How diverse brain cells make decisions together

Princeton researchers explain how billions of neurons make unified decisions, revealing the hidden order of the brain’s confusing chat.

Your brain makes thousands of decisions every day, from that route to peak hours, whether the half-second gap is enough to dart on busy streets. Not only are these choices happening so quickly, but they happen completely. Billions of very different brain cells, each firing in its own seemingly random pattern, somehow coordinating to produce a single unified decision?

Princeton University scientists have solved the puzzle by staring at the monkey’s anterior cortex as they make visual decisions. The challenges they found were fundamental assumptions about how the brain works and revealed elegant solutions to one of the most confusing problems in neuroscience.

Hidden democracy of the brain

A research team led by Tatiana Engel of Princeton College of Neuroscience trained rhesus monkeys to determine whether the red or green squares are patterns of checkerboards. While monkeys were deliberating, scientists documented individual neurons in the dorsal anterior cortex, a brain region that translated decisions into actions.

Even in the same trial, each neuron responded completely differently. Some people steadily improve their activity, some fire in complex bursts, while others reduce ignition rates in unpredictable ways. This heterogeneity has long troubled neuroscientists. Universal wisdom shows that this chaos reflects the inherent complexity of decision-making itself.

But Engel’s team was completely suspicious of other things. “Imagine a group of skiers going down the mountain,” she explained. “Everyone likes slightly different paths, but all paths are shaped by the same slope beneath them.”

Using advanced computational modeling published in nature today, they found that despite their very different personal responses, all neurons are actually encoding the same underlying decision variables. The obvious confusion is simply the result of each neuron having its own unique “preferences” or adjustment function, i.e. how to represent shared information.

Neurological ski slope

The breakthrough comes from the development of a new mathematical framework that can track population dynamics and individual neuron preferences simultaneously. This reveals something amazing: Neurons all slide down the same decision-making pattern, but everyone is on their own path.

In easier decisions, this landscape is steep and decisive, like a neat ski slope that makes everyone quickly push towards the right choice. In harder decisions, the terrain becomes flatter and more susceptible to noise, increasing the chances of mistakes, such as trying to navigate among white people.

The main findings of the study include:

• All neurons have the same latent dynamics
• The brain uses a “attractor” mechanism whose single barrier can separate the correct and incorrect choices
• Simple decisions create steeper neural landscapes, leading to faster, more confident choices
• Single neurons have nonlinear tuning curves instead of simple gradient responses

Better predict choices than the brain itself

The model not only explains neural activity—it predicts monkey choices with obvious accuracy. Even if you are trained only in neural firing modes without any information about the actual selection of animals, in some cases it correctly predicts 90% of the time.

Is it more surprising? Unsupervised models outperform traditional machine learning methods that are explicitly trained to predict choices. This suggests that the team does identify decision variables in the brain, not just statistical correlations.

“Every decision is unique,” Engel notes. “But by mining the levels of a single trial and a single neuron, we can start to understand it.”

From chaos to clarity

The discovery addresses the long-standing debate in neuroscience about whether the brain uses drift diffusion dynamics, attractor mechanisms or footprint patterns to make decisions. Previous studies have drawn contradictory conclusions, partly because they cannot separate geometric representations from fundamental dynamics.

The study also shows that errors are not caused by completely different neural processes, but by the occasional push of neural noise in the wrong direction, just as skiers lose control of cold plaques.

This discovery has implications beyond basic neuroscience. Understanding how a healthy brain coordinates decisions can provide insights into diseases like schizophrenia or bipolar disorder, and the decision-making process can be problematic.

Princeton now plans to explore how different types of neurons and their connections facilitate the various tuning patterns they observe. Their next goal is to understand exactly how the wiring of the brain creates these elegant computing solutions.

It seems like neural confusion is actually a complex democracy – most professional voices contribute their unique perspectives to reach the same collective decision, each following the mountain of their choice.

Related