Various Kinds of Learning

Before proceeding to discuss the process of trial and error learning, however, we should consider certain other mechanisms of neural impression, real or alleged. We have seen in the preceding chapter that the cerebral cortex is presumably able to receive and to retain impressions of exact patterns of afferent nerve excitation without linking these impressions or excitations with any particular motor innervations or reactions. The simplest kind of learning would seem to consist in the mere recording of such patterned impressions upon the cortex. Such learning does not involve the formation of habits and, in itself, cannot be detected by purely behavioristic observations. A large part of what passes for education consists in the laying down of purely afferent records of this sort, which never express themselves in the behavior of the individual. Their failure to gain such expression results from the fact that they have no natural relationship to the principles of motor control. In order for such "expression" to appear, an arbitrary association must be established between the afferent patterns and particular efferent ones. The means by which such connections are made constitute our most interesting problem. The laying down of afferent records in the cerebral cortex may be attributed to the principle of exercise or the opening up of a certain pathway as a consequence of the mere incidence of nervous energy.

The conception of learning which predominates both in common sense discussion and in psychological research is that of the process by which particular kinds of motor reaction become connected with specific stimuli. There is one case in which this can apparently be accomplished by the mere formation of an afferent excitation record. This is the case of the generation of a conditioned reflex. We assume in the beginning that there is an hereditary connection between a certain stimulus and a particular reaction. Then a purely afferent process can lay down a patterned record in the cerebral cortex which simultaneously involves or includes as parts, first, the innately operating stimulus in question and, second, certain other stimulus factors. The pattern which is thus formed makes it possible for the new afferent or sensory component to set off the efferent reaction because it has become associated with an afferent element which already has this specific motor connection. In this process, it is unnecessary to form any new liaison between the afferent and efferent sides. Learning in accordance with this principle plays a very important part in the development of response in the individual.

Another aspect of learning which is emphasized in popular thought seems to have a purely efferent character: the acquisition of skill. Such acquisition involves the perfection of incito-motor mechanisms and their accessories. However, it is to be doubted whether this process is ever exclusively efferent in nature. Certain kinds of skill very obviously involve the evolution of a close coördination between sensory and motor factors as, for example, in marksmanship. In other cases, as in playing a musical instrument, --without reference to a score--the mechanism may seem to be exclusively motor, but closer examination shows that it actually involves proprioceptive and tactual afferent nerve currents, in an intimate manner. When these afferent impulses are eliminated, the reactions become impossible. The skill of a white rat in running a maze is independent of visual, auditory or olfactory stimuli, but is believed by Watson to rest upon proprioceptive or kinaesthetic impressions; and in such case, is not an exclusively motor acquisition.

In the case of human beings, there are certain supposed methods of establishing specific response, or particular connections between stimuli and motor reactions which do not seem to follow the plan of trial and error learning, as observed in animal experimentation. One of the commonest methods of attempting to establish habits in children is that of verbal instruction, command, or "telling them what to do." The principle of this form of teaching appears to be as follows. First, certain words must be associated with the stimulus and with the reaction, respectively, so that the individual "knows the meanings of the words." Second, a sentence in the form of an admonition or command, is presented to the "pupil," which links together the name of the stimulus with that of the reaction. There must, also, as a rule, be a name for the individual who is being instructed and this must be included in the instructions. Ordinarily this name is "you."

Sometimes this verbal method of establishing a reaction appears to be very effective. Its effectiveness may be a measure of what we call the "suggestibility" of the individual, that is his tendency to be governed in his behavior by verbal formulae. However, in the case of children-and, particularly, very young children--the method is singularly ineffective. It obviously cannot be applied to infants or those who are incapable of appreciating the significance of words, because they have never successfully associated them with objects, situations, or the like. But, even when the instructions are "understood," it is ordinarily necessary to add further verbal formulae, which usually take the form of promises or threats, representing some sort of reward or punishment. Even with these accompaniments, the instructional method may fail entirely.

It is evident that the basis of this most commonly accepted method of human control is rather complicated and depends upon the establishment of a considerable amount of "experience," in the form of associations, before it can become at all effective. Consequently, we shall be justified in postponing a detailed analysis of this scheme until we have arrived at a satisfactory explanation of a means for teaching which will apply to infants or to animals.

Another somewhat more primitive device, which has been used somewhat successfully with animals, but with considerable less success in the case of children, is that of forcibly putting the organism through the movements, which it is desired to have executed, while in the presence of the appropriate stimuli. Thus, we may endeavor to teach an animal how to get out of a problem box, by dragging it or leading it to the right part of the box and then manipulating its limbs so that it operates the latch or other releasing device. As a rule, this scheme is not very effective, because the associations which it establishes in the animal or child are quite different from those which are desired by the instructor. The difficulty is that we may not be able to secure active motor innervations through the cortical centers, corresponding to the forced postures or movements. Even the proprioceptive currents which arrive at the cortex will not be the same as those which would be generated if the movements were active, instead of being passive. Nevertheless, there may be some degree of similarity between the proprioceptive impulses in the two cases, which will serve to aid in the establishment of the desired interconnections between afferent and efferent activities.

This leads us to consider the interesting relationship which necessarily exists between proprioceptive and the corresponding efferent nerve currents. On account of the nature and location of the proprioceptors, they must be set off in patterns of excitation which correspond with the form of the given motor innervation. These proprioceptive configurations arrive, in most cases, at the cortex while the motor innervations are still operative. Consequently, an association will inevitably be established between each specific form of motor incitation and the corresponding proprioceptive, afferent pattern. It is, undoubtedly, for this reason that the proprioceptive configurations become the keys to the selection of particular kinds of movements, from the afferent side. Just as soon as the organism begins to experiment with its motor apparatus, it starts to lay down proprioceptive or kinaesthetic records in the cortex, which may be presumed to have afferent-efferent connections with the corresponding motor incitations. Thus, it has ready for use by the more advanced forms of learning, a repertory of afferent cortical neurograms which are in dynamic correspondence with an equal number of motor control systems.

Two other methods of learning or teaching which are commonly employed in human education are those of imitation and reasoning, respectively. Imitation involves the duplication of movements which are seen or otherwise represented in a non-motor fashion (possibly only by verbal description), and therefore requires a relatively direct translation of afferent patterns to efferent ones. This may possibly take place through the intermediation of proprioceptive or kinaesthetic factors. Learning in this manner has been noted among the lower animals as well as in men. It is particularly frequent in the case of apes, and evidently plays a part in the fixation of characteristic songs in birds. Like the other popularly considered methods of instruction, the neurological basis is complex and consequently we must postpone its detailed consideration. Learning by reasoning is the most complex of all of the methods by which we perfect our forms of response. This method seems to be a peculiarly human device, which is probably due to the fact that it depends essentially upon language or symbolism. We shall consider some details of the process at the appropriate point in our argument.

Logic and Language

It will not fail to have struck the reader of the preceding paragraphs that logical theories suggest at each step remarks of a grammatical nature. And this is natural, for, to put it briefly, language is but the vulgar and imperfect though the most usual expression of the thought of which Logic seeks to determine the laws. Nevertheless, the relations between Logic and language have been generally neglected by philosophers. If we are to be guided by their scholastic programmes, they are occupied at most with one sole question, i.e. the origin of language. This preoccupation corresponds to an absolutely false and superannuated conception of Philosophy, according to which the object of the latter is "the beginning and the end of things." Such questions (in so far as they are at all soluble) evidently belong to the scientific and historical methods and have nothing really philosophical about them (unless by a confusion of ideas springing from the ambiguity of the word principium, "principle" is identified with beginning). It is equally childish to conceive the relations between Logic and language as do certain nominalists who maintain that Logic is based entirely on the forms of language and who do not even shrink from the extreme and absurd conclusion that there are as many logics as languages.

Philologists are generally too preoccupied with the material and physiological part of language (phonetics), and even when they study its intellectual side (in Semantics or the Science of meaning), they are inclined to dwell on the more or less bizarre and illogical particularities (which certainly abound and jump to the eye) rather than to disengage the general features which manifest, in spite of all appearances to the contrary, that there is a latent logic in the formation and evolution of our languages. Philology is too exclusively historical and descriptive, too much in subjection to particular facts; it regards all attempts at appreciation as heresy, and is even averse to all theory. Philologists lack the logical spirit which, essentially critical and normative, does not fear to criticise language by confronting it with its aim, i.e. the exact and complete expression of thought.

Words are signs for our ideas; they are signs like other signs, but more convenient than others, because they are at once oral and graphic, visible and audible; but still they have to satisfy the conditions which govern all signs. The first of these conditions evidently is that there should be a univocal correspondence between the sign and the idea signified; for every idea a single sign and for every sign a single idea.

This principle is so evident that it seems little more than a hackneyed truism. But its bearing becomes apparent directly we apply it to the critical analysis of our languages. Every notion ought to be expressed in language once and once only (mere economy would counsel this, even if Logic did not). Now the notion of "plural" is repeated five times in the following phrase: "Les bons enfants sont obéissants"; four times by the plural of the article, the adjectives, and the noun, and once again in the plural form of the verb. Similarly the notion of "feminine" is expressed four times in the following phrase: "Une bonne mère est diligente"; once in the idea of mother itself (which ought to be sufficient), and three times more in the article and the adjectives. Again the notion of "person" is always expressed twice in our languages, once by the pronoun (or noun) which is the subject, and a second time by the form of the verb. And here we light on the origin of these pleonasms: it resides in the evolution of our languages which proceeds (speaking roughly) from the synthetic to the analytic. Ancient languages, such as Latin, did not employ the subjectpronoun with the verb: the person was indicated by the verbal form itself (which had already absorbed a pronoun, witness to the primitive Greek endings: mi, si, ti. . .). As these verbal forms weakened and gradually became confused, it was felt necessary to indicate the person more precisely, and a separate pronoun was added, while, at the same time, the personal forms of the verb were preserved. Similarly, case-endings tended at one period (to a certain degree) to replace prepositions themselves, and came from older agglutinated prepositions. But their meaning gradually became confused and faded, and this is why in the classical epoch the Latin of ordinary speech employed prepositions, even with the cases which did not require them. The idea was expressed twice. Nowadays caseendings have nearly disappeared from the Romance languages, the daughters of Latin, and are replaced (advantageously) by prepositions. This is the final result of a logical evolution.

All this perfectly explains the pleonasms which encumber our languages, but does not justify them from the logical point of view. Moreover, we see that the popular and unconscious logic which presides over the evolution of our languages tends to eliminate progressively double uses and superfluities. Conscious logic, therefore, would only be anticipating natural evolution if it suppressed them from now onwards.

By an inverse phenomenon, but in virtue of the same interior logic, our languages tend to create special words to express certain ideas which lack proper expression. For example, interrogation has, in our languages, no proper expression (such as have negation, doubt, etc.), except the inversion of the subject, which is an inconvenient and insecure proceeding. This is why many languages have forged special words or locutions to give special expression to this idea; for example, the English do (they no longer say, "dream I?" but, "do I dream?"), the Danish mon, the French est-ce que. And in vulgar French a very convenient interrogative particle has made its appearance: ti, e.g. je sais-ti? j'ai-ti couru? (taken, by analogy, from the third person, est-il venu?).

Thus the immanent logic of our languages ceaselessly tends to apply the principle of univocity, or at least of approximating to it. But it is constantly impeded by custom and tradition, i.e. by the secular products of evolution which every language bears within it. Our modern languages, even those most highly evolved, carry profound traces of prehistoric (and prelogical) mentality, and they will only disengage themselves from these very slowly and very incompletely. It is only in an artificial language that we can wipe out the past; only there could we apply in all its rigour the principle of univocity, and hope to realise the desiderata of Logic. Few people have an idea to what a degree of simplicity such a language could be reduced, while at the same time it would provide as adequately, and even more than do our traditional languages, all the elements necessary for the exact and precise expression of thought.