Redemption of the Traditional Notion of Selection

Traditionally the notion of selection is specific to the domain of biological evolution:

It is about a cause of evolution. In addition, most philosophers as well as biologists nowadays also speak of selection as a process (if not a force). But, every case of evolution (by selection), and hence every case of selection, is wholly constituted by organismal-level occurrences, i.e. processes/events each involving only some (not limited to one), but not all, of the organisms of a population. According to Matthen and Ariew, this constitution thesis speaks against the reification of the term “selection”: It renders redundant the posit of something called “selection” which is over and above the organismal-level occurrences in a case of what is called “evolution by selection” and which caused the given evolutionary change. Further, although “variation in wing darkness caused the evolution of the moth population” is an acceptable causal claim, it doesn’t have the same meaning as “selection caused the evolution of the moth population” and doesn’t talk about certain process. Can the traditional idea of selection survive these challenges? Yes it can. Below we’ll see that the conventional talk can be understood in a way that is consistent with the constitution thesis. In addition, a careful assessment of the above attacks will reveal that they themselves rest upon several mistakes so that they pose virtually no threat to the traditional notion.

We start by establishing the grounds for reifying “selection” in the ordinary sense of a process such that variations in certain features amongst a population of organisms lead to the variation in their reproductive success. This latter idea is nothing but the Huxley-Darwin cliché and Lewontin’s well-known HVF formula (that the condition for selection is heritable variation in fitness) precisified and neutralised (by leaving unspecified what exactly those features are). It is also what underlies the two accounts

of selection to be examined in later chapters. At this stage, we can leave aside questions like whether the characterising features are causal properties or even properties at all, whether selection so characterised is a causal process, and whether the characterising sentence expresses a law or even a causal law. We just ask: In a particular case of evolution by selection, what is picked out (or can be picked out) by the term “selection”

so understood? This is essentially a question regarding the reference of “selection” in any given case. The term can be reified only if there is a process to which it can plausibly refer in any given case. Does such a process exist, then?

Of course it does. It is basically just the collection of every event/process involving any of the organisms of a given population within a given generation, which includes any sort of activity, suffering, mutual interaction, interaction with organisms of other populations and reaction to the environment. Or, equivalently, it is the congeries of life histories of any and all of those organisms. This is only natural because “selection”

doesn’t apply to any particular organismal-level occurrence; nor does it pick out any specific type of organismal-level occurrence or any fixed set of specific types of such occurrences in any given case. It obviously applies to the whole case. And a case of selection (as well as evolution) is constituted not only by the subject, i.e. any and all organisms of a population within a generation, but also by each and every organismal-level occurrence in which any of those organisms participated in any time-interval within that generation-time (i.e. from their births until the reproductive events/processes or their deaths in a discrete population). Or better put, a case of selection is constituted by all and only those organismal-level occurrences of each of which some organisms of a given population within a given generation are a part. So long as the attribution of

“being a case of selection” is based upon the connection between features of organisms of a population within a generation and their reproductive success, the reference to such

a collection of indeterminate organismal-level occurrences is natural and inevitable, for whatever effects of features of organisms are parts of organismal-level occurrences and those features as well as their immediate effects are connected to reproductive success only through those occurrences. Thus, “selection” picks out any such collection of organismal-level occurrences that also exhibits a certain connection between variation(s) in certain feature(s) of organisms and the variation in their reproductive success. It is (presumably) a type-term, and accordingly any particular collection of organismal-level occurrences picked out by it is an instance of selection. In any given case of selection,

“selection” just denotes such a whole collection of organismal-level occurrences.

The above shows that the term “selection” in the conventional sense is referential and therefore selection so understood exists. This fully establishes the grounds for reifying

“selection” in that sense. And it also implies that the reification of “selection” is fully compatible with the constitution thesis and, hence, Matthen and Ariew are mistaken in asserting otherwise. They go wrong in two aspects. First, they think that reifying a term amounts to posting something extra. This is not generally true and is not true of our case.

What is posited is not a separate force that acts upon a population of organisms, an additional something that can be metaphorically understood as a motor that drives them to exhibit a new allelic frequency distribution, or a tertium quid that mediates between variation(s) in certain feature(s) of organisms and a new allelic frequency distribution among them. Rather, it is a type of collection of certain pre-existing organismal-level occurrences as characterised above. Its posit is grounded in a new general (and presumably explanatorily useful) way of describing and classifying collections of those pre-existing occurrences. In fact, when Matthen and Ariew identify selection with a formal pattern of change, they are also attempting, though unsuccessfully, to posit a type of change or occurrence, any instance of which in the domain at issue is exactly

one such collection of pre-existing organismal-level occurrences. So, they are reifying

“selection” too. The difference between their account and the ordinary talk is only that the two (supposed) types being posited are different: While the one in the former is supposed to be defined by a mathematical theorem concerning number count, the one in the latter is based upon a regular connection between variations in features of organisms of a population.

Second, it is erroneously to think, from the very beginning, that the constitution thesis furnishes a reason for the non-existence of selection as traditionally understood.

What the constitution thesis expresses is, no more and no less, that selection is token-identical to a collection of organismal-level occurrences. This alone already asserts that selection-instances exist, which then implies that selection as a type of certain collection of organismal-level occurrences exists. Matthen and Ariew here seem to make a mistake similar to the confusion between (ontological) reduction and elimination. The fact that something is constituted by some other things doesn’t suggest the elimination of the former in the slightest. A ball doesn’t cease to exist because it is wholly constituted by a collection of atoms. Likewise, an occurrence that is a ball’s breaking a window doesn’t cease to exist because it is wholly constituted by a collection of atomic-level occurrences. And the same goes for selection.

Similar considerations also indicate that selection-instances are themselves processes and, arguably, causal processes. Given that an instance of selection is a collection of organisma-level processes, I see no reason why it cannot itself be an occurrence or process. It is surely not an organismal-level occurrence; we may say that it is a population-wide and basically generation-long process. Or, relative to those constituent organismal-level occurrences, it is a macro-process. Moreover, in so far as membership of a biological population is partially defined in interactive terms, the organismal-level

occurrences constitutive of an instance of selection are causally bound (and temporally confined if not also more or less spatially localised), and there should be no difficulty in accepting that a selection-instance is a causal process. Both of these reveal that, once we accept the constitution thesis, it is just a small step away from also recognising that a selection-instance, a said collection of organismal-level occurrences, is a process and a causal process for that matter. This suggests that the question of whether selection is a causal process is almost trivial. The debate over whether the various regularities regarded to characterise selection are causal or not is a totally different matter. Failure to distinguish between the two betrays the confusion between causation and causal efficacy of properties, a crucial distinction to be made clearer at the end of this chapter.

Now, in a given case of evolution by selection, did selection so understood cause evolution? In Matthen and Ariew’s opinion, the correct way of talking about the matter is to say that various organismal-level events like predations and matings severally caused the deaths and births of organisms (which are themselves also organismal-level events), which then collectively “constituted” evolution. This way of looking at the matter is flawed in two respects. First, such a picture makes the identification of selection in the ordinary sense particularly non-straightforward if not precluding it a priori. It makes no sense to say that a collection of births and deaths is an instance of

selection, and incorporating only births and deaths and their respective immediate causes into an instance of selection is too narrow-scoped. As seen above, the talk of selection applies to the whole of the lives of organisms of a population in a generation:

It is a regularity exhibited by some such collections of life-histories that is based upon certain (supposed) properties of organisms and their reproductive successes. Those property-instances don’t influence reproductive successes just in the immediate causes of births and deaths. Their influences in that respect are usually cumulative and indirect

and the causal conditions complex and holistic, so that the talk of selection should apply to the whole collection of causal chains featuring organisms of a population within a generation during all their lives. This suggests that we fuse all the births and deaths and their immediate causes and everything those organisms participate before those events up to their own births and talk, instead, of a macro-process that lasts generation-long.

And these considerations are missing in the picture given by Matthen and Ariew.

Second, saying that births and deaths constitute evolution doesn’t quite capture the notion of (biological) evolution as cross-generational change in allelic frequency distribution for a population of organisms. Such a “change” is actually a difference between two feature-tokens (two allelic frequency distributions) of two different collections of organisms (those in a given generation and those in the next generation).

So, in any case of evolution by selection, evolution is a matter of a collection of parent organisms’ reproducing a collection of offspring organisms having any of certain collective genetic features that is different from that of their parental generation.

Making this explicit unveils what causes evolution in such scenarios. For asking what caused evolution in a given case of evolution by selection is asking what caused the reproducing, by a certain collection of parent organisms, of a certain collection of offspring organisms having a certain collective genetic feature. And the answer is exactly the particular instance of selection in that case, i.e. the macro-process that is the whole of the organismal occurrences involving the parent-generation organisms, which extended from their births until at least the births of their offspring and which betokened a certain regularity that is based upon their certain properties and their reproductive successes (in virtue of the tokening of which it is an instance of selection). The traditional idea of selection being a cause of evolution is thus thoroughly intelligible and perfectly feasible.

There may be some worries about the talk of causation in this context. Standard metaphysics dictates that causal relata are events. Yet selection-instances are said to be processes. And a token of evolution, being basically a collection of certain reproductive events or results, may not be legitimately taken as a macro-event, for the reproductive events may not be spatiotemporally contiguous. Moreover, in so far as an instance of selection comprises everything the parent-generation organisms participate from their births until the last reproductive events at least, the collection of all the reproductive events itself is to be considered a part of the selection-instance. As a result, even if a process is allowed to cause, there is no separate (macro-)event to assume the role of the other causal relatum. Such worries, however, can be removed by appealing to the notions of process and product. A case of evolution by selection can then be regarded as a (macro-)process, a selection-instance as previously identified, producing a final product, a collection of offspring-generation organisms having certain collective genetic features different from those of the parental generation. This eliminates the need for segregating a collection of organisma-level occurrences into two non-overlapping aggregates or dividing the life-histories of the parent generation into two temporally distinct series. It also erases the demand that the proper talk of cause and effect apply only to event-pairs. And there is a perfectly good sense in which a collection of offspring-generation organisms having certain allelic frequency distributions dissimilar to the parental generation is a product of the parent-generation organisms’ being involved in the organismal-level occurrences during their lives so that their certain properties and their reproductive successes exhibit a certain pattern that is an instance of the regularity characterising selection. More crucially, the notion of production is so akin to that of causation that no one would deny that it is a causal notion. Consequently, the ordinary talk can still be properly understood as a causal talk, albeit a non-standard

one.

The correct understanding given here also indicates that it is a mistake for Matthen and Ariew to acknowledge that “variation in wing darkness caused the evolution of the moth population” is a legitimate causal claim, yet to insist that it doesn’t have the same meaning as “selection caused the evolution of the moth population” and doesn’t represent a process. No doubt variation in wing darkness is not identical to selection:

The former is a collective feature and the latter is a macro-process. But in the context of explaining a given case of evolution, those two sentences should be interpreted as talking about one and the same thing: the complex of the macro-process which is the particular selection-instance and its product which is the particular token of evolution in that case. Alternatively, we may literally take “variation in wing darkness” to mean a certain collective feature of the parent-generation organisms. In that case, one is just citing the most important characteristic or constituent of the macro-process to explain the presence of the product. In fact, so long as the first sentence is accepted as a legitimate causal claim, variation in wing darkness has to be taken as a property and its instance can be said to cause only on condition that it is a constituent of some occurrence, which in our case is simply an instance of selection. Hence, the claim that variation in wing darkness caused the evolution of the moth population just implies the claim that selection caused the latter, and both are talking about a certain macro-process producing the given result or product.

Matthen and Ariew’s fundamental mistake in thinking that evolution cannot have a macro-cause called selection but can only be constituted by micro-occurrences that are births and deaths is, I believe, that they fail to (fully) appreciate the fact that the same occurrence can be described in different ways. This is also what underlies their peculiar idea, as exposed above, that reifying “selection” is positing, redundantly, some extra

cause of evolution. Their objection to the posit of a separate force or the like is argued, essentially, by setting up a causal exclusion problem for any such extra cause of evolution and is well taken. There can be a causal exclusion problem only when the competing causes are distinct. In the case of positing selection as a cause of evolution that is over and above the organismal-level occurrences or the organismal properties, selection is distinct from the latter and there is indeed a causal exclusion problem. That exclusion problem can be readily avoided by eliminating altogether such a putative cause, i.e. by refraining from positing selection in this sense at the very beginning, because it is blatantly explanatorily superfluous. But the same problem cannot arise for the ordinary talk as explicated above since selection, as a type of macro-process, is token-identical to, and hence indistinct from, a collection of organismal-level occurrences. The macro-causal/productive relation between a selection-instance, a macro-process, and a collection of offspring organisms, a collective reproductive result or product, may be said to be constituted by a collection of organismal-level causal and productive relations. Surely this doesn’t imply that the macro-causation doesn’t exist,

“any more than the fact that [a] baseball is composed of microparticles entails its nonexistence” (Kim 2003: 167). It does imply, however, that there is in reality only one set of causations. The key point is: This single set of causations can be described in different ways. It can be described collectively as a single macro-causation or separately as a lot of organismal-level causations (or a myriad of atomic-level causations and so on). Although there is a sense in which some descriptions are more fundamental than others, all of them are equally legitimate in the sense that none of them is truer or more authentic than another. All of them are about the same things and occurrences. The same set of occurrences, causes and causations won’t mysteriously cease to exist just because they are described as a single macro-causation rather than as a lot of

organismal-level causations.

But selection is essentially a type (of process), and so is evolution (a type of product).

And they are not type-identical to any other types. More importantly, selection is not type-identical to the type that is the collection of all life-histories of all organisms of a population within a generation (if this is indeed a type), for it is further characterised by a regularity concerning variations in certain organismal properties and variation in reproductive success. That is, not all collections of all life-histories of all organisms of a population within a generation are selection-instances. Likewise, not all collections of all births of offspring organisms (and all deaths of the parent-generation organisms) are tokens of evolution. Then, the type of macro-causal relation the instance of which is between a selection-instance and a token of evolution is not identical to the type of macro-causal relation the instance of which is between the occurring of a collection of all life-histories of all organisms of a population within a generation and the occurring of a collection of all births of their offspring organisms. The latter is just reproduction

And they are not type-identical to any other types. More importantly, selection is not type-identical to the type that is the collection of all life-histories of all organisms of a population within a generation (if this is indeed a type), for it is further characterised by a regularity concerning variations in certain organismal properties and variation in reproductive success. That is, not all collections of all life-histories of all organisms of a population within a generation are selection-instances. Likewise, not all collections of all births of offspring organisms (and all deaths of the parent-generation organisms) are tokens of evolution. Then, the type of macro-causal relation the instance of which is between a selection-instance and a token of evolution is not identical to the type of macro-causal relation the instance of which is between the occurring of a collection of all life-histories of all organisms of a population within a generation and the occurring of a collection of all births of their offspring organisms. The latter is just reproduction