Patterns that persist: Heritable information in stochastic dynamics

TL;DR

This paper investigates how heritable information persists in stochastic dynamical systems, linking thermodynamics, system size, and information theory to understand the emergence of life-like patterns.

Contribution

It introduces a framework connecting slow-relaxing patterns, thermodynamic constraints, and information quantification in Markov processes, advancing origins of life research.

Findings

Slow-relaxing systems exhibit persistent patterns with long lifetimes.

Entropy production bounds the stability duration of these patterns.

Universal compression algorithms can quantify heritable information in experimental systems.

Abstract

Life on earth is distinguished by long-lived correlations in time. The patterns of material organization that characterize living organisms today are contingent on events that occurred billions of years ago. This contingency is a necessary component of Darwinian evolution: patterns in the present inherit some of their features from those in the past. Despite its central role in biology, heritable information is difficult to recognize in prebiotic systems described in the language of chemistry or physics. Here, we consider one such description based on continuous-time Markov processes and investigate the persistence of heritable information within large sets of dynamical systems. While the microscopic state of each system fluctuates incessantly, there exist few systems that relax slowly to their stationary distribution over much longer times. These systems, selected for their persistence, are further distinguished by their mesoscopic organization, which allows for accurate course grained descriptions of their dynamics at long times. The slow relaxation of these stable patterns is made possible by dissipative currents fueled by thermodynamic gradients present in the surrounding reservoirs. We show how the rate of entropy production within a system sets an upper bound on the lifetime of its persistent patterns. We also consider the probability of finding persistence within large sets of dynamical systems. We show that physical constraints based on continuity and locality can strongly influence the probability of persistence and its dependence on system size. Finally, we describe how heritable information can be quantified in practice using universal compression algorithms. We demonstrate this approach on an experimental system of magnetically-driven, colloidal rollers and discuss the application of these methods to origins of life research.