Demon with dementia - the deterioration of information transcription

TL;DR

This work reframes biological aging as the deterioration of information transcription fidelity within a stochastic thermodynamics framework using an autonomous Maxwell's demon that reads a template bitstream and writes to a copy stream while coupled to a heat reservoir and a lifted mass. The authors develop a minimal, solvable time-independent and time-dependent Markovian model with a 3-state demon and two bitstreams, analyzing fidelity $F(ε,τ)$, mutual information $I(N_T;N_C)$, extractable work, and entropy production; they implement a time-dependent protocol $ε(t)$ to mimic aging and study how performance decays. Key contributions include explicit demonstration of decay in fidelity, information, and work under aging, a decomposition of entropy production into adiabatic and non-adiabatic contributions under static and dynamic driving, and a concrete aging protocol with conditions for adiabatic tracking to instantaneous steady states. The results provide a compact, quantitative framework linking information processing and energy dissipation in microscopic biological-like machines, with potential extensions to error correction and stochastic environmental driving and broader biological relevance.

Abstract

In introductory biology, aging is typically explained as a result of mutations during the DNA replication process within cells. Upon abstraction, we recognize that cellular aging can be understood as the gradual decay in fidelity of information transcription. Since cellular processes are microscopic and inherently stochastic, the abstracted process of information transcription can be understood using Markovian dynamics. In our work, we model the process of information transcription with an autonomous Maxwell's demon (AMD) which interacts with two bitstreams, a lifted mass, and a heat reservoir. As main results, we analyze the steady-state properties of the system with both time-independent and time-dependent transition rates, focusing on the statistics of extractable work, bit transcription fidelity, and two-bit mutual information. Together, these results provide a holistic view of a simplified model for DNA transcription as an information-theoretic process.

Figures (13)

• Figure 1: Illustration of DNA transcription, adapted from Khan Academy, “Overview of transcription” (https://bit.ly/4pXt3nY), licensed under CC BY-NC-SA 4.0.
• Figure 2: Autonomous demon system to simulate information transcription.
• Figure 3: Markov chain representation of information replication process with AMD system. Note the change in directionality of the transitions connecting states A1 and C0 depending on the template bit.
• Figure 4: Transcription fidelity $F$ as a function of interaction time $\tau$ with each 'template' bit for 1,000 trajectories. Here, $|\epsilon|$ = 1 to simulate perfect erasure/anti-erasure processes.
• Figure 5: Transcription fidelity $F$ on a lin-log scale as a function of thermodynamic bias $\epsilon$ for 1,000 trajectories. Here, $\tau=30$ to minimize readout error.