Abstract
Matthen (Philos Sci 76(4):464–487, 2009) argues that explanations of evolutionary change that appeal to natural selection are statistically abstractive explanations, explanations that ignore some possible explanatory partitions that in fact impact the outcome. This recognition highlights a difficulty with making selective analyses fully rigorous. Natural selection is not about the details of what happens to any particular organism, nor, by extension, to the details of what happens in any particular population. Since selective accounts focus on tendencies, those factors that impact the actual outcomes but do not impact the tendencies must be excluded. So, in order to properly exclude the factors irrelevant to selection, the relevant factors must be identified, and physical processes, environments, and populations individuated on the basis of being relevantly similar for the purposes of selective accounts. Natural selection, on this view, becomes in part a measure of the robustness of particular kinds of outcomes given variations over some kinds of inputs.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Throughout this, I will write of “individual organisms” and “types of organisms” as these are the most common items of interest. However, the same points will apply to the evolution of e.g. groups and any other units of interest.
In the case of e.g. disruptive selection, the statistical expectation is that one of two outcomes will occur, each of which will have a mean value and an associated distribution of outcome frequencies.
Actually, the deterministic claim is a strong metaphysical assumption; in the case of a coin toss, under carefully controlled conditions, determinism of this sort might well be true (see Diaconis et al. 2007, for details). But even leaving aside quantum indeterminacy, it is unclear if this kind of determinism it is defensible more generally; Dupré for example has compellingly argued that this kind of determinism does not hold more generally (see Dupré 1993). In any event, even if in a particular population some organism-event interactions are truly non-deterministic, this kind non-determinism still does not capture the sorts of trends that e.g. selection refers to; it is in part for this reason that Matthen does not consider “true” under-determination to be something encompassed by statistically abstractive explanations.
As noted below, there is one set of inputs to which we do have easy epistemic access that is systematically related to the outcomes of interest—namely, the side of the coin that faces upwards when the coin is flipped is (somewhat) more likely to be the side that faces up when the coin is caught (and, possibly, even more likely to be the side that faces up if the coin is permitted to “bounce” off a surface and come to rest “naturally”) (see Diaconis et al. 2007, 218–219). If we cared about the pattern of heads and tails, this fact would not be of much interest to us; if we cared about the proportion of heads to tails, and if the side facing up when the coin was flipped was chosen at random, then, again, this fact would not be of much interest to us; if we cared about the chances that the coin would come up heads and the side facing up at the flip was known to us, this fact might be of interest to us (if the small bias mattered for our purposes). Dennett’s (1991) claim that there may sometimes be no grounds for determining whether a particular pattern is “real” or which of two possible patterns is the “correct” pattern may be of some interest in this context (see esp. pp. 33–36).
Note that selection in the “process” sense does not guarantee the presence of selection in the “formal” or “statistical sense” (other processes may work against that process, or particular features of the developmental system of the organisms in question may interfere); however, selection in the formal or statistical sense does imply that there is selection in the process sense, though it does not uniquely specify the process in question.
The key exceptions here are those cases where the traits in question are inevitably lethal; in these cases, the physical processes that make the traits inevitably lethal fully explain the formal selection against the trait in question.
To continue the coin-toss analogy, when a coin that is biased 60–40 in favors of “heads” comes up heads, we cannot explain the outcome by referring to the “bias.” Rather (assuming a reasonably deterministic set-up), the particular details of the coin’s toss—the angular momentum, time in the air, etc.—determine the outcome. The bias explains why some outcomes are more common than others, but cannot explain any particular outcome (see Fig. 2b, 3b) (The exception, again, is the case where the coin is “biased” such that it must always come up heads!).
Again, it is worth noting Dennett’s claim (1991) that there may be “no ground” for determining which of two (putative) patterns is the “correct” one—whether we “prefer” a relatively “noisy” but simple pattern, or a less “noisy” but more complex pattern, may, in some cases, be a matter “of taste” (pp. 35–36). As will become clear below, in this case, the more “complex” patterns are those that control for more rather than fewer kinds of variations in inputs. Relatively more detailed specifications leads to explanations that are more like actual sequence explanations, whereas relatively coarser specifications lead to explanations that are more like robust process explanations.
An example: In Sterelny’s (1996) elaboration of the hypothesis that sperm competition is responsible for the very large testicles of chimpanzees, he notes that certain changes in the environment (such as the introduction of “sharped-beaked dive-bombing efficient ballivorous birds,” p. 212) would prevent the evolution of large testicles, even in the presence of sperm competition. Sperm competition as an explanation for large testicles is not robust across every environment in which there is sperm competition!
This obscures the fact that we are generally unable to literally replicate populations at all; but conceptually, the issue of what kinds of replications matter is still an important consideration in thinking about what kinds of variations we should try to abstract away from.
It is worth noting that the above in some ways reverses Matthen and Ariew’s “hierarchical realization” approach; Matthen and Ariew (2009) start with “the set of all possible population histories” and work downwards towards particular populations by adding conditions, each of which limits the set of populations to a realized set of histories (221–222). Approaching the hierarchy from the “ground up”—starting with an actual population and considering what we must permit to vary—highlights how, in practice, the development of selective accounts must actually be approached. .
This is one of the reasons that Matthen and Ariew reject the ‘force’ model of evolutionary change (see 2009, esp. 217).
References
Dennett D (1991) Real patterns. J Philos 88(1):27–51
Diaconis P, Holmes S, Montgomery R (2007) Dynamical bias in the coin toss. Soc Indus Appl Math Rev 49(2):211–235
Dupré J (1993) The disorder of things. Harvard University Press, Cambridge
Futuyma DJ (2009) Evolution, 2nd edn. Sinauer Associates, Sunderland
Jackson Frank, Pettit Philip (1992) In defense of explanatory ecumenism. Econ Philos 8:1–21
Matthen M (2009) Drift and ‘statistically abstractive explanation’. Philos Sci 76(4):464–487
Matthen M, Ariew A (2002) Two ways of thinking about fitness and natural selection. J Philos 99:55–83
Matthen M, Ariew A (2005) How to understand causal relations in natural selection: reply to Rosenberg and Bouchard. Biol Philos 20:355–364
Matthen M, Ariew A (2009) Selection and causation. Philos Sci 76:201–224
Pigliucci M, Kaplan J (2006) Making sense of evolution. Chicago University Press, Chicago
Sterelny K (1996) Explanatory pluralism in evolutionary biology. Biol Philos 11:193–214
Author information
Authors and Affiliations
Corresponding author
Additional information
Kim Sterelny, Massimo Pigliucci, and an anonymous reviewer provided very helpful feedback on earlier drafts of this paper.
Rights and permissions
About this article
Cite this article
Kaplan, J.M. “Relevant similarity” and the causes of biological evolution: selection, fitness, and statistically abstractive explanations. Biol Philos 28, 405–421 (2013). https://doi.org/10.1007/s10539-012-9342-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10539-012-9342-2