Generalized Selected Effects Functions and Ecology

1 Introduction

In a series of publications, Justin Garson has developed and defended a generalized selected effects (GSE) theory of function, which, as its name suggests, consists in an expansion of the “classical” selected effects (SE) theory (e.g., Garson 2016, sec. 3.4; 2017, 2019, 2024). The SE theory states that a biological item’s function is an effect of that item that led it to be favored by natural selection (e.g., Millikan 1989; Neander 1991; Godfrey-Smith 1994; see also Wright 1973). The GSE theory broadens this picture, and states that functions may derive from any selection process operating on a population of the appropriate type, rather than from natural selection exclusively. Relevant selection processes may involve differential reproduction (just like in standard natural selection), but also differential retention or persistence. The GSE theory is among the most important recent contributions to the function literature in the philosophy of biology. Garson’s case for its superiority over the classical SE theory is strong, and, as he remarks, the theory addresses an issue of excessive restrictiveness that the classical SE theory has commonly been thought to raise. Its broadened picture of the selection processes from which biological items may acquire functions makes the GSE theory more easily applicable to biological domains that have seemed to fall outside the scope of function theories of the SE family (henceforth SE theories). In particular, it more straightforwardly applies to neurons and neural elements by means of neural selection, and to behavioral dispositions by means of trial-and-error learning (Garson 2016, 56–58, 92–94; 2017, secs. 3–4; 2018, sec. 5; 2019, chaps. 4–5).

In this paper, our focus will be on one particular domain that has been thought to fall outside the scope of SE theories: ecology. We will consider whether and to what extent an account of ecosystem natural selection in terms of differential persistence like the one advocated by Frédéric Bouchard and others provides vindication for a GSE account of ecological role functions (Bouchard 2008, 2013a, 2013b, 2014; Bourrat 2014, 2023; Doolittle 2014, 2017, 2019; Inkpen and Doolittle 2021; Papale and Doolittle 2024; Lenton et al. 2018, 2021; Boyle and Lenton 2022). In ecology, ecological role functions are often ascribed to organisms, populations, species, or their traits, in relation to the communities or ecosystems in which they are involved (for discussions, see Jax 2005; 2010, sec. 4.1; Nunes-Neto, Do Carmo, and El-Hani 2016). These role functions link the traits of organisms to their contributions to community- and ecosystem-level processes, such as nutrient cycling, energy flow, pollination, seed dispersal, etc. Philosophers who have discussed these functions have usually considered them incompatible with the SE theory, because interpreting them along its lines would commit one to the at best controversial view that communities and ecosystems are, on a general basis, units of natural selection (Maclaurin and Sterelny 2008, 114; Odenbaugh 2010, 250–51; 2019, 172; Nunes-Neto, Moreno, and El-Hani 2014, 124; Lean 2020, 9320). Given that skepticism with regard to community- and ecosystem-level natural selection is partly based on the observation that communities and ecosystems do not reproduce, the GSE theory’s recognition of functions based on differential retention (besides reproduction) makes it look more easily applicable to ecology.¹

Garson does not discuss ecology in detail, but he points to the hope for an application of the GSE theory to ecology that seemingly resides in Bouchard’s account of natural selection in terms of differential persistence (Garson 2019, 150). Bouchard (2008, 2013a, 2013b, 2014) argues that ecosystem evolution by natural selection can take place through the differential persistence of ecosystems and hence does not require ecosystem reproduction (see also Bourrat 2014, 2023; Doolittle 2014, 2017, 2019; Inkpen and Doolittle 2021; Papale and Doolittle 2024; Lenton et al. 2018, 2021; Boyle and Lenton 2022).² This suggests the possibility of grounding GSE ecological role functions in ecosystem differential persistence. Bouchard (2013a) himself defends an account of ecological role functions based on ecosystem differential persistence, though one aligning with Bigelow and Pargetter’s (1987) evolutionary forward-looking theory of function rather than an SE theory. Dussault and Bouchard (2017) further develop this account and label it the persistence-enhancing propensity (PEP) account of ecological role functions. One might ask, however, why could ecosystem differential persistence not also serve as a basis for a GSE account of these functions?³

This paper tackles this question. We will consider whether the GSE theory’s admission of functions based on differential retention (or persistence) makes it more applicable to ecology than the classical SE theory. Being able to encompass ecology would be a significant achievement for the GSE theory. It would reinforce its ability to evade the above-mentioned issue of excessive restrictiveness allegedly raised by the classical SE theory. We will argue, however, that there is little prospect for a GSE account of ecological role functions, even when ecosystems are conceded to undergo natural selection through differential persistence, and that for two reasons. First, it is unclear whether the kind of natural selection that can be shown to operate on ecosystems meets Garson’s conditions for GSE functions. Second, the explanatory purpose for which the concept of role function is used in ecology better aligns with theories of function under which functions explain the higher-level capacities of some systems their bearers contribute to, than with theories under which, like for SE theories, functions explain the presence of their bearers in these systems. The former theories mainly include the causal role (or systemic) theory (e.g., Cummins 1975), the goal-contribution theory (e.g., Boorse 1976), and the evolutionary forward-looking (or contribution-to-fitness) theory (e.g., Bigelow and Pargetter 1987), from which Bouchard and Dussault’s PEP account is derived. We will hence argue that an account that links ecological role function to ecosystem-level selection processes would more promisingly be a PEP than a GSE account.

Our goal is thus to use ecology as a test case for the GSE theory, and to assess whether its inclusion of differential retention (or persistence) among selection processes from which functions may arise enables it to bring ecology back within the scope of SE theories (as Garson argues it does for other biological domains that these theories have been thought to exclude). In this regard, the point of our focus on the GSE and PEP accounts’ comparative advantages and drawbacks is to untangle two key aspects that distinguish SE theories (including the GSE theory) from most other theories of function—(1) their linkage of function to selection, and (2) the explanatory purpose they assign to the concept of function—in order to show that the latter, not the former, is what ultimately restricts SE theories’ applicability to ecology. Since the PEP account shares the former aspect with SE theories, but differs from them precisely regarding the latter aspect, it is especially helpful for untangling those aspects. Hence, although the PEP account will occupy a central place in our discussion, our chief aim in this paper is not to argue for this account (though we indeed hope our discussion will provide some indirect support for it). We will assume, for the sake of the discussion, that there is sufficient provision for linking ecological role functions to ecosystem persistence, and focus on what this implies with regard to the prospects for a GSE account of ecological function.⁴

In section 2, we will examine the GSE theory, Garson’s reasons for supporting it, and its scope of application. In section 3, we will rehearse some key aspects of the use of “function” in ecology and make a first exploration of the applicability of the GSE theory to ecology. In section 4, we will discuss the applicability of GSE functions to ecology in more detail. We will argue that the PEP account is better applicable to ecology than the GSE account, first, because it is unclear whether ecosystem differential persistence as it can plausibly be thought to generally occur meets Garson’s conditions for GSE functions, and second, because the PEP account is more appropriate given the explanatory purpose for which the concept of role function is used in ecology. In section 5, we will link this conclusion to Garson’s (2016, sec. 5.3; 2018; 2019, chap. 9) remarks on function pluralism, and suggest some implications for his overall defense of the GSE theory.

2 The Generalized Selected Effects Theory

The GSE theory broadens the classical SE theory and considers that functions may derive from any selection process operating on a population. It states that a biological item’s function is an effect of that item that has led to its differential reproduction or differential retention in a population of the appropriate type (see Garson 2016, 58; 2017, 534; 2019, 93).⁵ As we remarked earlier, an advantage of the GSE theory over the classical SE theory is its ability to encompass uses of function unaccounted for by SE theories, thus dealing with an issue of excessive restrictiveness allegedly raised by the classical SE theory. Garson shows, for instance, that since relevant selection processes occur in the biological domains studied by behavioral sciences, neurosciences, and immunology, the GSE theory is more straightforwardly applicable than the classical SE theory to uses of function that occur in these disciplines (Garson 2016, 56–58, 92–94; 2017, secs. 3–4; 2018, sec. 5; 2019, chaps. 4–5).

Although this is a significant advantage, its comprehensiveness is not, according to Garson, the prime reason for accepting the GSE theory. The prime reason is parity of reasoning (Garson 2016, 59; 2017, 530–31; 2019, 73, 93–94). In Garson’s view, the classical SE theory’s main appeal lies in its ability to make sense of what are often considered as cardinal features of functions, and the GSE theory also makes sense of these features, but with fewer arbitrary restrictions. These cardinal features are: difference from accident, i.e., functions differ from lucky beneficial effects (e.g., filtering, warming and moistening the air that enters the body, not holding glasses, are the nose’s functions); explanatoriness, i.e., functions explain the presence of their bearers (e.g., pumping blood is not just an effect that hearts happen to produce, but an effect that explains why many types of animals have hearts); and normativity, i.e., function-bearing items may sometimes be unable to perform their function and/or function inadequately (e.g., a stomach that cannot digest food in the appropriate context is dysfunctional) (Garson 2016, 4–6, 35–36; 2017, 526–27; 2019, sec. 1.1, 2.1). The classical SE theory easily makes sense of these three features: filtering, warming and moistening the air, not holding glasses, are the nose’s functions because noses have been selected for the former, pumping blood explains the heart’s presence via natural selection for hearts that pump blood, and a non-digesting stomach is dysfunctional because it cannot produce the effect stomachs were selected for. The GSE theory, Garson points out, also makes sense of these features, but does so with fewer arbitrary restrictions. It recognizes as function-bestowing not just natural selection (narrowly construed), but any selection processes capable of generating functions that have these three features. Hence, according to Garson, the GSE theory should be adopted because it makes sense of the three (alleged) cardinal features of function in a more principled manner than the other influential theory capable of making sense of these features, namely the classical SE theory.⁶

In order to be precise about the scope of application of the GSE theory, Garson places important restrictions on the type of selection processes that can be considered to generate functions. As mentioned above, according to the GSE theory, functions may derive from any selection process operating on a population of the appropriate type. The issue of what a population of the appropriate type is has turned out to be a crucial one for the GSE theory. The restriction of relevant selection processes to ones that occur among members of populations is meant to prevent the theory from being overly inclusive (see Garson 2016, 52–54, 60–61; 2017, 535–39; 2019, 102–8; 2024). It intends to exclude seemingly trivial function ascriptions, such as ones applied to rocks differentially eroding on a beach. Since items like rocks on a beach might be regarded as members of populations if one were to adopt a broad enough notion of population, Garson needs to tie the GSE theory to a sufficiently restrictive notion of population.⁷ Garson has refined his characterization of populations in the course of developing the theory and responding to criticism (Garson 2016, 54–55; 2017, sec. 5; 2019, sec. 6.3), and his most recent take is that a population is “a collection of individuals, of the same type, that impact one another’s persistence, survival, or reproduction prospects by virtue of the fact that, by using some common resources, they impact each other’s ability to use those same resources” (Garson 2024, 964).⁸ Competition, cooperation, and parasitism are examples of interactions by which individuals may impact one another’s access to resources. According to Garson, this characterization of populations differentiates between selection processes that are GSE-function-bestowing and those that are not (e.g., ones that occur between rocks differentially eroding on a beach). These conditions for function-bestowing selection processes will be an important reference point in our below discussion of whether GSE ecological functions can be grounded in ecosystem differential persistence as characterized by Bouchard and others.

3 GSE Ecological Role Functions, a First Exploration

The term “function” is used in various ways in the ecological literature. Jax (2005, 641–42; 2010, 62–65) identifies four uses of the term: (1) function as an ecological interaction; (2) the overall functioning of an ecological system; (3) the functional role of an organism, species, or abiotic item in a community or ecosystem; and (4) ecosystem services. Philosophers discussing ecological function have focused on the third concept. We also take this concept as our focus here—and have referred to it with the phrase “ecological role function.” As Jax (2010, 65) remarks, this concept most closely corresponds to the concept of function on which philosophical theories of function have generally tended to focus, though, of course, these theories were developed in reference to the parts and traits of individual organisms rather than to organisms, species and abiotic items as functional components of ecosystems (for reviews, see Wouters 2005; Lewens 2007; Garson 2016). Using this concept, ecologists describe organisms and species, for instance, as primary producers, others as nitrogen fixers, still others as pollinators, and so on. When they do so, they are somewhat linking Jax’s third concept with his second concept, defining the functional roles that organisms fulfill in ecosystems with respect to ecosystem-level processes constitutive of their overall functioning, e.g., productivity, nutrient cycling, pollination, and so on. The aim is to explain how these processes are reliably achieved, and to predict whether and when changes in the species composition of ecosystems will affect the performance of these processes.

The concept of ecological role function occupies a central position in a research domain called biodiversity and ecosystem (BEF) research (Naeem 2002; Loreau 2010), a domain in which ecologists seek to establish explanatory relations between the amount and type of biodiversity present in ecosystems and their carrying out of ecological processes such as those just mentioned. In this research, ecological role functions underlie a particular type of functional classification of organisms and species: functional effect groups, which gather organisms according to how their traits dispose them to achieve similar contributions to such processes (Jax 2010, 54; see Lavorel and Garnier 2002; Hooper et al. 2002). The idea is that organisms whose traits dispose them to achieve different contributions to ecological processes are complementary to each other, whereas organisms whose traits dispose them to achieve similar contributions may, under some conditions, be substitutable (see Jax 2010, 53). Hence, ecologists use ecological role functions to delineate the contributions that components of ecosystems make to ecological processes and, in so doing, to formulate explanations and predictions that link changes in the species composition of ecosystems to eventual changes in their functioning.

Our discussion here will thus be centered on this concept of ecological role function. Our question will be: does Bouchard and others’ account of ecosystem selection in terms of differential ecosystem persistence provide vindication for a GSE account of the concept of ecological role function? That is, could an account linking ecological role functions to ecosystem persistence be a GSE account instead of (or as well as) a PEP account?

To approach this question, it will be useful to flesh out what PEP and GSE accounts of ecological role function might look like. Dussault and Bouchard’s (2017) PEP account, which, as we mentioned, is derived from Bigelow and Pargetter’s (1987) evolutionary forward-looking theory of function, can be formulated as follows:

[PEP-ECO-F] The ecological item I has the ecological role function F in ecosystem E₁ if and only if I’s presence in E₁ enhances E₁’s propensity to differentially persist over alternative ecosystems E₂, E₃, E₄, etc., which are also suited to maintain themselves for some time under the environmental conditions that prevail in the area in which E₁ is located.⁹

Drawing on Garson’s discussions, a GSE account of ecological role function can be formulated as follows:

[GSE-ECO-F] The ecological item I has the ecological role function F in ecosystem E₁ if and only if I is present in E₁ in part because its past presence in it contributed to E₁’s differential persistence over alternative ecosystems E₂, E₃, E₄, etc., which were also suited to maintain themselves for some time under the environmental conditions that prevail in the area in which E₁ is located.

At first glance, a GSE account would seem to share some of the advantages that Bouchard and Dussault claim for their PEP account over an account derived from the classical SE theory. The GSE account would seem to share the advantage that the PEP account gains from giving up the importance assigned to reproduction by the classical SE theory (Bouchard 2013a, 88–90; Dussault and Bouchard 2017, 1123, 1125). Since both accounts consider differential persistence as sufficient for the generation of functions, both theories seem prima facie equally able to ascribe functions to components of ecosystems even though they do not reproduce.

Another advantage that the GSE account would seem to share with the PEP account concerns the possibility of ascribing functions to components of transient ecosystems. As Bouchard (2013a, 93–94) notes, many ecosystems are like Donald Davidson’s (1987) “Swampmen,” that is, they appear and disappear much more transiently than other biological systems. This, he argues, implies that a nonhistorical account of function (like the PEP account) is more useful with respect to transient cases than a historical account (like one derived from the SE theory). For instance, a (happenstance) landslide initiates new ecosystems that will be subject to selective pressures and may eventually evolve, and likewise, a (one-off) hurricane rearranges marshes and establishes new relationships between their component species (Bouchard 2013a, 93). Along similar lines, Dussault and Bouchard (2017, sec. 3.2) maintain that the PEP account has the advantage of being applicable to “Gleasonian ecosystems” formed through the contingent assemblage of populations whose previous evolutionary trajectories are independent.

A GSE account would, it seems, be equally applicable to such transient ecosystems. Although, like the classical SE theory, the GSE theory is historical, it is not so in a way that excludes short-lived systems. However transient they might be, the ecosystems portrayed by Bouchard and Dussault presumably persist at least for some time—indeed, they otherwise could not achieve differential persistence. Hence, these ecosystems do eventually come to have histories, albeit possibly short ones. Moreover, if their differential persistence partly results from the presence of some types of organisms within them, and if the persisting ecosystems provide these organisms with favorable conditions, then the organisms will be in a position to acquire GSE ecological role functions in the ecosystems. They will acquire such functions whenever their presence in the ecosystems comes to partly result from their contributions the ecosystems’ differential persistence. In other words, provided that some ecosystems, although transient, have propensities to persist long enough for their differential persistence to be relevantly explained through the ascription of PEP functions to their components, then it may seem that, quite likely, they will persist long enough for their persistence-enhancing components to acquire GSE role functions within them.

So it seems, at first glance, that a GSE account of ecological role functions based on ecosystem differential persistence would share some important advantages claimed by Bouchard and Dussault for the PEP account over an account based on the classical SE theory. However, we will see in the next section that the issue is more complicated than it may seem at first glance.

4 Ecosystem Persistence and GSE vs. PEP Ecological Functions

5 Ecology, GSE Functions, and Pluralism

The considerations brought up in the previous section indicate that there are few prospects for a GSE account of ecological functions even when ecosystems are seen as subject to some kind of selection through differential persistence. What is the upshot of this for Garson’s overall defense of the GSE theory? We suggested above that encompassing ecology would be a significant achievement for the GSE theory, because it would reinforce its ability to evade the issue of excessive restrictiveness allegedly raised by the classical SE theory. Should the GSE theory’s inability to encompass ecology be a converse cause for worry for its proponents?

In a sense, no. As we highlighted above (sec. 2), Garson’s main rationale for the GSE theory is that it is as able as the classical SE theory to make sense of what are often considered as cardinal features of functions—difference from accident, explanatoriness, and normativity—and that it does so with fewer arbitrary restrictions. Garson, however, is well aware that this rationale for the GSE theory rests on the premise that functions do in fact have these three cardinal features, and that not all discussants of function accept this. He thus surveys examples showing that the concept of function is frequently used in biology in a way that attributes those features to it (Garson 2016, 37–38; 2019, sec. 1.1). He also acknowledges, however, that not all uses of “function” that occur in biology necessarily involve a concept that has these features. For this reason he, like previous proponents of selected effects theories (e.g., Millikan 1989; Neander 1991; Godfrey-Smith 1993), adopts a pluralist stance on function (Garson 2016, 11, 94; 2017, 528; 2018; 2019, 142).¹⁶ His intent when advocating for the GSE theory is thus not to defend it as the exclusive theory of function that will capture all biological uses of the concept. Rather, he advocates it as the theory that best makes sense of “function” when the concept is used in a way that involves the three above cardinal features. Garson thus recognizes that “function” is sometimes used by biologists in ways that do not involve these three features, and that these uses are to be accounted for by alternative theories, presumably ones that are more in line with alternatives to SE theories for which functions explain the higher-level capacities of systems rather than the presence of their bearers (he focuses on the causal role theory, but the evolutionary forward-looking theory, from which the PEP account is derived, also fits the bill here; see Garson 2016, 33–34, 90; 2017, 528; 2018, 11, 17).¹⁷

Hence, our above discussion should, in a sense, not be a cause for worry for proponents of the GSE theory, since this theory purports to account for uses of function in which an item’s function is explanatory of its presence, whereas what we showed above is that ecology tends to use function in a way that is not explanatory in this way. Ecology just happens to use function in precisely the way that the GSE theory does not purport to capture, and Garson readily recognizes that such uses are to be accounted for by alternative theories.

In another sense, however, we contend that the GSE theory’s omission of ecology should be at least a minor cause for concern for its proponents, especially if the theory is advocated in a function-pluralistic context. The concern ensues from a marginalization issue that, as Garson rightly notes, arises within pluralism (Garson 2016, 90; 2019, 143). As we have seen, Garson’s defense of the GSE theory is grounded in its ability to make sense of the three (alleged) cardinal features of function in a more principled manner than the classical SE theory. This ability, to be sure, demonstrates the superiority of the former over the latter. What it does not show, however, is that SE theories, overall, better capture the use of function in biology than alternative theories for which functions explain the higher-level capacities of systems. The GSE theory could be superior to the classical SE theory and still capture a use of function that remains very localized and marginal within biology. It would then be true, but unimportant. This, we contend, is in essence what critics of SE theories have been arguing when pointing out the excessive restrictiveness of the classical SE theory. The latter theory at best captures an interesting use of function, but one that remains overall marginal within biology. The GSE theory’s ability to bring purportedly non-SE disciplines back within the scope of SE theories tackles this threat of marginalization.

In his discussions of function pluralism, Garson argues that the GSE theory’s inclusion of purportedly non-SE disciplines invalidates a particular form of function pluralism: between-discipline pluralism. He defines between-discipline pluralism as the view that “different theories of function are appropriate to different branches of biology” (Garson 2018, 3, 5; see also 2016, 90; 2019, 142), and contrasts it with within-discipline pluralism, a standpoint that “seeks out and emphasizes the plurality of function concepts within any branch of biology” (Garson 2018, 4; see also 2016, 94; 2019, 151). As he recognizes, between-discipline pluralism need not be “a strict stand” or “hard-and-fast rule,” stating that researchers in a discipline always use function in line with one type of theory or the other. What it implies is that different disciplines “place overwhelming emphasis” on one theory of function or the other, and that there is a “strong correlation” between the use that prevails in a discipline and a particular theory of function (Garson 2018, 5; 2019, 149). According to what we have shown above, ecology appears to be a good candidate for a discipline with respect to which between-discipline pluralism still holds. Ecology places overwhelming emphasis on a use of function that diverges from SE theories, and this remains so even given a shift on the part of proponents of SE theories from the classical SE to the GSE theory. Ecology’s prevailing use seems prone to strongly correlate with a theory of function according to which functions explain the higher-level capacities of systems rather than the presence of their bearers. In our view, it correlates more particularly with the PEP account of function, but our current point about pluralism would still hold if another theory that assigns a similar explanatory purpose to functions, such as the causal role theory (e.g., Morrow 2023), turned out to be the more appropriate one.

Is this in itself sufficient to marginalize the GSE theory? Probably not. However, it indicates a possible cause for concern for the theory’s proponents. If more disciplines, or more biological uses within disciplines, turned out to line up with ecology and its non-SE use of function, then a more serious risk of marginalization of the GSE theory would start to emerge.

And Garson’s line of argumentation for the GSE theory’s applicability to purported non-SE biological domains leaves room for this risk to be more than just hypothetical. Certainly, his demonstration of the existence of relevant selection processes in biological domains with respect to which SE functions have been thought implausible compellingly demonstrates the possibility of applying the GSE theory to these domains. However, this in itself does not show that biologists’ most common use of function with respect to these domains aligns with this theory. It is at best a demonstration that, among biologists’ uses of function with respect to these domains, those that involve the three above-mentioned cardinal features of function (difference from accident, explanatoriness, normativity), accord with that theory. This is compatible with the possibility that some, many, or even most of these uses concur with theories that do not assign those features to function, and that overall, the use of function in biology more consistently aligns with these theories.

6 Conclusion

We argued that there are few prospects for a GSE account of ecological functions, because ecosystem differential persistence as it can plausibly be thought to generally occur among ecosystems is unlikely to meet Garson’s conditions for GSE-function-bestowing processes, and because such an account would be in tension with the explanatory purpose for which the concept of role function is used in ecology. Ecosystem differential persistence may more plausibly be thought of with respect to ecosystems diachronically succeeding one another than with respect to ecosystems synchronically contending for similar resources. Moreover, ecologists use the concept of function to explain ecosystem processes, whereas a GSE account would be more appropriate to the aim of explaining the presence of function-bearing items within ecosystems—as we illustrated with the cases of dormant species, sink populations, and abiotic items. Linking this verdict to Garson’s remarks on function pluralism, we suggested that, although the GSE theory’s omission of ecology is not in itself a cause for concern, it may become one should more biological uses of function turn out to line up with ecology’s non-SE use of the concept.

With regard to the question of which philosophical theory of function is most appropriate for ecology, and, in particular, that of the applicability of SE theories to this field, the above discussion yields an important observation. It indicates that this applicability hinges less on the presence or absence of relevant selection processes at work in ecosystems, than on the suitability or unsuitability of SE theories to the explanatory purpose for which the concept of role function is used in ecology.