Scientists Read Mice's 'Thoughts' from Their Faces (fchampalimaud.org)

Researchers reported in Nature Neuroscience that subtle facial movements can reveal mice’s problem‑solving “thoughts”: by recording high‑speed video of the face together with population neural activity while mice solved a changing‑reward two‑spout task, the team used machine‑learning decoders to infer which decision strategies the animals were using. Remarkably, facial signatures carried as much information about the animal’s internal strategy as recordings from dozens of neurons, and similar facial patterns mapped to the same internal strategies across different mice — suggesting stereotyped, cross‑subject “thought” expressions, even when multiple strategies were represented simultaneously in the brain. Technically, the work links synchronized behavioral video and neural population data with supervised decoding methods to show that micro‑movements of the face mirror parallel computations happening in cortex. The finding opens a non‑invasive route to monitor internal cognitive states for basic neuroscience and potential diagnostic tools, because video alone could index latent neural representations. At the same time, authors warn this proof‑of‑concept raises ethical and privacy issues: ubiquitous cameras might reveal mental contents, creating a need for discussion and regulation as similar methods scale toward other species or settings.