2. Theories and values in theory choice

Traditional discussions of scientific theory choice conceptualize the way in which scientists choose theories as a one-way process. In this process, scientists (or the infamous scientific community) actively choose scientific theories or alternative units of theory choice, which are the passive element of this process. This choice is usually assumed to rely on certain values that the chosen theory allegedly maximizes, at least relative to its rivals. Thus, the relationship between epistemic values and scientific theories is traditionally conceived as unidirectional, in the sense that the direction of selection is always from values to theories.

Kuhn’s (Reference Kuhn and Kuhn1977) seminal discussion of epistemic values and theory choice is usually considered a paradigmatic example of the standard picture of theory choice we just described. The original aim of Kuhn’s paper was to answer the many critiques of irrationality that were raised against his characterization of scientific theory choice as depicted in The Structure of Scientific Revolutions (Kuhn Reference Kuhn1962). To further explain his views on scientific rationality, Kuhn stressed that the ineliminability of values in scientific theory choice is compatible with a pragmatic kind of scientific rationality. According to Kuhn, the choices of a scientific theory are always dependent on certain epistemic values (like simplicity, accuracy, empirical adequacy, and the like). This is because, even if scientists could agree on which values to consider, the weighting and the application of each value fundamentally involve a subjective element that cannot be eliminated, thus precluding the possibility of a universal algorithm for theory choice. Discussions of values are thus always present in episodes of scientific theory choice, and these entities crucially guide how a new paradigm is adopted by a scientific community.

Philosophical discussions of scientific theory choice following on from Kuhn upheld the ineliminability of values when scientists choose theories. Several philosophers analyzed in detail the nature and the extent of the influence that values exert on the choice of a theory (e.g., Laudan Reference Laudan1984; McMullin Reference McMullin1983; Earman Reference Earman and Horwich1993; Sankey Reference Sankey1995; Okasha Reference Okasha2011; Morreau Reference Morreau2013; Bradley Reference Bradley2017). A further matter of philosophical discussion concerned whether the ineliminable role of values in science is restricted only to epistemic values. Several authors (e.g., Longino Reference Longino1990, Reference Longino1996; Lacey Reference Lacey1999; Laudan Reference Laudan, Machamer and Wolters2004; Douglas Reference Douglas2013) argued, in fact, that non-epistemic values, such as social, cognitive, and pragmatic values, also crucially influenced many episodes of scientific theory choice in the history of science. Related discussions concerned which notion of scientific rationality (Longino Reference Longino1990; Douglas Reference Douglas2009), progress (Laudan Reference Laudan1978; Shan Reference Shan2020), and objectivity (Daston and Galison Reference Daston and Galison2007) fits with such a value-laden picture of scientific theory choice.

A common feature of extant discussions of values in theory choice is their exclusive focus on how values influence the selection of scientific theories. Few discussions take into consideration, instead, the other direction of this relationship, namely, whether and how theories affect values. This is surprising, given that influential authors like Kuhn (Reference Kuhn1962, Reference Kuhn and Kuhn1977) and Laudan (Reference Laudan1984) emphasized how the epistemic values that guide a given scientist’s practice are often affected by the paradigm or research tradition within which she is working. Indeed, Kuhn (Reference Kuhn and Kuhn1977) explicitly stated that the relationship between values and theories is one of mutual influence. In the last part of his paper, in fact, Kuhn stressed that one fundamental caveat of his picture of scientific theory choice is that epistemic values should be considered, to a certain extent, as historically changing entities. This is because “both the application of these values and, more obviously, the relative weights attached to them have varied markedly with time and also with the field of application” (Kuhn Reference Kuhn and Kuhn1977, 335). In addition, Kuhn suggests that a specific pattern of covariance between theories and values can be observed, whereby changes in the role or weight of values often follow changes in scientific theories:

Kuhn goes one step further and characterizes this covariance in terms of a “feedback loop through which theory change affects the values which led to that change” (Kuhn Reference Kuhn and Kuhn1977, 336). ^{Footnote 2} Kuhn’s model for the covariance between values and theories is then the notion of a feedback loop, i.e., the dynamic by virtue of which the outcome of a certain system or process becomes the input for the same system or process at a later stage. We take Kuhn’s feedback-loop model of scientific theory choice to be the following: first, epistemic values operate a direct selection among theories within a scientific domain; then, the outcome of this selection feeds back into the values, by producing modifications on either their application or their weight.

By virtue of this feedback-loop characterization, Kuhn stresses that the relationship between scientific theories and epistemic values is one of mutual influence. Yet, as we saw, the influence that values exert on theories has been extensively discussed in philosophy of science, whereas the reverse influence, that theories exert on values, has not received much attention. In the remainder of this paper we focus on such hitherto under-discussed influence, and we analyze its role for our overarching perspective on scientific theory choice.

3. Theory choice as niche construction

In this section we further characterize the feedback-loop dynamic between values and theories envisaged by Kuhn. We conceptualize this dynamic as structurally analogous to the one between organisms and the environment described by niche-construction theorists. By virtue of this analogy, we offer a better-rounded picture of scientific theory choice, which characterizes the relationship between values and theories as one of mutual influence.

Before introducing our analogy, one terminological specification is in order. For the purposes of our analysis, we consider scientific theories to be certain sets of theoretical assumptions central to the practice of a given scientific community. Theories, understood in this way, are different from disciplinary/methodological frameworks, which comprise a much broader and more diverse assemblage of commitments related to the epistemic activities, goals, and values of a scientific community. ^{Footnote 3}

Our picture of theory choice is centered around a structural analogy involving the relationship between values and theories in theory choice and the relationship between organisms and the environment in niche construction. ^{Footnote 4} Both relationships are relationships of mutual influence structured around a feedback-loop dynamic, where a first, major selective influence (of values on theories in theory choice and of the environment on organisms in niche construction) is followed by a minor, reverse influence (of theories on values in theory choice and of organisms on the environment in niche construction).

In order to see this structural analogy better, let us briefly introduce the main tenets of niche-construction theory (NCT). In the classic neo-Darwinian framework, the environment is considered as the external factor that exerts selective pressures on organisms which, in response to these pressures, evolve adaptations via natural selection to “fit” the environment. However, organisms can also fit the environment by enacting behaviors that transform the environment. These niche-constructing activities emerge as responses to environmental selective pressures but, by transforming the environment, they contribute to changing those selective pressures. Therefore, while the environment exerts a direct selective pressure on organisms, organisms are responsible for an indirect selective pressure that happens through the medium of the selective environment, i.e., all those environmental factors on which organisms themselves exert their causal influence (Fig. 1). This, in turn, prompts further adaptive responses from the organisms that affect their fitness, thus generating looping effects that are at the basis of the co-creation of the selective environment of certain species, that is, of their environmental niches (cf. Odling-Smee, Laland, and Feldman Reference Odling-Smee, Laland and Feldman2003). Generally, the direct pressure from the environment to organisms is considered to have a major selective role compared to the minor influence exerted by organisms via the medium of the selective environment. However, as a result of this process, members of many species inherit the cumulative environmental changes induced by previous generations, which can echo in macro-evolutionary patterns (e.g., Danchin et al. Reference Danchin, Charmantier, Champagne, Mesoudi, Pujol and Blanchet2011; Erwin Reference Erwin2008).

Figure 1. The feedback-loop dynamics between organisms and environment in niche construction. Arrows indicate the directions of the selective influences, while their size represents the strength of the selective force involved.

Our analogy between theory choice and niche construction can then be further specified as follows. ^{Footnote 5} The mutual influence between scientific theories and epistemic values can be viewed as analogous to the relationship between organisms and the environment as conceptualized by niche-construction theory. Just like the external environment exerts selective pressures on competing organisms in the domain of life, each choice among scientific theories is the result of a selection process that is influenced by methodological and disciplinary commitments that crucially include epistemic values. However, as we have seen, according to NCT, organisms themselves can have an impact on certain parts of the environment, modifying it through their niche-constructing activities. More precisely, even though niche-constructing activities emerge as responses to environmental selective pressures, by transforming the environment they contribute to changing those selective pressures. In this way, we can see how certain “niche-constructing activities” of theories, that is, some of their traits that have emerged in response to selective pressures, can, in turn, modify part of the environment, that is, the epistemic values. Analogously to what occurs between organisms and the environment, in the case of theory choice the methodological or disciplinary framework also exerts a direct selective influence on the theories held by individuals embracing that framework. However, theories can have an indirect effect on those selective pressures by having an impact on certain components of the framework, more specifically, its epistemic values (Fig. 2).

Figure 2. The feedback-loop dynamics between epistemic values and scientific theories.

More specifically, we maintain that the weight and the application of epistemic values in the selection process, i.e., in the choice among theories, is partly determined by the niche-constructing activities of previous generations, viz., by the outcome of previous choices among theories. In a nutshell, the result of previous choices, i.e., a certain theory “winning” or “losing” a certain step of the selection process, influences the selective pressures exerted by a certain epistemic value during a later choice. Obviously, not all previous choices are relevant to determining the selective pressures of a current choice, as this will be the case only at a local level, depending on the history of the epistemic interactions within and among certain fields and disciplines. In this sense, each theory choice context “inherits” the cumulative effect of previous relevant choices that determines the selective environment within which that choice has to be made. Therefore, even though the direct influence of the disciplinary/methodological framework on theories at a specific time is generally more sizable than the influence of theories on the epistemic values belonging to the framework, in the long run the indirect selective pressure of theories can produce substantial effects, leading to the emergence of a niche of theories and values. This feedback-loop dynamic between scientific theories and epistemic values shows the significance of the diachronic dimension of scientific theory choice. Later choices among theories can be influenced by the outcome of earlier relevant choices, in that the weight and the application of the epistemic values contributing to later stages of selection can be affected by those theories having “won” or “lost” previous stages of selection. This effect can lead to the building of a niche co-constructed by a certain theory together with one or more epistemic values.

An additional aspect of the analogy is worth spelling out. ^{Footnote 6} In NCT, organisms are considered as agents, in that they actively contribute to the process of natural selection, because their activities produce environmental modifications that alter the relevant selective pressures. While, of course, it would be quite odd to claim that theories are agents in the same way, since the locus of agency is in an individual scientist or a scientific community, the analogy holds if we consider that theories have an active role in that they provide elements of the worldview within which scientists work. It is in this sense, by producing modifications to the scientists’ worldview which, in turn, provides fundamental commitments of their disciplinary/methodological framework, such as the weight and application of epistemic values, that theories can be considered as active. However, it should be remembered that our analogy is just an analogy and, as such, it leaves room for aspects of disanalogy between biological niches and epistemic niches. ^{Footnote 7}

The idea of a feedback loop between scientific theories and epistemic values defies standard approaches to theory choice, which usually treat this process as a static, unidirectional phenomenon. As we stressed in the last section, in fact, the relationship between scientific theories and epistemic values is traditionally modeled as involving only a direct selective influence of values on theories. Our contention is, instead, that the process of theory choice often also involves an indirect, although weaker, selective influence resulting from the impact of theories on values. Such an indirect selection should not be expected to have significant effects in every case of scientific theory choice, since it is far weaker than the traditionally highlighted selective influence of values on theories. Just like in the biological realm, where not every case of natural selection involves niche-constructing activities, not every episode of scientific theory choice involves the construction of an epistemic niche between theories and values. Nevertheless, as we see in the case study in the next section, this hitherto under-appreciated reverse influence of theories on values arguably plays a crucial role in explaining why in certain cases of scientific theory choice certain (applications of) epistemic values seem more important than others, and why resistance to scientific and methodological change might arise.

4. Case study: The disagreement over the nature of biological variation and evolution (1895–1904)

In this section we describe an example of the niche-construction process between scientific theories and epistemic values by relying on a case study from the history of biology. We focus on the disagreement involving continuous and discontinuous views of variation and evolution by natural selection at the turn of the twentieth century. More specifically, we examine three key episodes related to this disagreement: the 1895 report of the Royal Society Committee, the homotyposis controversy, and the resolution of the controversy over Mendelian inheritance. We show not only that certain epistemic values were relevant to determining which theoretical view would prevail as an outcome of each of these episodes, but also that the outcome of these episodes influenced the values which had motivated the prevalence of a certain theoretical view at an earlier stage. ^{Footnote 8}

The core of the theoretical disagreement under our scrutiny concerned whether natural selection operates in a continuous or discontinuous way. That is, whether the natural selection of small phenotypic variations among individuals of the same species is sufficient for evolution to take place or if, on the contrary, “big jumps,” i.e., unusual or novel traits that sometimes appear in certain individuals of a species, are required. The controversy over the continuous or discontinuous nature of variation and evolution started right after the publication of The Origin of Species in 1859, with Darwin himself holding a continuist position. This disagreement was highly entangled with another scientific controversy over the nature of the inheritance mechanism, whose resolution led to the emergence of classical and population genetics in the 1910s. A central part of the debate over the nature of inheritance has been referred to by several historians and philosophers of science as the “controversy between biometricians and Mendelians” (e.g., Frogatt and Nevin Reference Frogatt and Nevin1971; Mayr Reference Mayr1973; Olby Reference Olby1989a; Provine Reference Provine1971). Our historical reconstruction deals with the period in which supporters of biometrical methods, particularly Karl Pearson and W. R. Weldon, were engaged in sharp disputes with proponents of Mendelism, most notably William Bateson. Although these scientists were deeply involved in the inheritance controversy, we mostly consider their theoretical views concerning the continuous or discontinuous nature of variation and evolution and how these views interacted with the epistemic values belonging to their broader methodological commitments.

Since the nature of the opposition between biometricians and Mendelians has been extensively debated (Kim Reference Kim1994; Mackenzie Reference Mackenzie and Webster1981; Roll-Hansen Reference Roll-Hansen1980), and the adequacy of this dichotomy as a historical framework has recently been questioned (e.g., Ankeny Reference Ankeny2000; Vicedo Reference Vicedo1995; Pence Reference Pence2011; Stoltzfus and Cable Reference Stoltzfus and Cable2014; Shan Reference Shan2021), a few clarifications on our analytic perspective are in order. First, we follow Morrison (Reference Morrison2002) in her characterization of the main divergences between the biometrical and the Mendelian approaches as being methodological in character. Their contrast was, in fact, essentially rooted in the belief that the study of inheritance should be based, respectively, only (Pearson) or primarily (Weldon) on statistical analysis, or on experimental methods (Bateson). Second, it must be pointed out that multiple lines of inquiry were pursued and several competing theories of inheritance coexisted at the turn of the twentieth century, thus presenting us with a theoretical heterogeneity that is hardly reducible to the biometricians–Mendelians dichotomy (e.g., Müller-Wille and Rheinberger Reference Müller-Wille and Rheinberger2012; Shan Reference Shan2020). In fact, the diversity of theoretical and methodological commitments held by the scientists involved in these controversies prevents the identification of a consensus converging into two Kuhn-like competing paradigms, a biometrical and a Mendelian, understood as disciplinary matrices (Müller-Wille Reference Müller-Wille, Michael and Borello2021). Therefore, the historical stretch that we consider is best characterized in terms of a pre-paradigmatic period of genetics, rather than an episode building up to a scientific revolution (Shan Reference Shan2021). For these reasons, we limit our focus to the specific disagreement between continuous and discontinuous evolution (and not to the whole biometrician–Mendelian controversy). This, was, in fact, a relatively well-defined theoretical disagreement, ^{Footnote 9} which pre-dated the peak of the dispute between biometricians and Mendelians and yet was strongly influenced by the methodological stances held by the scientists involved. ^{Footnote 10} Most importantly, the epistemic values attached to these methodological stances are also clearly identifiable, despite the methodological heterogeneity characterizing the controversy over inheritance.

For the purposes of our reconstruction, we consider only Pearson and Weldon as representatives of the biometric methodological standpoint and supporters of the continuity view, and Bateson as a representative of Mendelism and supporter of the discontinuity view. ^{Footnote 11} Eventually, the apparent contrasts between biometrical methods and Mendelism were solved by the synthetic paradigms of classical and population genetics, while the theoretical controversy between continuous and discontinuous views of variation and evolution was found to be a spurious one. Yet, we show that a pivotal element that fueled this specific theoretical disagreement is the feedback-loop dynamic between the continuous view of variation and evolution held by Pearson and Weldon and the values of mathematical generality and breadth of scope central to the biometrical methodology.

Finally, we want to emphasize a further reason for our choice of this case study: as we mentioned above, the controversy between continuous and discontinuous views on evolution belongs to a pre-paradigmatic stage of genetics. This makes it an excellent candidate for our study of the feedback-loop dynamics between values and theories because, as Kuhn (Reference Kuhn and Kuhn1977, 331) stressed, it is exactly at these stages that values have most impact on theory choice and, therefore, the effects of the feedback-loop dynamic between theoretical views and epistemic values are more prominent.

4.3. From the homotyposis controversy to the 1904 meeting of the British Association

In 1900, Hugo de Vries put forward his theory of evolution by mutation, based on a great deal of empirical observation. His theory supported the discontinuous view of variation and evolution, and it became rapidly widespread, corroborating the unfounded association of Mendelism and discontinuous evolution. Weldon was reluctant to accept the evidence of mutations as provided by de Vries and warmly endorsed by Bateson. Therefore, in 1902 Weldon started an attack against Mendel’s laws of inheritance and his neglect of ancestry. Initially, he criticized the limited scope of Mendel’s laws (Weldon Reference Weldon1902a), while a few months later he criticized the empirical adequacy of Mendel’s classification of characters. He provided examples, some drawn from Bateson, showing that the Mendelian categories were inaccurate, and from that he concluded that the law of ancestral heredity was, in fact, working (Weldon Reference Weldon1902b). This criticism delayed the analysis of continuously varying characters in Mendelian terms, as it led experimenters to search only clear-cut characters.

As early as 1902, the mathematician G. Udny Yule had debunked this presumed incompatibility, and showed that Mendelism could account for continuous variation and was compatible with biometry and continuous evolution. However, his synthetic approach went largely unnoticed until 1918, by which time the unified framework of population genetics had emerged. In fact, in 1904, Pearson, when investigating the mathematical consequences of pure gamete theory, that is, that characters are inherited intact, still concluded that the pure gamete theory was “not elastic enough to account for the numerical values of the constants of heredity hitherto observed” (Pearson Reference Pearson1904). Therefore, he rejected Mendelian inheritance and suggested that the Mendelians should develop more general principles to start a new mathematical investigation. However, at the 1904 meeting of the zoology section of the British Association, Weldon and Pearson were not able to convincingly counteract the overwhelming evidence in favor of Mendelian inheritance.

In this third moment of collective evaluation, we can see how the impact of the empirical evidence in favor of Mendelian inheritance trumped any value-driven effect on the choice between continuous and discontinuous views. As such, this evidence dispelled the effect of the feedback-loop dynamic, and the related construction of the epistemic niche, between the continuous view of variation and evolution and the values of generality and scope. Despite the fact that generality still remained firmly a central defining value of biometrical methodology, considerations of scope and generality are pushed aside by the overwhelming empirical evidence in favor of Mendelism, still associated with the discontinuous view. By 1906, when Weldon died, the balance was clearly in favor of Mendelism, and Mendelian inheritance was still mistakenly associated with discontinuous evolution which, until the disagreement was found to be spurious a few years later, became dominant.

Our reconstruction of the controversy between continuous and discontinuous views of evolution has shown the significance of the mutual influence between scientific theories and epistemic values for understanding this historical episode. Specifically, we identified a feedback-loop dynamic between the continuous theory of evolution and the values of generality and scope that significantly affected the outcome of the second moment of collective evaluation (i.e., the homotyposis controversy) of our case study (Fig. 3). More specifically, we saw how this feedback-loop dynamic increased the weight and crystallized the specific application of the values supporting the continuous theory and, therefore, significantly contributed to delaying the acceptance of the Mendelian approach to the study of inheritance.

Figure 3. The feedback-loop dynamics between the continuous theory of evolution and the values of mathematical generality and breadth of scope.

Historically speaking, this reconstruction of the feedback-loop dynamic between the continuous theory of evolution and the values of generality and scope dovetails with Shan’s suggestion that “the Mendelian–Biometrician controversy can be viewed as a case of the resistance to the emergence of the Mendelian approach” (Shan Reference Shan2021, 161). This historical case of scientific disagreement shows the advantages of our picture of theory choice as niche construction. Thanks to our modeling of the relationship between epistemic values and scientific theories as one of mutual influence, we could explain why in this controversy the values of generality and scope had such a major impact on the decision of the community and why, in turn, this led to resistance to scientific change.