Embodied reinforcement learning in the primate cortico-basal ganglia system
By recording neuronal activity across the macaque cortico-basal ganglia system, this study demonstrates that value representations are motor-system-specific rather than invariant, thereby challenging existing biological accounts of reinforcement learning and supporting embodied frameworks where behavior is grounded in an agent's physical structure.
Original paper dedicated to the public domain under CC0 1.0 (https://creativecommons.org/publicdomain/zero/1.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
The Brain's Two-Track System: Why Your Hands and Eyes Learn Differently
Imagine your brain as a super-smart video game console trying to figure out which buttons give you the most points. For decades, scientists believed this "point system" worked like a single, universal currency. They thought that no matter how you played the game—whether you were using a joystick, a keyboard, or a voice command—the brain calculated the value of a reward in one central place and then just sent that same "good job!" signal to the specific muscle group needed to act. This idea, known as reinforcement learning, suggested that the brain's reward calculator was completely separate from the body parts doing the moving. It was a neat, tidy theory: one brain, one value, many ways to move.
But what if the brain doesn't work like a single currency exchange? What if, instead, the way you learn is deeply tied to how you move? This is the heart of "embodied cognition," a concept suggesting that our thoughts and learning are grounded in our physical bodies, not floating above them in an abstract cloud. If this is true, then learning to grab a cookie with your hand might feel and look completely different in your brain than learning to look at a cookie with your eyes, even if the cookie is the same. Understanding this is crucial because it changes how we view everything from how animals survive in the wild to how we might build smarter robots or help people recover from brain injuries. If the brain treats different body parts as separate learning channels, it means our minds are far more physical and less "abstract" than we once thought.
The Great Monkey Experiment: Eyes vs. Hands
In this study, researchers Franco Giarrocco and Bruno Averbeck decided to test this idea by turning two macaque monkeys into super-learners. They set up a game where the monkeys had to choose between two pictures to win a juice reward. Sometimes, the monkeys had to make their choice by looking at the right picture (a saccade, or eye movement), and other times, they had to reach out and touch the right picture (a reach, or arm movement). The catch? The monkeys had to learn which picture was the "good" one (80% chance of juice) and which was the "bad" one (20% chance) just by trying, trial and error.
While the monkeys played this game, the scientists recorded the activity of 4,843 neurons across eight different brain regions. They looked at everything from the amygdala (often linked to emotions) to the prefrontal cortex (the brain's CEO) and the basal ganglia (the brain's action center). They wanted to see if the brain used the same "neural language" to talk about the value of a reward, regardless of whether the monkey was using its eyes or its hands.
The Big Discovery: Two Different Languages
The results were a shock to the old theory. The scientists found that the brain did not use a single, universal language for value. Instead, it spoke two completely different dialects depending on which body part was doing the work.
Imagine the brain's neurons as a massive orchestra. If the "universal value" theory were true, the orchestra would play the same melody whether the monkey was using its eyes or its hands, just maybe a little louder or softer. But what the researchers found was more like the orchestra playing two entirely different songs at the same time. When the monkey used its eyes, the neurons fired in a specific pattern to say, "This picture is worth a lot!" When the monkey used its hand, a different set of neurons fired in a different pattern to say the exact same thing.
Using a mathematical tool called "population geometry," the researchers measured the "angle" between these two neural patterns. They found that the patterns were almost at right angles to each other—like a "T" shape. In math terms, this means they are orthogonal, or geometrically independent. The information about value when using the eyes was so different from the information when using the hands that if you tried to read the "eye" signal using the "hand" decoder, you would get very confused. The value information was essentially lost in translation.
This separation happened everywhere in the brain, even in the deep, emotional parts like the ventral striatum and amygdala, which scientists used to think were the "universal value centers." The only place where the brain kept a little bit of a shared language was in the ventrolateral prefrontal cortex (vlPFC), a region known for handling abstract rules. But even there, the separation was strong.
Why It Matters: The Body Shapes the Mind
The study suggests that the brain is "embodied." It doesn't just calculate a reward and then tell the body what to do; the way the body moves actually shapes how the reward is learned and stored. The researchers found that monkeys were actually better at learning when they used their hands (reaches) than when they used their eyes (saccades), learning faster and making better choices. This hints that our different body parts might have evolved to handle different kinds of learning tasks.
Crucially, the researchers ruled out the idea that these differences were just because the monkeys were better at one task than the other. Even when they looked only at the sessions where the monkeys performed equally well with both eyes and hands, the brain still used two separate languages. This proves that the separation isn't a mistake or a side effect of being bad at a task; it's a fundamental feature of how the brain is wired.
So, the next time you are learning something new, remember: your brain might be treating your hands and your eyes as two different students in the same class, each with their own notebook and their own way of understanding the lesson. The brain isn't a single, abstract calculator; it's a physical machine where the way you move is part of how you think.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.