XVII—ITS SCOPE

The derivation of Reason as given in the preceding sections may be summed up thus:—the meeting of Minds gives Perception or primary experience; Attention selects therefrom objects of special interest to the observer; Memory retains impressions of these in the mental plasma, by which ideas of them are recollected though the originating mind be not present; community with divergence of imprint gives rise to Comparison; from this are derived Imagination and Generalisation; from imagination emerge Reason and Art.

Generalisation is thus only a collateral relation of reason, not its immediate parent nor in the direct line of descent. It is not essential to reason, but may enter as a subsidiary process into an argument. If the things we argue about are numerous it will be more correct to generalise them and then argue from the general idea, than to argue from one concrete object to another. But innumerable inferences are drawn from one particular thing to another, and these involve no generalisation.

Reason is chiefly the art of predicting by means of the intellect what will occur to us in the future. Its use is to enable us to prepare for future events in so far as our resources permit. We never predict quite accurately, but general preconceptions are better than none at all. The same process by which we preconceive the future can be applied to the conception of what is actually taking place but not within our ken—as at the antipodes—and can be applied also to events that took place in the past and will never be experienced by us. It might be objected that as regards the past we can have no motive in imagining it, seeing we can never experience it. But a conception of the past is often a necessary condition of our conceiving the future, and is artistically interesting. It awakens pleasing emotions to be able to picture to ourselves, even imperfectly, states of the world and of society that have long been obsolete.

An investigation of the manner in which reason supplies us with ideas of the unknown, involves the consideration and arrangement of so many details that it may be regarded as a small science in itself—Dialectic.

A dialectician (logician in the narrower sense) is neither a grammarian nor an encyclopedia of the best information on every subject. His office consists in deciding whether certain theorems are arguments or not. An Argument is an act or product of rational imagination. Theorems which purport to be arguments, but are not, are Fallacies.

A fallacy is not merely a bad argument—it is no argument at all. Quite apart from fallacy there is a goodness and badness in arguments, but with this discrimination the dialectician (as such) has nothing to do. Only persons experienced in the matter are competent to decide between good and bad arguments. Hence when the quality of an argument is in question the dialectician takes no part in the debate: he is neither combatant nor umpire. He is at most an impartial president whose chief duty is to see that people do not debate about mere words and foregone conclusions. Granting that a theorem has the qualities of an argument, the dialectician is not competent to say that it is improper or too trivial to be discussed. He is not a judge of what people ought to be interested in.

From his better knowledge of what constitutes rational prediction, a dialectician may offer his services to disentangle and render explicit involved and partial arguments. Many people reason well who are yet unable to express themselves coherently. A dialectician should be able to reconstruct an argument from the slightest hint, as a naturalist imagines an animal from a single bone. In ordinary reasoning the arguments are seldom fully expressed, and the reasoners themselves are not always quite conscious of the premises from which they argue. All such suppressed and overlooked assumptions should be brought to light by dialectic, the aim of which is to render reason as self-conscious as possible.

Though a dialectician need not be an expert in any department of knowledge, he must know the facts on which an argument is built, otherwise he may be deceived by equivocal language. Reverting to the instance of the vase—the dialectician must have seen both the whole vase and the broken vase, but he need not have any opinion as to whether the proposed handle is the most suitable, or not. That must be left to those who are familiar with vases and who are interested in the restoration of the one in question.

The definition here given of the scope and office of dialectic may appear to some too modest. But in reality there is a great deal involved in it. Philosophers have been discussing Reason for twenty centuries or more, and have not produced a satisfactory definition of it. Consequently they cannot decide with any confidence whether a theorem is an argument or a fallacy. The cleverest of them give their sanction to theorems that are demonstrably fallacies. They are evidently judging more by ear than by rule. All this causes confusion of mind and waste of energy.

Dialectic takes its general idea of reason from the higher analysis of logic, and brings the general idea to bear on concrete arguments. A dialectician makes a collection of theorems for study just as a botanist makes a collection of plants. He sorts them out into convenient classes, separates the valid or useful from the erroneous and misleading, studies the relation of language to argument and the influence for good or ill that words have upon rational thought.

From the example of the vase cited above it will be seen that in every act of reason two principal things are requisite. There must be something wholly known (or comparatively well known) and something less well known, and the reasoning or argument consists in ideally completing the latter on the model of the former. If we would predict the coming of a future season of the year we must have a picture in the mind of all the seasons in the order in which they occur. If we would go straight to a place on the surface of the earth we must have a plan of the way in our imagination. If we would predict the effect of a drug on an animal body we must have previously noticed the effect it has produced—and so on. Neither the mind nor intellect supplies spontaneously any of these models; they are all formed out of actual experience remembered and recollected. When they have been refined into extremely general ideas they are apt to be taken for innate tendencies of the intellect, as Kant erroneously thought. They are not so; all we know of the intellect is consistent with the belief that it begins with pure plasm without a trace of idea, and is absolutely indifferent to the imprints it may receive. Doctrines of innate ideas—innate forms of thought or categories—innate 'principles' of various kinds—are devices of metaphysicians to cover the weakness of their theories.

The two main parts of an argument divide naturally into four subdivisions. There is the thing argued about (corresponding to the broken vase); there is the ideal extension or restoration; in the model we reason from there are the parts corresponding to each of these. I propose to take terms for these four parts from one of the most important, formal and correct modes of reasoning—the application of a precedent or statute to a case in Law.

XVIII—THE RATIONAL PARALLEL

Every argument, whatever be the matter of it, consists in bringing a Case under a Precedent, and applying to the case ideally the better knowledge possessed of the precedent. The Conclusion (also called Inference or Deduction) is the result of this application, and is always an addition to our stock of ideas.

A conclusion has never the same reality as actual experience. It is not 'true' in that sense, though it may be 'morally' true, that is, we are ready to act upon it without hesitation—to stake our life or fortune on it. As regards actual or experienced fact there can be no argument, since it is useless to 'predict' what we already know.

Academical logicians have a doctrine the reverse of this. They assert that their syllogisms yield conclusions that are always as certain as the premises. Grant their premises and you are obliged to accept their conclusions. This is so, because a regular syllogistic conclusion is simply a restating in other words of the information, or part of the information, already contained in the premises. If the syllogism has any use at all, it is merely as an aid to recollection; no new idea is generated by it. It is needless to insist on a fact so notorious as that ordinary rational conclusions—those that form the staple of our daily thought—are not by any means so certain as the data from which they are drawn. For example, the sky is red and lowering this evening, and we conclude therefrom that the weather will be bad to-morrow. There is no doubt about the present aspect of the sky, but much doubt about the inference.

The form of an act of reasoning or argumentation may be rendered plainer by a diagram.

S	A
C	I

S A represents the precedent. S is the Subject or body of the precedent; A (the Applicate) is one property, or a part, or a relation of S abstracted from the rest to illustrate a case. C is the case; I is the conclusion (or inference). I results from imagining C to be associated with a property or relation similar to A. The sum of our I's constitutes what we know of the world and man before we were born, of what is taking place in other parts of the world or universe, of what may take place in the future, and of the concealed and inaccessible parts of present objects. This is true not only of the results of our own reasoning but of what we have learned as verified knowledge from others, for the interpretation of language is, in the last analysis, a rational conclusion.

All the parts of an argument exist in the mind, but they are not always expressed in language. When treated dialectically the implicit members are expressed, and the terms arranged so as to show as clearly as possible the nature of the argument. The following are the points most necessary to be observed in constructing or analysing an argument.

(1) C must resemble S, for that is the basis of the argument. If C is not felt to be like S, or (as sometimes happens) is explicitly declared to be unlike S, there can be no conclusion. The precedent is not applicable to the case. A may, or may not, be associated with S; that is to say, a verbal negation may appear in the statement of the relation of S to A, but there must be no negation with respect to the relation of C to S.

The resemblance of C to S may, however, vary in degree from the faintest analogy to community of species. The difference between them may far outweigh their resemblance. There may even be no material likeness, but only a similarity of function, or position, or of any the most trivial attribute. Only it is to be observed that the kind and degree of resemblance between S and C determine the kind and degree of resemblance between A and I. We must not infer specifically unless the case is specifically like the precedent. In all other instances we can only infer proportionally or by transfusion.

(2) None of the antecedents must be a verbal or identical proposition, that is, a proposition which merely substitutes one name or nominal phrase for another; nor must the case be merely the precedent expressed in other words, or the precedent a paraphrase of the case. In any of these circumstances one of the elements of the argument is wanting; we have two names for one thing or two propositions giving the same information.

(3) The precedent may (as has been already remarked) be a general idea, or may be an individual idea or object. If S A has occurred frequently it is certain to be generalised, and so may form a maxim, a law, a rule, an induction, &c. But one well-observed precedent is enough to suggest a conclusion, if there has been no experience to the contrary. There is therefore no dialectical difference between arguing from a general idea (class notion) to an individual or subordinate idea, and arguing from one individual to another. Comparison and inference occur in both.

(4) After separating A from S care should be taken that it is A and not S that is used to generate I. Examples are plentiful of theorems in which S and A change parts, which invalidates the conclusion. Other errors in stating theorems intended to be arguments will be noticed under the head of 'Fallacies.'

The following is an argument conformable to the above rules.

Tyrants	deserve death
Caesar was a tyrant	no doubt he deserved death

This square mode of stating the argument is adapted from the general type, and brings out the mutual relations of the compared parts better than the three-lined arrangement. The word 'therefore,' which usually introduces a logical conclusion, is ambiguous. It may mean that the antecedents are the causes of the fact mentioned in the conclusion, or merely that the antecedents are the reasons why we believe the conclusion. The former is the scientific 'therefore,' the latter is the purely dialectical. I shall generally omit the illative word, and print conclusions in italics, besides entering them invariably in the fourth compartment of the parallel when this arrangement is adopted.

An idea once generated in the intellect is not to be erased at pleasure. It can be obliterated only by the process of forgetting. If after we have formed a dialectical conclusion we meet with evidence that contradicts it, the only result of that evidence is to affix a mark of falsity to the conclusion, so that as often as it is recollected the stigma is recollected too, and neutralises the effect of the idea. A negative or destructive argument is thus, plasmically speaking, a positive addition to the idea it seeks to efface. For the time being it renders the idea more conspicuous, as the word CANCELLED stamped in large letters across a document makes it more evident than it was before; but no doubt the stigmatising of an idea hastens the process of oblivion, for we thenceforth bestow less attention upon it. Stigmatic arguments are not another species, but merely the ordinary constructive arguments used for a particular purpose.

Suppose we have inferred from the general resemblance of the earth to the moon that the latter is inhabited, we stigmatise this belief by such an argument as—

Without air	animals cannot live
There is no air in the moon	therecanbenolifeinthemoon

There is an exception to the rule that argument is superfluous when the speaker has already verified the conclusion. It is when he is addressing a person who has not had the same experience as himself and who doubts his word. The speaker may then resort to arguments drawn from antecedents recognised by the hearer, if any such are applicable to the subject. But a fact may be truly reported though neither the witness nor a sceptical hearer can find dialectical antecedents to prove it, for there may be no relation between the fact in question and any prior knowledge they possess, or they may not be able to find the relation.

This brings us again to that view of the intellect which represents it as artificial and limited by experience. Man is rational only on matters familiar to him; in utterly novel circumstances he is irrational, and must fall back for guidance on his general mental sentiment, or the advice of persons more experienced than himself.

XIX—HYPOTHETICAL ARGUMENTS

It is allowable to imagine ourselves placed in circumstances not yet realised, or in possession of information not yet acquired, and to anticipate or rehearse the reasoning we should employ under the supposed conditions. Such arguments take in language a conditional or hypothetical phraseology.

The case may be entirely fictitious, but I cannot find a valid instance of a whole precedent being fictitious. Its dubiety turns on our knowledge or ignorance of the applicate. Has a subject such or such an attribute? Then it may be applied to illustrate a certain case. 'If it is true that Damon and Pythias are inseparable, then Pythias must be in town, for I have just seen Damon.'

It is more often the case that is dubious. 'If Caius is a European he is white, for all Europeans are white.' 'If Damon is in town Pythias is in town, for they are inseparable.' 'If I were you I should defer the voyage to the summer season, as I have always found winter travelling disagreeable.' But the word 'if' does not always mark a hypothetical thought. In the proposition, 'if children are neglected they will grow up ignorant,' we have a dogmatic or assertorial judgment—'neglected children grow up ignorant.' (Bain.)

The precedent may be suppressed in hypothetical as in dogmatic argument. 'If the crops are good, corn will be cheap' implies the unspoken precedent, 'good crops have been invariably followed by cheap corn.' 'If logic is useless it deserves to be neglected,' carries the mind to the more general thesis, 'all useless studies deserve to be neglected.' 'If Great Britain should be invaded the volunteers will be called out,' rests on the precedent judgment, 'it is the duty of the volunteer army to repel invaders.'

Arguments in which both applicate and case are hypothetical are so very dubious that they cannot be considered of any practical use. 'If opium is poisonous, and if this substance is opium, you will be poisoned by taking this substance.'

The Aristotelian hypothetical is almost invariably a fallacy, sometimes on more than one account. It usually consists of—first, a conditional or doubtful statement; next, a solution of the doubt by means of positive information; finally and by way of inference the first statement is given without the doubt. Here is an example from Jevons: 'If the barometer is falling, bad weather is coming; but the barometer is falling; therefore bad weather is coming.'

Where did the information that the barometer is falling come from? If we knew it before uttering the first proposition, we were affecting an ignorance that did not exist. The second proposition takes away all occasion for argument; it is an amendment of the first proposition, and what we get from the theorem as a whole is a case, followed by a prediction for which there is no precedent justification. We are arguing in a circle.

'If Aristotle is right, slavery is a proper form of society; but slavery is not a proper form of society; therefore Aristotle is not right.' If we knew for certain (as the second proposition indicates) that slavery is not a proper form of society, what is the use or meaning of treating the question as hypothetical (as is done in the first)? If we acquired the information after uttering the first proposition, there was no occasion to go on with the argument; we should have said simply, 'Slavery is not a proper form of society, though Aristotle said it was.' It is needless, except for verbal completeness, to say 'he was not right'—we have logically said so.

When two or more alternative data are presented, of which only one is valid or relevant to a proposed argument, but we know not at first which the valid datum is, we have the dilemma (trilemma, tetralemma, &c.) of logicians. In such conditions we have a double process to go through; we must first settle by observation or by an auxiliary argument which of the alternative data to select, and then work out the principal argument in the regular dogmatic form.

Suppose we have to determine dialectically the specific gravity of a piece of metal, but do not know whether it is gold or gun-metal. It is evident we must first somehow make up our mind as to its identity, and then proceed on the usual method of argumentation. The 'making up our mind' is probably itself an argument, and might be of this character—'A piece of yellow metal stamped with what appears to be a hall-mark, is more likely to be gold than gun-metal; this piece of metal has traces of such a stamp; so I conclude it is gold.' Then we proceed to the principal question—'The specific gravity of gold is 19·26; I have concluded that this object is gold; I conclude further that it has a specific gravity of 19·26.'

We may work out all the alternative conclusions first and fix on a datum afterwards, as in deciding how we shall invest our money. 'If I put my money in Consols I shall have a small return with good security; if I buy Patagonian bonds I may have a large interest for a time, but the security is bad.' The next thing to settle is whether in our experience or on accepted principles small profit with good security is, or is not, to be preferred to large profit and bad security: having decided in favour of the former alternative, we now choose our investment dogmatically—'A good security with small profit is to be preferred; Consols are of this character; they are a suitable investment for me.'

We may be unable to decide for any of the alternative data, but we work out all the possible arguments as hypotheses, and so are prepared in a degree for all the possible events. A person is seen approaching our residence, but we cannot discern whether it is A. B., who is a bore, or C. D., who is an entertaining companion. We argue rapidly—'If it is A. B. I shall have a bad half-hour, for he always wearies me; if it is C. D. I shall have an agreeable distraction, for he is very amusing.'

According to the syllogists, the dilemmatic premises are a statement of alternative data and the choice of one of them, and the inference is the rejection of the remainder: or the rejection may be given as matter of fact and the selection as conclusion. In neither case have we argument.

From the moment we select a datum the remaining data are of no import to us, and they need not be mentioned. The selection of one datum is logically identical with the rejection of the rest, and this is therefore not a conclusion from that.—'Do you take tea or coffee?'—'Tea, please.'—'Then I conclude you do not take coffee.'—A person who would 'conclude' in this fashion would be justly deemed irrational. The choice of the tea is a fact, and the rejection of the coffee is the same fact otherwise expressed, so that the rejection cannot be a rational conclusion.—'My doctor sends me off every winter to Nice, Algiers, or Egypt; but I never go to Algiers or Egypt.'—There is no occasion to say, 'therefore you go to Nice'; that has been already announced as a matter of fact and is not susceptible of inference. For the sake of verbal emphasis we might remark, 'So it is to Nice you go', but this is not logically requisite.

Whately's examples of this kind of theorem are exactly of the model just given.—'Either the earth is eternal, or the work of chance, or the work of an intelligent Being; it is not eternal, nor the work of chance, therefore it is the work of an intelligent Being.' This is put forward in all gravity as a specimen of reasoning. It is plain that if we know the premises as matters of fact, we also know the proposed conclusion as a matter of fact. There is no occasion to reason about it.

The Aristotelian hypothetical can be reduced to arithmetical subtraction. Suppose we put five balls into a bag and afterwards take out three without seeing the remainder: is the judgment that two balls remain in the bag a logical inference? No—it is matter of fact. Since we last perceived the objects they have undergone diminution, but that does not confer on what is left of them the imaginary character proper to a rational conclusion. What remains is as much fact—recollected but not imaginary fact—as before the subtraction.

Whately's next example is—'It is either spring, summer, autumn, or winter; but it is neither spring nor summer; therefore it is either autumn or winter.' This is aggravated fallacy. Not only is it mere subtraction, but the remainder is perceived—not recollected, as in the preceding case. The actual season of the year is a known fact, and is not rendered more certain by an inference drawn from the absence of some other season. Arguments have no validity as against matters of fact, and add nothing to their authority. Fact is above, and independent of, argument. The example just cited may be paralleled thus—'The cards in my hand are either spades, hearts, clubs, or diamonds; but they are neither spades nor hearts; therefore they are either clubs or diamonds'.—I see that they are either clubs or diamonds: the perceptual judgment renders the rational—imaginary—judgment superfluous. Reason is intended to supplement experience—not to supersede it.

XX—DEBATE

The purpose of debate is to determine the goodness or badness of an argument by general logical criticism and knowledge of the matter. This is not dialectic, but takes place after the dialectician has declared that a given theorem is valid argument. If then its conclusion is repugnant to us we may seek to stigmatise it—or remove a stigma as the case may be—by going behind the argument to the composition of the judgments that enter into it.

Let us take the case of Caesar being proved to be a tyrant in a society that punishes tyranny with death. There are two ways in which he may be saved or his punishment mitigated.

We are not bound to take the first precedent that is offered from which to generate a conclusion. We grant that Caesar resembles the general notion 'tyrant,' but we ask if he does not resemble in an equal or greater degree some other person or class in regard to whom capital punishment is no just treatment. Does he resemble a 'successful and patriotic general'—a 'benevolent monarch'—a 'wise legislator'—a 'virtuous man'? All these resemblances are compatible with his being a tyrant in some senses of the word. Let us not condemn Caesar for what may be a merely technical offence—the usurpation of authority—if in other respects he is an admirable man. So an opportunity must be given to Caesar or his advocate to suggest other precedents, yielding a different conclusion, by which to complete our imperfect knowledge of the case. Socrates, when he was brought under the class 'perverters of youth'—which also yielded the conclusion 'death'—suggested as an amendment that he should be classed under 'national benefactors,' with the conclusion 'maintenance for life at the public expense.'

It is not enough that we can say of a case that it 'is' this or that, and so proceed to draw the conclusion bound up in that classification. 'Is' in the case means likeness to the precedent, and one 'is' is good only when no better can be found.

If after having weighed the alternative precedents it appears clear that Caesar resembles tyrants more than any other class of persons, the prospect looks bad for him. But there is still a chance of escaping the worst penalty. It turns on the meaning of the word 'all,' which in logic generally introduces a proposition to which no exception has been found—the misnamed and misleading 'universal.'

Logicians do not hesitate to say that in this connection it means 'all possible, known or unknown, past or future individuals of the class.' They suppose, or talk as if they supposed, that at some fixed date in our life we enter into possession of our general ideas, and that no subsequent experience can modify them. Hence the moment it is admitted that Caesar is a tyrant, he is supposed to come under the rule of a stereotyped general idea with inflexible consequences.

This is not quite so. 'All' does not mean 'all possible' but 'all known up to the present time, exclusive of the case under discussion.' Our general or average ideas are the plasmic product of the individuals we have actually known—not a unit more. And as that idea is liable to be modified by every new individual examined, it is possible that on examining Caesar we may find reason to change our general idea, to the extent at least of dividing it into two species, the tyrants who deserve death and the tyrants who deserve some milder punishment, and that we shall resolve to bring Caesar under the latter species. Thus if the idea threatens to hang Caesar, on the other hand Caesar may burst the idea, and his case establish itself as a new precedent. That is how general ideas multiply—by a sort of fission.

In the proposition 'tyrants deserve death' as first proposed, we are dealing with the old general idea, and—as regards all individuals except those from which it was drawn—the proposition is little more than a hypothesis. The idea is itself on trial. Until Caesar is examined we do not fully know how the general tyrant is in future to be defined. Our examination of Caesar is a part of our education on the subject of tyrants. In judging we learn, and the general idea which remains after Caesar is examined is that by which he is to be judged.

If our idea of tyrant remains unshaken after the trial of Caesar, and if he is found to resemble that class more than any other, then—and not till then—are we compelled to pass on him the judgment associated with the definition of tyrant.

An argument based on a particular or solitary precedent is criticised on the same principles. We seek to prove either that the case is not sufficiently like the precedent to justify the application, or that the applicate is not a property of the precedent. If we make good either of these propositions, we prevent the suggested conclusion from being fastened on the case.

The syllogistic dialecticians do not admit alternative precedents or reconstruction of general ideas: their terms and figures are not adapted to express such notions. Hence they cannot evade a conclusion whose premises are correctly given. They have an axiom to the effect that a judgment must be absolutely true or absolutely false—a door must be open or shut, it cannot be ajar; every colour is white or black, it cannot be green or grey, and so on. Now in practical reasoning we may and constantly do admit premises and reject the conclusions they dialectically involve. We look at the question 'from another point of view.' This means that while admitting there is some ground for bringing a case under a certain precedent, we contend that on the whole it is preferable to bring it under another precedent with a different conclusion. The proposed handle may fit the vase somehow, but we think another sort of handle will suit it better. Or—rather than accept an objectionable conclusion—we will divide our idea. This is degree in truth. And that is the elastic method on which we reason in actual affairs. Logicians give a false account of reason, and so their systems are neglected and their authority is never recognised in real debates.