Learning to Theorize the World from Observation

TL;DR

This paper introduces a new learning paradigm and neural model that induce explicit, compositional theories from raw observations, aiming to emulate human-like understanding through explanation-driven generalization.

Contribution

The paper proposes Learning-to-Theorize and the Neural Theorizer (NEO), a probabilistic neural model that constructs and executes explicit theories as latent programs from observations.

Findings

NEO induces interpretable, executable programs explaining observations.

The model enables generalization based on explanations rather than just predictions.

Experiments demonstrate the effectiveness of theory-based understanding in complex scenarios.

Abstract

What does it mean to understand the world? Contemporary world models often operationalize understanding as accurate future prediction in latent or observation space. Developmental cognitive science, however, suggests a different view: human understanding emerges through the construction of internal theories of how the world works, even before mature language is acquired. Inspired by this theory-building view of cognition, we introduce Learning-to-Theorize, a learning paradigm for inferring explicit explanatory theories of the world from raw, non-textual observations. We instantiate this paradigm with the Neural Theorizer (NEO), a probabilistic neural model that induces latent programs as a learned Language of Thought and executes them through a shared transition model. In NEO, a theory is represented as an executable, compositional program whose learned primitives can be systematically recombined to explain novel phenomena. Experiments show that this formulation enables explanation-driven generalization, allowing observations to be understood in terms of the programs that generate them.