Back

Astrocytes and neurons encode natural stimuli with partially shared but distinct composite receptive fields

Lu, S.; Schultz, S. R.; Kozlov, A. S.

2025-10-07 neuroscience
10.1101/2025.10.06.680791 bioRxiv
Show abstract

Astrocytes are increasingly recognized as active participants in sensory processing, but whether they show selective responses to stimulus features, analogous to neuronal receptive fields, is not yet established. To address this, we used two-photon calcium imaging in the auditory cortex of anesthetized mice during presentation of natural ultrasonic vocalizations. Our aim was to compare astrocytic responses with those of neighboring neurons and to determine whether astrocytes exhibit feature-selective receptive fields. Event detection showed that astrocytic calcium activity is highly heterogeneous, but only a minority of events were consistently stimulus-linked. To examine this stimulus-driven subset, we estimated receptive field features using maximum noise entropy modeling and compared them with those of concurrently recorded neurons. Despite qualitative similarities in receptive-field features, analysis of modulation spectra and principal angles showed that astrocytic and neuronal receptive fields overlap but occupy distinct regions of feature space. This indicates that astrocytes and neurons encode partially shared, but not identical, dimensions of the sensory stimulus. Our findings indicate that astrocytes encode diverse sensory features, providing an additional contribution to neuronal encoding. This suggests that astrocytic calcium activity is not simply a reflection of neuronal firing, but instead represents a distinct component of cortical sensory processing. New and noteworthyWe used two-photon imaging to record calcium activity in astrocytes and neighboring neurons during presentation of natural ultrasonic vocalizations. We show that astrocyte activity is highly heterogeneous across spatial and temporal scales. Further analyses indicate that a subset of astrocyte calcium activity is stimulus-linked and encodes dimensions of the stimulus that partially overlap but are not identical to those encoded by neurons.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.