Why Energy Flowing Through Matter Makes It Lean Toward Order (Dissipative Adaptation)
Drive energy through matter and it does not drift at random. It leans toward the configurations that dissipate the drive best — and those look organized.
Under a sustained energy drive, matter is not directionless. The second law, in its fluctuation-relation form, makes trajectories that absorb and dissipate work more likely than their time-reverses, so a driven system drifts toward configurations that dissipate the drive well, and those good dissipators are taken to be the organized, gradient-processing structures. This is dissipative adaptation (Perunov, Marsland and England, 2016, building on England, 2013), grounded in earlier results that order is sustained by entropy production far from equilibrium (Prigogine) and that energy flow organizes matter (Morowitz, 1968). It points to a thermodynamic arrow toward organization that is prior to, and distinct from, Darwinian selection, and it comes with a falsifiable test.
A whirlpool over a draining plughole, a candle flame, a hurricane: each is a structure that exists only because energy keeps flowing through it, and each would vanish the moment the flow stopped. Why does flowing energy build such things instead of just spreading out? At equilibrium, with no flow, matter wanders without a preferred direction, drifting back and forth through its possible arrangements. Start driving it, pour energy through steadily, and the symmetry breaks. The second law, in its modern fluctuation-relation form, ties a trajectory's odds to how much entropy it produces (entropy being the spreading-out of usable energy into useless heat): paths that dissipate more become more likely than their exact time-reverses. A driven system gains a direction in the space of its arrangements, pointing toward configurations that process the drive.
That a path beats its own reverse is not yet enough to say the better dissipators win out over the worse ones. Perunov, Marsland and England (2016) supplied that stronger step: for matter coupled to a heat bath and pushed by a drive, there is a general tendency to arrive at the states that are reached through exceptionally reliable absorption and dissipation of work. The probability of landing in an arrangement scales with how much energy flowed through on the way there, so a driven population of configurations drifts toward the better dissipators, not just away from their reverses. England's (2013) earlier result is narrower: for something that copies itself, it sets a floor on the heat the replicator must shed to grow at a given rate. That is a cost on replication, a supporting special case, while the drift itself needs no copying at all. Now the key move, and it is an assumption, not a proof: the arrangements that dissipate a drive well are taken to be the ones with internal structure tuned to it, the organized, gradient-processing ones, so the thermodynamic favoring becomes a favoring of organization. It is not automatic (a plain resistor dissipates energy efficiently with no such structure), so it is offered as a conjecture and tested below. This rests on older results: that ordered structure is held together precisely by producing entropy far from equilibrium (Prigogine), and that energy flowing through a system organizes it (Morowitz, 1968). Because the drift is paid for by irreversible entropy production, it does not spontaneously run backward. That is a thermodynamic arrow aimed at organization, sitting underneath, and before, any biological selection.
Is this natural selection?
Is this selection in the Darwinian sense? Maybe, or maybe only the stage on which real replication-and-heredity selection later performs. I am flagging that question, not answering it. The drift described here needs no copying and no inheritance; it is a statistical lean of matter under drive. Whether it earns the name selection is open.
The claim is testable, and the test separates a driven system from an idle one. Fix one measure of organization in advance (the structure that couples to the throughput) and hold the drive steady. Averaged over many runs, that measure should trend upward over a chosen time window and not relax back while the drive is on, with single runs allowed to wobble around the trend; the bias buys an upward average, not a strictly uphill path every time. The same system with the drive switched off should show no such trend, only fluctuation. Two things can sink the claim, and each has its own test. The drive-off control catches a starting-condition artifact: if a driven system organizes no more than an idle one, the drive imposes no lean. The forward-versus-backward asymmetry catches the arrow: if the organization relaxes back to baseline once the drive is removed, or its trajectory shows no time-asymmetry, there is no arrow. Dissipation rate can be reported alongside as a secondary check, but it cannot stand in for the organization measure, since dissipation climbs with the drive on its own.
Sources
- Perunov, N., Marsland, R. A., and England, J. L. (2016). Statistical physics of adaptation. Physical Review X 6(2), 021036.
- England, J. L. (2013). Statistical physics of self-replication. The Journal of Chemical Physics 139(12), 121923.
- Crooks, G. E. (1999). Entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences. Physical Review E 60(3), 2721.
- Nicolis, G., and Prigogine, I. (1977). Self-Organization in Nonequilibrium Systems. Wiley, New York.
- Morowitz, H. J. (1968). Energy Flow in Biology. Academic Press, New York.
Comments