Skip to main content

The Age of Connectome: Q&A with Sebastian Seung

This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American


In 1949, a Canadian psychologist named Donald Hebb penned the following revolutionary words in his pioneering work, The Organization of Behavior:

"Let us assume that the persistence or repetition of a reverberatory activity (or 'trace') tends to induce lasting cellular changes that add to its stability... When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased."

Or, to put it more bluntly: "Cells that fire together, wire together."


On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.


Hebb's ideas have influenced many a modern neuroscientist, notably in the area of brain mapping. To date, most brain mapping efforts have been on more of a macroscale: identifying which parts of the brain are affiliated with specific functions, for example, or staining single neurons to track them in the mass of brain tissue, or looking at thicker "wiring" that connects different parts of the brain. Ideally, neuroscientists would like to trace the actual "wiring" of the brain: the dendrites and axons that form the synaptic connections between neurons.

All the cool kids call this the "connectome." So does MIT's Sebastian Seung, -- in fact, he has a new book out (his first) called Connectome: How the Brain’s Wiring Makes Us Who We Are. Jen-Luc Piquant devoured it and pronounces it a terrific read. She now has Seung's TED talk on a never-ending loop playing in her pixelated brain. Such a fangirl.

I heard Seung speak a few years ago at the Kavli Institute for Theoretical Physics in Santa Barbara, and was thoroughly riveted; I wasn't the least surprised when he was tapped for TED. He came to neuroscience by way of condensed matter physics theory, working on artificial neural networks (ANNs).

That early interest served as a natural segue into neuroscience, and he's drawn on that expertise in his current research. His goal is nothing less than to transform the field of neuroanatomy into a "high throughput, data-rich field of science," via the creation of automated systems that can take a sample of brain tissue as raw input and generate a complete circuit diagram as output.

This, says Seung, is "an image processing problem of unprecedented scale, because a sample of even modest dimensions yields a huge amount of data at nanoscale resolution." He's not kidding: we're talking terabytes and petabytes of data just to produce a wiring map of a fruit fly, never mind a human brain. Achieving the complete fruit fly "connectome" would constitute success beyond most people's wildest dreams, and make it more likely that neuroscientists could achieve, in our lifetime, circuit diagrams for certain critical locations in the human brain: the hippocampus, for example, or the olfactory bulb and retina.

If you want to see the connectome in action, there's a nifty online mini-movie on the Technology Review site. It’s a 3D animation of the detailed wiring map of part of the rabbit retina called the inner plexiform layer. That's the little piece of neural tissue at the back of the eye that senses light and sends visual information to the brain. A single neurite appears first, shown in green, followed by a larger subset of neurites in multiple colors. The result is a veritable "brain forest" as the animation traces the "wires" through the dense brain tissue.

Even better: Seung graciously found time to chat with Jen-Luc Piquant ("Squee!") about his new book, and why he believes the connectome may hold the key to the basis of personality, intelligence, memory, and maybe even mental disorders.

Jen-Luc: How, exactly, does a condensed matter physicist find his way into neuroscience?

Seung: When I was finishing my PhD I started thinking maybe the most interesting emergent properties are living systems. How do molecules come together to make living organisms, and how do neurons come together to make a smart brain? I built mathematical models of neural networks at Bell Labs for a number of years and I continued doing that when I went to MIT.

But by then I was growing a little disenchanted because we had to make so many assumptions in building those models. And new technologies were coming online that would in principle allow us to map the connections of neural networks, and potentially provide a much-needed constraint for our models.

Jen-Luc: Why is it so important to study neurons and the connections between them?

Seung: There is a tradition of dividing the brain into regions and ascribing functions to them. But the regional approach can’t answer the question of why a brain region might work really well in some people and not well in others. We can’t explain intelligence and many mental disorders. We can’t explain how a brain region changes when we learn something. For that we need to further subdivide the regions of the neurons. Anybody who has studied physics would be familiar with that idea. You learn a lot by dividing a piece of matter into atoms.

I believe the connectome is crucial to what we care about most: differences between people and change in ourselves. There is the joke “I quit smoking every day.” That is about a change in neural activity. I have made the motion and I put down the cigarette and some activity pattern has changed, but to quit smoking for a lifetime? That probably involves a change in the connectome.

Jen-Luc: What you're really talking about is changing the Self. But isn't that already a constantly emerging and evolving thing?

Seung: We need to distinguish between two notions of self. There is the conscious self, the one that is always changing and constantly shifting, and then there is the stable self, the one that changes only with difficulty – the core self. We are really in the dark about that.

We know that the connectome has all these mechanisms for change. In the book, I talk about the four R’s: Reweighting, Reconnection, Rewiring and Regeneration. But we don’t know exactly how those are involved in change – in learning a new skill, remembering something, recovering from an injury. We are not really sure how those processes serve those changes to the self.

Jen-Luc: How might we go about learning that?

Seung: It is important for us to figure out what I call code breaking. You can’t point to any single connection that is responsible for personal change. Presumably it is some kind of pattern of connections. Where memory is stored, there must be a pattern that is there and we have never been able to see patterns. That is why the technology is so important.

Another question about the four R’s is, to what extent do they continue throughout adulthood? And the last question is, what artificial means can we use to promote them if we want more of them? I don’t think you can ever take a pill that would just change your behavior, but you might be able to take a pill that will enable you to better change yourself.

Jen-Luc: What excites you most in terms of future challenges?

Seung: Right now I am very excited about the potential to see a memory. When we store memory, there is presumably some change in the connections of neurons. Can you see what happens? Can we see that actually happen inside the brain? Memory seems intangible, but according to the hypothesis of neuroscientists, it is a material structure and we should be able to see it.

The second challenge is the possibility of seeing connectopathies – the hypothetical miswirings of the brain that are associated with mental disorders. Those are the two challenges that I want to attack. There are plenty of other challenges to attack, but those are the two that excite me the most.

Jen-Luc: There is evidence that we don’t actually retrieve a memory; it's not stored in a file like in a computer. We rebuild them every single time, so a "memory" might be distributed across the entire brain. Is that going to make it doubly hard to finally "see" a memory?

Seung: Sure. One of the problems is that we can’t view the entire brain. We would have to do little pieces, and that means we need to have a good idea about where to look. The cognitive neuroscientists who study the brain on a coarse scale have given us some candidate areas. You might worry that we can’t capture all of the memories, but if memory is distributed than any portion of it already contains a memory. I am not worried so much about that. I think that even part of the pattern will tell us something.

Jen-Luc: You're currently working on mapping a mouse retina.

Seung: Yes. We have this new site: Eyewire.org. It is a citizen science project. Our AI is not accurate enough to map the connectome by itself. We still need human intervention. So we have now created this website that allows anybody to do it.

We want to motivate them with the same motivation that drives scientists to participate in discovery. This website allows people to learn about the retina and discover something about the retina at the same time.

There is a rhetorical question that the philosophers always ask: "Is the brain complex enough to understand itself?" Maybe if we unite our billions of brains and cooperate with AI, we can do the job.

Jen-Luc: Towards the end of your book, you tackle the inevitable question of the singularity, and the possibility of uploading an entire human brain. How big is the current gap between what we can do now with digital avatars, and the actual uploading of human consciousness?

Seung: In my book I define the quest: to deconstruct the brain. It is the inverse of artificial intelligence, which seeks to construct an artificial brain. To deconstruct an entire human brain, you can calculate how long that will take by extrapolating Moore’s Law. If progress is fundamentally computationally limited, then it will take, say, 40 years to get there. Everybody wants us to then simulate the brain, once we have that. I don’t know if that will be possible. I do believe that we can learn a lot about ourselves by having that connectome. but just because we have the genome doesn’t mean we have a simulation of a cell.

People are obsessed with simulation because it is an old dream, but simulation is much less important these days. In many ways the brain simulation people are out of touch with science. They are more obsessed with science fiction. And simulation is a lot harder than people think.

It is arguable that the connectome gets you closer to simulating a brain than the genome to a cell. If you look at our current theories about how things like perception and memory work, they are pretty simple actually, but maybe those theories are wrong. I don’t know what is going to come out of this. We are going to test all these theories that have accumulated over the past half century, but we don’t know what nature is going to give us when we look. I think it is important to go in with hypotheses. But it is also important to go in with an open mind.

Book cover art: Christopher Niemann.

Connectome image (top): Gigandet X, Hagmann P, Kurant M, Cammoun L, Meuli R, et al. (2008) Estimating the Confidence Level of White Matter Connections Obtained with MRI Tractography. PLoS ONE 3(12): e4006. doi:10.1371/journal.pone.0004006

Image of retina (bottom): Aleksandar Zlateski and Sebastian Seung