Excerpts From

A Textbook of Psychology

Donald O. Hebb, McGill University

W. B. Saunders Co., 1972

 

Chapter 4:

Mechanisms of Learning and Development

 

In Chapter 2 we saw that learning takes a number of forms.   Some learning seems simple and easily explained by direct S-R (stimulus-response) connections, but other kinds are more puzzling.   However, it turns out that even the simpler learned responses in mammals – a CR (conditioned reflex), for example, in which a dog lifts a paw to avoid shock – is unlikely to be explained by direct connections.    Even farther removed from such a simple explanation is perceptual learning, and what was called, in Chapter 2, the acquisition of knowledge.  It is possible that direct S-R connections may be the explanation of learning in certain lower forms, but it seems that the ordinary learning of mammals must be more complex.  

 

Also, learning is not the same at all stages in development, but changes with experience.  The infant is not at all capable of learning in the same way as an adult.  The remarkable learning capacities of the adult, so familiar to us that we do not see how remarkable they are, result from learning that went on in infancy and childhood.   We have to learn in order to learn:  just as a capitalist must have money before he can make money, so the student who has already learned a lot in his normal experiences before entering school is in the best position to learn more.  Early latent learning is the prerequisite for much that follows.   Learning people’s names, for example, requires having first learned to see faces as distinctive entities, learning to use words demands having learned to hear the distinctive sounds of the language.  Much learning during childhood is, obviously, motor learning, and lays the basis of motor skills later on; but much of it is the latent learning that is a foundation for intellectual development.

 

Here then we will proceed by first taking a look at neuron and synapse, since a change of neural connections (the basis phenomenon of learning) must involve the interaction of single neurons at the synapse.   We will also look at certain anatomical peculiarities of the nervous system, to see how these may help us to understand such things as latent learning, selective attention and the difficulty of concentration when one is studying. 

 

NEURON AND SYNAPSE

 

Examples of some of the different forms taken by neurons are provided in Figure 23 (see also Figs. 32 and 34).  What these have in common is that they are all one-way streets, each with a receiving and a sending end.   The dendrites are fibrils at the receiving end; the neuron may have more than one of these.  There is only one axon, the fibril that conducts away from the cell-body and toward the next cell, but it usually has a number of branches or collaterals.   Though the dendrites have the function of receiving excitation from other cells, the cell-body itself also receives excitation directly, by-passing the dendrites (Fig. 24).  Conduction by the dendrite may be slow and inefficient; it has been suggested that this is  the primitive arrangement, and that direct excitation of the cell-body is an evolutionary development which permits more efficient conduction.

 

Figure 23.  Different forms taken by neurons; a, axon. In neuron A, only part of the axon is shown; in B, C and D the whole cell is shown. B and C are short-axon cells from the CNS (note how the axon in B comes back toward the dendrites of the same cell, as if to form a closed loop). D is an afferent neuron, from a spinal-cord nerve.

 

The synapse is the point at which an axon makes contact with the dendrite or cell-body of another neuron (Fig. 24).  The enlargement of the axon fibril at the point of contact is known as a synaptic knob.   When synaptic transmission is improved in learning, one possible basis for the improvement is an enlargement of the knob, or a closer contact of knob with cell-body or dendrite.   This explanation is not a necessary one, for the change may be in the chemical functioning of the knob rather than its size.  Or, of course, both explanations may be correct. 

Figure 24.  Synapses: synaptic knobs (black) making contact with a cell-body (stippled). Only a few knobs are shown;  the cell-body and its dendrites may be completely covered by them. (From E. Gardner, Fundamentals of Neurology, Saunders).

 

 

 

Both dendrite and axon are conductors, but it appears that a large part of the dendrite has different properties from those of the axon.   The axon (like the cell-body itself) works on the all-or-none principle (except possibly at its very end).  The dendrite over much of its extent does not.  The all-or-none principle means that the axon, when it fires, expends all its accumulated energy (“firing” means that the cell is excited and begins to conduct, not that it goes off like a firecracker).   It is like a shotgun that either fires or does not fire, with no half-way measures; pulling the trigger gently does not produce a gentler explosion.   The dendrite is more like a bow-and-arrow system, in which a weak pull produces a weak effect, a strong pull a strong effect. 

 

The axon conducts “without decrement”:  since it burns all of whatever fuel is available at each point, the electrochemical disturbance does not decrease with distance as it travels along the fiber.   In this respect the axon is like a train of gunpowder; a large part of the dendrite is like a damp match in which the flame gets smaller and smaller as it moves along, so that it may go out before the end of the match is reached.   The dendrite conducts “decrementally.”  Often, therefore, a dendrite may be excited at some distance from the cell-body and not excite the latter; the greater the disturbance in the dendrite, the greater the probability that the disturbance will reach the cell-body and fire it.   The cell-body and the axon tend to act as a unit, both all-or-none in action; so that, if the cell-body is excited, the excitation sweeps over it and on over the axon, including all its branches. 

 

This event, the nerve impulse, is the fundamental process of neural transmission.   The important facts for our present purposes are summarized as follows.   (1)  The impulse is a change, both electrical and chemical, that moves across the neuron at a fast but limited speed, the rate varying with the diameter of the fiber (up to 120 meters per second in large fibers, less that 1 m./sec. in the smallest);  (2)  this disturbance can set off a similar one in a second neuron, across the synapse, or when it reaches a gland or muscle cell can cause it to secrete or contract;  (3)  the neuron needs a definite time to “recharge” itself after firing in this way;  (4)  immediately after firing nothing can fire the neuron again, but a little later, before recharging is complete, the neuron can be fired by a strong stimulation; and  (5)  when the neuron fires, its cell-body and axon fire completely – the all-or-none principle.  

 

The absolute refractory period is the first stage of recharging, about a millisecond in duration (0.5 – 2.0 msec.), when the cell is incapable of firing no matter how strong the stimulation.   The relative refractory period follows, in which a strong stimulation can fire the cell; this is about a tenth of a second (100 msec.) or longer.   The term limen or threshold refers to the strength of stimulation necessary to produce a reaction, so we can say that in the absolute refractory period the limen (or threshold) is infinitely high, and that it is higher in the relative refractory phase than when the cell is resting.   For large cells the refractory period is shorter and the resting limen is lower – that is, large cells can be re-excited sooner, and are more easily excited, than small ones. 

 

Next we can look at some of the elementary consequences of these facts.   Any nerve or bundle of neurons is made up of fibers varying in size, and hence in speed of conduction.   If, therefore, a strong stimulus fires all the neurons in a given bundle, the “volley” of impulses starts out at the same time but is dispersed, in time of arrival, at the other end.   A short sharp stimulation of the foot, for example, does not produce an equally brief excitation at the level of the cord, but a scattering of impulses extending over an appreciable part of a second.   (The dispersion in time is still greater at the level of the cerebrum.)  Next, the refractory period means that the fastest frequency of firing in a single fiber is of the order of 1000 per second, since it takes about a thousandth of a second (1 msec.) for the fiber to recover each time. 

 

The logical consequences of the all-or-none principle are quit clear, though students usually have some trouble with them.   A strong stimulation does not produce bigger impulses in a fiber.   It can, however, fire the cell more frequently, by catching it earlier in the relative refractory period.   Thus intensity of stimulation is translated into frequency in the CNS.   Furthermore, since different afferent cells have different limens, a stronger stimulation excites more cells, which again means an increase frequency of firing.   Thus the all-or-none principle applies to a single impulse in a single fiber, but not to the repetitive firing of the fiber nor to a bundle of fibers. 

 

 

Figure 25. A, polarization of resting neuron. B, passage of one impulse (shaded region) along the axon: showing that two or more impulses can occur at the same time in the neuron, since a second one can be started in the "recovered" region as soon as the first has moved along the fiber and the cell-body has recharged itself. The process is known to be far more complex than diagram A would suggest.

 

 

At this point in discussions such as this it is customary to introduce a simple diagram like Figure 25A, hallowed by long use, to explain the nerve impulse.  Let us introduce it by all means.   In the resting state the cell has more positive ions on the outside and more negative ions inside, separated by a membrane which is semipermeable (i.e., it allows some ions to pass through, but not others).  This is an unstable equilibrium; a very slight disturbance in the neighborhood of the membrane can upset the balance and allow the positive ions on the outside to pass through the membrane.  When this happens the outer surface of that part of he neuron becomes negative, an electrical effect referred to as the action potential.”  The polarization (i.e., the separation of positive and negative ions by the membrane) has disappeared; the depolarization then spreads by disturbing the of the region next to it, so that it travels along the axon.   No sooner is the equilibrium upset, however, than the cell begins to restore it by moving the positive ions outward, a process that takes altogether about 1 msec. at any one point in the cell (0.5 msec. in large fibers, 2.0 in small)  (Fig. 25b). 

 

The whole process is known to be more complex than this; the positive ions moving inward are sodium ions, but positive potassium ions are moving outward while this is going on (just to confuse the picture), and no one knows how the sodium ions are moved out in the process of recovery.   Something must do it, and this something is known as the sodium pump, an entity that has some relation to unicorn or phoenix, or the celestial spheres that move the stars in their courses.  Psychologists frequently have to give names to things they have not seen and do not understand; it is reassuring to observe that others must do so too.  

 

The all-or-none action of the axon makes possible a rapid conduction to distant points.   A further contribution to this end is made by the myelin sheath, a fatty covering surrounding many nerve fibers (this is what makes the white matter white.)  At intervals of a millimeter or so there are gaps in the sheath (at the “nodes”); the electrical potential at one gap produces an excitation in the next, starting what is really a second nerve impulse at that point.   The myelin sheath over the intervening part of the fiber appears also to prevent the impulse from occurring in the internodal region.   Thus the impulse does not travel continuously along the fiber, but jumps from node to node at a faster rate perhaps than continuous travel would permit.   (It may also demand less energy expenditure.) 

 

 

WHEN ARE NEURONS ACTIVE?

 

The neuron is a living cell and, being alive, must be active.  If it is not excited from outside it tends nonetheless to fire spontaneously:  that is, the cell stores up energy received from the blood stream until a point is reached at which the membrane polarization breaks down – so the cell fires.   Some neurons, it seems, will eventually die if not excited from outside, but they may be the exception.   It is true, however, that in normal circumstances the cells of the brain are always active, as shown by the EEG or by recording electrodes inserted in the brain.   The activity continues even in sleep, although the pattern of activity is changed.

 

 

Facilitation and Summation

 

Facilitation is the delivery of an excitatory impulse by one neuron to another – whether the excitation is strong enough to make the second neuron fire or not.   (“Facilitation” has the same meaning as “stimulation” except that it is customary to distinguish  (1) excitatory events from outside the nervous system, which are stimuli, from  (2) the excitation of neurons by other neurons, inside the nervous system:  facilitation.)

 

It is important for the student to realize that one impulse at the synapse is usually not enough to fire the postsynaptic neuron.   It is always possible that the neuron has been building up toward spontaneous firing, and then, when the neuron is almost ready to fire anyway, one impulse is enough to do the trick; but for reliable transmission across the synapse it may be necessary to have two or more neurons sum their effects. 

 

Summation is the reinforcement of the action of one stimulus, or one facilitation, by that of another.   If one touch on the skin or one slight sound or one glimmer of light is not enough to affect behavior, two together may sum their effects and be able to do so.   This is part of the reason why a strong stimulus is more likely to be effective than a weak one:  as we have seen, the strong stimulus cannot produce bigger nerve impulses, but it can excite more impulses which sum and are more likely to reach the limen of behavioral response. 

 

At the synapse, summation must normally be essential.   A single axon, delivering an impulse to another neuron, produces a slight depolarization that is not usually extensive enough to result in firing.   Two such axons side by side, however, will produce a greater area of depolarization (Fig. 26) which is more likely to be effective.   Especially with continued rapid firing, the postsynaptic neuron must be relatively refractory and the summation of impulses from a  number of presynaptic axons will be necessary if the firing is to be maintained. Since the impulses must arrive close together in time in order to sum, perhaps within a millisecond or so, timing becomes very important in neural functioning.

 

 

Figure 26.  Summation at the synapse. A and B, axons; C, cell-body. A alone may not be able to fire C; A and B together produce a greater area of breakdown and a higher probability that C will fire.

 

 

 

 

Fatigue and Inhibition

 

Fatigue in the neuron is, first, the refractory period.   The absolute refractory period lasts only for the duration of the impulse at one point, about 1 msec.   During the relative refractory period the cell can be fired again, but full recovery takes from 80 msec. (in large fibers) to a second or so (in small ones).   Secondly, a cell that fires at a rapid rate begins to have a kind of supply problem.   For example, the sodium ions that move inward when the cell fires are not excreted completely by the sodium pump for some time, and so accumulate when the cell is continuously active.   Full recovery may take an hour or more. 

 

Inhibition occurs in two ways.   In one the postsynaptic neuron is made harder to fire; in the other, the presynaptic impulse is made less effective.   The two mechanisms are diagrammed in Figure 27.  

 

 

Figure 27. Two mechanisms of inhibition; A, the "normal" postsynaptic mechanism, where the inhibitor acts directly to prevent depolarization in the postsynaptic neuron; B, presynaptic inhibition, where the inhibitor acts on the excitatory axon and decreases its capacity to deliver excitatory impulses.  In A, the postsynaptic cell is hyperpolarized; in B, the postsynaptic cell is not directly affected.

 

 

In the first mechanism an inhibitory neuron produces hyperpolarization – the opposite of the depolarization that takes place in firing – in the postsynaptic neuron, which makes it harder to fire.   In the second, it is not an inhibitory but an excitatory neuron that has the inhibitory effect, by delivering an apparently weak excitation to the presynaptic axonal ending.   This makes a partial depolarization of the ending, so that when a normal impulse arrives its effect is diminished and the postsynaptic neuron does not fire.

 

It is known that some neurons in the spinal cord (Renshaw cells) are specialized for inhibition, and it is believed that many of the short-axon cells found in the brain have the same function, with the sort of interaction between neurons that is diagrammed in Figure 28.   They are of the highest importance in learning, for learning is the elimination of wrong or irrelevant activity as well as the establishment of the right activity.   Learning must be both the formation of new associations and the elimination or suppression of previously existing ones that interfere.   As we saw above, however, not all inhibition is produced by inhibitory neurons, and H. Wachtel and E. R. Kandel describe neurons that are both excitatory and inhibitory.   The same neuron may excite one postsynaptic neuron and inhibit another; or may excite a second neuron when firing at a low rate but inhibit the same neuron when firing at a high rate. 

 

 

Figure 28.  The mechanism of inhibition. Neuron A excites X in the usual manner; a collateral also excites I, which is specialized  for inhibition. I then acts to hyperpolarize Y and prevent it from firing at the same time as X. Similarly, B excites Y, but (via J) inhibits X. (This is known as reciprocal innervation.)

 

SYNAPTIC CHANGES IN LEARNING

 

Now we come to the crucial question:  what happens at the synapse when something is learned?  And the answer is that after three-quarters of a century of research on the problem, we still do not know for certain.   This is an area in which we deal mostly in theory, not fact, and it is important for the student to remember that this is so.   We know much about the end results of learning and the conditions in which it is likely to happen, but fine details of the process?  No.   What we must do is see if we can arrive at theoretical ideas about neuron and synapse that would account for the phenomena we know.  

 

 

Figure 29. A mechanism of establishing synaptic connections. A, a group of active neurons, of which only one (a) is shown completely; B, a second group of neurons active at the same time, of which only two are shown completely. Since a and b are active together and an axon of a is close to b, a will become connected with better able to facilitate) b.

 

There is a long-standing idea that when two brain processes are active at the same time they tend to make connection with each other.   This explains the association of ideas and sensory-motor associations, and there is good reason to think that something of the kind does happen.  How might it work?   Figure 29 shows what is implied concerning individual neurons at the synapse.  Two brain activities A and B are excited at the same time.   This might be in the sensory preconditioning experiment [i], A perhaps being the perception of the light, B perception of the sound.   A neuron a, in activity A, happens to be one whose axon ends near b, a neuron that is part of activity B.   The impulses in a then arrive near b while b is being fired (by c and d), so the impulses from a help to fire b.   Theory says when this happens a connection is formed between a and b, or if there is one already it will be strengthened.   We can put the theoretical idea in general terms:

 

When an axon of a neuron x is near enough to help fire a neuron y and does so, some change takes place such that x becomes more effective at exciting y.

 

What is this change, or how does it work?  This is the question to which we have no final answer.   Transmission at the synapse is chemical, the axon tip or synaptic knob secreting a small quantity of an exciting substance (in the cortex, probably acetylcholine).   When transmission at the synapse becomes more effective, in learning, it may be because there is more transmitter substance accumulated in the presynaptic axon (x), or possibly because there is a closer or more extensive contact between the synaptic knob and the cell wall:  or both.  

 

Now the mechanism of association diagrammed in Figure 29 requires that the axon a must be close to b, so it can help to fire it, before the learning begins.   How often would this be so?  Not often, surely.   And when we talk about an association between two perceptions (e.g., sound and light in the sensory preconditioning experiment), we must suppose that a number of single-neuron contacts are required – a single neuron a could not reliably excite all the neurons making up the group B.   It is not likely that any brain activity has such contacts with any other brain activity, to make direct association possible.   But if A in Figure 29 does not have such potential connections with B, it will have many other connections already formed (the associations of common experience), including one with some process C or D which can connect with B.   When a direct A – B connection is impossible, an A – C – B or A – C – D – B connection may be formed instead (Fig. 30).   A can excite B, but indirectly.

 

 

Figure 30.  Direct and indirect associations. Left: some axons from neurons in group A end close enough to neurons in group B so that a direct A-B connection can be established when the two are active together. Right: no axons from A approach B, but among A's many connections (associations from common experience) is one with C, which is connected with D, whose axons do reach B and thus make a connection A-C-D-B possible.

 

 

The scientific literature tells us nothing about how inhibitory synaptic changes are made.  For excitation we have a theoretical idea, at least:  if x excites y and y then fires, x becomes more capable of exciting y.  No such rule has been suggested for inhibition, but we might consider one.   This is the converse of the excitatory rule:

 

When a neuron x delivers an impulse to a neuron y and y does not fire, further impulses from x make y less likely to fire. 

 

In other words, x now inhibits y.   It is Pavlov’s work that has given us the most direct and detailed information about learned inhibition, and this theoretical rule seems consistent with his data, though working it out exactly would involve us in too much detail.   The student must keep in mind that all this is theory and there are limits to the extent to which such theory should be elaborated in the absence of physiological data. 

 

Finally, to complete this account of what happens at the synapse in the course of learning, some mention of reinforcement and consolidation is necessary.   We will come back to this topic in discussing memory (Chapter 6).   Here it is enough to say that the synaptic changes of learning are temporary until they have had a chance to become consolidated.   Consolidation takes some period of time up to an hour or so, and the effect of reinforcement may be to promote consolidation.  

 

 

THE INFANT’S FIRST LEARNING

 

Some of the things that were said in the preceding section give us some clue to the nature of mental growth, and suggest why the growth depends on stimulation from the baby’s environment.  The stimulation must have two effects, according to the theoretical ideas we have been discussing.   One is perceptual learning, which enables the baby to perceive better and develops a capacity for images and ideas.   The other is to establish many associations, which provides a basis for making new associations by means of indirect connections, when direct connection are not available (Figure 30). 

 

First the perceptual learning:  common events in the baby’s experience – sight of a face or a hand or the milk bottle, the sound of mother’s footsteps, the taste of milk, the tough of a finger on the cloth – repeatedly excite groups of neurons in the cortex.   The neurons that are excited when one of these things happens are not the same every time, but there is a common core of ones that are excited every time.   The core neurons therefore tend to become connected with one another in a single system that we will call a cell-assembly.  Many of these neurons are in closed self-re-exciting circuits (Figs. 31, 32, and 33) and so, as we will see shortly, the system can continue to be active after outside stimulation has ceased.  Also, the system may be excited by another system, instead of by the sensory event that developed it originally.   While it is being excited by its own proper sensory stimulation, the activity of the cell-assembly is perception, theoretically; if it is active after the sensory stimulation has ceased, or if it is excited by another cell-assembly and not sensorily, the activity is imagery or ideation or a mediating process (for the meaning of this term, see below).   The increase of organization of the cell-assembly, with experience, is perceptual learning, which theoretically (none of the babies have told us about this) means increased clarity and distinctiveness of perception; but since it is also laying a basis for ideation, we can think of this as conceptual learning too. 

 

A word should be said about the increase in the distinctiveness of perception.   At an early stage in the learning many of the cortical neurons that are excited by seeing a face (e.g.) must be different from one time to another, so it would make little difference to the baby’s behavior if he sees the same face a second time, or a different face.   But the common core of neurons excited by one face are different from those excited by the second face, and if both faces are seen often the two core groups will be organized in two assemblies, becoming more distinct as organization goes on.   For behavior, the existence of two distinctive cortical processes – two different assemblies – means that two different responses become possible.   Inhibition presumably will play a part in this process, tending to suppress firing by neurons that are not part of the assembly and thus making the perception more clear-cut and distinctive. 

 

Secondly, establishing the network of associations:  cell-assemblies that are active at the same time become interconnected.   Common events in the child’s environment establish assemblies, and then when these events occur together the assemblies become connected (because they are active together).   When the baby hears footsteps, let us say, an assembly is excited; while this is still active he sees a face and feels hands picking him up, which excites other assemblies – so the “footsteps assembly” becomes connected with the “face assembly” and the “being-picked-up assembly.”  After this has happened, when the baby hears footsteps only, all three assemblies are excited; the baby then has something like a perception of the mother’s face and the contact of her hands before she has come in sight – but since the sensory stimulations have not taken place, this is ideation or imagery, not perception.   According to the work of the great Swiss psychologist Jean Piaget, ideation can be detected about the age of four to five months.[ii]    At this point the baby is ready for the development of a wide network of associations – he is no longer limited, in his brain activities, to ones that are excited by here-and-now sensory stimulations – which (again theoretically) must help in the formation of new associations.  

 

The variety of CRs that can be established in a baby’s first weeks of life is limited, and in part the reason may be the absence of that network to provide for indirect associations (Fig. 30).   (We know that indirect association is important at least for adult learning, in everyday life as in the laboratory.   In the laboratory, for example, a subject with the task of associating “rock” and “run,” in a long list of paired associates, tied this particular pair together in memory by thinking, “Throw rocks at a dog and he runs.”)   The characteristics of learning change greatly as a child grows – there is, for example, no immediate one-trial learning in early infancy – and this must be partly due to the fact that the brain is not yet fully developed, but part of the reason also seems to be that learning takes place increasingly against a broad background of common associations.  

 

 

CELL-ASSEMBLIES:   THE BASIS OF THOUGHT

 

Another way of putting what has just been said is that the development of thought processes is what changes infant-style learning to adult-style.   Though we do not know exactly how cell-assemblies are constituted, with respect to the details of their function, there is a good deal of empirical evidence to indicate that they exist and are the basis of thought. 

 

Figure 31. Diagrammatic representation of a re-entrant, closed or reverberatory pathway: when incoming excitation excites A, A excites B which again excites A, and so on. The continuing excitation may then be transmitted to motor organs via C.

 

Much of the CNS, but especially association cortex and certain closely connected subcortical structures, is filled with paths that lead back into themselves as well as leading on to other paths.   The simplest case is shown diagrammatically in Figure 31.  A and B form a loop circuit or re-entrant pathway; if A is excited it excites B, which can then re-excite A, and so on – the excitation chasing its tail round and round the loop, or “reverberating.”   Such a reverberatory mechanism can hold an excitation, at the same time sending out impulses to other central processes or to a motor path, via pathway C.

 

Figure 32. Diagram of closed path in the human cortex. This is more realistic than Figure 31 but still diagrammatic since it shows only about one one-thousandth of the connections that would actually be found in a block of cortex. The small diagram at the right is a simplification of the larger one. Arrows show the direction of transmission in different cells. (From Lorente de Nó, in J. F. Fulton, Physiology of the Nervous System, Oxford University Press.)

 

 

Figure 31 shows this in schematic form.   Figure 32 is a drawing by R. Lorente de Nó, the distinguished neuroanatomist and physiologist to whom we owe most of our knowledge of these matters (following the great neuroanatomist of another generation, S. Ramón y Cajal).  Figure 33 is a schematization of another such drawing, allowing the closed loops to be seen more clearly.   Figure 34 then shows the kind of photomicrograph from which such drawings were made, to give some idea of the actual complexity of the structures we are discussing.   The student, when he looks at the schematic diagrams which in this text represent closed circuits in the brain, should remember that Figure 31 stands for something much more complex. 

 

 

Figure 33. Diagram of relations between neurons actually observed by Lorente de Nó. The entering axon excites the dendrites of four neurons, A, B, C and D. Of these B and C send impulses out of the system to excite other systems, but impulses from A and D are delivered only within the system itself. A-B, B-E, and B-E-E'I (After F. A. Beach, et. al. (Eds.), The Neuropsychology of Lashley, McGraw-Hill.)

 

 

 

Figure 34.  Photomicrograph of a section of cat cortex, giving a better idea of the complexity of connections -- but still only about one neuron in 60 is shown here, stained by the Golgi-0Cox method which for some unknown reason is selective. If all were stained, no detail could be observed, only a solid mass of stained tissue. The outer layer of the cortex is at left. (From D. A. Sholl, Organization of the Cerebral Cortex, Methuen.)

 

 

For convenience, these paths will be discussed as if they were in the association cortex.   Actually, some of them are in the cortex, but many of the closed loops are “corticothalamic,” running from cortex to thalamus and back to cortex; and some of them must similarly involve other subcortical structures, such as the hippocampus, which in man lies inside the tip of the temporal lobe and appears to have an important part to play in memory.  

 

We assume that a cell-assembly consists of a number of these re-entrant paths that have become connected with each other by the processes of perceptual learning discussed above.  Thus the assembly is a system that is organized in the first place by a particular sensory event but is capable of continuing its activity after that stimulation has ceased.   Not only is it made up of self-re-exciting closed loops, neuron A exciting B and B exciting A, but two loops or sets of loops may also have that self-re-exciting relation to each other.   The whole system therefore is such as to maintain an internal activity for short periods of time. [iii]

 

Further, the system may be excited by other similar systems in the total absence of the sensory event that originally organized it.   It thus meets the requirements of an ideational process.   The very essence of an “idea” is that a brain activity is occurring in the absence of the environmental event it corresponds to.   You need not have an elephant present to think about elephants, the discomfort of being caught in the rain can be thought about long after your clothing has dried.

 

Note finally that latent learning (that is, learning without overt response at the time the learning occurs) now becomes intelligible.   It is a change of connection between cell-assemblies at a time when the assemblies are not exciting other assemblies that have direct motor connections.  The need of summation must mean that one assembly may be able to fire another only with support from other assemblies, so these systems may be active, in certain combinations, without producing a behavioral effect.   Other combinations result in behavior.   In the first case, when there is no motor outflow, interaction between assemblies makes changes at the synapse, which affects the relations between them.   For example, it may be proposed now that the two “central activities” in Figure 7 (p. 32) are cell-assemblies.   Also, the development of an assembly, as a form of perceptual learning, is itself latent learning.   Latent learning has been regarded as mysterious in the past, but it need not be. 

 

Let us see next why the sort of direct S-R connection that Pavlov and Thorndike talked about is unlike to exist.   The nervous system does contain straight-through pathways, especially from sense organ to cortex, and from motor cortex to spinal cord, but this seems not to apply to the cortex – and it is in the cortex that learning takes place.  

 

DIVERGENT VS. PARALLEL CONDUCTION:   ATTENTION

 

When dendrites and cell-bodies lie close together, they tend to be excited together; and if their axons also lie side by side, ending at about the same place, impulses reaching the ends of the axons at the same time can sum their effects on the dendrites or cell-bodies of neurons at the next level, past the synapse.   This arrangement of fibers is found all the way from the skin, for example, to the cortex, and the results is that skin stimulation produces an excitation in the corresponding region of the somesthetic cortex with high reliability.   But when we look at the way in which the excitation is carried from the sensory to the association cortex, we find divergent conduction instead.   The neurons lead in different directions, and transmission must be less reliable.   Is this a defect?  No, we will find that there is good reason for it.   It allows central processes to do things they could not do otherwise.   The “unreliability” of divergent conduction allows central processes to receive, and to be affected by, relevant sensory messages only.   If transmission in the association cortex was always effective, these processes would be bombarded by too many messages at the same time.   In effect, divergent conduction screens sensory input and allows the higher animal to respond selectively to events around him:   the fundamental mechanism of attention. 

 

 

Figure 35 represents the change from parallel to divergent conduction as a sensory excitation is transmitted first to the sensory area of cortex and then on to the association cortex.   In parallel conduction, branching at the axonal ends of the neurons (inset, Figure 35) allows each neuron at the next level to receive excitation from several axons, so summation can occur.  But where divergent conduction begins, at the right of the figure, summation is not provided for except when by chance two neurons starting in different places happen to converge.   In Figure 35, the stimulation S which fires a group of cells A will have a high probability of firing group B, and also the single cells C, D and E, but the excitation is likely to peter out at the level of F, G and H unless other neurons (not shown in the figure) provide supporting excitation.   There is a low probability, for example, that E will fire H, unless something else in this region is sending impulses to H at the same time.   So the region of divergent conduction will act like a screen, allowing through only the “messages” that can obtain some support from other activities that are already going on.

 

Figure 35. Neural conduction: parallel on the left, divergent on the right. S, stimulation of a part of a receptor surface, excites  a group of neurons A which converge at the next synaptic level and provide summation in a postsynaptic group B, which therefore fires reliably. At the next synapse there is divergence; B produces summation in C, D and E, and so fires them reliably, but  there is no summation for F, G and H (some of these diverging paths may meet by chance, when a very large number is involved). Inset: the convergence in greater detail, showing the overlap of branching fibers that produces summation.

 

 

These are the reasons why it seems most unlikely that there are direct S-R connections formed in the cortex during learning.   This conclusion was long ago reached by K. S. Lashley, who showed by means of brain operations that there is no clear-cut loss of single habits when parts of the cortex – even fairly large amounts, as much as 15 to 20 per cent of the total cortex – are removed.  Direct connections should be interrupted by such removals.   Also, there is plenty of other evidence to support the idea that the mammal’s cortex is constantly screening out sensory messages and not allowing them to be responded to behaviorally.   The cortex is always exposed to messages from the sense organs:  from the pressures on all parts of the body that are touched by clothing and the change in these pressures with every movement, from joints and muscle reporting to the brain the position of the limbs, the constant information being received from all parts of the visual field as long as the eyes are open, the constant low-level background noise received by the ears.   Not 1 per cent of this information is responded to, or enters awareness (that is, it does not affect the ongoing thought process). 

 

We may think instead of transmission via cell-assemblies, which may be relatively direct and reliable, if the assemblies are strongly connected, or very indirect and unreliable.   Figure 36 shows, schematically and theoretically, how selective transmission from sensory cortex would occur.   In the figure, X, Y and Z are excitations coming from sensory cortex into association cortex, in a region where a cell-assembly A is already active.   Facilitation from both A and X sum to make B fire, and now B can sum with Y to fire C.   B and C in turn fire D, which sends excitation back to help keep B active.   The effect is to produce a temporal series of central activities, A-B-C-D firing in that order.  

 

 

Figure 36.  Selective transmission from the sensory cortex. X, Y and Z, sensory transmissions; A, B, C and D, closed systems (cell-assemblies). A being active before the sensory input occurs, B, C and D becoming active thereafter; E and F, inactive systems. In these circumstances excitation X would have its effect, and then Y, but Z would not have an effect.

 

 

 

 

 

 

 

 

 

 

On the other hand, the assembly F is not active, so there is nothing to sum with the sensory input Z, so this will have no central effect.   If F was active and A not active, X and Y would have no effect, but Z would excite E; a different cortical activity – and therefore a different behavioral response – would excited by the same total sensory excitation (XYZ).   The response that is made to a given stimulation depends on what activity is already going on in the brain. 

 

This is the selective action made possible by divergent conduction in association cortex, and transmission via cell-assemblies instead of straight-through pathways with neurons organized in parallel.

 

 

ATTENTION AND CONCENTRATION

 

The selectivity of brain response is, from a psychological point of view, attention or set, to which we will return in the following chapter.   Before we conclude the present account of the development of perception and thought, there is one further point which is suggested by what we know of the anatomy and physiology of the brain. 

 

The large size of the human brain is needed to allow human beings to learn so many different things, but it also means that there are many more neurons in the brain than are needed for learning any one thing.   As we have already seen, neurons fire spontaneously if they are not being kept active.  The many neurons that are not necessary for a learning task may, when the task is prolonged, become active anyway, producing other thoughts besides the ones that concern the present task.  This is the student’s problem of attention and “concentration”:  to respond to, and think about, only the subject-matter before him.   How is it possible?

 

For one thing, the cell-assemblies that are active tend presumably to inhibit random activity by other neurons, and this inhibitory action may become more and more effective as mental growth goes on and assemblies become better and better organized.   Young children have a notoriously short “span of attention,” but this span increases as they grow older.   Secondly, the span of attention is longer for the more intelligent subjects at any particular age, provided the topics they attend to have many facets for one to think about.   We may suppose that the more intelligent person has developed more  cell-assemblies, which would provide more inhibition to control the random extra activity in the brain.   What the full story is here we do not know, but it seems that the “interesting” task – the one that is easy to concentrate on – involves both complexity and some level of arousal (Chapter 10).   The complexity makes it possible to find different ways of thinking about the material, making more assemblies active.   Another way to add somewhat to the interest of a dull learning task, as we will see in the discussion of study method in Chapter 6 (p. 108), is to make the learning active, though it is not clear theoretically why this should contribute to the inhibition of that random extra activity of the brain cells that are for the moment unused. 

 

 

SUMMARY

 

This chapter provides a short account of what is known about the functioning of the single neuron, and a more theoretical account of the changes that occur at the synapse in learning.   Inhibition is an important part of the process, eliminating irrelevant or conflicting response.   It is proposed that in mammals, especially man, an essential part of learning is the development of cell-assemblies – closed systems capable of briefly maintaining their own activity – and a background of interconnections between assemblies corresponding to familiar events.  It is also proposed that this development is what happens as the baby’s style of learning changes to a more adult style.   The structure of the cortex appears to rule out direct S-R connections in mammalian learning.   Cortical transmission thus appears relatively inefficient, from one point of view, but from another it has the advantage that much irrelevant excitation from the sense organs is screened out and not allowed to disturb behavior.  Finally, the chapter points to a disadvantage arising from the large size of the human brain, related to the student’s problem of concentration.

 

 

GUIDE TO STUDY

 

The student should be clear about the meaning of all the italicized terms in this chapter, but especially the following:  axon, dendrite, synapse, all-or-none principle, nerve impulse, relative and absolute refractory periods, facilitation (vs. stimulation) and summation.   He should be able to state a theoretical rule for the strengthening of connection at the synapse, and a converse rule for inhibition.  He should be able to explain why it is possible that a wide background of associations may make further associations easier to form.   How is a cell-assembly formed, and how can it maintain its own excitation?   Why is parallel conduction more “efficient” than divergent conduction?   If divergent conduction screens out some messages, what messages will it let through, and why?   And why may a large brain be a handicap to the learner – in some ways? 

 

 

 

Chapter 5:

THE CONTROL OF BEHAVIOR:   COGNITIVE AND NONCOGNITIVE

 

At several points in the preceding chapters the discussion has referred to higher behavior, or has implied a difference between behavior that involves thought and other behavior.   Ultimately all behavior is a reaction to environmental stimulation, but the relation between stimulus and response varies from direct to extremely indirect.   At one extreme is the unconditioned reflex, where the stimulus has its effect at once; the neural connections between receptor and effector are straight-through.   At the other extreme, the stimulus has the effect of exciting complex cortical circuits and the behavioral effect may be long delayed (when latent learning occurs, for example).

 

Knowing something about the way the nervous system works will now help us to find order in these behavioral phenomena.   Some psychologists have talked as if all behavior was cognitive:  meaning that it involves thought.   Others have talked as if all behavior fitted the S-R formula:  meaning that thought does not enter into it, that all behavior is fully controlled by the stimulation from the environment.    In the light of what has been said about the nervous system, the student will see that both kinds of behavior do occur and that the real question is the directness of the connections between stimulus and response.  There is no opposition between two kinds of behavior but a gradation from one to the other, though at the extremes they seem quite different.  

 

What is the difference between the behavior of a spider building a web and that of a man ploughing a field to plant potatoes?  Both will obtain food as a result of what they are now doing, but one is consciously planning, one is not; what is the meaning of this difference?  A girl at a dance is cold; she shivers reflexively, but then goes to get a scarf to put over her shoulders.  On what principles do we distinguish between her two kinds of response to the environment?  In general, the answer to these questions is that some behavior shows a close temporal relation between stimulus and response, and depends on direct, or relatively direct, connections in the CNS; other behavior does not show this close relation, and we must assume that the connections are more indirect.   The first kind is reflexive or sense-dominated; the second kind is cognitive behavior, dependent on mediating processes (ideas, thinking), which in this text are assumed to consist of the activity of cell-assemblies. The presence of cell-assemblies permits a delay between stimulus and response but also, when two or more stimuli are involved, may introduce other kinds of complications:  for example,  the phenomenon of set.

 

But having mediating processes, and so being less directly controlled by sensory events, does not mean that the higher animal has less need of sensory information or is less influenced by it.   All behavior is affected by sensation all the time.   Behavior fundamentally an adaptation to the environment under sensory guidance.  It takes the organism away from harmful events and toward favorable ones, or introduces changes in the immediate environment that make survival more likely.  Not all behavior is adaptive in such a narrow sense; sex and maternal behavior are not necessary to the behaver’s survival, nor is play.   But with these forms of behavior also (for example, in finding the mate, avoiding obstacles in moving to and from the nest, or maintaining bodily orientation in play), sensory guidance is always an essential factor.   No organized behavior is possible without it. 

 

A simple one-celled organism such as ameba does not have the specialization of parts, sensory and motor, found in higher animals.   In obtaining nourishment, for example, the same tissue must act to detect the presence of food, move toward it, ingest and absorb it, and excrete wastes:  a single cell must be nose, mouth, legs and alimentary canal.   As a result, the ameba has very limited ability to capture food and avoid destruction.   Only events in its immediate vicinity, at the present moment, affect its behavior.   In a higher animal, specialization of parts permits an extraordinary sensitivity of some cells (the receptors) to environmental events, so that food or danger is detected at a distance, and an equally extraordinary speed and precision of movement in others (the muscles).   But specialization means that the receptors and effectors are spatially separated, and there must be some means of communication from one to the other.   This is the first function of the nervous system:  a spatial integration or coordination of parts.   The specialization of effectors also means that they must be active in a definite sequence or at just the proper time in order to have their effect; this temporal integration is also achieved by the nervous system.

 

For example:  when a mosquito alights on the forehead and begins operations, the skin of the forehead has no adequate means of self-defense.   Nerve fibers in the skin transmit the excitation, originating in the skin, to the central nervous system, whence it is relayed to effectors at a distance, the muscles of a hand and arm.   The mosquito is swatted.   For successful defense, cells at a distance must be called upon, and they must be called upon in the proper order; the muscular contractions involved in a swift, accurate movement must have very precise timing or the hand will reach the wrong place.   Another example:  the nose of a hungry animal smells food but, though it needs nutrition as much as the rest of the body, it cannot obtain the food directly; what it must do is initiate a complex series of activities in other parts of the body, in definite order.   The end effect is that food gets into the stomach and the blood stream delivers to the olfactory cells of the nose (and of course to other parts) the proteins, salts, sugar and so forth that they need in order to keep on serving their function.  

 

The role of sensation is clear, not only in initiating the activity but in continuing to guide it throughout.   In swatting a mosquito, the muscular contractions to be made depend on the initial position of the hand, so they are determined by the sensory processes which, coming from muscle and joint, “tell” us where our limbs are at any moment.  Similarly, the predatory animal seeking food must change the course of his movements as the prey changes position.   Sensory control is involved in any form of adaptation to the environment, simple or complex, and we must recognize it as a first principle of behavior. 

 

The directness of control, however, and its complexity, vary.   Think of the nervous system as a communications network, and of the brain and spinal cord as a switchboard where messages coming in get passed on to the proper destination.   The primary function of the switchboard is to connect the sender (sensory) directly with the receiver (motor):  a simple routing function.   This is all the nervous system does in animals at the lower end of the evolutionary scale.  But in higher animals the switchboard has developed, one may say, a mind of its own and is no longer fully controlled by its input.  

 

What can such a statement mean?  Figuratively, the phrase “a mind of one’s own” means that one does not merely follow instructions, but may go counter to, or act without, instruction.   In the present discussion the phrase is applicable both figuratively and literally, and seeing why this is so should help the student to think psychologically.  The figurative meaning applies, for the complex network in higher animals does not merely transmit signals from receptor to effector; sometimes the signal is held and not transmitted at all, sometimes it is transmitted only after being changed, and sometimes signals originate in the switchboard itself, arising from the continuing activity that goes on within.   The brain does not merely “do what it is told” by sense organs. 

 

The literal meaning also applies, when mind is defined as the higher activities of brain; for the complex communications network of the higher animal has developed so that messages run to and fro within it, as well as into and out of it.  Such internal activity, infinitely more complex than these words can suggest, is mind; and possession of this internal complexity is what distinguishes higher from lower animal, making the behavior of the higher animal less directly under the control of sensory input. 

 

For example:  a barefoot boy steps on a sharp stone and pulls back his foot.  We may think of this as simple in-out (reflex) transmission:  in from the skin of the foot to spinal cord, out to leg-muscles.   But something else may happen also.   The boy stops, stands for a moment, then goes back for his sandals.   He has “thought” about it, and “decided” that the beach is too rough for barefoot walking.  Something has gone on in the closed loops of the switchboard, a complicating factor in the relation of sensory events to the concomitant behavior. 

 

Thus we have two main classes of behavior, roughly speaking, though one shades into the other.   One of them is reflexive, one involves a thought process to some degree.   In both, sensory guidance is essential; but in the first the guidance is a full control – sensory events by themselves take charge and elicit the complete pattern of response as long as no other event interferes.   The second class involves a much more complex process at the level of the association cortex in the cerebrum. 

 

 

CLASSIFYING BEHAVIOR BY MEANS OF THE S-R FORMULA

 

We can distinguish the two classes of behavior by means of the stimulus-response or S-R formula.   The formula describes the fundamental pattern of behavior:  each movement of the animal is a response to an immediately preceding stimulation, and is predictable from that stimulation.   Behavior that fits this formula is reflexive behavior, explained by the operation of through-routes from sense to motor organ.   This is the first class of behavior.   Behavior that does not fit the S-R formula belongs to the second class. 

 

As far as we know at present all the behavior of lower organisms, such as ant, bee, housefly, jellyfish, cockroach and spider, is comprised by the S-R formula and does not justify any reference to mental processes:  there is no mind, consciousness, emotion, purpose or perception.   Also, in higher animals – even man – there is a great deal of reflexive behavior which is likewise comprised by the formula.   To understand the problem of mind it is essential that we first separate lower from higher behavior and not make the mistake of seeing mental processes everywhere.   It is possible that mental processes occur in the spider but we have no evidence that this is so (and much evidence that it is not).   It is quite possible, even probable, that conditioned reflexes in man involve mental processes:  but there is no clear evidence that this is so, and until such evidence is obtained we will be on firmer ground by working with the assumption that the ordinary CR is the operation of a relatively simple through route, even though part of the route goes through the cortex.

 

So we assume that some (lower) behavior is fully controlled by direct S-R connections, and that other (higher) behavior is not.  Since neural conduction is rapid, the direct connection means that a response should occur promptly, unless some other process interferes with it, and obviously the response cannot occur when the stimulus is not present.   When analyzing some new piece of behavior we apply the S-R formula – that is, we ask whether it meets the above conditions – and if it agrees with the formula we conclude that it can be explained by the more direct S-R pathways, with no need to assume any self-maintained activity within the switchboard.   If the behavior does not meet these conductions, even if it still seems simple, the more complex switchboard activity (mediating processes, ideation) is involved and we are dealing with cognitive behavior.   Given the same sensory input, cognitive behavior varies from one time to another; what the response will be is not determined by the stimulus alone.  This point will be clearer below, where set is discussed. 

 

These are broad classes of behavior, and one merges into the other with no clear dividing line.   And to this it should be added that the fundamental principles of neural action are the same at both ends of the continuum, just as the fundamental principles of chemical union are the same in inorganic and organic chemistry.  Organic chemistry deals with far more complex molecules than inorganic chemistry, and hence with superficially different phenomena, but the atoms are the same in both cases and obey the same laws.   In psychology, we know more about the relatively simpler behavior corresponding to the S-R formula, and what we must find out is how to apply this knowledge to “higher” (i.e., more complex) mechanisms.

 

First, then, reflexive behavior, more closely controlled by the present sensory input. 

 

 

SENSE-DOMINATED BEHAVIOR:  UCR AND CR

 

In psychology as in other fields of scientific thought one prefers the simpler explanation rather than the more complex one, as long as the facts permit it.  We must prefer the S-R connection to ideation, as involving fewer steps of inference and thus being more nearly factual.   But in doing this, we must ask what properties of behavior are implied by S-R connections if the explanation is to be satisfactory.   For example, it was suggested in the preceding section that the conditioned reflex fits the S-R formula.   When we apply our criteria, however, we find that while some CR’s fit, others are doubtful cases. 

 

The essential idea is that behavior is produced by sensory stimulation.   This means that the response coincides in time with the stimulus, within a second or so; and that the same response follows the stimulus each time, provided there is no interference from background stimulation and that learning has not changed the S-R connection in the meantime (if, for example, pain followed the response on the preceding occasion, the response might not be repeated:  but this would mean that learning had occurred).  Neural conduction is fast (ranging from about a meter per second in small fibers to 120 m./sec. in large ones).   With a direct S-R connection, consequently, there can be little delay of response following stimulation, barring interference.   We saw in Chapter 4 that connections through the cortex, where mammalian learning occurs, can hardly be direct; but they may be relatively direct.   Cell-assemblies, though they make a delay of transmission possible, might also be connected so as to permit prompt and reliable transmission.   The essential question concerns the behavioral evidence:  is the response predictable from knowing (a) the physiological state of the animal (whether he is hungry or not, for example) and (b) what the environmental stimulation is?

 

The UCR meets the requirements of the S-R formula completely.   It begins and ends with stimulation and it is highly predictable.   If stimulation for two incompatible reflexes is given simultaneously, or course, only one of the two can occur.  A pinprick in the foot of a newborn infant produces a flexion (withdrawal) response of the leg, mild pressure on the sole of the foot produces an extensor thrust; if both stimuli occur at the same time, only one response can be made:  it is usually the flexion.  Also, especially in older subjects, the higher centers of the brain are capable of interfering with reflex processes.   The reflex response to a pain stimulus in the fingers, for example, is to pull back the hand; but if one is holding a valuable teacup which becomes too hot, the pain reflex is usually inhibited long enough for one to set the cup down before letting go of it.   Similarly, one can often inhibit a cough. 

 

Otherwise, however, the UCR is highly constant and predictable.   There is a long list of separate reflexes:  the pupillary response to increased light in the eye, producing contraction of the pupil; salivary reflexes produced by stimulation of the mouth; sucking reflexes in the baby, produced by stimulation of the lips; sneezing, coughing, eye-watering, produced by irritations of nose, throat or eyeball; reflexes of heart and arteries, regulating the flow of blood to different parts of the body; reflexes of the stomach and gut, controlling digestion and the movement of food through the alimentary canal; a large number of postural reflexes, producing maintenance of orientation of the body in space; and so on.   All these UCR’s are highly consistent in their action; there is no doubt that the responses are controlled sensorily, and depend on straight-through S-R connections. 

 

Now let us see how these considerations apply to the learned reflex-like responses, the CR’s or acquired S-R connections. 

 

In our first example, the CS is a buzzer; the UCS is an electric shock to the foot, delivered two seconds after the buzzer begins.   After a few trials, the animal raises his foot off the grid immediately when the buzzer sounds, and continues to do so on almost every trial.   We may then think of a fairly direct pathway from certain cells in the ear to the muscles of the leg (Fig. 37). 

 

 

Figure 37. Schematic diagram of an S-R connection via the mammalian cortex: not a straight-through connection, as Thorndike would perhaps have suggested, because the cortex does not seem to work that way, but still relatively direct.

 

If the connection does not work every time, we need not reject this conclusion; any path through the nervous system must thread its way through a tangle of other paths, and is exposed to possible interference from other processes, especially inhibition.   What we can ask is whether a given stimulus combination tends to arouse always the same response (any deviations from the response being referable to interference or fatigue); or whether, on the other hand, the same total pattern of stimulation produces systematically different responses on different occasions. 

 

Figure 38. If the primary pathway of Figure 37 is not capable of evoking a response outside the apparatus, this may mean only that in the apparatus it is supported by other pathways, from different receptors. The mechanism is somewhat more complex, but can still be comprised by the S-R formula.

 

Thinking not of single stimuli but of the total pattern of stimulation, as in the preceding paragraph, helps us deal with another possible difficulty.  Having established a CR to the buzzer, we take the animal out of the apparatus and sound the buzzer again.   The CR  does not appear.    This does not necessarily mean that the response is not controlled by S-R pathways, but shows that a pathway from ear to leg muscles is not enough, by itself, to account for the response.   But we can assume that the path is supported by others (Fig. 38) from eye, nose and skin – that is, by sight, smell and touch of the apparatus.   The S-R formula may have seemed overly simple to the student at first glance, but when we begin to deal with actual behavior it involves us in complexity enough.   But the principle still remains clear.  In the apparatus, at least, the experimenter can elicit the CR whenever he wishes by manipulating the animal’s environment:  thus the behavior is controlled by sensory stimulation.   

Another problem is this.   In Pavlov’s procedure the CS is presented for 15 seconds before the UCS.    Early in conditioning the animal secretes saliva as soon as the CS is presented; but then the CR is delayed and eventually is made only in the last two or three seconds before food appears.  The same thing happens in conditioned-avoidance experiments.   A buzzer is presented for 10 seconds, followed by shock if the rat does not move off the grid that delivers the shock.   Early in conditioning the rat jumps as soon as he hears the buzzer; but eventually a time comes when he does not move till 7, 8 or 9 seconds later.   Are these delays compatible with the S-R formula, which says that response should be prompt?  They are, perhaps, if the response is being inhibited – as Pavlov’s results indicate – and if the inhibition itself is under sensory control.   (Sense dominance may be inhibitory as well as excitatory.)    A more likely hypothesis, however, is that the inhibition is controlled by a mediating process – a combination of cell-assemblies – with limited duration.   With these data, no final decision is possible. 

 

A clearer case, however:  we change the experimental conditions further, and get a different kind of conditioning.   Instead of presenting the CS for the whole delay period of 10 seconds, we present it for 1 second, and given the UCS 9 seconds later.   Once again we obtain a 7- or 8-second delay in response.   What produces the CR, when it does occur – 8 seconds after cessation of the CS?  In some manner the brain holds the activity aroused by the CS, instead of transmitting it at once to the effectors.   Such behavior is not comprised by the S-R formula, but involves the question of the mediating process, to which we now turn. 

 

 

THE MEDIATING PROCESS

 

The typical problem of higher behavior arises when there is a delay between stimulus and response.  What bridges the S-R gap?  In everyday language, “thinking” does it; the stimulus gives rise to thoughts or ideas that continue during the delay period, and then cause the response.  And in fact, we are now talking about the thought process.  But the words “thought” and “idea” have been around for a long time and have acquired a number of meanings, so it is hard to use them precisely though they are still useful in a general sense.  

 

Mediating process has a more exact and more limited meaning.   It may be defined as an activity of the brain which can hold the excitation delivered by a sensory event after this event has ceased, and thus permit a stimulus to have its effect at some later time.  To “mediate” means to form a connecting link, and the simplest function of the mediating process is to connect S with R.   Theoretically, however, a mediating process can also be excited by other mediating processes as well as its own sensory event, and when a number of mediating processes interact in this way – being excited by each other as well as by sensory events – the result is thinking; so, theoretically, a mediating process might also be defined as the unit or elementary component of thought, replacing the term “idea.” 

 

We do not know, certainly, what a mediating process is, as a physiological mechanism.   By definition, however, it can hold an excitation for some short period of time, and the only way in which this can happen according to present knowledge is by means of the closed loops or reverberatory circuits discussed in Chapter 4 (Figs. 31, 32, 33).  One of these circuits consisting of two or three neurons could not hold an excitation long enough to correspond to an idea, because with steady firing fatigue would build up in 10 or 20 msec., but a number of them combined in a cell-assembly might do so.   Still longer holding could occur with several cell-assemblies, assembly A exciting assembly B, which excites C, which excites A again.   In this text therefore it will be assumed that a mediating process consists of one or more cell-assemblies.  The student must remember that the cell-assembly idea is a hypothesis of how a mediating process works.   We know from behavior that mediating processes exist; that they consist of cell-assemblies is theory, which may turn out to be wrong, or only partly right (that is, there may be some other mechanism of holding in addition to cell-assemblies).  Consequently, we can speak of mediating processes without committing ourselves to any definite idea of how they may work; when we speak of cell-assemblies we are talking about a specific theoretical construct.  

 

Figure 39. Diagrammatic representation of a reverberatory pathway; incoming stimulation excites A, which excites B, which re-excites A and so on. It is suggested that this is the mechanism of holding or trace activity, in principle. C represents other paths which may be excited by collateral fibers (branches) from B, and which might be excited each time the excitation travels round the closed pathway.

 

Now let us look at the behavioral evidence.  Figure 39 might represent the situation in which a schoolboy has been told to add some numbers, but before he has been told what numbers, so he cannot make the response yet.   Five seconds later the teacher says, “Four, seven.”  The pupil at once says, “Eleven.”  It is a simple task, but how did he do it?  He could not make the response to the first stimulation (“Please add”) alone, nor to the second (“Four, seven”) alone; the response can only be made to both, so the effect of the first must have been held for five seconds until it could be combined with the second, and Figure 39 shows how a reverberatory path could hold it (though the path would have to be complex – probably two or more cell-assemblies – to hold for as long as five seconds).   This kind of behavior, thoroughly familiar to everyone, is the simplest and clearest evidence of the existence of mediating processes.   The behavior takes us beyond what is comprised by the S-R formula.  The capacity for holding an excitation in the central nervous system is the primary mark of the higher animal.

 

The student is reminded that the simple closed paths of Figure 39 and the figures that follow in the present chapter are conventional representations, deliberately schematized so they will not be taken for reality (cf. Figs. 32 and 33).   The closed-loop diagram of one or two arrows is a symbol to represent a self-re-exciting system, just as the chemist has his conventional symbols for the improbable atomic structures he talks about, and the physicist has his to represent battery, condenser or ground in electrical circuits.  In other words the loops of Figures 39 and 40 are not pictures, but a kind of pictorial shorthand. 

 

SET AND DELAYED RESPONSE

 

Let us take as an experimental subject an intelligent student for whom simple arithmetical operations are automatic.   We seat him before a screen, tell him that pairs of numbers will be flashed on it, and instruct him to give their sums as quickly as possible.   We present then a series which is made up of combinations such as 8/2.  To each we obtain a correct and rapid response.  A given stimulus pattern always produces the same response, and the reaction time is short, of the order of a second.   This is a highly practiced form of behavior; no thought appears to be involved, and we might conclude that the behavior meets the criteria of the S-R formula:  promptness and reliability.   

 

But the response depends on the subject’s being set to add.   The 8, 2 combination produces the response “ten” every time – until we say to the subject, “Now subtract” (or divide, multiply); whereupon the same stimulus pattern produces, with equal speed, and reliability, the response “six” (or “four” or “sixteen”).   It is therefore clear that the response is not determined by the present stimulus pattern (8, 2) alone, so the behavior does not fit the S-R formula.   The response is determined by two stimulations, one of which has to be held and has its effect only after an interval.  The highly schematic diagram of Figure 40 illustrates how this might occur, developing further the idea presented in Figure 39.  (It shows also, in the broken lines, what would have happened if the earlier stimulation had been different and had produced a set to subtract.)

 

 

Figure 40. Diagram of a possible mechanism of a set to add. The excitation from the prior stimulus, "add," is held in a reverberatory loop. The second stimulus (8,2) is connected with two motor paths, and can evoke "ten" or "six"; but the reverberatory activity supports only one of these, and the response is "ten." If the prior stimulation had been "subtract," a different reverberatory circuit would have been active and would have determined the response "six." Needless to say, this diagram is entirely schematic (any resemblance to neural tissue is entirely coincidental.)

 

 

The paradigm (the clear, representative example) of set is as follows:  Stimulus A is applied to the organism, and then a different stimulus B; B elicits promptly a response C, but only if A was presented first.   A sets the switchboard, or prepares it, so B can have its effect. 

 

The delayed-response procedure provides us with another example of set, in the behavior of the monkey.   (The preceding discussion has referred several times to a delay in responding, but the term “delayed response” refers technically to W. S. Hunter’s method, about to be described.)  The monkey is allowed to see food put in one of two containers out of reach.   A screen is then put between him and the containers, so that he cannot later find the food simply by keeping his eyes fixed on the correct container.   After a delay of five seconds, ten seconds, or more, the screen is removed, the two containers are brought within reach, and the monkey is permitted to choose between them. 

 

The monkey is quite capable of success with this task, in a way that provides some evidence of the presence of mediating processes.   In one experiment, particularly, the evidence was decisive (O. L. Tinklepaugh).  In it the experimenter sometimes used lettuce, which the monkeys liked, as the food reward, and sometimes banana, which they liked better.   When the monkey saw lettuce put into one of the containers, chose the right one, and found lettuce in it, he took it and ate it.   But when he saw banana put in, and then found lettuce – the experimenter having deceitfully made a change during the delay period – the monkey did not take the lettuce, but showed surprise and searched in and around the container (apparently looking for the missing piece of banana).   On occasion the animal simply had a temper tantrum instead.

 

Here is our holding process again.   Seeing banana put into the food container had some lasting effect, as shown by the conflict that appeared when lettuce was found instead.   We know that when the monkey found lettuce without having an expectancy of banana he reached for it and ate it.   This is consistent with a direct sensory control of response.  But the tempter tantrum, and failure to take the lettuce, is not; this behavior must be jointly determined by a mediating process resulting from the earlier stimulation and the effects of the present stimulation.  

 

The problem of holding does not always arise when there appears to be a delay of response.    A lower animal may succeed in delayed-response tests by making a postural adjustment immediately and maintaining it.   For example, when the animal sees feed put into the right-hand container he may move over to that side of the cage and wait there until the screen is raised; then he simply chooses the near container.   Monkeys and chimpanzees do not solve the problem in this way; they usually move around during the delay period.    It would be possible, if we did not have any other data, to suppose that the monkey might tense the muscles of the hand nearest the food and keep them tensed while moving around; when the screen is raised and he turns back to face the containers, he could then choose the correct one by using the hand whose muscles were contracted.

 

However, there is usually no sign at all of the monkey’s “remembering” the location of the food in this way; and the lettuce-versus-banana experiment has special importance in ruling out such an explanation as far as the higher animal is concerned.  

 

 

SELECTIVITY IN BEHAVIOR:  ATTENTION RELATED TO SET

 

It was said above that the distinguishing mark of the higher animal is the capacity to hold an excitation for some time before it has its effect on behavior.   The mediating process that does the holding is apt to introduce selectivity into the behavior, in either or both of two ways, in the form of attention and set.  Accordingly, these also are marks of higher behavior.   Attention is selectivity in what is responded to, or sensory selectivity; set is a selectivity of response, motor rather than sensory.  Very often, however, attention and set go together. 

 

Notice that the selectivity is constant at any one time, but is easily changed from one time to another.   This has already been illustrated by the set to add.   Presented with a visual stimulation of a pair of numbers, the subject consistently produces one response to each pair as long as that set lasts; when the set is changed to subtraction, a different response to each pair is made – again consistently.   It is characteristic of such sets that they change readily, so with the same stimulation the response varies systematically from one time to another.   With the visual stimulus of 6, 3 we do not get a random variation of “nine” and “three,” but a response that is highly predictable, provided we know what the subject’s set is.  Thus the mark of higher behavior is not mere selectivity of response:  the lower animal is also selective, but he is always selective in the same way, because he is built to behave in that way only.   The higher animal is capable of responding in many different ways, and does so at different times, but at any one time he tends to respond in one of those ways only:   a changeable selectivity of response.  

 

Attention is closely related to set.   In Figure 40 is suggested a way in which a mediating process would support one response and not another.   Figure 41 shows how the mediating process would support the effects of one sensory input and not another.   A and B are two stimulus events whose effects are transmitted to the higher levels of the CNS.  C₁ is a mediating process which supports the input from A – that is, excites the same central paths that A does – so that the excitation  from A is transmitted farther.   A is, as we say, “noticed” by the subject, and is likely to affect behavior.   C₂ is a mediating process that would similarly support B, but it is not active, and so it is much less likely that B will have any effect.   It is not noticed.   This would be true especially if C₁ tends to inhibit C₂  and vice versa:  A or B will be attended to, but not both.  Attention may then be defined as an activity of mediating processes (C₁ or  C₂) which supports the central effects of a sensory event, usually with the implication that other sensory events are shut out.  

Figure 41. Schematic diagram of a mechanism of attention in which a central process, C₁, supports one sensory input (from A), C₂ supports another (B). Event 'a will be responded to if C₁ is active, event B if C₂ is active.

 

If the student will now compare Figures 40 and 41 he will see that we are talking about a process very similar to that of set.   The two terms really have almost identical meanings, but “set” is usually applied when the process is thought of as a selection of one response rather than another, and also when one can point to a specific preceding experience which excites the mediating process that does the selecting, or “sets” the animal.   The similarity of the two is such that we sometimes speak of a “perceptual set,” a set to perceive one way rather than another, and this obviously is a form of attention as defined above. 

 

 

TEMPORAL INTEGRATION IN BEHAVIOR

 

Next we can see how the selectivity (of set and attention) appears when there is an extended series of responses, rather than the single responses we have talked about so far.  This is the question of how the links are connected in a chain of responses, or how one step in behavior leads to the next – in general, the question of how it is that man or animal in responding to his environment appears to be doing one thing in a coordinated series of movements, rather than responding randomly, now to this stimulation, now to that, in a disorganized way. 

 

For the lower animal, as we have seen, the whole process is sensorily controlled, and the organization or integration of individual movements into a unified, directed whole is determined by the animal’s environment, for any given physiological state of the animal.   (A hungry animal, or a sexually active one, will respond differently from the way he would at other times, but this is because the low level of nutrient chemical substances in the blood stream, or the presence of sex hormones, changes the way in which certain synapses in the CNS function.   The S-R paths are changed, but they are still S-R connections and the animal is still under sensory control.)   Each leg movement of the animal, for example, produces a further stimulation, which leads to the next movement.   This may be by feedback from the muscles of the leg (sensation of movement); by changing the animal’s position in space, which changes visual or tactual input; by sensation from the foot as it meets the ground; and so on.   Each of these new stimulations can give rise to a new response, and the continuity of the behavior derives from the situation in which the animal finds himself. 

 

Feedback stimulation is equally important for the higher animal, though here mediating processes take part in producing the directedness and unity of behavior, sharing the control with sensory input moment by moment.   Figure 42 shows how one may think of the shared control.   C₁, C₂ and C₃ represent the combined central processes of the brain at three successive moments in time.   X and Y represent the central processes that might have occurred, but did not.   S is sensory input.   S₁, occurring at the same time as C₁, tends to excite the cell-assemblies of the C₂ group and also the Y₂ group (that is, it delivers facilitation to both).   C₁ tends to excite C₂ and X₂.  The one that is excited, therefore, is C₂, which receives the summed facilitations of two sources, C and S; X₂ and Y₂ are not excited, each receiving facilitation from one source only, without summation. 

 

Figure 42. To illustrate the selective process in thinking. C, central processes (simultaneously active assemblies) at three successive moments in time; S, corresponding sensory inputs; X and Y, subliminally excited assemblies. X receives excitation from one source (central) only, Y from one source (sensory) only, so these have lower probabilities of being active. C consists of the assemblies which receive excitations from both sources and which consequently are active. Thus, C₁ selectively determines which of the assemblies will be active, from among those that S₁ tends to excite; and contrariwise. This selective central influence is attention, represented by the horizontal arrows of the diagram.

 

There will be motor outflow – facilitation delivered to some part of the motor system – from these central activities, though it is not shown in Figure 41.   Thus behavior is determined by C₁, C₂ and C₃, and since each of these groups of activities is controlled by (a) sensory input and (b) facilitation from the preceding central activity, the behavior is under the joint control of sensory and central processes.   Both the environment in which the animal or human subject is, and the continuity of central processes – that is, thought processes – make for continuity and direction in what the subject does.

 

This describes a single unified train of thought, which we may perhaps think of as the normal state of affairs.   However, it is possible that the facilitation from C₁ on X₂ may sometimes be strong enough so that X₂ is activated.   It is also possible that when this happens there may be enough summation among the cell-assemblies making up X that it can excite another group, and this one still another – and so on.  That is, there might be a separate X series of activities that starts and continues at the same time as the C series, in parallel with it, provided that the cell-assemblies in question did not cause interference with one another.   This means that there would be two trains of thought at the same time, one of them in control of behavior.   Something of the sort seems actually to occur in man.  Sometimes, when one is reading aloud, from a not-too-interesting book and for an extended period of time, one finds that though one has kept on reading intelligibly (since the audience has not complained) one’s thoughts have wandered, so there must have been two separate series at the same time.   Again, lecturers commonly have the experience while speaking one sentence of thinking about the next one, or wondering whether an illustration is clear, or sometimes even thinking that the lecture is rather dull.   Speech is not interrupted while all this is going on, and it seems definite that two independent thought processes must be running in parallel.  

 

Speech is the most interesting example of temporal integration in behavior, and we will return to it in Chapter 13.

 

 

SENSORY DOMINANCE AND VOLUNTARY BEHAVIOR

 

“Volition” represents an old and troublesome philosophic problem, chiefly because the nature of the underlying psychological  issue was not clearly formulated.   The problem does not arise with the behavior of lower organisms such as the ant, in which there is no clear departure from direct sensory control; nor does it arise when the behavior of higher organisms remains under such control.   It does arise at other times, and it is, in short, the problem of understanding how mediating processes are involved in a response.  

 

In an earlier day the nervous system was thought of as simply a sensory-motor system.   As we have seen, this would imply a direct sensory control; so whenever the control was absent, the behavior became a mysterious business.  “Volition” or “will” was a power of some separate agency, which somehow could be exerted on the brain or on the motor system to make it behave in a way in which it would not otherwise behave.  “Will power” thus was something that one might have a lot of, or little.  “Free will” also might seem to mean that voluntary behavior was not subject to scientific law, not determined by cause and effect. 

 

But all this, in a much earlier day, was related to a very crude idea of how the bodily machinery operates, and especially the machinery of the brain.   If the higher animal responds in two different ways to the same total pattern of stimulation, it is because the activity of the central switchboard is not the same on the two occasions, but “set” differently; as a result the sensory input is routed to different muscles.   It is evident that we are as yet far from understanding these problems in detail, and must not be dogmatic about their eventual explanation in terms of brain processes; but at the same time there is no fundamental philosophic problem about voluntary behavior as such. 

 

Consequently, in modern psychology the terms “volition” and “will” or “will power” have disappeared.  “Voluntary behavior” still has a certain usefulness, as a rough classification; it is, in short, behavior that cannot be predicted from a knowledge of the present environmental stimulation alone because a systematic variability is introduced by mediating processes. 

 

 

SUMMARY

 

All behavior is under sensory guidance, through the switchboard of the central nervous system.   Reflexive or sense-dominated behavior is controlled by direct connections; higher behavior involves mediating processes (roughly, ideas or images).   The mediating process is an activity of the switchboard itself, not a straight-through transmission.   It can hold a sensory input for an appreciable time before transmitting it; it may also be excited by other central activities (i.e., other mediating processes), instead of by sensory input. 

 

Set is like closing one switch and opening others before current is applied to the line.   It prepares the switchboard for a particular kind of output.  Attention is closely related:  it prepares the switchboard for receiving a particular class of input.  Both involve mediating processes, and a holding of the prior stimulation that prepares the switchboard. 

 

The chief problem is to understand how mediating processes can hold an excitation.   An available theory (which may or may not be right) proposes that this is done mainly by complex closed circuits (cell-assemblies: Chapter 4) in which excitation can travel round and round without dying out immediately.   “Volition” appears to refer to the selective effect of mediating processes on behavior.  

 

Guide to Study

 

For review, the student might see whether he can produce schematic diagrams representing what happens in holding, set, and attention.   He might find, for example, that the delayed-response procedure, involving holding followed by a later stimulation (when the screen is raised so the animal can make his choice between the two containers), calls for the same diagram as that of set – and consequently he would be able to say why the delayed response is a special case of set, or set another form of delayed response.   He should understand how the holding process frees the subject from immediate sensory dominance, and how “free will” essentially means the absence of such dominance.  He should see what problem is raised by the CR which occurs only after the CS has lasted for 5 seconds, as distinct from a CR that appears at once, and how a delayed CR (5 sec. after the CS has stopped) raises the problem of holding or mediating process.   He should be clear about the meaning of feedback and be able to give examples of his own, and he should be able to explain how feedback contributes to organized sequences of behavior.  

---