Home Page || The Calvin Bookshelf || Science Surf magazine || Table of Contents
A book by
William H. Calvin
The Throwing Madonna
Essays on the Brain
Copyright 1983, 1991 by William H. Calvin.

You may download this for personal reading but may not redistribute or archive without permission (exception: teachers should feel free to print out a chapter and photocopy it for students).

Scanned, OCR'ed, and webbed -- but NOT proofread (14 Jan 97)


On June 8, 1866, the official Censor's Committee of St. Petersburg presented an indictment against the pioneering Russian neurophysiologist, Ivan M. Sechenov, for attempting to publish a small book, Reflexes of the Brain: "This materialistic theory reduces even the best of men to the level of a machine devoid of self-consciousness and free will, and acting automatically; it sweeps away good and evil, moral duty, the merit of good works and responsibility for bad works; it undermines the moral foundations of society and by so doing destroys the religious doctrine of life hereafter; it is opposed both to Christianity and to the Penal Code, and consequently leads to the corruption of morals."

        When the book was under investigation by the courts, a friend asked Sechenov (who eventually became Pavlov's mentor) why he did not seek the counsel of an attorney--to which he answered, "Why should I need a lawyer? I shall take a frog with me to court and perform my experiments in front of the judge; then let the State's attorney refute me!"
adapted from LEONARD A. STEVENS,
Explorers of the Brain

The most modest research worker at his bench, pushing a probe into a neuron to measure the electric response when a light is flashed, is enmeshed in a large and intertwined network of theories that he carries into his work from the whole field of science, all of the way from Ohm's law to Avogadro's number. He is not alone; he is sustained and held and in some sense imprisoned by the state of scientific theory in every branch. And what he finds is not a single fact either: It adds a thread to the network, ties a knot here and another there, and by these connections at once binds and enlarges the whole system.
JACOB BRONOWSKI, The Identity of Man


The Computer as Metaphor in Neurobiology

Either you sit there and just close off, or if you do become engaged in what is going on with other people, then you have lost the thread [of your concentration upon your writing]. You've turned off the computer, and it is not for that period of time making the connections that it ought to be making.
      I really started thinking of my mind mechanically. I almost heard a steady humming if it was working all right, but if it stopped for a couple of days, then it would take a while to get it back.


Someday in the distant future, a team of archaeologists will dig up a computer. But, alas, no one will have remembered to leave behind the instruction manual and blueprints. At a minimum, the specialists will open the cabinet doors and give Latin names to the various cable bundles. They may even embed a computer in clear plastic and slice it up with a bandsaw for closer examination of cross sections.
      But if the computer is in working condition, another archaeologist will undoubtedly try unplugging various cables, just to see what stops working. Through indulging such childlike curiosity, he or she would gradually identify the "head office" for various functions, such as the locations essential for input, output, memory, even addition. The psychological archaeologists would put the machine through its paces, demonstrating that it could play chess and do payrolls simultaneously. Such relatively crude knowledge would parallel our understanding of the brain only a few decades ago, which was based largely on anatomy, pathology, and descriptions of stroke victims.
      But locating the head office for various functions, whether in a brain or a computer, does little to reveal how the elementary computing units carry out their electrical machinations. The archaeologist might want to attach a hi-fi amplifier to various computer wires, to listen in on the internal communications. The nerve cells that comprise the brain also run on electricity, computing and signaling with small voltages. A few decades ago, neurophysiology entered its "wiretapping" era: we began listening in on the conversations of individual neurons, the brain cells that analyze the external and internal worlds, signal the muscles, store information, and somehow provide the underpinnings of intelligence.

The Impulse Arises
      The most eye-catching feature of a neuron's electrical repertoire is a sudden up-and-down-again swing in the voltage across the cell's membrane. Named the "impulse" by pioneering wiretappers, it shoots up about 100 millivolts (still, a thousand times smaller than household voltages) and then right back down again. Hence its nickname, "the spike." The whole event lasts about 1/1000 of a second, less than the time it takes most camera shutters to open and close. Making a light brighter or a sound louder won't change the neuron's impulse: in any given cell, they either set off a standard-size impulse or they don't. Nothing in between. And there is a threshold for setting it off, rather like pulling the trigger on a gun. A gentle pull and nothing happens; just enough, and bang; harder, and one still gets the same old response.
      This all-or-nothing attitude on the part of the brain cell has long reminded observers of the computer's elementary computing unit, the flip-flop: information isn't coded by grading the size of the voltage, but by whether it is on or off. And indeed the neural impulse does seem more like flipping an ordinary light switch on and off, rather than like adjusting a dimmer switch. But what serves to flick the switch? Each neuron receives small voltages from hundreds or thousands of other neurons; it generally requires a sum of many such inputs to set off an impulse. Those more familiar with computers than neurons immediately see an analogy (physicist's fallacy #1): The neuron must, so the extrapolation goes, be detecting simultaneous events as does ten AND gate. So the brain is a digital computer in disguise!
      But this digital view of the neuron's physiology is simply wrong. Many flights of fancy have, however, been based upon it. They may describe some hypothetical computer, but not the particular one inside our heads. Its building blocks work quite differently.

Neurons Without Impulses
      It comes as a heresy even to many neurophysiologists, but the fact is that many neurons get along fine without impulses-- such as 99 percent of the neurons in the eyes that are reading this essay. Like deposits and withdrawals from a savings account, it is the balance between excitatory and inhibitory inputs that counts. In a nonimpulsive neuron, the balance controls the signal sent to the next neuron in the chain--just as the savings account yields ("outputs," in computerese) interest proportional to its balance. Neurons even have idiosyncratic rules like banks, some requiring a minimum balance before starting to put out anything ("Everything over $1000 earns 7 percent!). So too with the product of neurons, the neurotransmitter molecules that are sent sailing on to the next brain cell: their output rate is controlled by the voltage balance, provided that it is above a threshold. Yet there is one big difference: Nonimpulsive neurons always send their "interest" on to another cell rather than adding it to their own balance. No compound interest.
      A nice feature of some neurons, which one wishes the banks would also learn to mimic, is to provide output with no input: they spontaneously leak neurotransmitter molecules, thus allowing them to adjust the leakage rate up or down in response to even the smallest input. But not all banks are the same--and certainly neurons are a varied lot.
      Inhibition ("withdrawals") is generally produced by a different neurotransmitter molecule than excitation ("deposits"). But neurons cannot add and subtract different kinds of neurotransmitter chemical. That would be like adding apples and oranges.
      The solution, discovered in evolution at least 500 million years before the invention of money, is to first convert everything into an artificial common currency, like dollars. The neurotransmitter merely alters the leakiness of the cell's membrane, which raises or lowers its voltage. Volts, in the case of the neuron, are the medium for the message. And volts is even a fairly universal currency (some regions of some cells happen to have some local currencies, such as internal calcium or hydrogen ion concentrations, but they accept volts as well).

The Long-Distance Impulse
      So, if variable volts are so useful, why are standard-size impulses ever used? Impulses seem to be essential for sending messages over a long distance (to the cell, that's perhaps a typewriter space or more) because voltage "balances" would be gradually lost by the leaky membrane. Just as one cannot use too long a length of leaky garden hose and still expect water to flow reliably out the far end, so lengthy cells may gradually exsanguinate the signal. Just as some lengths of garden hose are too short for the leaks to matter, so one talks of "short neurons" and "long neurons."
      But long neurons have invented a solution: They can locate booster stations every millimeter along their leaky "axon" between the input end of the cell and the output end, where the neuro-transmitter is actually released. The boosters aren't exactly high-fidelity amplifiers of the sort installed in underwater telephone-cable booster stations, but they are good enough to boost an impulse back up to its standard height. Thus the impulse arrives at the other end of the neuron none the worse for its travels.
      Now there is no limit to how long the cell can be, provided it keeps adding another booster station every millimeter (one neuron may extend from your toe up to your neck, using perhaps 2000 booster stations called "nodes of Ranvier"). Cells short enough to avoid exsanguination may not require impulses for long-distance communications, yet they may use an occasional impulse anyway. Just for emphasis, short neurons may produce an extra squirt atop the steady secretion of neurotransmitter released by nonimpulsive means.

Coding with Impulse Trains
      There are people who see codes and ciphers embedded in the impulses. True. One impulse, by itself, it like one dot or dash in Morse Code: usually meaningless, except in the context of its neighbors. Thus there is much more to neural signaling than just what sets off the first impulse; it is the rhythm of the impulsive neuron that carries the message. But even the textbooks are fond of halting the story after explaining the first impulse which is rather like a music critic listening to the opening strains of Beethoven's Fifth and then discoursing only upon the first note of dit-dit-dit-dah.
      In the neuron, the brightness of a light or the loudness of a sound is indeed likely to be coded by varying the rhythm at which impulses are produced. Physicist's fallacy #2: The timing of an impulse train must be analogous to the on-and-off codes we send to "serial line devices" such as computer terminals and printers. Ergo, the exact pattern of impulses must be very important, just as in computers. But some caution is in order: It would only have added insult to injury for the music critic's response to dit-dit-dit-dah to be instead a claim that the symphony was about the letter "V" in Morse code. (Besides, Beethoven came first by half a century and it seems more likely that Samuel F. B. Morse selected dit-dit-dit-dah for "V" based on the Roman numeral for Beethoven's symphony). There is, fortunately, a simpler encoding and decoding scheme to be found in most impulsive neurons.
      Neurons that use impulses seem to work on analog principles, similar to their nonimpulsive cousins. When the impulse reaches the far end of the cell, it releases a standard-size squirt of neurotransmitter. So how do the inputs' voltages encode the message to be sent by the impulsive neuron? They simply vary the rate at which it produces impulses, with higher voltages causing the neuron to "beat" faster (a principle called frequency modulation, more recently applied to FM radios) and thus release more squirts per second. This makes the total neurotransmitter released each second at the distant end vary up and down as does the original balance of input voltages. If this sounds familiar, it's because this scheme is just a roundabout way of doing what the nonimpulsive neuron did, without impulses as the middleman: vary transmitter output with voltage input.
      A better, more homely analogy for impulsive neurons is the sewing machine's pedal (rather than the gun's trigger). Press lightly and nothing happens. A little harder and the machine starts stitching rhythmically, but slowly. Harder still and it speeds up, proportionally to the pedal pressure --just like neurons speeding up and slowing down their beat as the balance of excitatory and inhibitory inputs changes. And just as some sewing machine models oblige you to press harder before anything happens at all, so some neurons have high thresholds, others low. Always diversity.
      But some neurons have additional tricks, such as occasionally giving two impulses for the price of one, even though the voltage balance hasn't changed. Imagine a sewing machine occasionally doing a double stitch in the midst of its regular rhythm, or a bank offering to briefly double your money. It would be almost as startling as your heart giving a quick double beat. But it is just another item in the neuron's repertoire, probably encoding for something which the next neuron in the chain can decode--but which neurophysiologists still can't.
      The moral of this little story? The workings of the brain are, at the level of elementary computing mechanisms, surprisingly simple--much more like balancing your checkbook and earning interest than like any binary mechanism inside a computer (or a physicist's fallacy). But there are enough variants to satisfy any free-market economist, to fill any ecological niche, to endlessly delight neurophysiologists.

Malignant Metaphor, Rampant Reductionism
      Yes, the brain is a computer--that underpowered metaphor will have to suffice until a more subtle technological analogy is invented--but surface similarities to existing digital computers can be deceptive. There is a simple-minded longing within biology and psychology for hard-science insights (Freudians might call it "physics envy") that reduce soft-but-sophisticated phenomena to "nothing but" hard physics and elegant mathematics. (Perhaps like reducing Beethoven's Fifth to nothing but a Fourier series of different tones?)
      Physics and mathematics can be most elegant, but the real world does not always cooperate with our aesthetic enthusiasms. Indeed, Ptolemy's belief that only perfect circles could be the building blocks of the universe sidetracked astronomy for fifteen centuries. Finally Kepler abandoned epicycles and tried out less elegant ellipses to represent the orbits of the planets (and then even Galileo didn't believe him--but Newton did).
      Hopefully the aesthetics of computers and the appeals of reductionism will not similarly sidetrack neurobiology. Neural reality is much richer, more diverse, with many levels between membranes and brains. A level above membranes lie cells, with their regional specializations of different kinds of membrane. Above cells lie circuits of different kinds of cells. There is even one known cortical "module" with stereotyped circuity (a cubic millimeter known as a hypercolumn) repeated hundreds of times within the visual cortex. Somewhere above that, somehow connected, lies the high level of the uniquely human language cortex, whose core is surrounded by a patchwork of specialized areas for semantics, syntax--even a short-term "holding" memory for words.
      Being able to explain one level in terms of the level below is a good test of the depth of your understanding, but it isn't everything (breadth, anyone?). While one might be able to reduce the chemist's valence to the physicist's electron waves, who would want to go to all that quantum mechanical trouble every time? Just as valence has a life of its own for predicting chemical compounds, so each level of a nervous system must be understood in its own right, in context, in all its rich variety--not just reduced to "nothing but" the level below. Nor to merely an analogy with computer hardware of recent design.
      Evolution has, after all, been tinkering around with thinking machines for a lot longer than computer science has. Brains are the most elegantly organized bundles of matter in the universe. Why be satisfied with substitutes when you can study the real thing? And, as Oscar Wilde said, "Truth is never pure, and rarely simple." But certainly more interesting than malignant metaphor and rampant reductionism.

The Throwing Madonna:
Essays on the Brain
(McGraw-Hill 1983, Bantam 1991) is a group of 17 essays: The Throwing Madonna; The Lovable Cat: Mimicry Strikes Again; Woman the Toolmaker? Did Throwing Stones Lead to Bigger Brains? The Ratchets of Social Evolution; The Computer as Metaphor in Neurobiology; Last Year in Jerusalem; Computing Without Nerve Impulses; Aplysia, the Hare of the Ocean; Left Brain, Right Brain: Science or the New Phrenology? What to Do About Tic Douloureux; The Woodrow Wilson Story; Thinking Clearly About Schizophrenia; Of Cancer Pain, Magic Bullets, and Humor; Linguistics and the Brain's Buffer; Probing Language Cortex: The Second Wave; and The Creation Myth, Updated: A Scenario for Humankind. Note that my throwing theory for language origins (last 3 essays) has nothing to do with the title essay: THE THROWING MADONNA is a parody (involving maternal heartbeat sounds!) on the typically-male theories of handedness.
Many libraries have it (try the OCLC on-line listing, which cryptically shows the libraries that own a copy), and used bookstores may have either the 1983 or the 1991 edition.

Email || Home Page || The Calvin Bookshelf || Table of Contents || End notes for this chapter || Continue to read next chapter