R. M. Hare, The Language of Morals (1952)
6
MEANING AND CRITERIA6.1. The argument of the preceding chapter establishes that 'good', being a word used for commending, is not to be defined in terms of a set of characteristics whose names are not used for commending. This does not mean that there is no relation between what have been called 'good-making' characteristics and 'good'; it means only that this relation is not one of entailment. What the relation is, I shall discuss later. But before that it is necessary to guard against a mistake into which it was easy to fall, when it had been shown that 'good' was not analysable in the way that naturalism suggests. This was the mistake of supposing that, because 'good' was not the name of a complex property ('good strawberry', for example, meaning 'strawberry that is sweet, juicy, firm, red, and large'), it must therefore be the name of a simple property. Of course, if all that is meant by 'property' is 'whatever an adjective stands for', then it is harmless to say that 'good' is the name of a simple property, except in so far as it suggests that there is for every adjective something to which it stands in this superficially simple, but philosophically baffling relation. But because 'property' is not normally used in such a broad sense as this, the use of the word in this connexion has led to serious confusion; it has led to comparisons between 'good' and typical simple property-words like 'red'. It is this comparison that must now be examined. Since it is, in fact, very difficult to establish a logical criterion for distinguishing simple from complex properties, I shall not confine the argument as narrowly as this comparison would suggest; the arguments which I shall use tell equally against the theory that 'good' is the name of a complex property, in the commonly accepted sense. They are complementary to another series of arguments marshalled with great skill by Mr. Toulmin against a similar theory.1
It is characteristic of the word 'red' that we can explain its meaning in a certain way. The suggestion that the logical character of words can be investigated by asking how we would explain their meaning comes from Wittgenstein. The point of the method is that it brings out the ways in which the learner could get the meaning wrong, and so helps to show what is required in order to get it right. Let us suppose that we are trying to teach English to a foreign philosopher who either deliberately or inadvertently makes all the mistakes that he logically can (for what mistakes anyone actually makes or avoids is irrelevant). We must assume that, when we start, he knows no English, and we know nothing of his language. At a certain stage, we shall get to the simple property-words. If we had to explain the meaning of the word 'red' to such a person, we might proceed as follows: we might take him to see pillar-boxes, tomatoes, underground trains, &c, and say, as we showed him each object, 'That is red'. And then we might take him to see some pairs of things that were like each other in most respects, but unlike in colour (for example pillar-boxes in England and Ireland, ripe and unripe tomatoes, London Transport trains and main line electric trains), and on each occasion say 'This is red; that is not red but green'. In this way he would learn the use of the word 'red'; he would become conversant with its meaning.
It is tempting to assume that the meaning of all words that are applied in any sense to things could be conveyed (directly or indirectly) in the same way; but this is not so, as is well known. The word 'this' could not be treated in this fashion, nor, perhaps, could the word 'Quaxo' -- if we can call the name of a cat a word at all. It is instructive to ask whether the meaning of 'good' could be explained like this, and if not, why not.
6.2. It is a characteristic of 'good' that it can be applied to any number of different classes of objects. We have good cricket-bats, good chronometers, good fire-extinguishers, good pictures, good sunsets, good men. The same is true of the word 'red'; all the objects I have just listed might be red. We have to ask first, whether, in explaining the meaning of the word 'good', it would be possible to explain its meaning in all of these expressions at once, or whether it would be necessary to explain 'good cricket-bat' first, and then go on to explain 'good chronometer' in the second lesson, 'good fire-extinguisher' in the third, and so on; and if the latter, whether in each lesson we should be teaching something entirely new -- like teaching the meaning of 'fast dye' after we had in a previous lesson taught the meaning of 'fast motorcar' -- or whether it would be just the same lesson over again, with a different example -- like teaching 'red dye' after we had taught 'red motor-car'. Or there might be some third possibility.
The view that 'good chronometer' would be a completely new lesson, even though the day before we had taught 'good cricket-bat', runs at once into difficulties. For it would mean that at any one time our learner could only use the word 'good' in speaking of classes of objects which he had learnt so far. He would never be able to go straight up to a new class of objects and use the word 'good' of one of them. When he had learnt 'good cricket-bat' and 'good idironometer', he would not be able to manage 'good fire-extinguisher'; and when he had learnt the latter, he would still be unable to manage 'good motor-car'. But in fact one of the most noticeable things about the way we use 'good' is that we are able to use it for entirely new classes of objects that we have never called 'good' before. Suppose that someone starts collecting cacti for the first time and puts one on his mantel-piece -- the only cactus in the country. Suppose then that a friend sees it, and says 'I must have one of those'; so he sends for one from wherever they grow, and puts it on his mantel-piece, and when his friend comes in, he says 'I've got a better cactus than yours'. But how does he know how to apply the word in this way? He has never learnt to apply 'good' to cacti; he does not even know any criteria for telling a good cactus from a bad one (for as yet there are none); but he has learnt to use the word 'good', and having learnt that, he can apply it to any class of objects that he requires to place in order of merit. He and his friend may dispute about the criteria of good cacti; they may attempt to set up rival criteria; but they could not even do this unless they were from the start under no difficulty in using the word 'good'. Since, therefore, it is possible to use the word 'good' for a new class of objects without further instruction, learning the use of the word for one class of objects cannot be a different lesson from learning it for another class of objects -- though learning the criteria of goodness in a new class of objects may be a new lesson each time.
It is strange that the use of 'good cacti' should not be a new lesson; for good cacti seem to have little in common with good chronometers, and good chronometers with good cricket-bats. Yet somehow we seem to be able to learn the use of the word without being taught what in a particular class of objects entitles us to apply it to a member of that class. Suppose that, in teaching the meaning of 'good', we determine not to be put off by the apparent dissimilarities of good chronometers, cacti, and cricket-bats; suppose that we go on trying at all costs to find something that we can point to in any class of objects whatever, and say 'There you are, that's what makes a thing good; when you've learnt to identify that elusive quality, you will know the meaning of the word'. This seems at first sight a natural procedure; for if the use of the word 'good' is common to all classes of objects, it must have a common meaning; and it is natural to suppose that if it has a common meaning, there is a common property to which it refers, like 'red'.
Such efforts are doomed to failure. But even if we fail to find a common property for all classes of objects, members of which are called good, we may try to carry out the programme in a less ambitious way; we may give up the struggle to find one common property, and content ourselves with dividing the uses of the word into a few groups, within each of which the word refers to a common property. Thus we may think that in the first lesson we may be able to teach the meaning of the word in its 'intrinsic' use, and then in the second lesson go on to teach the 'instrumental' use, and so on.
This procedure also runs into difficulties. People who have suggested it have usually been most interested in 'intrinsic' good; and therefore they have divided off 'instrumental' good only in order to ignore it. This has meant that they have also ignored the immense difficulties of dealing with 'instrumental' good in this fashion. I propose to adopt the opposite programme; I shall put 'intrinsic' good on one side for the moment, and ask whether it is possible to treat 'instrumental' goodness as a common ostensible property.
6.3. There are two different possible variants of the suggested procedure. One seeks to explain the meaning of 'good', as used 'instrumentally', on the assumption that the common property which we are looking for is that of being conducive to good in the 'intrinsic' sense. This will hardly do; for we call many things 'good so-and-sos' in an instrumental way which are not conducive to 'intrinsic' good; for example, good pistols (which are as good pistols in the hands of the murderer as in those of the police); here, granted the assumption that there is one 'instrumental' use of the word, it seems that the word 'good' is being used in just the same way as in 'good chronometer'. Good chronometers, too, are not always conducive to intrinsic good -- not if they are used for navigating aircraft which are going to drop atom bombs on the chosen people (whichever that is).
The other variant seeks to explain the meaning of the word 'good' as used 'instrumentally' on the assumption that it means the same as 'efficient', that is to say, conducive to the end that it is used for. Now it is possible that 'good' does sometimes mean this; I am not at the moment discussing whether it does or not, but whether, if it does, 'being conducive to the end that it is used for' is the sort of property that we could teach our foreigner to identify in one lesson. Let us suppose that we try. We take him round to a lot of people who are doing things with things, and say to him 'That is a good X', 'That is not a good Y', and so on. But suppose that he is, or pretends to be, rather obtuse. We take him to see cricket-bats and chronometers and fire-extinguishers, and point out, in each case, which are good and which are not. But he still refuses to admit that he can recognize a property common to good members of all these three classes. His difficulty is obvious. Good fire-extinguishers differ from bad fire-extinguishers in putting out flames quickly, without fumes, &c.; good chronometers differ from bad ones in giving Greenwich time, being easily readable, &c.; good cricket-bats differ from bad ones in hitting balls far and fast, not stinging, &c.; but there seems to be little in common to all these three performances that he can learn to recognize. We call them all 'the ends for which the objects are used'; but this common designation presents the same difficulty as we were having with the word 'good' itself. For unless we can teach him, in the case of any new class of objects, to recognize without assistance for what end they are being used, we shall still have to go on giving him a new lesson each time, though it will be not about the word 'good' but about the word 'end'. And the fact that the word 'end' presents the same problems as the word 'good' suggests that the problems have the same source in both cases. We remember that Aristotle, who made the word 'end' a technical term in philosophy, defined it as 'a good to be achieved by action'.2
6.4. There is a certain class of words which we may call, in a wide sense, 'functional words'. A word is a functional word if, in order to explain its meaning fully, we have to say what the object it refers to is for, or what it is supposed to do. Functional words include, not only the names of instruments in the narrow sense, but also the names of technicians and techniques. We do not know what a carpenter is until we know what a carpenter is supposed to do. Similarly with an auger; we do not know what an auger is, until we know, in the words of the Shorter Oxford English Dictionary, not only that it has 'a long pointed shank, &c.', but also that it is 'a carpenter's tool for boring holes in wood, &c.' We could not explain the meaning of 'auger' to our foreigner in this sense by showing him a lot of augers, and teaching him to recognize an auger when he saw one. He might be able to do this infallibly, and still not know what augers are for, and so not know fully the meaning of the word as the dictionary gives it. It is more conducive to clarity to regard the peculiarities of sentences like 'this is not a good auger', which we shall be considering, as due to this feature of the word 'auger' (the fact that it is a functional word) than to say that the word 'good' has a special meaning in this sentence. In this sentence we are handed on a plate, in virtue of the meanings of the words used, one of the necessary criteria of a good auger; but we are handed it by the word 'auger', not by the word 'good'. We saw above that it is possible to construct 'hypothetical' imperative sentences which are derivable from indicative minor premisses alone, and that this is done by including the required imperative major premiss as part of the conclusion, inside an 'if'-clause. We have here a somewhat similar operation. To know what an auger is for is to know the end that augers are supposed to fulfil; it is to know that being able to bore holes is a necessary condition of being a good auger, or that if any auger will not bore holes it is not a good auger. But if we define 'auger' in such a way that this major premiss is analytic, then by including the word 'auger' in the conclusion 'This is not a good auger' we make this conclusion derivable from the indicative minor premiss alone, 'This auger will not bore holes'.
But to know what an auger is for, is to have no more than a very rudimentary knowledge of the criteria of a good auger; it is to know one necessary condition only. Holes may be bored with very bad augers. We can indeed say that if an auger will not bore holes at all, then it is certainly a bad one; but this is as far as the definition of 'auger' by itself takes us. For this reason, 'good auger' means a lot more than 'auger which is conducive to the end that augers are used for, sc. boring holes'; it means at least 'auger which is conducive to fulfilling well the end that augers are used for, sc. to boring holes well'. And so even if our foreigner knew what an auger was, there would be a lot that we should have to teach him still about the criteria of a good auger. We should have to teach him, for example, that a good auger does not blister the hands, is not rusty, and bores holes that have clean edges.
Let us ask, however, what would be involved in teaching our foreigner this minimum, that an auger is for boring holes. We should have to take him to see people boring holes with augers. He would have to know what they were trying to do. If he thought that they were just trying to exercise their wrists, we should not be able by this demonstration to explain to him what an auger is for. Now to try to produce a result is to choose, subject to the limitations of our knowledge and power, to do those things which are conducive to that result. Thus to try to make a hole is to choose to do those things (including selecting those instruments) which are conducive to a hole's being made.
This same word 'choose' obtrudes itself if we seek to explain, not what an auger is used for, but what it is designed for. To design an instrument for boring holes is to choose to have it made in such a way that it is conducive to boring holes. The fact that the word 'choose' obtrudes itself in this way is of great interest. To choose is to answer a question of the form 'What shall I do?' The man who is designing an instrument for boring holes asks himself 'Of what design shall I have this instrument made?' and answers 'Of such a design that it will be conducive to boring holes'; the man who is trying to bore holes asks himself 'What sort of instrument shall I employ?' and answers 'The sort of instrument that will be conducive to boring holes'. Thus we have here an important tie-up between the present discussion and that of the first part of this book. But let us return to our foreigner. It is by now established that if we can explain to him what choosing is -- or if he knows already -- then we shall be able to explain how to find out, in the case of any instrument, what it is for; and that if we can explain this, we can also give him some rudimentary explanation of how to tell a good member of any class of instruments from a bad one. If, on the other hand, he does not understand what choosing is, he will not understand any of our explanations.
It is therefore apparent, that we have here a similar situation to that which we noticed earlier. To teach what makes a member of any class a good member of the class is indeed a new lesson for each class of objects; but nevertheless the word 'good' has a constant meaning which, once learnt, can be understood no matter what class of objects is being discussed. We have, as I have already said, to make a distinction between the meaning of the word 'good' and the criteria for its application. Even in the case of instrumental goodness, there is not one common criterion for all classes of objects. We still have to teach our learner something new each time. It is true that the words 'conducive to' will occur in all our explanations; but there will occur after these words some other expression, such as 'boring holes' or 'keeping exact time' which will be different in each case. If instead of all these different expressions we write the common expression '. . . what the instrument is for', we reintroduce an expression whose meaning is not explicable by the 'red' technique. It requires an understanding of what it is to choose; and this understanding is necessary, whether we make the reference to it in explaining the meaning of 'auger' or 'chronometer', or whether we explain the meanings of these words (inadequately) simply by showing examples, and leave the reference to choosing until we have to explain 'good auger' or 'good chronometer'.
Thus the notion of 'instrumental good', which was introduced in order to mitigate the difficulty of 'a new lesson for each class of objects', fails in this purpose. To sum up: there is no common property which is recognizable in all cases where a member of a class -- no matter what class -- is said to be 'instrumentally good'. Even if, therefore, we divide the uses of the word 'good' into certain broad classes, 'instrumental good', 'intrinsic good', and so on, we still cannot apply the technique of explanation within those classes that we apply with the word 'red'. We can teach the criteria for applying the word 'good' within a particular class; but this does not teach the meaning of the word. A man could even learn to tell good augers from bad, without in the least knowing what 'good' meant; he could, that is to say, learn to sort out augers into piles, good and bad, and do this quite correctly, but still not realize that this classification was for the purpose of selecting some augers in preference to others. Suppose, for example, that he was coming with us on a long voyage of exploration, and we said to him 'Don't forget to bring an auger', and he brought one of the bad ones; we should think that he did not know the meaning of 'good auger', although quite able to tell a good auger from a bad one.
6.5. I will now describe a way in which, provided that our foreigner knew the meaning of the word 'choose', I might indeed be able to explain the meaning of 'good' to him in one lesson, the paradoxical character of which will emphasize the point that I have been making. Suppose that I ask him to teach me one of the games of his own country, and he says that he will teach me about the game of smashmak. This game, he explains, is played with a thing called a shmakum. Before asking him to describe to me a shmakum, or to proceed with his account of the game, I say to him 'Where do you get these shmakums from?' and he answers 'From shmakum-makers; in our country every town has a street of shmakum-makers'. I then ask 'Suppose you are buying a new shmakum, and you go to this street, and all sorts of shmakums are offered you, all about the same price, what sort of shmakum would you choose?'; and he replies 'All other things being equal, I would choose the one that I could make the most smashes with'. I then make a bold venture, and say 'Ah! I see, then you think the best shmakum is the one that you could make most smashes with'.
Now it might well puzzle my learner that I can say this. We must assume that he has learnt, on the analogy with other adjectives, that 'best' is the superlative of 'good'. But the strange thing is that, although I do not know how the game of smashmak is played, or what a shmakum is like, or what it is to make smashes, I have, on the strength simply of his telling me that he would choose, all other things being equal, the shmakum with which he could make the most smashes, ventured to suggest that he thinks that this sort of shmakum is the best sort of shmakum. How, he may well ask, can I tell what properties he thinks shmakums, with which he could make the most smashes, have, except that he could make the most smashes with them -- I who know nothing about the game?
Now we must examine this opinion which I have attributed to him, to see some of its logical characteristics. The opinion is:
The best shmakum is the one that I could make the most smashes with.Let us call this sentence, A. And let us notice, first of all, that A does not mean the same as the following sentence, which I will call B:The expression 'the best shmakum' means 'the shmakum that I could make the most smashes with'.For if I said that he thought B, I should be attributing to him an opinion which, in his circumstances, it would be very odd for him to hold; for it is an opinion about the equivalence in meaning of a word ('best') and a phrase ('that I could make the most smashes with'); and since he does not know (nor even think he knows) the meaning of the word 'best', how can he have opinions about what phrases are equivalent to it ?Let us detail the position again; I know the meaning of the word 'best', but do not know the meanings of 'shmakum' or 'smash'; he knows the meanings of the latter expressions, but not of 'best'. And so neither of us is really in a position to say B. But I have said that he thinks A; that is to say, I have attributed to him an opinion, not about the meaning of words, but about what, as a matter of substance, is the best shmakum -- an opinion which, if either of us had the necessary knowledge of the meanings of the words to be used, could be put into words by saying A.
Moreover, I have now got, by this manoeuvre, into the position of being able to explain to him, in one lesson, the meaning of 'best', and therefore of 'good'. For I have caught him, as it were, having the thought about shmakums for which the appropriate linguistic expression is A. It is a thought which has something to do with choosing or being inclined to choose. The paradoxical feature of this explanation is, that it is conducted with reference to a class of objects (shmakums) the criteria for the goodness of which I do not know. This shows that to explain the meaning of 'good' is quite different from explaining any of the various criteria for its application. The explanation is not, of course, a logical analysis, for we are not concerned in this chapter with logical analyses; but it is at least a sketch of the sort of way in which a person who did not know the meaning of 'good' might be helped to pick it up.
6.6. At this point a superficial observer might misinterpret the procedure which I have used in explaining the meaning of the word 'good'. For it might be said, 'Surely it can now be seen that the word "good" is like the word "red" after all. It refers to a common property, only this common property has the characteristic of being, unlike redness, inaccessible in a peculiar way. It is in fact the property of producing or being in some way associated with certain inner experiences, which cannot be experienced except by the person who is having them; these experiences may be called purposive or preferential, and instances of them are what we refer to as "trying", "aiming at", "preferring", "choosing", &c. Of course', the objection goes on, 'if a word refers to a certain sort of experience, you cannot define it ostensively to someone who has never had that experience; but this is equally true of "red". You cannot define "red" ostensively to someone who has never had the experience of seeing a red object.' The effect of this objection would be to undo all my argument; for I have been maintaining that 'good' is unlike 'red' in that its meaning is independent of the criteria for its application; but if the criterion for the application of 'good' is the having of certain purposive or preferential experiences, it is no longer possible to distinguish meaning from criteria in the way that I have been trying to do. For it might be that it was possible to explain the meaning of 'good' to my foreigner by getting him to have these experiences and then telling him that the word 'good' was properly applied to the objects of them; and this would make 'good' just like 'red' -- for you explain the meaning of 'red' also by getting the learner to have certain experiences and telling him that the word 'red' is properly applied to the objects of them. I require, therefore, to destroy the hypothesis that the meaning of the word 'good' is fully explained by saying that it is properly applied to the objects of certain recognizable experiences. It is worth while noticing here that this is a familiar theory in relation to 'good' in moral contexts; for it is sometimes said that in these contexts we tell whether or not to apply the word 'good' to an object solely by observing whether or not we have certain experiences towards that object -- experiences, for example, of 'moral approval' or of a 'sense of fittingness'.
We must notice that at the crucial point in the last stage of my explanation of the word 'good' to my foreigner, what happened was this: I learnt from him that all other things being equal he would choose a shmakum with which he could make the most smashes; and on the strength of this, I told him that he thought that the best shmakum was the one with which he could make the most smashes. I did not tell him on the strength of it that the best shmakum was the one with which he could make the most smashes; and this point is of fundamental importance. For, while it may be the case that if I know that X would choose, all other things being equal, the shmakum with which he could, &c, I am more or less safe in saying 'X thinks that the best shmakum is the one with which he could, &c.', it is by no means the case that I am safe in saying 'the best shmakum is the one with which X could, &c.' For suppose that my learner misconstrued my remark and thought that he could correctly apply the word 'best' to any member of a class which he, as a matter of fact, would choose, all other things being equal. Then suppose that I ask him to tell me which is the best of a number of hockey-sticks; he may choose the one with which he, as a beginner in the game, could miss the ball least often, and say 'This is the best one; I know this is right after what you have told me about the word "best"; for this is the one that I would choose'. But then I have to explain to him that he has made a mistake; for that he would choose that hockey-stick shows, not that it is the best, but that he thinks that it is the best.
What the learner has done can be made clear as follows. He has gone on, in spite of adverse experiences, assuming that criteria and meaning are the same. Therefore, having quite correctly learnt, from my previous remark, that the thought that he had when he chose or was inclined to choose a certain sort of shmakum was correctly expressed by saying that the best shmakum was, &c, and that having learnt this, he had learnt the meaning of the word 'good', not only as applied to shmakums, but as applied to anything else, he naturally thought that he had also learnt something about the criteria for applying the word. But in fact he had learnt nothing at all about the criteria for applying the word. For about the criteria for shmakums he knew already, and about the criteria for other things he was no whit the wiser, since they are all different from the criteria for shmakums. What he had learnt was the meaning of the word, and nothing about its criteria. And since criteria are different from meaning, it would be perfectly possible for him to use the word in the full knowledge of its meaning, but, through ignorance of the right criteria, apply it to the wrong objects. Thus, even if he had not misconstrued my remark in the manner described in the last paragraph, and even if he had correctly learnt the meaning of 'good', he might still say 'The best hockey-stick is the one with which I can miss the ball least often'; and in so doing, he might be using the word 'good' correctly to express the thought that he had about hockey-sticks -- namely, his choosing or being inclined to choose such a hockey-stick; but he would, of course, be choosing a sort of hockey-stick which we, who know the criteria for choosing hockey-sticks, know to be a bad sort of hockey-stick to choose.
It is, moreover, not necessary to believe in 'inner experiences' in order to commit the confusion to which I have just referred. The same mistake could be made by one who interpreted the word 'choose' entirely in terms of 'preferential behaviour'. The fact that a person, or set of people, behave preferentially towards a certain member of a class is not in itself a necessary or sufficient condition for saying that it is a good member of the class; it is only the most important of the many things that might make us want to say that they think that it is a good member. Suppose that we are investigating the meaning of 'good drink'. We find that Americans behave preferentially towards coca-cola, and Russians towards vodka, and that they apply to these drinks, respectively, the word 'good' and its Russian counterpart. But this is no indication of a distinction in meaning between the English and the Russian word. It merely shows what sorts of drink Americans and Russians think good; it is a help towards discovering the criteria of good drink current in America and Russia respectively. Needless to say, this confusion is not confined to the question of good drink -- all those who think that they can find out what 'good' means by studying preferential behaviour are destined to discover the following infallible guide to conduct: that they should go on doing just as they are doing, or as most of the people whom they study do.3
It must be explained that I have ignored so far one common sense of the word 'mean' in which it is obviously untrue to say that meaning is different from criteria. Suppose that I had finally succeeded in teaching my foreigner what 'good' meant in the sense of that word that I have been using so far, and to celebrate the achievement we went for a tour to his country and turned out to watch a game of smashmak. Then suppose that he said to me 'That fellow just going out on to the field is the best smashmak player in our country'; I might ask 'How do you mean, the best player?' and he might reply 'I mean he always scores the largest number of smashes'. Here it is obvious that what I was asking for, and was given, were the criteria for calling him the best player; I might equally well have said 'What makes you call him the best player?' And it is also obvious that I could not ask 'How do you mean, the best player?' (in this second sense of 'mean' in which it has to do with criteria), unless I already knew what the expression 'the best player' meant (in the first sense of 'mean' in which it has nothing to do with criteria). I do not wish in the least to deny the existence of this meaning of the word 'mean' in such contexts; it has indeed, through being confused with the other sense of the word on which I have been concentrating, been responsible for most of the trouble that I have been trying to clear up.
Notes 1 Reason in Ethics, ch. ii.
2 Nicomachean Ethics, 1097a23, 114b12.
3 See, for further discussion of this point, my review in Mind, lx (1951), 430, of Value, a Co-operative Enquiry, ed. Ray Lepley.