P32 - A Scalable System for Social Neural Imaging: Integrating Multi-animal 3D Pose Tracking and Calcium Imaging in Freely Interacting Animals

Introduction: : Understanding the neural basis of social and motor behaviors in freely moving animals necessitates experimental platforms that minimize human interference while maintaining high-resolution neural and behavioral monitoring. We present an integrated system combining endoscopic calcium imaging, multi-camera 3D tracking (DANNCE-based), and automated cable management. Our setup enables synchronized, chronic neural recordings in large arenas with single and interacting mice, supporting investigations into naturalistic behaviors such as exploration, interaction, and orientation-based modulation of sensory and motor cortical activity.

Materials and

Methods: : We developed a commutator-compatible tethering system designed to reduce torque and extend naturally as animals move, facilitating continuous GCaMPcalcium imaging with minimal external interference. Recording arenas were equipped with 6 synchronized and geometrically calibrated cameras to reconstruct 3D body poses during social interaction. We used adult male mice implanted with a small cranial window to enable long-term cortical imaging. A calcium indicator was expressed in either primary motor (M1) or visual cortex (V1), and neurobehavioral measurements were made while mice engaged in lone and social behaviors. 6-camera behavioral videos were processed using s-DANNCE, a recent computer vision tracking algorithm that reconstructs 3D body dynamics in social groups of animals. To investigate correlations between activity in cortical neurons and social behavior, we reconstructed each animal’s field-of-view to investigate visual-social encoding in V1, and we examined M1 representations of forelimbs in lone vs. social contexts. As an exploratory analysis, we examined correlations between principal modes of V1 and M1 population activity and the expression of discrete behavioral motifs identified via clustering of 3D pose kinematics. We also examined neural encoding of 3D joint angles.

Results, Conclusions, and Discussions:: Our system enables long-duration recordings with stable calcium signals and high-quality 3D tracking under unsupervised conditions. We observed patterns in V1 activity correlated with the position and behavior of an animal’s social partner, establishing proof-of-concept for future investigations of visual encoding of conspecifics and modulation by social context. In M1, we found population activity modes that correlated with spontaneous movement patterns, including transitions and fine-motor events. Some clusters enriched for grooming-like behaviors exhibited distinct neural activation profiles. Relative stability and high-movement epochs showed differing population dynamics, suggesting that M1 activity may reflect internal state or behavioral transitions. These observations were made using unsupervised behavioral analysis, highlighting the potential of this platform for identifying subtle neurobehavioral relationships and naturalistic conditions. This platform’s scalability opens the door to large-scale, data-driven decoding of behavior-linked neural dynamics across individuals and conditions. In future work, we aim to expand recordings for machine learning analysis, and integrate additional modalities—including multi-fiber photometry, optogenetics, and audio capture during social interactions—to enrich our understanding of multimodal neural-behavior coupling.

Acknowledgements and/or References (Optional)::