This paper is based upon a lecture given at the 23rd Annual Meeting of the Society for Neuroscience in Washington, D.C., November 8, 1993
Dr. T.H. Bullock, Department of Neurosciences,
University of California, San Diego
La Jolla, CA 92093-0201
History seen by a professional historian, based only on the documented record, always incomplete and liable to bias, can be unreliable. Modern history seen by a protagonist must surely be among the most unreliable. My only excuse for this effort is that I was invited by the relevant Society committee. My reason for accepting is that I feel even the fragmentary part of neuroscience I can speak about is a human drama, romantic and exciting, and a flood on which we are floating, unable to dump the baggage of past biases. Our points of view, priorities, and positions on all the controversial issues and even the well established, noncontroversial ones, are not as rational as we would like to think but are strongly conditioned by where we came from.
I will depend mainly on selected vignettes of the way things looked when I was a student, a young postdoctoral fellow and an Assistant Professor, to compare with the way they look to me or to others now, in each of half a dozen mesoscopic domains. I mean by mesoscopic domains the middle levels - those in between the most basic subcellular or molecular and the higher levels of learning and cognition. The half dozen domains constitute of course, anything but a representative fraction of neuroscience. I believe, however, that they add up to a nontrivial segment of the big picture with respect to the integrative aspects of our science. Most of the fronts that grew into today's popular branches of our science are not represented but a small set of particular interest and probability of further surprises.
|The Doctrinal Neuron as a Functional Unit|
In 1936, when I was a graduate student at Berkeley, the neuron doctrine of Cajal (Jones 1994) was well established and accepted. There were still dissenters, mainly microscopists who said they saw fibrils or wisps of cytoplasm passing from one neuron into another. But, on the whole, they were not very successful in convincing people. At most their material was regarded as exceptions to a well documented rule. Incidentally, Rafael Lorente de No never accepted that the third order giant fiber of squid is a confluence of processes from many neuron somata of ordinary size, a view otherwise generally accepted since it was first described by J. Z. Young in 1936. The confirmation of Cajal that came with electron microscopy many years later - not with the first good pictures but gradually, as more and more examples were described, was no surprise or upset.
The main contrasts with current views were in the physiology. Ever since all-or-none spikes became easy to record, at first in fine twigs of sensory nerves in frog's toes in the '20s, and for a long time, even today in some quarters - spikes were all. The neural activity of neurons was the action potential; that is, the only potential of action known was the all-or-none impulse.
In order to begin a list of names and dates, let me select McCulloch and Pitts (1943) to represent this period when the whole neuron was a unit and acted together throughout, when we had no idea of multiple forms of spikes like the simple and complex spikes of cerebellar Purkinje cells, or of spikes that don't invade the soma or dendrites or of spikes that fail at branch points or short of the axonal terminal, or of modulation of transmission.
Now, the fact had already been found by Katz in 1937 and Hodgkin in 1938 that axons can show something besides all-or-none spikes. This is the local graded activity, above the electrotonic spread from a nearby impulse in the axon. Although this local graded activity is without or before the explosive event of a nerve impulse, it is highly nonlinear with increments of subthreshold stimulating current. It was recognized immediately as a major claim, basic to our understanding or working model. But it was not instantly accepted generally. It was disbelieved and resisted by some, for years (including C.A.G. Wiersma who worked, as Hodgkin did, with crustacean peripheral motor axons). Only slowly did it work its way into the normal picture of axonal activity, especially near the terminals. It is now widely believed that spikes die out, at least often, if not generally, and much if not most transmitter release is under the control of this graded, local activity. This is often forgotten, however, even today.
The first direct demonstration of graded junctional potentials was accomplished intracellularly, in the neuromuscular junction, by Paul Fatt and Bernard Katz in 1951, followed in a few years by Eyzaguirre and Kuffler who found them in 1955 in a neuron, the crayfish stretch receptor. In the same year Coombs, Eccles and Fatt found them in the mammalian CNS. These were exciting discoveries and were instantly accepted. People realized a new set of questions was now opened up.
Even earlier, in 1947, D.A. Parry claimed to see spikeless neurons in an insect ocellus, that is neurons which transmitted normal signals of light reception to the central ganglia but without ever firing a spike. His work was forgotten and a long controversy ensued over the absence of spikes in the Limulus retinula cells. H. Autrum and D. Burkhardt (1960) settled the question for insects by recording spikes intracellularly from visual sense cells in flies, followed by Ken-Ichi Naka (1961) in bees. The most unequivocal spikeless neuron was given to us by S.H. Ripley, B.M.H. Bush and my former student Allan Roberts in 1968 from a large stretch receptor in the base of the legs of crabs. Although the sensory region of this large neuron is some 10 mm from the nearest central ganglion (in a small crab) and the graded receptor potential spreads along its large diameter axon with a great decrement, it is still adequate to mediate normal reflex movements. This anomalous neuron turned out, over subsequent years, to be a model of many. Frank Werblin and John Dowling gave us spikeless bipolar cells in the vertebrate retina in 1969. We now know there are large numbers in the central ganglia of insects and in the retina of vertebrates. Spikeless neurons, it was recognized, opened up another can of worms; their discovery was cheered by some and grudgingly ignored by others, according to their predilections.
Years ago I placed my bet that they will be found in the brains of mammals, once we learn how to recognize them or to penetrate many cells with intracellular electrodes. That bet has not paid off yet, but it still holds. Even more common, I believe, will be found neurons that spike, but in between spikes exert a continuous and nonlinear graded influence on the next neuron or on neighbors. This can be either by continuous, graded release of transmitter which was suspected long ago and demonstrated by Katherine Graubard, John Raper and Dan Hartline in 1977 (Graubard et al. 1980), or it can be by electrical field effects.
Field effects are well established as a class of mechanisms but not well known in respect to requirements such as synchrony of sets of neurons or geometric relations with receiving neurons or effective portions of the frequency spectrum. Already early in the half century we are revisiting, electrical field effects were considered by some authors to be likely. In 1936 Alfred Fessard and in 1941 Ralph Gerard, one of the founders of the Society for Neuroscience, as well as Frederick Bremer in 1944 speculated that neurons can influence each other without impulses by graded signals and I agreed with them in 1945. Hansjochem Autrum (1952, 1953) in Göttingen, about the same time also invoked this idea in his theory of fast insect eyes. Psychologists thought they disproved the possibility of field effects by placing metal foil or grids over the rat cortex or rows of pins into the cortex, to shunt at least the gross fields. They reported no failure of rats to learn and to run mazes.
In 1956 Carlo Terzuolo and I imposed a very weak, subthreshold field onto a preparation of the stretch receptor neuron of a crayfish and saw changes in its tonic firing rate when the field was about as strong as some of the larger evoked potentials and spontaneous brain waves. This surprisingly high sensitivity was one of the main motivating factors that led me in 1960 to look into electroreception in electric fish, where still higher sensitivity, by orders of magnitude, must be found - small fractions of a microvolt per centimeter in the water around the fish. The fascinating story of the discovery of electroreception, which goes back before 1941, would take us too far afield in this essay (Moller & Fritzsch 1993). When my colleagues and I found the afferent axons for normal, physiological electroreception in the lateral line nerve of gymnotiform fish (Bullock et al. 1961), it soon became clear that specialized electroreceptor neurons can vary over an extremely wide range in their sensitivity to electric fields. Some can go down to a behaviorally significant gradient of 0.005µV per centimeter (Kalmijn 1988); others normally function with tens or hundreds of microvolts per centimeter (Bullock & Heiligenberg 1986). They are typically tuned to preferred frequencies from <10 Hz up to hundreds and even several thousands of cycles per second. The proposition that brain waves, evoked potentials and other kinds of normal electric fields might be, not only effects but also causes, not just the noise of the engine but in some degree and some places modulatory was put forward long ago, in the '30s and '40s (references in the preceding paragraph). It is still far from proven or delimited in terms of strength, frequency and places in the brain - but I am betting it will find its way into the popular working model of how the brain works any day now and it will become fashionable to study when, where and how!
I want to mention one other early claim that is closely related. Robert Gesell in 1940, in connection with his model of the neural control of respiration, proposed that the neuron has DC polarization along its axo-dendritic axis and that this exerts some influence in controlling firing rate; a kind of self-generated field effect. This idea is still not solidly confirmed but seems to me quite likely, in some degree or formulation - and I look for it also to work its way into our orthodox view of how the brain works. As you see, I am choosing examples that happen to underline how much of history consists of unfinished stories on hold until the next push achieves an apparent closure, or adds another complication. These stories also underline how subjective are the attitudes, priorities and bets of the day, in this enterprise we like to think of as objective and reasoned!
|The Controversy over Soup vs Sparks|
The controversy over chemical vs electrical transmission at the synapse raged for years. For a long time it was just over peripheral junctions such as the neuromuscular junction and only slowly spread to include central synapses as well. The reason it raged for years was not that the evidence for electrical transmission was strong but that the arguments for chemical transmission were indirect and came from drugged preparations, often in a subnormal state. Evidence for electrical transmission was positive and required preparations in very good condition, although not really normal junctions - the ephapse, including crosstalk in nerves and spinal cord. One should mention Herbert Jasper and A.M. Monnier in 1938, Bernard Katz and Otto Schmitt in 1940, Arturo Rosenblueth in 1941 and Angelique Arvanitaki in 1942, as well as the finding of reflection from cut ends of nerves and tracts, that is the synapse-like crosstalk from some stimulated fibers in a nerve to others (Granit & Skoglund 1945). These are all special cases, not ordinary, normal synapses.
The best evidence for excitatory chemical transmission was negative evidence, that is, the direct demonstration that transmission could not be electrical. This was first announced in a dramatic meeting of the Royal Society on February 21, 1952 (Brock et al. 1952) by John Eccles, previously a champion of electrical transmission. He and his colleagues had penerated the motoneuron and found the hyperpolarization caused by inhibitory input; this could hardly be understood to result from arriving impulses of the same electrical form as excitatory impulses that cause depolarizing synaptic potentials. A specially favorable example was the squid giant synapse. I had found already in 1946 (Bullock 1946) that this was uniquely suitable for studying the unit synapse. With Susumu Hagiwara we penetrated both pre- and postsynaptic sides of this giant junction and found the clear absence of any sign of the prespike during the synaptic delay in intracellular recordings in the postunit (Bullock and Hagiwara 1957). The much earlier and classical evidence, namely findings of so-called transmitters in the soup around junctions, recognized to be "hypodynamic," and only when they were protected from cholinesterase by drugs that inhibited the esterase, was suggestive and quite sufficient to start a large following, but was not convincing that normal, healthy junctions worked this way.
The controversy simmered down only as two independent lines of evidence improved, in a succession of small saltations. The evidence for two kinds of electron microscopically defined synapses gradually became convincing - gap junctions and cleft junctions with subsynaptic apparatus. The pharmacological evidence for normal chemical transmitters became convincing and the physiological evidence accumulated that gap junctions work electrically. Although chemical transmission had the upper hand, Furshpan and Potter's beautiful demonstration in crayfish giant-to-motor synapses in 1957 was instantly hailed as proving that electrical junctions exist.
Now, I will guess that most workers today look back with some amusement at this long debate and regard the matter as finally settled. We understand today that there are chemical and electrical synapses and we look for advances in working out the molecular machinery of each. Well, I am ready to wager that the last laugh is some time off and our students will remember this stage, with amusement, as a short plateau or lull in the controversy. I am not ready to say where I think it will go - in which direction it will break out and change the issue; I have some bets but this is not where they belong. Among the names we should remember, even though I cannot detail here just which advance each was responsible for, or list all the worthies, are Otto Loewi, Sir Henry Dale and W. Feldberg, Paul Fatt and Bernard Katz, Sir John Eccles and Steve Kuffler, Carlos Eyzaguirre, and David Nachmansohn - each for distinctive contributions.
|Integrative principles in small sets of neurons|
Moving up a level of complexity, it was just over a half century ago that we began to see the integrative properties of synaptic transmission analyzed parametrically in a space whose variables are still increasing today. This domain overlaps with the physiological aspects of what Pasko Rakic named local circuits, the intimate functions between consenting neurons.
As the first example I turn to another way in which the neuron doctrine began to be refined from the McCulloch-Pitts all-or-none element. It represents one of the long list of "personality traits" of neurons, the elementary integrative property called facilitation, a junctional trait that was measured long before synaptic potentials were known. At a gross, reflex level we can recognize it in Sherrington's experiments on reflexes of mammalian limbs. But Carl Pantin at Cambridge in 1935 - actually at the Plymouth Marine Laboratory, quantitatively demonstrated it at the unit level in sea anemone nerve nets and in crustacean neuromuscular junctions, followed immediately by Kees Wiersma and Anthonie Van Harreveld (1938) in Holland, shortly before they came to Pasadena. The concept of facilitation was required by the evidence of a graded and local, highly nonlinear, subthreshold residue after an arriving impulse that failed to fire the postsynaptic unit but left it closer to threshold. This invisible, inferred change, whether pre- or postsynaptic, was found to be tuned to a preferred interstimulus interval. The preferred interval differs widely among the junctions of different muscles within sea anemones and among different neuromuscular junctions in crabs and crayfish and it was shown to have a decay with, sometimes, two or more time constants. Since this is an integrative form of neuronal action first found in neuromuscular junctions, quite unlike the familiar vertebrate type that transmits 1:1, Pantin and others spoke of crabs "thinking with their claws." As a crucial variable in determining output as function of input, it is remarkable that only slowly did it become a part of the everyday model of neuronal action people use although it was soon shown to be widespread.
Its opposite is a common interval-dependent depression of the postsynaptic potential, sometimes called antifacilitation or difficilation; it was also found to be widespread. My colleagues and I in the late '40s and '50s had a hand in all this, especially Donald Maynard (1955, 1967), Susumu Hagiwara (Hagiwara & Bullock 1957), Carlo Terzuolo (Bullock & Terzuolo 1957), and Akira Watanabe (Watanabe & Bullock 1960) in intracellular recordings from the squid synapse and lobster cardiac ganglion cells. These were very heady days, with exciting new parameters of neuronal integration coming along one after the other. Presynaptic inhibition and excitation, chemical and electrical transmission, sometimes in the same cell, different chemical transmitters in the same cell (contrary to Dale's Principle, as it was then understood), slow postsynaptic potentials - lasting for seconds and even minutes (Laporte & Lorente de Nó 1950; Eccles and Libet 1961; Tosaka et al. 1968; Nishi and Koketsu 1968) not dependent on conductance increase and showing long-term modulation of one transmitter action by another transmitter (Libet & Tosaka 1970) are just some of them. Diversity in afterpotentials, in sensitivity to depolarization and in apparent neuronal codes are others. New integrative variables at the neuronal level are still being found in 1994. Besides these general neuronal parameters, back there in the 40s and 50s we were stimulated to widen our working model of the brain by related discoveries such as the first tonic receptors and sensory thresholds that showed hypersensitivity to light, sound, infrared, electric fields and olfactory stimuli.
Another of the principles that rose from the sea to astound us was lateral inhibition. This sprang full blown from the head of Limulus one day in 1953 thanks to discoveries of Keffer Hartline, Henry Wagner, and Tsuneo Tomita in a major step from their previous concentration on the primary receptor toward the central, multiunit interactions. It is difficult to exaggerate the excitement that followed the discovery of this process. It was immediately clear what an expansion in degrees of freedom throughout the nervous system it would represent, if lateral inhibition is not confined to the optic ganglion of a strange, living fossil for which there was still no evidence then that it can see. Fortunately evidence that lateral inhibition is indeed found widespread in the animal kingdom and in the brain came rolling in - not rapidly but steadily. The general feeling was, in a current phrase, "We needed that!" It reminds me of the similar sense of a sudden expansion of the degrees of freedom and an approach to the true complexity, compared with the simple and, so we thought, adequate mechanisms of a McCulloch-Pitts nervous system, when David Lloyd first gave us direct neural inhibition, in 1941, previously unknown except as refractoriness after excitation. And the same kind of saltation, multiplying all previous notions of the basic devices available to organized sets of neurons came with the demonstration of presynaptic inhibition, already touched upon. Eccles (1961) presented arguments for interpreting earlier observations of "remote inhibition" in mammalian spinal cord as direct transmitter action of inhibitory nerve endings near the synaptic terminals of excitatory axons and Dudel and Kuffler (1961) showed this in favorable crustacean preparations. In this as in other instances of major discoveries, it is humbling to read the literature just preceding, with extensive and sophisticated knowledge unaware of the vast change just around the corner. One wonders what surprise is in store for us tomorrow.
Two other principles are the functional refractory period and nuclear delay. These early insights have had less influence on everyday thinking than they should. They were due largely to Arturo Rosenblueth (1941, 1949a, b) or at least he articulated well what was a common understanding at the time. We need to be reminded of these concepts because many authors today tend to think in terms of the fastest recovery curve under maximal test stimuli and of the minimum synaptic delay of ca. 1 ms in the special case where a postspike can be attributed to a given prespike, usually requiring artificial synchrony of a group of presynaptic axons. The functional refractory period is the time it takes to recover excitability to a level where the normally available stimulus will excite. It is therefore not a fixed number but depends both upon the recovering membrane and the exciting event. And it is not a curve as the relative refractory period is, but a number that applies to that situation. Nuclear delay, as opposed to the minimum synaptic delay, is the time it takes for activity to emerge from a functional system, which is normally less than a brain nucleus but more than a monosynaptic junction receiving synchronized input. Nuclear delay is a useful concept when thinking about normal function even though it has no sharply defined input time or output time, both being barrages or bursts of spikes. The delay can be tens of milliseconds, even hundreds and thousands, when only one synapse is in series, obviating the common interpretation that, if 20 ms transpires, there must be something around 15-20 synapses.
My next example is one of the forms of extra-synaptic transmission between neurons, a still little-known list of mechanisms and pathways that I predict will become a major category in our everyday model of how brains work, instead of the fringe phenomenon that it is respectable to ignore - as was true in turn for a list of curiosities that slowly became standard options for neurons in real animals with hair. I am thinking now of slow electrotonic interaction. This is perhaps still not widely familiar. It was shown already 30 years ago in lobster cardiac ganglion cells that a subthreshold, intracellularly imposed depolarization in a follower cell can accelerate the upstream pacemaker cell, if the depolarization lasts long enough - something like 50 ms. If it is repeated at suitable intervals, it soon entrains the pacemaker. A brief but stronger depolarization that reaches threshold and fires the penetrated follower cell does not accelerate or influence the pacemaker at all. The electrotonic connection is not an ordinary electrical synapse but passes only slow or long-lasting current, as though it were a very thin, long process. Akira Watanabe in Tokyo first noticed electrotonic spread between follower cells by penetrating both of them. He came to California and together in 1960 we found such spread can be functionally significant by exerting an effect back upon the pacemaker. Besides being the first subthreshold electrotonic interaction, this was the first and only kind of feedback from followers upon pacemakers in the lobster cardiac ganglion. We still don't know how widespread such connections between neurons may be; they presently require exceptionally favorable conditions to test for them.
|Nerve nets and unpolarized synapses|
It is difficult today to talk about real nerve nets in animals, since the advent of the so-called neural nets, which would be better called neuroid nets, made of computer simulated neurons in "layers" with restricted inputs and outputs, not at all simulating the nerve nets in real nervous systems. Nerve nets were inferred in jellyfish >100 years ago, from the ability of excitation in the umbrella of the medusa to find a path around partial cuts, spreading without interruption even through spirally cut strips of tissue or doughnut preparations with alternating deep incisions that lengthen the path (Bullock & Horridge 1965). There must be a diffuse conduction system that does not depend on nerves or tracts but has widely distributed inputs, outputs and conducting elements. If it is not a continuous sheet but made of cells and processes, it must be net-like. This is the simplest level of complexity among nervous systems; judging by its survival value, it was a very successful one.
First one and then another example was eventually visualized anatomically and then found electrophysiologically to employ all-or-none impulses and synapses, perhaps some with electrical and some with chemical transmission. The synapses are either unpolarized, two-way transmitting junctions or there are side by side one-way junctions polarized in each direction. We know examples in many species in some detail, anatomically and physiologically, and which aspects of the behavioral repertoire of the various species they mediate. Commonly there are two separate nerve nets communicating only at restricted places. In addition, nonnervous conducting systems, operating through epithelia or muscle cells, may coexist with nerve nets. We should remember names like Romanes, Eimer, Parker, Bozler, Horridge, Spencer and Mackie. Yet we still don't know enough to explain all their behavior or the slow changes in mood or the differences between species in patterns of spread of response.
Some quite realistic models of nerve nets were programmed in computers in the early 60s - for example by my students, Robert Josephson (Josephson et al. 1961) and Les Fehmi (Fehmi & Bullock 1967). These permitted us to ask the relative values of different connectivities and temporal patterns of impulses and to explain some of the species diversity in propagation of excitation. Since then there has been a burst of new information about the coelenterates and their quite diversified behavior and nerve net physiology. The potential for new insight from modeling has only begun to be realized. I can predict confidently that we will see significant changes in our understanding of how they work. This is all quite unrelated to the "neural nets" of cognitive science, based on computer-generated adjustment in the connectivity and not on asking questions about how natural nets work.
I mention this topic partly to highlight the sad fact that we still know little about, and few people work explicitly to reveal just what differences there are, especially between major taxa, that might be associated with the greater and greater levels of complexity we see at least in anatomy and behavior, from jellyfish to worms, from snails to squid, from fish to mammals (Bullock 1993a)!
|Origins of patterned discharge|
The connection to my next subject may not seem obvious but it follows from the detailed studies of Mayer around 1906 on the origin of the rhythmic beating of jellyfish, which he found is intrinsic in each of the eight marginal bodies where a concentration of the nerve net amounts to a simple ganglion. The subject I want to bring up is the origin of patterned discharge, especially rhythmic pattern. I will not go back farther than about 1940, though a good deal of history took place long before that. Obviously Sherrington, his contemporaries as well as predecessors dealt with the problem before 1900.
When I was a graduate student in 1936-40 the usual teaching was that walking and similar repetitive motor actions were due to chains of proprioceptive reflexes. A theoretical alternative was recognized because it had been shown by Carlson in 1904 that some cardiac rhythms, such as that in Limulus, were neurogenic, arising intrinsically in a cardiac ganglion. Not only that; the bandwagon or hot insight of the 30s, reflected in titles of major lectures and book chapters, had been that the CNS is not just a telephone switchboard, waiting for input to react to, but was spontaneously active, even in the absence of episodic or timing stimuli - any but tonic inputs - and such inputs were not yet known. The first known tonic receptors, that is some of the vestibular and temperature receptors, came in the late 30s and early 40s.
The central origin of respiratory rhythms had been suggested for fish, by E.D. Adrian and F.J.J. Buytendijk, in 1931 but for mammals was still a few years in the future. Erich von Holst in 1935 pointed out evidence of a heavily central role in the locomotor pattern of certain species of centipedes as opposed to others that show a stronger influence of sensory input. He went on to teleost swimming in 1936 and argued for a basically central pattern, subject to modulation (Holst 1936a, b; 1939a, b). Von Holst's papers, however, were not widely read during his lifetime. The voice of Paul Weiss crying in the wilderness, beginning about 1935 (Weiss 1941), proposing that the CNS is issuing rhythmic motor commands, without any necessary sensory feedback, did not cause the sympathetic vibrations he thought it deserved. His main evidence came from the behavior of amphibian larvae. James Gray, later Sir James, produced evidence for both central and reflex origins of locomotor pattern in leeches, earthworms, sharks and frogs, beginning about 1938. In an influential symposium in 1950 (Gray 1950), attended by luminaries like Von Holst, Weiss, Lashley, Konorski, Tinbergen, Konrad Lorenz, J.Z. Young, W.H. Thorpe and others, Gray had, in effect, the last word, saying the problem is extremely complex but "the main conclusion must be that the existence of centrally controlled patterns of locomotion should be regarded as non-proven."
In 1961 I reviewed the issue and wrote "I would underline that it is difficult to find fault with the elegant work of Gray and Lissmann ... except in matters of emphasis and of selection of examples. But it is just here I believe where there has grown up a climate of opinion, not fully explained by a literal reading of these authors, which is so one-sided, emphasizing peripheral as opposed to central control, as to call for restoration of a balance." I went on to outline the main types of pattern formulation, with their combinations, boiled down to five, and claimed that all five can be found. I emphasized that in almost all cases, including non-rhythmic actions, the details of temporal and spatial distribution of impulses, as distinct from the triggering, are basically central. "Nervous systems do not give outputs predictable by their inputs or externally controlled 'instructions.' ... Central patterning is the necessary and often sufficient condition for determining the main characteristic features of almost all actions, whether stimulus triggered or spontaneous."
The prevailing emphasis today is quite different. It is eclectic and recognizes a range of combinations of central and peripheral influences. I think the balance has been restored or at least approximated, although surprises are still, no doubt, in store. Certainly the extensive work on central pattern generators reflects the fact that pattern determined in the brain or cord became once more recognized sometime after my review, 30 years ago. My former student Don Wilson started a parade of workers and new evidence for central pacemakers with his demonstration in 1961 that the flight motor rhythm continues in locust central ganglia after deleting all the timing signals or phasic components of the proprioceptor and exteroceptor afferent input. Some input is necessary just to keep up the frequency of the wing beat, but it can be tonic and free of any rhythm. Just as certainly, there has been recognized the crucial modulating role of sensory input on intensity or frequency of motor actions and in steering locomotion and shaping movements.
I would guess that at least the more comparative neurobiologists today have a real appreciation of the diversity in relative weight of the endogenic versus the exogenic components, as in the old case of von Holst's centipedes and the case of walking vs swimming in leeches and other groups. From my memory of the three protagonists, Erich von Holst, Paul Weiss and Sir James Gray, I am sure they would all be pleased with the present position because they were all broad gauge biologists and worshipped experimental facts. The issues have advanced now to such questions as how many types and intergrades of neural circuit and intracellular oscillators are there; how much of the fluctuation of period is true noise or slop and how much has physiological value or meaning; how is high regularity achieved in populations of irregular pacemakers - and the molecular mechanisms in each case?
|Identifiable cells, equivalent cell sets and circuits|
These interwoven topics together represent one of the most striking of the developments I have chosen and bring us to a period about 30 years ago. The concept of identifiable cells is familiar to invertebrate neurobiologists and those who work on Mauthner's or Müller's neurons in fish. In the late 60s, over a period of a few years, the concept was proposed, struggled for recognition and gradually became accepted - that many cells in major groups of higher invertebrates are individually consistent so that they can be identified and named or numbered in every specimen. This is now familiar in leeches, crustaceans, insects and some gastropod molluscs, as well as in lower groups such as nematodes. Furthermore, the experts appear to agree that there are doubtless many more such identifiable cells than we presently know of.
Ever since 1910 and the work on the giant nematode, Ascaris, by Richard Goldschmidt (he of the classical work on Lymantria and Drosophila and the first questioning of the classical gene-as-particle theory and the first book on physiological genetics), we had known that this strange phylum of round worms has not only constant cell number, called eutely, but also identifiable nerve cells, all 162 of them (Goldschmidt 1909, 1910). Invertebrate neurophysiologists also knew the work of Wiersma and Van Harreveld in the 30s on crayfish and crabs, showing that every specimen has less than a dozen motor axons in each leg, each axon identifiable by its consistent distribution and physiological influence on the muscle fibers it innervates (summarized in Bullock & Horridge 1965). We knew of Mauthner's cell in many teleosts, a single pair of cells that mediate a startle response. But these were all special cases. We knew that many invertebrate ganglia have only a few hundred neurons and some, like the cardiac ganglion of lobsters were shown by Alexandrowicz in 1932 to have exactly nine, setting off many years of happy hunting for their dynamic properties in my lab. But nobody had discerned even a moderately common incidence of unique neurons.
The poignant part for me was that my two volume treatise with Adrian Horridge on the nervous systems of invertebrates came out in 1965 without any suggestion that identifiable cells are a common phenomenon! Already in 1965 and gradually in the next few years the evidence began to pour in. Without the space for an accurate attribution of credit, let me just mention that a number of workers found, perhaps first in the gastropod, Aplysia and the garden snail, Helix and later in a list of other species of opisthobranchs and pulmonates, and almost in parallel in locusts and other insects, with crayfish, lobsters and leeches not far behind, that some central motor neurons and interneurons and even a few sensory neurons are consistent. They began to be named and numbered and then more and more neurons were identified until a few authors began to extrapolate that probably a large fraction or possibly the majority of the central neurons would turn out to be identifiable in these species. Relevant names I would mention, prior to a great burst of activity in the 70s, are Stefanelli (1951), Bodian (1952), Arvanitaki and Chalazonitis (1961), Tauc (1959, 1962), Rovainen (1967), Kandel and colleagues (Frazier et al. 1967), Cohen and Jacklet (1967), and Kerkut et al. (1970).
Most peripheral sensory neurons, generally very numerous, are thought to be, at best, addressable rather than individually identifiable. Identifiable cells are rare, very few or unknown in most marine gastropods, the prosobranchs, all other molluscs, arthropod groups such as arachnids, oligochaetes such as earthworms, and many other groups, like the flatworms, echinoderms and vertebrates, even though the count of identifiable cells in primitive fishes has increased from one pair to a dozen or two.
This trait seems to have arisen independently a number of times. The question that appeals to me now is how many neurons are not really unique but exist in small sets of fully equivalent cells - that is, in stes of 2, 3, 5, 10 or 100 cells? How many such sets are there? The concept of addressable cells is not adequate to handle the questions about gradations of uniqueness vs overlap because we have to add the physiological traits of modality, tonic vs phasic firing, facilitation, afterdischarge, rebound, tendency to burst and the long list of integrative variables as well as input and output connectivity before we can characterize the cells (Bullock 1993b, Chapt. 7). My conclusion, no doubt to be proved wrong or inadequate in some important way, is that taking all traits into consideration, including chemical, immunological, plastic capacities and vulnerabilities to all kinds of influences, we will find much less redundancy and more specification, than is usually assumed, even in mammals. Adding all these traits to the connectivities, both receptive and projective, I estimate we have a very large number of distinct kinds of cells, that is, equivalence sets, indistinguishable within the set on any criterion but distinguishable from other sets on some criterion, at least on a significant degree of incomplete overlap.
Since the main criterion for identifiability is connectivity, the idea of identifiable neurons brought with it the powerful and exhilarating idea that circuitry can be worked out. Much to our surprise, it turned out to be possible and a great deal of circuitry was worked out - and still is being worked out, at levels down to the individual identifiable cells and their synapses! Now I believe we have come to a stage when we need a new image and terminology to go beyond circuitry, in order to include in our working picture of the brain concepts not embraced by the terms connectivity and circuitry, such as the nearly unique personality of each neuron and each locus within it - the sum of the static and dynamic proclivities, chemical signatures, arbor geometries, local fields and cooperative interactions within ensembles.
|Compound field potentials|
The last topic in my list embraces event-related potentials (ERPs), evoked potentials (EPs) and ongoing background activity such as the electroencephalogram (EEG) in both vertebrates and invertebrates. Those readers who believe the only avenue to understanding higher levels of brain function is the study of single units and spikes may be turned off by now. I take the risk, on the ground that it is never too late to save a soul. This topic also underlines a common feature of each of the topics, namely that the history is still going on, the mysteries are still there, the opportunities for new ideas to be tested are not reduced by the enormous accumulated knowledge, but actually increased.
Compound field potentials, being the vector sums of extracellular currents from many generator cells, neural, glial and perhaps others, emphasize the slower components of the spectrum of activity potentials. EPs were discovered in the last century in rabbits and EEGs were found in unstimulated humans >60 years ago, but we have only slowly learned a few things about their nature and significance. One class of things is how far we are from understanding the relative contributions of the known classes of generators (somata, synapses, nonclassical approximations of membranes, axons, nodes, terminals, dendrite segments, glia of several kinds, vascular walls, ependyma and pia), the known classes of activity or relaxation (spikes, local, synaptic, plateau, pacemaker and dendritic potentials and currents attributed to or funneled by astrocytes or their equivalents (Galambos 1989), and the forms and means of cooperative interaction and their vector summation in a nonisotropic volume conductor. Another is how distinct this window is from the window of single cell and spike-only activity sampling. The historical conclusion I am most sure of is that this domain will look very different to us in the future, compared with the conventional wisdom today.
The EP, ERP and EEG signs of underlying cellular activity are generally quite unpredictable from a knowledge of unit activity. They can be virtually absent or vanishingly small in the presence of high unit activity; EEG waves can be present after drugs that block synapses. EPs can be the most sensitive objective sign that the brain is receiving sensory input, long before behavioral data or single unit data is available. This is most notable in the cases of unfamiliar receptors, such as electroreceptors or ultrasonic receptors or infrared receptors, but is also true for familiar modalities. They can be the most sensitive or the only objective sign that the brain is attending to the right ear and not the left or can distinguish two different stimuli, such as Monets from Rembrandts.
Parts of the CNS differ markedly. The cerebellar cortex has hardly any ongoing slow waves although exceptionally high in unit activity, but it has large EPs. The tectum has large EEG waves even in laminae that are very quiet in terms of spikes. Some acoustic signals cause large EPs in the cerebral cortex but traverse the midbrain without a detectable EP. Slow, compound waves are a neural option and we basically do not understand where they come from or what conditions make them happen. Even the relative roles of neuronal and glial and other membranes are little known and are likely to differ among species, states and parts of the brain. We do not yet know the biological meaning of compound electric fields, where and when they may be not only consequences but also causes of brain function. We certainly cannot conclude that they are insignificant byproducts, good only for diagnosis. They are prime examples of potentials of action in need of study to determine both their functional meaning and their information content as signs of neural processing.
If mysteries intrigue you, we are not much ahead of Lord Adrian, who in 1931 and 1937 saw slow waves in a water beetle indistinguishable from his own alpha waves. No one doubts he saw them but no one has seen them since. Invertebrate ganglia and brains were shown by Prosser in 1936 to have busy, spiking activity and virtually no slow waves. But Jahn and Crescitelli in 1938 could reliably get slow sinusoidal waves in insect optic lobes under special conditions. About the same time, somebody cited by Gerard got slow waves from the spinal cord of frogs by applying the right drug. In 1945 I asked the question what can it mean that the EEG looks the same, except for amplitude, from the tectum and forebrain of all vertebrates, whether large brain or small, whether they have a laminated cortex or not? And what does it mean that crustaceans, insects, worms, and snails all have similar, spikey activity almost without slow waves except for the few cases under special conditions? Now I can add that Octopus and Sepia are more like vertebrates than like other invertebrates (Bullock & Baar 1988).
The history of this subject is not like the others I have touched upon but one with fewer simplifying discoveries and continued mystery, expanding in dimensions. We now realize that EPs, ERPs, and EEG waves are not, as has often been claimed or implied, really global or congruent over large areas of the brain, as a general rule, when one records from the brain surface or interior rather than from the scalp, with some special exceptions. There is fine structure in the EEG of all animals examined, in space as well as time, in the fractional millimeter and fractional second ranges - but we know only a few general principles about this structure.
We now realize it is exceptional that the activity is rhythmic or oscillatory. Rhythms are clear only under special conditions such as those conducive to alpha waves in some people and a few other species - the subject awake but with eyes closed and not busily thinking. A number of such special states and places are known, where peaks in the power spectrum are not merely transient but most of the activity most of the time in most of the brain is very wide band, without consistent narrow peaks.
We think it is mostly deterministic and cooperative, by mechanisms not acting, or only weakly effective in most of the lower levels, below the midbrain and outside the vertebrates. Synchrony of slow, subthreshold membrane fluctuations may be one of these mechanisms and may be largely nonsynaptic or extrasynaptic. Certainly synchrony is an option that varies widely, presumably with some functional significance, not yet understood (more likely different kinds of significance in different places and states). It is not well correlated with good perception, good performance, or awareness. Perhaps there are various kinds of synchrony. We know little about them or their mechanisms. We are just beginning to measure synchrony after using eyeball analysis since the early 30s, for conclusions like "This record is relatively desynchronized and that one more synchronized." Confusing amplitude and synchrony, we are probably quite wrong most of the time (Bullock 1993b, pp. 288, 375).
We have learned that innovation in cooperativity occurs with many cognitive events and states, that is, after a whole range of mental experiences such as "There's another one," "What's that?" or "That's ridiculous". We are still in an early stage of distinguishing between "re-ordering" of phases vs new energy (Baar 1992) and between coherence increases due to phase shifts vs those due to power shifts. Non-linear analyses of the type of bispectrum and bicoherence have just begun.
Expectations learned in the last few seconds or over a lifetime can trigger internal events such as the complex sequence of slow, fast and oscillatory waves that follows the omission of an expected stimulus, or its tardiness by a few milliseconds. The omitted stimulus potential can have a fixed latency after the due-time, whether that due-time was longer or shorter after the last actual stimulus - as though the system has a precise temporal expectation learned in a few preceding cycles of a periodic stimulus or in a few minutes with self-generated reafference, like respiratory or locomotor movements (Bell 1981). This happens not only in human subjects, given a task to assure their attention but also, to our surprise, without any control of attention in stingrays and turtles and in lower brainstem levels, even in the retina.
The modern history of this subject compels me to say something else that is not so pleasant to speak of. The larger question of how organized assemblages work, to accomplish what we observe they do - things like recognition, selection among alternative responses and decision making - has been held back by a simple and understandable human failing. That is a mutual disparagement of those, on the one hand, who choose the strategy of searching for single units, usually extracellularly, and recording their spike activity during interesting states or events, throwing away the precious slow potentials their electrodes see, and those workers, on the other hand, who choose the strategy of searching for compound field potentials of a population, unable to account for it in terms of cellular activity. Each of these approaches is a window and a quite inadequate one. We need both and the combination of the two and still others to untangle this most complex of known systems.
To discover what is going on in organized neural systems, at least in the vertebrates and cephalopods, we need many electrodes, more closely spaced, recording with wideband amplifiers, to see both the spikes and the slow waves at once, under a variety of conditions, including external and internal events and states. We also need new methods of detecting, describing and assessing, not only the linear dependencies but the nonlinear higher moments of cooperative action! Even then, the mysteries are not going to unravel themselves, but until then, we are likely, not merely to oversimplify but to miss essential principles of neural integration.
If you think back over my human drama in seven acts, you might detect some leitmotifs. I will only point out a few by way of underlining their contribution or their lack of contribution. One is the attempt to choose systems for study that have some relevance to behavior important for the species. This goes back long before our 50-year period but it came to be called neuroethology in the late middle of the period. Another leitmotif is the use of a wide variety of species, lower and higher. It is commonplace to point out the value of model species, favorable for one or another reason in the study of general principles. That's a success story in terms of contributions and continued promise. Curiously, another facet of the same gem has so far yielded little and I think it is due to another human failing - to be overly influenced by what is popular. The facet I refer to has simply been neglected; that is the approach of comparing higher taxa - widely separate groups, such as classes, like reptiles and mammals or fish and reptiles. The comparison of species within the same order, like bats or insects or opisthobranch gastropods is much more often done and reveals interesting things about adaptations but we still need comparison of major taxa to uncover the differences that might be relevant to one group being obviously more complex in brain anatomy and behavioral repertoire than another, whatever the adaptations for particular habits of life. We need lists of physiological variables. The anatomists and chemists have no trouble listing the things they must look for, given a new taxon of interest. We physiologists have great difficulty - as have the behavior people who want to compare cognitive capacities.
The other leitmotifs I will leave to you to discern, if you replay in your mind these half a dozen sample topics of mesoscopic integration. For me, our history, as exciting as it has been, has just begun. A rich vein of new principles of brain dynamics, organization and function is there, waiting for new prospectors to find them.
Adrian ED (1931): Potential changes in the isolated nervous system of Dytiscus marginalis. J. Physiol. 72:132-151.
Adrian ED (1937): Synchronized reactions in the optic ganglion of Dytiscus. J. Physiol. 91:66-89.
Adrian ED, Buytendijk FJJ (1931): Potential changes in the isolated brain stem of the goldfish. J. Physiol. 71:121-135.
Alexandrowicz JS (1932): The innervation of the heart of the Crustacea. I. Decapoda. Q. J. Microsc. Sci. 75:182-249.
Arvanitaki A (1942): Effects evoked in an axon by the activity of a contiguous one. J. Neurophysiol. 5:89-108.
Arvanitaki A, Chalazonitis N (1961): Excitatory and inhibitory processes initiated by light and infra red radiations in single identifiable nerve cells (giant ganglion cells of Aplysia). In: Nervous Inhibition. E. Florey (ed.). Pergamon Press, Oxford.
Autrum H (1952): Über zeitliches Auflösungsvermögen und Primärvorgänge im Insektenauge. Naturwissenschaften 39:290-297.
Autrum H (1953): Elektrobiologie des Auges. Klin. Wschr. 1953:241-245
Autrum H, Burkhardt D (1960) Die spektrale Empfindlichkeit einzelner Sehzellen. Naturwiss. 47:527.
Baar E (1992): Brain natural frequencies are causal factors for resonances and induced rhythms. In: Induced Rhythms in the Brain. E. Baar and T.H. Bullock (eds.). Birkhäuser Boston Inc. Boston, MA, pp. 425-467.
Bell CC (1981): An efference copy which is modified by reafferent input. Science 214:450-453.
Bodian D (1952): Introductory survey of neurons. Cold Spring Harbor Symp. Quant. Biol. 17:1-13.
Bremer F (1944): L'activité "spontanée" des centres nerveux. Bull. Acad. Roy. Med. Belg. (Ser. 6) 9:148-173.
Brock LG, Coombs JS, Eccles JC (1952): The nature of the monosynaptic excitatory and inhibitory processes in the spinal cord. Proc. Roy. Soc., B, 140:169-176.
Bullock TH (1945): Problems in the comparative study of brain waves. Yale J. Biol. Med. 17:657-679.
Bullock TH (1946): A preparation for the physiological study of the unit synapse. Nature 158:555-556.
Bullock TH (1961): The origins of patterned nervous discharge. Behaviour 17:48-59.
Bullock TH (1993a): How are more complex brains different? One view and an agenda for comparative neurobiology. Brain Behav. Evol. 41:88-96.
Bullock TH (1993b): How Do Brains Work? Papers of a Neurophysiologist. Birkhäuser, Boston.
Bullock TH, Baar E (1988): Comparison of ongoing compound field potentials in the brains of invertebrates and vertebrates. Brain Res. Rev. 13:57-75.
Bullock TH, Hagiwara S (1957): Intracellular recording from the giant synapse of the squid. J. Gen. Physiol. 40:565-577.
Bullock TH, Heiligenberg WF (1986): Electroreception. John Wiley & Sons, New York.
Bullock TH, Horridge GA (1965): Structure and Function in the Nervous Systems of Invertebrates. W.H. Freeman and Company, San Francisco, pp 1-1610.
Bullock TH, Terzuolo CA (1957): Diverse forms of activity in the somata of spontaneous and integrating ganglion cells. J. Physiol. 138:341-364.
Bullock TH, Hagiwara S, Kusano K, Negishi K (1961) Evidence for a category of electroreceptors in the lateral line of gymnotid fishes. Science 134:1426-1427
Carlson AJ (1904): The nervous origin of heart-beat in Limulus and the nervous nature of co-ordination or conduction in the heart. Am. J. Physiol. 12:67-74.
Cohen MJ, Jacklet JW (1967): The functional organization of motor neurons in an insect ganglion. Phil. Trans. R Soc. Lond. B Biol. Sci. 252:561-572.
Dudel J, Kuffler SW (1961): Presynaptic inhibition at the crayfish neuromuscular junction. J. Physiol. 155:530-542.
Eccles JC (1961) The nature of central inhibition. Proc. Roy. Soc. B, 153:445-476.
Eccles RM, Libet B (1961): Origin and blockade of the synaptic responses of curarized sympathetic ganglia. J. Physiol. (London) 157: 484-503.
Eyzaguirre C, Kuffler SW (1955): Processes of excitation in the dendrites and in the soma of single isolated sensory nerve cells of the lobster and crayfish. J. Gen. Physiol. 39:87-119.
Fatt P, Katz B (1951): An analysis of the end-plate potential recorded with an intra-cellular electrode. J. Physiol. 115:320-370.
Fehmi LG, Bullock TH (1967): Discrimination among temporal patterns of stimulation in a computer model of a coelenterate nerve net. Kybernetik 3:240-249.
Fessard A (1936): Recherches sur l'Activité Rhythmique des Nerfs Isolés. Hermann, Paris,
Frazier WT, Kandel ER, Kupfermann I, Waziri R, Coggeshall RE (1967): Morphological and functional properties of identified neurons in the abdominal ganglion of Aplysia californica. J. Neurophysiol. 30:1288-1351.
Furshpan EJ, Potter DD (1957): Mechanism of nerve-impulse transmission at a crayfish synapse. Nature (Lond.) 180:342-343.
Galambos R (1989): Electrogenesis of evoked potentials. In: Brain Dynamics: Progress and Perspectives. E. Baar and T.H. Bullock (eds.). Springer-Verlag, Berlin, pp. 13-25.
Gerard RW (1941): The interaction of neurones. Ohio J. Sci. 41:160-172.
Gesell R (1940): Forces driving the respiratory act. A fundamental concept of the integration of motor activity. Science 91:229-233.
Goldschmidt R (1909): Das Nervensystem von Ascaris lumbricoides und megalocephala. II. Z. Wiss. Zool. 92:306-357.
Goldschmidt R (1910): Das Nervensystem von Ascaris lumbriocoides und megalocephala. III. In: Festschr. für R. Hertwig, Vol. II. Fischer, Jena.
Granit R, Skoglund CR (1945): Facilitation, inhibition and depression at the "artificial synapse" formed by the cut end of a mammalian nerve. J. Physiol. 103:435-448.
Graubard K, Raper JA, Hartline DK (1980): Graded synaptic transmission between spiking neurons. Proc. Natl. Acad. Sci. U.S.A. 77:3733-3735.
Gray J (1950): The role of peripheral sense organs during locomotion in the vertebrates. In: Physiological Mechanisms in Animal Behavior. Cambridge University Press, Cambridge, pp. 112-126.
Hagiwara S, Bullock TH (1957): Intracellular potentials in pacemaker and integrative neurons of the lobster cardiac ganglion. J. Cell. Comp. Physiol. 50:25-47.
Hartline HK, Wagner HG, Tomita T (1953): Mutual inhibition among the receptors of the eye of Limulus. Int. Physiol. Congr. 19:441-442.
Hodgkin AL (1938): The subthreshold potentials in a crustacean nerve fibre. Proc. R. Soc. Lond. B Biol. Sci. 126:247-285.
Holst E von (1935): Die Koordination der Bewegung bei den Arthropoden in Abhängigkeit von zentralen und peripheren Bedingungen. Biol. Rev. 10:234-261.
Holst E von (1936a): Vom Dualismus der motorischen und der automatisch-rhythmischen Funktion im Rückenmark und vom Wesen des automatischen Rhythmus. Pflügers Arch. Eur. J. Physiol. 237:356-378.
Holst E von (1936b): Über den "Magnet-Effekt" als koordinierendes Prinzip im Rückenmark. Pflügers Arch. Eur. J. Physiol. 237:655-682.
Holst E von (1939a): Die relative Koordination als Phänomen und als Methode Zentralnervöser Funktionsanalyse. Ergebn. Physiol. 42:228-306.
Holst E von (1939b): Über die nervöse Funktionsstruktur des rhythmisch tätigen Fischrückenmarks. Pflügers Arch. Eur. J. Physiol. 241:569-611.
Jahn TL, Crescitelli F (1938): The electrical response of the grasshopper eye under conditions of light and dark adaptation. J. Cell. Comp. Physiol. 12:39-55.
Jasper HW, Monnier AM (1938): Transmission of excitation between excised non-myelinated nerves. An artificial synapse. J. Cell. Comp. Physiol. 11:259-277.
Jones EG (1994): The neuron doctrine 1891. Proc. R. Soc. Lond. B Biol. Sci. 124:244-276.
Josephson RK, Reiss RF, Worthy RM (1961): A simulation study of a diffuse conducting system based on coelenterate nerve nets. J. Theor. Biol. 1:460-487.
Kalmijn AJ (1988): Detection of weak electric fields. In: Sensory Biology of Aquatic Animals. J. Atema, R.R. Fay, A.N. Popper and W.N. Tavolga. (eds.). Springer-Verlag, New York, pp 151-186.
Katz B (1937): Experimental evidence for a non-conducted response of nerve to subthreshold stimulation. Proc. R. Soc. Lond. B Biol. Sci. 124:244-276.
Katz B, Schmitt OH (1940): Electrical interaction between two adjacent nerve fibres. J. Physiol. 97:471-488.
Kerkut GA, French MC, Walker RJ (1970): The location of axonal pathways of identifiable neurons of Helix aspersa using the dye procion yellow M-4R. Comp. Biochem. Physiol. 32:681-690.
Laporte Y, Lorente de Nó R (1950): Potentials evoked in a curarized sympathetic ganglion by presynaptic volleys of impulses. J. Cell. Comp. Physiol. 36: Suppl. 2, 61-106.
Libet B, Tosaka T (1970): Dopamine as a synaptic transmitter and modulator in sympathetic ganglia: a different mode of synaptic action. Proc. Nat. Acad. Sci. 67:667-673.
Lloyd DPC (1941): A direct central inhibitory action of dromically conducted impulses. J. Neurophysiol. 4:184-190.
Maynard DM (1955): Activity in a crustacean ganglion. II. Pattern and interaction in burst formation. Biol. Bull. 109:420-436.
Maynard DM (1967): Organization of central ganglia. In: Invertebrate Nervous Systems. C.A.G. Wiersma (ed.). University of Chicago Press, Chicago, IL, pp 231-255.
McCulloch WS, Pitts W (1943): A logical calculus for ideas immanent in nervous activity. Bull. Math. Biophys. 5:115-133.
Moller P, Fritzsch B (1993): History of electroreception. From electrodetection to electroreception: the problem of understanding a non-human sense (Proceedings of a conference in honor of Thomas Szabo, "Contributions of Electrosensory Systems to Neurobiology and Neuroethology" edited by CC Bell, CD Hopkins, K Grant. J Comp Physiol A 173:657-763). J. Comp. Physiol. A 173:734-737.
Naka K-I (1961): Recording of retinal action potentials from single cells in the insect compound eye. J. Gen. Physiol. 44:571-584.
Nishi S, Koketsu K (1968): Analysis of slow inhibitory postsynaptic potential of bullfrog sympathetic ganglion. J.Neurophysiol. 31:717-728.
Pantin CFA (1935): The nerve net of the Actinoza. I. Facilitation. J. Exp. Biol. 12:119-138.
Parry DA (1947): The function of the insect ocellus. J. Exp. Biol. 24:211-219.
Prosser CL (1936): Rhythmic activity in isolated nerve centers. Cold Spring Harbor Symp. Quant. Biol. 4:339-346.
Ripley SH, Bush BMH, Roberts A (1968): Crab muscle receptor which responds without impulses. Nature (Lond.) 218:1170-1171.
Rosenblueth A (1941): The stimulation of myelinated axons by nerve impulses in adjacent myelinated axons. Am. J. Physiol. 132:119-128.
Rosenblueth A, Alanís J, Mandoki J (1949a): The functional refractory period of axons. J. Cell. Comp. Physiol. 33:405-439.
Rosenblueth A, Wiener N, Pitts W, García Ramos J (1949b): A statistical analysis of synaptic excitation. J. Cell. Comp. Physiol. 34:173-205.
Rovainen CM (1967): Physiological and anatomical studies on large neurons of the central nervous system of the sea lamprey (Petromyzon marinus). I. Müller and Mauthner cells. J. Neurophysiol. 30:1000-1023.
Stefanelli A (1951): The Mauthnerian apparatus in Ichthyopsida; its nature and function and correlated problem of histogenesis. Quart. Rev. Biol. 26:17-34.
Tauc L (1959): Sur la nature de l'onde de surpolarisation de longue durée observée parfois après l'excitation synaptique de certaines cellules ganglionnaires de mollusques. C. R. Acad. Sci. (Paris) 249:318-320.
Tauc L (1962): Identification of active membrane areas in the giant neuron of Aplysia. J. Gen. Physiol. 45:1099-1115.
Terzuolo CA, Bullock TH (1956): Measurement of imposed voltage gradient adequate to modulate neuronal firing. Proc. Natl. Acad. Sci. U.S.A. 42:687-694.
Tosaka T, Chichibu S, Libet B (1968): Intracellular analysis of slow inhibitory and excitatory postsynaptic potentials in sympathetic ganglia of the frog. J. Neurophysiol. 31:396-409.
Watanabe A, Bullock TH (1960): Modulation of activity of one neuron by subthreshold slow potentials in another in lobster cardiac ganglion. J. Gen. Physiol. 43:1031-1045.
Weiss P (1941): Does sensory control play a constructive role in the development of motor coordination? Schweiz. Med. Wochenschr. 71:591-595.
Werblin FS, Dowling JE (1969): Organisation of the retina of the mud puppy. Necturus maculosus II. Intracellular recording. J. Neurophysiol. 32:339-355.
Wiersma CAG, Van Harreveld A (1938): A comparative study of the double motor innervation in marine crustaceans. J. Exp. Biol. 15:18-31.
Wilson DM (1961): The central nervous control of flight in a locust. J. Exp. Biol. 38:471-490.
Young JZ (1936): The structure of nerve fibers and synapses in some invertebrates. Cold Spring Harbor Symp. Quant. Biol. 4:1-6.