levels of wholes thus generated, the stratification of the world, has an irreducible character that is a direct consequence of the from-to relation.
Stratification in the world can also be found in the realm of thought. Ryle speaks of many cases of second and higher order processes, such as the increasing levels of sophistication in which we attend to acts of attention. The second-order character of the recognition of the acting out of an action,  of the recognition of a person’s skill in recognizing the skill of another,  and of the imagination of a scene previously perceived,  are among the aspects of hierarchical level relations in The Concept of Mind. The joint use of several levels is the source of Ryle’s concept of a “thick description”. 
At the same time that the world is stratified, it is also unified. A holon and its set of particular features constitute one entity. While it is true that a part looked at focally is an entity distinct from the whole, subsidiary particulars are only such in being integral with the whole. This point is
19. This term is introduced by Koestler in The Ghost in the Machine (London, 1967), ch. III, p. 48.
20. CM, p. 191. 21CM, p. 171. 22CM, p. 266.
23. “On the Thinking of Thoughts “, University Lectures. No. 18, (University of Saskatchewan, 1968).
insisted on by Ryle on the numerous occasions on which he says that an action, for instance, can be described by multiple predicates in a thick description but cannot be treated as being or involving two distinct and independent entities, one physical and one mental.  The stratification of the world has no sharp boundaries between its levels.
The relation between stratification and unification could be elucidated by classifying Ryle’s multiple predicates into those characterizing an entity as a whole or in some holistic way, and those that refer to a part or a collection of parts of an entity. The nearest that Ryle comes to making this distinction is his reference to main and subordinate clauses in the description of an action, as exemplified by a person pretending to be cross.  The person’s acting cross may be described by the main clause of a sentence, and his doing it by pretence by the subordinate, or we can reverse the emphasis. While different levels of stratification can be jointly described in this way, the main-subordinate clause distinction fails to show the essential from-to character of our perception of hierarchical relations.
I turn now to indwelling, another fundamental Polanyian conception which will assist in my interpretation of The Concept of Mind. The conception is derived from the way we experience our bodies. A person’s body is the one thing or collection of things in the Universe that the person knows almost exclusively in the subsidiary mode. We rely on our bodies for all our doing and perceiving; our knowledge of our members is almost entirely subsidiary. At the same time, we are aware of our bodies not as identical with our conscious selves but rather as our dwelling places, using the term ‘dwelling’ to represent that partial and ambivalent way in which our bodies resemble edifices and yet are less rather than greater than ourselves and function more as instruments than as boundaries. Subsidiary particulars occur within us, and yet we live within them.
A fundamental feature of our dwelling within our bodies is that our reliance on them and our trust in them constitute a commitment. We are there, and cannot retreat in order to examine our position, for we use our position for all our examination as well as for anything else that we do. 
Polanyi uses the term ‘indwelling’ to describe in the first instance this experienced functional relation of self to body,  and extends it by considering particulars outside our bodies that we rely on for doing things and getting about. We say that we dwell within a set of particulars, for instance a tool that we are using, when we are attending from these particulars to something we are doing, or better, relying on the particulars for doing something.  In writing with a pen, I focus on the words, the
24. CM, p. 50.
25. CM, p. 262.
26. Personal Knowledge, ch. 10.
27. “The Logic of Tacit Inference”, pp. 148-9; Personal Knowledge, p. 59.
28. “Knowing and Being “, in Knowing and Being, pp. 127-8 (reprinted from Mind Vol. LXX (1961), pp. 458-70); “Tacit Knowing: Its Bearing on Some Problems of Philosophy “, in Knowing and Being, p. 160 (reprinted from Reviews of Modern Physics Vol. 34 (1962), pp. 601-16); Personal Knowledge, pp. 55-9.
page, and the pen-point, while I am only subsidiarily aware of the pen’s contact with my hand and the motions of my hand. These two sorts or particulars merge together in a way that can be expressed by saying that the pen functions as part of my hand, and my bodily habitation is extended to include the pen - I commit myself to it. A similar extension occurs in driving a car, and explains the ease with which we learn (tacitly, of course) where the boundaries of the car are located.
A most important example of the subsidiary-focal dichotomy and the extension of bodily indwelling is that of ordinary speech. We rely on sounds for attending to the words of which they are composed; we are subsidiarily aware of words as we focus attention on a sentence, and in fact as we attend to the meaning of that sentence. We commit ourselves to the conceptions involved in the words and grammar of a language as we come to understand it, and rely on it for thought and communication. In entrusting our mental existence to language, we dwell within it. Language is necessarily public language, so that our dwelling in it makes us inherently social creatures. Ryle’s book is basically an account of the language we dwell in and confidently use for describing many kinds of human and therefore mental activity. His very considerable ability to communicate this rich variety of language use is a measure of joint indwelling in language by his readers and himself.
The relation between a collection of instances of a disposition and their joint meaning can be fruitfully discussed in Polanyian terms. Ryle characterizes a skill or “knowing how” as a capability for an indefinitely large number of similar actions.  How is it that we can identify a skill from given instances of it, and continue to recognize it by perceiving further instances that were not predictable or specifiable in advance? Clearly there is a general coherence among all the actions we lump together as illustrations of a knowing-how. We can account for our recognition of a skill if we describe the latter as the focal centre of which the known examples are subsidiary particulars, and around which we can organize new examples as pointers to the same skill. In fact, unless we make some explicit or implicit reference to an organization of particulars into a coherence, we clearly fail to account for our conviction that there is a single skill rather than many or none involved in the actions in question. Ryle gives a hint of this kind of organization when he says of boredom that it is the “temporary complexion” of the totatility of all that the bored person is doing and undergoing. 
Similarly, learning something means being able to use concepts and information in a wide and systematic range of applications.  We recognize a person’s “knowing that” by attending to his knowledge from the particulars of his competent use of it. The coherence involved in the concept of knowing how to do something or knowing that something is the case can be generalized to all dispositional terms. If a person has a disposition to
29. CM, pp.44-8.
30. CM, p. 104.
31. CM, p. 312.
act angrily towards an acquaintance, the numerous occasions and types of angry behaviour only cohere into a disposition if we can responsibly integrate these occasions and types into a single comprehensive entity. To characterize a disposition as anger or vanity or any other conditional mental entity need not place it in a wrong “causal-mechanism” category as Ryle seems to think would be implied,  but will place it properly and explicitly in the hierarchy of the organizing principles of mental life.
It is evident that the same analysis applies to the meaning of words or concepts, and also to propositions which relate concepts. The indefinite range of uses, of a word or concept, and the indefinite variety of ways in which a proposition can be expressed, point to the meanings of these entities as their controlling coherences. The “from-to” aspect of meaning shows again that articulate, propositional knowledge is rooted in tacit knowing.
Ryle refers many times to the manner in which something is done - intelligently or obediently, for instance - as a characterization in one or another category of the thing done. This is his way of pointing out that we “read” this or that manner from the particulars. But why stop there? It seems evident to me that we read the presence of knowledge, intelligence, purpose, heed-taking, obedience, or whatever in just the same way.
Mind and body are seen in this light to have a focal-subsidiary relation. More explicitly, the relation of mind to its disposition is of the same kind as the relation of a disposition to its instances in action. Polanyi describes it by saying that the mind is the meaning of the body.  Alternatively one can say that the mind is the coherence perceived when we focus attention on the person, dwelling in the workings of his mind as subsidiary particulars.  The mind is “read” through its overt workings, and in suitable cases can be read quite transparently. Ryle is quite right in rejecting the idea of a completely hidden mind, carrying out a “second set of shadowy operations  that can only be inferred by some as yet unfathomed procedures from perceptual evidence. In the present view, mind is perceived provided we focus attention on it, attending to its workings only subsidiarily. This generality-to-particular relation for mind and body can be called a whole-part relation, since in Polyani’s terms we are relating real entities and not just their organizing principles.
If the workings of the mind were looked at focally, for instance in seeing a person put a piece of a puzzle into place, what is seen could equally well be blind, stupid, mechanical or any of several other kinds of non-intelligent action. Ryle grants that we can see the difference between intelligent and non-intelligent ways of doing things, of course. The only point I wish to
32. CM, p. 86.
33. M. Polyani, “The Body-Mind Relation “, in Man and The Science of Man (Columbus, Ohio, 1969), eds. W. Coulson and C. Rogers, pp. 85-102.
34. Ryle’s closest equivalent to Polanyi’s idea of indwelling is his reference to “merely thinking what the author is doing along the same lines as those on which the author is thinking what he is doing “, CM, p. 55.
35. CM, p. 50.
make here is that in order to see the difference we must focus on the meaning and manner of doing rather than on the doing itself, and that when we direct our attention this way, the particulars of how a thing is done become seen as pointing to the mental powers manifested in the doing. The extent to which a mind is perceived in this way is a separate question, which in the present context might be answered in terms of successive levels of attending from a particular set of mental occurrences to a higher coherence embodying them.
I think it is the combination of just this kind of wholly appropriate reading of mental attributes from human activity with the wholly inappropriate tradition of Cartesian dualism that accounts for the persistence of the ghost-myth that Ryle attacks so effectively. The Concept of Mind begins by asserting that a category mistake underlies this myth, characterizing mind and mental activities as parallel activities to body and bodily occurrences, two cases of the same category with a totally mysterious connection between them. We can now see that this persistent category mistake can be further specified as a hierarchy mistake of the kind that substitutes two collections of parts for a single collection of parts and the comprehensive whole which integrates them.  Minds, or more precisely persons, do not constitute counterpart existences to bodies but more comprehensive levels of reality than bodies.
I should like to make one more application of the focal-subsidiary relation before ending this section: a clarification of the problem of whether I can inherently know more about myself than you can. If we distinguish between subsidiary and focal knowledge, we can say that you and I have different sets of subsidiary clues bearing on me, and another range of different sets of clues bearing on you. Each of us can focus attention on me or on you, but the different subsidiary clues available to each of us does not entail that I can perceive coherences about myself that are closed to you.  In fact, it leaves open the question as to which kind of observation best provides evidence for which kind of occurrence or trait.
Another important application of the from-to relation is to the problem of sensation and observation. It is clear from the discussion in section 1 that sensations and sense-data fall into the category of subsidiary particulars. We do not observe them focally while observing something in the world, and thus we neither need to locate them nor are able to discuss and analyse them. Some of the particulars, such as those that constitute our visual field, can be reflected upon and occasionally brought into aware-
36. Ryle implies this hierarchical nature of the mistake in his example of the colleges at Oxford and the University of Oxford, but he does not describe it explicitly in such terms.
37. Ryle insists that the claim that a person has uniquely privileged access to information about himself is unfounded; CM, p. 181.
ness, but others, such as the physiological items, cannot be sensed at all. Thus there can never be a question of looking at an internal image of the object seen, as Ryle shows from the simple fact that words like ‘observe’ refer to things and not their sensory images.  Both the process of looking and the achievement of seeing can only be described in terms of the external focus of the attention of the perceiver. To attempt a description in some other terms, such as neurological ones, would not only destroy the evidence by changing it, but would also destroy the coherences which define the very processes under study. In the ordinary sense of the word ‘data’, there are no sense-data.
In my judgment, a better term than ‘sense-datum’ is ‘sensory clue’. Ryle says that having sensations is neither discovering nor using clues.  He has no other way to speak of clues than as focally observable, and thus does not consider that clues, whether or not we could focus on them, have the character of clues because we do not focus on them but rely on them for their bearing on their joint but possible hidden meaning. For this reason he is forced to reject explicitly the idea that sensations are used as tools  for perceptive achievement. In spite of this rejection, however, he effectively grants the subsidiary tool-like functioning of sensations in referring, several pages earlier, to paying heed to sensations without watching them. 
Ryle’s discussion of ordinary “sunlit” imagination hinges on the idea that when our imagination brings to mind a familiar picture, it is the act of seeing which differs from that of ordinary perception, whereas the picture may be the same one that has been actually seen. We “see” a picture; we do not see a “picture “. In terms of Polanyi’s analysis of subsidiaries, we can say that imagination relies only on a set of “inner” particulars in the nervous system, excluding the retinal and motor sensors that make contact with the invariances in our surroundings. This set of inner particulars is completely beyond the possibility of focal attention, so that it makes even less sense to say that we can look at them than to speak of looking at retinal images.
The language puzzles that develop concerning sensations like itches  arise in those few cases where we can actually shift our focus of attention to particular elements of sensitivity. Even then a from-to relation must hold. The knowledge that my finger itches must come from unconscious bodily clues, which point from themselves to a genuine or spurious source of irritation in my finger.
Ryle tells amusingly how we get into trouble if we try to chronicle the process of making a formal inference.  At what time did we first make
38. CM, pp. 222-4.
39. CM, p. 232.
40. CM, p. 233.
41. CM, pp. 206-7.
42. Ryle admits his own puzzlement concerning the language about sensations that arises from such cases; CM, pp. 240-4. The approach of this paper eliminates, in my opinion, most of the difficulties that Ryle is concerned with.
43. CM, p. 299.
the passage from premiss to conclusion? Did we repeat the trip, and did we do it hurriedly or did we dawdle? We can present a formal piece of logic in a report or in a lecture to a class, but this is always long after the time we first developed or discovered it. It is the informal process of originally arriving at a piece of deductive logic that can be chronicled, although it is not always easy to do so. The temporal difference between the formal and informal processes of reasoning is reflected in a more fundamental distinction between them, which is that the former are reversible and the latter irreversible. When we present a formal argument didactically, as to a class of students, we hold it up for examination from both ends. The argument can be traced from premisses to conclusion, or followed backward from result to beginning. On the other hand, the making of a discovery, insightful or experimental, is irreversible, for once we have perceived a coherence, its clues change their character by becoming subsidiary to that coherence, and we cannot go back to the state before we perceived the coherence when the meaning of the clues was uncertain and different. Only if a long enough time elapses for us really to forget the discovery can we return to our previous state of ignorance.
The term ‘inference’ is not used by Ryle, nor by many others, for the informal processes of discovery, largely I believe because of a widespread tradition among philosophers that discovery is a psychological and not a logical process. However, Polanyi’s analysis of seeing an object or forming a concept by means of an integration of subsidiary particulars into a coherent focus is a ground for examining the process of discovery in terms that transcend the logic-psychology distinction.  His use of the term ‘tacit inference’ for this method of arriving at truth  appears to me to be an appropriate and useful extension of the term ‘inference’. It also extends the notion of tacit knowledge in immediate perception to cases in which a set of clues is contemplated for some time before the coherence to which they point is discovered. While the clues after discovery have different aspects than those before, in each case they point to the same thing, once vaguely and uncertainly, and later with clarity.
Tacit inference, thus defined, is seen to belong to the set of mental powers so well described in Ryle’s chapter on “The Intellect”;  specifically, it is an irreversible one, capable of being chronicled. There are many specialized arts of tacit inference that are developed by schooling and apprenticeship and that take their place along with the more explicit learned skills to which Ryle applies the term ‘intellectual’. Among the trained tacit powers are those by which we recognize other intellectual abilities. Tacit inference, like other powers, operates in a succession of levels of sophistication.
44. M. Polanyi, “Logic and Psychology”, American Psychologist, Vol. 23 (1968), pp. 27-42.
45. Cf. “The Logic of Tacit Inference
46. CM, ch. IX.
One of the characteristics of tacit knowledge is its real or apparent vagueness. Some of this vagueness comes from matters we have not yet put into words, some from unclear, “moonlit” perception, and still more from the subsidiary character of much of our knowledge. But the widest range of indefiniteness is that which we have discussed above in connection with dispositions. All of the terms used to describe mental events or properties - manners of doing things, varieties of disposition, and so on - have an inescapable degree of indefiniteness in their meaning, for they are capable of application to a wide range of as yet unspecified occasions. Our recognition of a person’s skill or of his intelligent learning, as contrasted with learning by rote, rests on a sufficient indeterminacy in the concept recognized that it can be applied to novel and unexpected features of the asserted skill or knowledge.
If the ideal of strict exactness were pursued successfully in a description of mind of Ryle’s type, the entire edifice would collapse, for we can only be exact in predicting the results of rote-learning and its analogues, wherein we recognize the absence of intellect. 
A most important aspect of the indeterminacy of tacit knowledge lies in the process of discovery. In Ryle’s pamphlet “On the Thinking of Thoughts”  he describes how a person puzzling something out tries many candidates for the statements or arguments or insights that will solve the problem. The person reflects on each of these and decides whether it is an appropriate step or guide for moving towards the solution to the problem. All well and good, but what Ryle does not tell us is how a person can make such decisions before the solution has been found. The problem of the Meno does not appear to be resolved by Ryle’s description.
The answer to Meno that Polanyi gives, and that seems to me to be correct,  has been hinted at above in the discussion of tacit inference. We first recognize a problem by a vague integration from a set of clues, an integration that recognizes the clues as pointing to something yet to be found. It is in terms of this vague idea that steps on the way are tested, and their testing involves further steps of intuitive, clarifying integration. The vision that initially was worse than moonlit, fog-bound in fact, begins to clear as we move through the fog towards the landscape we are trying to see.
Since the integration of particulars is needed for each trial solution as well as in judgments as to how close the trial is to the correct answer, the continual guidance of the imagination is needed. The imagination is most
47. Correspondingly, if only exact prediction is allowed for validating a scientific discovery, we are reduced to theories of the type that could almost as easily be artful inventions as descriptions of what is really the case. It is the experience of unexpected new consequences of a discovery that gives us a true sense that something real has been found.
48. See footnote 23 above.
49. Personal Knowledge, pp. 120-31; “The Unaccountable Element in Science” in Knowing and Being, p. 117 (reprinted from Philosophy, Vol. 37 (1962), pp. 1-14); “The Creative Imagination “, p. 60.
active in the questing part of the process after starting on a problem and before finally being able to put its solution into formal terms.  But even in the statement of a solution, some of the imaginative and intuitive scaffolding  erected for purposes of discovery must surely remain, at least for the discoverer, in the tacit integration that allows him to keep track of and to make sense of his formal apparatus.
For the readers or hearers of a formal discovery, tacit components are also needed. The clues by which the reader or hearer constructs a coherent imaginative background that makes sense of the formal presentation come not from the original scaffolding but from a didactic substitute in the form of illustrations and examples, and more importantly although more subtly, from nuances of expression in the presenter’s speech or writing.
The innovative and unspeciflable way in which imagination assists the process of discovery is also the way it functions in the much simpler and more common case of ordinary speech. What we are about to say will form the coherence of clues that point to it, primarily clues in the imagination that knows incompletely what it is we mean and are trying to say. To paraphrase Ryle’s reference to thimble-seeing, we might say that we begin a certain sentence in an asserting-the-weather-to-be-good frame of mind. Imagination of this sort provides a type of anticipatory mental act that precedes speech and yet does not constitute that explicit private rehearsal which Ryle rejects.  It is perfectly consistent with Ryle’s monistic view of body and mind to include this function of the imagination along with the more explicit type which he describes, and it satisfies the unhappy feeling one gets in reading The Concept of Mind that speech comes as it were out of an empty head. Ryle does not mean that it does, but in his effort to attack the Cartesian ghost, he has not given enough attention to our indeterminate, imaginative, anticipatory powers to allow what I think he really means to become clear.
Ryle’s disposal of the ghost in the machine is so effective that the value or even the possibility of adding to the argument might well be questioned. Nevertheless, the from-to structure of knowledge does add a new dimension to the attack on the ghost. I described above our dwelling within any set of particulars as a generalization of the way we dwell within our bodies, and the way in which a tool such as a pen becomes a part of our body, in that the contact between self and outside world is transferred to the point of the pen. When we focus our attention on. a complex state of affairs, such as writing a paper, we become so involved in the many kinds and levels of subsidiary elements that we come to dwell in them in the same way as for
50. “Sense-giving and Sense-reading”, pp. 199-200; “The Creative Imagination” pp. 64-6. See also “Genius in Science “, Encounter, forthcoming.
51. Cf. CM, pp. 291-2, for a graphic description of this scaffolding.
52. CM, pp. 295-6.
simpler eases of tool-using. I have already described in more general terms our dwelling within the whole range of language that we use.
I see no way of actually drawing the line between such extended in-dwelling and the purely biological type. I can shift my boundary both outward and inward. If I look at my finger and contemplate the possibility of losing it, I use a narrower concept of myself than usual. The boundary shifts with a shift of the division between the focal and the subsidiary.
Now if the boundary between self and world can shift with the focus of attention, how about the boundary between mind and body? The difficulty in focusing attention on parts of our bodies, and the impossibility of objectifying them in toto, means that the limits of inward shifting of the self-body or mind-body boundary cannot be found. The ghost in the machine was a different object of the same category as the machine; clearly each half of the dichotomy had its own fixed boundary. With the boundary undefinable and movable, we have another argument against the old dualism.
I pointed out in Section 1 that indwelling is a measure of our commitment. In these terms, we can answer a puzzle that must appear to the more behaviouristically oriented in reading Ryle’s book. How can he be so sure and so convincing to us about all his ways of describing mental activity when every term has an indefinite range of meaning and almost none of them can be defined operationally without destroying the whole enterprise? The answer is that Ryle, or any of us, is able to describe mental occurrences with confidence because we dwell in the particulars and in the language used to describe them so thoroughly as to be completely committed to them. It is a measure of our common commitment that Ryle can elicit from us his readers our confident assent to nearly all of his descriptions.
I have tried to show that the Polanyian distinction between subsidiary and focal awareness enriches Ryle’s account of mind by allowing mind to be a comprehensive entity distinct from its overt workings and yet unified with them. The concept of attending from a set of particulars to a focal entity and its generalization in terms of tacit inference, indwelling, and commitment, provide a firm basis for validating Ryle’s extensive and illuminating account of mental activities, an account in language of indefinite range of applicability, which requires judgment in application and yet carries with it both clarity and conviction.
While Ryle’s account may be taken as part of the groundwork for a Grammar of Scientific Knowledge in the field of psychology,  the informal processes of discovery and insight by which we can affirm his account form the groundwork for a Grammar of Scientific Discovery. How we find things out and how we state our knowledge of them in the study of mind, both depend on the from-to relation and on the belief in comprehensive entities that is affirmed by that relation.
University of Nevada, Reno.
53. CM, pp. 317-8.