When Randal Koene was fourteen, he read a science fiction novel that would shape the rest of his life. The City and the Stars, written in 1956 by Arthur C. Clarke, is a story set a billion years in the future, where people store their minds in a city’s central computer and take turns living one thousand years in cloned bodies. To many, it must have seemed like another futuristic pipe dream, but to Koene, it presented an exciting prospect for human kind. “It gave me the notion that everything is information,” he said. “We’re all information.”
The son of a particle physicist, Koene grew up believing that anything could be built “from the atoms up” — all you needed was the unique composition of atoms and sophisticated tools to do the job. As a kid, he spent most of his time with his younger brother dreaming up imaginary worlds, and as a teenager, he wrote a four-hundred-page novel about a civilization that lived deep underground at the center of the earth.
As he got older, Koene’s belief in the fantastical endured, and he began to think that maybe, just maybe, he could bring to life the world from Clarke’s novel where so-called “mind uploading” was commonplace. But it wouldn’t be easy. He decided he needed to find a way to copy the mind in its entirety, and somehow capture consciousness — the sense of “I” that each of us feels — and save it forever.
Decades later, the 41-year-old is trying to make it happen. A neuroengineer at a small start-up in San Francisco’s Mission District, Koene has come up with a theory called “whole brain emulation,” which aims to make an exact copy of the brain — a replica so precise that all phenomena of the mind, including consciousness, would be contained. His goal is to take this copy and transfer it to another substrate, like a supercomputer or a silicon chip, much like how software is copied to a hard drive.
Mind uploading might seem at the very least hokey, and at the most, preposterous, but it has a growing number of followers and investors — many based in the Bay Area and in Silicon Valley, including some prominent members of the tech world. Just this year, Koene has traveled to five countries — including China and Russia — to spread the gospel of whole brain emulation, and people are listening. In August, he sat on a panel in Melbourne, Australia for the 2012 Singularity Summit, a conference on artificial intelligence and its future implications.
That said, there are certainly a few holes in Koene’s theory, and most brain experts contend that it’s built upon a fundamental misunderstanding of how the brain works. For the past twenty years, George Lakoff, a cognitive linguist at the University of California, Berkeley, has argued that the mind and body simply cannot exist without each other. Neurons aren’t housed solely in the brain, he points out, but they occupy every part of the body that connects the brain directly to bodily experience.
Furthermore, after thousands of years of inquiry, nobody really understands what consciousness consists of. Jack Gallant, a professor of neuroscience at UC Berkeley, says that even if we find a way to upload facts from our neural circuitry, like memories, the main roadblock remains. “If you can’t even define a phenomenon, how can you possibly expect to measure it and record it in a way that you could move it to another device?”
But Koene isn’t troubled by the challenge — he says consciousness is included in the whole brain package. “We don’t assume to understand the strategy that went into designing a software program,” he told me. “Instead, we seek to copy it line by line.” If the neural connections are mapped correctly, he posits, consciousness will simply show up.
Although Koene and others like him have long faced ridicule and skepticism from the greater scientific community, they’re determined to push ahead. And a groundbreaking study published last year that came from more than a decade of research suggests that recreating parts of the mind is not as far-fetched as we might think. It implies that the brain isn’t unique to biology, and that its functions could perhaps one day be carried out by a machine.
In which case, the future could be a lot closer than we think.
Koene was born in Holland in 1971, moved to Long Island when he was three years old, then to Winnipeg, Canada at five. His father was a physicist, so the young boy spent his childhood following his dad, who was chasing after experimental nuclear reactors and particle accelerators. This nomadic lifestyle never allowed him to call any place home. “I wasn’t uprooted, really. More like unrooted. I never really made any roots.”
In Long Island, Koene and his family lived on a scientific compound, inside the secured boundaries of Brookhaven National Laboratory, maintained by the US Department of Energy. Within the compound, the community hailed from around the world and imparted in Koene the idea that there are no national boundaries — that there is no real nation. He lived in a bubble where harebrained scientific theories were nurtured and explored, instead of judged or shot down.
Without a feeling of national identity, Koene has lived where he’s wanted to, instead of where he feels he belongs. He got a bachelor degree in physics from The University of Amsterdam, but it wasn’t enough; to build an emulation of a mind, he needed more. So, he went on to get a master’s degree in electrical engineering at Delft University of Technology, and then to McGill University in Montreal where he received a Ph.D in computational neuroscience — a field that’s emerged in the last twenty years that studies the brain using mathematical computations of neural networks.
In 2010, Koene and physicist Suzanne Gildert founded CarbonCopies.org, a nonprofit dedicated to advancing research and organizing projects in order to achieve “substrate-independent minds,” or minds that would be sprung free from their three-pound slab of grey matter and have their information stored longterm on more durable substrates (although Gildert is no longer involved with the project). By networking with other scientists and engineers around the world, Koene was better able to design projects that fit into the bigger picture of whole brain emulation.
Today, Koene calls himself a neuroengineer — a relatively new discipline (it got its own journal in 2004) that uses engineering techniques, like fine-scale modeling of structures, to understand neural systems. He founded NeuraLink Co., a small for-profit startup in San Francisco, where he and a handful of scientists and engineers are working to build “cyborcells,” tiny wireless neural probes, powered by infrared light, which could one day fit within a human circulatory system. This would allow researchers to put thousands, even millions, of probes inside the brain to collect a high-resolution map of neural activity, which they hope will get us one step closer to mind uploading.
When you think about it, there would be some interesting benefits to living in a world where mind uploading is mainstream. We could survive harsher environmental and political climates; we would be immune to bioweapons and nuclear wars; and, Koene points out, we would even be able to erase a traumatic event from our minds or selectively tune our emotional response to be okay. It’s an attractive thought. How many of us, after we watched the 2004 film Eternal Sunshine of the Spotless Mind, kind of wished we could smudge out a memory of a bad breakup or the death of a loved one?
In fact, Gallant says that current evidence suggests that erasing specific memories is far more likely than mind uploading. “Scientists think that memories are re-encoded each time they are recalled, so if some method is used to prevent this re-encoding during recall, then traumatic memories may lose their efficacy over time.” If this could be done, he says it might serve as a valuable tool for psychotherapy.
UC Berkeley philosophy professor John Searle disagrees, saying that our lives are formed by our memories. “If you systematically lose your memories, you lose your life,” he said. “It would destroy everything because your relationships are largely based on your memories. That’s why memory loss is such a tragedy.”
Koene’s team at NeuraLink Co. is only at the very beginning stages — they’re creating experimental designs to test and verify desired features of the probes and still need someone to build a prototype of a cyborcell. With these kinds of longterm, experimental endeavors, it’s hard to know if a project will receive funding to its completion. Right now, the team says it has enough investment to push forward, but to see the entire project through would be very expensive. Koene estimates that the cyborcell project, from start to finish, will be a multimillion-dollar investment.
That may sound like a lot, but there are some surprising (and wealthy) investors behind artificial intelligence projects.
It’s hard to pin down exactly how much money mind uploading research is getting. Koene estimates the amount isn’t much — perhaps around $1 million annually. And while he wouldn’t reveal how much money his startup is receiving, he said that 31-year-old Russian billionaire Dmitry Itskov, who founded his own futurist foundation last year, the 2045 Initiative, has provided NeuraLink Co. a chunk of money to get its work off the ground.
Other investors in the transhumanist cause include Martine Rothblatt, founder and CEO of United Therapeutics, and Paypal co-founder Peter Thiel, who has invested a couple hundred thousand dollars to a project called Nemaload, which is working to digitally replicate the functionality of the brain. There are many other projects, however, that aren’t directly related to uploading, but that are building tools that Koene says could be used to advance mind uploading, like The Human Connectome Project. If you count all of these supplemental projects, Koene says the amount of money involved is tens of millions of dollars.
The Human Connectome Project, led by David Van Essen at Washington University and funded by the National Institutes of Health, is the first real attempt to map every neural connection in the brain (an average adult has about 100 billion neurons and one quadrillion connections), a network known in its entirety as the “connectome.” The team believes that mapping every synaptic pathway in the brain, which it hopes to have done by the end of the century, will lead to a better understanding of myriad disorders, like schizophrenia, autism, and multiple sclerosis.
While the connectome project has won funding from one of the world’s leading medical research centers, a number of projects that aim to develop these ultramodern tools are funded by private investors, many of whom belong to a group who call themselves “transhumanists.” It’s a crew of about 6,000 techies, mostly white men, who believe that certain experimental practices, like cryonics, gene manipulation, brain-machine interfaces, and mind uploading, will allow us frail, biological beings to transform ourselves into stronger superhumans.
The formal name of the futurist organization is Humanity+, and although its members are scattered across the globe, the beating, intellectual heart of transhumanist activity happens to be right here in the San Francisco Bay Area, and, more specifically, in Silicon Valley.
“This is the only bubble where all the stuff that people talk about in transhumanist circles is completely normal and nobody even blinks,” Koene said. “It’s dangerous … because you get used to that.”
But not everyone wants Koene to succeed in his sci-fi pursuit.
For the past two decades, George Lakoff, a cognitive linguist at UC Berkeley, has argued — in books such as Metaphors We Live By — that mind and body need each other. Lakoff is best known for his work on the “embodied mind” theory, which claims that almost all of human cognition, up through the most abstract reasoning, depends on “low-level” facilities, like physical sensations and emotions.
Neurons can be found in every part of the body, not just in the brain. There are visual neurons that have connections to the retina and others that have connections to the auditory cortex, or hearing. There are neurons that connect to your muscles through the spine. The gut and the stomach have as many neurons as the brain.
“A computer doesn’t have a body,” he said. “It’s the body that allows you to understand anything. It gives meaning to those neural firings.” Even the structures of our ideas, which he calls “conceptual primitives,” are shaped by our physical experiences.
We think in terms of metaphors, which he describes as circuits in the brain that link to different embodied parts of the brain. “When you get angry, your skin temperature rises half a degree, so that’s why you’re ‘boiling mad,’ and when you’re scared, your temperature goes down half a degree, which is why you’re ‘frozen with fear.’ Metaphors depend on that physiology.” And, he says, we can’t think without them.
Koene is familiar with this critique. He acknowledges that people need bodies to understand the world around them, but insists that robot substitutes would be effective fill-ins. By uploading a mind to a device, practitioners could connect a person’s mind through radio waves to an android, which would continue to interact with the world, gaining knowledge through new experiences.
Also, he says that everything — every sensation and every emotion — is processed in the brain, and that if we perceive something to be real, then it is. We don’t, he asserts, need our Earthly bodies for our minds to process information. “Your mind is the arbiter of what is real,” said Koene. “That’s where it’s processed, and that processing is, ultimately, all that matters to you.”
Even if an emulation of a mind could gain and process bodily experiences, Lakoff says a copy is still just a copy. The cognitive linguist has been working with colleagues at the Berkeley-based International Computer Science Institute for two decades, where they’ve been measuring neural activity and making high-scale models of the brain. But models, he says, still aren’t the real thing.
“Suppose you can model the weather,” he said. “The model of the weather is not the weather. You can’t get wet from a computer program. The same is true for neural computation. You can model what the neurons are doing, but they’re not the neurons and they’re not in the body.” In other words, it’s a simulation of the brain, not a duplication.
While the potential for mind uploading may be in dispute, what’s not misunderstood is the fundamental behavior of neurophysiology at a cellular level. For example, we know that neurons send out electrical impulses to other cells through pathways called synapses, which helps the brain process information and sensations. But what we do not know is how these actions integrate in a sophisticated manner to produce cognitive behavior, like emotional responses or decision-making, or what’s happening in terms of activity in networks of neurons.
The whole-brain emulation approach acknowledges that we aren’t smart enough to recreate the brain in its entirety, but instead looks to build one from the bottom up. By copying the lower-level connections, Koene hopes that as they operate in parallel, the higher-level behavior will emerge.
“It’s like a bricklayer who has a good understanding of bricks and mortar,” he said, “but really has no overall grasp of architecture. He may not be able to design a new bridge from scratch, but if he does his work well, he may be able to copy one brick by brick.”
UC Berkeley neuroscientist Jack Gallant says that the biggest problem isn’t extracting information from our minds, but our inability to define the phenomenon of consciousness — the elusive sense of indiviudal identity — that each of us has. “It could be tomorrow that some brilliant neuroscientist, or even computational neuroscientist, will wake up and say, ‘Ah, I’ve got it. It’s the equation here.’ But … I just don’t think that’s going to happen.”
Philosophy professor Searle, a leading expert on consciousness, thinks transhumanists are simply living in a fantasy world where normal rules don’t apply. “People love technology. They want to find a technological solution to everything. You’re unhappy in love? We’ll get a computer for you. It’s bullshit what these guys are doing and it’s amazing they’re getting away with it.”
While mind uploading — if it were to ever become a reality — may help humans deal more effectively with disease and aging, Koene also has more personal reasons for pursuing his research. Because while forward thinking has always been his strong suit, living in the present has proven to be more of a challenge.
Last April, in a move to become more honest and emotionally present, Koene wrote a revealing blog entry titled, “What not to do in your personal life: My Two Years as a Liar.” In it, he chronicled a muddle of affairs he had during his marriage to a woman (they are now separated) with whom he has two children. He says by being dishonest about his feelings, he lost a lot of trust and ruined close relationships. That was only six months ago, and since then, he’s trying to pick up the pieces.
One way he’s trying to rebuild his personal life is by practicing a less hard-core form of “radical honesty,” a technique developed by W. Brad Blanton in 1990 that is based upon the belief that lying is the reason for modern human stress and that direct, blunt honesty will lead to greater intimacy in relationships and general peace of mind. Although Koene says he’s still sensitive to people’s feelings, he makes a daily effort to be truthful in all his interactions. “Now I feel like an alcoholic who has to watch every last drop,” he wrote in an email in April. “Except in my case, it’s every white lie.”
It’s these types of human flops and follies, which most of us have made or will make in our lifetime to some degree, that make mind uploading appealing to some. If you could simply rewind your mind or another’s mind after an unfortunate misstep, you could have another chance to make it right, do it the way you wanted without all the messy fallout caused by volatile human emotion.
Koene says along with being able to undo certain traumas or mistakes, mind uploading would allow us to be more aware of what is happening inside our minds, like knowing the intended emotion behind the words of a conversation partner. “It’s like empathy to the nth degree,” he said.
It would be a place where consequences didn’t exist and possibilities of discovery were endless. “Imagine if testing a medical procedure wouldn’t mean risking death of that patient,” said Koene, “because you could revive the patient from an archive, or just run processes in parallel until it was clear the procedure could do no harm.”
It’s almost a type of transcendence to a new realm, where you get to create your own universe. If what Koene says is true — that everything we experience is mental and the mind is the only place where we experience existence — then it opens up the world to become what we want it to be. Koene says the basic ideas behind mind uploading echo the teachings of one of the major religions of the world: Buddhism.
In fact, the Dalai Lama has even given his official blessing and support to Itskov’s 2045 Initiative’s avatar project, which involves the creation of an android and a brain-computer interface system to link the mind with it. It would mean creating a life support system for the human brain, and finally developing an artificial brain in which to transfer a person’s consciousness. The end goal? Cybernetic immortality.
Itskov is a “Buddhist Transhumanist” — yes, there is such a thing — and believes that by the year 2045, our minds will be capable of moving to new bodies with capabilities that far exceed that of humans. Koene says the billionaire sees it as a route to transcendence — to truly understanding ourselves — and that to reach the self-cleansing aspect of Buddhist masters, people need 1,000 years, so extending human life will make it much easier to get there. Itskov’s foundation has already held his first meeting in Moscow in February and the next one is slated for New York City in June.
While the possibilities that the concept of mind uploading offers are interesting — even fascinating — the idea still seems, well, like science fiction. After all, how can we begin to think we can recreate our most complicated organ, when no one has even come close, after thousands of years of inquiry and research?
Turns out a new study shows that the recreating the mind might not be as far out of reach as skeptics suggest.
For more than a decade, Dr. Theodore Berger — a professor of biomedical engineering and neurobiology at the University of Southern California — has been working to make a microchip that can learn and remember in the same way the hippocampus in our brains learns and remembers.
Once a memory is created, it has to be stored. A number of experts believe there are three ways we store memories: first, in the sensory stage; second, in short-term memory; then, in long-term. The hippocampus, a horseshoe-shaped structure with one foot in the left hemisphere and the other in the right, is part of the limbic system that consolidates these new memories by discerning which are important and sending them to the appropriate part of the cerebral hemisphere for long-term storage.
When a person’s hippocampus becomes damaged or has to be removed (sometimes it’s the only way to treat severe epilepsy), it can rob him or her of the ability to store long-term memories. A person without a hippocampus can still recall several events for a short period of time after they happen, but these memories fade after about thirty seconds. (Interestingly, when people have their hippocampuses removed, they keep memories they formed before the operation, even though they can’t store new ones.) In Alzheimer’s, a degenerative disease without a known cure, people also lose the ability to store long-term memories. It’s a tragic illness that afflicts an estimated 5.4 million people in the United States (about 500,000 in California alone) and costs the country $183 billion a year to treat, according to the Alzheimer’s Association.
Berger’s tireless efforts to recreate the function of the hippocampus proved to be worthwhile. Just last year, he was the lead author of an article published in the Journal of Neural Engineering that detailed the success of a neural prosthesis used to assume the function of the hippocampus in rats’ brains. (His team also worked with scientists from Wake Forest University in the study.)
In the experiment, researchers chemically severed the connection between the two major internal divisions of the hippocampus, which interact with each other to create long-term memory. Then, they implanted a tiny silicon microchip that duplicated the pattern of interaction between the two subregions and had the rats learn tasks, like pressing one lever instead of another for a reward.
To measure the activity between the two subregions of the hippocampus, the team implanted a bipolar stimulation electrode in the stratum radiatum of one subregion, which measured neural signals as the rats learned tasks. When the electrical activity peaked, it meant the new memories were sent to the hippocampus for long-term storage. To further prove that the microchip was functioning as a real hippocampus would, researchers disabled the chip and the rats lost their ability to store long-term memories. When they reactivated the device, long-term memory capability returned.
“Flip the switch on, and the rats remember,” Berger was quoted as saying in a PR Newswire article shortly after the study was completed. “Flip it off, and the rats forget.”
Basically, the study showed that a damaged part of the brain can be replaced with a silicon part that acts the same way. But the device can’t retrieve old memories; it can only store new ones on the microchip. It also relies on functional live tissue to teach it what it needs to know, so it couldn’t be used to replace tissue that was already damaged.
Koene says that while this work is impressive, it doesn’t mean that the chip stores meaningful memories. The cells, he says, aren’t actually storing a long-term memory; instead, they’re creating impermanent and shallow neural connections after an episode occurs.
There’s still a lot of work to be done. But when Berger started talking about his plans to create a synthetic hippocampus, the scientific and philosophy communities never thought it would happen.
Gallant was one of his skeptics. “I thought this was just silly, that it would never work,” he wrote in an email. “After all, no one had ever done this before. And there are lots of reasons to think that it is much, much easier to decode information from the brain than it is to encode new information and put it into the brain.” But since reading the paper Berger and his team wrote, Gallant gives them major credit for accomplishing such a complicated feat. “What they have achieved is indeed quite impressive. I think that it is very, very cool.”
Koene says that one of his biggest accomplishments is convincing prominent scientists, including Berger, to not only take whole brain emulation seriously, but to do so publicly. It’s forward thinkers like this that move theory into reality, but like every major feat, it takes hard work and perseverance.
“When you really think about it, things still have to make sense the way they always have. Time goes on and you want to achieve certain things,” said Koene. “To do them, you need to put the work in, and you don’t sit around and wait for trends to happen.”
Ultimately, however, Koene’s main obstacle may not be his detractors or funding; instead, it may be the very thing he seeks to understand. Perhaps the main problem isn’t that we don’t comprehend what creates our identities, but that our most advanced tool — the human brain — isn’t sophisticated enough to crack its own code.
Maybe the brain just isn’t designed to understand itself in its entirety. But that doesn’t mean Koene and other proponents of mind uploading will stop trying.
Editor’s Note: The original version of this story got wrong the year Randal Koene was born; it was 1971. We also got wrong the year Koene co-founded CarbonCopies.org; it was 2010. This version has been corrected.