Substrates of Selection

Thanks to Julia Persson and Clem von Stengel for related conversations.

Substrates of selection make selection non-trivial

The replicator-centric-view of evolution identifies evolution with differential replication and persistence of some small discrete-ish units. These are then the units on which selection operates, i.e. the replicators. Whatever agent-y/complex systems can emerge from this struggle for differential survival and replication, is instrumental to replicators' purposes and is called a vehicle (Dawkins) or interactor (Hull). In the context of (modern) Earth-originating life, replicators are genes.¹ But evolution doesn't have to be biological and replicators don't have to be genes. Dawkins himself proposed that memes — culturally transmissible viral-ish units of information — undergo some sort of evolutionary dynamic.

Importantly, this view differs somewhat from the one proposed by pre-genetic evolutionary theorists, such as Darwin and Wallace, the originators of the theory. According to them, selection was (roughly) a description of differential persistence of organismal traits, due to the differential advantage they were affording in those organisms' survival and reproduction. The replicator-centric view can be cast a lower-level implementation of this latter view. You don't need genes to have stable inheritance. All you need is a population of individuals with some degree of variability that has implications for differential survival-and-reproduction and a mechanism to (re-)generate that variability. (Although it's very plausible that something like a discrete-ish heritable information/blueprint increases the chance that evolution stumbles upon some interesting/complex form and speeds up that stumbling-upon.) See for example Darwinian Populations and Natural Selection by Peter Godfrey-Smith. Perhaps dually, the latter view can be cast as an abstraction/coarsening of the replicator-centric view.

One can go one step further and generalize the notion of selection to general struggle for persistence. Here we have, a world inhabited by various things (or processes or whatever) and some higher-level process (e.g. physics or a world-state transition function) that differentially favors/disfavors persistence of some things or types of things. Persistence of things is survival. Persistence of types of things is survival and/or reproduction with the caveat that reproduction is instrumental because things of the type in question tend to decay/rot and need to be regenerated for the type of things to continue to exist. The two blur into each other when we consider cases of imprecise/indeterminate/ambiguous individuality, such as clonally reproducing organisms, e.g. strawberries reproducing via ramets or Pando or even memes, if we accept memes as an example of evolution. This is what Frédéric Bouchard (2004, 2011) argues for.

Memetic "evolution" is very different from paradigmatic examples of Darwinian selection. One might argue that a problem is that it's not clear how to parse the cultural millieu into distinct replicators² but we have the same problem when trying to think about genes as replicators, though perhaps to a lesser extent because (1) we have more tangible access to genes, and (2) we are being routinely forced to adopt some kind of parsing scheme for the purpose of biological research that is much more mature than psychological/sociological/memetic research.

However, the bigger difference between memes and (e.g.) genes is that along "mere and dumb" "selection pressures", there are also intelligent (or at least adaptive) action of organisms/agents shaping memes and cultures to their will (even if not consciously/intentionally/deliberately, whatever that means).³ The same issues apply to the so-called cultural evolution, which is one might (as, for example, I do) prefer to call it "cultural learning".⁴

The gene-centric view of evolution, more inclusive views of evolution,⁵ and memetics/"cultural evolution" and various specific/particular examples or subcategories of differential persistence involve very different mechanisms.⁶ If one were to set out to develop a generalized theory encompassing all of them, this heterogeneity arguably condemns it to turn out to be essentially void of any interesting implications. While these concepts/processes/[process type]s share some family resemblance, they are too different to be encompassed within a framework.⁷

It is the detail and greater specification given by particular "instances"/"types" of "generalized evolutionary dynamics" that allow the development of theories suited to describing these "instances"/"types".

Here's a comparison that might turn out to be appropriate. We have a very dry, basic theory of sets that doesn't say many interesting things. However, it serves as a foundation for more interesting theories such as functional analysis, topology, or group theory. For another mathematical comparison, take abstract algebra. Wikipedia has pages on group theory and ring theory but there's no page on "monoid theory". Apparently, monoids are too dry, too basic, too underwhelming to "deserve" a meaningful theory. Similarly, the concept of differential persistence is too underwhelming to "deserve" a meaningful theory. To get an object "deserving" of a meaningful theory, we need a collection of phenomena with sufficient implicative capacity, such as genes, populations, individuality, etc.

Henceforth, I will call a collection of phenomena that specify the processes underlying a specific instance or category of (generalized) selection "a substrate of selection". There is perhaps a notion to be introduced here of something like a "meaningful theory-sufficient substrate of selection". I have an intuition that this is a good way to go though it first requires developing better concepts for this.

A skeletal theory of generalized selection?

Set theory (or some other foundation) in itself is rather impotent in itself but it does serve as a foundation on top of which we've developed abstract algebra, analysis, topology, &c. If this analogy is appropriate, we might need (or at least benefit from) some minimal/skeletal theory of generalized selection. It likely will seem mostly pointless and perhaps chasing its own tail but its value is meant to be holist, evaluated in the context of what specialized theories of selection(-like processes) can be built on top of it. This theoretical base is not going to be deep (in and of itself) but it should be something more than (some trivial restatement of) "whatever shit happens, it happens because it happens".

I don't have this theory. Not right now. But here are some concepts that it may involve if it turns out to be a Thing™.

The world is created already in motion. At the most basic level, selection is being done by (the illegible stuff of which are best current description is) physics. Here's a qestion #-/❔ can physics be thought of as selection in the Bouchardian sense of the word?

• world-state transition function / higher-level process
• persistence
• fitness
• measure of fitness
• individuality
• selection/control distinction

Changing substrates of selection

Sometimes the substrate of selection changes, e.g. arguably when human cultural learning became the main driver of human expansion, replacing genetics that so far has been the main driver. Could a generalized theory of selection (along with a good understanding of both the pre-transition substrate and the post-transition substrate of selection) be helpful in understanding this transition? Perhaps. But it may also turn out to be that there isn't much more to explain than something like the following

(Not claiming that this is the most joint-carving description schema in this particular case.)