Neuroscientist Kauê M. Costa redefines how the brain learns

The work of neuroscientist Kauê M. Costa isn’t just refining our understanding of dopamine—it’s quietly upending decades of dogma about how learning happens in the brain. For years, dopamine has been framed as the brain’s primary reward signal, a chemical messenger that reinforces behaviors by predicting future rewards. But Costa’s research suggests a far more nuanced role, one where dopamine’s influence is not just about pleasure or motivation but about how the brain dynamically adjusts its expectations in real time. This isn’t a minor tweak to an existing theory; it’s a fundamental challenge to the canonical model of reinforcement learning, which has shaped everything from AI algorithms to clinical treatments for addiction and Parkinson’s disease. What makes Costa’s findings particularly compelling is their origin in unexpected experimental results. Traditional models assume dopamine surges when rewards exceed expectations, but his lab has documented cases where dopamine drops in response to better-than-expected outcomes—suggesting that the brain’s learning mechanism is far more adaptive, even counterintuitive, than previously thought. This could explain why some people persist in behaviors that don’t seem to deliver tangible rewards, or why certain neurological conditions resist standard treatments. The implications stretch beyond neuroscience into fields like machine learning, where reinforcement learning models—often inspired by dopamine’s role—might need to be overhauled. The big question now is whether these findings hold up under broader scrutiny. If dopamine’s role is as fluid as Costa’s data suggests, how does that reshape our understanding of habit formation, addiction, or even psychiatric disorders like depression? And if the brain learns by constantly recalibrating expectations rather than chasing rewards, what does that mean for therapies designed to modulate dopamine pathways? Early responses from peers have been cautiously optimistic, but replication and deeper exploration will be critical. What’s clear is that Costa’s work sits at the intersection of two growing trends: the push to move beyond simplistic reward-prediction models in neuroscience, and the increasing recognition that the brain’s plasticity is far more sophisticated than once assumed. If validated, his research could force a paradigm shift—not just in how we study learning, but in how we treat conditions where that process has gone awry.