But in the child learning to distinguish letters from numerals or one letter from another, and generally in what is called 'training the senses,' differentiation and assimilation make one process of growth. The process is not one of construction, comparable to the manufacture of a watch: it is much more akin to the steady increase in clearness and distinctness of a landscape as morning breaks. At first sight the child may still confuse M with W, the cowslip with the primrose and the cat with the rabbit: only on closer scrutiny do the differences 'emerge.' When they do, the percept in question becomes more distinct and so more complex: but so far there may be no association. The fact is, great as are the advances that psychology owes to the doctrine of association, the time has come to question its finality and to circumscribe its range. The restriction here contended for is one which the earlier writers on association fully allowed: association was wholly confined to ideas that to begin with are distinct and that to the end are separable1. The process by which ideas arise from impressions cannot then be explained by association. And for long no such explanation was attempted, but the practice of regarding ideas as merely the residues of sensations prepared the way for such an attempt and the confusions to which it has led.

This remark brings us back to the first of the two questions above mentioned, that concerning the formation of ‘a continuous memory-record.' This we may now consider genetically.

Association and the Memory-Continuum.

§3. Great confusion has been occasioned, as we have found, by the lax use of the term 'association,' and this confusion has been increased by a further laxity in the use of the term association by similarity. In so far as the similarity amounts to identity, as in assimilation, we have a process which, as has been already pointed out, is more fundamental than association and presupposed in it. And when the reviving presentation is only partially similar to the presentation revived, the nature of the association does not appear to differ from that operative when one 'contiguous' presentation revives another. In association is essentially concerned, particularly when the specialisation exceeds the limits of a single sense.

1 Cf. Hume, Treatise of Human Nature, pt. i. § 4.

the one case we have, say, a+b+x recalling a+b+y, and in the other a+b+c recalling d+e+f. Now anybody who will reflect must surely see that the similarity between (a + b) + x and (a+b)+y, as distinct from the identity of their partial constituent (a + b), cannot be the means of recall; for this similarity is nothing but the state of mind-to be studied presentlywhich results when a+b+x and a+b+y, having been recalled are in consciousness together and then compared. But if (a+b), having concurred with y before and being now present in (a+b)+x, again revives y, the association, so far as that goes, is manifestly one of contiguity simply; albeit as soon as the revival is complete, the state of mind immediately incident may be what Bain loved to style 'the flash of similarity.' But, so far as the mere revival itself goes, similarity is concerned in it no further than it is concerned when a+b+c revives d+e+f. The actual a+b+c that there operates as the reviving presentation was obviously never in time contiguous with the d+e+f that is revived; if all traces of previous experiences of a+b+c were obliterated there would be no revival. In other words, the a+b+c now present must first be assimilated to the previous experience of a+b+c which alone was 'contiguous' with d+e+f, before the representation of this can occur. And this, and nothing more than this, we have seen, is all the 'similarity' that could be at work when a+b+x 'brought up' a+b+y.

On the whole, then, we may assume that the only law of association' we have to examine is the so-called law of contiguity, which, as ordinarily formulated, runs: Any primary presentations whatever, occurring (1) together or (2) in close succession, tend to grow together or to cohere, in such a way that when any one recurs it tends to revive the rest as secondary presentations -such tendency increasing with the frequency of the conjunction'. It has been often contended that any investigation into the nature of association must be fruitless2. But if So, it may at least admit of such a description as will reduce this inquiry to simpler terms. So long, however, as we are asked to conceive presentations as distinct and isolated originally and yet becoming eventually linked together, we cannot but feel the need of some

1 Cf. Bain, The Senses and the Intellect, 4th ed. 1894, p. 341.

2 So Hume, Treatise of Human Nature, pt. i. § 4 (Green and Grose's ed. p. 321); also Lotze, Metaphysik, 1st ed. 1879, § 265, p. 526, Eng. trans. 1884, p. 466.

explanation of the process. For neither the isolation nor the links are clear: not the isolation, for we can only conceive two presentations separated by other presentations intervening; not the links, unless these are also presentations, and then the difficulty recurs. But, if for contiguity we substitute continuity and regard the associated presentations as parts of a new continuum, the one immediately important inquiry is how this new whole was first of all integrated.

To ascertain this point we must examine each of the two leading branches of this supposed association by contiguity-that of simultaneous presentations and that of presentations occurring in close succession. The last, being the clearer, may be taken first. In a series of presentations that have become associated ABCD..., such as the movements made in writing, the words of a poem learned by heart, or the simple letters of the alphabet themselves, we find that each readily recalls its successor but not its predecessor. Familiar as this fact is, it is not perhaps easy to explain it satisfactorily. Since C is associated both with B and D, and apparently as intimately with the one as with the other, why does it usually revive the later only and not the earlier? B recalls C; why does not C equally recall B? We have seen that any reproduction at all of B, C or D depends primarily upon its having been the object of special attention, so as to occupy at least momentarily the focus of consciousness. Now we can in the first instance only surmise that the order in which they are reproduced is determined by the order in which they were thus attended to when first presented; since without attention there is no association at all.

The next question is whether the association of objects simultaneously presented can be resolved into an association of objects successively attended to. Now whenever we try to recall a scene noticed only for a moment we commonly find that not more than a few traits recur-those that specially impressed us, the rest being blurred and vague: what we do not find is the whole revived in equal distinctness or indistinctness. On seeing the same scene a second time our attention is apt to be caught by something unnoticed before, as this has the advantage of novelty; and so on, till we have lived ourselves into,' become familiar with, the whole', which may then, as a whole, admit of

W. P.

1 Cf. above, ch. iv, § 4.

13

simultaneous recall. Bain, who is rightly held to have given the best exposition of the laws of association, admits something very like this in saying: "So far as the mind is concerned, the generic fact is Succession. Co-existence is an artificial growth formed from a certain peculiar class of mental successions1." But, whereas it is easy to think of instances in which the associated objects were attended to successively, and whereas too we are all well aware that the surest-not to say the only-way to fix the association of a number of objects is by thus concentrating attention on each in turn, it seems hardly possible to mention a case in which attention to the associated objects could not have been successive. In fact, an aggregate of objects on which attention could be focused at once would either be already associated or would simply be a whole as yet psychically unanalysed. We seem justified, then, in substituting continuity of attention for contiguity of presentations and in talking of a secondary continuum, or 'memory-thread,' to which it gives rise. It is worth while to note that, though our acts of attention must always have a chronological order, the cases in which what we attend to is itself likewise chronologically ordered are of especial importance. Not only is the order in which we attend then objectively ordered, but the series to which we attend is more quickly and closely associated in consequence of this double correspondence. In view of our practical interest in such series-in relation to causation—the advantage of this more intimate association is obvious3.

The exclusively successional character of association has however recently been denied, and its exclusively simultaneous character maintained instead. It is at once obvious that this opposition of succession and simultaneity cannot be pressed so as to exclude duration altogether and reduce the whole process to an instantaneous event. Nor is there any ground for saying that there is a fixed and even distribution of attention to whatever is simultaneously presented: facts all point the other way. Still, though we cannot exclude the notion of process from

1 Mental and Moral Science, 1868, pp. 11 f. Cf. also James Mill's Analysis, 1878, i. pp. 80 f., Trautscholdt, “Experimentelle Untersuchungen u. d. Association." Wundt's Philosophische Studien, i. 1883, p. 244, et passim.

2 Cf. the current phrase 'thread of consciousness.'

3 Cf. Kant, Critique of the Pure Reason, “Second Analogy,” M. Müller's trans., pp. 166 ff.

consciousness, we may say that presentations attended to together become pro tanto a new whole, are synthesized or complicated. Where such synthesis is primary, it leads not to an association of images, but rather to the formation of one percept, which may become eventually a free idea. The disconcerted preperception which may later on set it free may likewise liberate a similar or contrasting idea; but it will not resolve either of them into the several ideas' of its sensory or motor constituents, with which only the psychologist is familiar'. The actual recurrence of some of these constituents may again reinstate the rest, not, however, necessarily as memories or as 'thoughts,' but only as tied ideas in a renewed perception. But we have advanced beyond such primary synthesis or complication in the yet more complex situations just now mentioned-the contemplation of a landscape or of an architectural interior, for example-which usually become familiar only after a time. In these the coexistence of the details leaves us more or less free as to the order in which we notice them. When at length familiarity has been attained, then-though the whole is past or absentthe ideal recurrence of any part may reinstate the rest in idea. This result is sometimes described as redintegration; but we must not forget that the successional associations, which have made it possible, were severally different, not many repetitions of one and the same order2.

It has become usual of late to distinguish the association of contiguous experiences and the so-called association of similar, or of opposite, experiences as respectively external and internal forms of association. The new terminology is illuminating: the substitution of forms for laws marks the abandonment of the old notion that association was by 'adhesion' of the contiguous and 'attraction' of the similar. We are thus left to find the cause of association in interested attention; and that, we may safely say, is an adequate, and apparently the only adequate, cause for the two commonly recognised forms of external association, the so-called simultaneous and the successive. But these two are certainly

1 Cf. above, § 1, p. 169 f.; § 2, p. 187.

2 Such redintegration thus pertains not to the memory-thread simply but to a new continuum of a higher dimension, so to say. This new continuum we may call the ideational tissue inasmuch as it is formed by partial redintegration or reduplication of the pre-existing memory-thread.