📄 other

Mechanical plasticity as a unifying axis for the thermodynamics of biological information machines

This paper introduces a unifying mechanical parameter based on the softness of functional coordinates relative to thermal noise to resolve biological information machines into collective modes, thereby deriving consistent thermodynamic limits for capacity and energy costs while successfully predicting empirical behaviors across ribosomes, ion pumps, and synapses.

Original authors: Pablo Garcia Tello

Published 2026-07-24

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Pablo Garcia Tello

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the inside of your body as a bustling, microscopic city. In this city, tiny machines are constantly at work: some are reading genetic instructions, others are pumping ions to keep your nerves firing, and some are storing memories in your brain. For decades, scientists have studied these machines using different rulebooks. One group looks at how accurately a machine copies a message, another measures how much energy it takes to move a load, and a third calculates how much heat is lost while storing a memory. The problem is, these rulebooks don't talk to each other. It's like trying to compare the speed of a race car, the fuel efficiency of a truck, and the storage capacity of a hard drive using three different languages. We know these machines are amazing, but we haven't had a single ruler to measure them all at once.

The missing ruler turns out to be "squishiness." In the world of physics, everything jiggles because of heat. A machine that is very stiff doesn't jiggle much, while a soft, squishy one wobbles wildly. This paper introduces a simple way to measure how "squishy" a biological machine is compared to the heat around it. By using this single measurement, the authors propose a new way to understand how these machines process information, how much energy they waste, and why some are built to last forever while others are built to change quickly. It's a new map that lets us look at a ribosome, a heart pump, and a memory synapse side-by-side to see what they have in common.

The Great Squishiness Scale

Meet the new hero of this story: a number called plasticity (pronounced pluh-STIS-uh-tee). Think of it as a "jiggle-meter." Every biological machine has parts that move to do their job. Some parts are like a solid steel beam (very stiff), and others are like a piece of cooked spaghetti (very soft). The authors created a formula to measure how much a specific part wobbles due to heat compared to how far it needs to move to do its work.

If the number is tiny, the machine is stiff and holds its shape tightly. If the number is big, the machine is soft and wobbles a lot. The paper suggests that if we look at all biological machines through this "jiggle-meter," we can finally sort them out. But here is the twist: you can't just pick one number for the whole machine. Just like a car has a stiff chassis but soft tires, a biological machine needs different parts to be stiff and soft for different reasons.

The Two-Mode Machine: The Writer and the Vault

The paper argues that the most interesting machines are the ones that need to do two contradictory things at once: write new information and keep (retain) old information.

Imagine you are trying to write a diary.

The Writer (Soft Mode): To write a new entry, you need a pen that moves easily. In a biological machine, this is a "soft" part that can wiggle and change shape quickly when a signal comes in. It's like a soft clay model you can reshape with your fingers.
The Vault (Stiff Mode): Once you write the entry, you want it to stay there for years. You can't leave it on a wobbly table; you need a steel vault. In the machine, this is a "stiff" part that barely moves at all, locking the information in place against the heat that tries to scramble it.

The paper shows that machines like the ribosome (which builds proteins) and the ion pump (which moves salt in and out of cells) mostly just do work. They are like assembly lines: they move, they finish, and they don't need to remember what they did. But the hippocampal synapse (the part of your brain that stores memories) is different. It must be a two-mode machine. It needs a soft part to accept new memories (adaptation) and a stiff part to hold onto them for a lifetime (retention).

If you try to describe a memory machine with just one number, you get confused. You might think the whole thing is too wobbly to hold a memory, or too stiff to learn anything new. By splitting the machine into its "soft writer" and "stiff vault," the authors show how it can do both.

The Cost of Keeping a Secret

The paper also introduces a new way to measure the "cost" of keeping a memory. In physics, there is a famous rule called the Landauer limit, which says that erasing a bit of information costs a tiny, tiny amount of energy. But keeping a bit of information? That's a different story.

The authors define a "Retention Burden." Imagine you have a secret.

Passive Memory (DNA): Your DNA is like a stone tablet. Once you carve the secret in, you don't have to do anything to keep it there. The stone just sits there. It costs almost nothing to maintain, as long as you fix the cracks when they appear.
Active Memory (Synapses): Your brain's memory is like a sandcastle. The waves (heat) are constantly trying to wash it away. To keep the castle standing, you have to keep adding sand and patting it down. This takes a lot of energy.

The paper calculates that keeping a memory in a synapse costs about 10,000 times more energy than keeping a memory in DNA. Why? Because the brain needs to be able to rewrite its memories. The ability to change (plasticity) is expensive. The paper suggests that this high cost is the price we pay for having a brain that can learn new things, rather than just a hard drive that only stores what it was programmed with.

What the Paper Predicts

The authors aren't just telling a story; they are making specific bets that can be tested in a lab. They suggest that if you look at the "jiggle-meter" of different machines, you will see a pattern:

The Stiffness Ratio: In a memory machine, the "vault" part should be about 1.5 times stiffer than the "writer" part. If scientists measure the stiffness of the brain's scaffolding and find it's wildly different, the theory might be wrong.
The Transfer Trick: There is a specific "transfer coefficient" (a number between 0.1 and 0.3) that controls how well a signal moves from the soft writer to the stiff vault. If you mess with the soft part (the actin) of a brain cell, the ability to write new memories should drop, but the ability to keep old ones should stay the same.
The Energy Law: The paper predicts that the more often a memory system is rewritten, the more energy it wastes per bit of information. If you compare DNA (which is rarely rewritten) with brain synapses (which are rewritten often), the brain should be much less efficient. This matches what we see: DNA is a super-efficient storage unit, while the brain is a high-energy, high-flexibility processor.

The Big Picture

This paper doesn't claim to have solved all the mysteries of the brain or to have invented a new law of physics. Instead, it offers a new lens. It suggests that the reason biological machines look so different from one another isn't just because they do different jobs, but because they are built with different balances of "stiffness" and "squishiness."

By separating the "soft" parts that do the work from the "stiff" parts that hold the memory, the authors provide a single language to talk about everything from the tiny machines in your cells to the complex networks in your brain. It's a reminder that in the microscopic world, the ability to be flexible is just as important as the ability to be strong, and that keeping a memory is a lot of hard work.

Technical Summary: Mechanical Plasticity as a Unifying Axis for the Thermodynamics of Biological Information Machines

Problem Statement
Biological information processing occurs in environments where the energy of functional events is comparable to thermal noise ( $k_B T$ ). Despite this, machines such as ribosomes, ion pumps, and synapses achieve reliable function. Current thermodynamic analyses of these systems are fragmented into disjoint programs: kinetic proofreading focuses on discrimination and turnover rates; synaptic plasticity literature focuses on quantal parameters and variance; and structural dissipation studies focus on abstract tasks versus physical implementation. There is no shared axis to compare these machines, making it impossible to distinguish whether differences arise from function or substrate. Existing frameworks often rely on single-parameter phenomenology, which conflates stiff structural modes (carrying persistent information) with soft functional modes (carrying mechanical work), leading to internally inconsistent numerical estimates.

Methodology
The paper proposes a unifying framework based on a single mechanical parameter: dimensionless plasticity, defined as $\gamma \equiv k_B T / (\kappa L^2)$ , where $\kappa$ is the stiffness of a functional reaction coordinate and $L$ is its physiologically relevant excursion.

Mode Decomposition: Rather than treating a machine as a single entity with one $\gamma$ , the authors model biological machines as a coupled harmonic Hamiltonian with $N$ degrees of freedom. By diagonalizing the Hessian matrix, they resolve the system into a spectrum of collective modes $\{Q_\alpha\}$ , each with an eigen-stiffness $\kappa_\alpha$ and a corresponding plasticity $\gamma_\alpha$ .
Spectral Partitioning: The spectrum $\{\gamma_\alpha\}$ ${γ_{α}}$ is partitioned into three distinct regimes:
- Structural modes ( $\gamma \ll 1$ ): Deeply pinned, carrying persistent information.
- Functional modes ( $\gamma \sim 0.1\text{--}1$ ): Soft coordinates used for updates and mechanical work.
- Thermal modes ( $\gamma \gtrsim 1$ ): Fully thermalized, indistinguishable from bulk solvent.
Derivation of Metrics: From this mode-decomposed Hamiltonian, the authors derive:
- Information Capacity ( $U_{max}$ ): Calculated as the sum of Shannon-Hartley capacities of independent Gaussian channels for each functional mode.
- Update Energy ( $E_{up}$ ): Derived as a quadratic form of the input coupling, showing that soft modes absorb more energy per unit input.
- Retention Burden ( $R$ ): A dimensionless index defined as $R \equiv P_{maint} \tau_{ret} / (u_{write} k_B T)$ . This measures the excess maintenance dissipation paid per retained bit relative to the Landauer floor, distinguishing passive substrates ( $R \sim 1$ ) from active, rewritable substrates ( $R \gg 1$ ).

Key Results and Applications
The framework is applied to four biological systems, demonstrating its ability to recover known measurements without adjustable parameters and resolve previous inconsistencies:

Ribosome (Throughput): The framework identifies two distinct modes: a stiff structural mode ( $\gamma_S \approx 0.02$ ) controlling structural error floors and a softer translocation mode ( $\gamma_T \approx 0.15$ ) controlling mechanical work. This resolves the inconsistency of single-parameter models. The analysis shows the ribosome is speed-optimized, with most energy dissipated in driving the cycle irreversibly rather than in proofreading.
Na+/K+-ATPase (Transport): This system is dominated by a single collective coordinate ( $\gamma \approx 0.1$ ), validating the single-parameter approximation in this specific case. The calculated thermodynamic efficiency is $\sim 75\%$ , consistent with direct measurements.
Hippocampal Synapse (Memory): This system requires a two-mode architecture: a soft adaptation coordinate ( $Q_A$ , $\gamma_A \sim 1$ ) for writing and a stiff retention coordinate ( $Q_R$ , $\gamma_R \sim 0.01$ ) for storage. The framework successfully reduces to the biophysical scaling laws of Malkin et al. ( $\sigma^{-2} \propto E^5$ ) and reproduces the efficiency-versus-conductance data of Harris et al. (via Stone's mapping) without adjustable parameters. It calculates a retention burden of $R \sim 10^4 k_B T/\text{bit}$ , highlighting the high thermodynamic cost of rewritability.
DNA (Passive Memory): As a passive substrate, DNA has a retention burden of $R \sim 10 k_B T/\text{bit}$ , close to the Landauer limit. The efficiency gap between DNA and synapses ( $\sim 10^3\text{--}10^4$ ) is identified as the thermodynamic price of active maintenance and rewritability.

Significance and Claims
The paper claims to provide a parameterization that places previously separate research programs (translation, transport, synaptic plasticity, genomic maintenance) on a single axis. The specific contributions are:

Unifying Axis: The introduction of mechanical plasticity $\gamma$ as a measurable, cross-machine unifying parameter.
Mode-Decomposed Architecture: The demonstration that machines requiring both writing and retaining fundamentally require a two-mode architecture (soft update, stiff storage) coupled by a transfer coefficient $\zeta$ . This resolves the conflation of structural and functional degrees of freedom found in single-parameter analyses.
Retention Burden ( $R$ ): The definition of a Landauer-normalized index that quantifies the cost of active storage, allowing for the comparison of diverse memory substrates.
Derivation from First Principles: The derivation of capacity ( $U_{max}$ ) and update energy ( $E_{up}$ ) directly from a coupled harmonic Hamiltonian, rather than treating them as phenomenological inputs.
Reconciliation of Empirical Laws: The framework contains the results of Malkin, Stone, and Harris as special cases, deriving their scaling exponents from the underlying mode structure and biophysical cost pairings.

The authors explicitly state they do not extend the Landauer bound, derive a new fluctuation theorem, or construct a formal classical quantum error correction (QEC) map. Instead, they offer a mechanical parameterization that contains explicit, falsifiable content regarding the mechanical stiffness of biological memory substrates and reduces to established results in known limits.

Empirical Predictions
The framework generates five specific, falsifiable predictions:

Stiffness Ratio: For two-mode memory machines, the ratio of retention to adaptation stiffness ( $\kappa_R/\kappa_A$ ) is set by the log-ratio of their timescales. For synapses, this predicts $\kappa_R/\kappa_A \approx 1.5$ .
Transfer Coefficient: Acute inhibition of actin dynamics (e.g., via cofilin) should degrade write efficiency ($LTP$ induction) but leave the retention of consolidated synapses intact, while PSD-95 knockdown should reduce capacity but leave update energy untouched.
Scaling Law: The retention burden $R$ should scale linearly with the write rate ( $k_{write}$ ) across different memory substrates (synapses, DNA methylation, immunological memory).
Efficiency Exponent: The exponent $\alpha$ in efficiency-versus-conductance curves is not universal (unlike Stone's $\alpha=5/2$ for synapses) but depends on the specific mode structure and cost pairing of the memory system.
Information Gain: Evolution should favor increasing the number of functional modes ( $|F|$ ) over signal amplitude to maximize information gain per unit energy, predicting a correlation between scaffold subunit diversity and per-event information content.

The Great Squishiness Scale

The Two-Mode Machine: The Writer and the Vault

The Cost of Keeping a Secret

What the Paper Predicts

The Big Picture

More like this