![[2602.18690] Neural Fields as World Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.18690] Neural Fields as World Models
Summary
The paper explores how neural fields can serve as world models, preserving sensory topology for better prediction of physical outcomes, with implications for intuitive physics and body schema.
Why It Matters
This research is significant as it proposes a novel approach to modeling how the brain predicts physical interactions, potentially enhancing our understanding of cognitive processes and improving machine learning applications in robotics and AI.
Key Takeaways
- Neural fields can maintain sensory topology for improved physics predictions.
- Local connectivity allows for more realistic modeling of physical interactions.
- Policies trained in simulated environments can transfer effectively to real-world scenarios.
Quantitative Biology > Neurons and Cognition arXiv:2602.18690 (q-bio) [Submitted on 21 Feb 2026] Title:Neural Fields as World Models Authors:Joshua Nunley View a PDF of the paper titled Neural Fields as World Models, by Joshua Nunley View PDF HTML (experimental) Abstract:How does the brain predict physical outcomes while acting in the world? Machine learning world models compress visual input into latent spaces, discarding the spatial structure that characterizes sensory cortex. We propose isomorphic world models: architectures preserving sensory topology so that physics prediction becomes geometric propagation rather than abstract state transition. We implement this using neural fields with motor-gated channels, where activity evolves through local lateral connectivity and motor commands multiplicatively modulate specific populations. Three experiments support this approach: (1) local connectivity is sufficient to learn ballistic physics, with predictions traversing intermediate locations rather than "teleporting"; (2) policies trained entirely in imagination transfer to real physics at nearly twice the rate of latent-space alternatives; and (3) motor-gated channels spontaneously develop body-selective encoding through visuomotor prediction alone. These findings suggest intuitive physics and body schema may share a common origin in spatially structured neural dynamics. Comments: Subjects: Neurons and Cognition (q-bio.NC); Computer Vision and Pattern Recognition (cs.CV); M...
