When a mouse brushes its whiskers against a virtual texture in a carefully controlled chamber, the decision about what it feels does not wait for signals to climb a one-way ladder to some distant executive centre. Instead, the primary sensory region itself begins to reflect the animal's accumulating choice. That finding, published in the Proceedings of the National Academy of Sciences, quietly dismantles a textbook assumption about how brains work.
Researchers trained mice on a tactile perceptual decision task inside a whisker-guided virtual reality setup. Dense electrophysiological recordings from the whisker region of the primary somatosensory cortex, often abbreviated as S1, captured neural activity as the animals gathered evidence. What they saw was striking. The cortical signals collapsed to a single latent variable that tracked the gradual accumulation of noisy sensory evidence toward a decision bound.
This activity was not merely sensory. The same region also showed categorical, all-or-none coding of the final choice. Such categorical signals, the team concluded, arrive through cortico-cortical feedback loops from higher areas rather than arising solely from within S1. The paper, received on 1 June 2025, accepted on 10 February 2026 and published on 29 April 2026, therefore points to rapid bidirectional communication instead of the purely feedforward hierarchy taught for decades.
The result matters because it replaces a tidy, sequential diagram with something messier and more lifelike. Decisions emerge from dynamic loops that knit early sensory machinery into the very process of judgment. A related press release summarising the work appeared on 13 July 2026.
Yurii Vlasov, professor of electrical and computer engineering at the University of Illinois Urbana-Champaign, sees broader implications. "We want to learn from a billion years of evolution. How is that biological intelligence organised 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? In the level of decision-making, that is where current AI is lacking."
He adds that the neural code remains largely undeciphered yet this systems-level view could reshape artificial networks. "The neural code of the brain is still mostly an unknown language. But this systems-level understanding can be viewed as a potential impact on how more efficient artificial neural networks can be built."
