The Sensory Order

3.37. In accordance with the distinction we have drawn between ‘effective’ and ‘potential’ connexions between neurons (3.24) we shall also have to distinguish between that part of the following of an impulse which will always occur whenever that impulse occurs, and that part which is merely ‘potential’ and which will appear only if the tendency towards an excitation of the neurons constituting the ‘potential’ following is supported by other impulses operating towards the same effect.

3.38. In the extreme case where the following acquired by any one neuron of a given class is completely identical with that of the other members of the class, their individual position in the whole system of connexions and therefore their functional significance would also be identical. This result is possible but not likely to occur frequently. When I was first working on these problems I thought I had found an instance of such undistinguishable sensations caused by stimuli operating on different receptors in the case of pressure on teeth standing opposite each other; I am no longer in a position to verify this. Such pressure seemed to me indistinguishable so far as immediate experience was concerned, and I was able to decide which tooth was concerned only by calling in further sensory experience, such as touching the teeth individually with my fingers. This case would, of course, satisfy the condition that the two stimuli almost always occur together.

3.39. There are undoubtedly other instances where stimuli, although they set up impulses in distinct fibres, remain indistinguishable. As a rule, however, we find that even very similar sensations caused by the stimulation of different receptors differ from each other, if in no other way, at least by an awareness of the different points at which they occur, or by what used to be called their ‘local sign’. If we are to account for these differences between the effects of impulses which produce sensation otherwise of the same quality, we shall, in addition to the common following which accounts for their similar quality, have to find differences which account for their assignment to different points in space.

3.40. If we examine this problem at first in connexion with vision we enter a field where the kind of explanation which we are attempting to apply generally was first used to account for a special problem: ever since Bishop Berkeley the connexions between the impulses registering the visual stimuli on the retina and the kinesthetic impulses recording the tension of the muscles used for focusing the eye have been employed for explaining the spatial order of sensations. A particular visual impulse may have acquired exactly the same connexions with other visual impulses and thus be classified as differing qualitatively in the same manner from all other visual stimuli, and yet it may differ from the former by being connected with a different set of kinesthetic impulses.

3.41. The use we shall make of this fact, however, will in two respects differ from that made of it in the Berkeley-Helmholtz-Mach theory of spatial vision. Firstly, the impulses registering the state of muscular tension will not be conceived as producing distinct sensations but will be considered merely as physiological events which are associated with, and evoked by, the visual stimuli, and which contribute to the peculiar effects which the latter are capable of producing.

3.42. Secondly, what we shall regard as connected with the visual impulses will not be the actual movements of the eye muscles but merely the sensory impulses which normally record such movements in the central nervous system but which may also occur, if they are associatively evoked by the visual impulses, without the eye movements actually occurring. It would, therefore, not be a valid objection against this interpretation if it were pointed out that the movements of the eye postulated by the traditional theory do not, in fact, take place.

3.43. The theory of spatial vision serves here merely as an example of the manner in which the spatial order of sensations can in general be accounted for. And even this in turn is significant for us mainly as an illustration of the even more general way in which most specific acts of sensation require particular postures or attitudes of the body in order that the characteristic quality of the sensation should be produced. This fact will have to be considered further in the next chapter (4.35–4.44).

3.44. At this point the artificial separation of the connexions existing between different sensory impulses from those between them and motor impulses becomes difficult to maintain. We have already been forced to take into account the motor responses usually accompanying the acts of perception—motor responses which are probably directed from some lower level of the central nervous system, and which in turn will give rise to kinesthetic and other proprioceptive impulses which will become part of the following of the initial sensory impulse. These matters, however, can only be considered further at a later stage.

3.45. The common spatial order,6 which is part of the common order of all sensations, serves here merely as an instance of the great variety of relations between different sensory impulses which will help to build up that order. The bundles of connexions or the following which any one of several sensory neurons may acquire can differ from those of others in an almost infinite variety of ways, ranging from complete identity of their following to complete absence of any common connexions; and every difference in the connexions which the individual neurons possess will have its peculiar functional significance.

3.46. To the possibilities of functional differentiation of the different impulses by connexions transmitting excitation, we have to add the effects of the second kind of connexions mentioned earlier, namely those transmitting inhibition (3.10). In whatever manner such inhibitory connexions may be acquired in the first instance and later transferred, their existence extends the range of possible differences in the position which any one impulse may occupy in the whole system of connexions: it adds the possibility of different impulses having effects which are directly opposed to each other. In such instances the evocation of certain other impulses normally following upon the occurrence of a given impulse would be prevented if still other impulses occurred at the same time. The range of functional differentiation of the impulses which may be determined by the differences in their following is thus extended, from equality through various degrees of similarity and difference, to contrariness and complete opposition.

3.47. The significance of the differences in the following which different neurons will have acquired will show itself in the differences between the effects which the impulses occurring in them will produce in different circumstances. The manner in which any newly arriving impulse will modify the existing excitatory state of the whole nervous system, and in which it will combine its effects with those of all other simultaneously arriving impulses, will depend on the different followings of all those impulses. The further course through which any one bundle of impulse-chains will run will be determined by the following of each successive impulse and by the manner in which the impulses of this following will combine with (i.e., reinforce or inhibit) other impulses proceeding at the same time (see 5.53).

3.48. In the complex interplay of many chains of impulses proceeding at the same time, the identity of the greater part of the followings which two or more neurons possess will bring it about that the occurrence of any one of them will in most situations produce the same or similar results, and that their simultaneous occurrence will tend to reinforce those parts of their following which they have in common.

3.49. Such a system, in which each of a set of events is connected with many others in such a way that the occurrence of any one or of any group of them causes (or contributes to bring about) the occurrence of certain others, evidently performs classifications in the sense in which we have defined this term. All the impulses or groups of impulses which evoke the same other impulses will belong to the same ‘class’ because they have this particular effect in common. Individual impulses or groups of impulses will of course almost always belong to a great many different classes, that is, multiple classification in the first of the different senses we have distinguished (2.40) will be the rule.

3.50. Since the different individual impulses will become members of a class through the fact that each of them evokes the same other impulses, it seems permissible to say that the latter represent the common attribute of the members of the class—though it would be more correct to say that they constitute that attribute. The classification is effected by the evocation of certain other impulses, and the latter serve, as it were, as the ‘signs’ or ‘symbols’ representing the class; the expression ‘representative processes in the brain’ which has been much used in recent physiological psychology,7 can therefore appropriately be applied to them.

3.51. It has been suggested above (2.20–2.31, 2.44) that the mechanism which we are considering can be conceived either ‘statically’, as an apparatus capable of performing classifications, or ‘dynamically’, as a process of classification. In the preceding discussion we have sometimes spoken in terms of the former, e.g., when we spoke of connexions between neurons through which impulses are transmitted, and sometimes in terms of the latter, when we spoke of the impulses evoking each other. These two aspects of the same phenomenon correspond to the two aspects of the system of sensory qualities which we discussed then. It should now be clear that it is the dynamic aspect which is the really relevant one and that the static view is merely a method which is sometimes convenient to use for describing the potential operations of the system.

5. THE CLASSIFICATION OF THE RELATIONS BETWEEN CLASSES

3.52. There is no reason why such connexions as we have been considering should be formed only between primary sensory impulses, i.e. between impulses arriving through afferent fibres at the higher centres; they can evidently be formed in a similar manner between the further impulses which are evoked by the former and which represent classes of them. Any impulse which occurs as part of the following of one or more other impulses will, on each occasion when it thus occurs, acquire or strengthen connexions with other impulses which form part of the same following. Connexions of this kind will therefore also be formed between impulses which as primary impulses rarely if ever occur at the same time, but which on different occasions have become connected with the same third impulse, in the following of which they have in consequence become included.

3.53. This acquisition of connexions between impulses in consequence of their simultaneous occurrence in a secondary or derived character is specially important in so far as those neurons in the cerebral cortex are concerned which are not directly served by sensory receptors but which appear to act solely as intermediaries between other sensory neurons or between sensory and motor neurons. Impulses in such neurons will occur, and in turn themselves acquire connexions, only in so far as they are part of the following of other impulses; but once they have acquired such a position in the system of connexions, they will in turn be able to acquire their own following and this will include impulses belonging to the following of all the different other primary impulses of which they form a part.

3.54. In the higher centres there occur undoubtedly a great many impulses which do not uniquely correspond to particular stimulations of sensory receptors but which represent merely common qualities attributed to the primary impulses; these representatives of classes of primary impulses will in turn become the objects of further processes of classification; the classes for which they stand will be further grouped into classes of classes, and this process can be repeated on many successive levels. We need, of course, not assume that these ‘levels’ are clearly separated or that the same impulse may not form part of the following of several other impulses which belong to different ‘levels’.

3.55. The process of classification which we are considering is therefore ‘multiple’ not only in the two senses which we have discussed before (2.39–2.43), but also in a third sense: it can take place on many successive levels or stages, and any one of the various classes in which an impulse may be included may in turn become the object of further classification. This third sense in which this process of classification may be multiple must not be confused with the second (2.41); the latter refers to the case where groups of simultaneously occurring impulses (a, b, c,), (e, f, g), (i, k, l), which, when they occur as groups, are as groups treated as members of the same class of groups. In the third sense multiple classification refers to the class A (of which individual impulses a, b, c, or groups of impulses (a, b, c), (e, f, g), etc. may be members) and the class B (of which the impulses m, n, o, or the groups of impulses (m, b, o), (p, q, r), etc. are members) and the similar class C, or rather to the ‘symbols’ representing the classes A, B, and C, which, by the common following they acquire, become members of a class of a higher order.

3.56. These different forms of multiple classification which it is necessary to distinguish conceptually, will, of course, occur in various combinations, and we obtain thus possibilities of classification of (or discrimination between) the different individual impulses and groups of impulses which are practically unlimited. The consequent differences in the influence which different impulses will exercise on the whole course of the nervous processes, varying from identity through various degrees of similarity to complete difference, would be adequate for building up an extremely complex system of relations among the millions of impulses.

3.57. The word classification scarcely conveys an adequate idea of the almost infinite wealth of variety and gradation of the discriminations which can be performed by such an apparatus. Since it is not merely a question of a particular impulse either belonging or not belonging to a particular class, but also of its belonging to it more or less ‘strongly’ (according as the connexions with the classifying impulses are ‘effective’ or merely ‘potential’, and therefore in the latter case requiring more or less support in order to become ‘effective’—3.24 and 3.37), it would be more appropriate to describe these complex processes by some such term as ‘evaluation’. We shall occasionally employ this latter term in the place of ‘classification’ in order to stress that the process is capable of making distinctions of degree as well as distinctions of kind.

3.58. The combination of the different kinds of multiple classification opens up the possibility of a still further organization of the order of the impulses, because through it the differences of the positions occupied in the whole system of classification by the impulses belonging to different classes may themselves become the object of classification and thereby acquire distinct qualities of their own.

3.59. It has been pointed out before (3.17) that it is a somewhat misleading and artificial approach to trace the effects of a single afferent impulse as if it ever occurred in isolation and as if its position were to be determined in an otherwise quiescent system; and that it is doubtful whether such a single isolated impulse, even if it ever occurred, could produce a sensory quality. It is probable that only groups of impulses as such can acquire that distinct position in the whole system which we call its quality. There exists, moreover, a good deal of physiological evidence which makes it probable that it is the so-called ‘gradients’ between different impulses rather than the individual impulses which are the significant features 8. We must, therefore, consider more fully this case where not a single impulse but only certain groups of impulses as groups acquire a distinct following of their own (the second kind of multiple classification) and where in consequence the specific following determining a class of groups of impulses will be evoked only if the whole of a group belonging to this class occurs.

3.60. The constituents of a following that will appear only if certain impulses forming a group occur together must be connected with the individual impulses forming the group by what we have called ‘potential’ connexions. The simplest instance of such a position would be provided by several primary impulses which possess potential connexions with the same other secondary impulse which will be called forth only if all the primary impulses forming the group occur at the same time (or in rapid succession). Several different groups of such individual impulses may evidently thus become connected with the same symbolic impulse (or following of impulses) which will then stand for a class of such groups of impulses. And the symbolic or secondary impulse (or following of such impulse) which stands for any one of this class of groups of impulses (performing thus the second kind of multiple classification) may then (by the third, or relay type, of multiple classification) become itself a member of some new and higher class of impulses representing classes of groups of impulses. This higher class will then be represented by impulses which are symbols of classes of symbols, and so on.

3.61. As a result of such combinations of the different kinds of multiple classification it is evidently possible that the simultaneous occurrence of members of several different pairs (or groups) of different classes of impulses, will be classed as similar events or, we might say, as different events related similarly to each other. Since in such a case the same classifying impulse or impulses will be evoked by different pairs (or groups) of impulses which separately do not belong to the same class, it is legitimate to speak here of a classification of the difference (or relations) between classes of the first kind.

3.62. In order to bring out distinctly the meaning of such a classification of the difference (or relation) between different classes, it will be useful to consider the different meanings of the expressions ‘respond differently to different impulses’, ‘show the same difference in the response to different pairs of impulses’, and ‘respond to a difference between impulses’. In these expressions ‘respond’, of course, does not necessarily refer to any peripheral response of the organism but to the symbolic or classifying responses in the central nervous system. ‘To respond differently to different impulses’ then corresponds to what we have called simple classification. ‘To show the same difference in the response to impulses of different pairs which in other respects are classified as equal’ means that although, e.g., the impulses a and b in most respects belong to the same classes and similarly the impulses e and f also belong in most respects to the same other classes, there is at least one reaction which a and e and another which b and f have in common. ‘To react to a difference’, finally, means that any member of the class A occurring with any member of class B will produce the same response. If this same classifying response is also evoked by the occurrence of any member of class E together with any member of class F, and by the occurrence of any member of class K together with any member of class L, we can say that the differences or relations between the classes (or qualities) A and B, E and F, and K and L are the same.

3.63. The impulses which in this manner come to stand for, or to represent, particular classes of relations between other impulses will in turn also acquire their own following and thereby obtain their own distinct functional significance: the qualities represented by their common following would attach to the relations between the primary impulses rather than to those impulses themselves. Or, to express the same idea differently, the various kinds of relations between different impulses may themselves become differentiated from each other and thus become capable of forming the starting points of distinct chains of further impulses.

3.64. The relations between impulses or classes of impulses may thus be ordered as a system, or be classified, in the same way and by the same kind of process by which the individual impulses or groups of impulses are arranged in an order. It is in fact only at this point that, strictly speaking, we are entitled to speak of different relations between the impulses (see 1.56–1.61 and 2.20). It will now be seen how, as a result of the hierarchical organization of the connexions between different impulses, the one kind of ‘relation’ from which we started (namely the causal connexion between the impulses) can be used to build up complex structures with regard to which it is legitimate to speak of different kinds of relations existing between the various elements.

3.65. This process, by which the relations on which the classification of the primary impulses is based, become in turn the object of classificatory processes, can evidently be repeated on many levels. Not only relations between impulses, but relations between relations between impulses, and so forth, may all acquire their distinct following and in consequence become capable of forming the starting point for distinct further proesses.

3.66. The complexity of the order which can be built up by means of this variety of relations is for all practical purposes unlimited. Given the number of separate neurons in the higher nervous centres and the number of the possible connexions between them, the problem is not one of the limitation of the number of possible differences between their respective positions in the whole system, but rather the inadequacy of our mind to follow out the full degree of complexity of the order which can thus be determined. It seems indeed that any conceivable order or structure of relationships could be reproduced within such a system.

3.67. The differences in the functional significance or in the ‘quality’ which different groups of impulses may acquire as groups, and which may be independent of the functional significance which the individual impulses forming these groups possess if they occur singly, is thus a problem of the same character as that of the functional differentiation of the individual impulses, and can be answered by recourse to the same principle. But although the processes which bring about these differentiations are in principle independent of each other, and while it is even possible that only the classification of the groups and never that of individual impulses is the significant phenomenon, these classifications on different levels will, of course, interact with each other.

3.68. There will thus exist as much justification for saying that the capacity of the individual impulses to combine with others into groups possessing distinct functional significance will contribute to the distinct character which these impulses possess individually, as there is for saying that the latter will contribute to the distinctive following possessed by the group as group. Neither of these two aspects of what is a single process can in any sense be regarded as more fundamental. Both contribute in the same way to the organization of the whole system of sensory qualities; and it is the whole complex order thus produced which determines the characteristic position within this order of individual impulses as well as of groups of impulses.

3.69. The fact that chains of further processes (‘associations’) can be evoked not only by the ‘elementary’ sensory qualities (which were supposed to correspond to the occurrence of particular primary impulses), but also by certain ‘abstract’ attributes of different groups of sensations (such as figures, tunes, rhythms, or abstract concepts), has usually been regarded as an insurmountable obstacle to any physiological explanation of mental processes.9 For the approach followed here no such difficulty arises: the problem of the equivalence of ‘similar’ complexes of stimuli is not different in principle from the problem why the same associations should become attached to different impulses which correspond to the same ‘elementary’ qualities. The problem of equivalence in both these instances is basically the same and can be solved by the application of the same general principle of explanation.

3.70. Once a given impulse has acquired a definite following in common with other impulses, any new connexion which it acquires will become attached also to the impulses of its following and will be evoked, therefore, also by the other primary impulses with which it shares part of its following, although those other primary impulses may never have occurred at the same time with those others with which they become in this indirect manner associated. If all the different qualities which different impulses have in common are represented by certain symbolic impulses standing for these qualities and included in the following of all the impulses possessing that quality, there is no difficulty about the manner in which associations will become attached to such common qualities of different impulses rather than to the individual impulses. The phenomena of transfer and generalization of learning (1.50) are a direct consequence of the fact that identical mental attributes are represented by identical physiological impulses.

6. THE UNIVERSAL CHARACTER OF THE PROCESS OF CLASSIFICATION: GESTALT PHENOMENA AND ABSTRACT CONCEPTS

3.71. The fact that relations between the parts of the total sensory situation, which individually may be quite unlike each other, may yet be recognized as similar, of course, is the most general aspect of the problem of gestalt. But while the significance of the phenomenon has come to be generally appreciated mainly as the result of the work of the gestalt school, it is by now recognized by practically all schools of psychology. That in perception we do not merely add together given sensory elements, and that complex perceptions possess attributes which cannot be derived from the discernible attributes of the separate parts, is one of the conclusions most strongly emphasized by practically all recent developments in psychology.

3.72. As we have seen, it is, in fact, no more difficult to explain why different impulses caused by different combinations of stimuli, which singly would occupy altogether different positions in the whole system of relations, should as combinations occupy similar positions in that system, than why different single impulses produced by different physical stimuli should acquire the same or a similar functional significance. That the problem of gestalt perception was singled out as a special problem was largely due to the fact that it was still widely believed that the ‘elementary’ sensory qualities were somehow originally, and in a manner either not requiring or not capable of explanation, attached to the elementary nervous impulses. The fight which even before the rise of the gestalt school some psychologists had conducted against the ‘mosaic psychology’, which conceived the more complex phenomena as built up from mental elements corresponding to the physiological elements,10 was, however, bound to be unsuccessful so long as the purely relative character of all sensory qualities was not recognized.

3.73. With regard to the more complex sensory phenomena our theory leads indeed to conclusions very similar to those of the gestalt school. This, however, is so because our approach leads us to raise with regard to all sensory qualities, even those presumed to be the most ‘elementary’, the same question which the gestalt school raised with regard to configurations. Once we are led to account even for what used to be regarded as ‘simple’ or ‘elementary sensory qualities by the principles outlined here, gestalt phenomena and ‘abstractions’ do not raise any fundamentally new or different problem.

3.74. As a result of the work of the gestalt school the view has now become widely accepted that sensory qualities must not be regarded as atomic fact but should be conceived as determined by the ‘organization of the field’. It may be suggested that the theory of the determination of sensory qualities here developed gives this somewhat vague conception of the ‘organization of the field’ a precise meaning; and, at the same time, that it takes this whole approach some steps further by making it clear, firstly, that the ‘organization of the field’ is based on, and is in principle capable of explanation in terms of, causal connexions between physiological impulses; and, secondly, that this organization of the field is not additional to the qualities of any kind of atomic sensations (as most of the discussion of ‘perceptual organization’ still implies), but that it is the structure of that field which determines the peculiar functional significance of the individual impulse, or groups of impulses, which we know as their sensory qualities.

3.75. The conception of the ‘organized field’ is usually applied to the system of qualities belonging to one particular sense or modality. For our purposes it will be necessary to interpret its meaning more widely and to include in the conception not only the relations between the different qualities belonging to the same modality, but also the relations which exist between the qualities belonging to different modalities (1.56–1.67). The fact that the whole system of sensory qualities must in this sense be regarded as one organized field need not prevent us, however, from occasionally speaking of different fields as sub-systems of the more comprehensive system—sub-systems within which the elements are differentiated by a more dense and complex system of relations.

3.76. In treating the so-called elementary sensations and the more complex sensory phenomena as instances of the same process, and, therefore, as being capable of being explained by recourse to the same principle, we arrive (again in agreement with the views of the gestalt school) at the conclusion that there is no substantial difference between the acts of ‘sensation’ and of ‘perception’: both appear as essentially similar and, as we shall see later, they constitute merely different stages in an even more comprehensive range of processes, all of which can be interpreted as acts of classification (or evaluation) performed by the central nervous system. We shall therefore henceforth use the terms ‘perception’ and ‘perceiving’ in their popular meaning in which they include the experiencing of ‘elementary’ sensory qualities as well as the perception of shapes, objects, etc.11

3.77. It will be shown later (6.44–6.50) that the principle used to explain these phenomena applies also to the so-called ‘higher’ mental processes such as the formation of abstract concepts and conceptual thought. With regard to those we are, of course, more familiar with the interpretation as processes of classification in which classes of events, or classes of such classes, interact in a complex manner. It should be noted, however, that if what are called abstractions are most easily accounted for as classes of classes, etc., this does not mean that they must always be secondary, in the sense of being derived from previous conscious experience. The perception of an abstract feature of a situation may in some measure be independent of the perception of the ‘concrete’ elements of which that situation may seem to be made up (6.40).

3.78. The processes of classification and re-classification on successive levels, and the ‘higher’ mental processes corresponding to them, will have to be considered further (Chapters V and VI) in connexion with the whole process of the building up of the system of connexions as a whole. Before we can turn to this, however, we must consider another source of classification which, in consequence of the simplifying assumptions made, we have so far disregarded.