Mice find target sounds better after listening to the background
Trained mice detected target vocalizations more successfully after hearing the background for longer. A peer-reviewed experiment linked that improvement with neural representations in auditory cortex and the structure shared across frequency bands. Stimulating one inhibitory cell population helped or harmed detection depending on the background, revealing a context-dependent contribution within controlled mouse experiments.
Science··Morning
Listening time improved target detection
Mice trained to find vocalizations in noisy backgrounds detected them more successfully when they could first listen to the background for longer. The target was also encoded more accurately by populations of neurons in auditory cortex. The peer-reviewed experiment varied when the target began, separating the effect of preceding listening time from the sounds themselves. The improvement developed over several seconds, consistent with the animals integrating the background's statistical structure.[1]
Correlated frequency bands made targets easier to find
The animals learned a rewarded detection task with their heads held still. Researchers combined behaviour with neural recordings and light-based stimulation of selected cells. Stronger correlations between changes in different frequency bands improved detection. Stimulating parvalbumin-expressing inhibitory cells helped in low-correlation backgrounds but impaired performance when correlations were high. A separate inhibitory population expressing somatostatin provided another comparison. The same intervention therefore had opposite behavioural consequences depending on the acoustic context.[1]
Better target coding did not mean every sound improved
Background reconstruction did not improve comparably as listening time increased. Active and passive listening also differed in neural responses and pupil measurements; passive recordings helped reduce contamination from licking in decoding analyses. This single study concerns controlled sound textures and trained mice. It does not test freely moving animals in natural surroundings or human speech understanding, and it establishes no hearing treatment. Its circuit finding is specifically conditional on the background, without an unconditional benefit from changing inhibition.[1]
