First-ever brain activity map captures the decision-making process

A new study recorded brain activity from 621,733 neurons in 139 mice as they made decisions, building the first activity map of a mammal’s entire brain at single-cell resolution.

The team saw where signals for seeing, choosing, moving, and getting feedback show up and how they unfold together.

The work goes beyond the usual hunt for a single decision center and shows a coordinated, brain-wide pattern. It also tests a simple idea about how past experience shapes a new choice without us noticing.

Brain map of activity

One of the researchers, Alexandre Pouget of the University of Geneva, helped lead an international group that trained mice to turn a small wheel to center a striped target for a reward. When the target was faint or absent, the mice leaned on their recent history to guess.

Across 12 labs, the group inserted 699 probes and sampled 279 brain areas while the same task ran everywhere.

They used Neuropixels to record spikes from hundreds of neurons per probe with millisecond precision, which made a fine-grained map possible.

“This is the first time anyone has produced a full, brain-wide map of the activity of single neurons during decision-making,” explained Pouget.

Activity and decisions in the brain

Vision signals appeared first in classical visual areas, then spread into midbrain and hindbrain regions that also carried choice information. 

Earlier work in mice had already hinted that choice, action, and engagement are distributed, not isolated in one cortical spot. The new map strengthens that view with coverage of hundreds of regions and clear timing.

Feedback mattered, too. When trials were correct and reward was delivered, wide swaths of the brain lit up together, reflecting both the liquid reward and the licking movements that came with it.

Expectations shape perception

A companion paper tested how prior expectations, built from recent trials, are encoded. Mice estimated the block’s left-right probabilities and performed better when the stimulus was weak, including zero-contrast trials.

The authors decoded signals for this subjective prior in 20 percent to 30 percent of recorded regions spanning early sensory areas like the thalamus, motor areas, and higher cortex. 

Crucially, prior signals were present even before a new stimulus appeared and were detectable very early after it arrived.

That result explains how expectations can nudge perception without crossing into conscious awareness.

What the data showed

Across the project, the team analyzed 75,708 well-isolated neurons with the highest quality metrics.

Just before a movement began, about 4 percent of neurons had activity correlated with the direction of the upcoming choice after controlling for stimulus and block.

Movement variables were strong predictors of activity, and reward delivery was one of the largest contributors to explained variance.

Those patterns tell us that action and outcome signals are broadly shared, while basic sensory signals stay more localized and brief.

Why the brain activity map matters

The map shows that decision-making is not a handoff from sense to thought to action. It looks more like many regions voting together, each carrying partial information that becomes aligned near movement and feedback.

That layout matters for medicine and technology. If priors and choices are distributed, measuring or modulating a single region may miss the story or cause side effects elsewhere.

The data come from head-fixed mice in a specific visual task, so the conclusions are about how mouse brains act under those conditions.

The authors also note that licking and other movements explain a large share of the variance during reward, which means the hedonic piece is tangled with motor signals.

The map is correlational, not causal. It shows where information is present, not which sites are required for a decision, and it sets up future causal tests.

Reading the technical terms

Neuropixels are high-density silicon electrodes that record spikes from hundreds of neurons along a thin shank. They offer single-spike timing, which is needed to separate fast sensory and motor events.

The thalamus is a relay hub that routes sensory signals to the cortex and receives strong feedback from the cortex.

Seeing prior signals there tells us that expectations are changing the initial formatting of incoming inputs, not only the later decision layers.

Mapping brain activity in humans

The dataset is public, and the analysis code is standardized. That enables other labs to check the findings, try new models, and test questions about learning, attention, and internal states.

A natural next step is to combine this map with targeted perturbations and to follow priors as they update in different contexts.

If the same network rules hold across tasks, that will sharpen theories of how brains implement probabilistic inference.

The study is published in Nature.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–