The process of argument; a contribution to logic

Preface

This book, like my former ones, has for its aim the extension of a knowledo^e of the more useful parts of Logic. It is written for those who are interested rather in the war against fallacy than in the grammatical inquiries which form so large a part of the Logic taught in the text-books.

Some care has therefore been taken to use words as far as possible in their everyday sense. Wherever it has seemed more convenient to depart at all from the commonest custom, reasons are given and the departure is left optional. No attempt is made to force the reader to accept hard doctrines or strange definitions, which are not yet his own.

vi THE PROCESS OE ARGUMENT

It is specially in regard to the meaning of technical terms that this negative mode of treatment shows itself. In Logic, as in other subjects, the leading terms are capable of better and worse definition, and it is not unusual to find that doctors differ on the question which definitions are best. In all such cases the aim of this book is to help the beginner to improve his own first notions for himself, rather than to get him to accept anyready-made ones which happen to seem satisfactory to some particular school of thought.

Appendix A is part of an article contributed to Mind, which the Editor kindly allows me to reprint. Special thanks are due to Mr. Carveth Read for the many improvements he has helped me to make throughout.

Contents

CHAP. PAGE

1. Introductory ...... i

2. False " Facts " AND False Inference . 12

Appendix

A. Notes on the Technicalities of Logic

B. Structure of the Syllogism

C. The Universality of the Major Premiss

D. The Typical Form of Syllogism

E. Essential Resemblance and Difference

PAGE 199

226 228 230

Chapter I

In the following inquiry we shall purposely avoid all direct search for the best definition of the word "Logic," Though this book is certainly meant as a contribution to Logic, it is not meant to contribute to it in that particular way. Interesting though such a question may be, its importance is of a very special and limited kind ; and at any rate there is no need to wait for a perfectly comprehensive and final idea of the limits of a subject in order to make a beginning with it. Knowledge of Logic, like other knowledge, may be acquired by degrees.

Some general guiding idea, however, will probably be of service, for the sake of putting our views together and regarding them as means to an end. For this purpose the broadest and commonest notion of Logic is quite sufficient. Let us say, for instance, that Logic is all about arguments, and helps us to distinguish sound arguments from unsound ones ; or that Logic attempts to supervise our methods of judging, or of convincing ourselves that a given assertion is true or false. No one can quarrel v^ith these loose definitions except on the ground of their incompleteness, and to us their possible incompleteness does not matter. The process of argument is just what is here to be discussed, and especially with a view to the distinction of sound argument from unsound, We are to inquire in what way any reasoned belief lies open to attack ; and so to survey objections generally, — the objections that can be brought against the truth of any disputable belief, or "judgment," or assertion.^

1 The words "belief," "judgment,"' and "assertion" may here

The process of forming judgments is, for human beings, almost inextricably entangled with the process of criticising the judgments as they are formed. The question whether the lower animals "judge" at all depends on what we choose to mean by the word. It is plain that many of them distinguish persons, and foods, and recognise signs of danger, or signs of what they are seeking. But at any rate they judge less deliberately and critically than we do ; so that, to speak broadly, they do not stop to weigh their judgments or to reflect upon the likelihood of error. And apart from any other and deeper difference between ourselves and the beasts, it is the process of conscious reflection upon our judgments that chiefly makes our mental operations distinct from theirs. For good or ill, our judgments are on the whole more

be taken as synonymous. There would be no harm in carefully making differences of meaning between them, but we are not here concerned to do so. Our plan, as already said, is to adopt the wide ordinary usage of words wherever we are not actually driven to de^Dart from it.

guarded, less impulsive, more conscious of the victory over doubt.

This weighing of judgment, this balancing of reasons for and against the truth of a belief, this awakeness of our critical faculty, is here to be included under the notion of "argument." We shall not restrict the word argument to mean only disputation between two parties, but shall take it in the widest V possible sense. At any rate the process which is here to be discussed occurs in the mind of an individual, as well as where two individuals are disputing ; and we may call it the process of argument in default of a better name. So understood, there is argument wherever an inference is critically drawn, or wherever a judgment is critically formed, even when the criticism against which it stands firm proceeds from our own critical faculty in the absence of any opponent. Still, the process can best be observed in cases where there is a conflict of opinion, and therefore there is some convenience in drawing illustrations chiefly from argument in the narrower sense.

Objections against an assertion, on the score of its truth, may be divided under two heads, — the objection that it is based upon false " facts," and the objection that it involves a false inference from facts that perhaps are true. For a reason which is explained in the next chapter, it is enough for Logic to concern itself with the latter form of objection only.

Our plan therefore will be, first to discuss the nature oi Inference generally, — meaning by inference the reading of signs ; that is to say, the inferring of one supposed fact from another or others. There is also a different process of " Inference" usually recognised in books on Logic, namely, that of reaching a conclusion by means of merely verbal transformations. From a given sentence — say, "All men are mortal" — we may infer certain other sentences to be true ; for in-

Stance, that " No immortals are men," or that " Some men are mortal," " Some mortals are men," and so on. Or, again, we may often take two sentences, and by leaving out a part of both of them, arrive at another sentence (called the " conclusion ") whose truth is involved in theirs. To take again the hackneyed example, we may draw from the two sentences "all men are mortal," and "Socrates is a man," the one sentence "Socrates is mortal." There is room for much ingenuity in following out the laws of this kind of inference, and in speculating on the grammatical and other questions suggested by the study of them. But this function is admirably performed already by a host of books too numerous to mention, and too complete to leave room for much improvement. We shall therefore here turn our backs resolutely upon all inquiries into the proper meaning of forms of sentence, singly or combined, and assume in the reader just that knowledge of ordinary grammar which the general reader is likely to have. Some error does, no doubt, arise in practice from our occasional failure to see equivalent meanings in different forms of words, but the chief source of actual errors in judgment certainly lies elsewhere, — namely, in our limited acquaintance with the facts and the laws of Nature. We infer one "fact" wrongly from another, for the most part not through failing to understand the full meaning of a sentence, or tangle of sentences, but through failing to understand the full meaning of what we observe, or take as true. We look at the surface of facts and judge of their nature hastily ; we leap to conclusions without testing them sufficiently ; we generalise on scanty data. The question then arises. What can be done towards combating errors in our interpretation, not of sentences, but of facts that are taken as true ?

So comprehensive a question admits of various treatment, according as we make it more or less elementary. The line of treat-

8 rilE PROCESS OF ARGUMENT

mcnt here adopted is as follows : inference, we fmd, always involves generalisation. That is to say, we never leap from observed fact to unobserved, except by the help of our knowledge of Nature generally. If the sight or smell of smoke leads us to infer the presence of fire, that is because we believe fire and smoke to be causally connected ; if the falling barometer leads us to infer the coming storm, that is because we believe there is some general rule, loose though it be, as to their concomitance. In short, to connect any two facts as sign and signification is to believe that they are connected by natural la\v. Inference is only an aspect of generalisation, though no express and conscious generalisation may, in a given case, be formulated. In the entire absence of supposed general rules, any fact would be wholly meaningless to us ; it would be strictly siii generis, and therefore unintelligible or monstrous. Thus an apparition is feared because we cannot explain it or foresee what will happen next.

Since, then, our inferences from fact to fact depend upon our belief in general rules of connection between fact and fact, — generalisations about the way things happen in Nature, — the work of criticising inferences resolves itself into that of criticising generalisations. Of any given inference we have first to ask what generalisations underlie it, and then what can be done towards establishincr these as true.

The former aim — the discovery of the underlying theories — is discussed in Chapters V. and VI., and indirectly in other parts of the book. It has been my aim to show how meagre a conception of it is provided in the traditional logic, which assumes that the words used are free from ambiguity, and that every conclusion rests on a single generalisation. In actual argument ambiguity is seldom or never wholly absent, except just where no y dispute or doubt arises. As we never infer one fact from another except by virtue of generalisation (tacit or express), so we always

10 TI/E PROCESS OF ARGUMENT

(dimly or clearly) sec the fact from which the inference starts as such and such a kind of fact ; and in seeing it so we put interpretations upon it. It is only so far as we have a perception (true or false) of the nature of the fact observed, that it leads our imagination backward to past times or forward to the future. Therefore, to find what theories underlie a given inference is to find at the same time how the observed fact is conceived, — what sort of description of its nature is regarded as satisfactory. False conception of the facts of a case is perhaps the subtlest of all sources of error.

Besides, it is very far from true that every judgment rests on a single generalisation. It is only true of the least disputable judgments. As a rule, a judgment rests on a large number of pieces of generalisation, better and worse in quality. Thus the notion chiefly to be guarded against, in the search for underlying theories, is that we can find them at once by mere inspection of the words used,— just as, in the traditional logic, we can find a missing "premiss" with certainty. The reasons we can expressly put forward for our beliefs give always a somewhat incomplete account of the sources of our conviction ; in order to find them fairly we must treat the record leniently, and give some play to our imagination. The search for underlying theories is thus always tentative and open to some correction as we learn more and more to read between the lines of verbal argument.

From Chapter VII. onward we are free to turn undivided attention to the nature of the process by which we generalise from facts observed and criticise our generalisations as they are formed. The results are summarised in Chapter XIII., and a few notes are afterwards appended for the use of students of Logic.

CHAPTER II FALSE "facts" AND FALSE INFERENCE

If we take any plain example of an error of "fact," it is easy to see what was meant, a few pages back, by saying that faulty inference from fact is all we need here discuss.

The clearest and simplest of all such examples may be found in conjuring tricks. We see, let us say, a watch destroyed by the conjuror, and the same watch is afterwards restored uninjured. Those who are acquainted with the ways of conjurors would of course not take the destruction seriously ; they would know there was some deception. But it requires no great imaginative effort to put ourselves into the position of a less sceptical person — say a child — and to perceive how

FALSE ''FACTS'' &- FALSE INFERENCE 13

stronof mio-ht be that child's false belief that the watch was irretrievably damaged. The false belief would rest upon a " fact " that occurred before the very eyes of the child. From the fact as seen, the inference is correctly drawn. But the " fact " was as false as a fact can be — that is to say, it was misconceived. There is no other possible kind of falsity from which a fact can suffer ; in every false fact there is a basis of real fact and a superstructure of faulty theory, — faulty inference. The watch was really handed to the conjuror ; some object like it was perhaps really destroyed ; but the falsity of the fact consists in our false inference (or theory) that these two objects were the same.

It is because of the intricate way in which fact and theory are mingled in all our so-called facts that we find it so difficult sometimes to make a satisfactory use of the common distinction between assertions of fact and assertions of theory. Whatever value this division has, it does not enable us to set aside a distinct class of " assertions of fact " as above dispute. A vaguely-distinguished class of this sort we are all accustomed to recognise. There are plenty of cases where we all in practice agree that a settlement of a question may be reached by some short and easy method such as using our eyes, or referring to an authority. But to admit the existence of such cases, or even their frequent occurrence, is a very different thing from specifying exactly which they are. That no man can do, even to his own satisfaction. No clear line can be drawn between questions that are easy to settle and those where a settlement is not so easily reached. Whatever is seen or heard, is seen or heard in spite of obstacles either in the observer himself or in his surroundings ; increase these obstacles — distance, for example — and the facts as seen or heard need more and more interpretation. Our state of mind, or of body, or the amount

FALSE ''FACTS" fi- FALSE INFERENCE 15

and kind of our previous information, will often largely influence our sight and hearing. What one person will call the sound of clanking chains may be described by another, better informed beforehand, as due to a rusty weathercock ; what one person will call the sound of a muffled footfall on the stairs behind him, may be described by another, less alarmed, as due to a trailing tassel. And from these weakest '' facts " upwards the same possibility must be faced, so far as we care to criticise the judgment carefully. As soon as a fact is named, or described, or conceived in this or that way, it is seen in the light of theory, and the theory may be mistaken.

There is no need for us here to press this doctrine to its utmost possible extent, if that should seem unpractical. We may freely admit, for instance, that a normal person's eyesight and hearing are for the most part trustworthy. We are only concerned to notice the impossibility of drawing

1 6 rilE PROCESS OF ARGUMENT

any clear line of division between the class of questions which are easy to settle by appeal to the senses, and those which are less easy. This being granted, we meiy go on to observe that, so far as we aim at discussing the objections that may be raised against assertions generally, our interest lies rather with those "assertions of fact" which are nearest the doubtful line than with those that are farthest away ; for It is precisely where truth and doubtful matter — fact and theory — are most intimately mingled that different opinions come into closest collision. Almost any case of divided opinion, on a matter of history, would serve as a fairly typical example ; for instance, the question whether Edison invented the microphone, or whether Byron ill-treated his wife. Such assertions evidently contain a mixture of admitted truth and doubtful matter. They have always a more and a less disputable part. There is no dispute, for instance, that Edison did invent something very like a microphone, and that

FALSE 'TACTS" &- FALSE L\FERENCE 17

Byron acted in such and such a way. The doubt in the one case is whether Hughes' microphone was a " real advance '" upon Edison's instrument, and in the other case how far Byron was to blame. The fair apportionment of blame in a quarrel is notoriously difficult ; and since one invention always grows out of another, and contains the germ of many future developments, it is often hard to apportion the honour fairly. If we wish to discuss such questions, we must first separate their disputable from their indisputable part. \\^ must admit somethino- within them as "fact'' while we class the rest as mere opinion or theory.

Precisely that is true of any assertion, whether its maker happens to call it a " fact " or not. In order to dispute it, we must regard it as theory based on fact ; we must separate within it a more and a less disputable part, — a part which we are at any rate willing to admit to be fact fo7' the purpose of argument, and another part which we declare to be mere faulty opinion. This does not mean, of course, that we can never reject an assertion as "wholly untrue," but that when we do so we are no longer arguing. For the sake of argument a basis of fact must always be admitted. Before this is done, our objection is like the celebrated objection to Dr. Fell. It may be just or not, but our opponent has no opportunity of learning its justice — no opportunity of knowing even what the objection is. In practice, assertions are seldom or never wholly untrue, especially when they correspond to a genuine belief. It is sometimes difficult, but seldom impossible (if it be thought worth while) to find a basis of agreement even with those whose view seems most opposed to our own. Some part of the way they go with us, and then the roads branch off. Why did they leave our road, or why did we leave theirs ? That question and its answer is the beginning of argument. The matter of argument is always matter of opinion ; not fact but theory ; not fact but inference from fact.

FALSE ''FACTS" &- FALSE LNFERENCE 19

That, then, is why a discussion of inference, its nature and its dangers, includes a discussion of all the definite objections that . can be brought against a judgment. The process of objection is the same in every case. So far as we find any definite fault with a judgment, we regard it as a case where admitted facts are mistakenly seen or interpreted. Sometimes the assertion comes before us openly as a theory, but the process of criticism is the same where the assertion professes to tell us simple facts. It often happens that a so-called fact contains some highly disputable matter, and the whole examination may be stultified if we are content to pass this disputable matter as simple fact. The believer in ghosts, for instance, will sometimes in telling a marvellous story profess to keep entirely to facts, and " leave you to find an explanation of them for yourself," and yet his facts may be full of hidden theory. So far as we criticise them we criticise not the facts themselves but the theories involved in them, — involved, that is, in the manner in which they are described, or conceived, or seen.

Whatever other subdivisions we may choose to make, for one purpose or another, within the class of assertions generally, the division into assertions of fact and assertions of theory covers all the ground. No assertion can be imagined which is not one (or both) of these. And, as we have seen, the so-called "assertions of fact " are a class that exist only on sufferance. Many assertions, disputable and disputed, are indeed called by their makers "assertions of fact," because mankind are prone to mistake their own theorised facts for simple facts. And then what we find is that within the assertion itself a more and a less disputable part are always to be distinguished. This disputable part is the pith and centre of interest of any question. The first step in argument is to separate a kernel of admitted fact from a husk of unadmitted theory.

Chapter III "

All our disputable judgments thus contain a mixture of fact and inference. They are only partly disputable. The disputable part we regard as based upon the indisputable part, as an " inference " either warranted or not warranted by the fact on which it rests.

The use of facts. to warrant inferences is very familiar — our habit of taking certain facts as signs of certain other facts ; smoke as a sign of hre, for instance.^ Our minds are full of such "generalisations," or grounds of inference, ranging from those of the most vague and tentative character upwards. And

1 This example has tlie merit of antiquity. It was used in the typical syllogism of the Nyaya (Hindoo) philosophy.

always, in regard to any generalisation that we use, the critical question is, How far may we safely trust it, or under what conditions and limitations may it be taken as true ?

Generalisations vary very much in the use that may safely be made of them. With some — like " where there is smoke there is fire" — hardly a fault can be found, for ordinary purposes ; but with plenty of others it becomes a really difficult matter to decide how far we may trust them. Take, for example, the knowledofe we have about classes of people, — say dissenters, or Jesuits, or peers, or working men. There is always some truth in the generalisations that are made about such classes, but their value as grounds of inference is often exaggerated ; merely because a man belongs to a certain class we cannot with any safety infer that he possesses all the usual class-characteristics. Our ordinary loose generalisations have many exceptions, and in the absence of reasons for knowing- the contrary any given case may be an exception and not come under the rule.

Logic sometimes sets out from the fact that only a perfectly " universal " generalisation can properly serve as ground of inference ; that if a rule be admitted to have even a single (unspecified) exception, its value for inference is lost, since any given case may be that one exception in the absence of knowledge to the contrary. And for some purposes ^ this view is useful. But we are now to look at another side of the truth, and one that has a closer connection with the actual process of argument.

It is comparatively seldom in actual argument— never, perhaps, where a really disputed or difficult question is raised — that we are able to rest our case on a single faultless generalisation, like " all men are mortal " or "where there is smoke there is fire." Inferences so supported are not in practice the kind that encounter opposition. Where any

^ See Appendix, C.

doubt exists, our express or apparent ground of inference is, nearly always, a looser kind of generalisation ; we are obliged to make what use we can of broad truths which we know to be incompletely universal. Yet we are not, as a matter of fact, compelled to stop at the fact of their incompleteness. There is a roundabout way by which we can and do to a great extent remedy their defects.

This consists in the claim that for some reason the exceptions to the rule do not include the particular case in question. Although there are plenty of exceptions to the rule, for instance, that extravagance leads to ruin, or familiarity to contempt, yet now and then in a special case such faulty generalisations are used with confidence and safety, because we see both the rule and the special case in the lioht of other knowledoe. The case in question is not merely " extravagance," but a kind of extravagance which specially deserves the name ; the money spent is not, we believe, in this particular instance a means (say) of supporting credit and so of producing more income ; it is not a form of insurance ao-ainst ill- health ; and so on. And in a similar way it is only under certain conditions that familiarity breeds contempt, and yet these conditions are capable of being approximately known.

Whenever we appear to use something less than a perfectly universal rule as ground of inference, that is, in effect, the way in which we justify our procedure. The generalisation as stated, or the s^eneralisation that seems on the surface to be implied, is not by itself relied upon as sufficient, but we use it merely to suggest the true ground of inference, which is perhaps too complex to state conveniently in any short and telling form. The great majority of our commonest actual judgments, true and false alike, are based on evidence that cannot well be stated fully and yet concisely, if at all. What satisfactory account can be given, for instance, of the marks by which we recognise a particular

26 THE PROCESS OF ARGUMl'.NT

key on a bunch, or a friend in the distance, or somebody's footstep on the stairs ? Such judgments as these are not, indeed, commonly matter of argument between two persons, but precisely the same sort of difficulty occurs even in the most openly argumentative matter. It is hard, for instance, to give a complete account of the reasons why we distrust our political opponents. The reasons we give, on a question so tangled, can only be a fraction of the real grounds of our judgment. When pressed to account for our faith, we often find it necessary to confess that even the best reasons we can put forward are sketchy, suggestive, and incomplete.

To this defence, however, the critic has an easy reply. If the ostensible ground of inference is not the real one, he says, he would be glad to know what the real one is. Till that is shown, the inference lacks necessity — may be right or may be wrong. It is unfortunate if the real oround cannot be given, for then how can we possibly judge its value ?

Often enouQrh a deadlock is reached in this way. Often no convincing reason for a belief can be stated, and yet the believer remains convinced, rightly or wrongly. But in the fortunate cases where some progress is made by means of criticism, such progress depends on the fact that more and more of the real ground of inference is stated, or made ostensible ; and this may take place in several different ways.

One way is closer specification of the rule ; that is to say, a more definite statement of the class to which the generalisation is meant to refer. Not every kind of X, we grant, permits the inference of Y, but only some kinds ; or X under certain conditions only. Our notion of the character of any class is nearly always too sweeping at first, and gradually becomes less so, as we learn to discriminate within the class. Sometimes, even after the process of discrimination has begun, we continue to use the wide class-name for brevity, as where we s-^y familiarity breeds contempt, though we only mean some kinds of familiarity, or familiarity under certain conditions. But we may also use the wider name through ignorance or carelessness. For we often get an idea — it may be only from novels, or caricatures, or plays — of the typical form of a class ; say, the typical worldly woman, or family lawyer, or curate, or pa^'venu ; and then we are apt to use the wide class-name when we are thinking only of these selected and pungent specimens. And the criticism which forces us to recoQ-nise the number and kind of exceptions to our ostensible generalisation performs in this way a useful service to the cause of truth. The more careful statement of the rule helps us to see whether the case in question belongs to the rule or not.