Top positive review
67 people found this helpful
Reconciling brain science and human concern: a timely addition to one of the most distinguished bodies of work in neuroscience
on May 5, 2007
Will knowing how the brain works--in particular, what consciousness is--transform our view of human knowledge itself? This is the question that looms large in Second Nature, Gerald Edelman's latest book. Though compact at 157 pages (excluding preface, footnotes, and index), this work represents Edelman's ambitious consideration of the implications of his view (likely the correct view) of the brain and mind for the broader world of human concern. Edelman seeks to understand the nature of knowledge as it is generated within a biological entity--the brain--that is shaped both by individual history and evolutionary forces. Astonishingly, in this little book, he succeeds in this quest marvelously. The result is no less than a new type of epistemology--what Edelman refers to as "brain-based epistemology."
Gerald Edelman is no mere dilettante or interloper in neuroscience. Since the publication of The Mindful Brain (a volume he co-edited and co-authored with Vernon Mountcastle) nearly thirty years ago, Edelman has diligently toiled in the theoretical vineyards to construct a comprehensive theory of higher brain function that is consistent with the latest available neuroanatomical, neurophysiological, and behavioral data. Perhaps the most significant fruit of these labors, the Theory of Neuronal Group Selection, or Neural Darwinism, proposes that, during neurogenesis, a vast "primary repertoire" of physically connected populations of neurons arises. Later, in a process akin to Darwinian selection, a "secondary repertoire" of functionally defined neuronal groups emerges as the animal experiences its world, and that world in turn selects patterns of connectivity (the so-called neuronal groups) that provide a good enough fit in a given moment to engender some kind of positive outcome. Underlying this selection is a neural "value system," established over the course of evolution and believed to comprise small populations of neurons within deep brain structures, that assigns salience to particular stimuli encountered by the animal. When the response to a given stimulus leads to a positive outcome (i.e., eating satisfies hunger), the value system will reinforce, or strengthen, those synaptic connections between neurons that happened to be firing at that particular moment. There is now a greater likelihood that, when the animal encounters similar stimuli in the future, many of the same neurons that fired the first time will fire together again. When a stimulus is noxious, the value system will similarly strengthen the connections between neurons that happened to be firing at the time the stimulus was encountered, thus increasing the salience of that stimulus. When a stimulus has no salience, synaptic connections between neurons that fired upon first exposure to that stimulus will become weaker with successive exposures. Simply stated, neurons that fire together wire together. Keep in mind that the mapping of the world to neural substrate is degenerate; that is, no two neuronal groups or maps are the same, either structurally or functionally. Nor are the populations of neuronal groups that make up the neural mappings of the world exactly the same each time similar stimuli are encountered. These maps are dynamic, and their borders shift with experience. And finally, since each individual has a unique (and privileged) history, no two individuals will express the same neural mappings of the world. Indeed, from the establishment of the primary repertoire during development, no two brains are wired in exactly the same way, not even those of identical twins.
Notwithstanding any of the various attempts at historical revisionism that you may have encountered if you've read broadly across neuroscience and the philosophy of mind, the selectionist view of the nervous system begins with Edelman's highly original work. What follows from others making selectionist arguments is (whether they like it or not) purely derivative.
Although Edelman's theoretical framework has expanded to include the Dynamic Core hypothesis, a proposed mechanism for consciousness (See Edelman and Tononi's A Universe of Consciousness) that he discusses throughout Second Nature (and I will not unpack here), I believe that Neural Darwinism is his most fundamental contribution to modern neuroscience. To this day, it remains the most detailed and comprehensive theory of higher brain function ever proposed. Perhaps most importantly, and likely to the great consternation of those critics capable of lobbing only ad hominem attacks at Edelman himself, the theory is, in the best traditions of empirically grounded science, eminently testable. I have laid out a brief (and wholly inadequate) sketch of Neural Darwinism here because many of the critiques of Edelman's work are colored either by misapprehensions about this theory or the unrealistic expectation that its underlying mechanism can and should be easily described and readily digested. But unless you can appreciate the vast complexity of a biology shaped by evolutionary principles that are not well understood by the lay public (or even some scientists, for that matter), you will probably struggle to understand much of what Edelman has to say, even in this little book. The fault lies not in Edelman's prose, but rather in the nature of the subject matter he seeks to describe (contrary to the complaints of a few critics--see below). Persevere; if you love biology, are fascinated by the mysteries of the brain, and are curious about the implications of modern brain science for the nature of human knowledge and endeavors, then this book should be your touchstone.
I'm not going to give you a detailed rundown of the contents of Second Nature here; I'll simply recommend that you read it. In the remaining paragraphs, I hope to provide you with something I think will be of even greater value: a discussion of some of the most commonly raised criticisms of Edelman and his work. I hope that this will allow you to read the book--if not totally free of misconceptions--at least less encumbered by what I believe to be unfair attacks on one of the most constructive and distinguished bodies of work in modern theoretical neuroscience.
It is curious that Edelman's work engenders as much vitriolic reaction as it does. If you've read my review up to this point, you've certainly concluded that I'm firmly in Edelman's camp. That said, what follows are the most common claims about Edelman and his ideas from his most vocal critics. These can be clearly stated and quite easily dispensed with. In no particular order, here they are:
1) There is nothing original in his ideas.
2) Natural selection is not an apt analogy for what the brain does.
3) His models are instantiated on computers even though he claims that the brain is not a computer (look up the review by George Johnson).
4) He doesn't understand, or mischaracterizes, the views of modern philosophers.
5) He denigrates philosophers and their work.
6) He omits the work of others.
7) He doesn't communicate his ideas effectively, i.e., he does not write clearly or well.
Now, my rejoinders to the above claims:
Claim #1: Quite simply, those who make this claim need to practice better scholarship. Edelman first suggested the idea of neuronal group selection nearly thirty years ago. Back in the late 1970s, no one else in neuroscience ventured any such selectionist ideas. Moreover, early on, Edelman took quite a lot of heat for this notion. His transition from immunology to neuroscience, though logical from a theoretical perspective (moving from one selectional domain to another), may have offended stalwarts of the neuroscientific establishment. In any case, later, when the evidence suggested that Edelman was indeed correct about competition among groups of neurons (see, for example, the work of M. Merzenich), the attitude of many within and outside of neurobiology was something along the lines of, "oh yeah, but of course there are competitive interactions between functional neuronal assemblies; everybody knows that!" Well, clearly not everybody, and certainly not back in 1978. Over three decades, an original idea had thus been unfairly relegated to derivative status. It wasn't derivative; it was the source.
Claim #2: There is much evidence to suggest that neural representations of the world are dynamic and based on the competitive interactions between functionally defined and degenerate (e.g., non-identical) groups of neurons. Many alternative views of the central nervous system (CNS) have invoked formal computational principles. But everything we know about the CNS suggests that it functions nothing like a computer. If it were a sort of Turing machine, it would represent the only such example known to biology. Most modern biologists steeped in evolutionary principles (whether strictly Darwinian or of the Punctuated Equilibrium variety championed by Stephen Jay Gould and Niles Eldredge) would probably balk, first, at the notion of the emergence of organized populations of cells (or proteins or molecules, for that matter) capable of executing computations in the same manner as a digital computer, and second, at the idea that this sort of arrangement, if it had appeared at all, would have appeared only once over the course of evolutionary history. Finally, a challenge to those who too easily dismiss Edelman's Theory of Neuronal Group Selection and all that has followed from it: Go ahead and TRY to formulate a detailed, testable theory of brain function that takes account of the underlying biology of the central nervous system. Any takers? No? Enough said.
Claim #3: A number of Edelman's critics, such as the science writer George Johnson (Miss Leavitt's Stars), see little distinction between Edelman's characterization of the workings of the brain and computation-based information processing. But there is one profound difference. In selection-based systems such as the immune system or the CNS, meaning or "information" is imposed from within; in instruction-based systems such as digital computing, meaning is imposed from without; there is no internal meaning--a lot of lights may be on, but nobody's home. Often, traditional digital computers fail in tasks that involve discriminating novelty in a changing environment or generalizing across categories; brains excel in such tasks. But brains built like computers would be neither flexible nor adaptive. Moreover, a computer built like a brain, with little or no specific point-to-point wiring, would not be a functional computer. Precise instructions could not be implemented on such a machine in the absence of point-to-point wiring.
Some critics perceive something of a contradiction in the fact that, while Edelman has strongly rejected the notion of brain-as-computer, he and his colleagues have created simulations of the brain using massively powerful supercomputers. This point is either a red herring or simply represents a woeful ignorance of the nature of computer-based modeling and its applications in biology. When one models biological structures and their interactions on a computer--whether these are proteins folding a certain way, bones reacting to mechanical forces, or brains that can interact with, and adapt to, a world of novelty--one essentially uses software to approximate the analog and not infrequently stochastic behaviors of elements within the biological system being modeled. So, in the case of a biologically based brain simulation, the software instantiates on the computer a functional approximation of neurons with firing thresholds which shift in a circuit interaction- and context-driven fashion. The computer's overt behavior--or that of the device it controls--is not binary when this software is being run. The computer--or more properly, the simulation running on it--does not behave like a classical Turing machine. Why is this so hard to understand?
Claim #4: Actually, Edelman's descriptions and characterizations of various philosophical stances are generally detailed and accurate, and show a depth of understanding that could only have come from a thorough and voracious reading of much of philosophy, not just the philosophy of mind. Edelman has obviously taken in and "gets" the bulk of what philosophers have to say about the nature of knowledge.
Claim #5: In fact, I think Edelman has pulled his punches when it comes to taking on modern philosophical approaches to brain, mind, and the nature of knowledge. Although in his review of Second Nature, David Papineau clearly took offense at Edelman's characterization of philosophical approaches to epistemology as "armchair operations" (Nature, 2007, 446(5):614-615), it is not at all clear that Edelman actually meant this as an attack. When he makes this statement, though, I think he is clearly on the right track; he just doesn't follow that track far enough. Like it or not, these are armchair operations, and few philosophers have ventured beyond such musings to explore the actual neural substrate that generates knowledge in the first place. There are notable exceptions; the efforts of some philosophers, including most prominently John Searle, Hilary Putnam, Ned Block, and Thomas Metzinger, demonstrate a truly deliberative and concerted effort to incorporate what is known about the biology of the brain into thinking about the nature of cognition in general, consciousness in particular, and human knowledge. But many modern philosophers, I believe, are not merely armchair theoreticians; they are intellectually lazy. They think that, when considering the nature of mental processes, it is actually possible to do an "end-run" around neurobiology. Why bother actually relating organic structure and function to cognition? This stance is, quite simply, bizarre; it seems to be akin to a sort of a holdout syncretism of the ideas of Fodor and Skinner. Whatever the roots of this particular philosophical strain, it is wrong and intellectually dishonest. That Edelman has never actually expressed this in print I can only ascribe to some sense of old school propriety and intellectual fairplay. Would that his critics could exercise the same measure of propriety and fairplay.
Claim #6: Nothing obligates Edelman to give a précis of the state of the art of all of theoretical neuroscience (such as it is), particularly in such a compact book. In his review of Second Nature, David Papineau takes Edelman to task for the absence of "scientific comparisons" and suggests that "[a] naïve reader could easily form the impression that Edelman and his associates are the only people trying to use scientific information to cast light on the human mind." (p.615) Well, this is a rather silly point, as a book of this size is clearly not intended to serve as a reference text or primer. Moreover, had any other neuroscientists actually offered competing comprehensive and testable theories of higher brain function and/or consciousness, I have no doubt that Edelman would have felt obliged to take full account of these in Second Nature. So far, they haven't. David Papineau offers that the book presents a senior scientist's "potted cultural history." (p.615) For what it's worth, I eagerly await Prof. Papineau`s version of the cultural history of the science and philosophy of mind. What would such a [presumably] unexpurgated historical landscape actually look like, Prof. Papineau, and precisely who and what, in the way of deep theoreticians and theory, would populate this landscape? Offer some examples and I might even relent and recant my denigration of your odd and useless proclamation.
Claim #7: This is a very old criticism, dating way back to the publication of Neural Darwinism in 1987. In his thick body of work, Edelman has tried to explain nothing less than the workings of the most complicated object in the known universe. Moreover, early on, he attempted this at a time when there were no commonly accepted terms for the interactions he sought to describe ("reentry" and "degeneracy" are examples of terms Edelman coined more than twenty years ago to describe phenomena and properties not previously recognized by neuroscientists). There are many biological properties, principles, and concepts that, by their nature, don't lend themselves to simple descriptions or easy explanations. In Second Nature, Edelman's prose and its organization are clear and amazingly methodical for such a brief book. The book is densely packed, and the subject matter is obviously difficult. Unlike some philosophers, whose abstractions of cognitive properties resemble nothing more or less than a functionalist's "black-box," offering [biologically] context-free and meaningless thought experiments and little depth or intellectual satisfaction, Edelman has gone to great pains in previous works to describe very complex neural properties in the clearest possible manner. With Second Nature, he has taken on the additional task of reconciling his view of brain function--specifically consciousness, that most mysterious of all neural processes--with the nature of human knowledge itself. Edelman addresses the question of whether the highest expressions of human concern--creative pursuits such as art, poetry, and music, or the ethical and moral codes that glue human societies together--can ultimately be " . . . reduced to a series of epigenetic rules of brain action." (p. 156) Unlike Patricia Churchland, Edelman is not a reductionist, so his answer to this question is a resounding "no." (playing or listening to the Chaconne from Bach's Partita #2 cannot be boiled down to an orderly, reproducible code of neuronal firing; and, contrary to the view offered in Churchland's Neurophilosophy, the terms that refer to complex neural function will not simply fall away as neural mechanism reduces to the description of the electrochemical properties of firing neurons; nor, finally, will consciousness come down to the subatomic states of microtubules, as Penrose has suggested). But within the subtext of Second Nature is another, very provocative, question that few before have posed in earnest: Would knowing how the brain works down to the finest detail fundamentally alter the nature of human concern? Although he offers no explicit answer to this question, I suspect Edelman's answer would be "probably not."
So, in sum, don't be put off by the acerbic musings of Edelman's critics (or the length of this review); go ahead and read Second Nature. It may change profoundly your perspective on the nature of human knowledge and its ultimate creator and locus, the human brain.