Thread by @yoginho

A little thread on maybe *the* false dichotomy in evolutionary biology: self-organisation vs. natural selection… 👇🏻

Let’s start with Lewontin’s minimal conditions for Darwinian evolution: (1) phenotypic variation, (2) inheritance, (3) differential fitness. To get evolution, these conditions must be met. /1

Note: you must have ontogenesis (a life cycle or “the acquisition of the capacity to reproduce") to get variation & inheritance. If you don’t buy this, read Griesemer’s excellent “Genetics from an Evolutionary Process Perspective.” /2

How do you get a closed life cycle? You (a) need organisational closure (autopoiesis, self-maintenance, self-repair) to live, and (b) an excess of self-production over decay. This implies reproduction (see Saborido et al. 2011). /3

From this account it not only follows that you need organismic organization with closure (and therefore organismic agency), but also a closure of the life cycle by reproduction. Self-organisation during development needs to be seen as acting under these constraints. /4

What you get from this simple agential perspective on evolution is (a) necessarily biased variation (under the constraints described above), and (b) an organism that can actively pursue its goals & alter its environment. /5

In other words, you get Lewontin & Levins’ dialectic dynamics: organisms as both product and source of evolution. This is not just “reciprocal causation” but reciprocal co-generation or co-emergence of the organism and its perceived environment. See Denis Walsh’s 2015 book. /6

This is what Denis means by situated (as opposed to fractionated) *Darwinsm*. He argues that this is very close to Darwn’s original notion of evolution fundamentally based on the individual’s “struggle for life.” /7

What follows is that any dichotomies between an independent environment and the activity of the organism, or between the generation of variation (self-organisation) and its consequences (natural selection) vanish. You get what Walsh calls a symmetric theory of evolution. /8

There is no longer an asymmetry between the organism “solving problems” posed by an independent and external environment, as Lewontin put it. Evolution is all just one, radically emergent, process grounded in the agency of the individual within its perceived environment. /9

This provides a simple and robust metaphysical grounding for evolution as a process, going straight back to Darwin and fully complying with Lewontin’s minimal conditions. But there are two practical problems with this perspective. /10

The 1st is that, for most circumstances, we cannot (& do not have to) follow each individual struggle for life to understand how evolution works. There are higher-level (e.g. population-level) regularities, that can be approximated mathematically. Hence, population genetics. /11

Evolutionary theory is a set of heuristic approximations or models that work for certain problems under certain circumstances. Each gives us a particular perspective on the process of evolution. These perspectives are complementary. But do not add up to a unified “synthesis.” /12

This is not a problem, but a good thing! The “causal structure” of evolution is so complex, that we cannot capture it in all its detail. But there are many useful heuristics, many useful perspectives, that do good work to explain specific phenomena/solve specific problems. /13

Two such perspectives are looking at the source vs. the consequences of phenotypic variation (Sober, 1984; Amundson, 2005). One deals with self-organisation & other aspects of ontogenesis, the other with natural selection & other population-level processes. They complement. /14

Thus, *not* either self-organisation or natural selection, but self-organisation *and* natural selection… /15

The 2nd problem with taking the agential perspective serious is that we do not have the mathematical tools to model its dynamics! The dialectic mutual co-generation of processes is materially impredicative (this is called Rosen’s conjecture). It can’t be modelled. /16

It can only happen. With unpredictable results. It is what enables truly open evolution. It is like a model with equations that write themselves. We have no idea how to even begin to understand this. /17

This is why I sometimes get upset abut the shallow discourse on “reciprocal causation” (e.g. in the #EES debate). That’s not what the “causal structure” of evolution is. Dialectic co-generation is ontologically fundamentally different. And much more radical. /18

More on that another time. It’s too complex to discuss on Twitter, really.

Bottom line: the dichotomy between self-organisation and natural selection stems from two different perspectives on evolution. These perspectives complement each other, they do not compete! /19

Both perspectives are based on limited heuristics, the underlying “causal structure” of evolution is unbelievable complex & at heart way beyond what we can tackle with today’s mathematical tools.

It’d be nice to see a more serious conversation on these topics in the field… /20