Neural Changes Involved in Learning

Practically nothing is known of the structuro-functional changes in neural tissue which accompany learning. Almost every known phase of nervous action has been linked with learning by some theorist, but the evidence which is advanced for these alleged relations is of negligible worth. We can distinguish roughly four types of speculation, based, respectively, upon evidence for (1) growth of new connections between neurones, (2) changes in the conductivity of the synaptic membrane, (3) alterations in the size and structure of the cell body, and (4) establishment of potential gradients. A theory of growth changes in learning which has received considerable attention from psychologists and physiologists alike is Kappers' principle of neurobiotaxis. The general doctrine was discussed in the first chapter, and we have had occasion to refer to it several times sinces. Stated briefly, it asserts that protoplasmic processes give rise to differences in electrical potential, and that neurones develop in relation to such gradients; the dendritic processes grow stimulopetally and contracurrently toward the active anode pole, and the axones grow with the current toward the cathode pole. Something like this seems to occur in embryo. The growing neuroblasts make their way through tissue very much as the roots of plants turn from light to moisture, select certain substances, and avoid others. Kappers attempts to extend his law beyond birth, however, and to account for all later nervous patterns in terms of continuous growth processes. He supposes that when neurones begin to conduct, differences in potential are set up by the passage of nervous impulses. These serve as further stimuli to neuronic growth and determine new interneural connections. If two neurones are stimulated simultaneously or in close succession, the potential from the weaker or ensuing excitation will discharge along the stronger or previously excited path; and the dendrites and axones of the respective neurones will be attracted to each other.

It is inadequate, however, to explain the direction of association, for it is forced to presuppose that this take place toward pathways of high conductivity, the existence of which the theory does not explain. Furthermore, the speed with which many associations are formed seems too rapid to be accounted for in terms of cell growth. Kappers tends to overemphasize the importance of interneural activity and to underemphasize more general factors. Child's principle of physiological gradients is to be preferred, therefore, although it offers no explanation of the selective connection of neurones in response. In so far as Kappers holds to the broader aspects of neural growth and action, his theory is undoubtedly sound; and many suggestive extensions are possible.

The most widely discussed theories of the neural processes involved in learning look to changes at the point of contact between nerve cells--the synapse. These contacts are supposedly fixed at birth, but the "resistance" between axones and dendrites may be altered through exercise. The dendrites are ameboid processes, capable of extending to form better anatomical connections with axones and other dendrites. Differential response is determined by the movements of neuroglia cells, which serve as insulators between neurones until they are moved away more or less permanently by learning. Neither of these views seems acceptable at the present time. The shortness of reflex latency seems to preclude any preparatory seeking of the path by the ameboid movement of dendrites, and the only evidence is the fact that the neuroglia cells assume various forms.

A much more acceptable view of changes brought about at the synapse by learning looks to alterations in the membranes of nerve cells. We are already familiar with the notion that a semipermeable dentritic membrane (or some intervening synaptic membrane) resists the passage of neural impulses. It is, of course, only a step beyond this to conceive that the changes which constitute learning or fixation are localized in this membrane. Several ways in which the structure of the membrane might be modified as a result of the passage of nervous impulses. For example, persistent electronic bombardment might cause a reorientation of the molecules of a membrane and so increase its permeability. He describes how the passage of a nerve impulse over a synapse might well cause a migration of hydrogen ions towards the end-brush, thereby rearranging the molecules of the dendritic membrane so that electrons could pass more easily from axone to dendrite. As a blow can rearrange molecules in magnetized iron only temporarily, so the membrane structures constituted by the passage of nerve impulses would tend to disappear.

This oscillatory change is due to the transfer of ions through the membrane as a result of the ordinary metabolic processes of the cells. In order for nervous impulses to pass from cell to cell, the periods of oscillation of their surface film must be synchronous. Asynchronous periods are rendered synchronous during learning by simultaneous stimulation of the membranes which are later to become tuned to each other. This theory is of interest chiefly because it supplies a hypothetical basis for chronaxie alteration, which is sometimes suggested as a basis for association.
What is known of nervous conduction certainly suggests that the alterations due to learning occur at or in the intercellular junctions. Propagated disturbances in the nervous system pass from neurone to neurone and are capable of modification only at the point of contact between cells. But we do not know whether the alteration occurs in the phase relationships (chronaxie) of impulses, the permeability of cell membranes, or their resistance to transmission.

Both increments in size and changes in internal structure of neurones have been suggested as the neural basis for learning. Exercised muscles increased in size, and that the visual cortex in blinded animals atrophied; upon such bases he argued that nerve cells, increasing in size due to use, also produced nervous impulses of increased intensity. Changes in the chemical structure of the protoplasm of the nerve cell offer a more plausible type of explanation. The chief chemical constituents of the neurone may be colloids, or non-liquid substances, dispersed in a liquid. Continued exercise might change the structure of the protoplasm by coagulating certain colloids, the effect being to produce swelling of the neurones and so to bring the dendrites and axones of adjacent neurones closer together. Differences between individuals in "impressability" and retentiveness might then be explained by the presence of various colloids in different proportion, and the failure of memory in old age might be referred to the aging of these colloids. Other suggestions are that learning curves approximate the rate of decomposition of the "autocatalytic monomolecular reaction" in the nerve cell, and that learning resembles the hysteresis of linseed oil because of the presence of linoleic acid in nerve cells.

In an attempt to get away from the strict anatomical localization of the engram fixation may take the form of potential gradients in cerebral fields. In applying his concept to the problem of discriminative reaction, he assumes that a given ratio of stimulus intensities at two points on the periphery establishes a difference in electrical potential between two corresponding cortical points. The direction of polarization will remain constant, even though there is considerable alteration in the actual position of points of excitation; that is, the excitability of cells which are influenced by this system will be modified in a constant direction as long as the properties of the potential gradient are the same.

Most of the theories reviewed above deal with the character of the engram rather than with the details of the method of fixation. With the exception of Lashley, the assumption seems to be that changes in the conductive capacity of specific neurones is the basic phenomenon of learning, and that these changes are brought about through exercise or the forced passage of impulses over paths later connected functionally as well as anatomically. Little of this is established fact; and if it is true, as Lashley maintains, that "a learned response may be carried out by the activation of conduction paths which were not employed during the process of learning," these assumptions are no longer relevant. At our present stage of knowledge the only value of learning theory is to define a possible relationship between some modification in neural structure and functional changes in behavior.