The Extraordinary Link Between Deep Neural Networks and the Nature of the Universe

Nobody understands why deep neural networks are so good at solving complex problems. Now physicists say the secret is buried in the laws of physics.

In the last couple of years, deep learning techniques have transformed the world of artificial intelligence. One by one, the abilities and techniques that humans once imagined were uniquely our own have begun to fall to the onslaught of ever more powerful machines. Deep neural networks are now better than humans at tasks such as face recognition and object recognition. They’ve mastered the ancient game of Go and thrashed the best human players.

But there is a problem. There is no mathematical reason why networks arranged in layers should be so good at these challenges. Mathematicians are flummoxed. Despite the huge success of deep neural networks, nobody is quite sure how they achieve their success.

Today that changes thanks to the work of Henry Lin at Harvard University and Max Tegmark at MIT. These guys say the reason why mathematicians have been so embarrassed is that the answer depends on the nature of the universe. In other words, the answer lies in the regime of physics rather than mathematics.

First, let’s set up the problem using the example of classifying a megabit grayscale image to determine whether it shows a cat or a dog.

Such an image consists of a million pixels that can each take one of 256 grayscale values. So in theory, there can be 2561000000 possible images, and for each one it is necessary to compute whether it shows a cat or dog. And yet neural networks, with merely thousands or millions of parameters, somehow manage this classification task with ease.

In the language of mathematics, neural networks work by approximating complex mathematical functions with simpler ones. When it comes to classifying images of cats and dogs, the neural network must implement a function that takes as an input a million grayscale pixels and outputs the probability distribution of what it might represent.

The problem is that there are orders of magnitude more mathematical functions than possible networks to approximate them. And yet deep neural networks somehow get the right answer.

Now Lin and Tegmark say they’ve worked out why. The answer is that the universe is governed by a tiny subset of all possible functions. In other words, when the laws of physics are written down mathematically, they can all be described by functions that have a remarkable set of simple properties.

So deep neural networks don’t have to approximate any possible mathematical function, only a tiny subset of them.

To put this in perspective, consider the order of a polynomial function, which is the size of its highest exponent. So a quadratic equation like y=x2 has order 2, the equation y=x24 has order 24, and so on.

Obviously, the number of orders is infinite and yet only a tiny subset of polynomials appear in the laws of physics. “For reasons that are still not fully understood, our universe can be accurately described by polynomial Hamiltonians of low order,” say Lin and Tegmark. Typically, the polynomials that describe laws of physics have orders ranging from 2 to 4.

The laws of physics have other important properties. For example, they are usually symmetrical when it comes to rotation and translation. Rotate a cat or dog through 360 degrees and it looks the same; translate it by 10 meters or 100 meters or a kilometer and it will look the same. That also simplifies the task of approximating the process of cat or dog recognition.

These properties mean that neural networks do not need to approximate an infinitude of possible mathematical functions but only a tiny subset of the simplest ones.

There is another property of the universe that neural networks exploit. This is the hierarchy of its structure. “Elementary particles form atoms which in turn form molecules, cells, organisms, planets, solar systems, galaxies, etc.,” say Lin and Tegmark. And complex structures are often formed through a sequence of simpler steps.

This is why the structure of neural networks is important too: the layers in these networks can approximate each step in the causal sequence.

Lin and Tegmark give the example of the cosmic microwave background radiation, the echo of the Big Bang that permeates the universe. In recent years, various spacecraft have mapped this radiation in ever higher resolution. And of course, physicists have puzzled over why these maps take the form they do.

Tegmark and Lin point out that whatever the reason, it is undoubtedly the result of a causal hierarchy. “A set of cosmological parameters (the density of dark matter, etc.) determines the power spectrum of density fluctuations in our universe, which in turn determines the pattern of cosmic microwave background radiation reaching us from our early universe, which gets combined with foreground radio noise from our galaxy to produce the frequency-dependent sky maps that are recorded by a satellite-based telescope,” they say.

Each of these causal layers contains progressively more data. There are only a handful of cosmological parameters but the maps and the noise they contain are made up of billions of numbers. The goal of physics is to analyze the big numbers in a way that reveals the smaller ones.

And when phenomena have this hierarchical structure, neural networks make the process of analyzing it significantly easier.

“We have shown that the success of deep and cheap learning depends not only on mathematics but also on physics, which favors certain classes of exceptionally simple probability distributions that deep learning is uniquely suited to model,” conclude Lin and Tegmark.

That’s interesting and important work with significant implications. Artificial neural networks are famously based on biological ones. So not only do Lin and Tegmark’s ideas explain why deep learning machines work so well, they also explain why human brains can make sense of the universe. Evolution has somehow settled on a brain structure that is ideally suited to teasing apart the complexity of the universe.

This work opens the way for significant progress in artificial intelligence. Now that we finally understand why deep neural networks work so well, mathematicians can get to work exploring the specific mathematical properties that allow them to perform so well. “Strengthening the analytic understanding of deep learning may suggest ways of improving it,” say Lin and Tegmark.

Deep learning has taken giant strides in recent years. With this improved understanding, the rate of advancement is bound to accelerate.

Ref: arxiv.org/abs/1608.08225: Why Does Deep and Cheap Learning Work So Well?

ORIGINAL: TechnologyReview

by Emerging Technology from the arXiv

September 9, 2016

Inside Vicarious, the Secretive AI Startup Bringing Imagination to Computers

By reinventing the neural network, the company hopes to help computers make the leap from processing words and symbols to comprehending the real world.

Life would be pretty dull without imagination. In fact, maybe the biggest problem for computers is that they don’t have any.

That’s the belief motivating the founders of Vicarious, an enigmatic AI company backed by some of the most famous and successful names in Silicon Valley. Vicarious is developing a new way of processing data, inspired by the way information seems to flow through the brain. The company’s leaders say this gives computers something akin to imagination, which they hope will help make the machines a lot smarter.

Vicarious is also, essentially, betting against the current boom in AI. Companies including Google, Facebook, Amazon, and Microsoft have made stunning progress in the past few years by feeding huge quantities of data into large neural networks in a process called “deep learning.” When trained on enough examples, for instance, deep-learning systems can learn to recognize a particular face or type of animal with very high accuracy (see “10 Breakthrough Technologies 2013: Deep Learning”). But those neural networks are only very crude approximations of what’s found inside a real brain.

Illustration by Sophia Foster-Dimino

Vicarious has introduced a new kind of neural-network algorithm designed to take into account more of the features that appear in biology. An important one is the ability to picture what the information it’s learned should look like in different scenarios—a kind of artificial imagination. The company’s founders believe a fundamentally different design will be essential if machines are to demonstrate more human like intelligence. Computers will have to be able to learn from less data, and to recognize stimuli or concepts more easily.

Despite generating plenty of early excitement, Vicarious has been quiet over the past couple of years. But this year, the company says, it will publish details of its research, and it promises some eye-popping demos that will show just how useful a computer with an imagination could be.

The company’s headquarters don’t exactly seem like the epicenter of a revolution in artificial intelligence. Located in Union City, a short drive across the San Francisco Bay from Palo Alto, the offices are plain—a stone’s throw from a McDonald’s and a couple of floors up from a dentist. Inside, though, are all the trappings of a vibrant high-tech startup. A dozen or so engineers were hard at work when I visited, several using impressive treadmill desks. Microsoft Kinect 3-D sensors sat on top of some of the engineers’ desks.

D. Scott Phoenix, the company’s 33-year-old CEO, speaks in suitably grandiose terms. “We are really rapidly approaching the amount of computational power we need to be able to do some interesting things in AI,” he told me shortly after I walked through the door. “In 15 years, the fastest computer will do more operations per second than all the neurons in all the brains of all the people who are alive. So we are really close.”

Vicarious is about more than just harnessing more computer power, though. Its mathematical innovations, Phoenix says, will more faithfully mimic the information processing found in the human brain. It’s true enough that the relationship between the neural networks currently used in AI and the neurons, dendrites, and synapses found in a real brain is tenuous at best.

One of the most glaring shortcomings of artificial neural networks, Phoenix says, is that information flows only one way. “If you look at the information flow in a classic neural network, it’s a feed-forward architecture,” he says. “There are actually more feedback connections in the brain than feed-forward connections—so you’re missing more than half of the information flow.”

It’s undeniably alluring to think that imagination—a capability so fundamentally human it sounds almost mystical in a computer—could be the key to the next big advance in AI.

Vicarious has so far shown that its approach can create a visual system capable of surprisingly deft interpretation. In 2013 it showed that the system could solve any captcha (the visual puzzles that are used to prevent spam-bots from signing up for e-mail accounts and the like). As Phoenix explains it, the feedback mechanism built into Vicarious’s system allows it to imagine what a character would look like if it weren’t distorted or partly obscured (see “AI Startup Says It Has Defeated Captchas”).

Phoenix sketched out some of the details of the system at the heart of this approach on a whiteboard. But he is keeping further details quiet until a scientific paper outlining the captcha approach is published later this year.

In principle, this visual system could be put to many other practical uses, like recognizing objects on shelves more accurately or interpreting real-world scenes more intelligently. The founders of Vicarious also say that their approach extends to other, much more complex areas of intelligence, including language and logical reasoning.

Phoenix says his company may give a demo later this year involving robots. And indeed, the job listings on the company’s website include several postings for robotics experts. Currently robots are bad at picking up unfamiliar, oddly arranged, or partly obscured objects, because they have trouble recognizing what they are. “If you look at people who are picking up objects in an Amazon facility, most of the time they aren’t even looking at what they’re doing,” he explains. “And they’re imagining—using their sensory motor simulator—where the object is, and they’re imagining at what point their finger will touch it.”

While Phoenix is the company’s leader, his cofounder, Dileep George, might be considered its technical visionary. George was born in India and received a PhD in electrical engineering from Stanford University, where he turned his attention to neuroscience toward the end of his doctoral studies. In 2005 he cofounded Numenta with Jeff Hawkins, the creator of Palm Computing. But in 2010 George left to pursue his own ideas about the mathematical principles behind information processing in the brain, founding Vicarious with Phoenix the same year.

I bumped into George in the elevator when I first arrived. He is unassuming and speaks quietly, with a thick accent. But he’s also quite matter-of-fact about what seem like very grand objectives.

George explained that imagination could help computers process language by tying words, or symbols, to low-level physical representations of real-world things. In theory, such a system might automatically understand the physical properties of something like water, for example, which would make it better able to discuss the weather. “When I utter a word, you know what it means because you can simulate the concept,” he says.

This ambitious vision for the future of AI has helped Vicarious raise an impressive $72 million so far. Its list of investors also reads like a who’s who of the tech world. Early cash came from Dustin Moskovitz, ex-CTO of Facebook, and Adam D’Angelo, cofounder of Quora. Further funding came from Peter Thiel, Mark Zuckerberg, Jeff Bezos, and Elon Musk.

Many people are itching to see what Vicarious has done beyond beating captchas. “I would love it if they showed us something new this year,” says Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence in Seattle.

In contrast to the likes of Google, Facebook, or Baidu, Vicarious hasn’t published any papers or released any tools that researchers can play with. “The people [involved] are great, and the problems [they are working on] are great,” says Etzioni. “But it’s time to deliver.”

For those who’ve put their money behind Vicarious, the company’s remarkable goals should make the wait well worth it. Even if progress takes a while, the potential payoffs seem so huge that the bet makes sense, says Matt Ocko, a partner at Data Collective, a venture firm that has backed Vicarious. A better machine-learning approach could be applied in just about any industry that handles large amounts of data, he says. “Vicarious sat us down and demonstrated the most credible pathway to reasoning machines that I have ever seen.”

Ocko adds that Vicarious has demonstrated clear evidence it can commercialize what it’s working on. “We approached it with a crapload of intellectual rigor,” he says.

It will certainly be interesting to see if Vicarious can inspire this kind of confidence among other AI researchers and technologists with its papers and demos this year. If it does, then the company could quickly go from one of the hottest prospects in the Valley to one of its fastest-growing businesses.

That’s something the company’s founders would certainly like to imagine.

ORIGINAL: MIT Tech Review

by Will Knight. Senior Editor, AI

May 19, 2016

First Human Tests of Memory Boosting Brain Implant—a Big Leap Forward

“You have to begin to lose your memory, if only bits and pieces, to realize that memory is what makes our lives. Life without memory is no life at all.” — Luis Buñuel Portolés, Filmmaker

Image Credit: Shutterstock.com

Every year, hundreds of millions of people experience the pain of a failing memory.

The reasons are many:

• traumatic brain injury, which haunts a disturbingly high number of veterans and football players; 
• stroke or Alzheimer’s disease, which often plagues the elderly; or 
• even normal brain aging, which inevitably touches us all.

Memory loss seems to be inescapable. But one maverick neuroscientist is working hard on an electronic cure. Funded by DARPA, Dr. Theodore Berger, a biomedical engineer at the University of Southern California, is testing a memory-boosting implant that mimics the kind of signal processing that occurs when neurons are laying down new long-term memories.

The revolutionary implant, already shown to help memory encoding in rats and monkeys, is now being tested in human patients with epilepsy — an exciting first that may blow the field of memory prosthetics wide open.

To get here, however, the team first had to crack the memory code.

Deciphering Memory

From the very onset, Berger knew he was facing a behemoth of a problem.

We weren’t looking to match everything the brain does when it processes memory, but to at least come up with a decent mimic, said Berger.

“Of course people asked: can you model it and put it into a device? Can you get that device to work in any brain? It’s those things that lead people to think I’m crazy. They think it’s too hard,” he said.

But the team had a solid place to start.

The hippocampus, a region buried deep within the folds and grooves of the brain, is the critical gatekeeper that transforms memories from short-lived to long-term. In dogged pursuit, Berger spent most of the last 35 years trying to understand how neurons in the hippocampus accomplish this complicated feat.

At its heart, a memory is a series of electrical pulses that occur over time that are generated by a given number of neurons, said Berger. This is important — it suggests that we can reduce it to mathematical equations and put it into a computational framework, he said.

Berger hasn’t been alone in his quest.

By listening to the chatter of neurons as an animal learns, teams of neuroscientists have begun to decipher the flow of information within the hippocampus that supports memory encoding. Key to this process is a strong electrical signal that travels from CA3, the “input” part of the hippocampus, to CA1, the “output” node.

This signal is impaired in people with memory disabilities, said Berger, so of course we thought if we could recreate it using silicon, we might be able to restore — or even boost — memory.

Bridging the Gap

Yet this brain’s memory code proved to be extremely tough to crack.

The problem lies in the non-linear nature of neural networks: signals are often noisy and constantly overlap in time, which leads to some inputs being suppressed or accentuated. In a network of hundreds and thousands of neurons, any small change could be greatly amplified and lead to vastly different outputs.

It’s a chaotic black box, laughed Berger.

With the help of modern computing techniques, however, Berger believes he may have a crude solution in hand. His proof?

Use his mathematical theorems to program a chip, and then see if the brain accepts the chip as a replacement — or additional — memory module.

Berger and his team began with a simple task using rats. They trained the animals to push one of two levers to get a tasty treat, and recorded the series of CA3 to CA1 electronic pulses in the hippocampus as the animals learned to pick the correct lever. The team carefully captured the way the signals were transformed as the session was laid down into long-term memory, and used that information — the electrical “essence” of the memory — to program an external memory chip.

They then injected the animals with a drug that temporarily disrupted their ability to form and access long-term memories, causing the animals to forget the reward-associated lever. Next, implanting microelectrodes into the hippocampus, the team pulsed CA1, the output region, with their memory code.

The results were striking — powered by an external memory module, the animals regained their ability to pick the right lever.

Encouraged by the results, Berger next tried his memory implant in monkeys, this time focusing on a brain region called the prefrontal cortex, which receives and modulates memories encoded by the hippocampus.

Placing electrodes into the monkey’s brains, the team showed the animals a series of semi-repeated images, and captured the prefrontal cortex’s activity when the animals recognized an image they had seen earlier. Then with a hefty dose of cocaine, the team inhibited that particular brain region, which disrupted the animal’s recall.

Next, using electrodes programmed with the “memory code,” the researchers guided the brain’s signal processing back on track — and the animal’s performance improved significantly.

A year later, the team further validated their memory implant by showing it could also rescue memory deficits due to hippocampal malfunction in the monkey brain.

A Human Memory Implant

Last year, the team cautiously began testing their memory implant prototype in human volunteers.

Because of the risks associated with brain surgery, the team recruited 12 patients with epilepsy, who already have electrodes implanted into their brain to track down the source of their seizures.

Repeated seizures steadily destroy critical parts of the hippocampus needed for long-term memory formation, explained Berger. So if the implant works, it could benefit these patients as well.

The team asked the volunteers to look through a series of pictures, and then recall which ones they had seen 90 seconds later. As the participants learned, the team recorded the firing patterns in both CA1 and CA3 — that is, the input and output nodes.

Using these data, the team extracted an algorithm — a specific human “memory code” — that could predict the pattern of activity in CA1 cells based on CA3 input. Compared to the brain’s actual firing patterns, the algorithm generated correct predictions roughly 80% of the time.

It’s not perfect, said Berger, but it’s a good start.

Using this algorithm, the researchers have begun to stimulate the output cells with an approximation of the transformed input signal.

We have already used the pattern to zap the brain of one woman with epilepsy, said Dr. Dong Song, an associate professor working with Berger. But he remained coy about the result, only saying that although promising, it’s still too early to tell.

Song’s caution is warranted. Unlike the motor cortex, with its clear structured representation of different body parts, the hippocampus is not organized in any obvious way.

It’s hard to understand why stimulating input locations can lead to predictable results, said Dr. Thoman McHugh, a neuroscientist at the RIKEN Brain Science Institute. It’s also difficult to tell whether such an implant could save the memory of those who suffer from damage to the output node of the hippocampus.

“That said, the data is convincing,” McHugh acknowledged.

Berger, on the other hand, is ecstatic. “I never thought I’d see this go into humans,” he said.

But the work is far from done. Within the next few years, Berger wants to see whether the chip can help build long-term memories in a variety of different situations. After all, the algorithm was based on the team’s recordings of one specific task — what if the so-called memory code is not generalizable, instead varying based on the type of input that it receives?

Berger acknowledges that it’s a possibility, but he remains hopeful.

I do think that we will find a model that’s a pretty good fit for most conditions, he said. After all, the brain is restricted by its own biophysics — there’s only so many ways that electrical signals in the hippocampus can be processed, he said.

“The goal is to improve the quality of life for somebody who has a severe memory deficit,” said Berger. “If I can give them the ability to form new long-term memories for half the conditions that most people live in, I’ll be happy as hell, and so will be most patients.”

ORIGINAL: Singularity Hub

BY SHELLY FAN

ON NOV 15, 2015

Will Computers Redefine the Roots of Math?

When a legendary mathematician found a mistake in his own work, he embarked on a computer-aided quest to eliminate human error. To succeed, he has to rewrite the century-old rules underlying all of mathematics.

Hannes Hummel for Quanta Magazine

Mathematics is being rebuilt on top of new foundations.

On a recent train trip from Lyon to Paris, Vladimir Voevodsky sat next to Steve Awodey and tried to convince him to change the way he does mathematics.

Voevodsky, 48, is a permanent faculty member at the Institute for Advanced Study (IAS) in Princeton, N.J. He was born in Moscow but speaks nearly flawless English, and he has the confident bearing of someone who has no need to prove himself to anyone. In 2002 he won the Fields Medal, which is often considered the most prestigious award in mathematics.

Now, as their train approached the city, Voevodsky pulled out his laptop and opened a program called Coq, a proof assistant that provides mathematicians with an environment in which to write mathematical arguments. Awodey, a mathematician and logician at Carnegie Mellon University in Pittsburgh, Pa., followed along as Voevodsky wrote a definition of a mathematical object using a new formalism he had created, called univalent foundations. It took Voevodsky 15 minutes to write the definition.

“I was trying to convince [Awodey] to do [his mathematics in Coq],” Voevodsky explained during a lecture this past fall. “I was trying to convince him that it’s easy to do.”

The idea of doing mathematics in a program like Coq has a long history. The appeal is simple: Rather than relying on fallible human beings to check proofs, you can turn the job over to computers, which can tell whether a proof is correct with complete certainty. Despite this advantage, computer proof assistants haven’t been widely adopted in mainstream mathematics. This is partly because translating everyday math into terms a computer can understand is cumbersome and, in the eyes of many mathematicians, not worth the effort.

For nearly a decade, Voevodsky has been advocating the virtues of computer proof assistants and developing univalent foundations in order to bring the languages of mathematics and computer programming closer together. As he sees it, the move to computer formalization is necessary because some branches of mathematics have become too abstract to be reliably checked by people.

“The world of mathematics is becoming very large, the complexity of mathematics is becoming very high, and there is a danger of an accumulation of mistakes,” Voevodsky said. Proofs rely on other proofs; if one contains a flaw, all others that rely on it will share the error.

This is something Voevodsky has learned through personal experience. In 1999 he discovered an error in a paper he had written seven years earlier. Voevodsky eventually found a way to salvage the result, but in an article last summer in the IAS newsletter, he wrote that the experience scared him. He began to worry that unless he formalized his work on the computer, he wouldn’t have complete confidence that it was correct.

But taking that step required him to rethink the very basics of mathematics. The accepted foundation of mathematics is set theory. Like any foundational system, set theory provides a collection of basic concepts and rules, which can be used to construct the rest of mathematics. Set theory has sufficed as a foundation for more than a century, but it can’t readily be translated into a form that computers can use to check proofs. So with his decision to start formalizing mathematics on the computer, Voevodsky set in motion a process of discovery that ultimately led to something far more ambitious: a recasting of the underpinnings of mathematics.


Set Theory and a Paradox

Set theory grew out of an impulse to put mathematics on an entirely rigorous footing — a logical basis even more secure than numbers themselves. Set theory begins with the set containing nothing — the null set — which is used to define the number zero. The number 1 can then be built by defining a new set with one element — the null set. The number 2 is the set that contains two elements — the null set (0) and the set that contains the null set (1). In this way, each whole number can be defined as the set of sets that came before it.

Olena Shmahalo/Quanta Magazine  Set theory constructs all of mathematics by starting with literally nothing — the null set — which is defined to be zero. The set that contains a single element — the null set — becomes the number 1, the set that contains two elements — the null set and the set containing the null set — becomes the number 2, and so on

Once the whole numbers are in place,

• fractions can be defined as pairs of whole numbers, 
• decimals can be defined as sequences of digits, 
• functions in the plane can be defined as sets of ordered pairs, and 
• so on. 

Luke Muehlhauser on Singularity 1on1: Superhuman AI is Coming This Century

Last week I interviewed Luke Muehlhauser for Singularity 1 on 1.

Luke Muehlhauser is the Executive Director of the Singularity Institute, the author of many articles on AI safety and the cognitive science of rationality, and the host of the popular podcast “Conversations from the Pale Blue Dot.” His work is collected at lukeprog.com.

I have to say that despite his young age and lack of a University Degree – a criticism which we discuss during our interview, Luke was one of the best and clearest spoken guests on my show and I really enjoyed talking to to him. During our 56 min-long conversation we discuss a large variety of topics such as:

• Luke’s Christian-Evangelico personal background as the first-born son of a pastor in northern Minnesota; 
• his fascinating transition transition from religion and theology to atheism and science; 
• his personal motivation and desire to overcome our very human cognitive biases and help address existential risks to humanity; 
• the Singularity Institute – its mission, members and fields of interest; 
• the “religion for geeks” (or “rapture of the nerds”) and other popular criticisms and misconceptions; 
• our chances of surviving the technological singularity.

My favorite quote from the interview:

“Superhuman AI is coming this century. By default it will be disastrous for humanity. If you want to make AI a really good thing for humanity please donate to organizations already working on that or – if you are a researcher – help us solve particular problems in mathematics, decision theory or cognitive science.”


ORIGINAL: Singularity 1 on 1
from Nikola Danaylov

The Brain’s Inner Language

Video|2:25 Credit Probing the Parliament of Neurons Clay Reid and colleagues are going deep into

SEATTLE — When Clay Reid decided to leave his job as a professor at Harvard Medical School to become a senior investigator at the Allen Institute for Brain Science in Seattle in 2012, some of his colleagues congratulated him warmly and understood right away why he was making the move.

Others shook their heads. He was, after all, leaving one of the world’s great universities to go to the academic equivalent of an Internet start-up, albeit an extremely well- financed, very ambitious one, created in 2003 by Paul Allen, a founder of Microsoft.

Related Coverage 

Mapping the Highways of the Brain
Deanna Barch and her colleagues are trying to map connections in the human brain. The study is part of the Human Connectome Project.

Still, “it wasn’t a remotely hard decision,” Dr. Reid said. He wanted to mount an all-out investigation of a part of the mouse brain. And although he was happy at Harvard, the Allen Institute offered not only great colleagues and deep pockets, but also an approach to science different from the classic university environment. The institute was already mapping the mouse brain in fantastic detail, and specialized in the large-scale accumulation of information in atlases and databases available to all of science. Photo

When neurons in the brain of a live mouse, top, are active, they flash brightly. Dr. Clay Reid, above left, and colleagues at the Allen Institute for Brain Science are working with mice to better understand the human mind. Above center, areas of the mouse cortex related to vision, and connected to other parts involving visual perception. Credit Zach Wise for The New York Times

Now, it was expanding, and trying to merge its semi-industrial approach to data gathering with more traditional science driven by individual investigators, by hiring scientists like Christof Koch from the California Institute of Technology as chief scientific officer in 2011 and Dr. Reid. As a senior investigator, he would lead a group of about 100, and work with scientists, engineers and technicians in other groups.

Without the need to apply regularly for federal grants, Dr. Reid could concentrate on one piece of the puzzle of how the brain works. He would try to decode the workings of one part of the mouse brain, the million neurons in the visual cortex, from, as he puts it, “molecules to behavior.”

There are many ways to map the brain and many kinds of brains to map. Although the ultimate goal of most neuroscience is understanding how human brains work, many kinds of research can’t be done on human beings, and the brains of mice and even flies share common processes with human brains.

The work of Dr. Reid, and scientists at Allen and elsewhere who share his approach, is part of a surge of activity in brain research as scientists try to build the tools and knowledge to explain — as well as can ever be explained — how brains and minds work. Besides the Obama administration’s $100 million Brain Initiative and the European Union’s $1 billion, decade-long Human Brain Project, there are numerous private and public research efforts in the United States and abroad, some focusing on the human brain, others like Dr. Reid’s focusing on nonhumans.

While the Human Connectome Project, which is spread among several institutions, aims for an overall picture of the associations among parts of the human brain, other scientific teams have set their sights on drilling to deeper levels. For instance, the Connectome Project at Harvard is pursuing a structural map of the mouse brain at a level of magnification that shows packets of neurochemicals at the tips of brain cells.

At Janelia Farm, the Virginia research campus of the Howard Hughes Medical Institute, researchers are aiming for an understanding of the complete fly brain — a map of sorts, if a map can be taken to its imaginable limits, including structure, chemistry, genetics and activity.

“I personally am inspired by what they’re doing at Janelia,” Dr. Reid said.

All these efforts start with maps and enrich them. If Dr. Reid is successful, he and his colleagues will add what you might call the code of a brain process, the language the neurons use to store, transmit and process information for this function.

Not that this would be any kind of final answer. In neuroscience, perhaps more than in most other disciplines, every discovery leads to new questions.

“With the brain,” Dr. Reid said, “you can always go deeper.”

‘Psychoanalyst’s Kid Probes Brain!’ Photo

A diamond-tipped slicer is used to prepare a piece of a mouse’s brain for examination with a modified electron microscope at the Allen Institute. Credit Zach Wise for The New York Times

Dr. Reid, 53, grew up in Boston, in a family with deep roots in medicine. His grandfather taught physiology at Harvard Medical School. “My parents were both psychoanalysts,” he said during an interview last fall, smiling as he imagined a headline for this article, “Psychoanalyst’s Kid Probes Brain!”

“I pretty much always knew that I wanted to be a scientist,” he said.

As an undergraduate at Yale, he majored in physics and philosophy and in mathematics, but in the end decided he didn’t want to be a physicist. Biology was attractive, but he was worried enough about his mathematical bent to talk to one of his philosophy professors about concerns that biology would too fuzzy for him.

The professor had some advice. “You really should read Hubel and Wiesel,” he said, referring to David Hubel and Torsten Wiesel, who had just won the Nobel Prize in 1981 for their work showing how binocular vision develops in the brain.

He read their work, and when he graduated in 1982, he was convinced that the study of the brain was both hard science and a wide-open field. He went on to an M.D.-Ph.D. program at Cornell Medical College and Rockefeller University, where Dr. Wiesel had his lab (he would go on to be president of Rockefeller).

As his studies progressed, Dr. Reid began to have second thoughts about pursuing medicine rather than research. Just a week before he was to commit to a neurology residency, he said, “I ran into a friend from the Wiesel lab and said, ‘Save me.’ ”

That plea led to postdoctoral research in the Rockefeller lab. He stayed as a faculty member until moving to Harvard in 1996.

Mathematics and physics were becoming increasingly important in neurobiology, a trend that has continued, but there was still a certain tension between different mind-sets, he recalled. He found that there were intangible skills involved in biological research. “Good biological intuition was equally important to chops in math and physics,” he said.

“Torsten once said to me, ‘You know, Clay, science is not an intelligence test.’ ”

Though he didn’t recall that specific comment, Dr. Wiesel said recently that it sounded like something he would have said. “I think there are a lot of smart people who never make it in science. Why is it? What is it that is required in addition?”

Intuition is important, he said, “knowing what kind of questions to ask.” And, he said, “the other thing is a passion for getting to the core of the problem.”

Dr. Reid, he said, was not only smart and full of energy, but also “interested in asking questions that I think can get to the core of a problem.”

At Harvard, Dr. Reid worked on the Connectome Project to map the connections between neurons in the mouse brain. The Connectome Project aims at a detailed map, a wiring diagram at a level fantastically more detailed than the work being done to map the human brain with M.R.I. machines. But electron microscopes produce a static picture from tiny slices of preserved brain.

Dr. Reid began working on tying function to mapping. He and one of his graduate students, Davi Bock, now at Janelia Farm, linked studies of active mouse brains to the detailed structural images produced by electron microscopes.

Dr. Bock said he recalled Dr. Reid as having developed exactly the kind of intuition and “good lab hands” that Dr. Wiesel seemed to be encouraging. He and another graduate student were stumped by a technical problem involving a new technique for studying living brains, and Dr. Reid came by.

“Clay got on this bench piled up with components,” Dr. Bock said. “He started plugging and unplugging different power cables. We just stood there watching him, and I was sure he was going to scramble everything.” But he didn’t. Whatever he did worked.

That was part of the fun of working in the lab, Dr. Bock said, “not that he got it right every time.” But his appreciation for Dr. Reid as a leader and mentor went beyond admiration for his “mad scientist lab hands.”

“He has a deep gut level enthusiasm for what’s beautiful and what’s profound in neuroscience, and he’s kind of relentless,” Dr. Bock said.

Showing a Mouse a Picture

That instinct, enthusiasm and relentlessness will be necessary for his current pursuit. To crack the code of the brain, Dr. Reid said, two fundamental problems must be solved.

The first is: “How does the machine work, starting with its building blocks, cell types, going through their physiology and anatomy,” he said. That means knowing all the different types of neurons in the mouse visual cortex and their function — information that science doesn’t have yet.

It also means knowing what code is used to pass on information. When a mouse sees a picture, how is that picture encoded and passed from neuron to neuron? That is called neural computation. 

Nuno da Costa of the Allen Institute prepared a slice of mouse brain for the modified electron microscope at Dr. Reid’s lab in Seattle. “With the brain, you can always go deeper,” Dr. Reid said. Credit Zach Wise for The New York Times

“The other highly related problem is: How does that neural computation create behavior?” he said. How does the mouse brain decide on action based on that input?

He imagined the kind of experiment that would get at these deep questions. A mouse might be trained to participate in an experiment now done with primates in which an animal looks at an image. Later, seeing several different images in sequence, the animal presses a lever when the original one appears. Seeing the image, remembering it, recognizing it and pressing the lever might take as long as two seconds and involve activity in several parts of the brain.

Understanding those two seconds, Dr. Reid said, would mean knowing “literally what photons hit the retina, what information does the retina send to the thalamus and the cortex, what computations do the neurons in the cortex do and how do they do it, how does that level of processing get sent up to a memory center and hold the trace of that picture over one or two seconds.”

Then, when the same picture is seen a second time, “the hard part happens,” he said. “How does the decision get made to say, ‘That’s the one’?”

In pursuit of this level of understanding, Dr. Reid and others are gathering chemical, electrical, genetic and other information about what the structure of that part of the mouse brain is and what activity is going on.

They will develop electron micrographs that show every neuron and every connection in that part of a mouse brain. That is done on dead tissue. Then they will use several techniques to see what goes on in that part of the brain when a living animal reacts to different situations. “We can record the activity of every single cell in a volume of cortex, and capture the connections,” he said.

With chemicals added to the brain, the most advanced light microscopes can capture movies of neurons firing. Electrodes can record the electrical impulses. And mathematical analysis of all that may decipher the code in which information is moved around that part of the brain.

Dr. Reid says solving the first part of the problem — receiving and analyzing sensory information — might be done in 10 years. An engineer’s precise understanding of everything from photons to action could be more on the order of 20 to 30 years away, and not reachable through the work of the Allen Institute alone. But, he wrote in an email, “the large-scale, coordinated efforts at the institute will get us there faster.” He is studying only one part of one animal’s brain, but, he said, the cortex — the part of the mammalian brain where all this calculation goes on — is something of a general purpose computer. So the rules for one process could explain other processes, like hearing. And the rules for decision-making could apply to many more complicated situations in more complicated brains. Perhaps the mouse visual cortex can be a kind of Rosetta stone for the brain’s code.

All research is a gamble, of course, and the Allen Institute’s collaborative approach, while gaining popularity in neuroscience, is not universally popular. Dr. Wiesel said it was “an important approach” that would “provide a lot of useful information.” But, he added, “it won’t necessarily create breakthroughs in our understanding of how the brain works.”

“I think the main advances are going to be made by individual scientists working in small groups,” he said.

Of course, in courting and absorbing researchers like Dr. Reid, the Allen Institute has been moving away from its broad data-gathering approach toward more focused work by individual investigators.

Dr. Bock, his former student, said his experience suggested that Dr. Reid had not only a passion and intensity for research, but a good eye for where science is headed as well.

“That’s what Clay does,” he said. “He is really good in that Wayne Gretzky way of skating to where the puck will be.”

A version of this article appears in print on February 25, 2014, on page D1 of the New York edition with the headline: The Brain’s Inner Language.

Go to previous

Go to next

Loading...

ORIGINAL: NYTimes

By JAMES GORMAN 

FEB. 24, 2014