Let me be that I am and seek not to alter me.
William Shakespeare, Much Ado About Nothing
Abstract
There is a tension between, on the one hand, the view that natural selection refers to individual-level causes, and on the other hand, the view that it refers to a population-level cause. In this article, I make the case for the individual-level cause view. I respond to recent claims made by McLoone that the individual-level cause view is inconsistent. I show that if one were to follow his arguments, any causal claim in any context would have to be regarded as vindicating a form of population-level cause view. I show why this is implausible and how a consistent individual-level cause position can be held within the interventionist account of causation. Finally, I argue that there is one sense in which natural selection might be said to refer to population-level causes of evolutionary change. The upshot is that, as noted by others, natural selection can be regarded as referring to a population-level cause in the context of frequency-dependent selection and other situations of fitness-altering interactions between the individuals of a population. But whether this statement is true will depend on the empirical case investigated, not some a priori conceptual distinction. Thus, even though situations of frequency dependence might be ubiquitous, it is orthogonal to the conceptual question of whether frequency-independent natural selection—McLoone’s target—refers to individual- or population-level causes.
Similar content being viewed by others
Notes
For instance see Matthen and Ariew (2002), Matthen and Ariew (2009), Walsh (2000, 2010), Walsh et al. (2002), Walsh et al. (2017), Bouchard and Rosenberg (2004), Rosenberg and Bouchard (2005), Millstein (2006), Millstein (2013), Reisman and Forber (2005), Shapiro and Sober (2007), Huneman (2012), Stephens (2004), Krimbas (2004), Hitchcock and Velasco (2014), Otsuka (2016b), Bourrat (2018).
By “individual” I will refer to any entity that is part of a Darwinian population after Godfrey-Smith (2009). To be clear, I do not necessarily refer to a biological organism when I use this word.
Note that by distinguishing individual- and population-level properties, I do not mean “population-level” in a sense that would invalidate the supervenience of population-level properties on individual-level properties. I simply mean that our knowledge about a particular evolutionary dynamics could come from properties that we are only able to characterize at the population level.
Note that by “fitness” Bouchard and Rosenberg have in mind an individual-level property that has some effect on reproductive output. Thus, importantly, fitness (which they characterize as “ecological fitness”) is different from reproductive output in their view. For more on the distinction between fitness and reproductive output see Bourrat (2015a, 2017, 2018, 2019a).
As pointed out by Pocheville et al. (2017), there is some ambiguity in the literature surrounding the notion of invariance. The term invariance has been used to refer to, on the one hand, whether, and if so to what extent, a relationship holds as the value of the causal variable is intervened upon; and on the other hand, whether, and if so to what extent, the relationship holds as diverse variables in the background of the relationship are changed. Following Pocheville et al., by “invariance” I mean the first and by “stability” the second.
For a precise information-theoretic measure of stability see Pocheville et al. (2017). Information theory is best suited for categorical variables. To my knowledge, a quantitative measure of stability for quantitative variables does not exist in the literature.
Note also that other individual-level properties in the background of the relationship might be relevant for the stability of the relationship, such as whether the cell in which the mutation occurs is somatic or germinal. In the former case, this mutation would be associated with no evolutionary change, demonstrating another way in which the relationship is unstable.
Note that one could describe a heap of sand collapsing as a sum of individual grains moving, but that would still make the explanation given of the change refer to all the grains of sand constituting the heap at once. Consequently, if one is interested in the fate of all the grains at once, describing the collapse from the perspective of individual grains would still mean that the effect variable refers to the population of sand grains. Note also that one might be interested in providing a causal explanation about one single sand grain moving in the heap following the addition of another grain to the heap. In this explanation, the other grains of the heap would only be considered as the background of the causal relationship, not as part of the effect variable. In consequence, the resulting explanation would not be an explanation of the heap collapsing, but of a sand grain moving during the collapse. Analogously, one might be interested in a given individual-level event (e.g., the fate of a particular organism) caused by an individual-level variable during an evolutionary process. Although this would be a perfectly valid causal explanation, it would not be an evolutionary one.
Again, I am not claiming that to understand a virus outbreak one need not take into account population-level parameters, just that the level at which the causal relata are described can be different.
Note that even though a given token mutation is an instance of a type of mutation and that some might regard this difference as vindicating two different sorts of causation, I follow Woodward (2003, p. 40) in his view that “a claim such as ‘X is causally relevant to Y’ is a claim to the effect that changing the value of X instantiated in particular, spatio-temporally located individuals will change the value of Y located in particular individuals.” In other words, type-causal claims are generalizations of token-causal claims but both refer to the same hierarchical level.
Frank does say that selection and transmission are “causes,” but he uses this word in a vernacular sense.
We will see in the next section, however, that there is some ambiguity surrounding the notion of “variable” when Otsuka makes his move against the population-level cause view.
Note that Sober (2013) makes a similar point.
Russo (2009) provides a defense of causation as variation, especially in a scientific context, against the view that regularities are enough to establish causation.
I believe that the distinction made by Glennan (2009), to defend the individual-level cause view, between productive and relevant cause is on a similar track.
Recently, Griffiths et al. (2015) have implemented this distinction using a causal variant of mutual information.
Note that strictly speaking this is not the fitness of this individual since it is the outcomes of all factors influencing reproductive output, including those that have nothing to do with natural selection.
As was suggested to me by Jun Otsuka, an alternative way to deal with \(\beta\) being a parameter and not a variable is to think of \(\beta\) as a “constant variable” b with all the probability mass at one value and zero elsewhere (aka Dirac delta distribution). Then the causal graph is \(b \rightarrow w \leftarrow z\), with the structural equation \(w = bz\). This will effectively make the linear parameter a manipulable variable.
More on the significance of this statement in the next section.
Hard interventions are opposed to “soft” interventions, the latter of which do not break incoming arrows in causal graphs but change the probability distribution of the variable intervened upon.
Note also that heritability might be considered as positive for intragenerational changes in line with the point made in Bourrat (2015b).
Note that leaving aside \(\beta\) which is supposed constant and thus only modulates uniformly what each individual contributes, this is almost literally what a variance is. In fact, in words, the variance of a variable is the expected squared difference between an individual value and the mean value in the population (for a formal definition see the main text). This difference is squared in order to obtain the magnitude of this difference, since some deviations will be positive, others negative, and on average will cancel out each other. But rather than the squared difference one might decide to take the standard deviation, which is the square root of the variance, to talk about the average effect of an intervention. Standard deviation is often considered a better measure of variation since it has the same unit as the variable to which it refers.
I thank Peter Takacs for this point.
Note that if y is not perfectly heritable one will have to compute its heritability \(h^2\), which, like \(\beta\), is a parameter of the model. Following our hypotheses, we assumed here that reproduction is perfect, so that (\(h^2=1\)).
This is why the notion of ideal intervention is often associated with that of a miracle. For instance, Woodward (2003, p. 135) considers an ideal intervention to be synonymous with a “localized miracle,” borrowing the notion of miracle from David Lewis. Pearl (2009) talks about “surgical procedure.” He is very clear that ideal interventions can do things that might not be possible physically when he writes “[s]ymbolically, one can surely change one equation without altering others and proceed to define quantities that rest on such ‘atomic’ changes. Whether the quantities defined in this manner correspond to changes that can be physically realized is a totally different question that can only be addressed once we have a formal description of the interventions available to us” (p. 365). For more on the view that ideal interventions need not be physically possible see Woodward (2016).
Woodward (2003, pp. 329-330) distinguishes modular interventions from “level-invariant” interventions. The former refers to invariance of the parameters of different equations in a system of structural equations, while the latter refers to invariance of the parameters of the equation in which the intervention is performed. For my purpose, I will refer to these two types of interventions as “modular.” Note also that modularity is intimately linked to the notions of stability and invariance presented earlier; see Footnote 6.
In general, any fitness-altering interaction between the individuals of a population will satisfy this phenomenon.
Note that this would be a particularly bad model precisely because the two characters are correlated.
References
Abrams M (2009) Fitness “kinematics”: biological function, altruism, and organism–environment development. Biol Philos 24(4):487–504. https://doi.org/10.1007/s10539-009-9153-2
Ayala FJ, Campbell CA (1974) Frequency-dependent selection. Annu Rev Ecol Syst 5:115–138
Birch J (2017) The philosophy of social evolution. Oxford University Press, Oxford
Bouchard F, Rosenberg A (2004) Fitness, probability and the principles of natural selection. Br J Philos Sci 55(4):693–712
Bourrat P (2015a) Distinguishing natural selection from other evolutionary processes in the evolution of altruism. Biol Theory 10(4):311–321
Bourrat P (2015b) How to read ‘heritability’ in the recipe approach to natural selection. Br J Philos Sci 66(4):883–903
Bourrat P (2016) Generalizing contextual analysis. Acta Biotheor 64(2):197–217
Bourrat P (2017) Explaining drift from a deterministic setting. Biol Theory 12(1):27–38
Bourrat P (2018) Natural selection and drift as individual-level causes of evolution. Acta Biotheor. https://doi.org/10.1007/s10441-018-9331-1
Bourrat P (2019a) In what sense can there be evolution by natural selection without perfect inheritance? Int Stud Philos Sci 32(1):39–77
Bourrat P (2019b) Natural selection and the reference grain problem. Stud Hist Philos Sci A. https://doi.org/10.1016/j.shpsa.2019.03.003
Bourrat P (forthcoming) Causation and SNP heritability. Philos Sci
Coulson T, Benton T, Lundberg P, Dall S, Kendall B, Gaillard JM (2006) Estimating individual contributions to population growth: evolutionary fitness in ecological time. Proc R Soc B 273(1586):547–555. https://doi.org/10.1098/rspb.2005.3357
Dieckmann U, Ferrière R (2004) Adaptive dynamics and evolving biodiversity. In: Ferrière R, Dieckmann U, Couvet D (eds) Evolutionary conservation biology. Cambridge University Press, Cambridge, pp 188–224. https://doi.org/10.1017/CBO9780511542022.015
Fisher RA (1930) The genetical theory of natural selection: a complete variorum edition. Oxford University Press, Oxford
Frank SA (1998) Foundations of social evolution. Princeton University Press, Princeton
Frank SA (2014) The inductive theory of natural selection: summary and synthesis. arXiv:14121285 [physics, q-bio] 1412.1285
Glennan S (2009) Productivity, relevance and natural selection. Biol Philos 24(3):325–339. https://doi.org/10.1007/s10539-008-9137-7
Godfrey-Smith P (2009) Darwinian populations and natural selection. Oxford University Press, Oxford
Griffiths PE, Pocheville A, Calcott B, Stotz K, Kim H, Knight R (2015) Measuring causal specificity. Philos Sci 82(4):529–555. https://doi.org/10.1086/682914
Hitchcock C, Velasco J (2014) Evolutionary and Newtonian forces. Ergo 1(2):39–77
Huneman P (2012) Natural selection: a case for the counterfactual approach. Erkenntnis 76:171–194
Illari P, Russo F (2014) Causality: philosophical theory meets scientific practice. Oxford University Press, Oxford
Krimbas CB (2004) On fitness. Biol Philos 19:185–203
Lande R, Arnold SJ (1983) The measurement of selection on correlated characters. Evolution 37:1210–1226
Lewontin RC (1974) Annotation: the analysis of variance and the analysis of causes. Am J Hum Genet 26:400
Luque VJ (2017) One equation to rule them all: a philosophical analysis of the Price equation. Biol Philo 32(1):97–125. https://doi.org/10.1007/s10539-016-9538-y
Lynch KE, Bourrat P (2017) Interpreting heritability causally. Philos Sci 84(1):14–34
Matthen M, Ariew A (2002) Two ways of thinking about fitness and natural selection. J Philos 99:55–83
Matthen M, Ariew A (2009) Selection and causation. Philos Sci 76:201–224
McLoone B (2018) Why a convincing argument for causalism cannot entirely eschew population-level properties: discussion of Otsuka. Biol Philos 33(1–2):11. https://doi.org/10.1007/s10539-018-9620-8
Millstein RL (2006) Natural selection as a population-level causal process. Br J Philos Sci 57(4):627–653. https://doi.org/10.1093/bjps/axl025
Millstein RL (2013) Natural selection and causal productivity. In: Chao HK, Chen ST, Millstein RL (eds) Mechanism and causality in biology and economics. Springer, Dordrecht, pp 147–163
Nowak MA, Tarnita CE, Wilson EO (2010) The evolution of eusociality. Nature 466:1057–1062
Okasha S (2006) Evolution and the levels of selection. Oxford University Press, Oxford
Okasha S (2016) The relation between kin and multilevel selection: an approach using causal graphs. Br J Philos Sci 67(2):435–470
Otsuka J (2016a) Causal foundations of evolutionary genetics. Br J Philos Sci 67(1):247–269. https://doi.org/10.1093/bjps/axu039
Otsuka J (2016b) A critical review of the statisticalist debate. Biol Philos 31(4):459–482. https://doi.org/10.1007/s10539-016-9528-0
Pearl J (2009) Causality: models, reasoning, and inference, 2nd edn. Cambridge University Press, New York
Pence CH, Ramsey G (2013) A new foundation for the propensity interpretation of fitness. Br J Philos Sci 64(4):851–881. https://doi.org/10.1093/bjps/axs037
Pocheville A, Griffiths PE, Stotz K (2017) Comparing causes – an information-theoretic approach to specificity, proportionality and stability. In: Leitgeb H, Niiniluoto I, Sober E, Seppälä P (eds) Proceedings of the 15th Congress of Logic, Methodology and Philosophy of Science, College Publications, London, pp 260–275
Price GR (1970) Selection and covariance. Nature 227:520–21
Ramsey G (2006) Block fitness. Stud Hist Philos Sci C 37:484–498
Ramsey G (2016) The causal structure of evolutionary theory. Australas J Philos 94(3):421–434. https://doi.org/10.1080/00048402.2015.1111398
Reisman K, Forber P (2005) Manipulation and the causes of evolution. Philos Sci 72(5):1113–1123. https://doi.org/10.1086/508120
Rice SH (2004) Evolutionary theory: mathematical and conceptual foundations. Sinauer, Sunderland
Rosenberg A, Bouchard F (2005) Matthen and Ariew’s obituary for fitness: reports of its death have been greatly exaggerated. Biol Philos 20:343–353
Russo F (2009) Causality and causal modelling in the social sciences: measuring variations. Methodos Series. Springer, Dordrecht
Sarkar S (2008) A note on frequency dependence and the levels/units of selection. Biol Philos 23(2):217–228. https://doi.org/10.1007/s10539-007-9092-8
Sesardic N (2005) Making sense of heritability. Cambridge University Press, Cambridge
Shapiro L, Sober E (2007) Epiphenomenalism—the do’s and the don’ts. In: Wolters G, Machamer P (eds) Thinking about causes: from Greek philosophy to modern physics. University of Pittsburgh Press, Pittsburgh, pp 235–264
Snijders TAB, Bosker RJ (1999) Multilevel analysis: an introduction to basic and advanced multilevel modeling. SAGE, London
Sober E (2013) Trait fitness is not a propensity, but fitness variation is. Stud Hist Philos Sci C 44(3):336–341. https://doi.org/10.1016/j.shpsc.2013.03.002
Sober E, Wilson DS (1998) Unto others: the evolution and psychology of unselfish behavior. Harvard University Press, Cambridge
Spirtes P, Glymour CN, Scheines R (2000) Causation, prediction, and search, vol 81. MIT press, Cambridge
Steel D (2006) Methodological individualism, explanation, and invariance. Philos Soc Sci 36(4):440–463. https://doi.org/10.1177/0048393106293455
Steel D (2008) Across the boundaries: extrapolation in biology and social science. Environmental ethics and science policy series. Oxford University Press, Oxford
Stephens C (2004) Selection, drift, and the “forces” of evolution. Philos Sci 71(4):550–570. https://doi.org/10.1086/423751
Walsh DM (2000) Chasing shadows: natural selection and adaptation. Stud Hist Philos Sci C 31(1):135–153
Walsh DM (2010) Not a sure thing: fitness, probability, and causation. Philos Sci 77(2):147–171
Walsh DM, Lewens T, Ariew A (2002) The trials of life: natural selection and random drift. Philos Sci 69(3):429–446
Walsh DM, Ariew A, Matthen M (2017) Four Pillars of Statisticalism. Philosophy and Theory in Biology 9(20170609), https://doi.org/10.3998/ptb.6959004.0009.001
Waters CK (2007) Causes that make a difference. J Philos 104(11):551–579
Woodward J (2003) Making things happen: a theory of causal explanation. Oxford University Press, New York
Woodward J (2010) Causation in biology: stability, specificity, and the choice of levels of explanation. Biol Philos 25(3):287–318
Woodward J (2016) Causation and manipulability. In: Zalta EN (ed) The Stanford encyclopedia of philosophy, winter 2016 edn. https://plato.stanford.edu/archives/win2016/entries/causation-mani/
Acknowledgements
I am thankful to Mathieu Charbonneau, Paul Griffiths, Jun Otsuka, Peter Takacs, and two anonymous reviewers for comments on previous versions of the manuscript. I also thank the members of the Theory and Method in Biosciences group at the University of Sydney and in particular Stefan Gawronski who proofread the final manuscript. This research was supported by a Macquarie University Research Fellowship and a Large Grant from the John Templeton Foundation (Grant ID 60811).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Bourrat, P. Evolution is About Populations, But Its Causes are About Individuals. Biol Theory 14, 254–266 (2019). https://doi.org/10.1007/s13752-019-00329-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13752-019-00329-3