Evolution Icon Evolution
Physics, Earth & Space Icon Physics, Earth & Space

Evolution Is Not Like Physics

Photo credit: basykes, CC BY 2.0 , via Wikimedia Commons.

Darwinian evolution is not like physics. In chapter 6 of his new book, Animal Algorithms, Eric Cassell explains why: evolutionary “laws” lack the physical deterministic process that underlie laws of physics. He quotes Elliott Sober saying “many of the generalizations in evolutionary theory are tautologies.” There’s a difference, Sober says, between physics and biology. “Physical laws are often empirical, but general laws in evolutionary theory typically are not.” Even Ernst Mayr agreed, “The so-called laws of biology are not the universal laws of classical physics but are simply high-level generalizations.” (p. 156)

Most importantly, Cassell continues, attempts to portray natural selection as a “law” like those in physics (gravitation, Boyle’s Law, Einstein’s E = mc2, Newton’s laws of motion) fail to account for the origin of the information found in biological systems, as William Dembski (No Free Lunch) and Stephen Meyer (Signature in the Cell, Darwin’s Doubt) have explained at length.

One would think these distinctions to be common knowledge among scientists, but another glaring instance of the fallacy (equating evolutionary “laws” with physical laws) has just been published in a leading journal, the Proceedings of the National Academy of Sciences. The proponents include Vitaly Vanchurin, Yuri I. Wolf, and Eugene V. Koonin from the National Institute for Biotechnology Information at the National Library of Medicine in Bethesda, Maryland, and Mikhail I. Katsnelson from the Institute for Molecules and Materials in the Netherlands. Vanchurin also works for the Duluth Institute for Advanced Study in Duluth, Minnesota. 

Koonin Should Know Better

Casey Luskin has pointed out that Koonin is well aware of discontinuities in the Darwinian tree of life. The facts of biological stasis led this “occasional gadfly to Darwinian conformism” to embrace punctuated equilibria as the “default mode of evolution” in certain conditions. Kirk Durston notes that Koonin is also painfully aware that “the probability of evolving RNA replication for the origin of life is so small that it is unlikely to occur anywhere in the universe, over its history to date.” Yet Koonin lent his name to the two PNAS papers advancing the idea that evolution is like physics, including, of all things, the laws of thermodynamics. Granville Sewell will enjoy that suggestion. 

To these four, the major transitions in biology, including the origin of life and the origin of multicellularity, and presumably the Cambrian Explosion, reduce to “phase transitions” like water turning into ice or steam. Surely you jest, Dr. Koonin? Not really; he has already shown his willingness to enter fantasyland by endorsing the multiverse as a way out of improbabilities.

Fallacy Upon Fantasy

The authors add fallacy to fantasy by describing evolution as a “law of learning” as well as a physical law. Every time a major transition in evolution (MTE) occurs, the universe learns something and gets better. It begins a history. This learning, moreover, occurs at multiple levels. To support this notion, they embrace the controversial notion of “multilevel selection” (MLS) and expand it out to encompass the whole universe.

Let them dig their own hole without our help. In a two-part evaluation of this new proposal, let’s look at their first paper, “Toward a theory of evolution as multilevel learning,” by Vanchurin et al. Be alert to the possibility that these leaps into fantasyland have been necessitated by failures of standard neo-Darwinian theory.

Modern evolutionary theory gives a detailed quantitative description of microevolutionary processes that occur within evolving populations of organisms, but evolutionary transitions and emergence of multiple levels of complexity remain poorly understood. Here, we establish the correspondence among the key features of evolution, learning dynamics, and renormalizability of physical theories to outline a theory of evolution that strives to incorporate all evolutionary processes within a unified mathematical framework of the theory of learning. According to this theory, for example, replication of genetic material and natural selection readily emerge from the learning dynamics, and in sufficiently complex systems, the same learning phenomena occur on multiple levels or on different scales, similar to the case of renormalizable physical theories. [Emphasis added.]

“Poorly understood.” How sad that after all these years of Darwinist saturation, the chief promise of the Origin — understanding — languishes in poverty. With great hope and rhetorical flourish, Darwin had led the scientific community to envision all the variety of life coming into emergence via natural laws.

It is interesting to contemplate an entangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent on each other in so complex a manner, have all been produced by laws acting around us.

That Grandiose View of Life 

The “grandeur in this view of life” followed directly “from the war of nature, from famine and death,” he wrote. (Richard Weikart revisits the consequences of that grandiose view of life in his new book, Darwinian Racism.) Evolution happened “whilst this planet has gone cycling on according to the fixed law of gravity,” Darwin said, in a not-so-subtle attempt to link his law of natural selection to physics. He used the words law and laws 158 times in the Origin.

The authors recount Schrödinger’s emphasis in What Is Life? The Physical Aspect of the Living Cell (1944) that life seems to run on “negentropic” processes, “and frustrations at different levels are necessary for these processes to take off and persist.” This is the basis of their speculation that physical laws underlie evolution.

At least since the publication of Schroedinger’s book, the possibility has been discussed that, although life certainly obeys the laws of physics, a different class of laws unique to biology could exist. Often, this putative physics of life is associated with emergence, but the nature of the involved emergent phenomena, to our knowledge, has not been clarified until very recently [referring to Vanchurin’s own idea of “the world as a neural network”]. Here, we outline a general approach to modeling and studying evolution as multilevel learning, supporting the view that a distinct type of physical theory, namely the theory of learning, is necessary to investigate the evolution of complex objects in the universe, of which evolution of life is a specific, even if highly remarkable form.

From there they proceed to describe “frustrations” in the development of the universe that caused the “emergence” of matter, stars, galaxies, worlds, and life. These emergences are all “major transitions in evolution,” entirely physical, with life just being a special case. But is it really physical? If the universe can “learn” at multiple levels, it seems to have a purpose or direction. At best, this is a form of the participatory anthropic principle, in which the universe needs to develop the properties that allow for observers to emerge. At worst, it is a resurrection of vitalism or animism. Let them explain in seven principles required for an observable universe:

    P1. Loss function. In any evolving system, there exists a loss function of time-dependent variables that is minimized during evolution.

    P2. Hierarchy of scales. Evolving systems encompass multiple dynamical variables that change on different temporal scales (with different characteristic frequencies).

    P3. Frequency gaps. Dynamical variables are split among distinct levels of organization separated by sufficiently wide frequency gaps.

    P4. Renormalizability. Across the entire range of organization of evolving systems, a statistical description of faster-changing (higher-frequency) variables is feasible through the slower-changing (lower-frequency) variables.

    P5. Extension. Evolving systems have the capacity to recruit additional variables that can be utilized to sustain the system and the ability to exclude variables that could destabilize the system.

    P6. Replication. In evolving systems, replication and elimination of the corresponding information-processing units (IPUs) can take place on every level of organization.

    P7. Information flow. In evolving systems, slower-changing levels absorb information from faster-changing levels during learning and pass information down to the faster levels for prediction of the state of the environment and the system itself.

An Atom, a Star, a Trilobite 

A brief attempt to summarize their thinking goes like this. If an evolving system has processes in operation at different scales and speeds, there will automatically be “frustrations” that adjust (renormalize, reset) at multiple levels, leading to losses of information. These losses can be minimized by the optimizing process of natural selection, they claim, which is good; it leads to progress, reducing the bad options and encouraging innovation. If the system has a memory of some sort, accurate enough to avoid Eigen catastrophe, the system learns from the rest — whether an atom, a star, or a trilobite.

The detailed correspondence between the key features of the processes of learning and biological evolutionimplies that this is not a simple analogy but rather, a reflection of the deep unity of evolutionary processes occurring in the universe. Indeed, separation of the relevant degrees of freedom into multiple temporal classes is ubiquitous in the universe from composite subatomic particles, such as protons, to atoms, molecules, life-forms, planetary systems, and galaxy clusters. If the entire universe is conceptualized as a neural network, all these systems can be considered emerging from the learning dynamics.

Natural selection as a “law” is key to their proposal: “All seven fundamental principles of life-compatible universes (P1 to P7) are involved in enabling evolution by natural selection.” Their praise of natural selection goes beyond the usual: “Evolution by natural selection is the central tenet of evolutionary biology and a key part of the NASA definition of life.” But they commit the foul of taking natural selection into the physical realm, beyond life. If certain atoms outcompete others, that is natural selection, they allege: nature has “learned” and made progress. “With the onset of Darwinian evolution, the system can be considered to cross the threshold from prelife to life,” they say, calling any kind of “asymmetric information flow” an instance of natural selection. This idea is contrary to most involved in evolutionary biology, who insist that accurate replication is a prerequisite for natural selection. 

The Fatal Flaw in the Proposal 

That’s enough to get the gist of the paper. “We apply the theory of learning to physically renormalizable systems in an attempt to outline a theory of biological evolution, including the origin of life, as multilevel learning,” they promise. “We formulate seven fundamental principles of evolution that appear to be necessary and sufficient to render a universe observable and show that they entail the major features of biological evolution, including replication and natural selection.”

The National Academy of Sciences likes this proposal because it is naturalistic. At least, it seems to be so, if one accepts the premise that the universe is a “neural network” and that mindless objects like atoms and stars like to learn things. The fatal flaw in the proposal, though, is the concept of “concept” (a word they use 14 times); e.g., “evolutionary concepts, such as natural selection.” Concepts are non-physical. So much for naturalism. Nice try anyway.

Next time, we’ll see how they apply their proposal to the origin of life.