Every once in a while, former computing and information science dean Bob Constable will have what he calls a “Hawkins event.” He’ll reach for his keychain, which contains two nearly identical keys, and inadvertently stick the wrong one into his front door lock—then realize his mistake with a flash of alarm. Or he’ll be walking on an icy sidewalk and strive to stay upright by putting one foot carefully in front of the other—something humans don’t generally worry about after toddlerhood.
The common denominator is that the normal suddenly becomes abnormal. The world we expect to encounter after years of experience—how a key slides into a lock, or how it feels to walk—suddenly throws us for a loop. Something we’ve long done on autopilot requires our active attention.
Constable’s nickname for such incidents is in honor of Jeff Hawkins ’79, the pioneering tech entrepreneur who has devoted considerable amounts of his time, energy, and fortune to understanding the human brain. Trained as an electrical engineer, Hawkins is a Silicon Valley legend: inventor of the Palm Pilot, founder of Palm Computing and Handspring, creator of the Treo line of smartphones. But in addition to mobile computing, Hawkins has another passion: he has itched to understand how the brain works ever since he was an adolescent who went to his local library looking for the definitive book on the subject and came up empty. Over the course of decades, he has developed a working theory of the brain—specifically its neocortex, which makes up 80 percent of its volume—as a system that makes sense of the world via memory, prediction, and pattern-recognition. (Hence the “Hawkins event.”)
Though Hawkins may not have a PhD in neuroscience, he’s nobody’s idea of a dilettante; rather, he’s what an earlier age might have called a gentleman scholar. Grown wealthy through his business ventures, he’s using his resources to study a subject that has long fascinated him. “I felt that this was a grand challenge,” he says, “and one that could sustain interest for a lifetime.” And as Hawkins himself happily admits, he’s a man on a mission: he wants both scientists and laypeople to get excited about figuring out what’s going on between their ears. “Understanding the brain and how it functions is to me, and to many people, probably one of the greatest scientific and intellectual quests,” Hawkins says. “Everything we’ve ever done—our art, music, culture, politics, science—are all products of our brain. To understand who we are as humans, you have to understand how our brains work.”
For a long time, Hawkins was something of a voice in the wilderness on the subject. Early in his career, he tried to pursue study of how the brain works—both as a fundamental question and for its potential real-world applications in computer science and engineering—but got the cold shoulder from academia and industry. As a result, Hawkins is largely self-taught; he devoured dense papers on neuroscience even while building successful tech companies. “He’s one of those people who’s smart in a really scary way; he knows so much about such a wide range of fields, it’s mind-boggling,” says Subutai Ahmad ’86, a senior employee at Hawkins’s current company, Numenta. “He knows the neuroscience extremely well—probably better than most neuroscientists I’ve spoken with. He has a deep understanding of computer science, and of how to create new businesses and successful products. There aren’t too many people who can be a keynote speaker at the Society for Neuroscience as well as at a business conference.”
In 2002, Hawkins founded the Redwood Neuroscience Institute, a nonprofit research center devoted to exploring how the neocortex works (it later became part of the University of California, Berkeley). And in 2005, in an effort to spread the word about his theories, Hawkins published a widely respected book on the subject: On Intelligence: How a New Understanding of the Brain Will Lead to the Creation of Truly Intelligent Machines. “In an understated way, he’s a compelling visionary,” says Cornell computer science professor Ramin Zabih, who has visited Redwood. “You can see how he’s succeeded in his commercial ventures. He has a gift for conveying enthusiasm about what he’s interested in. He’s a big-picture thinker.” Or, as Nobel laureate James Watson has said of Hawkins: “He has changed the lives of many people—and I have a feeling he’s going to change the lives of many neuroscientists who don’t yet think like him.”
Co-authored with New York Times science writer Sandra Blakeslee (but written in the first person), On Intelligence offers a detailed account of Hawkins’s take on how we learn. Though the book does contain its share of technical jargon—Hawkins warns nonscientists to take a deep breath before plunging into Chapter Six: “How the Cortex Works”—he intended it for a broad audience. “I am crazy about brains,” Hawkins writes on the first page. “I want to understand how the brain works, not just from a philosophical perspective, not just in a general way, but in a detailed nuts and bolts engineering way. My desire is not only to understand what intelligence is and how the brain works, but how to build machines that work the same way.”
Among the early readers of On Intelligence was Constable, who offered feedback during the writing process. He notes that just as Hawkins has a knack for starting companies, he was able to see that there was a gaping hole in the field of neuroscience—that no one was trying to present a complete model of how the brain works. “He’s an extremely creative and smart person, but humble; great ideas, very imaginative,” Constable says. “I’d take notes at his lectures at Cornell or in Silicon Valley, because he always said something about the brain that was worth carrying around.”
These days, Hawkins isn’t alone in believing that the topic is a societal imperative. In early April, the White House unveiled the BRAIN (Brain Research through Advancing Innovative Neurotechnologies) Initiative, which it called “a bold new research initiative designed to revolutionize our understanding of the human brain.” To be launched with a planned $100 million in initial federal funding, the effort—described as neuroscience’s answer to the Human Genome Project in scope and potential economic payoff—will involve myriad public and private sector organizations including the NIH, DARPA, the NSF, and the Howard Hughes Medical Institute; its academic arm is co-chaired by officials from Stanford and the Rockefeller University. “History will mark humanity in two epochs, before and after understanding how brains work, or before and after machine intelligence,” Hawkins wrote in a blog post when the project was first announced in February. “National focus and money on this important problem makes sense and will accelerate progress.”
Although Hawkins is interested in how the entire brain works, he has focused on the neocortex—evolutionarily, the newest part of the brain, and the center of intelligence. Containing some 300 trillion synapses (the connections between neurons) the neocortex is a mass of scrunched-up, wavy matter that sits atop the evolutionarily ancient “old brain,” which governs such fundamental things as blood pressure, hunger, sex, and emotion. If spread out, the neocortex would be two millimeters thick and about a foot and a half square—roughly the dimensions of a cloth dinner napkin, which Hawkins has been known to bring to lectures as a visual aid. “The seat of intelligence is the neocortex,” he writes in On Intelligence. “Even though it has a great number of abilities and powerful flexibility, the neocortex is surprisingly regular in its structural details. The different parts of the neocortex, whether they are responsible for vision, hearing, touch, or language, all work on the same principles. The key to understanding the neocortex is understanding these common principles and, in particular, its hierarchical structure.”
So what is Hawkins’s theory? In a nutshell, it’s all about memory and prediction. As Hawkins describes it, the brain stores knowledge in a layered hierarchy: at the top is new information, while way at the bottom are tasks and facts that have become second nature. As humans learn, we’re constantly taking in new information, comparing it to what we already know, and using that knowledge to predict the immediate future. When you’re learning to drive, for example, the task requires all of your concentration—changing lanes, observing traffic signals, keeping a steady speed. But once you’ve got thousands of miles on your odometer, you can drive for chunks of time without really thinking about it.
The power of prediction also plays an important role in Hawkins’s premise: each time you take a step, as in Constable’s example, you unconsciously anticipate what a footfall will feel like. Similarly, once you learn to clap your hands, you don’t think about it anymore; your brain has expectations about what clapping entails. “I expected my hands would stop and not go through each other,” Hawkins told an audience at the New York Academy of Sciences during his book tour for On Intelligence. “I expected they wouldn’t turn into potatoes. They’d make a specific sound”—he clapped again to demonstrate—”and not squeal like a pig.”
But when something unexpected happens—a car suddenly cuts you off, say, or you sit in a chair and it wobbles badly—your brain jumps back up to that top level of awareness and information gathering. “Brains are not computers,” Hawkins observes. “Brains are learning and memory systems. We’re not programmed to do things; we don’t know about the world when we’re born, we have to learn about it. For the most part, in the beginning of the artificial intelligence era, people ignored the learning part. They tried to codify knowledge—write it down and put it into something. That was the fundamental flaw. They’d make progress in certain areas, like a computer that can play chess, and people said, ‘That’s great; we’re on our way.’ But we really weren’t.”
Another basic tenet of Hawkins’s theory is that the brain is not a static organ: it functions across time. In his book, he asks readers to imagine their homes, recalling various details. What does the front door look like? What items do you keep in your shower? “You might say these things are all part of the memory of your home,” he writes. “But you can’t think of them all at once. They are obviously related memories, but there is no way you can bring to mind all of this detail at once. You have a thorough memory of your home; but to recall it you have to go through it in sequential segments, in much the same way as you experience it.”
Hawkins grew up on the North Shore of Long Island, in a family enamored of building things—boats in particular. Although he says he wasn’t a “science kid,” he was deeply curious and a voracious reader. At fifteen, he drew up a list of four fundamental questions about the world. They went something like this:
1. Why does the universe exist?
2. Why does it have the laws that it does?
3. What is life and where did it come from?
4. What is intelligence?
In the interest of addressing question number four, he did what he always did: went to the library and looked for a book on the subject. “As a teenager I had become accustomed to being able to find well-written books that explained almost any topic of interest,” he writes. “There were books on relativity theory, black holes, magic, and mathematics—whatever I was fascinated with at the moment. Yet my search for a satisfying brain book turned up empty. I came to realize that no one had any idea how the brain actually worked. There weren’t even any bad or unproven theories; there simply were none.”
Fast-forward to September 1979 and the dawn of the microcomputer age. After graduating from Cornell, Hawkins had taken a job with Intel in Portland, Oregon. That month, Scientific American published a special issue dedicated to the brain—a volume that remains beloved among many neuroscientists of his generation. In addition to addressing such topics as brain organization, development, and chemistry, the issue included an essay by Francis Crick, Watson’s partner in discovering the double-helix structure of DNA. In the piece, entitled “Thinking About the Brain,” Crick noted that despite all the accumulated knowledge about the organ, its workings remained a profound mystery. For Hawkins, Crick’s piece was a rallying cry. “He was like the boy pointing to the emperor with no clothes,” Hawkins writes. “According to Crick, neuroscience was a lot of data without a theory. His exact words were: ‘what is conspicuously lacking is a broad framework of ideas.’ To me this was the British gentleman’s way of saying, ‘We don’t have a clue how this thing works.’ “
While at Intel, Hawkins tried to convince his bosses that a greater understanding of the brain could lead to the development of better microprocessors; the company took a pass. After transferring to Intel’s Boston campus to be near his future wife, Hawkins applied to MIT’s artificial intelligence lab, offering the same proposition. A.I. was all the rage—but its researchers were looking to surpass the human brain, not emulate it. Hawkins’s application was rejected.
Hawkins eventually relocated to Northern California, where he went to work for GRiD Systems, designer of the first laptop computer. Still compelled to study the brain, he quit his job and enrolled as a full-time grad student in Berkeley’s biophysics program. He studied there for two years in the mid-Eighties, but left when his thesis proposal on neural pattern recognition was rejected—because there was no one on the faculty studying that topic, and therefore no one Hawkins could work under. (As Watson has joked: “At Berkeley, he found out that his professors weren’t that bright.”) Hawkins returned to GRiD, becoming vice president for research and pioneering the GRiDPAD, the progenitor of the Palm Pilot. After licensing the technology, Hawkins founded Palm in 1992.
Although Hawkins’s brain research wasn’t an active part of developing the Palm, it did play a role. In struggling to solve the problem of handwriting recognition—which so bedeviled Apple’s failed Newton—it occurred to him that people are happy to learn new skills, such as touch-typing, as long as their tools work as expected. “It was about predictability, not writing style,” he says. “If a tool is predictable then people will like it, even if you have to learn something new. But most people thought the idea of a new alphabet was really stupid.” The result of Hawkins’s stupid idea: Graffiti, the easy-to-master interface that changed the way people manage personal information.
In 2005, Hawkins launched his latest commercial venture: Numenta, founded with longtime business partner Donna Dubinsky. Named after the Latin word for mind (“mentis”), the company develops technology stemming from his neuroscience research. “We’re trying to understand how the brain works from a computational perspective and translate that into software algorithms that can operate under the same principles,” says Ahmad, Numenta’s vice president of engineering and one of its earliest hires. “We always go back to the neuroscience when we come up against specific problems. If we’re butting our heads against something, sometimes for months or years, we go back to the biology to see, ‘How was this problem solved in the brain and how can we translate that into a computer program?’ “
Numenta’s early clients include EnerNOC, a Boston-area designer of smart-energy systems; Numenta is adapting its software, dubbed Grok, to help the company project energy demand on a minute-to-minute basis. Numenta has also created software to aid a wind energy company in scheduling preventive maintenance; other potential applications range from financial services—a field perennially needing ways to predict market performance—to anticipating inventory demands in retail stores. “The way the world is now, we’re generating data at a huge rate,” says Ahmad, who holds a PhD in computational neuroscience from the University of Illinois, Urbana-Champaign. “Phones, computers, cars, cameras—there are streams of sensory information coming in everywhere. To have an intelligent system, it must be able to take that information and act on it contextually, to look at patterns and make predictions.”
While Hawkins’s company—located in Redwood City, halfway between Palo Alto and San Francisco—is eight years old, he and Ahmad say that in many ways it’s still in its early days. Even when it was founded, it didn’t have a start-up mentality. “Most start-ups are very short-term focused,” Ahmad says. “Their investors typically want to see returns quickly, within a couple of years, or they’re not going to fund you again. Numenta has been different in that we’ve taken a long-term approach. We spent several years focusing on the fundamentals of the algorithm, and Jeff has been extremely patient. We’ve purposely not gone toward venture capital funding, because we know that what we’re doing is going to have a long-term impact, and we want to make sure the fundamentals are solid.” Almost all of Numenta’s two dozen employees work on one big, open floor; Hawkins has no private office, sharing space with the CEO. “My favorite place to work,” he says, “is either sitting in the lounge or chatting with people in a conference room.”
Ultimately, Hawkins says, his model of brain function holds promise for addressing a variety of long-standing tech challenges—not just in artificial intelligence, but also in computer vision, robotics, and more. For former Engineering dean Dan Huttenlocher, such real-world applications are essential to evaluating the worth of Hawkins’s theories. “To me, as a computer scientist, the proof is what these systems do, how they work for solving problems,” he says. “The fact that Jeff and his team are producing software that seems to be very useful for clients who have particularly challenging data-mining problems is exciting, because it ties the theoretical framework that he’s developed to practical applications that are really challenging.”
Huttenlocher left the Engineering deanship last year to lead the University’s nascent technical campus in New York City. That project, known as Cornell Tech, has dual academic and economic aims: Mayor Michael Bloomberg conceived it as a way to drive the city economy by uniting researchers with companies and investors. Huttenlocher notes that Hawkins, as a successful entrepreneur who has long married the practical and the theoretical, exemplifies what Cornell Tech hopes to accomplish. “I think of Jeff as one of the great visionaries in the technology field of the last twenty-five years or so—he towers above a lot of other people in envisioning things that are really new,” Huttenlocher says. “Think back to what the personal digital assistant or smartphone looked like in the time of Palm and Handspring. They didn’t exist; it was a total reconceptualization. He’s a guy who comes at things and completely redefines them.”
Think About It
In On Intelligence, Hawkins outlines his quest
The question of intelligence is the last great terrestrial frontier of science. Most big scientific questions involve the very small, the very large, or events that occurred billions of years ago. But everyone has a brain. You are your brain. If you want to understand why you feel the way you do, how you perceive the world, why you make mistakes, how you are able to be creative, why music and art are inspiring, indeed what it is to be human, then you need to understand the brain. In addition, a successful theory of intelligence and brain function will have large societal benefits, and not just in helping us cure brain-related diseases. We will be able to build genuinely intelligent machines, although they won’t be anything like the robots of popular fiction and computer science fantasy. Rather, intelligent machines will arise from a new set of principles about the nature of intelligence. As such, they will help us accelerate our knowledge of the world, help us explore the universe, and make the world safer. And along the way, a large industry will be created.
Fortunately, we live at a time when the problem of understanding intelligence can be solved. Our generation has access to a mountain of data about the brain, collected over hundreds of years, and the rate at which we are gathering more data is accelerating. The United States alone has thousands of neuroscientists. Yet we have no productive theories about what intelligence is or how the brain works as a whole. Most neurobiologists don’t think much about overall theories of the brain because they’re engrossed in doing experiments to collect more data about the brain’s many subsystems. And although legions of computer programmers have tried to make computers intelligent, they have failed. I believe they will continue to fail as long as they keep ignoring the differences between computers and brains.
What then is intelligence such that brains have it but computers don’t? Why can a six-year-old hop gracefully from rock to rock in a streambed while the most advanced robots of our time are lumbering zombies? Why are three-year-olds already well on their way to mastering language while computers can’t, despite half a century of programmers’ best efforts? Why can you tell a cat from a dog in a fraction of a second while a supercomputer cannot make the distinction at all? These are great mysteries waiting for an answer. We have plenty of clues; what we need now are a few critical insights.
From On Intelligence: How a New Understanding of the Brain Will Lead to the Creation of Truly Intelligent Machines. Copyright © 2005 by Jeff Hawkins and Sandra Blakeslee. Reprinted by permission. All rights reserved.