W.S. McCulloch


The experiments that gave rise to the following notions were all performed on the primates, Macaca mulatta and Pan satyrus, although the counts of neurons are usually on man. In all of them the cerebral cortex has come to outweigh and, in neurons, to outnumber the rest of the central nervous system. As in all such evolutions of the cephalic end of the dorsal plate, mere size permits greater distinction of functionally dissimilar portions and of their various paths of communication. But with its majority has come not only an undertaking of selected functions of more caudal portions of the dorsal plate and a control of their activity by impulses projected upon them (1), but also interference with the execution of their functions by impulses upon one or more of the systems relaying their signals toward the final common path (2).

Some of those systems, say the cortico-ponto-cerebello-dentato-rubro-spinal (3), have lost so much of their projection upon the cord that the output of their remnants returns in no small measure to the cortex (4), so relating its events indirectly. In primates, the plurality of all lower neuron are thus re-entrantly involved. Among remaining caudad systems, I know none which does not receive, directly or indirectly, signals from the cortex, nor any sector of the cortex that does not signal to some of them. But, as they still receive their proper peripheral afferents their output is the result of at least two streams - one from the cortex and one or more from elsewhere in the dorsal plate (5, 6).

To that great extent to which its afferents inform it of the peripheral consequences of the action of its own afferents (7, 8, 9), any system is part of the path of a reflex. And most if not all purely reflexive action is negative feedback (10) tending to reduce all deviation from an intended peripheral state measured by the afferents (11). Left to itself each such circuit is homeostatic (9). But, to that extent to which its central nervous components can be influenced from elsewhere, notably the cerebrum, it can be made to seek other states or, to use an engineer's term, be reduced to a servo-mechanism (11). Nothing but the name is new.

In 1826 Sir Charles Bell read his famous paper on the nervous circle (7) before the Royal Society; and in 1868 Clerk Maxwell read his on the governors of steam-engines (12), with much of the mathematics required for these negative feedback, or reflexive, systems (13). Had Hughlings Jackson founded his hypothesis (14, 15) upon their ideas instead of resting it on Herbert Spencer's Epicurean psychology (16) he would certainly not have pictured the second level (17), especially the motor cortex, as the base of a piano-accordion whose every button sounds a chord (17, 18). Nor does he save his theory in admitting cooperation or competition from other structures, notably the cerebellum (19), for it makes no great part of his account of the motor cortex and the pyramidal tract (14).

Let experiment have frozen proper activities of servo-mechanisms by fixing bones, muscles, joints, etc., then excitation of one or few Betz cells will evoke what Jackson called movements (16), that is to say contraction of few muscles (20) and, perhaps, relaxation of antagonists a complexity not allowed by the classical description of the motor cortex as a harpsichord with a cell to pluck each string (21, 22, 23, 24). No better is Walshe's notion (25) that these cortical efferents, like the push-buttons of some monstrous juke-box, serve each to evoke entire some innate or learned pattern of movement, (26, 27) which dwells in one such efferent at least and is not else effectable (27).

These inept conceits accordion, harpsichord and juke-box alike failed to predict the disparate consequences of stimulation at diverse frequencies (28). Percival Bailey has shown us that from all parts of Face Area 4, where twelve or more impulses per second regularly evoke contraction of facial muscles innervated by the seventh nerve, ten or less per second elicit only movements of the tongue, innervated by the twelfth alone (29). As he excited the efferents directly, and impulses descending conserve frequency (30), only filters in lower structures accomplish their diversion.

In familiar phrase, we might concede that frequency as well as place must enter our idea of the mode of representation of motion, and conceive the cortex an English horn. This opinion, as sweet, equivocal and perfidious as its three precedents, is as legitimate. For the lexicographer (31) will allow all four, and twenty others like them, quite impartially, each with a good sense of to represent. All are metaphorical, compatible, assert little, predict nothing. It were less deceptive to have cortex represent that internuncial ocean where its axons end better, to do nothing of the kind (32).

Now nothing from the cortex partakes in the functional organization of the cortex unless the recipient internuncials affect the cortex again, and only so much as the corticifugal activity informs the return. We shall treat such feedback under Indirect Functional Organization, merely remarking that the cortex does thus affect its sensory input (33, 34).

The number of corticipetal fibers has been variously estimated, but I can find no one but Earl Walker brave enough to make it a hundred million, even in man. From the relative density of thalamic to extrinsic cortico-cortical fibers we may admit of the latter as many more, i.e, 10. The number of cortical neurons to be connected we may likewise round off at 1010 again in man (35). Thus indirect and extrinsic fibers together are no more than a fiftieth of the associators of cortical events, for we can estimate the actual number of intrinsic cortico-cortical axons by recalling that less than one in a hundred leaves the cortex. This gives us again a hundred intrinsic to one extrinsic and one indirect cortico-cortical connection. To the ramification of axons we look for the multiplication of synapses required for summation.

We turn, therefore, to the grey mass of the cortex. Histological pictures by Cajal (36) and Lorente de Nó (37) show we can separate its vertical from its horizontal constitution. Conceive, then, the cortex as a mass of cells, say a hundred in depth (38, 39), divisible into ill-defined layers containing principally neurons of given kinds, so connected from layer to layer that there are skip-distances of one layer in their vertical synapses except in the outermost and innermost where cells of the odd and even series associate (37). In the intermediate layers are chiefly neurons whose axons are vertical with only subsidiary horizontal ramifications. Such vertical systems of neurons can reverberate (39, 40, 41) longer than if all layers were equally thoroughly interconnected by single axons. They can easily tolerate rhythms whose period is determined by a full circulation through the hundred cells, for this, with reasonable allowance for slow conduction and synaptic delays, demands but ten impulses per second in each neuron. And we know some neurons that fire for long periods at twenty times that frequency (42, 43).

Yet, it has become increasingly evident that this rhythm usually recorded from idle cortex may not be attributable to it alone (44, 45), but rather to its ability to follow without loss that frequency imparted to it by unspecific afferents from undifferentiated thalamic structures (46, 47, 48, 49, 50, 51, 52). As yet there is nothing to suggest that it is anything more than a happy coincidence of the natural periods of intrinsic and indirect functional organizations, which may become unhappy when three per second impulses appear (53, 54), leading to petit mal, with its spike and dome (55) that block all proper function of cortex and thalamus (56). These gross fluctuations of voltage through the cortex, having properties largely dependent simply on distance, resemble all disorders which introduce fields (57, 58, 59), be they physical or chemical or both, into mosaics of cells each capable of a discrete decision impulse or none and thereby consume the effective degrees of freedom without contributing to the precise ordering of their decisions, that is, to information (60). In 1936 Bishop showed that even the mild fluctuations, or alpha rhythm, of idle brain have a somewhat similar effect (52).

To him we are yet more indebted for the first proof of any function of the largest single bundle of extrinsic cortico-cortical connections, the corpus callosum, for he showed that, on cutting the visual cortices loose from the geniculate bodies, it swept them into phase (52). Van Wagenen and Akelaitis (61, 62, 63, 64, 65, 66) have proved its unimportance by exhibiting the relative normality, including the continuity of visual perception (65), after its section in epileptics, in whom the section frequently prevented spread of grand mal convulsions across the midline. Its ability to sweep the other hemisphere into such seizures, Erickson proved (67), and I can confirm, with all paths through lower structures interrupted. Had Van Wagenen included the anterior commissure in his transections probably all his operations would have succeeded (68). Although I have spent much time strychninizing one square millimeter of cortex in primates under Dial and recording the impulses over the extrinsic system (69), I for one am convinced that, although it alone can keep the cortex promptly informed of its own remote activity, no more specific task can yet be surely ascribed to any part of it. To generalize convulsions is scarcely its duty. Our inability to discover its role is exasperating, for with 1010 neurons organized vertically by hundreds, there are some 108 such vertical groups and hence nearly enough extrinsic axons for one to go to each group.

But before we return to these groups whose numbers and arrangement ultimately determine the cytoarchitecture of the cortex, let me say that Bailey and von Bonin have completed their study of this aspect of Macaca Mulatta. (70) and are well along on Pan Satyrus (71), that histological facts have compelled them, to adopt von Economo's cartography (35) and that all other neuronographic findings on these extrinsic connections when so plotted show regularities we sought in vain in Brodmann's scheme (72). Sooner or later we shall adjust nomenclature to the ineducable vertical structure of the cortex cerebri.

We have seen that this structure could support rhythms requiring say 100 synaptic delays but we know that the time for reflections through the thick of the cortex is often less than 10 synaptic delays (52, 73, 74) with due allowance for conduction. An incoming volley of specific afferents initiates a surface-position wave of say 2 milliseconds (52). This is followed by 4 or more milliseconds of surface-negativity (52) while it inhabits and spreads in superficial layers. Then it descends, taking perhaps 2 more milliseconds (52). A primary sensory area bombarded by such afferents properly timed at lower levels might then be driven at 120 cycles per second, if output and input did not interfere. Presumably they do, for its highest recorded frequency is that at which flicker fuses, about 53 per second (75, 76, 77, 78). Now there can be but one corticifugal fiber for the whole 100 cells of which some or all have come into play to determine, in a matter of say 5 synaptic delays, whether that efferent shall fire or no.

The maximum information (60) that may enter such a decision of the corticifugal fiber in this short time is the time in synaptic delays multiplied by the number of neurons, i.e., 5 x 100, or as much information as can be conveyed in twenty five-lettered words. Since these neurons are arranged in reverberating chains (39, 40), a second impulse, following a first within any period during which their reverberations are still informed by the first, has a fate determined in part by that active memory (78). Moreover, these columns are not merely linked with remote columns by one extrinsic fiber, but with adjacent columns by numerous fibers (39, 40, 79) and with others in the neighborhood by fewer, the number decreasing with increasing distance. Thus, the information brought to bear in determining the twenty words-worth of information is many times more. It increases with the duration of a figure in the impinging world; and all our perceptions of tone, chords, positions, shapes, etc. become blurred (43, 80) when one figure of excitement succeeds another at say 10 per second.

This demands a process which shall, from information radiating through the cortex out of the many channels that report the world, abstract one or more universals, or ideas (81). These may be found by scanning the recipient volume of cortex to discover figures of excitement regardless of position or size. To scan that space we shall detect and relay to fixed points the coincidence of the information with a horizontal plane of excitement that sweeps up and down the volume. Let it step but a cell's height per synaptic delay, and the time for clear perception, be it form seen or chord heard, becomes the required tenth of a second. Large electrodes on a cortex uninformed detect the sweep of scansion (82, 83), but on a cortex informed, the sweep is lost in the twinkle of details comprised in the perception (82, 83).

This fancy cleaves to fact, prescribes experiment, predicts outcome, invites refutation. It does not strain credulity, for mere statistical order suffices for nets to embody this process, inasmuch as small perturbation of threshold, of excitement, even of synapses has little or no effect on the average that is the figure, or idea. These nets are random in detailed synapses, for genes do not predestine termini to all neuronal ramification.

If connections in the felt-work of the cortex be random they are suitable to explain that surface-negative wave which expands in a circle about a mildly stimulated point in the cortex. Rosenblueth measured its rate of propagation from a strychninized spot under chloralose (84). The circle expands too slowly to be spread by anything but repeated synaptic relay. It diminishes as it spreads. But weak strychninization of an area near its origin renders that area hyper-excitable and then radial symmetry is lost, for the wave, with greater size and steeper front, races across the strychninized area (85). I could make little of this until Walter Pitts set up the mathematics for conduction in the random net (86). Now it appears as a verification of his theory that its velocity of propagation is determined by the rate at which spatial summation reaches a critical value in a random net and, therefore, by the threshold of the cells.

We have theories now that will force us to many experiments that may reveal to what extent the nets of the cortex are ordered, but none that will account for the ordering of these nets by learning, for this, like crystallization, is a change of state, and for such stochastic processes the mathematics is not yet (87, 88).

In closing, let me rehearse the general order that can be seen. Sectors of cortex, defined by their connections with thalamic nuclei (89), severally exhibit a receptive area or a pair juxtaposed but oriented reversely (90, 81, 82) each surrounded by a zone to which it gives many extrinsic axons. From each of these composite fields arises a projection whose activity affects that part of the body in which its organs of sense are situate. Thus the area striata (93) and its surrounding zones (94) move the eyes toward the place in the visual field corresponding to the position of excitation. The first and second somaesthetic areas and the zone anterior to the former and deep to the latter (5) in the Sylvian fissure, extending on to the Island of Reil, direct the carriers of their appropriate receptors, and the superior surface of the second temporal convolution pricks up the ipsilateral ear (96).

This close cortical coupling of each sensory to output affecting that sensory input by moving the organ of sense is a mode of functional organization whose circuit leaves and returns to the body. It is more indirect than the interception of sensory impulses to cortex which area 4s achieves via nuclei caudati by blocking thalamic relay (97, 98, 99). Yet circuits through the external world are but extensions of what Bell proposed merely through muscles in The Nervous Circle. On these re-entrant activities Cannon founded his conception of homeostatis (9), and Rosenblueth and Wiener, their theory of purposive behavior (100).

In primates these appetitive circuits pervade the cortex wherein their dominance is ordered and whence their dominion is effected. Yet the general plan of the cortex makes its own functional organization imperfect without control of input by output, over links beyond the present scope of physiology. They are to be sought wherever appetitive circuits traverse the public world. Unlike unicorns some of these links are always to be found in Parliament Square (101).



Le Gros Clark, W.E., Physiol. Reviews 22: 205, 1942.

Ward, A.A., Jr., and Reed, H.L., J. Neurophysiol. 9: 4, 329, 1946.

Bucy, P., J. Neuropathy, and Exp. Neurol. 1: 2,224, 1942.

Crescitelli, F. and Gilman, A. Am. J. Physiol. 147: 127, 1946.

Lorente de Nó, R., J. Neurophysiol. 2: 5,402-464, 1939.

Lorente de Nó, R., J. Neurophysiol. 1: 3.195, 1938.

Bell, C., On the Nervous Circle, etc. (Read before the Royal Society, February 16,1826.)

Magendi, T., Memoire sur quelques decouvertes recentes relatives aux factions due system nerveux. Lu a la seance publique de lAcadamie des Sciences, Paris, 1823.

Cannon, W., Physiol. Reviews 9: 399, 1929.

Black, H.S., Elec. Eng. 53: 114, January, 1934.

McColl, H., Servo Mechanisms. 1945: Van Nostrand.

Maxwell, C, Proc. Royal Soc. 100: 1868 (On Governors).

Heaviside, O., Electromagnetic Theory, 1,2 and 3: Electrician Pub. Co., 1893, et seq.

Hughlings Jackson, J., Med. Press and Circ. 10: 1, 329, 1881.

Hughlings Jackson, J.. Lancet. 1: 79, 1898.

Spencer, H., Principals of Psychology. 1855 (publisher not listed).

Hughlings Jackson, J., Med. Press and Circ. 2: 411, 1882.

Hughlings Jackson, J., Croonian Lecture. Part 1, March, 1884 (Lancet, 1: 535,1884).

Hughlings Jackson, J., Brain 22: 621,1899.

Chang, H., Ruch, T. and Ward, A.A. Jr., J. Neurophysiol. 10: 1, 39, 1947.

Fritsch, G. and Hitzig, E., Arch. Anat. Physiol. Swiss Med. 37: 300-332, 1870.

Hitzig, E., Untersuchungen ueber das Gehirn. Berlin, Hirschwald, 1874.

Beever, E.C. and Horsley, V., Phil. Tr. 181: 129, 1890-91.

Sherrington, C. and Leyton, A.S.F., Quart. J. Exp. Physiol. 11: 135, 1917.

Walshe, F.M.R., Oliver Sharpev Lecture No. 2, Royal Coll, of Physicians, of London, May 2,1929.

Walshe, F.M.R., Brain 50: 377, 1927.

Walshe, F.M.R., Brain 56: 104, 1943.

Wyss, O.A.M. and Crozier, W.J., Holv. Physiol. Pharmacol. Acta 1: 89, 1943.

Bailey, P., Personal Communication.

Adrian, E.D. and Moruzzi, S., J. Physiol. 97: 153-199,1939.

Craigie, W., Oxford English Dictionary (Represent) ca. 1910.

Barnard, R., J. Comp. Neurol. 73: 235, 1940.

Head, H. and Holmes, G., Brain 34: 102, 1912.

Dusser de Barenne, J.G. and McCulloch, W.S., J. Neurophysiol. 1: 2, 177, 1938.

Von Economo, G and Koskinas, G.N., Die Cytoarchitektonik. J. Springer, Vienna, 1925.

Ramon y Cajal, S., Histologie du system nerveux. Paris, Maloine, 1909-1911.

Lorente de Nó. R., Chap. XV in Physiology of the Nervous System. Fulton: Oxford University Press, 1943.

Von Bonin. G., Personal Communication.

Lorente de Nó, in Fultons Physiology of the Nervous System. Oxford Med. Pub. 1943.

Lorente de Nó, R., J. f. Psych. u. Neur. 45: H 6, 381, 1934.

Lorente de Nó, R., J. f. Psych. u. Neur. 46: H 2 u 3, 113, 1934.

Lorente de Nó, R., J. Neurophysiol. 1: 3, 207, 1938.

Stevens, S.S. and Davis, H., Hearing. Wiley, 1938.

Dusser de Barenne, J.G. and McCulloch, W.S., Am. J. Physiol. 127: 620, 1939.

Dusser de Barenne, J.G. and McCulloch, W.S., J. Neurophysiol. 4: 4, 304, 1941.

Lewy, F.H. and Cannon, W., J. Neurophysiol. 3: 388, 1940.

Morison, R. and Dempsey, W., Am. J. Physiol. 135: 2, 281, 1942.

Dempsey, W. and Morison, R., Am. J. Physiol. 135: 2, 293, 1942.

Dempsey, W. and Morison, R., Am. J. Physiol. 138: 42, 283, 1943.

Morison, R. and Dempsey, W., Am. J. Physiol. 138: 42, 297, 1943.

Morison, R. and Bassett, D.L., J. Neurophysiol. 8: 304, 1945.

Bishop, G., Symposia on Quan. Biol. Cold Spring Harbor 4: 305, 1946.

Ward, AA Jr., J. Neurophysiol. 10: 2, 104, 1947

Jasper, H, Penfield, W. and Droogleever-Fortuyn, J., 26th Annual Meeting Assn. for Research in Nerv. and Ment. Dis., December 13, 1946.

Gibbs, F., Lennox, W. and Gibbs, E., Arch. Neurol, and Psychiat. 46: 613, 1941.

Penfield, W. and Jasper, H., 26th Annual Meeting, Assn. for Research in Nerv. and Ment. Dis., December 13, 1946.

Gerard, R. and Libet, B. Livro de Homenagem. Rio de Janeiro, Brazil, 1939.

Libet, B. and Gerard, R., J. Neurophysiol. 4: 438, 1941.

Kohler, W. and Wallach, Proc. Amer. Phil. Soc. 88: 269, 1944

Wiener, N., Time, Communication and the Nervous System. Mss, in preparation for Northrop.

Akelaitis, T., Am. J. Psychiat. 98: 409, 1941-42.

Akelaitis, T., Arch. Neurol and Psychiat. 48: 914, 1942.

Akelaitis, AJ., Risteen, W.A, Herren, R.Y. and Van Wagenen, W.P., Arch. Neurol. and Psychiat. 47: 971, 1942.

Akelaitis, AJ., Risteen, W.A, Herren, R.Y. and Van Wagenen, W.P., Arch. Neurol. and Psychiat. 49: 820, 1943.

Akelaitis, AJ., J. Neuropath. Exper. Neurol. 2: 226, 1943.

Akelaitis, AJ., J. Neurosurg. 1: 94, 1944.

Erickson, T., Arch. Neurol. and Psychiat. 43: 429, 1940.

McCulloch, W.S. and Garol, H., J. Neurophysiol. 4: 7, 555, 1941.

McCulloch, W.S., Cortico-cortical Connections - Chapter VIII in The Precentral Motor Cortex. P. Bucy (Ed.), University of Illinois Press, 1944.

Bailey, P. and Von Bonin, G., Neocortex of Macaca Mulatta, University of Illinois Press, 1947.

Bailey, P., Von Bonin, G. and McCulloch, W.S., Studies on Neocortex of the Chimpanzee. Mss. in preparation. University of Illinois Press.

Bradmann, K., Vergleichende Loakalisationslehre der Grosshirnrinde, etc. Leipzig, Barth, 1909.

Marshall, Woolsey, C and Bard, P., J. Neurophysiol. 4: 1, 1941.

Bartley, S.H., OLeary, J.L. and Bishop, G., Am. J. Physiol. 120: 604, 1937.

Crozier, W.J., Wolf, C., Zerrahn, G. and Wol. J. Gen. Physiol. 21: 203, 1937.

Hecht, S. and Schlaer, S., J. Gen. Physiol. 19: 965, 1936

Bartley, S.H., Vision. Van Nostrand, 1941.

McCulloch, W.S., Finality and Form. Fifteenth James Arthur Lecture, N.Y. Acad. of Science, May 2, 1946.

Ramón y Cajal, S., Die Sehrinde. Leipzig, Barth, 1900.

Walls, G.L., The Vertebrate Eye. Cranbrook Inst. of Science, August 1942 (pp. 356-365)

McCulloch, W.S. and Pitts, W., Bull. Math. Biophys. September 1947.

Berger, H., Das Elektrenkephalogramm des Menschen. Nova Acta Leopoldina 6: 38, 173, 1938, Halle (Saale).

Gibbs, F. and Gibbs, E., Atlas of Electroencephalography.

Rosenblueth, A. and Cannon, W., Am. J. Physiol. 135: 3, 690, 1942.

McCulloch, W.S., mss. in preparation.

Pitts, W., Ph.d. Thesis in preparation. MIT Dept. of Mathematics.

Wiener, N., The Homogeneous Chaos, Am. J. Math, 1938.

Wiener, N. and Wintner, A., The Discrete Chaos, Am. J. Math, 1940.

McCulloch, W.S., Physiol. Reviews 24: 3, 390, 1944.

Woolsey, C. and Walzl, E.M., Bull. Johns Hopkins Hosp. 71: 315, 1942.

Talbot, C.N., Woolsey, C. and Thompson, W.G., Fed. Proc. Am. Soc. Exper. Biol. 5: 103, 1946.

Woolsey, C. and Fairman, Surgery 19: 684, 1946.

Walker, A.E. and Weaver, T., J. Neurophysiol. 3: 353, 1940.

Foerster, O., Lancet. 2: 309, 1931.

Chusid, J. and Sugar, O., demonstration.

Sugar, O. and Chusid, J., demonstration.

Dusser de Barenne, J.G. and McCulloch, W.S., J. Neurophysiol. 1: 364, 1938.

Gellhom, E., J. Neurophysiol. 10: 2, 125, 1947.

Barker, S.H. and Gellhom, E., J. Neurophysiol. 10: 2, 133, 1947.

Rosenblueth, A, Wiener, N. and Bigelow, J.. Philosophy of Science 10: 1, 18, 1943.

Walshe, F.H.R., Brit. Med. J. 1: 514, 1931.

Delgado, J.M.R., Fulton, J. and Livingston, R.B., Fed. Proc. 6: 1, 1947.

For further research:

Wordcloud: Activity, Afferents, Akelaitis, Arch, Area, Axons, Cells, Circle, Connections, Cortex, Delays, Determined, Excitement, Extrinsic, Fibers, Frequency, Functional, Hughlings, Hundred, Impulses, Information, Jackson, Layers, Lorente, McCulloch, Morison, Nervous, Nets, Neurol, Neurons, Neurophysiol, Nothing, Number, Order, Organization, Per, Physiol, Press, Psychiat, Sensory, Structures, Synaptic, System, Theory, Vertical, Whose, Wiener

Keywords: Cortex, System, Neurons, Systems, Cortico-Ponto-Cerebello, Mulatta, Activity, Afferents, W.A, Axons

Google Books:

Google Scholar:


1 Reprinted from Federation Proceedings. Vol. 6,2:448-452, June, 1945.
2 This work was aided by grants from the Joseph and Maty Markle Foundation, Josiah Macy, Jr. Foundation and the Rockefeller Foundation. The author wishes to thank Mr. Walter Pitts for his careful revisions of the manuscript and Dr. Jerome Lettvin for the preparation of the references.