Basis logotype

Collaborative Intelligent Systems

Challenge Project

Collaborative Intelligent Systems

Studying animal social behavior in natural habitats

Multimodal sensing and modeling tools for studying social behavior in real-world animal groups: inferring social values/preferences, and how environments shape cognition, coordination, and affect.

Code Contact

Overview

Collaborative Intelligent Systems studies how intelligent behavior emerges in groups: How animals communicate, and how their movements reflect their preferences relative to the environment and other animals. We work with data from many species and environments, including urban rodents and birds, and collaborator datasets from other species.

We build systems for collecting high-resolution multimodal data, and computational tools for inferring strategies, testing hypotheses in simulation, and reasoning about social behavior in real-world ecological settings. A central goal is to understand not only what animals do, but what their behavior reveals about cognition, social values/preferences, stress, and adaptation.

Current Work

Rats

We study social communication, coordinated movement, and urban adaptation in New York City rats using thermal video, ultrasonic audio, field observation, and computational models.

Generalizable modeling framework

We develop species-general tools for inferring latent strategies, preferences, and interaction rules from trajectories, testing models in simulation, and reasoning about how local decisions produce group-level behavior.

Multimodal field systems

We build hardware and software for making sense of multimodal recordings of animal social groups, including video, ultrasonic and audible audio, environmental context, and 3D structure.

Partnerships

We partner with academic groups, municipal partners, nonprofits, and for-profit companies on projects where better measurement and modeling of animal social behavior can support research, policy, management, or welfare. Please reach out at contact@basis.ai if you are interested in working together.

Publications

Recent publications and preprints.

Code

collab-creatures

Open-source tools for analyzing animal collaboration with Bayesian and causal inference.

collab-environment

Integration package across projects for representing, modeling, and simulating behavior within 3D environments; code for Aitsahalia et al., “Inferring cognitive strategies from groups of animals in natural environments,” presented at the NeurIPS Workshop on Data on the Brain & Mind Findings, 2025.

Team and Contributors

Ralph Peterson

Urban rat fieldwork, vocal communication, and computational neuroethology.

Nick Jourjine

Bioacoustics, social neuroscience, and automated behavior tracking in the wild.

Dima Batenkov

Applied mathematics, signal processing, and computational methods.

Sreela Kodali

Embedded systems, robotics platforms, and field instrumentation.

Matthew Levine

Scientific machine learning, dynamical systems, and uncertainty quantification.

Dan Waxman

Bayesian machine learning, causal inference, and dynamical systems.

In the News