A Physical Basis for Information

TL;DR

This paper introduces a causal-physical framework that defines information through heritable structures in dynamical systems, enabling analysis across biological, cultural, and digital domains.

Contribution

It presents a novel causal-physical approach to defining information based on metastable structures and motifs, with algorithms for identifying and analyzing informational entities.

Findings

Mapped cultural example into causal structure set

Developed algorithms to identify information motifs

Proposed conditions for the emergence of hereditary informational motifs

Abstract

What is information, physically, and why does it so reliably emerge in living, cultural, and technological systems? Existing theories quantify uncertainty, cost, or compressibility, but do not identify which physical structures count as information or how informational entities arise from dynamics. Here we introduce a causal-physical framework that defines information as a heritable causal role played by persistent (metastable) structures in a dynamical system. We represent long-lived structures as almost-invariant sets and assemble them into causal structure sets that encode how such structures generate, transform, and maintain one another. Within this representation, informational entities are singled out by three generative motifs: replication, heritable variation, and translation under shared templates, which together define when a collection of structures constitutes an information family. We demonstrate the full pipeline by mapping a concrete cultural episode (fruit-salad recipe sharing and modification) into a causal structure set, and show how the motifs and information families can then be identified algorithmically. The framework yields computable quantities, including informational fitness and informational entropy, directly from causal structure, enabling informational variants to be detected, compared, and tracked across biological, cultural, engineered, and digital domains. Finally, motivated by analogies to random directed graphs and catalytic networks, we propose testable conditions under which hereditary informational motifs become statistically generic in sufficiently large causal-physical systems.