Scientists at the University of Illinois Urbana Champaign have discovered that our brains start making decisions much earlier than researchers ever suspected — and this surprising finding could help engineers build smarter, greener artificial intelligence systems.
Led by Professor Yurii Vlasov from The Grainger College of Engineering, the team published their findings in the journal Proceedings of the National Academy of Science (PNAS). Their work challenges a long-held view about how the brain works, and it points to ideas that could change how we build AI.
For decades, scientists believed the brain processed information in a simple, one-way sequence. According to that model, sensory information — what you see, hear, or feel — travels upward through increasingly complex brain regions until it reaches the front of the brain, where decisions finally get made. Vlasov and his colleagues suspected this picture was incomplete.
To investigate, the researchers recorded neural activity in mice as the animals navigated a virtual reality corridor and made simple decisions. They focused on the primary somatosensory cortex, or S1 — one of the brain's earliest sensory processing areas, located close to where sensory information first enters the brain.
What they found surprised them: S1 showed clear decision-related activity. This suggests that the brain starts making decisions much earlier in the process than traditional theories propose.
The key insight came from noticing something called feedback loops — channels through which information flows back downward from higher brain regions to earlier ones. Rather than a strict hierarchy where only the frontal cortex decides, the team found that decision-making involves constant back-and-forth communication across multiple brain areas.
"We want to learn from a billion years of evolution," Vlasov said. "How is that biological intelligence organized architecturally? Can we learn from the architectural side of the brain and emulate that to make AI more effective, less power hungry, and more intelligent than it currently is?"
This matters because today's most powerful AI systems use enormous amounts of energy to function. The human brain, by contrast, performs incredibly complex tasks while sipping power — roughly the equivalent of a dim lightbulb. If researchers can understand how biological intelligence achieves this efficiency, they might be able to build AI systems that do the same.
The researchers are quick to caution that this study does not provide a ready blueprint for new AI. Instead, it offers a fresh framework for thinking about how intelligence — both natural and artificial — might be organized. Next, they plan to examine the precise timing of brain signals and develop new tools for measuring neural activity.
"By looking at the fast temporal dynamics of neural activity, maybe we can understand better how these feedback loops are engaged in making decisions," Vlasov said. "Maybe that can be implemented in new architectures for AI."
In other words, by studying the oldest problem in the known universe — how the brain works — scientists may finally unlock smarter machines that use far less energy to run.