W.S. McCulloch


When one eyeball is infected, unless it is promptly excised, the sight of the other eye fails. This has been known for 3000 years. About 2500 years ago Alcmeon of Croton first removed an eye from a living man. There were no anesthetics to prevent doctor and patient from talking to each other. Perhaps Alcmeon cut one structure after another until only the optic stalk remained, and found that the bad eye could still see, but when he cut the optic stalk it went blind. He certainly believed that vision was transmitted along the optic nerve, which he dissected to the chiasm. Thus began neurophysiology and psychology as experimental sciences. Alcmeon formulated his findings by saying that the eye, ear, nose and tongue contained opposites light and dark, up and down, shrill and base, sweet and sour in proportions matching those in the world, and shipped those opposites themselves by way of the nerves to the brain, which made of them the harmony he called health. We call it normal function. Aristotle calls Alcmeon unsystematic in the matter of opposites, and says he threw them out at random, but it was in terms of the opposites, associated by similarity and dissimilarity, continuity and contiguity, that Aristotle founded the first and now the oldest school of psychology, Associationalism.

Today we know much of how some organs make opposites and of how they ship them to the brain, much less of how the brain makes the harmony of health, and nothing of why one eye goes blind unless we remove its infected fellow. The best the surgeon can do today is to imitate Alcmeon of Croton.

Man is predominantly eye-minded. Most of his information of the world about him comes by way of his two eyes, each of which is connected to his brain by as many axons (about a million) as all the rest of his sense organs combined. The retina contains two substances necessary for vision, bleached by visible light, rejuvenated in the dark at such a rate as to account for our slow adaptation to the general level of illumination. But how these substances or their changes initiate or prevent impulses in the rods and cones themselves is unknown. It is generally agreed that the rods, which inhabit chiefly the periphery of the retina and have a lower threshold to light, have little to do with color vision. That is the business of the cones, which are most concentrated at the fovea centralis. Even if both photochemically active substances have something to do with color vision, their sensitivities as a function of wavelength have too broad and flat a maximum for us to attribute color vision to their differential sensitivities to sundry wavelengths, and only two such substances have been found where we should need or like three in current theory. Stroud has pointed out that if we had only one such substance to work with, we could still build a lens into our receptors to secure differential sensitivity to different wavelengths, and that the cones with their refractive globules may do just that. Unfortunately, we have insufficient knowledge of the refractive indices of cones, globules and ambient fluid. We do not know whether they, like the crystalline lens and aqueous humor of the eye, have the optical properties necessary for the functions we would assign to them.

How hue is transmitted to brain has been most carefully studied by Granit, who found that particular axons of the optic stalk responded more to some wavelengths than to others. Whether because of unfortunate choice of animal or because of the very unhappy conditions under which such explorations can be made, or both, his findings suggest to me little more than that there are such differential activities not that they convey hues as we sense them.

These axons end in the lateral geniculate, whose point-wise anatomical connections with the retina are now beautifully mapped. According to Le Gros Clark, roughly, the six-layered portion of the geniculate receives the projection of that part of the retina whose excitation gives us the full spectrum: the four-layered portion, the blue-yellow distinction only; and the two-layered portion, just black and white. On the basis of destructive changes in certain layers of the geniculate produced by raising monkeys in a red light, and on the basis of the destruction of their layers by toxins producing a loss of sensitivity to blue light, Le Gros Clark thinks that colors are separated into separate channels in the lateral geniculate, and so, in their projection to the visual cortex, notably the area striata. But to my mind the evidence is not yet conclusive.

Still another theory with no evidence so far as I can see would make the distinction in the area striata on the basis of excitation of different layers. I have only mentioned it for the sake of completeness.

If the eye codes hue by which fibers are activated, there must be differences in the eye which might be anatomically detected. But no one, so far as I have found, has yet published the histological picture of normal versus color-blind eye, geniculate or cortex. People have collected statistics of color-blindness in England recently. These confirm the frequency of deficiency at the red end of the spectrum, the hereditary nature of the recessive red-green blindness and of the rare blue-yellow blindness, which is apparently not sex-linked. It is devoutly to be hoped that these known cases will be followed to the autopsy table. So long as we are ruled by undertakers it will be impossible in the United States.

If hue is not coded by the eye space-wise, into separate axons, it must be time-wise, by pulse-interval modulation of impulses over those axons; that is, a change in the frequency of pulses, each of which is an all-or-none signal traveling in an axon. Such coding is known in on-effect, off-effect and during-effect, discovered by Hartline in the horse-shoe crab, in which frequency is clearly related to intensity of illumination, the frequency being roughly proportional to the log of the intensity after accommodation in fibers firing during illumination. There is nothing in this to suggest any relation to hue, but Alexander Forbes, working on the excised frogs eye, has now clear evidence of dissimilarities in the slow potentials of the eye which are related to the wavelength of the light and are not attributable to any change in the intensity. Such changes in slow potentials may modulate the frequencies of firing of the ganglion cells.

The eye is not only the most important of sense organs. It is the most complicated, being in reality an invaginated evagination of the brain itself. It is certainly not the simple three-layered structure of rods and cones, bipolars and ganglion cells that appear in our familiar diagrams. It receives axonal terminations from cells we know not where in the brain. It has many kinds of horizontal cells, and some that probably conduct upward toward the rods. Poljaks published work is sufficient to prove all this. But the bulk of his lifes work, now ready and waiting for publication, contains so much that is new and important that we can scarcely go forward without it. To get it published would cost a few thousand dollars, and those dollars had better be spent soon, for they will push the science of vision ahead faster and farther in that publication than by any projected research proposed to date.

But no anatomy will answer the great remaining question. The eye of the scallop, as Hartline has shown, has a layer of cells that fire continuously in the dark, and are inhibited by light. Today we do not know whether or not the mammalian eye has cells similarly active. If it has, it is not unique among our sense organs. If it has, then our sensation of light of any kind may be conceived not as mere all-or-none impulses merely triggered by light, slowing to a frequency proportional to the log of the intensity, but as a true interval-pulse modulation. To tell this we must work on unanesthetized and almost undamaged eyes, and that work with micropipettes has only just begun.

We come at last into the central coding of vision with a frank admission that we do not know how hue is coded possibly spatially, that is into separate axons. On, off and during are certainly in part thus spatially coded. Some axons carry only on effects, others carry only off effects, still others during, and some, two or three of these. Coding of intensity can easily be preserved through the geniculate to the input to the visual cortex, or, by-passing the geniculate, it may go to the superior colliculi of the midbrain and to a nucleus controlling the pupillary diameter to decrease it as the illumination increases. Spatial relations in the retina are spatially coded in the geniculate, in the cortex and in the superior colliculus. In the geniculate they are cut and stacked like maps in an atlas. In the cortex, each half of the visual field, transmitted through one lateral geniculate, is mapped on the cortex, inverted and reversed. It is distorted so that to a first approximation the distance along the cortex from foveal representation is proportional to the logarithm of the angle subtended at the eye. Thus a disproportionately large portion of the cortex is devoted to central and near-central vision. In the superior colliculus the distortion plots the central visual area anteriorly, the upper field, medially, the lower field, laterally. This, with its efferent connections going to the oculo-motor system makes sense out of its function of turning the eyes to the source of a light, as has been beautifully demonstrated by Julia Apter. Of the visual function of a superior colliculus in primates there can be little doubt. Heinrich Klüver and Paul Bucy destroyed the visual cortex of monkeys. They lost all form vision, could not distinguish a small bright area from a larger that was less bright if the total luminous flux at the eye was the same for both, but they could distinguish a greater from a smaller total luminous flux. This is vision of a kind and, since most of the geniculate proper degenerates in such a preparation one in which the visual cortex is gone one is almost compelled to look to the superior colliculus for the remaining function. Anyway we know that it does turn the eye to a source of the illumination. It must, therefore, sense it.

I have a sneaking suspicion that it is here that we must look for a prolonged burst of activity in migraine. In the first place the typical scintillating scotoma that moves slowly and fairly symmetrically outward from the fovea toward the periphery of one half-field would require a very carefully prescribed spread in the visual cortex, and one without a substantial basis in anatomical connections. Were it cortical one would expect to find an associated abnormality, a real fit, at the occipital pole. I and several of my friends who suffered from migraine have sought it often found it never. In the second place such a spread could scarcely be supposed to occur in the geniculate, which lacks the proper topography and connections. In neither of these places would the hemicranial headache accompany the scotoma. In the superior colliculus this spread of the scotoma would be easily accounted for by a march along known anatomical lines. The headache should occur, for excitation in that superior colliculus might catch axons or collaterals, chiefly those representing the head, swinging from the spinothalamic tract (which carries pain and temperature) to end in the fibers that surround the periaqueductal grey matter under the superior colliculus.

In primates form vision is a cortical function. In saying that I do not intend to exclude its partner in the pulvinar. Cortex and pulvinar probably work in continuous give and take. But without the cortex there is no form vision and no color vision in man. It is difficult to conceive any mechanism that will account for our ability to perceive shape regardless of size. Certainly Gestalt notions will not work. I do not merely mean that they violate Ohms law, and conserve currents, where they are actually moving sources and sinks. I mean they yield, even in theory, only particular images. Norbert Wiener proposed that some sort of scanning by planes through volumes underlay form vision. In a paper with Walter Pitts I made use of this notion, supposing a plane of scansion to move up and down through the layers of the cortex through which information concerning the visual world rose along the input channels, branching widely as it rose. The scanning sweep would then sum with the incoming impulses to excite cells at successive levels. Thus for a centered apparition it would produce in the output of the visual cortex expanding and contracting replicas of the original input. By summing such transforms one would obtain shape, regardless of size, in terms of a set of average values of suitable functionals to each transform as a distribution of excitation in the space of that matrix of relays. Such a collection of invariants depends in no way on the original size. The notion fits a lot of facts but Lashley believes that he has anatomical evidence that the branching of the afferent channels is inadequate to account for this process. This does not seem to me crucial, for if it is not in the primary afferent, there are plenty of internuncials whose axons do scatter through the recipient layers in the visual cortex. Lashleys second objection arises from destroying large areas of the visual cortex in the rat, leaving only small fragments; he finds that the animal can still make differential responses to dissimilar figures. I am not sure that this objection is conclusive. To know form, regardless of size, is to have a universal notion triumph so to speak, quite apart from the size of the object in question. It may be that while the rat sees an angle and can distinguish it from a curve, it has not a size invariant stimulus equivalent. But Donald MacKay, while collaborating with me in my own laboratory in Chicago, produced a square on the screen of a very large oscilloscope. This square could be made to expand and contract at set frequencies around 10 per second or could be coupled to the brain waves. Viewing such a changing figure in no way disturbed our form vision. This does challenge my theory which now requires a hypothesis ad hoc to defend it. I know of none that is scientifically warranted.

Men born with congenital cataracts, which prevent form vision from dominating either the growth or learning of the visual cortex, when those cataracts are removed, tend to trace outlines and count curves and corners. A theory of visual perception of form can be conceived along these lines. It has been independently proposed by Strauss and by MacKay but I doubt that this is the normal procedure. Such a theory does not match several properties of vision as we know it. Such a perception of form would be hopelessly upset by holes punched in the visual cortex whereas the mechanism Pitts and I proposed would not be, and of course, recognition of visual form is not. But, perhaps more importantly, it is not related to the peculiarity of vision first noted by Craik and elaborately studied by Stroud, namely, that we see in snapshots, each devoid of motion taken at a rate of about 10 a second. We judge that motion has occurred on a basis of two such snapshots, and acceleration on the basis of three. This will fit well with a sweep of scansion of 10 per second or so the so-called alpha rhythm of idling brain. To this frequency we shall return later. It seems to be related to the length of those messages which underlie most perceptions and much in the course of motor performance. It was first noted by Babinski as that frequency for which the threshold of faradic induction of motor response by stimulation of the motor cortex has the lowest threshold. He noted also that it was the maximum rate for the vibrato of the violin and for the rate of pronouncing of syllables. As yet we have little direct evidence that it is the alpha rhythm of idling cortex that is directly related to these things. Bishop first noted its relation to the cortical response to stimulation over geniculo-cortical afferents. Its fixed phasic relation to voluntary movement has been studied chiefly in Cardiff. But in vision in the waking state its relation to the snapshots is not yet established. None the less I still think that our theory fits more facts better than any other yet proposed.

There are two more points about the eye itself which are apt to be significant to us soon in investigating central mechanisms, particularly the central coding of information.

In reading, the line of gaze shifts more than the eyeball rotates in each saccadic movement of the eye. The excess may amount to as much as 6 degrees. This used to be attributed to a shift in attention to a different part of the retina. It is now clear that it is due to the action of smooth muscles, which either slide the iris or distort the lens, for it is lost or paralyzed by local application of homatropine. Any of you who are interested in recording optokinetic nystagmus, to discover whether our predictive filters can handle translation as well as rotation, will naturally record the motion of the orbit electrically. It is a good dipole and a good direct coupled amplifier and an oscilloscope are all you need to record it. But your answer may be in error by 6 degrees unless you use homatropine.

The second point is equally exciting, particularly to the electroencephalographer who is interested in perception. For some time oculists have know that when a man fails to recognize a word or other symbol on a page in front of him his focal plane moves outward. When he recognizes it, it moves inward, coming much closer than the page. In babies this shift may amount to 7 diopters in all. It is easily measured by the relation of the reflections from the two surfaces of the lens itself. Now every neurologist also knows that when the eyes accommodate for near objects, they also converge. Christine Kris, a psychologist who has been working on metabolic conditions affecting repetitive activities, had been studying the frequency of the reversal in the aspect of the cube, or the flight of stairs drawn in outline, and found a maximum rate of 5 per second. She came to my laboratory with this question, Is there an associated shift in convergence that can be electrically recorded? Dr. Antoine Remond and I put electrodes on all three of our heads by turns and watched the cube reverse, marking each reversal on the moving paper. There was always a record in the electroencephalogram corresponding to the eye movement there; that is, convergence did occur with each perception of the cube. Now the eye is a strong dipole with a centralized mass rotated rapidly by strong muscles. Consequently it is an excellent tell-tale for the perception. I believe that Dr. Remond has worked on the next step in the problem, looking for a corresponding change in the brain waves in the occipito-parietal cortex. There may be something there, but to find out how it is related, if it is there, to a movement of the eye, we shall require a type of electrical correlator that can detect their phase relations. We know how to build them, and if all goes well we may have the first indication of when and where perception occurs by its electrical signs in the cerebral cortex. I think that work should be pushed ahead.

We come next to our second great source of knowledge of the world and of our relation to it: all of those receptors that detect acceleration, whether through water, like the lateral line organ of the fish; by gravity, like the otoliths; by linear acceleration, like the utricle; by rotation, like the semicircular canal system, or by air-borne vibration, like the cochlea. In each case De Vries has found that nature employs the same device, namely, a glob of piezoelectric jelly in which is embedded the hair of a hair cell which initiates the nervous impulse. To what extent the distortion produced in the jelly, giving rise to the piezoelectric voltage, is responsible for the second electrically-discoverable step in the process, the so-called augmenting potential, is not known. It is conceivable that the second, in some cases, is but a rectification of the first. Fortunately for the experimenter, its axis of maximum voltage is rarely the same as that of the piezoelectric potential, and so they are easily separated. When the augmenting potential reaches a critical value, the nervous impulses are initiated and passed by the nerves to the brain-stem. This sequence has been most beautifully studied in the organ of Corti and the spiral ganglion and along the auditory nerve by Davis et al.

From a study of lesions in the ears of boilermakers, from direct stimulation of the turns of the cochlea in animals, from electrical detection of the signals of the spiral ganglion and the nerves and the central pathways, it is now clear that pitch is coded spatially all the way through to the cerebral cortex which pitch being a matter of which fibers are active. These, though topologically twisted, retain the topology given pitches in the spiral ganglion, through the tracts, the medial geniculate, and the auditory cortex. The highest pitches, from the lowest turn of the cochlea, arrive in the inner end of Heschls gyrus on the superior temporal plane; the low pitches, from the apical region of the cochlea, arrive in the outer end of Heschls gyrus. In primates there is a second auditory cortex adjacent to the first, wherein the pitches are represented in reverse order; but of its function we know relatively little.

In the ear, as in the eye, intensity is conveyed by frequency of impulses. The frequency after brief accommodation is nearly proportional to the logarithm of the intensity, that is the energy of vibration, at any one frequency. The envelopes of impulses in the nerve preserve their phase relation to the sound waves. The consequence of higher frequency in the sound is higher frequency in the envelope which persists through successive relays, but the exact phase relations are lost in the cortex and perhaps in the geniculate, and our hearing is notably deficient in detecting phase-shifts in pure and in complicated musical tones. However, we do utilize the phase-difference between our two ears to detect the direction whence a sound comes. Helmholtz notes that the difference of 30 microseconds is enough for us to be right two-thirds of the time about whether a sound arises to the right or to the left of the midsagittal plane of the head. This has been amply confirmed in recent years. Now 30 microseconds is about the standard deviation of the time a nervous impulse arises after stimulation and it is about the time that a neuron could resolve or discriminate, if signals were timed to fall at the end of the period of latent addition. So, clearly, a judgment depending upon a difference of 30 microseconds requires that the signals responsible for it must be no more than a step or two from the receptors. It could first be achieved when the impulses from the two ears first come together. It would require somewhat more than the diameter of a single cell along the region where impulses are coming in opposite directions along their respective axons. One has merely to suppose that only those cells fire on which impulses from the two ears converge synchronously to get this judgment of interval. These relays probably play on structures turning the head and eyes toward the source of the sound, eventually reach the cortex, enabling us to say that it was to the right or left of straight ahead.

But let us return to the cortex which detects intervals regardless of pitch. Here the incoming axons slant up through the konio-cortex, some hundred relays deep. The sweep of scansion, like that proposed for vision, would carry the output up and down the axis of pitch. Since in this cortex octaves span approximately equal distances, as they do on the piano keyboard, the translation of two tones up and down the axis would give an output-conserving interval. A servomechanism that stops that sweep of scansion when the interval had been brought to some canonical pitch would yield an output-detecting interval regardless of pitch. Or the invariants under these transformations of pitch might be made as proposed for visual perception of shape regardless of size. From my limited experiments in this field I think the latter more likely but much work needs to be done before one can make a flat-footed assertion touching the mode of action of the auditory cortex. Again, the rate of scansion would need to be about 10 per second to account for our appreciation of music or understanding of syllabic speech. Under optimum conditions human audition is so accurate that the thermal agitation and the souffle of the blood in the vessels of the ear are just about audible. So, under normal circumstances, there are many impulses coming over the auditory nerve; but, except for the directional effect, they are not to be thought of as pulse-interval modulated. They are just due to the noise always present.

It is otherwise with the vestibular portion of the eight nerve. Almost any fiber of this nerve, whether it arises in the vestibule, in the utricle, or in the otolith, shows in the resting state a remarkable constant frequency of impulse. This frequency is modulated by acceleration in the manner appropriate to that particular receptor. Cold water chilling the ear, or warm water heating it, starts circulation of the fluid in whichever canal is in the vertical position and this modulates the interval between the impulses in the axons coming from that canal and only in them. The same is true for the action of gravity as a function of the position of the head. Roberts in Glasgow has made a submammalian otolithic preparation which can be removed from the head and it will continue to work normally for a long time with no blood supply. It contains the otolith and the jelly hung by hair cells in a chamber full of fluid. The rate of discharge of its axons is found to be a function of the inclination angle of the structure to the vertical.

Not only is this input space-coded for the kind and direction of acceleration, and pulse-modulated for intensity and angle, but the recipient nuclei in the brain stem behave in the same manner. So also does the cerebellum, which was originally derived from the more dorsal portions of these nuclei.

Moruzzi was the first to place microelectrodes on the Purkinje cells and find them popping at very constant rates in the absence of significant changes in their input. He was then working with Snider, who has studied the modification of this rate by volleys afferent to the cerebellum from several other structures. The effect of such volleys is to increase the rate, then at the end to decrease it again briefly, and return it to normal; or else there is a decrease followed by a transient increase and return to normal. This may have the obvious relation to the function of the cerebellum, whose business it is to bring to rest whatever is put in motion. This it does by shadowing the rise in tension of an agonist putting the mass in motion by a rise in tension of the antagonist to stop it. In order to do this in time to stop it at the proper place, it must be informed beforehand where the part is to stop, and informed continuously of its motion and acceleration; from all of which it must precompute the necessary sequence of orders to stop it at the time it arrives. A failure of this action we commonly call ataxia, or, more carefully, cerebellar ataxia. Normally, if you pull against a weight and your hand slips, it stops before it breaks your ribs. In the absence of the cerebellum, a check is wanting, and the consequence disastrous.

Just as the parts of the body map in several parts of the cerebral cortex, for sensation and motion, so do they map on the cerebellum; and the maps are connected point-to-point, certainly from the cerebrum to the cerebellum by the cortico-ponto-cerebellar system, and probably from the cerebellum to the cerebrum by the cerebello-rubro-thalamo-cortical or the cerebello-thalamo-cerebral systems. Such representations are truly topological codings, like that of the visual system to its geniculate and cortex and to the superior colliculi; and the cochlea to its nuclei, geniculate and cortex. To work properly, the cerebellum must be in collaboration with the cortex and must receive its instructions from the basal ganglia, which program our movements, including our voluntary activities. This is done by partially known pathways. But for governance of position in motion, the cerebellum must be continuously informed of position, motion and, above all else, of the accelerations of the parts in question, as I have already indicated.

It has its own vestibular input and receives impulses over at least three systems from the spinal cord: the dorsal and ventral spino-cerebellar tracts, and the arcuate nuclei. Each of the former systems is topographically organized. Of the arcuate system I know almost nothing. Unfortunately, the work on the spino-cerebellar systems has been done under barbiturates, which so decrease the rate of recovery of neurons and thus the excitability of their cells of origin at any particular time that we can say almost nothing of how active they are in the normal state in the absence of stimulation. Hence we cannot tell whether they also are pulse-interval-modulated by stimulation.

Lloyd and Mackintyres work indicates, notably for the dorsal spino-cerebellar system, that the most important afferents arise in stretch-receptors. These are small muscle fibers enclosed in sacks imbedded in muscles. Afferent peripheral axons wind around them. When the muscle is stretched, these terminations send in streams of impulses whose frequency increases with the intensity of stretch. Kuffler has shown that these little muscle fibers in stretch-receptors innervated by fine fibers from the ventral root, when activated by these efferents, cause a greater rate of afferent impulses for the same stretch. This kind of tuning of receptors gives both flexibility and precision of control by inverse feedback. In fact, the muscle may be set for proper tone and for a given intensity in response to a given stretch. The nature of these trains of impulses should give us pause in interpreting the knee-jerk as a clinical manifestation of the neurophysiologists most useful artifact the monosynaptic reflex. It exists. But the knee-jerk is almost certainly a response to a barrage, not a single volley, of impulses from stretch receptors. Here again intensity is coded into frequency.

While we are considering the motor system, let me add that stimulation of the basal ganglia, at say forty pulses per second, elicits repetitive movements like those lost to the Parkinsonian. At threshold voltages they recur at but one or two movements per second, but as the frequency, the pulse duration or the voltage of stimulation is increased, they increase in frequency to a maximum of about ten per second. The particular variety of movement is space-coded, different locations in the basal ganglia giving repetitive movements. One point will yield a sequence intended to convey food to the mouth and its reception by the mouth; another will produce walking; a third, boxing; and a fourth, one well adapted to catching a mouse or a fish.

And we should note this also: stimulation of the so-called motor-face region of the cerebral cortex at fourteen or more pulses per second produces contraction of the lips and cheek on one side, thus baring the teeth; but at ten or less pulses per second it produces discrete unilateral twitches of the tongue. One may destroy the cortex and stimulate the descending fibers, and obtain the same result. The impulses are thus obviously descending from the cortex over the same fibers, and we know that these impulses conserve their frequencies. So it must be that frequency filters at lower levels accomplish the diversion of excitation in one case to the seventh nerve nucleus, in the other to the twelfth. This is a clear example of one variety of pulse-interval modulation and its detection. It was first demonstrated by Percival Bailey.

One more interesting item in the motor system must be considered. Swallowing is practically an all-or-none macroscopic reaction. It can be invoked by stimulation of the recurrent laryngeal nerve. What is strange about this circuit is that it is responsive to the number of volleys it has received. Doty has shown this in several species. There is no response until the total number of volleys reaches the threshold number. Then it occurs in toto. Moreover Doty has found the limits of its reliability. Like all counting circuits, there is a maximum rate at which it can count and there is a lower limit to the rate at which it can count as if it had forgotten the previous number. The details of how this is coded centrally are unknown, and neither he nor I would like to hazard a guess.

So much for the motor system and its most significant afferents. It is of more interest to the neurologist than to the psychologist, for it runs itself beautifully, smoothly, and automatically, so long as the brain, cord and nerves are intact. Normally we are no more aware of its doings than the cat who lands on its feet, regardless of how it is tossed into the air without bothering consciously to compute what it ought to do on the basis of any theory of the conservation of angular momentum.

Concerning taste I know nothing that throws any light on central coding. On the basis of the distribution of points sensitive to its four modalities, its impulses are thought to travel over separate axons. The relay stations are fairly well known all the way to the cortex, but there is no evidence concerning their distribution in time or space within these stations.

Until Adrians work appeared in the Journal of Electroencephalography and Clinical Neurology the same was true of smell. There was a controversy over its proper stimulus. The older schools attributed dissimilarity of odors to dissimilar polar ends of volatile molecules; the school of psychologists and chemists at Yale, to the dissimilar absorption bands in the near infrared. But we knew nothing of its physiology, despite our immense knowledge of the central connections of its receptors. Thanks to Adrian, we now know that in the waking state an odor introduced to the nose at each breath produces a sudden brief intermission on the otherwise regular discharge of the olfactory bulb. Presumably, then, smell is mediated by this change in frequency; and what the smell is probably depends on which fibers are affected. We should know something of this in the next few years.

This leads us to consider touch. So far as the periphery is concerned, the story of this modality is more confusing than it was heretofore thought to be. Each sensation had been neatly attributed to one or two specialized types of end-organs, except burning pain, which was attributed to bare and diffuse endings. Each such ending was supposed to be connected to the central nervous system by an axon of appropriate diameter. It was thought that there might be several specialized sense organs of some one kind, employing a single afferent neuron. Now it turns out that a single such afferent axon may be, often is, connected to several dissimilar endings. Burning pain was attributed to one and only one variety of afferent fibers, namely, C fibers, which are supposed to import no other information. But in the pulp of the tooth there are no such afferents. Instead, there are only myelinated fibers of relatively large diameter and appropriately fast conduction and low threshold to electrical stimulation. Yet their endings are bare, and they do mediate both burning pain and so-called quick or sharp pain. Moreover the pinna of the human ear detects cold, cool, warm, hot, pressure, touch and pain, but is now thought to have only the diffuse types of endings in the dermis and around hair follicles. Under these circumstance, it is obviously foolish to look to the endings as uniquely related to the modality. We must seek such a modality either in the temporal sequence of signals along particular axons or in the spatio-temporal configuration of such signals in the spinal cord or higher. This is the way it looks to me today. It is not a guarantee for tomorrow. We shall soon have better evidence from microelectrodes touching the position and motion of sources and sinks in the cord after many varieties of stimulation. If one had to guess today he would put his money on this: that enough afferent channels excited often enough might start up persistent activity in the substantia gelatinosa of Rolando. It is the only part of the internuncial pools of the spinal cord in which there can be enough closed paths to sustain reverberation. Such a reverberation may determine which of the relay nuclei of the cord are to respond to a given input. For instance, it might determine that a subsequent efferent barrage excited the cells of origin of the spino-thalamic tract which mediates pain and temperature.

The dorsal column is but upward prolongations of relatively large afferent peripheral axons, and its stimulation in man does not result in sensations of pain and temperature, but may give rise to some sensations of touch or motion. So the spatial separation of these sensations begins in the cord and is well preserved in the tracts to the thalamus. Beyond this, pain and temperature, except pain from the tooth, have shown signs of arrival of their impulses in the cerebral cortex. But there is a second spatial segregation in the case of pain. As the spino-thalamic fibers traverse the midbrain, axons or collaterals leave its bundle to pass under the deepest layers of the colliculus, where they end in the mesh of fibers surrounding the central grey matter. Section of the spino-thalamic tract in the mid-brain below their divergence removes the sensation of pain and the accompanying affect called suffering, whereas sectioning above the divergence eliminates the sensation but leaves the affect. So much for what we know of how sense organs make Alcmeons opposites, and of how they are coded in their respective nerves and in the central nervous system. Originally I thought that the central nervous system handled this information as a digital computing machine would, were it made out of simple telegraphic relays. The striking and relaxation time of the relay limit the rate at which information can be relayed. In the nervous system, these times, called synaptic delay and the duration of the impulse, are about a millisecond, so that the maximum rate obtainable would be about one bit of information per millisecond. The highest sustained rate of repetition is, let us say, a third or a fourth of the maximum, for fatigue raises thresholds and effectively limits the duty cycle. When I said that the brain behaves like a digital computer I did not mean that it employs a positional nomenclature for ascending powers of a radix, say two or ten. It employs position for other purposes, notably for those aspects of the information which are space coded in its input. Rather I had in mind that it uses neurons as coincidence detectors, so that its output, in any one millisecond, namely, an impulse or none, implies the configuration of impulses in the previous millisecond which are sufficient to excite or inhibit it. It was to keep track of impulses so considered that Pitts and I originally proposed the use of the calculus of propositions, subscripted for the time of occurrence of the signal of a given neuron. We quantized that time, taking as our unit a duration about equal to synaptic delay or nervous impulse. Such a simple calculus makes it possible to describe the circuit action of elaborate computers and to provide, relatively simply, the converse of Turings thesis: Let us say that a man using a pencil and paper can compute any number that is computable by a Turing machine; or, what amounts to the same thing, that he can deduce any deducible consequence of a finite set of premises; or again, that he can detect and react to any figure of excitation in his input; or, in still other words, that he can have appropriate ideas of the world about him in so far as it affects him. All of which is an elaborate way of saying that all general Turing machines are, in this sense, equivalent. This makes it clear that the physiologist who is interested in the functions of brain can always set up some mechanistic hypothesis as to how it works and that his principal business on the theoretical side is to make those hypotheses which conform to what is known of the anatomy and the physiology of the parts of the brain subserving the functions he is studying. It is the business of his experiments to make him guess again.



For further research:

Wordcloud: Active, Afferent, Along, Axons, Brain, Cells, Central, Cerebellum, Change, Coded, Connections, Cortex, Detect, Ear, End, Excitation, Eye, Fibers, Form, Found, Frequency, Function, Geniculate, Impulses, Information, Input, Intensity, Light, Motion, Movement, Nerve, Normal, Organs, Pain, Per, Pitch, Produced, Rate, Relation, Relays, Response, Sense, Stimulation, Studied, Superior, System, Theory, Vision, Visual, Work

Keywords: Fails, Croton, Function, Life, Nerve, Wiring, Brain, Looms, Today, Brains

Google Books:

Google Scholar:


1 Published in Synthese, International Journal for the Logical and the Psychological Study of the Foundation of Science, Vol. IX, 3: 3-5, F.G. Kroonder, Bussum, The Netherlands, 1954.