Information field theory for collective behavior

Proposed project for an Emmy Noether research group at the Goethe University Frankfurt


Julian C. Schäfer-Zimmermann


Max Planck Institute of Animal Behavior

Department for the Ecology of Animal Societies

Communication and Collective Movement (CoCoMo) Group


Collective behavior, phase change, and criticality


Recruitment events in spotted Hyenas

© Eli Strauss | Michigan State University

Approaches to quantitative collective behavior


Lagrangian

Tracks individuals (particles)

  • Vicsek-, Couzin-, Cucker-like models
  • Behavioral heuristics
  • Reproduce swarming, milling, and schooling states

Eulerian

Treats group as fluid

  • Toner-Tu-, Mogilner-Edelstein-like models
  • Continuous hydrodynamic equations
  • Can be linked to RG theory via path-integral methods

These models are only concerned with movement

The animal collective


© Ariana Strandburg-Peshkin | Max Planck Institute of Animal Behavior

Goal of this project


My research focuses on understanding the internal motivations of animal behavior directly from multimodal data. I model the dynamics of entire animal groups using an approach that renders these internal states visible.


Thank you for you attention



Information field theory for collective behavior

Proposed project for an Emmy Noether research group at the Goethe University Frankfurt


Julian C. Schäfer-Zimmermann


Max Planck Institute of Animal Behavior

Department for the Ecology of Animal Societies

Communication and Collective Movement (CoCoMo) Group


Phase 1:

From the animal collective to the behavioral neural network


* Bialek, W.. "On the dimensionality of behavior." Proceedings of the National Academy of Sciences 119.18 (2022): e2021860119.
* Zhang, Y. et al. (2023). Meta-Transformer. arXiv.2307.10802
** Schäfer-Zimmermann, J. C., et al. (2025). animal2vec and MeerKAT. Methods in Ecology and Evolution, 2041-210x.70218
Vali, M. H. et al. (2025). Diveq: Differentiable vector quantization using the reparameterization trick. arXiv:2509.26469.
* Brown, Tom, et al. "Language models are few-shot learners." Advances in neural information processing systems 33 (2020): 1877-1901.
** Khosla, Prannay, et al. "Supervised contrastive learning." Advances in neural information processing systems 33 (2020): 18661-18673.

Phase 2:

From the neural network to information field theory

© Ariana Strandburg-Peshkin | Max Planck Institute of Animal Behavior

Phase 2:

From the neural network to information field theory


* Merbis, W., & de Domenico, M. (2023). Emergent information dynamics in many-body interconnected systems. Physical Review. E, 108(1–1), 014312
* Merbis, W., & de Domenico, M. (2023). Emergent information dynamics in many-body interconnected systems. Physical Review. E, 108(1–1), 014312

The approach and novelty of this project


  1. Learning behaviors directly from data by training a behavioral neural network (BNN), which assumes nothing
    • This is a hypothesis-free abductivist approach to behavioral science
  2. Define biologically relevant situations that show phase transitions or tuning in and out of a state of criticality, such as foraging or predator evasion
    • Infer the behavioral states for the situations from the BNN
  3. Learn a social Hamiltonian for each of these situations
    • Pairwise and, later, higher order interactions via Graph Neural Networks
  4. Recreate the situation using the information-field theory approach
    • How is each behavioral state from each individual affecting every other behavioral state from every other individual
    • These are predictive models, with which we can also predict behaviors for new and/or rare situations (Hypothesis generation)

The approach and novelty of this project


  1. Once this is established, we can:
    • Explore the collective behavior of other species / systems (Max Planck Institute of Animal Behavior (MPIAB) as collaborator)
      • I will remain a guest scientist at MPIAB
    • Derive world models with agents based on the learned behaviors of the neural network
    • Backwards engineer analytical - biologically interpretable - models from the learned Hamiltonians
    • Go back to the field and test hypotheses in playback experiments

Timeline


JSZ: Myself | P1: first Phd-student / Post-Doc after 3rd year | P2: second Phd-student / Post-Doc after 3rd year | SE: student employee

Risk mitigation


  1. What if the BNN learns nothing
    • We can kickstart phase two using:
      • Classical ethological descriptors (e.g., "sleeping" or "foraging")
      • Video as additional input modality for the BNN (this substantially decreases dataset temporal coverage)
      • Synthetic data from established movement models
      • Modify the collars worn by the meerkats (sampling frequencies, modalities, ...) and do an experimental campaign
        • I followed up a comment by Prof. Meyer in terms of funding for this to ensure this remains a possibility
  2. What if the modelling in the configuration space yields nothing
    • We can fallback to node-based mean-field modelling
    • Directly derive an analytical model from the behavioral states of the BNN

Final words


My research focuses on understanding the internal motivations of animal behavior directly from multimodal data. I model the dynamics of entire animal groups using an approach that renders these internal states visible.


We will gain an understanding of the internal motivations of each animal within a group before, during and after collective action, and derive testable, biologically relevant rules for complex social interactions within animal societies.

The team


Thank you for you attention



Information field theory for collective behavior

Proposed project for an Emmy Noether research group at the Goethe University Frankfurt


Julian C. Schäfer-Zimmermann


Max Planck Institute of Animal Behavior

Department for the Ecology of Animal Societies

Communication and Collective Movement (CoCoMo) Group