Particle.news
Download on the App Store

Video and AI Decode Mice's Decision Strategies From Subtle Facial Movements

The peer-reviewed study reports facial cues carry information about latent cognitive states comparable to dozens of neurons.

Overview

  • Researchers recorded mice’s facial movements and brain activity simultaneously during a shifting-reward two-spout foraging task.
  • Machine-learning analyses decoded internal decision variables from facial kinematics with information content on par with neural population recordings.
  • Facial features reflected both active and unexpressed strategy variables, with components originating in secondary motor cortex.
  • Consistent facial patterns corresponded to the same strategies across different animals, indicating stereotyped cognitive signatures.
  • The authors present a non-invasive approach for studying brain function and urge consideration of regulations to protect mental privacy.