Map of 600,000 brain cells rewrites the textbook on how the brain makes decisions

Researchers have completed the first-ever activity map of a mammalian brain in a groundbreaking duo of studies, and it has rewritten scientists' understanding of how decisions are made.

The project, involving a dozen labs and data from over 600,000 individual mouse brain cells, covered areas representing over 95% of the brain. Findings from the research, published in two papers in the journal Nature, suggest that decision-making involves far more of the brain than previously thought.

The mammoth project was led by the International Brain Laboratory (IBL), a collaboration of experimental and theoretical neuroscientists from across Europe and the U.S. These scientists were united by a familiar, nagging feeling.

"We had a problem with the way science was done," said Matteo Carandini, a neuroscientist at University College London and a core member of the IBL.

In previous studies of the brain, many separate labs set out to answer big questions about the organ, exploring how brain activity relates to behavior, for instance. However, each lab studied this question in different mice's brains, and performed slightly different behavioral tasks with each set of rodents. Once you added in uncertainties around how each research group defined distinct regions within the brain, these inconsistencies muddied the results.

"We wouldn't know whether we actually agree or disagree, because so many things were different," Carandini told Live Science.

Related: Most detailed human brain map ever contains 3,300 cell types

So the IBL came together to design a single, robust, standardized experiment on a scale that no individual lab could tackle alone. They then paired this megatest with precision brain measuring tools and preset analysis methods to make the results as reproducible as possible. The aim of the experiment would be to overcome an enduring obstacle in the field.

"One of the longest-standing challenges in neuroscience is to decipher how variation in neural systems — both structural and functional — maps onto variation in behavior," Federico Turkheimer, a neuroscientist at King's College London who was not involved in the study, said in a statement to the U.K. Science Media Centre.

This project ultimately included 139 mice, spread across 12 labs around the world, that were implanted with brain-recording devices called Neuropixels probes. The probes can record up to 1,000 individual neurons simultaneously. The researchers tested the mice with a simple behavioral task that each of the dozen labs could reliably replicate: Researchers placed mice in front of a screen, and a black-and-white striped marker would flash either on the right or left. If the mice moved a small wheel in the same direction as the flash, they received a reward.

Based on what you'd read in a neuroscience textbook, said Carandini, you'd expect the brain activity that occurred during the experiment to follow a linear path. First, cells in the visual cortex that recognize images would fire up, followed by neurons in a different part of the brain, such as the prefrontal cortex, known to be involved in abstract decisions. This information might then be combined with additional activity that represented the mouse's prior experiences — in other words, memories — before being sent to motor regions of the brain that control muscle responses.

The researchers' findings supported some of this chain reaction; the visual cortex was the first thing to activate, for example. Yet other findings clashed with the team's expectations.

"We found decision signals and signals related to the prior information in way more brain regions than we might have thought," Carandini said. Taken together, the activity across nearly all of the brain regions studied could be used to deduce whether or not the mouse had received a reward.

In some of the experimental trials, the researchers made the on-screen marker incredibly faint, so the mice essentially had to guess which way to move the wheel. The second Nature paper focused on how the mice used prior expectations — based on where the marker had been in previous tests — to inform their guess. The brain activity that flashed up when the mice guessed in these tasks was also far more widely distributed in the brain than the team anticipated it would be.

The IBL modeled its approach to understanding the brain on similar initiatives, such as the particle physics experiments conducted at CERN or the Human Genome Project's work to understand our DNA. To describe the project's impact, Carandini reaches for yet another field: astronomy.

He noted that the earliest astronomers could look up at the night sky and see every star, but in very poor detail. With the advent of the telescope, individual celestial bodies could be explored. Previous work in neuroscience, he said, was "as if somebody had pointed a telescope only to one galaxy, and then different astronomers had pointed their telescopes at different galaxies, and said, 'My galaxy does this!' or 'No, my galaxy does that!" The new project, he explained, was like being able to view all the features of the night sky at once and up close.

Such work has only been possible with recent technological advances and improved collaboration across labs, but Carandini hopes that it can now be used to address other big questions about the brain. The current paper's findings are only correlational, so it is currently not possible to say whether the observed brain activity directly causes a decision to be made or is only associated with the process.

"I think that's the next frontier," he said, "is to add causality to the study."