Natural Building Blocks for Structured World Models: Theory, Evidence, and Scaling

TL;DR

This paper introduces a modular framework for structured world models based on fundamental stochastic processes, emphasizing interpretability and scalability, and demonstrating competitive performance in passive and active tasks.

Contribution

It proposes a hierarchical, modular approach to world modeling using natural building blocks like HMMs and sLDS, facilitating interpretability and scalability.

Findings

Competitive performance in generative modeling and planning tasks

Supports passive and active world modeling within a unified architecture

Highlights the challenge of scalable joint structure-parameter learning

Abstract

The field of world modeling is fragmented, with researchers developing bespoke architectures that rarely build upon each other. We propose a framework that specifies the natural building blocks for structured world models based on the fundamental stochastic processes that any world model must capture: discrete processes (logic, symbols) and continuous processes (physics, dynamics); the world model is then defined by the hierarchical composition of these building blocks. We examine Hidden Markov Models (HMMs) and switching linear dynamical systems (sLDS) as natural building blocks for discrete and continuous modeling--which become partially-observable Markov decision processes (POMDPs) and controlled sLDS when augmented with actions. This modular approach supports both passive modeling (generation, forecasting) and active control (planning, decision-making) within the same architecture. We avoid the combinatorial explosion of traditional structure learning by largely fixing the causal architecture and searching over only four depth parameters. We review practical expressiveness through multimodal generative modeling (passive) and planning from pixels (active), with performance competitive to neural approaches while maintaining interpretability. The core outstanding challenge is scalable joint structure-parameter learning; current methods finesse this by cleverly growing structure and parameters incrementally, but are limited in their scalability. If solved, these natural building blocks could provide foundational infrastructure for world modeling, analogous to how standardized layers enabled progress in deep learning.