The Man Working To Reverse-Engineer Your Brain

Feb 29, 2012
Originally published on February 29, 2012 11:06 am

Our brains are filled with billions of neurons, entangled like a dense canopy of tropical forest branches. When we think of a concept or a memory — or have a perception or feeling — our brain's neurons quickly fire and talk to each other across connections called synapses.

How these neurons interact with each other — and what the wiring is like between them — is key to understanding our identity, says Sebastian Seung, a professor of computational neuroscience at MIT.

Seung's new book, Connectome: How the Brain's Wiring Makes Us Who We Are, explains how mapping out our neural connections in our brains might be the key to understanding the basis of things like personality, memory, perception and ideas, as well as illnesses that happen in the brain, like autism and schizophrenia.

"These kinds of disorders have been a puzzle for a long time," says Seung. "We can look at other brain diseases, like Alzheimer's disease and Parkinson's disease, and see clear evidence that there is something wrong in the brain."

But with schizophrenia and autism, there's no clear abnormality during autopsy dissections, says Seung.

"We believe these are brain disorders because of lots of indirect evidence, but we can't look at the brain directly and see something is wrong," he says. "So the hypothesis is that the neurons are healthy, but they are simply connected together or organized in an abnormal way."

One current theory, says Seung, is that there's a connection between the wiring that develops between neurons during early infancy and developmental disorders like schizophrenia and autism.

"In autism, the development of the brain is hypothesized to go awry sometime before age 2, maybe in the womb," he says. "In schizophrenia, no one knows for sure when the development is going off course. We know that schizophrenia tends to emerge in early adulthood, so many people believe that something abnormal is happening during adolescence. Or it could be that something is happening much earlier and it's not revealed until you become an adult."

What scientists do know, he says, is that the wiring of the brain in the first three years is critical for development. Infants born with cataracts in poor countries that don't have the resources to restore their eyesight remain blind even after surgery is performed on them later in life.

"No matter how much they practice seeing, they can never really see," says Seung. "They recover some visual function, but they are still blind by comparison to you and me. And one hypothesis is that the brain didn't wire up properly when they were babies, so by the time they become adults, there's no way for the brain to learn how to see properly."

At birth, he says, you are born with all of the neurons you will ever have in life, except for neurons that exist in two specific areas of the brain: the dentate gyrus of the hippocampus, which is thought to help new memories form, and the olfactory bulb, which is involved in your sense of smell.

"The obvious hypothesis [is] that these two areas need to be highly plastic and need to learn more than other regions, and that's why new neurons have to be created — to give these regions more potential for learning," says Seung. "But we don't really have any proof of that hypothesis."

But not everything is set in stone from birth. The complex synaptic connections that allow neurons to communicate with one another develop after babies have left the womb.

"As far as we know, this is happening throughout your life," he says. "Part of the reason that we are lifelong learners — that no matter how old you get, you can still learn something new — may be due to the fact that synapse creation and elimination are both continuing into adulthood."

Connectomes: Reverse-Engineering The Brain

Only one organism has had its full connectome — or neural map — mapped out by neuroscientists. It's a tiny worm no bigger than a millimeter, but it took scientists more than a dozen years to map out its 7,000 neural connections. They started out by using the world's most powerful knife and slicing the worm into slices a thousand times thinner than a human hair. They then put each slice in an electron microscope and created a 3-D image of the worm's nervous system. That's when the true labor started, says Seung.

"That's when [neuroscientists had to] go through all these images and trace out the paths taken by all of the branches of the neurons and find the synapses, and compile all that information to create the connectome," he says.

Each of the worm's 300 neurons had between 20 and 30 connections. In comparison, humans have 10,000 connections of neurons — and billions of neurons. And scientists still aren't sure what the various pathways in a worm's nervous system mean.

"We're still far away from understanding the worm," says Seung. He says that scientists would like to eventually map a 1-millimeter cube of a human brain or a mouse brain, which contains 100,000 neurons and a billion connections.

"The imaging of all of those slices of brain can be automated and made much more reliable," he says. "And now we have computers that are getting better at seeing."

So far, though, neuroscientists have only mapped the neural connections of a piece of a mouse retina, which is very thin.

"What we know in the retina is a catalog of the types of neurons," he says. "The next challenge is to figure out what are the rules of connection between these types of neurons. And that's where we still don't know a whole lot."

Mapping more of these connections, he says, will tell us a lot about brain function and possible pathways that can be treated.

"I don't want to promise too much, and my goal right now is simply to see what is wrong," he says. "That's not in itself a cure. But obviously it's a step toward finding better treatments. The analogy I make is the study of infectious diseases before the microscope. You could see the symptoms, but you couldn't see the microbes — the bacteria that caused disease. We're in an analogous stage with mental disorders. We see the symptoms, but we don't have a clear thing we can look at in the brain and say, 'This is what's wrong.' "

Interview Highlights

On connectomes

"A connectome is a map between neurons inside a nervous system. You can imagine it as being like the map that you see in the back of the pages of in-flight magazines. Imagine that every city in that map is replaced by a neuron and every airline route between cities is replaced by a connection."

On the Jennifer Aniston neuron

"Sometimes people with seizures don't respond well to medications, and the only way for them to respond is for surgeons to remove the part of the brain from which the seizures originate. So [a computational neuroscientist] got permission to also record the signals of single neurons inside human subjects before doing the operating. So what the experimenters did was they showed the people pictures of celebrities and places and other kinds of objects, and they found that the neurons in the areas that they recorded from, which is in the medial temporal lobe ... responded highly selectively. They would respond to only a few pictures out of a large collection of many pictures. And in particular, there was one neuron in one person that responded only to pictures of Jennifer Aniston — not to Halle Berry, not to Julia Roberts, and one great finding said that this neuron did not respond to pictures of Jennifer Aniston with Brad Pitt. ... It would be overstating the case to say this neuron only responds to Jennifer Aniston because the experimenters didn't have time to show the person all possible celebrities. But it seems safe to say that this neuron responds to only a small fraction of celebrities."

On neural networks

"Your brain is this vast network of neurons, communicating through signals. And as far as neuroscientists can tell, these signals that are passed around the network are reflecting the processing of all of our mental processes — your thoughts, your feelings, your perceptions and so on."

On regenerative neurons

"If you have brain damage, and lots of neurons are killed, those neurons won't grow back except in [the dentate gyrus of the hippocampus, which is thought to help new memories form, and the olfactory bulb, which is involved in sense of smell]. So you could view it from a very pessimistic viewpoint. On the other hand, it's entirely possible that medical advances in the future will somehow activate regenerative powers in the brain. If these regenerative powers exist in [those] two areas, why not awaken them in other areas of the brain? So there's also an optimistic kind of spin on this."

Copyright 2014 NPR. To see more, visit



Scientists have known for a long time that different regions of the brain have different functions. Our guest, neuroscientist Sebastian Seung, says what really makes one brain different from another - why you have a special fondness for tacos while your spouse loves foreign films, or perhaps while some people become schizophrenics - has to do more with how our brains are wired than anything else.

The brain has billions of neurons, and Seung says the ways they are connected with one another make spying much about brain function and dysfunction. This wiring pattern, which Seung calls the connectome, changes constantly as we acquire new memories and experience emotions. It's dizzyingly complex, since every one of those billions of neurons may have a thousand connections. Sueng says everybody's wiring is different and it's at least theoretically possible to diagram it.

Sebastian Seung is a professor of computational neuroscience at MIT and an investigator at the Howard Hughes Medical Institute. His new book is called "Connectome: How the Brain's Wiring Makes Us Who We Are. "

Well, Sebastian Seung, welcome to FRESH AIR. One of the fascinating things that your book presents is the effort to actually map the neurons of the brain and their connections. I mean the human brain is enormously complex, a hundred billion neurons. But there is this case where scientists have managed to map the connectome, the connections of a very, very tiny worm. You want to just describe that effort?

SEBASTIAN SEUNG: Yes. So a connectome is a map of connections between neurons inside a nervous system. You can imagine it as being like the map that you see in the back pages of in-flight magazines. Imagine that every city in that map is replaced by a neuron, and every airline routes between two cities is replaced by a connection. There is only one organism for which the entire connectome has been mapped, and that's a tiny worm called Sea Elegance. It's just one millimeter long, and in the 1970s and '80s scientists mapped out all 7,000 connections between its 300 neurons.

DAVIES: So how did they do it?

SEUNG: You can take a worm and treat it like a tiny sausage and cut it into extremely thin slices using the world's sharpest knife, the diamond knife. These slices are a thousand times thinner than a hair. Now, take each one of these slices and put inside the most powerful microscope, the electron microscope, and image the nervous system of the worm at extremely high resolution. Once you're done, you have a whole stack of images of each of the slices. Effectively, it's a three-dimensional image. It's like a virtual worm. There's an image of every neuron inside this worm. Every synapse inside this worm has all been captured inside these images.

DAVIES: All right. So - so in contrast to this millimeter-long worm, I mean the human brain seems impossibly complex. I mean what are some of the challenges in trying to map the human connectome? Has it been tried? Can you do a piece of it?

SEUNG: Well, the simple answer is the progress in technology. First of all, the imaging of all those slices of brain can be automated now and made much more reliable. And the second point is that, remember that we have to analyze those images to find the neuron branches and the synapses.

There's no way that any person, any single person could look through all those images and do that task for a cubic millimeter of brain. There's so much data. But now we have computers that are getting better at seeing. It's true that robots don't see nearly as well in real life as they do in science fiction movies. But the capabilities are getting much better.

DAVIES: So 100 billion neurons in my head?

SEUNG: Yes. That's the estimate.

DAVIES: And we have each of these hundred billion neurons with thousands of connections to other neurons, is that right?

SEUNG: That's right.

DAVIES: All right. So let's take an idea, an image of a flower. Or one that you use in the book is a memory of a first kiss. How is that idea or that memory physically represented in the brain?

SEUNG: Right. Great question. People have wondered about that for millennia, at least. And we have tantalizing hints that we might be close to an answer to that question. So I will have to describe some of the theories and some of the evidence that we have so far.

Let's start out with your experience of the kiss first. Not the memory of the kids, but your actual experience. As I said before, that is represented in your brain as some kind of pattern of activity of your neurons. Neurons produce electrical pulses. Those are the signals that they use to pass information throughout the brain. And so let's imagine that some configuration of neurons, some constellation of activity is evoked by the experience of the kiss. And this might represent all kinds of information, not just the kiss itself but maybe the surroundings, the person that you're kissing and so on and so forth.

So somehow that experience has to become deeply imbedded in your brain in a manner that - well, I guess most of us never forget our first kiss. So it has to stay with you for a lifetime. And so the question is: How can an experience that is ephemeral become imbedded in your brain in a way that is almost permanent?

So the hypothesis has been that all those signals, all those transient signals, somehow modify the material structure of your brain as if - I guess the metaphor that Plato used was leaving an impression on a wax tablet. There's some kind of material thing that gets changed.

DAVIES: OK. So there is some assembly of connections that cement my memory of my first kiss, and part of it is the physical sensations, part of it is the moonlight shimmering off the water, part of it was what happened that day.

SEUNG: Why, you're very - this is a wonderfully romantic description. I guess you can remember your first kiss very well.


DAVIES: Well, I'm embellishing a little. But I guess what I'm asking is: Is there a neuron that associates with shimmering water, and another neuron with the color of her hair and another neuron that, you know, connects to another piece of it?

SEUNG: You're getting at the question of what is the representation of perceptions, which is - you have to address before memories. And yes, the evidence would be that maybe it's not a single neuron. It might be some population of a number of neurons, but they're activated by each of these properties. So any experience is somehow represented by this kind of combination of neurons, each of which is responding to some particular feature in the particular experience.

DAVIES: And the experience sort of established connections, which then remain, and so the brain is a little different than it was before the kiss.

SEUNG: Yes. And the crucial fact here is that neural activity can lead to changes in connections, that if two neurons are activated simultaneously or in quick succession, then the connections between them are modified. That's a crucial fact. There's empirical evidence for it. But the extension of that fact to an entire experience, as opposed to one connection between two neurons, that has not been adequately investigated.

DAVIES: If you're just joining us, we're speaking with Sebastian Seung. He's a neuroscientist at MIT. His latest book is called "Connectome: How the Brain's Wiring Makes Us Who We Are." Let's talk a little bit about how the brain develops. When a baby is born, the wiring, if you will, is not complete. What happens in those early months?

SEUNG: So the first thing that has to happen in brain development is happening in the womb. These 100 billion neurons have to be created, and then they have to migrate to their appropriate positions in a very complex dance. Then they're sending out their branches, and the branches are intertwining with each other. And finally, those branches are forming the synaptic connections with each other. And that extension of branches and a lot of the synapse formation is still happening after birth.

DAVIES: And how long does it go on? Do we know?

SEUNG: Well, as far as we know, this kind of creation of synapses is happening throughout your life. Part of the reason that we are lifelong learners, that no matter how old you get you can still learn something new, may be due to the fact that synapse creation and elimination are both continuing into adulthood.

In the 1960s, many neuroscientists thought that this process of creating and eliminating connections actually stopped by the time you were an adult. But now there's plenty of evidence that it keeps on going.

DAVIES: You know, there's some fascinating evidence that you describe, that the wiring of the brain in the first three years is critical. And there's this phrase: The first three years last forever. Can you give us some examples of behavioral observations that suggest that, in fact, things happen in those first three years which are critical, in some cases, can't be reversed?

SEUNG: One of the most outstanding examples is patients who are born blind because of cataracts. In poor countries, they can't get the simple cataract surgery that would restore their eyesight, and they grow up blind. If you give them cataract surgery when they're adults, it turns out that they still can't see. So this suggests that - and no matter how long they practice seeing, they can never really see.

They recover some visual function, but they are still blind by comparison to you and me. And one hypothesis is that the brain didn't wire up properly when they were little babies, and so by the time they become adults, there's no way for the brain to learn how to see properly.

DAVIES: Right. And it didn't wire up because it wasn't getting the proper feedback from the, I guess, the optic nerves?

SEUNG: That's right. So here, the assumption is that brain wiring is guided not only by genes, but also by experiences, that we need experiences for the brain to wire up and connect properly. When deprived of those crucial formative experiences, the brain doesn't develop properly. That's the best explanation we have at this point.

DAVIES: And explain the feral children and what they reveal, if anything, about this.

SEUNG: Sure. The feral children are the legendary cases of kids who were orphaned, abandoned in the wilderness and raised by animals. These kids, when discovered later on - let's say in adulthood - attempts were made to civilize them, but they never learned how to speak. They never learned normal social behaviors, never learned language.

And so, again, this suggests the existence of a so-called critical period, a period during which you have to have experiences for your brain to develop properly, and after which nothing can be done.

DAVIES: Does that make sense to you, that not being around other human beings who were speaking might've affected the brain connections in such a way so that it simply isn't possible?

SEUNG: Sure. It's entirely possible. These are not really science, because they're kind of anecdotal stories. They weren't studied carefully by scientists. But it's entirely possible. On the other hand, I wouldn't want to read too much into this. So sometimes, parents get scared when they hear stuff like this: the first three years last forever, if I screw up with my kids, my kids will be damaged for life, so on and so forth.

I think that proponents of this idea have gone too far in stressing the idea of irreversible damage. And you may be aware there's another kind of popular trend in neuroscience, which is to talk about neuro-plasticity, about the amazing power of the brain to recover in adulthood. And these are actually diametrically opposed takes on neuroscience. One stresses the inability of the adult ability to change, and the other stresses the amazing unlimited power of the adult brain to change. They can't both be correct.

DAVIES: Well, where are you on this?

SEUNG: Well, the truth, of course, has to be somewhere in-between. That's a boring kind of answer. But with science, we want to rigorously investigate what kinds of changes specifically are possible in adults. And the next point I would say is that we shouldn't regard these as fundamental limits. They're not like the laws of physics. These questions have changed really dependant on the conditions.

There may be new learning methods that enable adults to change when they couldn't change before and, more futuristically speaking, there could be new drugs, new neuro-technologies that will enhance the brain's capability for change.

DAVIES: Sebastian Seung's book is called "Connectome: How the Brain's Wiring Makes Us Who We Are." We'll talk more after a break. This is FRESH AIR.


DAVIES: If you're just joining us, we're speaking with neuroscientist Sebastian Seung. His new book is called "Connectome: How the Brain's Wiring Makes Us Who We Are." I mean, one of the interesting ideas that you mentioned in the book is the notion that when a baby is born, there are certain functions that, if they don't develop in the first three years, might never come, because they were really critical to our ancestors being able to function, I mean, you know, their capacity to interpret visual signals or auditory signals, whereas a cultural kind of enrichment, you know, from reading is not something that maybe our prehistoric ancestors needed and therefore is not quite so critical in the early wiring of the brain.

SEUNG: Yes. Again, that's a cautionary note. The examples we know of irreversible damage to the brain because of deprivation all have been associated with severe deprivation, severely abnormal deprivation. For example, being deprived of sight by cataracts is an abnormal kind of deprivation. Or never being exposed to language, that is very strange and abnormal, almost never happens.

So you shouldn't jump to the conclusion that depriving your kid of the chance to hear Mozart is going to damage them irreversibly. That seems like a big stretch.

DAVIES: So if we were able to map the neural connections in the brain, what are some of the implications, say, for treating neural disorders?

SEUNG: Since an entire human connectome is far away, I'd like to emphasize that even doing small parts of the brain now is going to tell us a lot about brain function. So one kind of study we would like to do is to search for these connect-topathies(ph), these abnormal patterns of connection that are hypothesized to underlie mental disorders like autism and schizophrenia.

We can do this by taking a little piece of brain and seeing whether the neurons inside that piece are connected in a way that's different from a piece of normal brain. Now, how would you do this kind of experiment? I should emphasize that this chopping up of brains and so on is a procedure that's done on dead brains. It can't be done on live brains.

So we can do these kinds of experiments on animals, because scientists are creating animal models of mental disorders. They take the genetic abnormalities that are associated with autism and schizophrenia, and they put those genetic abnormalities in mice. And it turns out that sometimes, these mice have abnormal behaviors, indicating that they have something analogous to the human mental disorder.

The next obvious question is: What's different about the mouse brain? And we can address that by finding a connectome of a particular part of the brain that is involved in these abnormal behaviors. Now, what about human brains? Well, you can imagine doing experiments on pieces of brain that are removed during neurosurgical procedures - for example, epilepsy. Neurosurgeons sometimes remove pieces of epileptic brain tissue, and the obvious question is: What's different about that piece of brain tissue that leads that tissue to generate epileptic seizures? We could address that through connectomics.

DAVIES: So it might be at least theoretically possible, in decades, to treat an abnormal connection in some way?

SEUNG: Well, I don't want to promise too much. My goal is simply to see what's wrong. That's not, in itself, a cure, but obviously it's a step towards finding better treatments. The analogy I make is the study of infectious diseases before the microscope. Imagine what that was like. You could see the symptoms, but you couldn't see the microbes, the bacteria that caused disease.

We're in an analogous stage with mental disorders. We see the symptoms. Most of these disorders are defined only by their symptoms. But we don't have a clear thing that we can look at in the brain and say: This is what's wrong.

DAVIES: Well, Sebastian Seung, it's been really interesting. Thanks so much for speaking with us.

SEUNG: Thanks a lot, Dave. I really enjoyed it.

DAVIES: Sebastian Seung's new book is called "Connectome: How the Brain's Wiring Makes Us Who We Are." Transcript provided by NPR, Copyright NPR.