Zero-shot World Models Are Developmentally Efficient Learners

2026-04-11

AI Summary

This work addresses the poor data efficiency and limited generalization of current AI systems in physical scene understanding by drawing inspiration from children’s remarkable ability to rapidly acquire intuitive physics from minimal experience. The authors propose the Zero-shot World Model (ZWM), which formalizes developmental cognitive mechanisms into a computable framework. ZWM achieves efficient learning from a single child’s first-person experiences through three key components: disentangled appearance and dynamics via sparse temporal prediction, zero-shot estimation grounded in approximate causal inference, and modular compositional reasoning. Evaluated across multiple physical reasoning benchmarks, ZWM rapidly attains high performance, successfully replicates characteristic child-like behaviors, and generates internal representations that align with neural activity patterns observed in the human brain, thereby substantially enhancing few-shot generalization capabilities.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Computer Vision: Low Level & Physics-based Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Young children demonstrate early abilities to understand their physical world, estimating depth, motion, object coherence, interactions, and many other aspects of physical scene understanding. Children are both data-efficient and flexible cognitive systems, creating competence despite extremely limited training data, while generalizing to myriad untrained tasks -- a major challenge even for today's best AI systems. Here we introduce a novel computational hypothesis for these abilities, the Zero-shot Visual World Model (ZWM). ZWM is based on three principles: a sparse temporally-factored predictor that decouples appearance from dynamics; zero-shot estimation through approximate causal inference; and composition of inferences to build more complex abilities. We show that ZWM can be learned from the first-person experience of a single child, rapidly generating competence across multiple physical understanding benchmarks. It also broadly recapitulates behavioral signatures of child development and builds brain-like internal representations. Our work presents a blueprint for efficient and flexible learning from human-scale data, advancing both a computational account for children's early physical understanding and a path toward data-efficient AI systems.

Problem

• physical scene understanding
• data-efficient learning
• cognitive development
• zero-shot generalization
• artificial intelligence

Innovation

• zero-shot learning
• visual world model
• causal inference
• compositional reasoning
• developmental AI