Learning Task-Agnostic Motifs to Capture the Continuous Nature of Animal Behavior

2025/06/18 by Jiyi Wang, Wang, Jiyi, Jingyang Ke +5 · 1 citation
Psychology · #FOS: Biological sciences #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neurons and Cognition (q-bio.NC) #Primate Behavior and Ecology

paper · pdf · doi:10.48550/arxiv.2506.15190

openalex publication_date 2025/06/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Animals flexibly recombine a finite set of core motor motifs to meet diverse task demands, but existing behavior segmentation methods oversimplify this process by imposing discrete syllables under restrictive generative assumptions. To better capture the continuous structure of behavior generation, we introduce motif-based continuous dynamics (MCD) discovery, a framework that (1) uncovers interpretable motif sets as latent basis functions of behavior by leveraging representations of behavioral transition structure, and (2) models behavioral dynamics as continuously evolving mixtures of these motifs. We validate MCD on a multi-task gridworld, a labyrinth navigation task, and freely moving animal behavior. Across settings, it identifies reusable motif components, captures continuous compositional dynamics, and generates realistic trajectories beyond the capabilities of traditional discrete segmentation models. By providing a generative account of how complex animal behaviors emerge from dynamic combinations of fundamental motor motifs, our approach advances the quantitative study of natural behavior.

Citations

• SkillBlender: Towards Versatile Humanoid Whole-Body Loco-Manipulation via Skill Blending
• Inverse Reinforcement Learning with Switching Rewards and History Dependency for Characterizing Animal Behaviors
• Diffusion Spectral Representation for Reinforcement Learning
• Foundation Policies with Hilbert Representations
• Multi-intention Inverse Q-learning for Interpretable Behavior Representation
• Self-Supervised Reinforcement Learning that Transfers using Random Features
• Behavior Contrastive Learning for Unsupervised Skill Discovery
• Spectral Decomposition Representation for Reinforcement Learning
• Making Linear MDPs Practical via Contrastive Representation Learning
• TRAIL: Near-Optimal Imitation Learning with Suboptimal Data
• Demonstration-Guided Reinforcement Learning with Learned Skills
• IQ-Learn: Inverse soft-Q Learning for Imitation
• Provable Guarantees for Self-Supervised Deep Learning with Spectral Contrastive Loss
• Learning One Representation to Optimize All Rewards
• Dynamic movement primitives in robotics: A tutorial survey
• Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
• OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
• Accelerating Reinforcement Learning with Learned Skill Priors
• Imitation Learning via Off-Policy Distribution Matching
• Dynamics-Aware Unsupervised Discovery of Skills
• Universal Successor Features Approximators
• Noise Contrastive Estimation for Scalable Linear Models for One-Class\n Collaborative Filtering
• Directed-Info GAIL: Learning Hierarchical Policies from Unsegmented Demonstrations using Directed Information
• Noise Contrastive Estimation and Negative Sampling for Conditional Models: Consistency and Statistical Efficiency
• Diversity is All You Need: Learning Skills without a Reward Function
• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• Mapping the stereotyped behaviour of freely-moving fruit flies

Cited by

• Fitting Reinforcement Learning Model to Behavioral Data under Bandits

Related

• Behave Your Motion: Habit-preserved Cross-category Animal Motion Transfer
• Relax, it doesn't matter how you get there: A new self-supervised approach for multi-timescale behavior analysis
• Learning recurrent representations for hierarchical behavior modeling
• Learning Behavior Representations Through Multi-Timescale Bootstrapping
• See and Copy: Generation of complex compositional movements from modular and geometric RNN representations