Selection

Tonight in an email exchange with Ian, the subject of selection came up.  Or perhaps more accurately, the problem of defining selection pressure came up.  Over the past couple of entries, I've been excited about thinking of group tightness and looseness as responses to differential selection pressures acting at the group and individual levels.  There is something intuitive about the idea that just "seems right."  As is so often the case, though, translating that gut intuition into something meaningful and practical is proving more difficult.

One stumbling block in this is that the entire scheme relies on understanding selection pressures that might act on either the group or the individuals comprising the group.  But what exactly do we mean by this?  How can we operationalize it?  How can we measure it?  My tack so far has been to come up with as many examples as I can of different groups falling into different categories of tightness and looseness and to think about the selection pressures at work.  While this has been seductive, I've started uncovering examples where the whole scheme seems to break down.  These have been instructive, though, inasmuch as they haven't been able to completely refute the scheme; they've left some room for the overarching idea to still wiggle free.  It's in this ambiguous space that I've come to appreciate where more precision is required.

I think one place to begin is to define what we mean by selection.  Without a firm conception of selection, any talk of selection pressure become rather silly.  I'd like to have a more exhaustive inventory of accepted definitions, but it's already past my bedtime so for now I'll just share the definition given by Hodgson and Knudsen in Darwin's Conjecture:

Selection in a complex population system involves an anterior set of entities that is somehow being transformed into a posterior set, where all members of the posterior set are sufficiently similar to some members of the anterior set, and where the resulting frequencies of posterior entities are correlated positively and causally with their fitness in the environmental context.  The transformation from the anterior to the posterior set is caused by the entities' interaction within a particular environment.  (pp. 241-242)

Interestingly, they differentiate between two different kinds of selection:  subset selection and successor selection:

Subset selection is defined as selection through one cycle of environmental interaction and elimination of entities in a population, structured so that the environmental interaction causes elimination to be differential, and where survival outcomes are correlated positively and causally with fitness in that environment.


Successor selection is defined as selection through one cycle of replication, variation, and environmental interaction, which leads to differential replication, novel entities, and a changed distribution of population properties that correlates positively and causally with the fitness of entities in that environment.  (p. 242)

Fitness plays a role in all of these definitions, so it's probably a good idea to pin what they mean by that down, too.  This turns out to be a bit thornier than you might expect:

In biology, fitness is most usefully defined as the propensity of a genotype to produce offspring (DeJong 1994).  Survival of the fittest is no longer a tautology:  it is possibly false.  The fitness of a replicator is the propensity to increase its frequency (relative to other replicators).  In the social domain, this definition of fitness translates into the propensity of a social replicator (such as a habit or a routine) with a particular feature to produce copies and increase the frequency of similar replicators in the population.  The fitness of an interactor is the propensity of its replicators to increase their frequency. (p. 238)


Here I sense some slippage between these concepts.  Much, I'm sure, to Ian's delight Hodgson and Knudssen define selection in terms of elimination of entities due to their interactions with the environment.  However, their notion of fitness seems to be more in line with conceptions of  selection for entities.  Perhaps that's neither here nor there, but it is curious to note all the same.

So what to make of these definitions?  At this point, I'm not entirely sure.  More than anything, I just wanted to get them out there for us to begin playing with.  However, some preliminary thoughts are that if we define selection in a similar way, then selection and existential security are in many ways interchangeable.  Selection pressure, which oddly isn't defined at all by Hodgson and Knudsen, would seem to be a property of the environment which affects how much variation can be present within a given population interacting within a particular environmental context.  If this is so, then it would seem to be even more directly related to notions of tightness and looseness than I first thought.  At the very least, it's not hard to see how I could come to the idea that selection pressures are key to understanding these phenomena.