May 22, 2014 | 2
Biologist Gerald Edelman–one of the truly great scientific characters I’ve encountered, whose work raised profound questions about the limits of science—has died. I interviewed Edelman in June 1992 at Rockefeller University in New York. Edelman subsequently left Rockefeller to head a center for neuroscience at the Scripps Institute in California. Edelman, 84, died in his home in La Jolla. The following is an edited version of my profile of Edelman in my 1996 book The End of Science.
Gerald Edelman Postures Around the Riddle
Gerald Edelman’s career, like that of his rival Francis Crick, has been eclectic, and highly successful. While still a graduate student, Edelman helped to determine the structure of a protein molecule crucial to the body’s immune response. In 1972 he shared a Nobel Prize for that work. Edelman moved on to developmental biology, the study of how a single fertilized cell becomes a full-fledged organism. He found a class of proteins, called cell adhesion molecules, thought to play an important role in embryonic development.
All this was merely prelude, however, to Edelman’s grand project of creating a theory of mind. Edelman has set forth his theory in three books: Neural Darwinism, The Remembered Present and Bright Air, Brilliant Fire. The gist of the theory is that just as environmental stresses select the fittest members of a species, so do inputs to the brain select groups of neurons–corresponding to useful memories, for example–by strengthening the connections between them.
Edelman’s tumescent ambition and personality have made him an alluring subject for journalists. A New Yorker profile called him “a dervish of motion, energy and raw intellect” who is “as much Henny Youngman as Einstein”; it mentioned that detractors consider him “an empire-building egomaniac.”
In a New York Times Magazine cover story in 1988, Edelman referred to himself as God. Twice. Discussing his work in immunology, he said that “before I came to it, there was darkness–afterwards there was light.” He called a robot based on his neural model his “creature” and said: “I can only observe it, like God. I look down on its world.”
I experienced Edelman’s self-regard firsthand when I visited him at Rockefeller University on a warm June day. Edelman is a large man. Clad in a dark, broad-shouldered suit, he exuded a kind of menacing elegance and geniality. As in his books, he kept interrupting the conversation to dispense stories, jokes or aphorisms whose relevance was often obscure. The digressions seemed intended to demonstrate that Edelman represented the ideal intellectual–both cerebral and earthy, learned and worldly. No mere experimentalist, he.
Explaining how he became interested in the mind, Edelman said: “I’m very excited by dark and romantic and open issues of science. I’m not averse to working on details, but pretty much only in the service of trying to address this issue of closure.”
Edelman wanted to find the answer to great questions. His Nobel prize-winning research on antibody structure had transformed immunology into “more or less a closed science”; the central question, which concerned how the immune system responds to invaders, was resolved. He and others helped to show that self-recognition happens through a process known as selection, which posits that the immune system has innumerable different antibodies; the presence of foreign antigens spurs the body to accelerate the production of, or “select,” antibodies specific to that antigen and to suppress the production of other antibodies.
Edelman’s search for open questions led him inexorably to the development and operation of the brain. He realized that a theory of the human mind would represent the ultimate closure for science, for then science could account for its own origin.
Consider string theory, Edelman said to me. Could string theory explain the existence of its chief proponent, Edward Witten? Obviously not. Most theories of physics relegated issues related to the mind to “philosophy or sheer speculation,” Edelman noted.
“You read that section of my book where Max Planck says we’ll never get this mystery of the universe because we are the mystery?” Edelman asked. “And Woody Allen said if I had my life to live over again I’d live in a delicatessen?”
Describing his approach to the mind, Edelman sounded, at first, as resolutely empirical as Crick. The mind, Edelman emphasized, can only be understood from a biological standpoint, not through physics or computer science or other approaches that ignore the structure of the brain.
“We will not have a deeply satisfactory brain theory unless we have a deeply satisfactory theory of neural anatomy, okay? It’s as simple as that.” To be sure, “functionalists” such as the artificial-intelligence maven Marvin Minsky say they can build an intelligent being without paying attention to anatomy. “My answer is, ‘When you show me, fine.’”
But as Edelman continued speaking, it became clear that, unlike Crick, he viewed the brain through the filter of his idiosyncratic obsessions and ambitions. He seemed to think that all his insights were totally original; no one had truly seen the brain before he had turned his attention to it. He noted that when he started studying the brain, or, rather, brains, he was immediately struck by their variability.
“It seemed to me very curious that the people who worked in neuroscience always talked about brains as if they were identical,” he said. “When you look at papers everybody talked about it as if it were a replicable machine. But when you actually look in depth, at every level–and there are amazing numbers of levels–the thing that really hits you is the diversity.”
Even identical twins, he remarked, show great differences in the organization of their neurons. These differences, far from being insignificant “noise,” are profoundly important. “It’s quite scary,” Edelman said. “That’s something you just can’t get around.”
The vast variability and complexity of the brain may be related to a problem that philosophers from Kant to Wittgenstein have wrestled with: how do we categorize things? Wittgenstein, Edelman elaborated, highlighted the troublesome nature of categories by pointing out that different games often have nothing in common but for fact that they are games.
“Typical Wittgenstein,” Edelman mused. “There is a kind of ostentation in his modesty. I don’t know what that is. He provokes you and it’s very powerful. It’s ambiguous, sometimes, and it’s not cute. It’s riddle, it’s posturing around the riddle.”
A little girl playing hopscotch, chess players, Swedish sailors doing naval exercises, rugby players are all playing games, Edelman continued. To most observers, these phenomena seem to have little or nothing to do with each other, and yet they are all members of the set of possible games.
“This defines what is known in the business as a polymorphous set. It’s a very hard thing. It means a set defined by neither necessary nor sufficient conditions. I can show you pictures of it in Neural Darwinism.”
Edelman grabbed his book off his table and flipped through it until he found an illustration of two sets of geometrical forms that represented polymorphous sets. He then pushed the book away and transfixed me once again. “I’m astonished that people don’t sit and put these things together,” Edelman said.
Edelman, of course, did put these things together: the polymorphous diversity of the brain allows it to respond to the polymorphous diversity of nature. The brain’s diversity is not irrelevant “noise” but is “the very basis on which selection is going to be made, when encountering an unknown set of physical correspondences in the world! All right? Well, that’s very promising. Let’s go one step further. Could the unit of selection be the neuron?”
No, because the neuron is too binary, inflexible. But groups of linked, interacting neurons could do the job. These groups compete in responding to the infinite variety of stimuli entering from the world. Groups that respond successfully grow still stronger while other groups wither.
Edelman continued asking and answering his own questions. He spoke slowly, portentously, as if trying to physically impress his words on my brain. How do these groups of linked neurons solve the problem of category that had troubled Kant and Wittgenstein? Through reentry. What is reentry?
“Reentry is the ongoing recursive signaling between mapped areas that are made by these first few properties, so that you map maps by massively parallel reciprocal connections. It’s not feedback, which is between two wires, in which I have a definite function, instruction–sine wave in, amplified sine wave out.”
He was grim, almost angry, as if I had suddenly become the symbol of all his puny-minded, envious critics, who said that reentry is merely feedback.
He paused a moment, as if to collect himself, and began again, speaking loudly, slowly, pausing between words, like a tourist trying to make a presumably dim native understand him.
Contrary to what his critics said, he continued, his model is unique; it has nothing in common with neural networks, he added, lacing the term “neural networks” with scorn. To gain his trust–and because it was true–I confessed that I had always found neural networks difficult to grasp. (Neural networks consist of artificial neurons, or switches, linked by connections of varying strength.)
Edelman smiled triumphantly. “Neural nets involve the stretching of a metaphor,” he said. “There is this yawning gap, and you say, ‘Is it me, or am I missing something?’” His model does not suffer from that problem.
I began asking another question about reentry, but Edelman held up his hand. It was time, he said, to tell me about his latest creature, Darwin 4. The best way to validate his theory would be to observe the behavior of neurons in a living animal, which is of course impossible. The only solution, Edelman said, is to construct an automaton that embodies the principles of reentrant loops. Edelman and his co-workers had built four robots, each named Darwin, each more sophisticated than the last. Indeed, Darwin 4, Edelman assured me, is not a robot at all but a “real creature.” It is “the first nonliving thing that truly learns. Okay?”
Again he paused, and I felt his evangelical fervor washing over me. He seemed to be trying to build a sense of drama, as if he were pulling back a succession of veils, each of which concealed a deeper mystery.
“Let’s go take a look,” he said.
We headed out of his office and down the hall. He opened the door to a room containing a huge, humming mainframe computer. This, Edelman assured me, is Darwin 4′s “brain.” Then we walked to another room, where the creature itself awaited us. A pile of machinery on wheels, it sat on a plywood stage littered with blue and red blocks. Perhaps sensing my disappointment–real robots will always disappoint anyone who has seen Star Wars–Edelman reiterated that Darwin 4 “looks like a robot but it’s not.”
Edelman pointed out the “snout,” a bar tipped with a light-sensitive sensor and a magnetic gripper. A television monitor mounted on one wall flashed some patterns that represented, Edelman informed me, the state of Darwin’s brain.
“When it does find an object it will poke up to it, it’ll grab it, and then it will get good or bad values… That will alter the diffuse relationships and synaptology of these things, which are brain maps”–he pointed at the television monitor–”that weaken or strengthen synapses that alter how muscles move.”
Edelman stared at Darwin 4, which remained stubbornly immobile. “Uh, it takes a fair amount of time,” he said, adding that “the amount of computation involved is hair-raising.”
Finally the robot stirred, to Edelman’s evident relief, and began rolling slowly around the platform, nudging blocks, leaving blue ones, picking up red ones with its magnetic “snout” and taking them over to a big box that Edelman called “home.”
Edelman gave me a running commentary. “Uh oh, it just moved its eye. It just found an object. It picked up an object. Now it’s going to search for home.”
What is its end goal? I asked.
“It has no end goals,” Edelman reminded me with a frown. “We have given it values. Blue is bad, red is good.” Values are general, goals specific.
When he was a teenager, Edelman elaborated, he desired Marilyn Monroe, but Marilyn Monroe was not his goal. He possessed values that led him to desire certain feminine properties, which Marilyn Monroe happened to exemplify.
Brutally repressing an upwelling image of Edelman and Marilyn Monroe, I asked how this robot differed from all the others programmed to learn. The difference, Edelman replied, his jaw setting, was that while Darwin 4 possessed values, or “instincts,” other robots needed specific “instructions” to accomplish any task.
But don’t all neural networks, I asked, eschew specific instructions for general learning programs?
Edelman frowned. “But all of those, you have to exclusively define the input and the output. That’s the big distinction. Isn’t this correct, Julio?” He turned to a dour young post-doc who had joined us and was listening to our conversation silently.
After a moment’s hesitation, Julio nodded. Edelman, with a broad smile, noted that most artificial-intelligence designers try to program knowledge in from the top down instead of having knowledge arise naturally from values. Take a dog, he said. Hunting dogs acquire their knowledge from a few basic instincts.
“That is more efficacious than any bunch of Harvard boys writing a program for swamps!” Edelman guffawed and glanced at Julio, who joined in uneasily.
But Darwin 4 is still a computer, I persisted, a robot running on a computer; Edelman was using language metaphorically when he called it a “creature” with a “brain.”
As I spoke, Edelman muttered, “Yup, all right, all right,” while nodding rapidly. If a computer, he said, is defined as something driven by algorithms, or effective procedures, then Darwin 4 is not a computer. True, computer scientists might program robots to do what Darwin 4 does. But they would just be faking biological behavior, while Darwin 4′s behavior is authentically biological.
If some random electronic glitch scrambles a line of code in his creature, Edelman informed me, “it’ll just correct like a wounded organism and it’ll go around again. I do that for the other one and it’ll drop dead in its tracks.”
Rather than pointing out that many programs have this resilience, I asked Edelman about the complaints of some scientists that they simply do not understand his theories.
Most genuinely new scientific theories, he replied, must overcome such resistance. He had invited those who have complained about the opacity of his writings–notably biologist Gunther Stent, who told the New York Times that he found Edelman incomprehensible–to visit him so he could explain his work in person. No one had accepted his offer.
“The opacity, I believe, is in the reception, not the transmission,” Edelman said. (Actually, Stent, whom I interviewed shortly before Edelman, had reached his decision about Edelman’s work after sitting next to him on a trans-Atlantic flight.)
By this time, Edelman was no longer trying to conceal his irritation. When I asked about his relationship with Francis Crick, Edelman abruptly announced that he had to attend an important meeting. He would leave me in the capable hands of his young colleagues.
“I have a very long term relationship with Francis, and that’s not something one can answer–Boom! Boom!–on the way out the door. Or, as Groucho Marx said, ‘Leave, and never darken my towels again!’” He departed on a wave of hollow laughter.
Edelman has admirers, notably the neurologist Oliver Sacks. Crick, on the other hand, spoke for many of his fellow neuroscientists when he accused Edelman of hiding “presentable” but not terribly original ideas behind a “smoke screen of jargon.”
Edelman’s Darwinian terminology, Crick told me, had less to do with any real analogies to Darwinian evolution than with rhetorical grandiosity. Crick suggested that Edelman’s theory, which he called “Neural Darwinism,” should be renamed “Neural Edelmanism.”
“The trouble with Gerry,” Crick said, is that “he tends to produce slogans and sort of waves them about without really paying attention to what other people are saying. So it’s really too much hype, is what one is complaining about.”
The philosopher Daniel Dennett of Tufts University visited Edelman’s laboratory and remained unimpressed. In a review of Edelman’s Bright Air, Brilliant Fire, Dennett argued that Edelman had merely presented rather crude versions of old ideas. Edelman’s denials notwithstanding, his model is a neural network, and reentry is feedback, according to Dennett.
Edelman also “misunderstands the philosophical issues he addresses at an elementary level,” Dennett asserted. Edelman may profess scorn for those who think the brain is a computer, but his use of a robot to “prove” his theory shows that he holds the same belief, Dennett explained.
Some critics accuse Edelman of deliberately trying to take credit for others’ ideas by wrapping them in his own idiosyncratic terminology. My own, somewhat more charitable interpretation is that Edelman has the brain of an empiricist and the heart of a romantic. He seemed to acknowledge as much, in his typically oblique way, when I asked if he thought science is in principle finite or infinite.
“I don’t know what that means,” he replied. “I know what it means when I say that a series in mathematics is finite or infinite. But I don’t know what it means to say that science is infinite. Example, okay? I’ll quote Wallace Stevens: Opus Posthumous. ‘In the very long run, even the truth doesn’t matter. The risk is taken.’”
Edelman added that Einstein, when asked whether science is exhausted, answered, “Possibly, but what’s the use of describing a Beethoven symphony in terms of air-pressure waves?” Einstein, Edelman explained, was suggesting that physics alone cannot address questions related to value, meaning, and other subjective phenomena.
One might respond, What is the use of describing a Beethoven symphony in terms of “reentrant neural loops”? How does the substitution of neurons for air-pressure waves or atoms or any physical phenomenon do justice to the magic and mystery of the mind?
Edelman cannot accept that, as Francis Crick put it, we are “nothing but a pack of neurons.” Edelman therefore obfuscates his basic neural theory–infusing it with terms and concepts borrowed from evolutionary biology, immunology and philosophy–in order to lend it added grandeur, resonance, mystique. He is like a novelist who risks obscurity–even seeks it–in the hopes of achieving a deeper truth.
But for Edelman, finally, the search for truth mattered more than the truth itself.
Photo: Anders Långberg, Wikimedia Commons, http://en.wikipedia.org/wiki/File:Professor_Gerald_M._Edelman.jpg