Abstract
This paper investigates the conception of causation required in order to make sense of natural selection as a causal explanation of changes in traits or allele frequencies. It claims that under a counterfactual account of causation, natural selection is constituted by the causal relevance of traits and alleles to the variation in traits and alleles frequencies. The “statisticalist” view of selection (Walsh, Matthen, Ariew, Lewens) has shown that natural selection is not a cause superadded to the causal interactions between individual organisms. It also claimed that the only causation at work is those aggregated individual interactions, natural selection being only predictive and explanatory, but it is implicitly committed to a process-view of causation. I formulate a counterfactual construal of the causal statements underlying selectionist explanations, and show that they hold because of the reference they make to ecological reliable factors. Considering case studies, I argue that this counterfactual view of causal relevance proper to natural selection captures more salient features of evolutionary explanations than the statisticalist view, and especially makes sense of the difference between selection and drift. I eventually establish equivalence between causal relevance of traits and natural selection itself as a cause.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Among many instances: Ridley (1999, 90): “natural selection causes both evolutionary change and the evolution of adaptation”; Stern and Orgogozo (2008, 2155): “Natural selection causes predictable changes in phenotypic variation”; Nevo et al. (1997): “Natural selection causes microscale allozyme diversity”; Fox and Wolf (2006, 49): “Natural selection causes change in the phenotypic composition of a population through the differential birth and death of its members”.
Odling-Smee et al. (2003) investigated the processes according to which activities of the organisms change the milieu, hence the selective pressures that bear upon them and supposedly shaped their traits. They claim that niche construction is “a neglected process in evolution” that has to be considered along with natural selection when explaining some organismal or ecological features. Yet one could object that niche construction is not another process than natural selection, it is only natural selection acting in a sophisticated manner; therefore, the whole debate relies on what one means by “causal process” when talking about selection and niche construction.
“Dynamical”, used first by Walsh, Matthen or Ariew, refers to the force talk used by population geneticists. Some authors prefer to talk of “causalists”, but this could be larger than the scheme described here. Nothing depends much upon these labels, anyway.
Stephens (2004) defends the force conception by analyzing this precise pattern of explanation.
Several arguments have been used by the statisticalists: the non-commensurability of selective pressures in a same fitness measure, which distinguishes selection from an addition of forces like in Newtonian mechanics ; the impossibility of distinguishing, among offspring in one generation and given the fitness values, those offspring that have been selected and those which are here by drift ; and therefore the non-commensurability of selection and drift which precludes one from treating evolutionary change as an addition of two forces. Objections have been proposed by Millstein (2005, 2006) considering selection as causal but taking place at a population level. Later Shapiro and Sober (2007) will use the same line of criticism as her: those who endorse a statistical view neglected the difference between “selection as an outcome” and “selection as a process”. As Shapiro and Sober (2007) say, “even if selection the product is not a cause of evolution of trait, selection the process might be”, but the statistical view seems to only consider “selection as an outcome”. However, the main point of the tenants of a statistical interpretation is that selection is not a causal process, so arguing against them about selection “as a process” seems to beg the question. Bouchard and Rosenberg (2004) answered by a defense of ecological fitness, Stephens (2004) rehabilitated the “force talk”, Brandon and Ramsey (2006) based their reply on considering how proceed selectionnist explanations in ecological context, and Riesman and Forber (2005) used Woodward’s manipulationnist view of causation to support a causal view of selection. Matthen and Ariew (2009), Walsh (2007) and Matthen (2009) refined the metaphysical and epistemological foundations of the statisticalist view.
Glennan (2009) has developed a view very similar to the present one. Especially, like in this paper, he relies on a discussion between two meanings of causation, “relevance” and “production”—while I distinguish “difference-making” or counterfactual aspect of causation, from the “process” aspect, and then we apply this difference to the analysis of selection’s causal nature (see Hall (2004) for a defense of this duality in meaning). Similarly, he accepts that, as to production, there is nothing more than the individual interactions, so that selection is not a cause in this sense; but that in terms of causal relevance there is something causal about natural selection, which is not exhausted by the analysis of causal (sensu production) interactions at the level of individuals. The difference between our views is in the way I understand this causal relevance, and the counterfactual analysis I make about the connection between trait values and trait frequencies. However, the bulk of the present argument has been written before its publication, so I did not know the paper and owe the knowledge of its existence to an anonymous referee. Given that it’s a case of independent convergence of views, I acknowledge here the proximity and will not detail at length the points of agreement, which would be useless.
This word means that the selective pressures which are at work and determine which traits are likely to be selected will always depend upon the ecological framework.
The issue of force had been considered by Endler (1986); his analogon was erosion, and he forcefully rebutted the force-talk. More recently Lewens (2009) has shown the exact measure within which Sober says selection is a force; he distinguishes then the “force of selection”, which can be equated with variation in fitness and does not exactly allow one to capture what causes evolutionary change, and “selection-for property”, which is causal but may actually depart from the force of selection and its predictions. As Lewens acknowledges, this detailed understanding of Sober’s view considerably decreases the difference with the statisticalist view, because all of them agree that what actually causes evolutionary change, at the ecological level (“selective forces” sensu Lewens (2004)), differs from the “force of selection”, captured in population genetics models. However, the contrast between a dynamical and a statistical view of selection is still useful to deal with the problem, and corresponds to the contrasts between statistical and causal properties advocated by Walsh (2007).
This commitment to local, process-based causation has been recently made clear by Matthen and Ariew (2009)—discussed in a companion paper to this one.
The counterfactualist view is more complex, because Lewis elaborates upon this concept of counterfactual dependence a concept of causation which excludes from all counterfactual relations these which would lead to backtracking counterfactuals, problems of preemption, etc. These subtleties are not relevant for my point here, so I skip them; I will later on take from Lewis’ analysis what is necessary to make my case regarding selection.
Endler (1986) gave an early critique of the idea of selection as a force, arguing for a view of selection very close to the one articulated later by the statisticalists. In this case, the statisticalist analysis of course fits the case of the Poeciliid no less than it fits the other cases of selection; however, I emphasize a consequence of this study, namely the possibility of building higher level causal patterns of selection, to which the general statisticalist view seems prima facie inappropriate.
The authors note that this asymmetry is important enough to be a “key innovation” since it “initiated the adaptive radiation of pareatine snakes through south east Asia as dextral snails eaters” (170).
Achinstein (1979) All these preliminary considerations on explanations are controversial but since this paper does not address explanations in general, I have to rely on some extant theories to sketch a difference between explanation and causation.
In fact, some explanations are not answers to why questions, as when we make clear what we said earlier, etc. But about natural selection expected explanations are naturally likely to be casted into why questions (why does pandas have opposable thumbs?, etc.).
Arguing for this claim would start with the fact that a demonstration of X being an attractor relies on some mathematical properties of X as an object in the phase space. The argument for the detachment of causal stories and attractor explanations is that two different systems might have the same or similar attractors X and X’ as mathematical objects in the phase space, which will be both attractors because they display the same mathematical properties, so they are attractors for the same reasons; yet those reasons are not causal stories, because, the systems being of different natures, the causal stories they involve are wholly distinct.
Of course, one must take here into account the background condition constituted by the fact that currency is divided into bills of $1, $5, $10 etc. If you change this distribution, if bills are of $1, $2, $3 and $100, the results about crumpledness will be different, however the same reasoning about the explanatory role of the value of the bills will still hold.
I do not enter here into complications proper to the real cases in a given economy, since the US Federal Reserve System can decide to create immensely more $1 bills than $100 bills, which would oppose my demonstration. I will not either consider the fact, which is true, that bills that are too crumpled are eventually replaced (but they have to be really very crumpled). My example is a bit like examples in combinatory theory, when one first treats the simple cases of urns with no replacement of balls, before considering the with-replacement case. Nevertheless, the empirical appearance of our various bills in real life seems very close to the ideal case I consider, making plausible that the two facts I just mentioned are in real life as negligible as they are neglected in my ideal case.
First of all because no conserved quantity is transmitted between the economical property “$1” and one given exchange—while the process-view of causation requires conserved quantities.
In fact, we should have to allow very different values for things, and another set of bills which would range up to something like $10,000, which seems further from our world than simply multiplying the $1 bills in order to keep the structure $1/5/10/50/100.
Another example would be the last individual in a small population of mimetic insects; all the good ones (good camouflages) have been eaten, there is only this (low fitness) one, and it gets eaten. The counterfactual “if it had been of the other color, it would have survived”, intuitively expressing the fact that it has been selected against, won’t hold (well, it’s the last insect left, it probably would have been eaten anyway).
Lewens (2009), developing Sober’s account, see cases of selection for the low fitness variants as example of a divorce between selection as a force (defined by variation in fitness), and selection for properties (being actually the cause of evolutionary success). However, here there is no “selection for properties”, for reasons explained in the next paragraph.
Computer simulations can often indicate the behavior of a set of hypothetical populations, because we can keep fixed the parameters (size, initial frequencies) and run the dynamics many times.
If one accepts objective chances (Mellor 1995), then this formulation could easily be turned into chances talk, chances being the hypothetical limit frequencies.
This consideration solves the “last living insect” case too.
This increase, more rigorously, has to be understood in the context of sets of hypothetical populations as noted above.
Notice also that those counterfactuals are not backtracking because an evolution by drift is always possible.
Lewens (2009) and Walsh (forthcoming) actually emphasized cases of Gillespie fitness, i.e. cases where defining fitness includes the variance of the trait distribution, and hence changing fitness implies changing population size, which makes impossible to distinguish an intervention on fitness (hence selection) and an intervention on population size (hence drift), so that no legitimate intervention, hence no causation in the manipulationnist sense, can be defined.
This is a rather crude model, not even a Fisher-Wright model for single locus diploid population, since we consider only genotypes and not alleles; all the subtleties of population genetics are here not at stake, it’s like a population of asexual haploids. However, the conceptual point made here still concerns real population genetics, because as we saw the values of the variables in the counterfactual statement can be understood as competing alleles in a one locus model.
Note that if there are several traits, it is highly probable that any reliably occurring factors will at least differentially affect one type of trait so in practice, reliably occurring factors suffice to define selective pressures.
Here one can define drift with Millstein (2006) in terms of biological processes like gamete sampling, etc.; the point is that in various possible worlds as close as one wants, those events won’t produce the same outcome, unlike what we call selection.
“Habitat conditions in this area are very stable and most resident fish populations are highly persistent with respect to their density” (Mori 1993, 217).
This value d will also define the variable to manipulate in the Reisman and Forber’s account.
Some could find Lewis’s modal actualism a high price to get some causal meaning of selection, but nothing in the present argument is tied to this interpretation of possible worlds; Stalnaker also made sense of counterfactuals with possible worlds but with a modally weaker metaphysics.
The account is refined above (Sect. 2) in order to take into account stabilizing selection, which could be objected to the present formulation.
Lewens (2009) pointed out that in some cases “selection for” cannot be equated with the force of selection, based on the variations in fitness values. Yet “natural selection” here is not what he calls “force of selection”, but just the causal scheme subsuming the specific explanations by natural selection for properties. The full account of this difference needs another paper aiming at distinguishing how selection differs from drift in each viewpoint. At the most general level, here I agree with Lewens’ conclusion that defining the nature of selection cannot avoid to specify the explanatory context conditioning which adaptations occurred, and especially to decide whether one explains rather the ecological context or the variations in allele frequencies.
References
Achinstein, P. (1979). The nature of explanation. New York: Oxford University Press.
Bouchard, F., & Rosenberg, A. (2004). Fitness, probability and the principles of natural selection. British Journal for Philosophy of Science, 55, 693–712.
Brandon, R., & Ramsey, G. (2006). What’s wrong with the emergentist statistical interpretation of natural selection and random drift. In M. Ruse & D. Hull (Eds.), The Cambridge companion to philosophy of biology. Cambridge: Cambridge University Press.
Dretske, F. (1972). Contrastive statements. Philosophical Review, 81(4), 411–437.
Dretske, F. (1986). Misrepresentation. In R. Bogdan (Ed.), Belief, form, content and function. Oxford: Clarendon Press.
Endler, J. (1982). Convergent and divergent effects of natural selection on color patterns in two fish faunas. Evolution, 36, 178–188.
Endler, J. (1986). Natural selection in the wild. Princeton: Princeton University Press.
Fox, C., & Wolf, J. (2006). Evolutionary genetics: Concepts and case studies. Oxford: Oxford University Press.
Glennan, S. (2009). Productivity, relevance and natural selection. Biology & Philosophy, 24(3), 325–340.
Hall, N. (2004). Two concepts of cause. In C. J. Hall & L. A. Paul (Eds.), Causation and counterfactuals (pp. 225–276). Cambridge: MIT Press.
Hall Ned, C. J., & Paul, L. A. (Eds.). (2004). Causation and counterfactuals. Cambridge: MIT Press.
Harvey, P. H., & Pagel, M. D. (1991). The comparative method in evolutionary biology. Oxford: Oxford University Press.
Hori, M. (1993). Frequency-dependent natural selection in the handedness of scale-eating cichlid fish. Science, 260, 216–219.
Hoso, M., Asami, T., & Hori, M. (2007). Right-handed snakes: convergent evolution of asymmetry for functional specialization. Biological Letters, 3(2), 169–173.
Lande, R., & Arnold, S. (1983). The measurement of selection on correlated characters. Evolution, 37(6), 1210–1226.
Lewens, T. (2004). Organisms and artefacts. Design in nature and elsewhere. Cambridge: Cambridge University Press.
Lewens, T. (2009). The natures of selection. British Journal of Philosophical Science, 1–21.
Lewis, D. (1973). Causation. New York: Oxford University Press. Philosophical papers II.
Matthen, M. (2009). Drift and ‘Statistically abstractive explanations’. Philosophy of Science, 76, 464–487.
Matthen, M., & Ariew, A. (2002). Two ways of thinking about natural selection. Journal of Philosophy, 49(2), 55–83.
Matthen, M., & Ariew, A. (2009). Selection and causation. Philosophy of Science, 76, 201–224.
Mellor, D. H. (1995). The facts of causation. London: Routledge.
Menzies, P. (2001). Difference-making in context. In C. J. Hall Ned & L. A. Paul (Eds.), Causation and counterfactuals. Cambridge: MIT Press.
Millstein, R. (2002). Are random drift and natural selection conceptually distinct? Biology and Philosophy, 17(1), 33–53.
Millstein, R. (2006). Natural selection as a population-level causal process. British Journal for the Philosophy of Science, 57, 627–653.
Millstein R., Skipper R., & Dietrich M. (2009). (Mis)interpreting Mathematical Models: Drift As A Physical Process. Philosophy and Theory in Biology, 1.
Neander, K. (1990). Dretske’s Innate Modesty. Australasian Journal of Philosophy, 74(2), 258–274.
Nevo, E., Apelbaum-Elkahert, I., Garty, J., & Avigdor, B. (1997). Natural selection causes microscale allozyme diversity in wild barley and a lichen at ‘Evolution Canyon’, Mt. Carmel, Israel. Heredity, 78, 373–382.
Odling-Smee, J., Laland, K., & Feldman, M. (2003). Niche-construction: Teh neglected process in evolution. Princeton: University of Princeton Press.
Putnam, H. (1975). The meaning of ‘Meaning’. Minnesota Studies in the Philosophy of Science, 7, 215–271.
Ridley, M. (1999). Evolution. Oxford: Oxford University Press.
Riesman, K., & Forber, P. (2005). Manipulation and the causes of evolution. Philosophy of Science, 72, 1113–1123.
Salmon, W. (1998). An At–At theory of causal influence. Causality and explanation. New-York: Oxford University Press.
Shapiro, L., & Sober, E. (2007). Epiphenomenalism—The Do’s and the Don’ts. In G. Wolters & P. Machamer (Eds.), Studies in causality: Historical and contemporary. Pittsburgh: University of Pittsburgh Press.
Sober, E. (1984). The nature of selection. Cambridge: MIT Press.
Stephens, C. (2004). Selection, drift and the “forces” of evolution. Philosophy of Science, 71, 550–570.
Stern, D., & Orgogozo, V. (2008). The loci of evolution: how predictable is genetic evolution? Evolution, 62(9), 2155–2177.
Takahashi, R., Watanabe, K., & Nishida, Hori. M. (2007). Evolution of feeding specialization in Tanganyikan scale-eating cichlids: a molecular phylogenetic approach. BMC Evolutionary Biology, 7, 195.
Wade S., & Kaisz S. (1990). The causes of natural selection. Evolution, 44(8), 1947–1955.
Walsh, D. (2000). Chasing shadows–Natural selection and adaptation. Studies in the History and Philosophy of Biology and the Biomedical Sciences, 31, 135–153.
Walsh, D. (2003). Fit and diversity: Explaining adaptive evolution. Philosophy of Science, 70, 280–301.
Walsh, D. (2007). The pomp of superfluous causes: The interpretation of evolutionary theory. Philosophy of Science, 74, 281–303.
Walsh, D. (forthcoming). Variance explanations.
Walsh, D., Lewens, T., & Ariew, A. (2002). Trials of life: Natural selection and random drift. Philosophy of Science, 69, 452–473.
Waters, C. K. (2007). Causes that make a difference. Journal of Philosophy, 104, 551–579.
Woodward, J. (2003). Making things happen. New York: Oxford University Press.
Acknowledgments
The materials from this paper have been presented at the CSHPS meeting in York (2006) and the University of Toronto workshop on the statistical interpretation of selection (2008). The author is grateful to the audiences for their suggestions and critiques. Many thanks in particular to Jean Gayon, Jim Griesemer, Françoise Longy, Francesca Merlin, Roberta Millstein, Michael Strevens, Denis Walsh, for their useful comments on earlier versions. The text has been carefully language-checked by Denis Walsh, whom I warmly thank for that. Finally, I thank two anonymous referees for having greatly improved the paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Huneman, P. Natural Selection: A Case for the Counterfactual Approach. Erkenn 76, 171–194 (2012). https://doi.org/10.1007/s10670-011-9306-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-011-9306-y