An animal foraging for food is engaged in a complex array of mental calculations. It needs to evaluate the smells and sounds in its environment, balancing signs of food with the possibility of predators. It has to recall past foraging jaunts to note which spots have been fruitful or dangerous in the past. It has to weigh the potential reward against risks, given its current state of hunger. And, finally, it has to formulate and execute a plan of action, darting across an open field to grab a desired morsel.
Such computations unfold simultaneously in millions of neurons that interact both locally and across multiple brain regions. Individual neuroscience labs have gained glimpses of how different aspects of cognition function, mostly within specific brain structures. But how multiple brain regions coordinate and interact to produce behavior is still largely a mystery.
Understanding how the brain produces thought requires approaches that are beyond the scope of a single laboratory. The Simons Collaboration on the Global Brain (SCGB) aims to tackle multidimensional problems such as these by bringing together researchers with the diverse expertise needed to decipher the complexities of the brain.
In 2017, the SCGB launched 20 new projects in which teams of experimentalists, theoreticians and computational experts explore some of the biggest questions in neuroscience. The largest of these is the International Brain Laboratory (IBL), a collaboration among 21 labs in four different countries. The scale of the IBL breaks new ground for a field that has typically been the domain of individual labs or a few labs working together.
“The deep problem of how activity across the entire brain produces cognition through neural coding and dynamics is a difficult one that is unlikely to be solved by any single lab,” says David Tank, director of the SCGB. “The sheer scale of the number of distinct areas to be investigated and diversity of skills required to do these kinds of experiments — expertise in behavior, animal training, imaging and optics, electrophysiology, and statistical analysis of data — favors collaborations between multiple labs.”
Perhaps the most profound impact that the IBL will have on the field is in providing a framework for standardizing and sharing data.
The IBL, jointly funded by the Simons Foundation and the Wellcome Trust, will focus its collective talent on the process of decision-making, starting with a task in the lab that mimics foraging. Different labs have expertise in specific brain regions and will record neural activity from those regions using a variety of tools, such as electrophysiology and calcium imaging, as a mouse makes a decision. To get a more cohesive picture of how the decision-making process works across the brain, the team will synthesize data from different groups, illuminating how information is transformed as it is transmitted from place to place. Ultimately, the group hopes to reveal how the brain integrates information from the environment, past experience and the animal’s internal state to arrive at the most appropriate action, an objective impossible to achieve by studying individual parts of the brain in isolation.
That’s an ambitious goal. To collate and compare data from different groups, researchers will need to train mice in four countries to perform exactly the same task. They’ll need to standardize the vast amounts of data they collect and figure out how to share it, a challenging prospect for a field that has little infrastructure for such things. Though none of these issues are new to neuroscience, the IBL and other large projects are forcing neuroscientists to tackle them.
“I think the IBL, and the Global Brain collaboration in general, is heralding a partial shift in culture, providing a laboratory for how to make larger-scale collaborations work,” says Loren Frank, a neuroscientist at the University of San Francisco and an SCGB investigator. “Creating groups of people who are there to work as a team and need to share data focuses energy on actually solving the problem.”
One of the first major challenges the IBL faced was choosing what behavioral task to use in its experiments. With input from theoreticians in the collaboration, the group decided to focus on decision-making, in part because it already has a strong theoretical framework.
Researchers developed a task in which mice rotate a wheel according to the detection and position of a visual cue. The reliability of the visual cue can vary, mimicking the complexity of real-world decisions. “It’s rare that you’d have all the facts on hand when making a decision,” says Alexandre Pouget, a computational neuroscientist at the University of Geneva and an investigator with the IBL. He and others have developed computational methods for analyzing these types of decisions, which they will apply to the IBL experiments.
Researchers in different labs across two continents now have the task up and running. The next step is to monitor activity across large populations of neurons as animals make decisions. They’ll use a variety of methods, including a new electrode recording technology called Neuropixels, which can simultaneously record from hundreds of cells in different parts of the brain.
Each of the IBL’s decision-making experiments will produce reams of data: terabyte-scale records of neural activity, behavior and other factors. One of the biggest challenges the project faces is how to make these data easily accessible to other labs. Indeed, data sharing is a huge issue for the field as a whole and only grows more urgent with the development of new data-intensive techniques.
“Creating groups of people who are there to work as a team and need to share data focuses energy on actually solving the problem.”
“Data sharing in neuroscience is rare and primitive,” says Liam Paninski, a neuroscientist at Columbia University and an investigator with the SCGB and IBL. Perhaps the most profound impact that the IBL will have on the field is in providing a framework for standardizing and sharing data.
The volume of data that IBL researchers are collecting is too large to share in its raw form, so Paninski’s team is developing ways to process data without losing important details, isolating essential signals from calcium-imaging and electrophysiology data so that they can be transferred to the cloud. The IBL plans to eventually make the data public so that theorists around the world can probe them for insight. Paninski hopes the tools his team develops will be adopted much more broadly than just within the IBL. “We want to develop solutions so that no one else has to worry about these problems,” he says.
The data the IBL collects will be stored on servers at the Simons Foundation’s Flatiron Institute, whose Scientific Computing Core has expertise in handling large datasets from high-energy physics, astrophysics and biology. The Flatiron’s neuroscience group is also developing new tools for processing large volumes of data, particularly for electrophysiology and calcium-imaging experiments of the type used in the IBL.
Less than two years in, the IBL has only just begun its efforts. The most exciting outcomes — the first scientific results — are expected soon. But how the project solves data-sharing and other problems could be equally important for the field, providing a model for how to work closely with many labs. “People are reaching out to us all the time about how we use these tools,” says Anne Churchland, a neuroscientist at Cold Spring Harbor Laboratory and an investigator with the IBL. “I’m glad we have the opportunity to be leaders in this field.”