William H. Calvin and George A. Ojemann's CONVERSATIONS WITH NEIL'S BRAIN (chapter 18)
Home Page || Public Bookmarks || Science Surf magazine || Table of Contents
Conversations with Neil’s Brain
The Neural Nature of Thought & Language
Copyright  1994 by William H. Calvin and George A. Ojemann.

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).

William H. Calvin, Ph.D., is a neurophysiologist on the faculty of the Department of Psychiatry and Behavioral Sciences, University of Washington.

George A. Ojemann, M.D., is a neurosurgeon and neurophysiologist on the faculty of the Department of Neurological Surgery, University of Washington.

In Search of the Narrator

NEIL FOUND US back in our usual quiet alcove of the hospital cafeteria and carried his tray over. Other than the baseball cap he’s wearing while his hair grows back — and also serving to hide the U-shaped scar — Neil looks as he did during our pre-op conversations. It’s only a month since the tip of his left temporal lobe was removed.
      “Well, I guess I gave you a piece of my mind!” Neil said as he unloaded his tray.
      I’m afraid George and I groaned so loudly that several people turned around to look our direction.
      “But so far as I can tell, I’m not missing any marbles,” he said, looking up at us with a grin.
      No, only a piece of brain. Anatomy. The mind is the brain in action, its physiology. Mind is like what the computer produces, rather than the computer itself.
      But Neil didn’t buy my computer analogy. “Now if it were my computer,” he observed, “knocking out a piece of it would cause it to crash. And it could never reboot itself.”
      Yes, but sometimes mental processes aren’t so sensitive to crippled hardware. Just imagine computer software that can reconfigure itself to run on various combinations of hardware, after testing to see what’s available.
      “With consciousness as the operating system?” Neil was teasing me again. “The original user interface for the little man inside the head?”
      He inspected his salad, then went on. “I think that the analogy you want is not a computer, but rather a computer network,” he said. “One that can share work around among the idle processors. You ought to figure out how the brain can reorganize itself, somehow reassign space. Dynamic Reorganization, you can call it. Then patent it quickly, before a computer designer does!”
      “Except for the little problem that we don’t know enough about how the brain works,” George replied. “Not nearly enough to be able to build a working model. One that can walk, talk, juggle cafeteria trays, worry about tomorrow, and tell bad jokes about missing marbles.”
      Neil grinned. “But just think what nice machines you could design with neural principles. Isn’t that what the neural networks researchers are doing?”
      They’re trying to design machines — electronic circuits, at least — that don’t have to be designed after a certain point. That instead can be tuned up or trained to perform specialized functions, such as speech recognition. The usual hyperbole, even repeated by the New York Times in a headline, is that the principles they use are “based on the human brain.”
      Which is gross overstatement. The everything-connects-to-everything wiring principles they utilize are found even in jellyfish nervous systems. Other principles for wiring up neural circuits have been superimposed on them, even by the time you get to the complexity of a crab nervous system — and neural-like networks haven’t gotten as far as crabs yet. I wrote a letter to the Times suggesting that a more appropriate headline, considering the current state of the art, would have been “based on slug nerves.”
      “Isn’t artificial intelligence trying to use them for fancier things,” Neil argued, “such as pattern recognition — handwriting, fingerprints, and stuff like that?”
      Yes, and maybe the simplest principles will suffice for fancy functions, without reverse-engineering the brain. I wish them well. But I’ll bet they’re going to have to find some fancier principles first, probably borrowing them from neurophysiology and evolutionary biology.
      George had finally finished layering all possible additions atop his hamburger. Even salad dressing. He looked up at Neil. “The problem, you see, is that you’re talking to a couple of biological chauvinists here. Who think that real brains are far more interesting than anything likely to be invented in silicon very soon.”
      “Yes, but biological brains have several drawbacks,” Neil persisted. “Wouldn’t you like the perfect traffic cop directing vehicles at every intersection on your way home from work, one that never got bored by the task, or demented from breathing the fumes? You’d never have to wait for the light to change, even though no traffic was coming. And it could wave you through a gap in the oncoming traffic for a quick left turn, just the way traffic cops do.”
      George and I allowed as how that might be nice.
      “And the other problem is that human brains seem to gradually self-destruct with age. Don’t you like that science-fiction idea of making a silicon working model of your brain, so that your mind can live forever?”
      The disembodied mind? But the brain is intimately a part of the body, evolved to be the ultimate hand-in-glove combination. You might be able to design a thinking machine with minimal sense organs and output devices, but it’s hard to imagine a human mind perking along without at least a head and a hand. Remember all those sensory isolation experiments that caused hallucinations after a little while? External reality is what keeps chaos in check — the brain has a perpetual battle between stability and flexibility, and external reality is a major arbiter.
      Replicating your brain in silicon sounds like a recipe for psychosis in perpetuity. And that even assumes that they got the silicon workalike circuits tuned up right, so that the whole system didn’t freeze or oscillate endlessly. No thanks.
      Reverse engineering the brain, however, is much more promising — you’d create a more stable machine, only parts of which flirted around on the edge of chaos to be creative. And then you would educate it like a child, over a period of many years. The good ones could presumably be cloned, if you’d built in readout facilities from the beginning. Different schools could have competitions to see whose star pupil got cloned, memories and all.
      Neil shook his head in resignation. “Well, in any event, I’ve got another data point for you, in your search for the seat of the soul.”
      “What’s that?” George responded, bemused.
      “Consciousness isn’t in the tip of the left temporal lobe — I can still think without it!”
      “All too well,” George observed. “Of course, consciousness wasn’t in your brain stem either, unless you want to equate it with mere arousal. Selective-attention circuits up in the thalamus and the cortex have a lot more to do with it. Consciousness is more like a searchlight that moves around from one part of the cerebral cortex to another.”
      That’s the problem with using the same word, consciousness, to describe things as different as arousal, focused attention, and talking to oneself. If you want to include selective attention within consciousness, then a three-month human fetus isn’t conscious but cats are. All cells, plant and animal, have irritability — and so by some overinclusive definitions, they must be “conscious.”
      “Yes, I remember our first discussion of consciousness terminology,” Neil said. “It was that day we first discovered this corner table where I could get my coffee fix by merely inhaling.”
      If you set the consciousness threshold at talking to yourself, I continued, then you’ve said that only humans are conscious — unless you want to believe in cartoon-strip depictions of animals that talk only to themselves and never out loud.
      “Of course,” George added, “if you’re going to set the threshold for consciousness that high, you’ll probably leave out some essential considerations. Such as that changing focus of selective attention, why we get bored after a while even when satisfied.”
      The hierarchy of mental functions builds up from the lowest level, that of deep anesthesia where very little is working, to the level of deep sleep — functionally, that’s like the late part of seizures where slow electrical oscillations dominate the action.
      Above that is a level of functioning like stupor and dementia, where things aren’t working very well but at least aren’t stuck in the limit cycle of oscillations. They’re now up into the realm of chaos, at least in the mathematical sense of the word.
      Expectancy is what characterizes our better mental operations, both perception and guiding movements in novel situations.
      Concept formation is up there a little higher — all those new categories. And we build atop it for language and intuition, where we’re always dealing with novelty and trying to apply our detailed memories to generating novel sequences of words and actions.
      And then, of course, there’s the issue of where in the brain these processes take place — as they could all be happening at once, but in different regions of the cortex.
      “I need some examples. So let’s take my train of thought during the drive to the hospital this morning,” Neil proposed. “In anticipation of being at the wheel again, I’ve been practicing back-seat driving on all my cabbies. As I was sitting there paying attention to the red light, waiting for it to change to green, what part of my brain was working harder than usual? The visual cortex?”
      “Not compared to looking around in general,” George replied. “It’s that right frontal area, plus that right parietal area, which is especially involved in vigilance tasks. They’re what keep the driver behind you from having to honk.”
      “I remember drumming with my fingers, getting impatient.”
      That’s your left supplementary motor area, plus the motor cortex, producing a nice limit-cycle oscillation.
      “And, of course, I was thinking over my schedule for the day. Where’s that?”
      “Maintaining a mental agenda seems to be in the frontal lobes,” George replied. “It’s the patients with damage to the left dorsolateral frontal lobe that have trouble monitoring progress in a mental agenda.”
      “And then the light changes, and the landscape starts flashing past on both sides of the car. What now?”
      The magnocellular pathways of the visual system, the ones that don’t work as well in dyslexics. It may only be 10 percent of the visual system, but it specializes in movement and depth perception. You probably stirred up lots of activity in an area known as V5.
      “A house that I noticed reminded me of a rather different house, one of those Frank Lloyd Wright houses built around a little waterfall. How did I bring up that fancy set of images from memory?”
      “Probably you activated some parietal-lobe areas specializing in complex forms,” answered George. “And those in turn reactivated a lot of the higher-order visual areas that participated in analyzing the house when you first learned about it.”
      “The cab driver pointed out an Italian sports car, and so I started keeping my eyes open for another Maserati. Where’s that?”
      The front part of the cingulate gyrus, in the midline on both sides of the brain. At least that’s what lights up when you’re staying alert for particular classes of objects. I don’t know where your Maserati memories are stored — maybe in the front of your temporal lobe.
      “It sure isn’t stored in my left temporal lobe!” grinned Neil. “That’s gone now! And good riddance. Must be here,” he said, pointing at his right temple.
      Warming to his tour, he continued. “Then I saw this poster on the back of a bus, and tried to read it.”
      That ought to have stimulated more activity in the parvo parts of your visual pathways, all of those areas in the back of the temporal lobe that are involved in reading.
      “But the poster had one of those tricky sentences that advertising agencies love to create,” he related. “It was intentionally missing the verb, just to force you to re-read it several times to figure out what you’d missed.”
      “That certainly ought to have stirred up your left hemisphere,” George laughed. “All of the language areas. And especially the left frontal lobe, down near your eyebrow. It gets very active when you have to supply a verb.”
      “And then another bout of right-hemisphere activity as we waited for a light to change,” Neil continued, “followed by our grand entrance onto the freeway. But we soon slowed down to a snail’s pace. So I started thinking about getting off the freeway and trying an alternate route.”
      Alternative courses of action are a frontal lobe function, par excellence. And surely the right parietal lobe was stirred up too, what with all those spatial tasks.
      “Then I started worrying about being late, checking my watch constantly. So does the temporal lobe really have something to do with time?”
      It’s called the temporal lobe because it’s behind the temple.
      “But isn’t that the same word root? For ‘time’?”
      The Latin word tempus.
      “Why was the back side of the forehead named after the Latin word for ‘time’?”
      Because the hair there is the first to turn gray.
      “Aggh! So the temporal lobe doesn’t really have anything to do with time.”
      Well, actually it does. It just wasn’t named on account of it, as the timing role was discovered much more recently. When neuropsychologists ask you to tap a finger as fast as you can, they’re checking up on how well the tips of the temporal lobes are working. Trouble there, or in the frontal lobe on the other side of the sylvian fissure, tends to slow down finger-tapping rate. It takes a lot of coordination to tap rapidly.

CATEGORIES, CONCEPTS, WORDS. They too require a lot of coordination between different regions of the cerebral cortex. You’ve got to be able to memorize things in versatile ways, invent new categories. Guess what another person might be thinking about. Confuse concepts, sometimes creatively. We mix and match to make metaphors.
      “What about those people who seem to have categories for animal, vegetable, and mineral?” Neil asked. “Or at least, that’s how I remember them. The newspaper reports say they lose all their vegetable names after a little stroke.”
      “Each patient shows something a little different,” George said. “Some can name tools, but not animals. Others have been missing their plants, or body parts, or verbs, or the combination of food, fruits, and vegetables — as if that category was stored in the brain destroyed by the stroke. They can usually recognize the word, and write the word — what they’re missing is the visual representation of that category. Sometimes they can use a word like crack as a verb but not as a noun. Electrical surface stimulation has shown some effects like that in the strip electrode studies. We don’t know if any of these reported categories are repeatable from one patient to the next — they could just be idiosyncracies related to how you learned words back when you were two years old.”
      “So do other animals have categories?” Neil asked. He was happily munching on his sandwich.
      Birds can certainly learn categories. You can show pigeons pictures of sad people and happy people, and they’ll learn to spot the sad ones, even in a batch of pictures they’ve never seen before.
      But we humans can do very fancy categories, such as the various connotations of a single word. Comb must allow for “combing” your house in search of a lost book, as well as what a comb looks like, how it feels in your hand, the spelling c-o-m-b, the movement for combing one’s hair, the sound sequence /km/, and so on. Combs even have a characteristic smell — if I were to ask you to close your eyes and held one up under your nose, you’d probably be able to identify it.
      Rather than storing a comprehensive record of comb everywhere, the visual representation of a comb developed in extrastriate visual areas seems to be stored there. The auditory association areas have the stored record of what comb sounds like — and probably a separate record of the characteristic sound made by running your fingernail along the teeth. Other areas probably have the record of the movement sequences needed to use a comb or pronounce the word.
      “But how are they tied together so one can evoke another? Seems like we’re back to Pavlov’s dog, associating the sound of the bell with the food.”
      “Good point,” George said. “Associations between and among things seem to be what the cerebral cortex is particularly good at. Things like habit and skills are thought to involve subcortical structures such as your basal ganglia.”
      “Subroutines are made in the subcortical structures?”
bk7p274.jpg 58.1 K
[FIGURE 79 Strokes affecting color perception and imagery]

      “That’s possible, but let’s stick to categories and committees for a minute,” George said. “It isn’t just the cortex that does committees. For color, it starts way out in the retina. Color is a much simpler case than a comb — there’s no sound associated with red, nor any feeling in the hand, nor any smell or taste.”
      And we know a lot about color, that it’s a committee property of three types of photoreceptors peaking in short, medium, and longer wavelengths. Equal amounts in all channels and we call it white. Long channel only, and it’s red.
      “I remember your Purple Principle. Short and long, but not much activity in the middle, and we say it’s purple.”
      “There are higher-order visual areas on the underside of the temporal and parietal lobes,” George resumed, “that seem essential for making use of those activity ratios. Patients with a stroke damaging those areas can’t even imagine colors anymore, though they’re otherwise pretty normal. They can’t name the paint sample that you show them. If you ask them to match up similar paint samples, they are hopeless. Their whole concept of color is gone, replaced by the twilight world of grays. It must be like living in a world of bright moonlight.”
      And, of course, there are patients with strokes in the rear of the lateral language areas who can’t name a color very well because they get the phonemes wrong — maybe they say “bluh” for blue, but you can get the idea that they indeed have the concept of color and can use it.
      “Then there are the rare patients who have the concept of color,” explained George, “but they can’t give you the name. They can pronounce the name from memory just fine, if you ask them what color a banana is. They can match up paint samples correctly. But they can’t name them for you. What’s missing seems to be the association between the color concept and the color name. The strokes that do this are on the underside of the left temporal lobe, adjacent to the ‘color concept’ area.”
      “There are all sorts of other examples like that,” George continued, “regions where certain kinds of knowledge seem to come together in a specialized way. There are patients with temporal lobe damage whose naming problems are worst when shown pictures of animals, and far less when trying to name inanimate objects such as screwdrivers.”
      “What about actions rather than objects?”
      “Verbs give them no problems,” George answered. “It’s patients with frontal lobe damage that have trouble with verbs. But concepts and their nouns seem to have something to do with the lower portion of the temporal lobe — in the rear, strokes are more likely to damage a general concept. Closer to the front of temporal lobe, it’s more likely to be proper names and unique episodes. That’s Tony Damasio’s idea of how the middle temporal lobe is organized.”
      “From the general to the particular,” said Neil, rephrasing. “You know, that’s probably the first time I’ve heard either of you mention some logical progression in brain organization. A principle at last.”
      Yes, I conceded. It’s too logical, in fact — too good to be true. Brains are fundamentally irrational. They’re patchworks. But we’ll see — Damasio could be right.

MAKING A DECISION to say “banana” when seeing one would seem to be pretty simple, so long as it is the only object around. Neil had gone back to get dessert and returned with the banana we were discussing. He seemed in no hurry to consume it.
      “So ‘banana’ is the activity in a collection of feature detectors,” Neil speculated, “a Hebbian cell assembly. Some of which like yellow things, some of which like curved things, some of which like fruits, and so on?”
      Yes, it’s a spatiotemporal pattern again. A rather widespread one, extending over many millimeters or even centimeters of brain.
      “And pronouncing ‘banana’ is another spatiotemporal pattern in the movement-control pathways up in frontal lobe,” George observed, “that orchestrates the mouth and tongue and vocal cords.”
      But there is no need to funnel down the spatiotemporal pattern to just one command neuron for “banana.” Nor to make the decision at some point in time. In fact, especially in brains less specialized than ours, there might not need to be two different spatiotemporal patterns — the sensory one might also be the movement one, because so many of the higher-order sensory neurons are also movement-associated neurons. In more complicated brains like ours, there probably are two or more distinct spatiotemporal patterns associated with “banana” — but it doesn’t have to be separated in time, like some sort of ‘he sings’ and then ‘she sings’ duet. The sensory and movement spatiotemporal patterns can overlap in time — real parallel processing.
      “So there’s no need to have a place and a time at which the decision is made,” mused Neil. “And that means no little person inside the head.”
      He thought a minute as he slowly peeled his banana. “So what’s that ‘binding problem’ I’ve read about, that postulates some sort of synchronous activity to tie together all those feature detectors in different places? Why do they need to be tied together? Doesn’t performing the action suffice to tie them together?”
      It does seem regressive, doesn’t it? Going backward toward what Dan Dennett mockingly calls the cartesian theater, a stage on which everything is played out, somewhere in the brain — with an implicit observer who decides.
      Neil threw up his hands, smiling mischievously. “Consciousness as the graphical user interface! Didn’t I tell you so?”
      But the need for some limited sort of binding is real enough, once you begin to worry about several objects at the same time, as Hebb’s colleague Peter Milner realized in 1974. Consider driving a car. There are dozens of moving objects in your field of view, each of which has a shape and color and distance away. We know from all those specialties of the higher-order visual areas that each object will set up activity in a number of the specialty areas. Some areas specialize in motion, others in shapes, others in distance away. And, of course, color.
      Each object is doing that, and the positional information gets blurred in the feature detectors that are doing the motion and color tasks — they register position only crudely compared to the neurons in the parvo pathways. There is some danger of assigning the wrong color to a moving object, that of an adjacent object.
      “So the traffic light appears to be moving — though it’s really me, driving along — and I might think that the top light was green, if I didn’t know better from memory?” Neil commented.
      Synchrony could save you from that mistake, or so the theory goes. The synchrony-for-binding theory postulates a tendency for the red light’s shape neurons, its color neurons, its distance neurons, and its motion neurons to all fire in near-synchrony. And for the green light’s committee to all fire together, probably at a somewhat different time in a cycle.
      “For example,” George said, “neurons in the motor cortex tend to get synchronized when the monkey has to concentrate on a particularly difficult task, such as groping for a raisin behind its back.”
      “Clapping in unison, no doubt, to signal impatience.”
      Unlikely. But I wouldn’t rule it out. There is certainly a lot of synchronized firing around, waiting to be explained — and it will probably turn out to be useful in various ways.
      “So how do you create categories, if the code for an item is a Hebbian cell assembly?” Neil wondered out loud. “Or, rather, your message-board amendment that says it’s a pattern, no matter where it’s located in the brain?”
      The Hebbian committee can be pretty sparse — maybe the neurons in a dozen active minicolumns, out of a few hundred neighboring minicolumns which otherwise keep quiet. That means that several cerebral codes, say the ones for apple and orange, can be superimposed to give you a category such as fruit. If you tried superimposing several letters from a dot-matrix printer, you’d get a black mess. But if the matrix is sparsely filled, you can probably recover the individual members because they each produce such distinctive spatiotemporal patterns.

WHAT SORT OF BRAIN ORGANIZATION makes for a good mechanic, or a good pianist, or a good composer? Neil and George had been comparing opinions of operas I’d never heard of.
      “We don’t know what the optimal brain organization for a musician is,” George said, “but functional mapping techniques ought to tell us someday.”
      There are certainly some temporary interactions between music and intelligence that make one wonder about short-term optimal arrangements. Someone did intelligence tests after subjects had listened to a highly-structured Mozart piano sonata and got 8-9 percent increases — even if they didn’t like Mozart! But the increase was only temporary, as if music was a good warm-up exercise for the rapid language processing and problem-solving that IQ tests require. Many researchers think that music is very similar to language — just a variation on the basic theme.
      “But I still don’t understand the basic theme — the long-term arrangement of the average brain,” complained Neil. “Oh, I now know a lot about the departments of the brain. Reminds me a lot of modern companies, actually.”
      Really? A hierarchy it isn’t.
      “No, no. You’ve got a military model in mind, and most new organizations really aren’t like that anymore.” Now Neil was the instructor. “Taken together, the brain ‘departments’ remind me of an organization of knowledge specialists. Like the ones that advise communities on how to solve their garbage problems. The organizations in this building seem to practice differential diagnosis. And there are consulting firms that will tell you how to do rooftop solar power, perform all the cost calculations for you, jump the regulatory hurdles for you, and force the local electric utility to buy your surplus on the weekends. The brain’s a lot more like such a knowledge-based organization — the knowledge society rather than the industrial society.”
      “But surely there’s some management somewhere,” he concluded, pointing with his fork. “Even a modern organization of equals, like an orchestra or a consulting firm, has some managers. If decisions don’t get made, nothing gets done.”
      “Selective attention might at least qualify as middle management,” George said, grinning. “But if we were to concede a vice president, you’ll probably want to know where the president’s office is located.”
      George is kidding Neil about the rotating presidency of his new company. Neil had to give up running his new company when the seizures got worse a year ago, and his partners are anxious for him to come back full time and relieve them of the burden.
      “Well, let me tell you the big difference between being a vice president and a CEO,” Neil said, warming to his topic. “And it isn’t the size of the office or the best view. It isn’t even leadership, in the sense of getting people to follow your lead. The junior people all know how to do that, by the time that they get very far up the ladder.”
      He spread his hands wide. “But what chiefs have to do is to provide the direction. There’s no one else to follow anymore. It takes real imagination to be a chief exec. All those possible different directions that the company could go — you’ve got to explore them in preliminary ways, and think about how the company might look in five or ten years, if you followed that path. And then decide.”
      “Of course, that disappoints the vice presidents who planned the other alternatives. That’s why my partners and I jumped ship three years ago and started our own company. We imagined a path to some new products, saw a niche that surely was going to be filled by someone. And we had a head start on it.”

A SENSE OF THE FUTURE is perhaps the most special of the higher intellectual abilities. Imagine someone — and the brighter of the autistic kids would be a good example — who has the basics of language but cannot make mental models of someone else’s point of view. And so he interprets everything very literally — metaphor is beyond his abilities. He can’t imagine anything very well, and so even when he plays by himself, he’s not creative. Because he cannot imagine another person’s reaction, he gets into all sorts of trouble with social relationships. Top executives show you the other end of this spectrum.
      “Oh, I agree,” Neil said with mock modesty. “But you’ve probably got an old-fashioned idea about how top management works — it’s really more like a string quartet or jazz combo, where the leadership shifts around and has little to do with rank or title in a hierarchy. That’s why top management is usually called the ‘President’s Office’ in an American company, the ‘Vorstand’ in a German one.”
      “But what it’s really like is the partners in tennis doubles matches,” he continued. “A player may have a preferred position, close to the net or in the back court, but they cover for one another. And not just for the situation, like getting drawn out of position, but according to their relative strengths. A good back court player will start running to cover for the weak backhand of the partner at the net, even before the ball leaves the racket on the other side of the net.”
      A half century ago, Kenneth Craik proposed that the brain was making a small-scale working model of external reality and of possible actions. That the brain could thus try out various alternatives, conclude which was best by utilizing stored knowledge of somewhat analogous events of the past. By having imagined alterative actions beforehand, you can react in a much fuller, safer, and more competent manner to the emergencies that come along. And you can also react to future situations before they arise, as in making a business plan that anticipates reactions by your competitors.
      The brain’s ability to reorganize isn’t merely a matter of adapting to the loss of a piece of brain. It’s also a matter of reassigning space from minute to minute, of setting up optimal arrangements in the way that listening to the Mozart sonata might have done. Of being able to marshall your resources, as you concentrate on a difficult task. Or to relax and allow those subconscious thoughts to meander — and occasionally pop some relevant old memory into your stream of consciousness. Or come up with a newfound relationship between something you saw a few minutes ago and something old, buried in the recesses of memory, as when you finally recall someone’s name.
      “I like it — but you’re missing my point,” Neil complained. “Of course you can reorganize without central coordination. Happens all the time in a decentralized economy. But there’s got to be some management for other tasks, even if there isn’t a hierarchy.”
      He gestured upward with the palm of his hand. “What shifts my attention from the traffic light to looking around for another Maserati? From emphasizing concerns of areas on the outside of right frontal and parietal lobes, then switching over — now emphasizing the stuff in the cingulate gyrus of both sides, up front of the corpus callosum?”
      Maybe you’re doing all those things at once, in parallel.
      “It certainly seems as if I switch my attention from one thing to another. I’ve got a stream of consciousness, and only one thing happens at a time,” he emphasized. Then he paused for a moment, looking perplexed.
      “You’re telling me that my stream of consciousness seems to be one thing after another — but that’s because it is sampling, one after the other, the parallel activities of lots of areas in my brain?”
      Could be something like the way that the television newscast samples the activities of its part of the world. And ignores most of the world, most of the time. Whatever it reports is only a small part of what’s going on. And of course the newscast isn’t causing those activities, only shining a spotlight on them, the reporters and camera crews having been drawn by other reports.
      “So the one-mental-topic-after-another is effect, rather than cause?”
      To some extent, at least. The spotlight might, of course, influence an outcome by allowing a decision to be made sooner. Orchestrating media coverage is a big part of building a consensus in politics these days. Surely there are some pathologies of such systems, just as television crews are sometimes accused of causing riots to happen by their very presence — they can act as a catalyst. Conscious attention, focused on a problem, is also a catalyst.

THE STREAM OF CONSCIOUSNESS surely works in parallel with selective attention most of the time, but they need not be identical. Selective attention may bias the activities of the subconscious as well. Schizophrenics suffer intrusive thoughts or hallucinations despite their attempts to focus on the external world. For a lot of people, the train of thought probably corresponds to what could be talked about, if you chose to speak aloud.
      “You can obviously have a stream of consciousness without talking aloud,” George pointed out. “And it would appear that some people can talk without much “consciousness” — those frontal lobe patients, for example, and especially the Williams syndrome kids, that talk and talk but without much content to the sentences.”
      So perhaps the little person inside the head is just the stream of consciousness. A serial stream sampling parallel processes. And it, in turn, may be biased strongly toward the verbal happenings, leaving out all sorts of more mundane things like blood pressure.
      “Where does this leave Julian Jaynes?” Neil asked after a moment. “His theory that talking to ourselves wasn’t evolved until recent times, back between the writing of the Iliad and the Odyssey? Before then, we heard voices telling us to do things, afterward we narrated our life’s story ourselves. And our mental lives became considerably more lively.”
      Surely language, whenever it arose during prehuman evolution, set up a big increase in mental life. But did it occur so recently? All his evidence is from literary styles — which are notoriously prone to fads. And there’s an old saying in anthropology, that absence of evidence is not evidence of absence. Why assume a sudden jump if you don’t have to?
      “I suppose there just isn’t an answer to my question, about the management of knowledge, about what focusses attention? About how ideas compete?”
      No, but there are some testable theories. Since you insist, I suppose I’ll have to tell you about Darwin Machines.

DARWINIAN PROCESSES not only evolve new species of plants and animals, but also shape up new antibodies for the immune system. It may take millennia to get a new animal species, but it only takes a week or two for a new antibody to specialize in destroying a novel invading molecule. Shifting your attention, or evolving a new thought, may involve a similar process that, because things happen much more rapidly in the electrical machinations of neurons, takes only milliseconds to minutes.
      “All because of the survival of the fittest?” Neil asked skeptically.
      That’s a well-known aspect of darwinism, but a darwinian process has at least six essential features. First of all, you’ve got to have a pattern of some sort.
      “Like a gene? A string of DNA bases?”
      Right, and in the case of brains we’re probably talking about the Hebbian cell assembly, that spatiotemporal pattern of neuron activity that represents an apple or an orange. This cerebral code is quite arbitrary — as I said, like the bar code for representing apples on a can of fruit. What’s important about the pattern, so far as mental darwinism is concerned, is that copies can be made of it. For example, you can presumably send the code for comb over long distances in the brain, through the corpus callosum.
      “Of course, you aren’t literally sending the message,” George clarified, “in the manner of mailing something. What you do is to make a copy of it at the other end, just as when using a fax machine.”
      And the original pattern isn’t destroyed in the process of “sending” it. So if you feel the comb with your left hand but need to send a message from your right brain’s sensory hand map over to the left-brain language areas, you’ll need to make a distant copy of the comb spatiotemporal pattern. Transmission of information is one reason Hebbian cell assemblies have to be copied, but there are others too.
      “So there might be many spatial patterns of neuron firing in the brain — but only some of them are capable of making copies of themselves, somehow? But why else might you want to copy the pattern?”
      To produce variations on the pattern. Intentional errors, if you like. Variations over long distances can likely be minimized with the equivalent of an error-correcting code, and I’ve already discovered one way to do that. But sometimes you might want to turn off the error correction.
      “I assume you want errors for the same reasons that mutations are sometimes useful?”
      Small errors can be handy. That’s certainly how the immune system produces an antibody molecule that is an even better fit to the invader. When an antibody kills an invader, reproduction is stimulated. But it isn’t just a clone of the successful defender: it’s a whole spread of variations around that successful defender’s pattern. Some of which are even better fits to the invaders, and so reproduce more. And variations about that new, better pattern produce some variants that are better still.
      “By the time all the invaders are wiped out,” George explained, “the antibodies have gotten to be pretty good fits. And this type hangs around for years, ready to immediately counter a fresh invasion.”
      “That’s the immune response. But what happens in brain circuits that’s similar to that?”
      Suppose that a Hebbian pattern made a copy of itself right next door. And that, just the way a crystal grows, a whole territory develops where that basic pattern repeats, over and over. As in wallpaper.
      “Hundreds of copies of the Hebbian pattern. But why? The biological imperative? Because the vacant territory is there, waiting to be organized somehow.”
      That’s actually a good possibility. Most physical systems with a big throughput of energy tend to self-organize into patterns, so why should the electrical activity of the brain be an exception? Other patterns might be trying to organize the territory too, at the same time, so you might expect some competition.
      “The way that bluegrass and crabgrass are competing in my back yard, I suppose,” Neil said.
      Yes, and your backyard is another essential aspect of a darwinian process: there’s got to be a workspace of some sort for the competition.
      “But what determines who wins? Or when a standoff occurs?”
      For your backyard, it’s the many-faceted environment. Rainfall and watering are important. And the cropping schedule, how often you mow it. Sunshine and shade. Nutrients. Some DNA patterns in grasses reproduce better with one cluster of environmental variables than with others. The environment in the brain includes the current inputs and the memories stored in synaptic connectivity patterns, those resonances.
      So pattern, copying, variations, competition for a work space, and a multifaceted environment that biases the competition are five of the six essentials of a darwinian process. The sixth is to close the loop, having many repeats on the variation-and-selection steps. Just as in the immune response, there has to be a ‘next generation’ where the new variants are based on the more successful of the ones in the previous generation. Most of the new variants will be worse than their parents, but some will be better — and they are mostly the ones that will reproduce with yet again more variations.
      “So that’s what all the neural darwinism is about, those six essentials?”
      That’s what I mean by it, but the word ‘darwinism’ is used almost as loosely in science as it is by journalists. In most cases, all someone is trying to convey is that random variations are being shaped up by selective survival into a meaningful pattern — say, the way the synapses in the infant’s brain are being edited by environmental experience. It’s a carving process. There’s often no closing of the loop, that sixth essential. Or if there is, there’s no copying of patterns involved.
      Selective survival is what’s usually meant when someone speaks of darwinism — and it’s very important all by itself. But that’s a cardboard view of darwinism — the real thing goes several steps further. It’s a repeating process that produces even fancier results by using multiple generations, repeatedly copying with variation from parent patterns that were successful.
      “A real engine of progress, it sounds like. That’s a Darwin Machine then?”
      That name refers to the whole class of computing machines, each of which uses those six essentials of a darwinian process. I named it that on the analogy to Turing Machines. It’s no one particular machine or computer simulation, but the whole class. It even includes biological machinery — the immune response is a particular example of a Darwin Machine.
      And if the cerebral cortex can make copies of those Hebbian spatiotemporal patterns, and have competitions between them that are biased by resonance with memories, then some of our cognitive processes could be products of a Darwin Machine in our head.
      “I think I need an example,” Neil said, shaking his head.
      Suppose that something round went whizzing past us, I said, gesturing. And then disappeared under the tables, so you couldn’t verify your first impression — that it was something like a tennis ball. Or maybe an orange or an apple. How do you come up with candidates like that, and how do you eventually decide which one it was?
      “I don’t know, but I trust you’re going to tell me,” Neil said, smiling while he rubbed an itch amid his regrowing hair.

THE SENSORY IMPRESSION creates a committee of feature-detection neurons firing away, but their spatiotemporal pattern doesn’t really resonate with a stored pattern. Copies are made of the fuzzy-round-object pattern and, in the process, errors are made. Suppose an error gets the pattern close to the stored pattern for apple.
“I thought that stored patterns were purely spatial patterns. And that active patterns were spatiotemporal, more like a melody?”
      But the active pattern resonates with the stored pattern, just as your car resonates with a washboarded road surface to enhance the teeth jarring. That does two things: it brings forth the complete details of the stored pattern, even though you provided only an approximation. Once the cortical circuits start oscillating near the chaotic attractor — that’s a fancier way to describe the resonance — you start getting a more complete spatiotemporal pattern, the stored one replacing the one which prompted it.
      “The way that I can supply the details of someone’s face, even though I’m only looking at a cartoonist’s sketch.”
      Right, a ‘pop out’ of details. Second, the resonance makes copying easier, so that the resonant pattern makes copies next door. So you start to get a territory of the active apple pattern even in areas that don’t have the resonance from long-term synaptic patterns. Copycats.
      But when copying in some other direction in the cortical sheet, perhaps the fuzzy-round-object pattern made errors in the direction of the stored resonance for tennis balls. Now you have two candidates for the unknown object. And maybe a third or fourth comes along, too.
      “So now that you have candidates, I suppose that you can have a competition for the work space. Just like my bluegrass and crabgrass, fighting for my backyard. And may the best resonance win?”
      It’s not only the stored pattern that affects copying success, but also the present situation. If you’re in a cafeteria, you’re more likely to guess an orange or apple than you are a tennis ball. It’s a multifaceted environment that biases the competition and sets up the next generation of variants.
      That apple does better in occupying territory doesn’t mean that it wipes out the competition, any more than you ever get rid of all the crabgrass in your backyard. A little change in the environment for the competition — suppose someone says the fuzzy-round object bounced — and suddenly the boundary between apple and tennis ball starts shifting so that the latter soon becomes dominant. And therefore the best guess, the one most likely to be named as the object.
      “So is that the subconscious?” Neil asked. “The alternatives lingering in the background? When I say I’m conscious of its having been a tennis ball, I’m just reporting on the currently dominant pattern?”
      That’s the way I’d formulate it. And the experience of consciousness could reflect the winner, as regions from all over the brain compete with each other for dominance.
      “Something like the basketball championships? Where there are regional winners, and then quarter-finals, semi-finals, and finally a championship match?”
      Yes, but with the competition going on constantly, so that the “spotlight” of your “conscious awareness” shifts from one part of the brain to another, as the current winner changes around. The stream of consciousness, in this formulation, is just the history of the meandering spotlight. But, of course, it really isn’t a spotlight shown by someone else — it’s one region of the cortex lighting up because it has become temporarily dominant over other regions.
      “That’s why there’s no head office for consciousness,” Neil mused. “Also why no stroke eliminates it?”
      “A stroke might eliminate a customary concept like color from the stream of consciousness,” George noted, “but seldom the switching around that produces the stream.”

THERE ARE SURELY MIDDLE MANAGERS in the brain that help regulate competitions like that, just as climate change serves to stimulate evolution more generally. And I expect that they actually create fluctuations.
      “Management as the equivalent of climate change, did you say?” Neil was smiling. “They didn’t mention that in business school. So these managers try to sharpen the competition?”
      Yes. And that creates empty niches, where a local population goes extinct. That can create a golden opportunity for a variant that, at first, might not have been able to survive a head-to-head competition. These spatiotemporal patterns are ephemeral, the work space easily wiped clean by a little inhibition that momentarily quiets things down. Most of them aren’t around for long enough to change the synaptic strengths permanently and thus create a long-term memory.
      That’s one aspect, and fragmentation is another. In biology, the way to evolve new species quickly is in an archipelago like the Hawaiian islands. So management that segregates the cortex temporarily, to make the equivalent of lots of little islands, would be another good management strategy.
      “A little inconsistency as a good management tactic? Or budgets that fluctuate unpredictably? Wait until I tell my buddies!”
      In the brain, all you should have to do is to change the background level of excitation or inhibition, something that those widespread arousal and attention mechanisms ought to be good at. Just spread around some norepi or serotonin or dopamine or acetylcholine.
      That way you could create cycles of climate change, and cycles of fractionation and integration — the same way that a group of islands becomes one big island during an ice age when the sea level drops. The species get a period of isolation, to go their separate ways, and then a period of competition that results in one species becoming widespread. Then when sea level or excitability rises again, that winner becomes the basis of the new fragmentation.
      “Sounds like the trust-busters at work, breaking up the phone company. So what about the economic cycle? Does that fit your metaphor?”
      It’s a good example of how whole continents might be affected together. The economic cycle analogy might be to mood, like the manic-depressive cycling which some creative people exhibit. But suppose the neural equivalents of climate change were much more regional, like the rainfall coming off those Pacific Ocean storm systems shifting around for a few years to give California an edge over Oregon, but then it shifts back. In neural terms, that could be a management technique, both for giving the various areas a chance to show their stuff, but also to speed up evolution of new ideas — since they percolate better in the smaller copying competition work spaces.
      “I’m beginning to think you’re serious about this.”

SEQUENCES ARE THE REAL TEST of a concept like this, however, not individual item recognition and recall. With sequences, we keep track of the order of perceived events. But most importantly, we produce representations that are coded by sequence, such as sentences. We generate scenarios. Because of our sequential abilities, we experience a stream of consciousness that we can recall as a sequence, that we can mull over some more, and sometimes solve problems.
      The stream of consciousness that’s really important is the autobiographical one, the narrative we create — telling ourselves the story of our own life. We try to explain what’s happened to us, to get us where we are. We imagine scenarios of success, we worry about scenarios of failure or threat. We’re always poised at a choice between alternative futures. You can hear four-year-olds narrating things out loud, but thereafter most of us narrate our mental lives silently. We are narrators of our own inner lives, thanks to our sequential abilities.
      “So you want a cerebral code that’s good at sequences. Isn’t a sequence just a fancy form of category?”
      True, and there are some strange attractors in chaos theory that seem good at doing one spatiotemporal sequence for awhile, and then shifting to another spatiotemporal sequence. That Lorenz butterfly-shaped attractor is one example.
      But there are also some ways of maintaining stable sequences of cerebral codes, just by lining them up in order across the cerebral cortex in a work space. So that they could be scanned in the right sequence. The chunking limit, those seven plus or minus two items that you can hold in working memory, might correspond to the limits on such scanned mapping. And perisylvian cortex might be particularly good at maintaining such sequences. Or storing the long-term form of them in synaptic connectivity changes.
      “I don’t sense that an organization chart is going to describe this very well,” Neil said. “Not any more than you can meaningfully map the functions of a president’s office or a doubles team.”
      “Certainly not so far,” George responded. “Just remember that the history of cerebral mapmaking has been one of discovering more and more underlying specialization, that the ‘association cortex’ has proved to have many maps hidden in it. Remember we’ve seen a lot of specialization so far, neurons active during one task but not another. Your Darwin Machine may run in the cerebral cortex as if it were just an unspecialized work space, but there may be lots of specializations, anyway.”
      Right. On the side of flexibility are those higher-order visual neurons that change the size of their receptive field, depending on whether the monkey’s behavioral task is responding to a color or responding to shapes. In my theory, there is even more flexibility because of all the copying. The resonances for apple aren’t everywhere, even though copying can temporarily maintain the apple spatiotemporal pattern without them in many other places.
      “And that satisfies the other important constraint from the physiology,” George went on. “There are sites that are essential for a task such as naming or reading, and then there are many other sites that are merely active during the task — but they don’t seem to be essential, in the sense that the task works fine with them temporarily confused by the electrical buzz.”
      What my kind of dynamic reorganization theory suggests, of course, is that the region of cortex specializing in fruit — if there is one — probably spends 99 percent of its time helping out with nonfruit tasks. Something like a neurosurgeon who spends most of the time acting as a general practitioner.
      “That’s terrible,” George groaned. “A neurosurgeon who only operates 1 percent of the time is scarcely a neurosurgeon at all!”
      “So here we have,” observed Neil with a tour-guide’s expansiveness, “the traditional conflict in science. Between ‘context is everything’ on the one hand and ‘specialization is essential’ on the other — with, of course, its ‘practice makes perfect’ corollary.”
      The experimental neurophysiologists seek specialization, and they find it. And the theoretical neurophysiologists seek schemes for generality — dynamic reorganization and flexible assignments.
      “But have to find schemes that don’t conflict with all the known specialties. Right?”
      We’re very complementary. Any brain theory — or at least, any theory worth wading through — needs to be able to explain a lot of versatility, plus all this specificity we see. At a low level, it needs to explain those cerebral codes for apples and oranges, how Hebbian cell assemblies are created and stored — and how the full spatiotemporal pattern is recreated during recognition or recall.
      At a middle level, it needs to explain how categories are formed and how metaphor might work. It needs to explain why listening to Mozart might be a good setup for problem-solving tasks.
      At the high end, it needs to explain creative confusion, multiple levels of abstraction, making models of what might happen, and utilizing fancy sequential rules such as syntax and logic. Remind me to tell you about hexagons sometime.
      Now there’s no reason why the brain couldn’t have invented some processes fancier than darwinism. But it’s something of a default process in nature, simply because nature is so good at copying. At some level — and probably multiple ones — the brain is going to operate on darwinian principles.

PRODUCING NARRATIVES for one’s mental autobiography and career plan is certainly one of the better candidates for “Where’s the real me?” But, as Neil pointed out before leaving, a train of thought needs good nutrition.
      “I’m still going fishing in my stream of consciousness,” he explained. “I had a professor who used to say that you’ve got to feed a stream of consciousness with a variety of new facts, keep feeding it all the time — if you want to get something new and interesting to pop out, every now and then. You’ve got to keep reading, keep trying out ideas for size, keep rearranging them to find something better.”
      He resettled the baseball cap on his head. “That’s about the only thing that worried me about this operation, that my memory wouldn’t be good enough afterwards to keep it fed with new facts. Or to fish among the old ones.”
      “But so far, so good. I still have trouble with the Sunday crossword puzzle, but I’m getting faster and faster.”

Conversations with Neil's Brain:
The Neural Nature of Thought and Language
(Addison-Wesley, 1994), co-authored with my neurosurgeon colleague, George Ojemann. It's a tour of the human cerebral cortex, conducted from the operating room, and has been on the New Scientist bestseller list of science books. It is suitable for biology and cognitive neuroscience supplementary reading lists. ISBN 0-201-48337-8.
AVAILABILITY widespread (softcover, US$12).
The Darwinian thought story is continued in both 1996 books:
How Brains Think
(Science Masters), a Book of the Month Club Selection,
The Cerebral Code
(MIT Press)
are now in US bookstores.
Home Page || Science Surf magazine || Table of Contents || End Notes for this chapter || Continue to Postscript