ANYONE WHO STEREOTYPES video gaming as the pastime of slackers might be surprised by how Princeton professor David Tank and his research team delve into the neuroscience of navigation. Two floors below the entrance to the new Princeton Neuroscience Institute (PNI) building, behind a heavy black curtain, lies a virtual-reality game fit for a mouse. During a typical experiment, researchers project a maze, similar to what appears in 1990s-era video games, onto a small curved screen. The mouse, alone in its very own IMAX theater, becomes the star of the game, navigating the moving maze by scuttling on a Styrofoam sphere about the size of a bowling ball. As the mouse turns every which way, the ball follows, while the animal’s brain is viewed using a specialized microscope.
The microscope zeroes in on specific groups of cells. A favorite target is a type of neuron in the hippocampus region of the mouse’s brain that fires when the animal is in a particular location in its environment. The readout appears on a nearby computer screen: a flurry of small white circles, each a single nerve cell, lighting up and going dim, each to the beat of its own drum. Now, for Tank’s team, comes the fun part: translating the dynamic patterns of nerve-cell firing into the mouse’s sense of place as it moves through the maze.
Scientists have known about the importance of these so-called place cells for decades. In fact, the researcher who discovered them, John O’Keefe of University College London, was just awarded a share of the Nobel Prize in Physiology or Medicine. Yet it is only recently that neuroscientists like Tank have demonstrated that it’s possible to record patterns of chemical activity across a whole network of these neurons at once — a finding that ultimately could help us understand motion, navigation, and other complicated mental activities.
Neuroscience is undergoing a transformation, enabled by new tools and new collaborations. The field has been hot for some years, but since 2013, it has been positively sizzling. The European Commission debuted its 10-year Human Brain Project. Similar efforts emerged in Japan and other nations. In the United States, President Barack Obama announced his BRAIN Initiative, a major research effort aimed at exploring how the brain works. BRAIN stands for Brain Research through Advancing Innovative Neurotechnologies, and Princetonians are working at the center of the effort. Princeton might not have a medical school, but it is drawing upon two traditional strengths: fundamental experimental research and cutting-edge theoretical work about the brain and mind. Tank, a physicist by training and a co-director, with Jonathan Cohen, of the PNI, helped plan the BRAIN Initiative’s scientific strategy. He compares the current state of affairs to the first times astronomers peered at the sky through a telescope. “My graduate students are doing experiments today that I only dreamed about doing 10 years ago,” he says.
Ever since humans have understood that the brain is the body’s information processor, they have sought answers about how it directs everyday activities: behaviors such as forming new memories, making decisions, and planning things. For all the current excitement about neuroscience, nobody is saying that researchers are close to figuring out the brain, or that cures for diseases like schizophrenia are in the offing. Still, scientists now realize that their questions are within reach of being answered. It’s like being a physicist or chemist at the turn of the 20th century, explains Cohen, a psychiatrist and psychologist. “There were principles that were starting to reveal themselves that were beautiful and cool, and everyone knew that if they committed to understanding things, there would be useful applications in the future,” even though they couldn’t necessarily envision nuclear power or the Internet.
The human brain is a fiendishly complicated organ, a three-pound mass of tissue containing roughly 86 billion neurons. Scientists understand some levels of its organization much better than others. At the highest levels of complexity are the large sections of the brain — the familiar names in anatomy textbooks, such as the cerebellum or the cerebral cortex. Students who take introductory psychology are almost always told the tale of Phineas Gage, the 19th-century railroad worker who lost much of the frontal lobe of his brain in a freak accident, reportedly becoming more impulsive and far less sociable as a result. How much the damage truly affected Gage’s personality remains a matter of debate, but there’s no disputing that his story is about connecting a large brain region to human behavior.
Inside the three-pound mass are the molecules that relay signals from one neuron to another, traversing the microscopic gaps, or synapses, between nerve cells. In the early and mid-1900s, scientists began identifying and understanding these messenger molecules: acetylcholine, dopamine, and many others. Psychiatric drugs as we know them today would not exist without this level of brain understanding.
It is the levels of brain organization in the middle that have remained the biggest mystery. Networks consisting of anywhere from tens of thousands to 1 million of the brain’s neurons work together in navigation, in vision, or in decision-making. Neuron networks are wired together with synapses to form what neuroscientists call circuits. One estimate puts the number of connections in the brain at a staggering 100 trillion. But the electronics analogy of wires and circuits only goes so far in a living brain — the “wiring” connections have varying strengths, and they might change as a result of a new experience. Neuroscientists use words like “mind-blowingly complex” to describe this fluid connectivity.
This middle level — the brain’s networks — is the focus of the BRAIN Initiative, and the focus of neuroscience’s snazzy new toolkit. These new technologies are monitoring nerve-cell networks, manipulating them, and mapping their connections. And importantly, the tools aren’t generating data in a vacuum — they are feeding information to ever-improving models that attempt to link brain activity to human behavior. (See article page 38.)
The Tank lab’s video-gaming mouse is the star of a monitoring experiment: watching what circuits of nerve cells do during a behavior or action and trying to discern patterns. The nerve cells going off like flashbulbs under the microscope contain an important tool that debuted in the mid-2000s — a glowing protein called GCaMP. This protein glows brightly when calcium floods into nerve cells, which happens every time a nerve sends a signal. In other words, the protein helps neuroscientists watch a proxy of nerve-cell chatter — it’s not quite the chatter itself, but it’s close enough to be useful. With some genetic tricks, the lab can selectively produce the luminescent protein only in the neurons it wants — for example, those neurons that control navigation. The end result is like being able to eavesdrop on a specific conversation amid the din of the brain’s proverbial cocktail party. The tool works in several of the animals that neurobiologists study, including mice, fruit flies, and roundworms, so labs worldwide use it in their experiments.
This is a sea change from traditional neuroscience approaches. Terrence Sejnowski *78, who, like Tank, is both a physicist by training and an architect of the BRAIN Initiative’s strategy, worked a postdoctoral stint surrounded by colleagues studying one neuron, or a handful of neurons, at a time. Animal experiments relied on injected chemicals or tiny electrodes to follow changes in electrical charge in active neurons. Though tremendously exciting at the time, the approach had limitations, explains Sejnowski, a professor at the Salk Institute in California. “Imagine you had to look at the world through a soda straw,” he says. “You can move the straw around, but you can only see a tiny piece of what’s going on. That’s what neuroscientists experienced by studying one neuron at a time.”
The picture wasn’t much different when Mala Murthy was an MIT undergraduate in the 1990s, nearly 20 years after Sejnowski’s postdoc days. Murthy, today an assistant professor at Princeton, says that even though electrical probes can be powerful, they lack the resolution to associate brain structure with brain function. The most dextrous surgeon can’t consistently poke the same neuron with an electrode, but the neuroscientist armed with genetic techniques will know exactly what neurons or circuits she’s watching. Murthy’s domain is the fruit-fly brain, and she says the glowing GCaMP protein has made it easier for her lab to determine, for example, how fruit flies sing courtship tunes to potential mates.
If monitoring neural circuits is like eavesdropping on cocktail-party conversations, then manipulating circuits is like butting in, changing the subject, and seeing how guests react. Ilana Witten ’02 uses a technology called optogenetics for manipulating nerve-cell activity, which was pioneered by her postdoctoral adviser, Karl Deisseroth of Stanford. Now back at Princeton as an assistant professor, Witten and her lab prod specific types of neurons with light to learn how a reward, such as a tasty treat, might motivate a rodent to start a good behavior or break a bad habit. The approach is powerful, she says, because it allows her team to manipulate neurons quickly, at the same speed at which they’d normally converse. Many labs have begun to use this approach in animals to investigate the underpinnings of epilepsy, addiction, depression, and other conditions, though many hurdles would have to be overcome for the technique to be used in humans.
When Witten is manipulating nerve circuits, she focuses on a particular area in the brain — for example, a reward center called the ventral striatum. She uses genetic tricks to label particular neurons in a living rodent’s ventral striatum with specialized light-sensitive proteins. A slender fiber-optic cable delivers bursts of light at the appropriate wavelength to the animal’s brain. When triggered by the light, the special proteins generate nerve impulses on demand. Along with Tank and other Princeton researchers, Witten has been awarded a share of the first round of BRAIN Initiative funding to study short term “working” memory, which employs some of the same brain circuitry as reward behavior.
For all the talk of networks or circuits of neurons, scientists have only a piecemeal understanding about how all the neurons in the brain are connected to one another. Mapping the human brain’s connections is so complex that it has its own major research initiative, the Human Connectome Project. The goal is to build maps of the functional connections in human brains with the help of brain-scanning technologies. The key word is “functional” — just because neurons are anatomically connected doesn’t guarantee they work together.
If there is a mapping technique that has whipped researchers into a frenzy, it is Clarity, the method that allows scientists to peer into an entire animal brain — removed from the body and no longer living — and generates stunning three-dimensional views of nerve-cell connections. Neuroscientists used to have no choice but to digitally piece together images from microscopically thin slices of a brain to get that kind of information.
Like Witten before her, Christina Kim ’11 works with Karl Deisseroth at Stanford, who announced Clarity to great fanfare in 2013. The aptly named Clarity technique renders brains transparent: It replaces the brain’s fatty molecules, which block the passage of light, with a clear, mesh-like material that holds the brain together. The result is a three-dimensional, transparent brain with all its neurons in place, which can be tagged with markers for specific nerve-cell types and reused again and again. In other words, Clarity is an opportunity to examine the entire brain at once and see that crucial middle level of brain organization: neuron networks. “The simplicity of Clarity — the ability to do a visual reconstruction of the whole brain — is something everyone can appreciate,” Kim says.
Clarity has yet to be performed on an entire human brain because the relatively large volume slows down the process. And it’s a stretch to claim that Clarity can map each neuron’s connectivity in the brain. With glowing labels, Clarity can map where certain nerve-cell types reside in the brain, and then trace connections between those groups.
The technique’s potential has not been lost on the U.S. Department of Defense, which is interested in understanding how traumatic brain injury and extraordinary stress in soldiers affect nerve circuitry. The department’s Defense Advanced Research Projects Agency (DARPA) aims to speed up the Clarity process and to combine its mapping ability with information from other technologies. To accomplish that goal, the agency is funding a project called Neuro-FAST — which stands for Neuro Function, Activity, Structure, and Technology.
Both Kim, who conducted her thesis research with Tank, and Daniel J. O’Shea ’09 are involved in the project. With Stanford professor Krishna Shenoy, O’Shea studies how neurons in the brain’s motor cortex produce the patterns of activity that drive movement. “Clarity and other new tools open a new class of experiments up,” says O’Shea, “but you still need to think of the right questions to ask.”
That, says PNI co-director Cohen, is where theoretical work comes in. “Neuroscientists are getting excited about all the new things they can measure, and that is fantastic,” he says. “If you think about the brain as a massively complex computer, you have to know about transistors and circuits to understand it. On the other hand, it’s not just how the transistors are connected, but it’s the computer’s overall architecture, and the software it runs.”
The experimental monitoring, manipulating, and mapping become even more powerful when guided by a testable theory — about how brain function gives rise to mind and behavior. “If you think about what science is, it’s about having a model, a hypothesis, that you can test,” Cohen says. “What that says to me is that neuroscience needs theoreticians side-by-side with the experimentalists.”
Many psychology and neuroscience researchers, including Cohen and Sejnowski, are employing mathematics and computer science to develop theories about how the brain links to behavior. Professor Matthew Botvinick, for example, is one of a cadre of researchers using computer models to recognize patterns of brain activity in collections of whole brain images obtained by placing living, thinking human volunteers inside a specially configured MRI scanner. His team is getting better and better at analyzing the scans obtained with this technique, known as functional magnetic resonance imaging, and figuring out what words people have in their minds, such as the words “furniture” or “table,” based on what the volunteer was thinking about during the brain scan. This kind of elementary brain-reading exercise is very far from practical applications, but it is already useful for researchers determining how patterns of neural activity encode language, whether word-by-word or letter-by-letter. “Now that experimentalists have handed us a telescope,” just as Galileo got centuries ago, “we can ask burning questions our computer models have been feeding us for years,” Botvinick says.
“Thirty or 40 years ago, scientists didn’t know what the best approach was to break down the brain,” Cohen says. They tended to split off into psychology, studying behavior; or neuroscience, studying the brain. “And rarely did the twain meet,” he adds. The cultural divide between psychology and neuroscience has been softened, Cohen says, but it isn’t totally gone. PNI’s gleaming new building connects to the Department of Psychology’s new home, and is steps from the departments of molecular biology, physics, and chemistry. That, Cohen says, is by design. “We want the Princeton Neuroscience Institute to embrace everyone’s tools,” he says, “whether they come from psychology, biology, engineering, or physics.”
Back at Princeton’s mouse virtual-reality headquarters, physics graduate student Alexander Song and imaging specialist Stefan Thiberge are developing a next-generation microscope. Song points out that the movie of flashing neurons, while unimaginable 10 years ago, keeps tabs on only a few hundred nerve cells at a time. That’s just a fraction of the tens of thousands of neurons that likely are involved in controlling a behavior. So if Song can increase the number of neurons that can be imaged at once, and increase the number of areas in the brain that can be monitored, he will have more comprehensive data. Also coming online is an experiment that not only will monitor mouse-brain circuits, it also will manipulate circuits to test whether they truly do control the behaviors scientists think they do.
“Ideally, you’d like to monitor a whole brain at once, while manipulating one neuron at a time,” Song says. “For now, that’s still a pipe dream.”
Tank and his fellow neuroscientists have bigger dreams still: “If we could do that mouse-brain movie in the human brain,” Tank says, “that would be revolutionary.”
Carmen Drahl *07 is a Washington, D.C.-based writer. She most recently worked at Chemical & Engineering News.
Crowd-Sourced Science: Sebastian Seung