Neural integration of acoustic statistics enables mice to detect acoustic targets in noise.

Mice trained to detect vocalizations in sound textures show better auditory-cortex encoding and detection when they can sample acoustic statistics longer, especially with higher cross-frequency correlations.

• Why it matters: Natural backgrounds like rain and wind are statistically predictable sound textures, yet the neural computations that let animals use those statistics for target detection are unclear.
• What they did: Mice were trained on vocalizations embedded in textures with different statistical predictability while auditory cortex activity was recorded and PV interneurons were optogenetically modulated.
• The result: Longer sampling improved performance and neural representation, higher cross-frequency correlations boosted both, and PV activation helped detection for low correlations but impaired it for high correlations.

AI summary of the abstract · may contain errors · not medical advice · read the paper.