Particle.news

Download on the App Store

Retrosplenial Cortex Simultaneously Encodes Competing Spatial Hypotheses

This finding shows the retrosplenial cortex performs probabilistic reasoning through separate neuronal patterns that hold different location hypotheses until one is confirmed.

New research finds a brain region critical for navigation uses distinct neural activity patterns to encode multiple hypotheses that help distinguish between ambiguous landmarks.

Overview

  • The study published in Nature Neuroscience demonstrates for the first time that the retrosplenial cortex holds concurrent neural representations of multiple spatial hypotheses as mice navigate ambiguous cues.
  • Researchers used a circular arena with 16 ports and identical light dots that only appear when mice are nearby to force reliance on short-term spatial memory rather than scent or dead reckoning.
  • Neural ensemble recordings revealed distinct low-dimensional activity patterns corresponding to individual hypotheses that collapsed into a single representation when sensory information clarified the correct port.
  • The work builds on computational models by Ila Fiete showing artificial recurrent neural networks develop similar dynamic attractor states, suggesting a conserved computational principle for spatial reasoning.
  • Researchers now plan to examine how the prefrontal cortex and other brain regions support naturalistic navigation and decision making beyond laboratory tasks.