Evolutionary Transitions in Individuality: much more than you wanted to know

I.

For the first half of the 20th century, evolutionary biology was dominated by the Modern Synthesis—the triumphant, mathematically rigorous fusion of Darwinian natural selection with Mendelian genetics. It was a beautiful theory, clean and legible in the exact way that neoclassical economics is clean and legible. Allele frequencies exist in a population; the environment applies a selection pressure; the frequencies update via a localized algorithm.

But the Modern Synthesis treated the actual, physical organism as a black box. A genotype goes into the box, a phenotype comes out, selection acts upon the phenotype, and the ledger of allele frequencies is adjusted. It offered almost no mechanistic explanation for the box itself. If you asked a mid-century population geneticist how a linear string of nucleotides actually reliably constructs a three-dimensional metazoan, they would politely change the subject to fruit fly mutation rates.

Evolutionary developmental biology (evo-devo) emerged in the 1980s and 1990s to pry open that box.

The foundational shock of evo-devo was the discovery of “deep homology.” Before the 1980s, biologists mostly assumed that morphologically disparate animals were built using entirely different genetic blueprints. Instead, they discovered that almost all bilaterian animals share the exact same core toolkit of master regulatory genes.

The canonical example is the Hox gene cluster, which determines the anterior-posterior axis of a developing embryo. A fruit fly and a mouse are separated by roughly 600 million years of evolutionary divergence. Yet, their spatial organization is dictated by strictly homologous genetic switches. This compatibility is so absolute that you can take the Pax6 gene—the master switch for eye development in a mouse—insert it into the genome of a Drosophila embryo, trigger it in the fly’s leg tissue, and the fly will successfully execute the cascade to grow a structurally functional (albeit completely useless) fly eye on its leg.

Evo-devo elegantly resolved the agonizing tension between the gradualism of the Modern Synthesis and the fossil record’s sudden morphological shifts. Critics of Darwin had long pointed out the “half a wing” problem: what is the selective advantage of an intermediate, non-functional limb?

Evo-devo answered: biology is object-oriented programming. Macroevolution is not driven by the agonizingly slow invention of novel structural proteins. It is driven by tweaking the regulatory networks that control when and where existing modules are deployed. To evolve a snake from a lizard, you don’t wait for a million point mutations to gradually degrade the leg; you just alter the embryonic expression gradient of the Hox genes that suppress limb formation, copy-pasting the thoracic rib-module down the length of the body. You don’t rewrite the code; you just swap a variable in the UI framework.

The problem with classical evo-devo is that it is a victim of its own restricted scope. It is an exceptionally powerful explanatory framework for understanding how to alter an animal’s morphology, provided you already have an animal.

Evo-devo essentially assumes the existence of a perfectly obedient command economy. It assumes a centralized organism, executing a developmental program, where trillions of cells flawlessly coordinate to build a liver, a femur, or a neocortex. It explains variations on a body plan, but it cannot explain the origin of the organismal paradigm itself.

And this is a massive problem, because if you take Richard Dawkins’ The Selfish Gene seriously, the organism shouldn’t exist.

Evolution is a Hobbesian war of all against all, driven by the ruthless maximization of inclusive fitness. So how do you get trillions of independent, reproducing entities (cells) to permanently suppress their own reproductive potential without defecting? Why does a somatic skin cell accept a fitness of exactly zero, voluntarily dying just so a tiny privileged minority of cells (the germline) can be passed on?

Evo-devo ignores the principal-agent problem of biology. It assumes the existence of unified evolutionary agents. It cannot map the transition from a population of free-living, single-celled protozoa to an obligate multicellular metazoan, because that isn’t a shift in morphological space. That is a game-theoretic nightmare. For that, biology needed a different paradigm.

II.

In 1995, John Maynard Smith (the biologist who, alongside George Price, formally introduced game theory to evolutionary biology) and biochemist Eörs Szathmáry published a blockbuster book titled The Major Transitions in Evolution. It was a sweeping attempt to provide a unified theoretical framework for the structural leaps that evo-devo took for granted.

Maynard Smith and Szathmáry proposed that the history of life is not a smooth curve of increasing complexity, but is instead punctuated by a series of distinct, radical reorganizations in the very nature of how evolution operates.

They identified a recurring pattern across these transitions:

1. Smaller entities that were previously capable of independent replication come together to form a larger, higher-level entity.
2. The smaller entities specialize, and irreversibly lose their ability to replicate independently (obligate interdependence).
3. The transition introduces a fundamentally new mechanism for storing and transmitting information.

Their canonical list consisted of eight transitions:

1. Replicating molecules → Populations of molecules in compartments (protocells).
2. Independent replicators → Chromosomes.
3. RNA as both gene and enzyme → DNA as genetic storage and protein as enzyme.
4. Prokaryotes → Eukaryotic cells.
5. Asexual clones → Sexual populations.
6. Single-celled organisms → Multicellular organisms (plants, animals, fungi).
7. Solitary individuals → Eusocial colonies (ants, bees, termites).
8. Primate societies → Human societies with language.

The brilliance of this framework was that it reframed evolutionary history as a series of solved coordination problems. The transition to multicellularity or eusociality was no longer viewed simply as the acquisition of a new trait. It was the establishment of a new biological individual. The lower-level units successfully navigated the Molochian trap of defection, and in doing so, the unit of selection shifted upwards.

The Major Transitions in Evolution essentially created a subfield overnight, but almost immediately, the philosophers of biology started tearing it apart.

The primary issue was that the list of eight transitions was an absolute mess. In a brutal 2011 critique, Daniel McShea and Carl Simpson accused the Maynard Smith and Szathmáry framework of being a “heterogeneous grab bag” that was “philosophically muddled and scientifically casual”.

The core problem was that the 1995 book conflated two completely different categories of evolutionary events:

1. Technological upgrades in information processing: The invention of DNA, the invention of sex, and human language.
2. Evolutionary Transitions in Individuality (ETIs): The origin of chromosomes, eukaryotic cells, multicellularity, and eusociality.

If the defining feature of a major transition is that independent entities obligately lose their reproductive autonomy to form a higher-level unit, then human language (Transition 8) is an incredibly embarrassing inclusion. Humans do not permanently surrender their independent reproductive capacity to the tribe just because they acquire syntax. We are not a eusocial hive mind. Including language made the entire framework look suspiciously like a Great Chain of Being teleology, conveniently designed to culminate at the exact apex of Homo sapiens writing books about evolution.

As philosopher Peter Godfrey-Smith later pointed out, “inventing DNA” is a profound shift, but it is fundamentally a hardware upgrade for an existing Darwinian population. Conversely, “inventing a multicellular organism” is a change in the nature of the player itself. You are taking a population of Darwinian individuals and fusing them to create a single, macro-level Darwinian individual. Grouping these two phenomena under the same umbrella term rendered the theory incoherent.

III.

Over the last two decades, the field has undergone a rigorous process of clarification, resulting in what is sometimes informally called “MET 2.0”. The modern consensus has largely abandoned the vague “information processing” criteria and focused squarely on Evolutionary Transitions in Individuality (ETIs).

The most crucial theoretical update came from theorists like David Queller, who pointed out that the 1995 framework lumped together two completely different pathways to higher-level individuality: Fraternal transitions and Egalitarian transitions.

Fraternal transitions occur when a single lineage of genetically identical (or highly related) individuals stays together. Multicellularity and eusociality are fraternal transitions. Because the individuals (the cells in an embryo, or the workers in a beehive) are highly related, the principal-agent problem is largely solved by inclusive fitness (kin selection). It’s a family business. A somatic cell accepts a fitness of zero because it shares its alleles with the germ cell. The primary threat to a fraternal transition is a mutation that causes a unit to defect, abandon the family business, and replicate selfishly. We call this defection cancer.

Egalitarian transitions, on the other hand, occur when entirely different, unrelated species merge to form a new individual. The origin of the eukaryotic cell (the merger of an archaeon and a bacterium to form the mitochondrion) is the archetypal egalitarian transition. Inclusive fitness cannot explain this, because the relatedness between the two original species is exactly zero. This is a corporate merger between two hostile alien species. They stay together because of extreme metabolic synergies (trade), but egalitarian transitions are notoriously difficult to stabilize because they are highly vulnerable to parasitism. They require draconian policing contracts—such as enforcing strict vertical transmission (which is why your mitochondria are strictly maternally inherited, preventing deadly competition among mitochondrial lineages within your cells).

Theoretical biologist Richard Michod provided the mathematical rigor for this new paradigm with his concept of the “export of fitness”.

Michod argued that you can mathematically define exactly when an evolutionary transition is complete. A transition is finished when the lower-level entities literally cannot be assigned a standalone fitness value anymore. Their independent fecundity drops to zero, and the viability of the whole completely depends on their specialized somatic labor. Fitness has been successfully exported to the macro-level.

We have moved far beyond classical evo-devo’s assumption of a smoothly unified organism, and we have discarded the 1995 framework’s poetic but messy fixation on human language. The modern study of major transitions is a rigorous, game-theoretic analysis of multi-level selection.

It leaves us with a profound, slightly unsettling view of biology. An “individual” is not a fundamental unit of nature. An individual is a treaty. It is a highly policed coalition of former rivals, a monument to a coordination problem that we managed, against all odds, to permanently solve.