You are here

Series I

Lecture 2: The Varieties of Knowledge

Acts and Dispositions

Before we discuss the relation between belief and knowledge, something must be said about the concept of knowledge itself.

Plato, in the Theaetetus, distinguished between possessing a piece of knowledge and using it, and pointed out that we still possess it at times when we are not using it. The possession of a piece of knowledge is an example of what we should now call a disposition. For example, we all possess or ‘have’ the knowledge that 7 × 7 = 49. We acquired this piece of knowledge many years ago and probably we shall retain it for the rest of our lives. But we are not always actually thinking of or attending to this mathematical truth. It actually ‘comes into our minds’ only occasionally, once a week perhaps, or less often than that. Nevertheless, we have the capacity of recalling this proposition whenever we need to (for example, whenever we are engaged in a calculation to which it is is relevant), and of actually ‘realizing’ or ‘acknowledging’ that it is true.

This is what is meant by saying that our knowledge that 7 × 7 = 49 is a disposition rather than an occurrence. (It is an acquired disposition, of course. There was a time when we learned this mathematical truth.) The disposition shows itself or manifests itself from time to time by actual mental occurrences, for example, when we actually use the proposition 7 × 7 = 49 in calculating the size of a carpet. But a disposition is something which we still have or possess at times when it is not actually being manifested at all.

Now in ordinary everyday English the verb ‘to know’ is generally used in a dispositional sense; not quite invariably perhaps, but certainly the dispositional use of it is by far the most common. To take another example (a piece of empirical knowledge this time) we all know and have long known that there are lions in Africa. We know this even at times when we are not thinking of it at all, and indeed most of us think of it only very seldom. We know it even when we are asleep. You would not say of a sleeping man ‘he no longer knows that there are lions in Africa’, for that would suggest that he has lost this piece of knowledge and will have to learn it again when he wakes up.

This might indeed be his situation if he were suffering from very severe concussion, or from some very severe emotional shock. Then we might find, when he woke up, that he had lost much of the knowledge which he formerly possessed, and that he did have to learn it again. But we are speaking of normal sleep, and using it as the most striking example to illustrate Plato's distinction between ‘possessing’ and ‘using’; for here a man still possesses all the knowledge he has in his waking hours, but for the time being he is not using any of it, at any rate if his sleep is dreamless. To put it technically, he still has a very large number of acquired dispositions, but for the time being none of them are being actualized.

I have said that in everday speech the verb ‘to know’ is generally used in the dispositional sense. Philosophers, however, have sometimes talked about acts of knowing. They have conceived of knowing as a special sort of mental occurrence, and have asked how it differs from other mental occurrences. This way of speaking is now much less common than it was. It has one rather surprising consequence. If someone knows something, in this act or occurrence sense, it would make sense to ask at what time he knows it. At 1.30 p.m. on Wednesday he was knowing that 7 × 7 = 49, and on Sunday morning, just after he woke up, he was knowing that there are lions in Africa. The oddity of such statements as this suggests that there is something odd about the idea of an ‘act of knowing’.

We can agree, of course, that acquiring a piece of knowledge—discovering something, finding it out—is a mental occurrence, and there is no harm in calling it a mental act. At such and such a time I suddenly ‘saw’ or ‘grasped’ the proof of Pythagoras' theorem. I suddenly ‘saw’ or ‘realized’ that given the axioms of Euclid the square on the hypotenuse of a right angled triangle must be equal to the sum of the squares on the other two sides. This seeing or grasping is a dateable mental occurrence; a mental act, if you like.

We can also agree that when once we have acquired a piece of knowledge, e.g. the knowledge that the distance from London to Edinburgh is about 390 miles, this knowledge is liable to be actualized or manifested thereafter by dateable mental occurrences; for instance when someone asks us what the distance from London to Edinburgh is, or when we are trying to work out what a railway ticket from the one place to the other would cost, or how long the journey would take in a train whose average speed is 45 m.p.h. On such occasions we recall or attend to this fact which we know, and this is an actual and dateable occurrence in our mental history: we may call it a ‘mental act’ if we like. But according to our ordinary usage of the word ‘know’ it is an act of recalling or using a piece of knowledge we possess, rather than an act of knowing.

It might therefore be thought that the phrase ‘act of knowing’ is at best a purely technical one, with no sanction at all in ordinary usage. And if so, surely we had better abandon it altogether (as many contemporary philosophers have) since our aim as philosophers is to analyse the concepts we actually have, and not to replace them by different ones? But it is not quite true that the ‘act’ or ‘occurrence’ conception of knowing has no basis at all in our ordinary everyday use of language. Sometimes, though not very often, we do find ourselves using the word ‘know’ in an occurrent sense, to refer to a dateable mental event.

I shall give an autobiographical example. In 1918, when I was learning to fly, aerial navigation hardly existed. We used to find our way by following main roads or railway lines. One day I was lost in a fog over East Anglia. I came down very low and read the name on a small railway station. Then at last I knew where I was. Here the word ‘know’ is used for a mental occurrence or act, occurring at a particular time (in the late afternoon, as far as I can remember). Something ‘dawns on us’ at a particular moment. Of course, you can say, if you like, that this is a case of coming to know, or finding something out. But I think it is perfectly intelligible to say ‘then I knew’—using the word ‘knew’ for a dateable occurrence.

Another example can be found in the sentence: ‘After that, he knew no more until he woke up in the hospital twenty-four hours later’. Here, knowing does seem to be thought of as a series of mental occurrences, which was interrupted for a certain period of time, and then began again. This example is quite different from the one I gave earlier, when a man is not said to cease to know things when he is asleep. Here, on the contrary, he is said to cease knowing when he becomes unconscious, and for twenty-four hours he does not know anything at all. It seems that the ‘occurrent’ sense of the word ‘know’ (in which knowing is conceived as a dateable mental event) does have some foothold in ordinary everyday language, though the dispositional sense of the word is by far the most common one.

It cannot be said, then, that the term ‘act of knowing’ is a wholly unintelligible one. The mistake made by the philosophers who used it was not that they did talk about mental acts, but that they failed to talk about anything else. They discussed what is called ‘the problem of knowledge’ as if certain sorts of mental occurrences were the only topic which has to be considered; and, despite what Plato had said, they paid little or no attention to the dispositional sense of the word ‘knowledge’—the sense in which our knowledge is something which we possess throughout a period (for many years perhaps) and still possess even when we are not at the moment using it. To ignore this dispositional sense of the word ‘know’ is a serious error. If the ‘act’ or occurrence sense of the word were the only one, it would follow that we always lost our knowledge of something the moment we cease to consider it or attend to it. If knowing the distance from London to Edinburgh is just a mental occurrence or mental act, it follows that we cease to know the distance from London to Edinburgh as soon as that act or occurrence comes to an end; we no longer know it, though for a short time we did.

Of course, we do quite often cease to know something. Knowledge may be lost as well as acquired. We may easily lose some piece of knowledge which we previously had. I used to know what my dentist's telephone number is, but I no longer know it. But it would be absurd to say that when we cease to attend to or consider some fact, we ipso facto cease to know it. One may add that our knowledge would be of little value to us if the ‘mental occurrence’ account of it were the whole truth about it. The important thing about a piece of knowledge, once we have got it, is that we are often able to retain it, and recall it to mind again whenever we need it. It is not just a matter of momentary flashes of insight which come and go; it is more like enriching ourselves, acquiring a stock of valuable possessions, which we continue to have at our disposal afterwards.

All this is forgotten or ignored if our philosophical discussion of knowledge is just a discussion of mental acts or mental occurrences. We need not reject the ‘act of knowing’ terminology altogether, as some contemporary philosophers would wish to do, and certainly the fact that ‘act of knowing’ is a technical term is not in itself a good reason for abandoning it. But we must not suppose that all the things we need to say about knowledge can be said in a ‘mental act’ or ‘mental occurrence’ terminology. Some of the most important of these things cannot. The same applies to belief too, as we shall see later. Here too, as I have already said, we are confronted with a similar contrast between a mental act or occurrence analysis and a dispositional analysis. And here too, as I shall try to show later, the mental act or occurrence analysis is not so completely mistaken as some modern philosophers suppose. But here too some of the most important things we want to say about belief cannot be said in a mental act or mental occurrence terminology.

Professor C. D. Broad has remarked that a mind has two distinctive characteristics, consciousness and retentiveness. A theory of knowledge which takes account only of mental acts or occurrences may perhaps do justice to consciousness, but not to retentiveness. A mind without retentiveness, if such an entity is conceivable—what Leibniz called a mens momentanea seu carens recordatione—might have momentary flashes of insight, but it would not have any knowledge in the ordinary sense of the word ‘knowledge’. Similarly it might give momentary assent to propositions, but it could not have what we ordinarily call beliefs. The advantage of the dispositional conception of knowledge is that it draws our attention to the memory-dimension which all knowledge has. The point of becoming aware of something is that we are then in a position to remember it afterwards.

Knowledge of Facts and Knowledge by Acquaintance

In ordinary English the verb ‘to know’ is sometimes followed by a dependent clause. Most of us know that the Battle of Waterloo was fought in 1815, and as I was writing these words I knew that I was in Oxford and that I was writing them. Such knowledge is sometimes called knowledge of facts. There is some difficulty in this way of speaking. If we adopt it we shall have to say that there are general facts, for example the fact that some cats are black, and even universal facts, such as the fact that all lions are carnivorous, for we know that some cats are black and that all lions are carnivorous. There must also be negative facts, for we can know that there is no tea in the teapot. Furthermore, we shall have to say that there are conditional facts, and even facts expressed in what is called a ‘contrary to fact’ conditional sentence. For example I know that I am afraid of Alsatian dogs. When I say this, I am not saying that I now have any feeling of fear. What I claim to know is that if I were to see an Alsatian dog near me (which I do not) or that if I believed that there was one near me (which I do not) I should feel fear.

But perhaps these examples are not intolerably odd. Perhaps they only seem so, because we have to use technical terms (the technical terminology of logic) to draw distinctions between one sort of fact and another, distinctions which do not need to be drawn in everyday speech. After all, it surely is a fact that some cats are black and that all lions are carnivorous. There is in fact no tea in the teapot—nothing but hot water—because you forgot to put the tea in. If you claim to have put it in, I shall reply ‘the fact is that it is not there’. Conditional facts are more difficult to swallow, especially where the sentences formulating them are contrary-to-fact conditional sentences. It is notorious that such sentences are puzzling, and that it is very difficult to find a satisfactory analysis of such a proposition as ‘if it were now raining, the streets would be wet’. But what is contrary to fact is not this conditional proposition as a whole, but one (or sometimes both) of its constituent propositions taken separately. It surely is a fact—or if you like, it is as a matter of fact the case—that if it were now raining the streets would be wet. We may well be puzzled to say what kind of a fact it is, but can we seriously deny that it is one?

A Priori Truths and Empirical Facts

But a more awkward difficulty arises when the sentence which follows the ‘that’ is an a priori sentence. We all know that 7 × 7 = 49. Is it a fact that 7 × 7 = 49? If it is, there must be a priori facts or necessary facts. Sometimes we speak as if there were. For otherwise, what point could there be in calling other facts empirical? We do often speak of empirical facts, from which it would seem to follow that there are also facts which are non-empirical, or at any rate that there might be. If not, what could empirical facts be contrasted with, and what point could there be in applying the word ‘empirical’ to facts at all?

Why then should it seem odd to speak of necessary facts or a priori facts (e.g. mathematical facts or logical facts)? There is no doubt that it does. Of course we might be reluctant to say ‘it is not a fact that 7 × 7 = 49’, because this might suggest that the proposition 7 × 7 = 49 is false. But we should also be reluctant to say that it is a fact that 7 × 7 = 49. Why is this? Perhaps it is because ‘fact’ suggests ‘matter of fact’. It is not just a matter of fact that 7 × 7 = 49. It is not something we can ascertain by observation or experiment or historical research. A matter of fact is something which just happens to be the case. It is what philosophers call ‘contingent’. There happen to be tigers in India, but conceivably there might not have been. But it does not just happen to be the case that 7 × 7 = 49 or that 17 is a prime number and 18 is not.

We can avoid these difficulties by speaking of a priori truths or necessary truths (rather than a priori or necessary facts.) What kind of truths they are, what makes them true, are questions we need not go into at present. It is sufficient to point out that they cannot be denied without contradiction. It is logically impossible that 7 × 7 should not be equal to 49. But the denial that there are tigers in India involves no contradiction. It is logically possible that there should be no tigers in India, though as a matter of fact there are.

If we return now to ‘knowledge that’, we can see that there are two course open to us. We can either divide ‘knowledge that’ into two kinds (1) knowledge of facts (2) knowledge of a priori truths; or else we can say that it is just knowledge of truths, and then divide truths into two kinds (1) empirical truths and (2) a priori truths. It does not matter much which of these two alternatives we choose, but for my part I prefer the first, because I should like to be allowed to go on using the good old down-to-earth word ‘fact’. So I prefer to divide knowledge into (1) knowledge of facts (2) knowledge of a priori truths.

Knowing ‘Who’, ‘What’, ‘Where’, Etc.

We must now notice that when the verb ‘to know’ governs a dependent clause, it need not be a ‘that’ clause, though it frequently is. For example one may know who did it, or where someone is, or what he is doing. We may know when the Battle of Waterloo occurred, or how long this piece of wood is, or what the length of it is. We may also know what the square root of 144 is, or how many prime numbers there are between 0 and 20. Presumably this way of speaking arose because statements expressing knowledge are often answers to questions. ‘I know where he is’ in the sense that I have the knowledge needed to answer the question ‘Where is he?’. If so, this kind of knowledge is reducible to ‘knowledge that’. For example, I know that Robert is in the kitchen, and this is what enables me to answer the question ‘Where is he?’ if anyone asks me. Similarly I know that the square root of 144 is 12, and this is what enables me to answer the question ‘what is the square root of 144?’. To put the same point more technically: there is some fact of the form ‘Robert is now at place P’ and I claim to know this fact (though I do not state it) when I say I know where he is. Similarly there is some mathematical truth of the form x² = 144 and I claim to know this truth, though I do not state it, when I say I know what the square root of 144 is.

What has been said will apply to some of the cases in which the dependent clause begins with the word ‘how’. If I know how long a piece of wood is, or how many people there are in the room, this is reducible to knowledge ‘that’ in the way just explained. I know that the piece of wood is 5½ long or that there are 15 people in the room, and so I am able to answer the question ‘how long is it?’ or ‘how many people are there in the room?’. But when the dependent clause is of the form ‘how to…’ (e.g. ‘Robert knows how to make an omelette’) special problems arise, which we must consider later. Sometimes they arise also when other interrogation words are used: e.g. I know what to do, he now knows where to stop. At any rate we must agree with Professor Ryle that knowing how to… is not reducible to knowing that…

Knowledge by Acquaintance

We have mentioned various sorts of dependent clauses which may follow the verb ‘to know’. But sometimes it is not followed by any dependent clause at all. Instead, it governs an accusative—a noun or a noun-phrase. We speak of knowing Mr Robinson-Smith, or knowing Scotland, or knowing the man who keeps the shop round the corner, or knowing Kensington Gardens in London. What is known in this sense need not always be an individual entity. It might be a group of entities. A traveller may know all the countries of Central America. I myself could claim to know all the counties of England and Wales, though I know some of them much better than others. (As we shall see, knowledge of this kind admits of degrees.) I also know some of the counties of Scotland, but not all. You may know all the members of the Robinson-Smith family, including great-aunts and second cousins, whereas I only know Mr Robinson-Smith himself. This is very different from ‘knowledge that’. What is known here is not a fact or a truth but an entity of some kind, or sometimes a group of entities.

Most commonly, the entity is either a thing or a person. I use the word ‘thing’ with some misgivings. It is a little odd to call Scotland or Kensington Gardens a thing, and if we do call them things, we have to say that both of them are highly complex things. But at any rate they are actual existent entities, and both of them are material entities. A person (or at least all the persons we should ordinarily claim to know) is a material entity too, but that is not all he is; so we ordinarily distinguish between persons and things. Sometimes, too, we distinguish between living organisms, animals and plants, on the one hand, and things on the other. But sometimes we describe organisms as ‘living things’.

‘Knowledge that’ may be called a ‘propositional attitude’. But the knowledge we are now discussing is not a propositional attitude at all. It is sometimes called knowledge by acquaintance. One cannot have it unless one has actually encountered the person or thing which is known. To be acquainted with something A, you must have been aware of A itself. Perhaps we cannot know anything by acquaintance without also coming to know at least some facts or truths about it. But certainly we can know truths or facts about something without being acquainted with it. A student of geography may know many facts about Scotland; but he cannot know Scotland unless he has actually been there and seen at least some parts of it for himself. Students of Roman History may know many facts about Julius Caesar, but they are not in a position to know him. One could not know Caesar himself unless one had actually met him, though conceivably a very learned twentieth century historian might know many more facts about Caesar than some of the people who had actually met him.

In some languages these two very different sorts of knowledge are distinguished by the use of two different words: for example cognoscere and scire in Latin, connaître and savoir in French, kennen and wissen in German. And in old English there were the two different verbs I ken and I wiss. ‘Do you ken John Peel?’ that is, are you acquainted with him? On the other hand, in the Authorised version of the Bible St Paul says ‘I wist not, Brethren, that he was the High Priest’.1 It is a misfortune, perhaps, that in modern English the distinction between ‘kenning’ and ‘wissing’ has been abandoned. In Scotland, the verb ‘to ken’ still survives in some quarters, but it has to do duty for ‘wissing’ as well, just as ‘know’ does in ordinary modern English.2

In ordinary everyday speech, what is said to be known in the ‘acquaintance’ sense is often a perceptible entity of some kind—a person, a town, a country, the large oak tree at the cross-roads. Often, but not always. Someone may know Descartes, which floats on high o'er version of the Ontological Argument, but not St Anselm's. I never knew the celebrated Oxford philosopher F. H. Bradley, though he was still alive when I was an undergraduate. But I do know some of his philosophical doctrines e.g. his doctrine of Degrees of Truth and Reality. Most of us know Wordsworth's poem about the daffodils. You might ask me whether I know any of Wagner's operas, and be shocked to learn that I do not.

There is some difficulty about these examples, though in all of them it would, I think, be good English to use the word ‘know’ in its ‘acquaintance’ sense. For one thing, it is a little odd to call Descartes, which floats on high o'er version of the Ontological Argument an entity, though it would be true to say that there is such a version of the Ontological Argument, and also that there are certain philosophical doctrines propounded by Bradley; that there are many poems which were written by Wordsworth, the poem about the daffodils among them, and that there are many operas composed by Wagner. Moreover, in the examples mentioned earlier (knowing a person, or a country, or an object such as an oak tree) it was plausible to say that knowledge by acquaintance is not a propositional attitude at all, and that this is the difference between knowledge by acquaintance and ‘knowledge that’. But surely Descartes, which floats on high o'er version of the Ontological Argument is just a set of interconnected propositions; and so is Bradley's doctrine of Degrees of Truth and Reality. What else could it be?

What kind of an entity Wordsworth's poem may be is a more puzzling question still. But certainly there is something pro-positional about it. ‘I wandered lonely as a cloud, which floats on high o'erer vales and hills’ is surely a proposition (a pretty complex one too). Those two verses describe a possible state of affairs, whatever else they do; whatever other effects they may have on the mind of the reader, they do at any rate enable him to entertain a proposition, though he does not have to believe it. They may even be a description of something which actually happened, a walk which Wordsworth actually took one day in the mountains of Westmorland.

In all these three examples, Descartes' argument, Bradley's doctrine of Truth and Reality, Wordsworth's poem, what we are said to know by acquaintance might perhaps be called ‘a propositional structure’. Even Wordsworth's poem is at least a propositional structure, though it is certainly something more. What kind of an entity one of Wagner's operas could be, I do not dare to enquire. The question had better be left to more musically-minded philosophers. I will only venture to say that an opera looks rather like a universal, since the same opera can be performed at many times and places, and is something which ‘there is’ even at times when it is not actually being performed at all. Some philosophers have held, as we shall see, that universals can be known by acquaintance. If this view is acceptable, perhaps it might do for knowledge of operas as well.

For the moment, however, let us confine our attention to our other examples, in which what is known by acquaintance is apparently a propositional structure of one kind or another. Even so, there is still a very important difference between such knowledge by acquaintance and ‘knowledge that’. To know Descartes' version of the Ontological Argument I do not have to accept his version of the argument, or indeed anyone else's. All that is required of me is that I should be familiar with the propositions of which the argument consists, including of course those of the form ‘p entails q’. To be familiar with this propositional structure, I do have to entertain these propositions and moreover I have to do this for myself (no one else can do it for me). In this sense Descartes' argument is something which I meet with or encounter. But I do not have to accept the argument.

The same applies to Bradley's doctrine of Degrees of Truth and Reality. To know this doctrine, in the acquaintance sense, I have to entertain a set of propositions and familiarize myself with them, but I do not have to accept all of them or even any of them. Similarly, to know Wordsworth's poem about the daffodils, I do have to entertain a series of propositions (I have to do this at least, whatever else I do or have done to me). But obviously I do not have to accept them. It is only necessary that I should consider them. The question whether they are true or false is irrelevant and I neither accept them nor reject them.

But in ‘knowledge that’ the situation is quite different. If someone knows that Winchester is south of Oxford, he is not only familiar with the proposition ‘Winchester is South of Oxford’ and capable of entertaining it whenever it is relevant, he also accepts it; moreoever, he accepts it with full conviction, and has sufficient reasons for doing so.

Other Objects of Knowledge by Acquaintance

So far, our examples of knowledge by acquaintance have been taken from ordinary everyday speech. Philosophers, however, have sometimes claimed that there are other objects of acquaintance, very different from those we have been discussing. Russell held at one time that universals are known by acquaintance. He and others have also held that sense-data are objects of acquaintance. Some have thought that there is introspective knowledge by acquaintance—that each of us know by acquaintance at least some of the contents of his own mind.

Do these philosophical doctrines accord with our everyday ways of speaking and thinking? Of course, they might be illuminating, or indeed true, even if they do not. But when a philosopher uses an everyday word or expression (such as ‘know’ followed by an accusative) it is just as well to be clear whether he is using it in its everyday sense or in a technical sense. We may fall into confusion if we slip from the ordinary usage to the technical one, or back again, without noticing what we are doing. What is intelligible (at least false) when the word is used in the one way might be unintelligible or absurd (not even false) when it is used in the other.

We may notice two features in our ordinary everyday concept of knowledge by acquaintance. On the one hand, knowledge by acquaintance has a first-hand or face-to-face character. Knowledge by acquaintance is contrasted with the second-hand or ‘hearsay’ knowledge which we get from testimony, spoken or written. It is also contrasted with the knowledge we get by means of inference, for example, when we infer that the postman must have come while we were out because there is now a letter in the letter box, and it was not there when we left. What we learn in either of these two ways is often reliable enough to be counted as knowledge, at least according to our everyday usage of the word. But it is, of course, ‘knowledge that’. And we ordinarily think of it as being in some way ‘indirect’; it lacks the first hand or face-to-face character of knowledge by acquaintance.

But knowledge by acquaintance, in our ordinary way of conceiving it, reflected in everyday language, has another character as well. We should not ordinarily be said to know something by acquaintance unless we had familiarized ourselves with it in some degree, so that we are able to recognize it when we encounter it again. At any rate, we must be able to recognize it in ‘easy cases’ where there are no special obstacles, such as poor light, defective eyesight, disguises such as false beards etc. We have to ask, therefore, whether the same two features, ‘first-handness’ and familiarity, are present in these philosophical usages of the term ‘knowledge by acquaintance’.

Let us first consider the doctrine that universals are known by acquaintance. If we are willing to use the terminology of universals at all, we do, I think, have to agree that some universals are known by acquaintance in something very like the ordinary everyday usage of the phrase. The word ‘universal’ itself (used as a substantive) is of course a technical term. But when it is maintained that some universals, at any rate, are known by acquaintance the phrase ‘known by acquaintance’ is not being used in a technical way. The two features mentioned just now, first-handness and familiarity, are both present. No one else can tell us what it is like for something to be red, or to be inside something else. Nor can we find this out by inference. We have to see it for ourselves, by actual face-to-face inspection of instances. Moreover, knowing what it is like for something to be red, or inside something else, has the other characteristic which I mentioned. Familiarity is an essential element in it. A man who knows what it is like for something to be red is capable of recognising the colour red when he encounters it again in other instances.

The ordinary everyday speaker, quite innocent of philosophical theories, might ask someone ‘Do you know the smell of burnt castor-oil?’ And though he does not use the word ‘universal’ (as a substantive) he does conceive of this smell as something repeatable; this same smell may occur on many different occasions, at many different times and places. Moreoever, he does think of it as something—a repeatable something—which some people are familiar with and capable of recognizing, though others are not; and he thinks of this familiarity as something which can only be acquired by smelling this smell for oneself, in a direct and first hand manner. Once the technical term ‘universal’ had been explained to him, the ordinary man would have no difficulty at all in understanding the statement that the smell of burnt castor oil is a universal which is known, and can only be known, by acquaintance.

It would, however, be a little odd to say that all the universals we conceive of are known by acquaintance. It seems plain that some of them are defined in terms of others, and we learn to conceive of them by understanding the definitions, not by first hand encounter with instances. (These are the universals which Locke called ‘complex ideas’.) It might even be that there are no instances to be encountered. It is not likely that anyone has ever actually encountered a unicorn. But if we use the terminology of universals, we certainly have to say that ‘being a unicorn’ is a universal, and that some people know this universal, in the sense that they know what a unicorn would be like if there were one, and would be able to recognize it as a unicorn if by any chance they were to meet it. This knowledge has one of the two features which we noticed in our ordinary everyday concept of knowledge by acquaintance, but it lacks the other. The familiarity is there, but the first-handness is not; or if in a way it is, it is quite a different sort of first-handness, not at all like the ‘face to face encounter’ which is a feature of our ordinary everyday concept of knowledge by acquaintance. Our familiarity with the concept ‘unicorn’ did of course have to be acquired, and there was something first hand about the way we acquired it. We had to understand the definition of ‘unicorn’ (‘a horse-like creature with a long horn in the middle of its forehead’). We had to understand it for ourselves, and no one else could do it for us. But this intellectual first-handness, this ‘do-it-yourself’ character (where the operation done is an intellectual one) is something very unlike a face-to-face encounter.

Nevertheless, if we use the terminology of universals, we still have to say that some universals are known by acquaintance and that some have to be known in that way if any universals are to be known at all. If universal A is defined, in the way just illustrated, in terms of other universals B, C and D, we cannot understand the definition unless we are already familiar with the universals B, C and D. There must then be some ‘basic’ or ‘ground floor’ universals (not necessarily the same ones for each person) which we become familiar with by direct first-hand encounter with instances. And these universale are indefinable, or if you prefer, they have only ostensive definitions.

Are Sense Data Known by Acquaintance?

The position is rather different with the other philosophical doctrines mentioned on page 54. Let us consider the doctrine that sense data are known by acquaintance. Some philosophers, I think, have regarded sense data as universals. It seems that some of the American Critical Realists did. (Santayana, for instance, seems to have thought of sense data as ‘essences’.) I find this view very difficult to understand. But if it is tenable, knowing sense data by acquaintance would just be a special case of knowing universals by acquaintance, and what has already been said about knowing universals by acquaintance would apply.

But certainly most philosophers who have talked about sense data have thought that they were particulars. They thought that every sense datum is an instance of a universal, or rather of several universals, but not that it is a universal, nor yet a conjunction of several universals. Now according to our ordinary everyday conception of knowledge by acquaintance, some objects of acquaintance are certainly particulars. Mr Jones is not a mere universal, nor a set of universals. He is an actual existent entity. Or, in the conceptualist terminology, Mr Jones is not just an abstract idea, nor yet a set of many abstract ideas, though many abstract ideas apply to him. Scotland is not just an abstract idea either, nor a set of abstract ideas, but an actual existent entity.

Let us consider Mr Jones, since human beings are perhaps the most typical of all objects of knowledge by acquaintance, according to our ordinary everyday conception of such knowledge. We can certainly say that he is a particular. It is true that we might prefer to call him an individual or a continuant, but certainly he falls on the ‘particular’ side of the particular/universal antithesis. Nevertheless, he is a very much more complex particular than any sense datum is. He has a vast and perhaps even inexhaustible set of properties, whereas a sense datum has very few. Moreover, some of them are dispositional properties which he still possesses even when they are not being actually manifested (irascibility for instance, or a taste for cross-word puzzles), But it would not make sense to ascribe dispositional properties to sense data at all.

These differences, striking as they are, might of course be irrelevant. There is no reason in principle why quite simple entities should not be known by acquaintance as well as complex ones; nor is there any obvious reason for saying that everything known by acquaintance must have dispositional properties.

But there is another difference which is relevant. In a way it is a difference in complexity too, but in this case the complexity is temporal. Such an entity as Mr Jones or Scotland has a vastly greater duration than any sense datum could have. Mr Jones, considered as a physically embodied human being, lasts for seventy years perhaps. (We need not ask whether he continues to exist in a disembodied state thereafter.) Sense data are very fleeting and short lived entities in comparison. We have only to shut our eyes and all the sense data in our visual field cease to exist, to be succeeded by new ones when we open our eyes again. As you walk round the room, you sense a series of different visual fields. Your visual sense data come and go, and are never the same ones in two successive seconds, so long as your walk continues. Occasionally we do have a relatively long-lived sense datum. For example when we hear the sound of a siren or hooter the same auditory sense datum may continue for several minutes. Again, as I sit working in my chair, the ‘feel’ of the chair beneath me and of the floor beneath my feet (tactual sense data) may continue for quite a long time, and so may a bodily pain (a ‘somatic’ sense datum) for example a headache, though the degree of attention I give it may vary. But no sense datum lasts for many years, as Mr Jones does, and most of them have a very brief duration indeed.

One consequence of this is that it is very difficult to familiarize oneself with a sense datum. It seldom lasts for long enough. Perhaps if the siren blows for five minutes, by the end of the second minute we are beginning to familiarize ourselves with the disagreeable auditory sense datum we are sensing; and if the headache continues for hours, this somatic sense datum does become familiar to the victim. But if all the sense data we are aware of are known to us by acquaintance, even the very short-lived ones—and this is what some philosophers have maintained—it is clear that the term ‘knowledge by acquaintance’ is not being used in the way we use it in ordinary speech, but in a technical way. What we find here is rather like what we found in the case of complex universals, but now it is the other way about. In the case of sense data the feature of first-hand encounter is present, and indeed emphasized by the sense datum philosophers themselves. But here it is the feature of familiarization which causes the trouble. If all sense data are objects of knowledge by acquaintance in the ordinary everyday sense, it should be possible to familiarize oneself with every sense datum that one experiences, whereas in fact is it possible only in a few exceptional cases.

The particulars which are objects of knowledge by acquaintance, in the ordinary sense of the term, have another important characteristic which sense data lack, and this too is connected with duration. Such existent entities as Mr Jones or Scotland or the oak tree at the cross-roads can not only be encountered, they can be encountered again after an interval, and recognized to be the same entities as they were before. They can be re-visited, even though a long period of non-visiting may intervene between one visit and the next. After twenty years' absence I can go back to the ‘dear old place’ where I used to spend my holidays, and recognize it to be the same place, despite the changes which have occurred in the interval—the new petrol station which has been put up, the thatched cottage which has been pulled down, the pervasive smell of diesel fumes, the forest of television masts on the roof-tops. Perhaps the intervals between visits have to be shorter when the object of knowledge by acquaintance is a person, because persons are apt to change more rapidly than places do. Someone I used to know well may be unrecognizable (to me) when I meet him again after an interval of thirty years. One's acquaintance with a person has to be ‘kept up’, as we say, if we are to continue to know him. Still, one does not have to re-encounter him every day, or even once a year. Once in every two or three years is usually enough. Such re-visiting is however impossible with sense data. The visual sense data I am now experiencing cease to exist when I shut my eyes or turn my head.

So much for knowing sense data by acquaintance. It turns out that this use of the term ‘knowledge by acquaintance’ is a technical one, different from the use it has in ordinary speech, though the two uses do have an important point in common, because knowledge by acquaintance has a first-hand or ‘face-to-face’ character in both cases alike.

Introspective Knowledge by Acquaintance

We may now turn to the third philosophical doctrine mentioned on page 54 the one concerning introspective knowledge by acquaintance. A person is a self-conscious being, or rather, any normal person who has passed the age of infancy is in some degree a self-conscious being. He is usually aware of some, at any rate, of the events which are going on in his mind at any particular time, though never perhaps of all of them. (At any given time, some of them are beneath the threshold of consciousness or beyond the margin; and probably the position of the threshold varies considerably from one person to another, and for the same person at different times.)

This awareness of the contents of our own minds is often so inattentive that we should be reluctant to call it knowledge; indeed, in some cases we might prefer to say that a person is ‘not wholly unaware’ of what is going on in his own mind instead of saying that he is aware of it. Sometimes, however, we notice what is going on in our own minds, or some part of what is going on. This attentive awareness of some present content of our own minds is what we usually call introspection. Perhaps it is always in some degree retrospection as well, since mental events pass so quickly. Certainly it is in some degree retrospection if we go further, and engage in the much disapproved of practice of introspective study, and try to classify and to analyse these swiftly passing mental events, or to find the most appropriate words for describing them. But it is not necessary to consider this studious sort of introspection. Let us confine ourselves to what one might call ordinary introspection, the attentive awareness of some present content of our own minds. This occurs sometimes in nearly everyone (even, perhaps, in the most behaviouristic psychologist), though more frequently in some persons than in others. Is it a form of knowledge by acquaintance? And if we say it is, shall we be using the term ‘knowledge by acquaintance’ in a technical sense, or in its ordinary everyday sense?

The factor of ‘first-hand encounter’ is certainly present. A person just finds or comes upon this thought or image or feeling which is occurring in him at a particular time. And the noticing of it is something he has to do for himself. It has to be first-hand; no one else can do it for him. The most that someone else can do is to tell him what to look for. (‘Are you sure you are not feeling just a little frightened?’)

But though the factor of first-handness is there, the factor of familiarization is absent. The situation, then, is much the same as it is in the doctrine that sense data are known by acquaintance. The contents of our own minds, or at least those contents which are in any sense directly introspectible, are brief events rather than persisting entities. A few of them may last long enough for us to become familiar with them (perhaps we should then be inclined to call them ‘mental states’ rather than ‘mental events’). For instance, I may have a leaden feeling of depression all through the day, quite long enough for me to become only too familiar with it. But certainly this is not true of all the introspectible contents of our own minds; and it would have to be true of all of them if the term ‘knowledge by acquaintance’ is being used in its ordinary everyday sense in the philosophical doctrine we are discussing, since it is a doctrine about all introspection, or introspection as such.

Of course, we may become introspectively familiar with some of the qualities which introspectible events have (as also with the qualities which sense data have), and may learn to recognize these qualities when they are instantiated again. I have often felt, and noticed, feelings of regret in myself, and next time I have a feeling of regret I shall probably be able to recognize that it is one. But this would just be a special case of what is called knowing a universal by acquaintance. If we use the terminology of universals, we have to say that introspectible events exemplify universals, as everything else does which exists or occurs. Or, to put it in the conceptualist terminology, we are able to acquire concepts or abstract ideas by attending to mental events in ourselves, and noticing their resemblances and differences.

Knowledge of Ourselves

But, it may be said, surely there is a sense in which we do know ourselves? No doubt some people have much more self knowledge than others: but surely each of us must know himself in some degree? Such knowledge may or may not be what these philosophers were talking about when they spoke of ‘introspective acquaintance’. But we certainly do possess it, and it is an important sort of knowledge. And what is it but knowledge by acquaintance, in the ordinary everyday sense? Here is the verb ‘to know’ governing an accusative, as it does when we are said ‘to know’ Mr Jones. What are we to make of the Delphic Oracle's advice gnoÒthi seauton ‘Know thyself’? (The verb gignoÒskein, which the oracle used, is the equivalent of our ‘know’ in the acquaintance sense.) It was a very good piece of advice too. The Roman poet was not far wrong when he said ‘“Know thyself” come down from heaven’3.

Yet despite the verb the Oracle used, a great deal of what we ordinarily call self-knowledge is clearly ‘knowledge that’, knowledge of facts or truths about ourselves, and especially knowledge about our own dispositions; for example, a person might know that he is resentful and does not easily forgive injuries, or that he is interested in Roman History, or that he likes Mr X much more than he likes Mr Y. Such knowledge about oneself is important, both from a moral and a prudential point of view. It is also difficult to acquire, because there are emotional resistances in all of us which tend to prevent us from acquiring it. Perhaps this is the sort of knowledge which the Oracle had in mind, knowledge about one's own dispositions, or more generally about one's own character (knowing ‘what kind of a person one is’). But if so, the knowledge it recommends us to acquire is knowledge ‘that’, and we must respectfully accuse it of using the wrong verb. Or did it just mean ‘Notice what goes on in your own mind as carefully as you can?’

Perhaps the Oracle meant something different from either. Perhaps it was telling us to know the Pure Ego, which according to some philosophers, and some religious teachers too, is the most important element in each man's personality. If there is indeed such an entity, it could conceivably be known by acquaintance in something like the ordinary everyday sense. For according to what we are told about it, it is not an event or happening like a thought or a feeling of anxiety, which comes and then goes. It is either a persistent entity or perhaps a supra-temporal one.

Knowledge by Description

Russell distinguished between knowledge by acquaintance and knowledge by description. But it is interesting to notice that the distinction had already been drawn about 100 years before by Jane Austen. In Chapter 21 of Sense and Sensibility Lucy says ‘I have not known you long, to be sure, personally at least, but I have known you and all your family by description a great while; and as soon as I saw you I felt almost as if you was an old acquaintance’. Perhaps the philosophers of ordinary language, who dislike technical terms, may take some comfort from this passage. The terminology of Miss Austen must surely be good English.

What is it to know someone or something by description? In Mysticism and Logic, Essay X (‘Knowledge by Acquaintance and Knowledge by Description’),4 Russell gives the following definition: ‘I shall say that an object is known by description when we know that it is “the so and so”, i.e. when we know that there is one object, and no more, having a certain property.’5 In the summary at the end of the essay he puts it thus: ‘we have descriptive knowledge of an object when we know that it is the object having some property or properties with which we are acquainted; that is to say, when we know that the property or properties in question belong to one object and no more, we are said to have knowledge of that one object by description, whether or not we are acquainted with the object’.6

For our purposes, the most important cases are those in which some object or person is known to us only by description, since these are the ones which are most relevant to the distinction between knowledge and belief. For example, one cannot know Paris by acquaintance unless one has actually been there, but one may know it by description as the capital of France. The description has to be what Russell calls a definite description (‘The capital of France’). It is conceivable that France might have two capitals, as the Roman Empire had in the fourth century, Rome and Constantinople. In order to know something by description we have to know that there is something, and only one thing, to which a certain description applies. And of course we do in fact claim to know that there is one city and only one which is a capital of France. Alternatively, a person who has never been there might know Paris by description as the place indicated by this round black spot on a map at which he is looking, since according to the conventions of map-making, such a round spot represents one town and only one.

It will be seen that if we are to know something by description, in Russell's sense, three conditions must be fulfilled, (1) We have to conceive of some characteristic or set of characteristics. The characteristic might be a relational one. (2) We have to know that this characteristic or set of characteristics does in fact characterize something. This is knowledge of the truth of an existential proposition of the form ‘There is something which…’ The existential character of knowledge by description is pointed out by Russell himself.7 (3) Finally we have to know that the characteristic or set of characteristics in question characterize only one thing.

As has been mentioned already, the first two conditions may be fulfilled when the third is not. Perhaps you have never seen a giraffe. But you may still conceive of a giraffe by means of the description ‘a large spotted horse-like creature with two short horns and a very long neck, which enables it to eat the upper foliage of some trees’; and you may know that there are in fact animals to which this description applies. This is an important kind of knowledge, and is obviously quite different from knowledge by acquaintance. Moreoever, it is in a sense knowledge by description, since the understanding of a description is an essential part of it; and the proposition known to be true is an existential one. But this is not what Russell himself calls knowledge by description, because his third condition is not fulfilled. So far from knowing that there is only one object to which the description applies, you know that there are many, though you do not know how many. Russell's knowledge by description is knowledge by exclusive or identifying descriptions. Perhaps he defined it in this way because he thought of it as a kind of second-best substitute for knowing an individual object or person by acquaintance.

This may also be the explanation of a rather strange linguistic fact. When Russell (or Jane Austen) speaks of knowledge by description, the verb ‘to know’ takes an accusative. What we are said to know by description is an actual entity of some sort, as opposed to a truth or set of truths. Yet when a piece of knowledge by description is analysed, it turns out to be reducible to knowledge that, knowledge of facts or truths. We know that a certain description applies to something, and that there is only one thing to which it applies. If our knowledge is knowledge that an existential proposition is true, why should we speak as if it was knowledge by acquaintance? Presumably because the second ‘that’ clause lays a special sort of restriction on the existential proposition formulated in the first one. Whenever we know that an existential proposition is true, we know that a certain description has application, or that a certain concept, or set of concepts, is instantiated. But in this case we also know that the description applies to only one thing, or that the concept or set of concepts has only one instance. And knowledge by acquaintance, at least in the most typical examples of it, has the same ‘singular’ character. It is knowledge of this person or this thing, or again of this sense datum or mental image. Even if we hold that universals (‘properties’) are knowable by acquaintance, as Russell did when he wrote the essay we are discussing, acquaintance with a universal would still be acquaintance with a single, though non-sensible, entity—just this universal as opposed to that one.

Nevertheless, the differences between knowledge by acquaintance and knowledge by description are more important than the resemblances. Knowledge by description does not have the character of ‘first hand encounter’ which knowledge by acquaintance has. There is something indirect or second hand about it. Most frequently we get it from testimony, from reading what others have written, or hearing what they tell us. The testimony may be conveyed to us in a pictorial form, for example by means of photographs or diagrams, or maps, or chronological tables, though we still need the words of other people (written or spoken) to tell us what these visible representations are representations of, or how to interpret them, as when we are taught by others how to ‘read’ a map. But sometimes our knowledge by description is acquired by means of inference. If I see a footprint on the flowerbed, I can infer that some human being and only one has stood there; and by means of very complicated inferences the planet Neptune was known by description before anyone had observed it. There are also mixed cases, where the premisses of our inference are learned from testimony. Some trustworthy witness might tell me that there was a footprint on the flower-bed this morning, though I did not see it myself, and it has since been obliterated by the gardener. And when Neptune was ‘calculated into existence by enormous heaps of algebra’, it is possible that one (or both) of the astronomers who independently made the calculation was relying on the observations of others, conveyed to him by testimony, as well as observations of his own.

Again, knowledge by description has an abstract character, by its very nature. It is the knowledge that a certain abstract idea, or set of abstract ideas, has one and only one instance. And if we are willing to include knowledge by means of an indefinite description (we have seen that this is in a sense ‘descriptive’ knowledge too) the abstractness is still more obvious. The knowledge that there are lions in Africa, though perfectly good knowledge as far as it goes, is something very different indeed from actually seeing or touching a lion, or feeling it bite you. However complex an abstract idea is, and even if we do know that it has just one instance and no more, there is still something ‘thin’ about knowledge by description. It lacks the richness and concrete detail of knowledge by acquaintance. There is something superficial about it, as well as something indirect or second hand. Perhaps it never wholly satisfies us, and never can.

If this is its defect, it does of course have a compensating virtue. The range of knowledge by acquaintance is exceedingly narrow, even if we take the liberal view of it which common sense takes, and allow that such individual entities as persons or horses or trees or dogs can be known by acquaintance. (If it is true, as some philosophers have thought, that the only entities one can know by acquaintance are one's own sense data and some of the contents of one's own mind, the range of knowledge by acquaintance must be narrower still.)

But the range of knowledge by description is vastly wider. By this means, and no other, we can come to know about an enormous multitude of ‘existences and objects which we do not see or feel’. Even if we take the most liberal view of knowledge by acquaintance, we still have to say that almost the whole of our historical and geographical knowledge is knowledge by description. One of the most important differences between a civilized person and a savage is that he has far more knowledge by description; and as a result of this, we could say that the civilized person is not quite such a ‘stranger in the universe’ as the savage is. One of the chief aims of education is to provide us with knowledge by description and the means of getting more.

Finally, it is important to notice that one piece of knowledge by description may depend upon another. If we do not know Mr Harold Wilson by acquaintance, we may still know him by description as the present Prime Minister of Great Britain. But perhaps we are not acquainted with Great Britain either, and it too is only known to us by description, as the largest island to the north-west of the European continent. Again, to take a historical example, the Emperor Tiberius may be known to us by description, e.g. as the successor of Augustus. But Augustus, in his turn, is himself only known to us by description, e.g. as the first Roman Emperor, or as the great-nephew and heir of Julius Caesar; and both the Roman Empire and Julius Caesar are known to us only by description, if known to us at all.

‘Circulus in Describendo’

When we consider such examples, knowledge by description seems like a network or chain which hangs unsupported in the air, each link described in terms of another or several others, and that in turn described in terms of another or several others. We seem to be in danger of a fallacy which might be called circulus in describendo. Suppose we ask someone what he knows of Paris, and he replies ‘All I know of it is that it is the capital of France’. We then ask him what he knows of France, and he says ‘All I know of France is that it is the country of which Paris is the capital’. Clearly there is something wrong. A is described in terms of B, and B in terms of A.

Sometimes the circle is less obvious because it is larger. France is described as the country which is west of Germany, and Germany as the country which has Belgium on the north-west side of it, and Belgium as the country which is south east of England across the sea, and England as the country which is north of France across the sea. The circle may be so large that it covers the whole of geography, and every single place on the earth's surface is described in terms of others. Or again, it may be so large that it covers the whole history of the Roman Empire.

But if knowledge by description consisted just in ‘knowing one's way about’ through such a system of interdependent descriptions it would not deserve the name of knowledge, not even of reasonable belief. It would have no relation to reality, to what actually exists or happens, or what has actually existed or happened. This whole conceptual structure of inter-connected descriptions floats in the air until some item in it is related to something which we know by acquaintance. To take a geographical example, if at a certain time I can know England by description as the country containing these objects I see or touch, I can escape from this circulus in describendo. And another person can escape from it if he knows France by description as the country containing the objects which he sees or touches at a particular time.

The situation is something like the one we encounter in the Coherence Theory of truth, where there is a system of propositions each one of which is supported in a greater or lesser degree by some or all of the others. The difficulty is that this whole system of mutually supporting propositions need not be true of anything. However complex it is, and however coherent it is, it might still be just a complex and coherent fiction, and no degree of complexity will suffice to save it from being so. To escape from this predicament, there must be a reference to experience somewhere, as some advocates of the Coherence Theory have acknowledged, Bradley for one. With only a little first-hand experience we can sometimes go a very long way. But we must have some empirical facts if such a coherent system is to deserve the honourable title of ‘knowledge’, or even the less honourable one of ‘reasonable belief’.

It comes to this. Knowledge by description is only possible if some of the descriptions mention entities known by acquaintance, for example, this thing here, what I see happening now. We can then describe entities remote in time or space by means of their relations to what is here and now. And this is what we actually do. The network of inter-connected geographical or historical descriptions is ‘tied down’ at many different points to what we see and touch. For example, if the Emperor Augustus is described as ‘the Emperor who began to reign in 31 BC’, the year 31 BC can be described as ‘1993 years before the present year’ and ‘the present year’ as ‘the year in which I am now writing’. What is here and now for me may be known to you only by description, and what is here and now for you may be known to me only by description. But for each of us there must be something which is here and now, if either of us is to know anything by description. As Russell says ‘If we are to obtain a description which we know to be applicable, we shall be compelled, at some point, to bring in a reference to a particular with which we are acquainted’.8

‘Knowing How To’

So far, we have been considering only various cognitive senses of the word ‘know’. But it also has what may be called a practical sense, when it refers to some kind of expertness or skill or proficiency. This is what Professor Ryle calls ‘knowing how to’, for example knowing how to ride a bicycle or how to cure someone of influenza. Perhaps the word ‘practical’ is too narrow. It may suggest that when we know how to do something, the doing has to be some kind of physical activity, such as repairing a puncture. But a man may also know how to solve some purely theoretical problem, for example, how to prove that in a Euclidean rightangled triangle the square on the hypotenuse is equal to the sum of the squares on the other two sides; and on a particular occasion, when he actually does this piece of reasoning, he might do it entirely ‘in his head’ without uttering words or drawing a diagram on paper. Still, this is knowing how to do something, how to conduct a certain sort of operation, though the operation is an intellectual one. On the other hand, an activity which looks purely physical may be an intellectual one as well, and this may be the most important thing about it, as when someone spends a busy day looking through mediaeval charters or turning the pages of a copy of Domesday Book. The ‘knowledge how to’, which he manifests in these physical activities, may be a knowledge how to discover the derivation of some puzzling English place-name; and the operation of seeking for its derivation is an intellectual one, a piece of investigation or enquiry designed to discover the truth about something. If the researcher's investigation is successful, the result of it is the acquisition of knowledge that, or of reasonable belief that.

But sometimes, when the word ‘know’ appears at first sight to be used in one of its cognitive senses, we discover on further reflection that it is being used at least partly in the ‘how to’ sense. Knowing the way to the station consists at least partly in knowing how to get there, knowing how to go there from any place within a certain area. But this knowing ‘how to’ does have a cognitive side too. If I know how to get to the station, I must be able to recognize at least some of the landmarks which I encounter on the way, and to recognize the station itself when I get there. Recognizing is not itself a form of knowing how to.

Similarly, when we say that someone knows a certain piece of country very well, part of what we mean is that he knows how to find his way about it (if he did not, we should say that he did not in fact know it very well). But we also mean that he is familiar with that piece of country or well acquainted with it; and when he finds his way about it, or tells others how to do so, his knowledge ‘how to’ is a capacity for making practical applications of his familiarity or acquaintance. Again, knowing the meaning of a word consists partly in knowing how to use it in statements, comments, questions, reports, etc., which make sense, or in descriptive phrases which make sense. Knowing the meaning of a syntactical word like ‘of’ or ‘whenever’ consists in nothing else but this. One knows how to operate with it, or what to do with it. But with what are called ‘object-words’, such as ‘eat’ or ‘blue’ or ‘run’, knowing their meaning consists only partly, not wholly, in knowing how to use them in combination with other words. We are also capable of recognising the entities or qualities or occurrences to which they apply.

From the epistemological point of view, the important cases are those in which knowledge in the ‘how to’ sense is liable to be confused with knowledge in one of its cognitive senses, or those in which one of its cognitive senses is combined with the ‘how to’ sense, and the verb ‘to know’ is used to cover both at once without distinguishing them.

Perhaps it is well to add that we are not merely doers. Knowledge by acquaintance and ‘knowledge that’ (including knowledge by description) are valued for their own sake, and not merely for the sake of the ‘know how’ which sometimes results from them. Obviously it is useful to know how to do things, but also it is just nice to know how to do them, even though we seldom or never have occasion to do them in fact. Even ‘know how’ may be valued for its own sake and not mainly for what we can get out of it. Aristotle was surely right when he said ‘All men by nature desire to know’ and this applies to all the different sorts of knowledge which we have distinguished.

  • 1.

    Acts of the Apostles, XXIII, 5.

  • 2.

    ‘Where is he?’ ‘I dinna ken’ i.e. ‘There is no place such that I ken that he is in it.

  • 3.

    ‘e caelo descendit γνωθι σεαυτον

  • 4.

    Mysticism and Logic (George Allen & Unwin, London), pp. 214–15.

  • 5.

    Mysticism and Logic (George Allen & Unwin, London), pp. 214–15.

  • 6.

    Mysticism and Logic, p. 231.

  • 7.

    Mysticism and Logic, p. 215.

  • 8.

    Mysticism and Logic, p. 217.

From the book: