Zero-shot World Models Are Developmentally Efficient Learners
cs.AI, cs.CV
Submitted: 2026-04-11
Updated: 2026-09-09
License: http://creativecommons.org/licenses/by/4.0/
The gist: Young children demonstrate early abilities to understand their physical world, estimating depth, motion, object coherence, interactions, and many other aspects of physical scene understanding.
Terminology
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 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 shows progressive, staged emergence of capacities during learning 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 of children's early physical understanding and a path toward data-efficient AI systems.
Sources
- Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination
- A Simple Framework for Contrastive Learning of Visual Representations
- Bootstrap your own latent: A new approach to self-supervised Learning
- VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training
- Emerging Properties in Self-Supervised Vision Transformers
- Revisiting Feature Prediction for Learning Visual Representations from Video
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- Self-supervised learning through the eyes of a child
- Self-supervised learning of video representations from a child's perspective
- The BabyView dataset: High-resolution egocentric videos of infants' and young children's everyday experiences
- Assessing the alignment between infants' visual and linguistic experience using multimodal language models
- Call for Papers -- The BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus
- Unifying (Machine) Vision via Counterfactual World Modeling
- Masked Autoencoders Are Scalable Vision Learners
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- The Kinetics Human Action Video Dataset
- World Modeling with Probabilistic Structure Integration
- DINOv3
- TAP-Vid: A Benchmark for Tracking Any Point in a Video
- Kubric: A scalable dataset generator
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection