So far as I know, it has been assumed that, if you are not able to swallow down this dogma, you are forced to accept an intolerable alternative. You are given a choice between naked universals, existing as such, and bare particulars. You can not stomach the first, and so you take the last. But why should you take either? Why not adopt the view that the real is the concrete individual, and that the bare particular and abstract universal are distinctions within it, which, apart from it, are only two forms of one fiction? You say, This is unintelligible. But perhaps you never heard of it, or heard of it too late, when you were already compromised, and had no inclination to begin life again. Let it then be unintel ligible; but permit me to add that the view you have adopted calls for something stronger, to back it against facts, than an a priori deduction from a metaphysical alternative.
§ 29. We have shown so far that, in the extension of our experience, there is a synthetic construction by virtue of identity, and that association by similarity has no part in it. We have shown that the test which we bring to inferences, in order to examine their validity, is also the principle which operates in all extension of experience. On our view the origin of the fact is explained, and its existence is at the same time justified. But, on the fashionable theory of Asso ciation, early inferences are made by what afterwards we find to be the essence of bad reasoning. And, to explain the origin of this unjustifiable fact, open fictions have had to be invented.
But not only is Association by Similarity a fictitious ac- CHAP. I THE THEORY OF ASSOCIATION OF IDEAS 33!
count of the reasoning process. It is a fiction altogether; there is no evidence for it at all. And it is to the final proof of this point that we must now address ourselves.
Our previous objections have raised at least a presumption against the alleged phenomenon. Let us now ask, Is there any evidence of any kind which tends to confirm it? I know of none whatever.
We are told (J. S. Mill, Hamilton, p. 315, note) that the elementary case of the suggestion of similars will not come under the head of redintegration. But the answer to this is very simple. Reproduction by mere similarity is a fact which, if real, would certainly stand by itself. Who doubts it? But then the existence of this fact is just what we deny. The general fact that ideas and perceptions give rise to others which are like them, is of course admitted. But this not only can be reduced to redintegration, but long ago it has been so reduced. I will exhibit this in a concrete instance.
§30. I am walking on the shore in England and see a promontory A, and then suddenly I have the idea of another promontory B which is in Wales, and I say How like is A to B. This is the fact which is to be explained. The false theory tells us to explain the fact by postulating a direct connection between A and the idea of B, for it says The suggestion is perfectly simple. But in the first place the postulate de mands an absurdity, and in the second place the suggestion is certainly not simple. If instead of asserting we are willing to analyze, we soon find the true explanation c fact The content of A, like the content of every other percep tion, is complex, and has several elements. Let us say that it has an element of form which is p. Now let us look at B, the idea which is to come up. That also possesses a complex content, and we find in it the same element p, in connection with others, q, r, s, t. These are the conditions, and let us; what follows.,., In the first place A is presented, and so presents p, which by redintegration stimulates the mind X to produce qr. Wl happens then? 1 i-rr.^u Several things may happen, and it is exceedingly difficult to work out the minute psychological conditions which settle 332 THE PRINCIPLES OF LOGIC BOOK II. PT. II the result. But this is a question with which we are not here concerned. One result would be the identification of qr with Ap. A would then be qualified as Apqr, and this would be an unconscious inference. In the present case we are to sup pose that this can not happen; for we suppose that q, r (say a certain colour and a certain size) are discrepant with A. What then may we expect? We might expect that qr would be simply dropped. It might not catch the attention, and the mind might be arrested by a new sensation. We might expect again that, if qr is not dropped, it might be used as a means for a wandering course through a train of ideas, foreign to both A and B, and which might take us anywhere. But we are to assume that none of these possibilities become real; and that instead the idea B rises in the mind. How do we explain this?
Very simply. B (we remember) had a content pqrst, and now we have A which has brought in p, and so introduced qr. But qr will not coalesce with A. Let them then instead go on to complete the synthesis pqrst, a synthesis which by its discrepancy with A is freed from union with it. But an inde pendent pqrst is B, and may be recognized as B. And now, B being there along with A, the perception of its resemblance calls for no special explanation. This account of the matter appears to me simple and natural and true.
§ 31. It may be objected, in the first place, that, if the sensation is simple, this theory will not work. I admit it, and I should be sorry if in such a case it did work. I would rather that any theory, which I adopt, did not explain im possibilities. And that any actual presentation should be simple is quite impossible. Even if it had no internal charac ters, yet it must be qualified by the relations of its environment. And this complexity would be quite enough for the purpose. For the identity of the simple internal character, over against the difference of two sets of external relations, would give rise to redintegration and to the perception of the resemblance. I think a sober antagonist will hardly deny this. And if it should be denied, then I am inclined to reply with a reductlo ad absurdum. If the suggestion is quite simple, perhaps there is no difference between the similars, or perhaps they are quite different. But on either alternative they can not be CHAP. I THE THEORY OF ASSOCIATION OF IDEAS 333 similar; and again, if neither alternative is true, then the sug gestion is now admitted not to be simple, because the elements have a complex content.
I can think of another case where mistake is possible, and where suggestion might seem to dispense with redintegration. If an idea before the mind is unsteady and wavering, it tends to pass into something different. This difference may be recognized, and may appear as an idea, which is not the first idea, and yet is seen to resemble it. But the unsteadiness will in no case be reproduction by similarity. If the new idea, which is similar to the other, is produced by a change in the actual impressions, then this of course is not reproduc tion at all. But if the alteration takes place apart from the stimulus of a fresh sensation, it will still be a case of redinte gration. For that will be the principle which determines the direction of the idea's unsteadiness.
We must pass next to an objection which I feel bound to notice, though I confess I am not able to understand it. We are told that the form, say of a triangle, is not one single feature among others, which therefore could call up the other features; and that yet a triangle may call up another which is similar in nothing but form (J. S. Mill, on James Mill, I. 113). But why the form of a figure is not to be a " feature " of it we are not told, and I at least can not imagine. I was glad to find when, after forgetting this passage, I came on it again, that accidentally (§30) I had chosen to work out an instance where the form is the base of the redintegration. And I will say no more.
And there is another misunderstanding which we may remove in conclusion. After pointing out that " in the very heart of Similarity is an indispensable bond of Contiguity; showing that it is not possible for either process to be ac complished in separation from the other," Professor Bam, i: L understand him rightly, goes on to argue that, notwithstand ing this, at least a partial reproduction by pure Similarity does actually take place.
" It might, therefore, be supposed that Similarity is, after all, but a mode of Contiguity, namely, the contiguity or asso ciation of the different features or parts of a complex whole^ The inference is too hasty. Because contiguity is a part 334 THE PRINCIPLES OF LOGIC BOOK II. PT. II the fact of the restoration of similars, it is not the entire fact. There is a distinct and characteristic step preceding the play of this mutual coherence of the parts of the thing to be recovered. The striking into the former track of the agreeing part of the new and the old, is a mental movement by itself, which the other follows, but does not do away with. The effect above described, as the consciousness of agreement or identity, the flash of a felt similarity, is real and distinct. We are conscious of it by itself; there are occasions where we have it without the other, that is to say, without the full re-instatement of the former object in its entireness. We are often aware of an identity without being able to say what is the thing identified; as when a portrait gives us the impression that we have seen the original, without enabling us to say who the original is. We have been affected by the stroke of identity or similarity; but the restoration fails from the feeble ness of the contiguous adherence of the parts of the object identified. There is thus a genuine effect of the nature of pure similarity, or resemblance, and a mode of consciousness accompanying that effect; but there is not the full energy of reproduction without a concurring bond of pure contiguity. A portrait may fail to give us the consciousness of having ever seen the original. On the supposition that we have seen the original, this would be a failure of pure similarity " (Bain on James Mill, I. 122-3).
Before I criticize this passage, let me show how easily the fact which it mentions comes under our theory. When the promontory A by means of p calls up q, r, these are not referred to A. And, unless the synthesis p, q, r, s, t is com pleted, they can not re-instate B. The uneasiness of partial but incomplete recognition is caused by the presence of con nected elements, such as p, q, r, s, which, by actual incom pleteness and by vague suggestion of completeness, give us the feeling that at every moment another object is coming. But, although the whole pqrs keeps calling in other elements such as u, x, y, w, yet none of these makes up a totality we are able to subsume under any head which we know. Should, however, t be called in, then B comes at once. In this case we have the feeling of discovery, while in the former case we have the feeling of search. And all is consistent.
CHAP. I THE THEORY OF ASSOCIATION OF IDEAS 335 In Professor Bain's account we have no consistency.11 His view, as I understand it, is that though, for the full reproduction of B, contiguity is required, yet partial repro duction takes place without it. In other words, the stroke of similarity affects us enough for us to strike into a former track, but the adhesion of the contiguous bond is too feeble to drag on the mutual play of the parts. The hammer of simi larity comes down, but the flash of agreement is a flash in the pan, which fails to explode the barrel of contiguity. But in this place again, I think truth has been sacrificed to imagina tion.
If anything is brought up which suggests agreement, then this must involve what is called contiguity. For apart from such contiguity there would be nothing to recognize. This is readily shown. In the first place let the similarity " amount to identity": let the differences, which went along with and qualified B, be none of them called up. Then what is there? Why nothing but one part of the content of A, say p. And p agrees with nothing; for what can it agree with? There is nothing save itself. But in the second place, if the differences which qualified B and made it B, are called up, then obviously we have contiguity at once; for p by contiguity has re-instated pqrst. " Oh but," I may hear, " we do not go on to t, and so we never do get so far as B. We go only as far as fqrs, so that we are not able to recognize the result. It would be contiguity if we went from p to t: but if we stop at s, it is not contiguity at all.
But this would surely be no less feeble than arbitrary. ^ the whole of the differences between a portrait and the idea of the original can not be given by contiguity, why then should any of them? Why not all be given by similarity? And if any are given by contiguity, why should not all be given, fo all of them are demonstrably " contiguous "? In oth words, if similarity will not bring up all the differences why should it bring up any? Why should not all be contiguity?
Because as before we do not start from the fact, but start from a vicious theory of that fact. In the perception A/> the * is not really a particular image; and if you said q,r,s, t were associated with this mere adjective p, you would have THE PRINCIPLES OF LOGIC BOOK II. PT. II deserted your vicious theory. You try to save it by inventing a fictitious substantival image p, which then can be brought in by similarity. But the result is a system of compromise and oscillation. You will not boldly say that A brings up all of B by similarity, and your theory forbids you to say it does so by contiguity. To satisfy both the fact and your theory you say, One arbitrary part is done by one agency, and the rest by the other. And you satisfy neither your theory nor the fact. For what is actually contiguous is not like, and what is supposed like could never have been contiguous. The par ticular image, which on your theory is called up, has never been contiguous to anything whatever. And the actual ele ment, which does re-instate qrst by contiguity, is not anything we can call like A at all. It is an universal which is part of A's content. Into this confusion we are led by forcing on the facts our bad metaphysics; and the confusion at once gives place to order when we recognize that Association by Simi larity has no existence.
§ 32. We have seen that reproduction of a similar idea comes under the general head of Redintegration. And if the English votary of Association, instead of declaiming against the blindness of Germans, had been willing to learn from them, he might long ago have amended his theory.
" Si quod nunc percipitur specie vel genere idem est cum eo, quod alias una cum aliis perceptum fuerat, imaginatio etiam horum perceptionem producere debet. Quae enim specie vel genere eadem sunt, ea sibi mutuo similia sunt, quatenus ad eandem speciem, vel ad idem genus referuntur (§233, 234, OntoL), consequenter qusedam in iisdem eadem sunt (§ 195, Onto!.). Quare si nunc percipimus A specie vel genere idem cum B, quod alias cum C perceperamus; qusedam omnino percipimus, quae antea simul cum aliis in B percepimus. Quamobrem cum perceptio ceterorum, quae ipsi B inerant et in A minime deprehenduntur, vi imaginationis una produci debeant (§ 104); imaginatio quoque producit perceptionem " Idem confirmatur a posteriori. Ponamus enim nos in convivio simul vidisse hospites et vitra vino plena. Quodsi domi die sequente oculos in vitra convertis, quibus vinum infundi solet; extemplo tibi occurrit phantasma hospitum ac CHAP. I THE THEORY OF ASSOCIATION OF IDEAS 337 vitrorum vino plenorum rerumque ceterarum in convivio prsesentium. Vitra, quae domi conspicis, specie saltern eadem sunt cum vitris, quae videras in convivio." * Let us hear now what Maas has to say. I translate from the second edition of his Versuch iiber die Einbildungskraft, " The first of these rules we have mentioned is the so- called Law of Similarity: All ideas which are like are asso- ciated.f I am aware that many psychologists give this law a place co-ordinate with the law of partial perception " [redinte gration] " and consider it independent. But on this view the former stands too high, and the latter too low. Similar ideas can not be associated unless, and so far as, either they or their marks form part of one total perceptive state. But this holds good without exception. Two ideas, a and b, are like one another in so far as they have a common mark /3. Suppose now that it is a fact that b has associated itself with a." [The explanation of this fact is that] " b contains the marks ft tf, f,and a the marks /?, or, y." [On the presentation of b] " the marks «, y associate themselves with the 0," [which appears in &, and Pay is then recognized as a.]:' The association which takes place is thus between connected ideas, which are parts of one perceptive state." s. 55.$ I admit that the passage is so brief and cramped that ] have been obliged to interpolate a commentary. But there * These quotations are from § 105 of Wolff's Psych. Emp. Ed. Nova, 1738. First published in 1732.
t" Ideas" here includes perceptions.
$"Die erste von den eben erwahnten Regeln ist das sogenannte Gesetz der Aenlichkeit: alle ahnlichen Vorstellungen associiren sich. Es ist mir nicht unbekannt, dass diese Regel von vielen Psychology dem Gesetze der Partialvorstellungen koordinirt, und fur em, von diesem unabhangiges Gesetz gehalten wird. Allein das heisst dem erstern einen zu hohen, dem andern einen zu niedrigen Rang anweisen. Aehnliche Vorstellungen konnen sich nur in sofern associiren als sie, Oder ihre Merkmale, zu einer Totalvorstellung gehoren, welches aber bei ihnen ohne Ausnahme der Fall ist. Zwei Vorstellungen a und sind einander ahnlich, sofern beide das gemeinschaftliche Merkmal haben. Wenn also b, der die Merkmale fi,*,' zukommen, sich mit worin die Merkmale ft a, y angetroffen werden vergesellscha tel: so associiren sich a, 7 mit ft sind also zusammengehonge Partialvorste lungen."
338 THE PRINCIPLES OF LOGIC BOOK II. PT. II are other passages, which I need not quote, which would settle the meaning even if it were doubtful.
From these extracts it will be plain that the school oi Association have had something to learn which they never have learnt.* § 33. There is a possible objection we may here anticipate. " Admitted," it may be said, " that your theory explains the suggestion of similars, yet it does so indirectly. We explain it directly and by a simple law. And the simpler explanation is surely the better one." Anything more unscientific than such an objection I can hardly conceive. It proposes to give a simple explanation of a complex case; in other words, to decline analysis, and to reassert the fact as a principle. And it proposes in consequence (as we have shown at length) to treat the simple as a complication of the complex. But the price you pay for turning a derivative law into an ultimate principle is somewhat ruinous. You have to import into the simplest processes a mass of detail which is demonstrably not there. And this is surely a procedure which science will not justify.
And if I am told, " At all events the process of suggestion, as you describe it, is much too complex for a primitive mind," that objection once more only serves to strengthen me. For the process does not exist in a primitive mind. Similarity is a somewhat late perception, and hence can not appear at an early stage. For a rude understanding, if things are not the same, they are simply different. To see, or to feel, that two things are not the same and yet are alike, are diverse and yet in part identical, is a feat impossible for a low intelligence. It demands an advance in reflection and distinction which no sane psychology can place at the beginning of mental evolu tion. No doubt you may say that from the very first mental * Sir W. Hamilton not only refers to the true account of Associa tion by Similarity, but even criticizes it. Unfortunately he had not the least idea of its meaning. He tells us first that we are to discount " Wolff who cannot properly be adduced." I have no notion what "properly" stands for here, and perhaps Sir W. Hamilton did not really know what Wolff says. He then proposes an emendation in the passage from Maas, which reduces it to nonsense, and his criticism shows that he had no idea of the real meaning of either Wolff or his followers (vid. Reid, 913-14).
CHAP. I THE THEORY OF ASSOCIATION OF IDEAS 339 elements are alike, although the mind does not perceive it. But in saying this you open a question not welcome I should judge to the disciples of Experience. For if states of mind can be alike, and yet not like to the mind, what is such simi larity but the identity of elements within these states? The distinction on the one hand between what is or was in the mind, and, on the other hand, that which is felt by the mind or is now before it, is, if admitted, quite fatal to the orthodox English creed. We should have an attempt to purchase con sistency by suicide.
If the school of Association desired to be consistent, it might find perhaps in the " mechanism of ideas," apart from consciousness, a way of propping its tottering beliefs. But that mechanism implies metaphysical doctrines as to the unity of the soul and the permanence of ideas, which in themselves would be somewhat difficult to maintain, and which would give the lie to our most cherished prejudices.
But if consistency can be reached by no way but suicide, something after all may be said for the admission of the doctrine we have adopted — that all association is between uni- versals, and that all consists in redintegration by identity.
§ 34. The answer no doubt will be the old " Non possumus. No two states of mind can have anything in common; for, if so, they would be the same, and that is impossible." On this rock of obstinate metaphysical prejudice our explanations are broken. It would be useless to point out, as we have already pointed out, to the disciple of Experience that his own theory has been wrecked on this same iron dogma. He would say, I suppose, "Let the facts go unexplained, let miracles be invoked and fictions multiplied, let analysis be neglected and experience contemned— only do not ask me to be false to my principles, do not ask me to defile the grave of my fathers. An advanced thinker once, an advanced thinker always." And I could not answer or reproach. I respect a fidelity which I can not imitate.
But to those whose honour is not yet pledged I may per haps in conclusion be permitted to address myself. Do you wish I should like to ask in the first place, to speculate on first principles, or are you content to engage yourself on special subject matter? In the first case I would beg you 34O THE PRINCIPLES OF LOGIC BOOK II. PT. II seriously to examine the question for yourself, and not to take any assertion on trust. I can not venture to anticipate the result you will then reach (if indeed you reach any), but I feel sure that any conclusion you do come to, will not be quite the same with the orthodox doctrine as handed down in England. And to those who are not prepared for metaphysical enquiry, who feel no call towards thankless hours of fruitless labour, who do not care to risk a waste of their lives on what the world for the most part regards as lunacy, and they them selves but half believe in — to all such I would offer a humble suggestion. Is it not possible to study the facts of psychology, without encumbering oneself with beliefs or disbeliefs as to the ultimate nature of the mind and its contents? You can not have metaphysical disbeliefs without corresponding beliefs; and, if you shrink from becoming a professional metaphysi cian, these beliefs must be dogmas. Would it not be better to study the facts, and to let metaphysics altogether alone?
If this can be done in the other sciences, it surely can be done in psychology too. In the other sciences we know how it is done. The so-called principles which explain the facts are working hypotheses, which are true because they work, and so far as they work, but which need not be considered as a categorical account of the nature of things. The physicist, for example, is not obliged to believe that atoms or ether do really exist in a shape which exactly corresponds to his ideas. If these ideas give a rational unity to the knowledge which exists, and lead to fresh discoveries, the most exacting demand upon the most exact of sciences is fully satisfied. The ideas are verified, and the ideas are true, for they hold good of the facts to which they are applied. And to suppose that the metaphysician should come in, and offer to interfere with the proceedings of the physicist or to criticize his conclusions, is in my judgment to take a most wrong view of metaphysics. It is the same with psychology. There is no reason why in this science we should not use doctrines which, if you take them as actual statements of fact, are quite preposterous. For the psychologist, as such, is not interested in knowing if his principles are true when taken categorically. If they are useful ways of explaining phenomena, if they bring unity into the subject and enable us to deal with the fresh facts which