Hostname: page-component-7b7567c8c-tnzcf Total loading time: 0 Render date: 2026-11-14T09:56:19.835Z Has data issue: false hasContentIssue false
  • English
  • Français

Finding Normality in Abnormality: On the Ascription of Normal Functions to Cancer

Published online by Cambridge University Press:  16 February 2023

Seth GoldwasserOpen the ORCID record for Seth Goldwasser[Opens in a new window]
Seth Goldwasser*
Affiliation:
University of Pittsburgh, Pittsburgh, PA, USACenter for the Neural Basis of Cognition, University of Pittsburgh and Carnegie Mellon University, Pittsburgh, PA, USA
*
Email: SEG111@pitt.edu
Rights & Permissions [Opens in a new window]

Abstract

Cancer biologists ascribe normal functions to parts of cancer. Normal functions are activities that parts of systems are in some minimal sense supposed to perform. Cancer biologists’ finding normality within the abnormality of cancer poses difficulties for two main approaches to normal function. One approach claims that normal functions are activities for which parts are selected. However, some parts of cancers that have normal functions aren’t selected to perform them. The other approach claims that normal functions are part-activities that (typically) contribute to the survival or reproduction of the relevant system. However, cancers are too heterogeneous to establish what (typically) contributes to their progression across a type.

Information

Type
Contributed Paper
Information
Philosophy of Science , Volume 90 , Issue 5 , December 2023, pp. 1214 - 1223
Copyright
© The Author(s), 2023. Published by Cambridge University Press on behalf of the Philosophy of Science Association

1. Introduction

Cancer biology features the ascription of normal functions to parts of cancers. Normal functions are activities that parts of biological systems are in some minimal sense supposed to perform. In this article, I argue that the ascription of normal functions to parts of cancers—finding normality within the abnormality of aberration—poses difficulties for the two main approaches to normal function in the philosophy of biology. One approach claims that normal functions are activities that the function-bearing part of a system is selected to perform. I identify these with selected effects accounts. The problem with selected effects accounts is that at least some parts of cancers that have normal functions aren’t selected to perform those functions. The other approach claims that normal functions are activities by which the function-bearing part (typically) contributes to the survival or reproduction of the system. Following Garson (Reference Garson2016), I call these “fitness-contribution accounts.” The problem with fitness-contribution accounts is that cancers aren’t uniform in the way required to establish what (typically) contributes to their progression. The failure of both approaches leaves open a gap in the philosophical literature on function. Though I don’t attempt to close the gap in this article, I argue that the ascription of normal functions to parts of cancers is legitimate. Ascribing normal functions to cancers provides therapeutic benefits by allowing cancer biologists to identify standard activities, dispositions, and structure of parts of cancers that can be undermined by clinical intervention. I claim that one countervailing intuition, namely, that the ascription of normal functions should be reserved for the activities of healthy tissue, is the result of conflating what’s normal with what’s healthy.

Section 2 briefly discusses different types of function and introduces function pluralism. Section 3 presents a representative example of cancer biologists ascribing a normal function to a part of a cancer. Section 4 considers two philosophical approaches to normal function and raises difficulties for them in explicating the ascription of normal functions to parts of cancers. Section 5 concludes by considering an objection to the claim that cancer biologists ascribe normal functions to parts of cancers.

2. Normal function and function pluralism

Function ascription is pervasive in biology. To a first approximation, functions are activities that parts Footnote 1 of biological systems perform and that, by being performed, contribute in some way to those systems (for a similar formulation that generalizes Cummins’s causal role account, see Weber Reference Weber2017). Consider the philosopher’s favorite example: The function of the heart is to pump blood. This ascription tells us, first, that hearts pump blood and, second, that pumping blood contributes in some way to biological systems with hearts, for instance, by transporting nutrients and waste to and from various tissues in the body. Ascribing a function explains by drawing our attention to the dispositions and structural features of systems that are (supposed to be) causally relevant for system-level phenomena of interest. Biologists are keen to understand how and why biological systems persist and propagate. Functions are indicative of how those systems do so and, often, why they have those dispositions and features which, in the good case at least, allow them to do so.

Different function concepts are applied within and across different subdisciplines of biology. Take cladistic systematics, the branch of biology that studies common descent and changes in phenotype as a function of descent. When studying a phenotypic trait, systematists ascribe a function to the trait either to mark continuity in the activity performed by that trait with that of traits in ancestral systems or as evidence of innovation in that trait or its activity (Griffiths Reference Griffiths2006). For instance, a systematist might ascribe to the tail of Crocodilus the function of propelling the animal through its aquatic habitat in recognition of the fact that its ancestor, Mystriosuchus, made the same use of its archosaur tail (Griffiths Reference Griffiths1994, 218–19). Or the systematist might ascribe to the carapace of Proganochelys (a genus of proto-turtle) the function of protection in recognition of its novelty as a trait. In this case, functions are activities that traits perform. Their ascription doesn’t necessarily tell us what a trait is supposed to do, only what it does or did or the causal role it plays or played.

By contrast, take physiology, the branch of biology which is said to study the normal functions of parts of organisms (Roux Reference Roux2014, 2245). When physiologists say of the heart that its function is to pump blood, they do so in full awareness that not all hearts pump blood. In this case, the functions referred to as normal are normative in the minimal sense that they embody a standard for trait-activity (Roux Reference Roux2014, 2248; Garson Reference Garson2016, 5–6, 36, 48). Their ascription indicates not what a trait does but what it’s supposed to do and, in turn, what it’s supposed to be disposed to do and the features it’s supposed to have so that it can perform its function. Standards for trait-activity, disposition, and structure guide identification of instances of traits as being of the same type despite variation between individuals, system types, and environments. A deformed or diseased heart that cannot pump is still recognized as an instance of the type “heart” at least in part by appeal to its function to pump blood (cf. Amundson and Lauder Reference Amundson and Lauder1994). Ditto for radically morphologically distinct hearts across species and environments.

I don’t take these two function concepts to exhaust those applied in biology. Rather, following function pluralism, the takeaway is that different types of function are ascribed in different subdisciplines biology for different explanatory purposes. However, I’ll be concerned primarily with normal functions throughout the rest of the article, especially as ascribed to parts of cancers by cancer biologists.

3. The normal functions of cancer

3.1. Cancer biology: Finding exploitable sameness in aberrant variation

A consistent challenge for those working in cancer biology is dealing with treatment relevant variation among cancers. Here’s a nonexhaustive list of dimensions of treatment relevant differences that individual cancers can exhibit: anatomical site and tissue type of origin, genome, mutation rate, growth rate, tumor formation, incidence and rate of metastasis, cancer microenvironment, and initiation, for example, environmental carcinogens and pathogens. Like inquiry in any domain, a central task in cancer biology is finding within all of this variation sameness that’s of causal and explanatory relevance. For instance, cancers have historically been classified by anatomical site, tissue type, stage, and grade (Plutynski Reference Plutynski2018, ch.1 and Appendix). A stage I, grade 1 lung adenocarcinoma is a cancer originating in glands (tissue) in the lung (site) that has yet to form a tumor (stage) and whose cells still resemble healthy, somatic cells (grade).

The standard classificatory scheme is effective at grouping cancers together and bears explanatory fruit. For instance, other properties relevant for treatment often cluster around tissue type, stage, and grade. Only some types of tissue form solid tumors, that is, clumps of cancer-associated cells. Size is a property of solid tumors that’s predictive of disease progression. And the degree of apparent similarity between cancer cells and healthy cells is predictive of growth rate and metastatic potential—grade 4 cancers with cells very unlike their healthy kin are likely to grow and metastasize more quickly and aggressively.

However, the standard classificatory scheme isn’t perfect (Plutynski Reference Plutynski2018, Reference Plutynski, Massimi and McCoy2019). For instance, cancers originating in the same organ can be genetically more similar to those originating in a different organ than to each other. Precision oncology depends on targeting specific mutated genes and proteins. So, sameness in anatomical site of origin isn’t always explanatory or helpful for guiding treatment. Luckily, this scheme represents only one of many tools for finding treatment relevant sameness among cancers. Another tool that cancer research shares with several subdisciplines of biology is normal function. Or so I now argue by example.

3.2. Case study: The normal function of melanoma-derived sEV

A widely cited paper, Peinado et al. (Reference Peinado, Maša Alečković, Irina Matei, Moreno-Bueno and Hergueta-Redondo2012), claim to have “explored the function of melanoma-derived [small extracellular vesicles (sEV)] in the formation of primary tumors and metastases” (883). And Zhang and Yu (Reference Zhang and Yu2019), reporting their results, say “[Peinado et al. (Reference Peinado, Maša Alečković, Irina Matei, Moreno-Bueno and Hergueta-Redondo2012)] have advanced our understanding of the novel function of [sEV] in pre-metastatic niches” (458). The (novel) function explored and of which our understanding is advanced is the delivery of a signaling protein to cells in bone marrow using small membrane-bound packages produced mostly by late-stage melanomas (Figure 1).

Figure 1.

(a) Melanoma-derived small extracellular vesicles (sEV) (here labeled “exosomes”) carry mesenchymal-epithelial transition factor (here labeled “MET”) to bone marrow progenitor cells (b) as well as sites of metastasis (here represented by the lungs) (d). The function ascribed to melanoma-derived sEV in Peinado et al. (Reference Peinado, Maša Alečković, Irina Matei, Moreno-Bueno and Hergueta-Redondo2012) is the delivery of Met to bone marrow progenitor cells (b), which mobilizes those cells (c) to inflame sites of metastasis, induce vascular leakiness (here labeled “extravasation”), and promote vascular growth (here labeled “proangiogenic”) altogether facilitating tumor growth and metastasis (d). Adapted from Matsumoto et al. (Reference Matsumoto, Masataka Umitsu, Silva, Roy and Bottaro2017).

Through a series of experiments, Peinado et al. identify an activity that melanoma-derived sEV perform and which results in greater primary tumor growth and more aggressive metastasis. Melanomas produce sEV carrying mesenchymal-epithelial transition factor (Met), an oncoprotein that can trigger several signaling pathways in cells (Organ and Tsao Reference Organ and Tsao2011). Melanoma-derived sEV carrying Met travel through the blood to cells deep in bone marrow that haven’t yet differentiated. Receiving Met sets off a cascade of signaling in those cells that mobilize them to inflame distant organs, causing those organs to exhibit vascular leakiness and produce vascular tissue. The result is premetastatic niche formation, which facilitates greater primary tumor growth and metastasis (Quail and Joyce Reference Quail and Joyce2013; Mashouri et al. Reference Mashouri, Hassan Yousefi, Aref, Ahadi and Alahari2019; Gonzalez et al. Reference Gonzalez, Garrie and Turner2020).

Peinado et al. (Reference Peinado, Maša Alečković, Irina Matei, Moreno-Bueno and Hergueta-Redondo2012) features the ascription of a normal function. Beyond use of the definite article, both Peinado et al.’s and Zhang and Yu’s talk of the (novel) function of melanoma-derived sEV generalizes over them without distinguishing between later stages of melanoma, sEV that successfully deliver Met, sEV that are deformed or fail to carry Met, or melanomas that fail to produce sEV at all. Generalizing over these differences allowed Peinado et al. to effectively type-individuate melanoma-derived sEV, affording them the ability to identify a standard applicable to melanoma-derived sEV activity in relation to its contribution to premetastatic niche formation (see also Zebrowska et al. Reference Zebrowska, Widlak, Whiteside and Pietrowska2020).

In the process of identifying the function, Peinado et al. examined sEV production and Met delivery across early- and late-stage melanoma patients as well as low and highly metastatic melanoma mouse models. They also examined sEV production in melanoma mouse models designed to produce sEV lacking Met, fewer sEV, or no sEV. The point was to home in on the mechanism(s) responsible for sEV mediated premetastatic niche formation. This in turn required Peinado et al. to reidentify sEV or mark their absence and to identify and relate in a systematic way the effects of their presence or absence on niche formation. Some of this was accomplished by tracking sEV-related proteins in blood. However, at least some of it was accomplished by hypothesizing the activity melanoma-derived sEV are supposed to perform, positing the dispositions and structure that in the good case (for the cancer) allow them to perform it. Their hypothesis drew Peinado et al. to look for sEV and Met in bone marrow and potential sites of metastasis in patients and mouse models. It also drew Peinado et al. to infer from a lack of sEV, a lack of Met, reduced tumor growth, and reduced metastasis that they had successfully disrupted the functional dispositions and structure of melanoma-derived sEV in mouse models designed to produce sEV lacking Met, fewer sEV, or no sEV.

In confirming their hypothesis, Peinado et al. show that deformed sEV and sEV that don’t carry Met are, in some minimal sense, supposed to deliver Met to bone marrow and are, in some minimal sense, supposed to have the dispositions and structure that allow them to do so. I discuss in what sense sEV are supposed to have these dispositions and that structure in section 5. For now, the generalization over both defective and nondefective sEV in the process of discovering their function and possible clinical interventions suggests that the ascription identifies a standard for part-activity, disposition, and structure. In which case, delivering Met to bone marrow is a normal function of melanoma-derived sEV. It’s normal for melanoma-derived sEV to deliver Met to bone marrow within the abnormality of melanoma progression.

4. Difficulties for current accounts of normal function

Cancer biologists ascribe normal functions to parts of cancers. This poses difficulties for the two main approaches to explicating the ascription of normal function in the philosophy of biology. These approaches roughly divide on whether normal functions are (a) activities whose performance by the function-bearing part results in that part being selected for (say, by natural selection) or (b) activities that confer a benefit to (inclusive) fitness. The first approach is often identified with selected effects accounts (Neander Reference Neander1991; Buller Reference Buller1998; Garson Reference Garson2017b). The second comprises a number of accounts that, following Garson (Reference Garson2016), I group under the heading “fitness-contribution accounts” (Boorse Reference Boorse1977; Bigelow and Pargetter Reference Bigelow and Pargetter1987; Piccinini and Garson Reference Piccinini and Garson2014). I consider each in turn. Footnote 2

4.1. Selected effects accounts

Selected effects accounts claim that an activity of a part is a normal functions iff that part is selected for performing that activity. Thus, the heart has the normal function of pumping blood because hearts are selected for pumping blood, in this case, by natural selection (cf. Garson Reference Garson2017b). However, at least some functional parts of cancers aren’t selected for.

For a trait to be subject to selection, the system in which that trait is present has to meet at least the three following conditions. First, the system has to exhibit heritable variants of the trait or trait-type. Second, the fit between system and environment has to favor some trait(s) over others. Third, the selected trait has to be retained over variants.

Satisfying these conditions is a matter of degree (Godfrey-Smith Reference Godfrey-Smith2009). The more paradigmatically Darwinian a population is—the more thoroughly it satisfies these conditions—the more likely the functional traits are present in that population as a product of selection. The less paradigmatically Darwinian a population is, the more likely at least some functional traits are present in that population as a product of some other evolutionary process. For instance, a system’s exhibiting traits with a low degree of heritability, there being little to no differential fit between environment and system, or there being weak differential retention of a trait may be indicative of convergence, genetic drift, or genetic hitchhiking.

Unfortunately, several cancers fail to meet these conditions except minimally (Germain Reference Germain2012; Germain and Laplane Reference Germain and Laplane2017; cf. Lean and Plutynski Reference Lean and Plutynski2016). At the cellular level, parts of cancers that are ascribed normal functions are often the product of drift or genetic hitchhiking without necessarily being fully co-opted (Germain Reference Germain2012, 806). At the tumor level, these parts are often neither heritable or recapitulated in metastases nor the product of competition between tumors (Germain and Laplane Reference Germain and Laplane2017). In which case, at least some parts of cancers have normal functions despite not being subject to selection (cf. Greaves and Maley Reference Greaves and Maley2012; Plutynski Reference Plutynski, Boniolo and Nathan2017, Reference Plutynski2018; Bozic and Wu Reference Bozic and Wu2020; Hausser and Alon Reference Hausser and Alon2020). But, on selected effects accounts, a necessary condition on a part’s having a normal function is that part’s being selected to perform the relevant activity. The ascription of normal function to parts of cancers therefore poses a difficulty for selected effects accounts that may prove intractable (cf. Garson Reference Garson2017a, 1100; Neander Reference Neander2017, 21).

4.2. Fitness-contribution accounts

Fitness-contribution accounts claim that an activity of a part of a system is a normal function only if performance of that activity by that part increases the (inclusive) fitness of the system. Thus, the heart has the normal function of pumping blood because its doing so increases the (inclusive) fitness of the vertebrate. Unfortunately, bare appeal to contribution to fitness fails to establish a standard for the activities, dispositions, or structural features of parts. Moreover, attempting to establish a standard in a way that fits with fitness-contribution accounts proves difficult by threatening to undermine the point of ascribing normal functions, at least in the case of cancer.

Bare appeal to contributions to fitness cross-cut instances in which performance of an activity by a part is of fortuitous benefit and those in which it contributes to fitness in the normal way. Excluding conferral of fortuitous benefit from counting as constitutive of standard part-activity is taken to be a basic desideratum on accounts of normal function (Wright Reference Wright1973; Garson Reference Garson2016). Barring appeal to the normalizing force of selection to set the standard, a type of normality that’s thought to be scientifically respectable is conditional statistical typicality (Boorse Reference Boorse1977; Piccinini and Garson Reference Piccinini and Garson2014). The normal function of a part then becomes what that part does to contribute to the (inclusive) fitness of an individual system which, conditional on that part’s so contributing, is typical for systems of the type. Thus, the normal function of the heart is pumping blood because, given that the heart does something to contribute to the (inclusive) fitness of some vertebrate, pumping is how it typically contributes to vertebrates.

The move to conditional statistical typicality doesn’t work in the case of cancers. Establishing what’s statistically typical for a type of system requires specifying a reference class for the systems the type comprises. But it’s not clear that reference classes can be specified for cancers without significant overlap. Given their rank variability, no one classificatory scheme has proven effective at unifying cancers into types without overlap or remainder (Plutynski Reference Plutynski2018, Reference Plutynski, Massimi and McCoy2019). This leaves reference classes for cancers severely underdetermined.

Without appeal to the standard classificatory scheme, one could try to stipulate a reference class for a type of cancer ad hoc. But doing so threatens to undermine the normal function of ascribing normal functions, namely, identifying standards of the activities, dispositions, and structural features of parts that allow us, in turn, to identify deviations from those standards. Ad hoc stipulation of a reference class threatens to make those standards arbitrary or otherwise interest-relative. In which case, differences between parts of members of the class might not reflect any genuine deviation, deformity, or failure. And differences that are of genuine casual relevance between members might go unnoticed. The explanatory payoff of finding genuine standards for part-activity among a type of system and figuring out how deviations from those standards impair those systems is constitutive of figuring out how that type of system works. This is the point of ascribing normal functions. We threaten to lose sight of that payoff if we elect to stipulate reference classes ad hoc. If this is right then the ascription of normal function to parts of cancers poses a difficulty for function-contribution accounts that may prove intractable (cf. Hausman Reference Hausman2012, 538–40).

5. The objection from loose talk

There’s at least one objection to the claim that cancer biologists ascribe normal functions to parts of cancers. The objection amounts to this: There’s no reason to think that the ascription of “the (novel) function” to a part of some cancer identifies a standard for its activity, dispositions, or structural features. That these phrases show up in cancer biology is likely loose talk. These functions are no more than the actual or typical causes of disease progression, for example, metastasis. Talk of function signals only that these activities are functions of the kind ascribed in, for example, cladistic systematics. Moreover, cancer biologists appear to reserve talk of “normal function” for the activities, dispositions, and structure of healthy tissue when contrasting them with aberrant tissue.

In response, there’s at least one reason to try to identify standards for part-activity, dispositions, and structural features among types of cancers. Namely, doing so successfully provides viable avenues for treatment. The ultimate aim of cancer biology is the development of effective clinical interventions. This aim is problematized by the variability that cancers exhibit (Section 3.1). And making good on this aim is further problematized by cancers effectively exploiting that variability. In particular, cancers that escape immune response and resist treatment are more likely to effectively maintain themselves (at least until the patient’s death). The more thoroughly a cancer establishes itself in its host the more specialized the knowledge required to root it out becomes.

At the same time, the variability that a cancer can exhibit and exploit is constrained by the evolutionary history of the organism, the mutations driving the cancer, the rate at which the cancer mutates, the organism’s environment, and so on (cf. Neander Reference Neander2017, 62–63). These limits on variability guide cancer biologists to commonalities across cancers. Hence the use of a plurality of classificatory schemes. Knowledge of how to undermine particular traits or mechanisms of types of cancer is often gained by identifying the normal functions of their parts. Identifying these activities as standard cuts across instances in which the relevant traits or mechanisms fail to behave as predicted. Idiosyncratic failures might not be a focus of cancer biology because clinicians might not want to rectify them. Nonetheless, in seeking to undermine the disease, cancer biologists identify standards for the activities, dispositions, and structural features of parts of cancers that many cancers do embody. In seeking to undermine the disease, cancer biologists try to figure out how those cancers work. They find what is normal within the abnormal.

I want to stress that giving up on normal functions here threatens to deprive us of an extremely useful tool in the cancer biologist’s toolbox. I also want to grant that cancer biologists might sometimes speak in confused ways, running together normality with health. Even if their ascription is an in principle dispensable part of cancer biology, the case presented by Peinado et al. (Reference Peinado, Maša Alečković, Irina Matei, Moreno-Bueno and Hergueta-Redondo2012) shows us that normal functions serve as an effective guide to at least some commonalities of clinical significance among cancers. The resistance to allowing normal functions to be ascribed to parts of cancers, rests, I suspect, in conflating what’s normal with what’s healthy. This conflation is exacerbated by the fact that cancer biologists will often refer to the activities of healthy variants as “normal.”

To back up this up, I need to disambiguate normality from health. It is good to be in a state of health. And what’s healthy is conducive to being in that state. The heart’s pumping blood (efficiently) is healthy. But health isn’t synonymous with normality where normality merely sets a standard (cf. Boorse Reference Boorse1977). Consider an Olympic sprinter whose leg muscles are quickly atrophying. The state of her leg muscles is unhealthy despite crossing into what’s normal for the species on a number of dimensions, for example, volume or mass, as they wither. By contrast, an Olympic sprinter in their prime will have leg muscles that differ from what’s normal on those dimensions and many besides. Footnote 3 Normality and health can come apart.

To conclude, normality sets a standard and some parts of cancers have standard activities, dispositions, and structural features. There can be normal ways for an abnormal process to unfold. That cancers are pathological and that cancer biologists often talk of normal functioning when they mean healthy functioning doesn’t undermine the legitimacy of ascribing normal functions to their parts. If so, we need an account of normal function that explicates their ascription in cancer biology (see Goldwasser Reference Goldwasser2023).

Footnotes

1 I use “activity” for both processes and continuous states, e.g., the presence of the ventricular septum. I use “part” and “trait” interchangeably to cover system-level and subsystem traits, parts, components, phenotypes, characters, items, and genotypes except where clarity dictates.

2 There’s another approach in the philosophy of biology that aims to give accounts of how normal function ought to be ascribed, most notably Millikan (Reference Millikan1984, Reference Millikan1989). Addressing Millikan’s technical notion of Normality and other, similarly prescriptivist accounts of normal function is beyond the scope of this article. Thanks to Colin Allen for pushing me to clarify this point.

3 It’s partly for this reason that Boorse restricts focus to a negative or minimal conception of health. Olympians are paradigms of positive health—they’re literally extraordinary.

References

Amundson, Ron, and Lauder, George V.. 1994. “Function without Purpose.” Biology and Philosophy 9 (4):443–69.Google Scholar
Bigelow, John, and Pargetter, Robert 1987. “Functions.” Journal of Philosophy 84 (4):181–96.Google Scholar
Boorse, Christopher. 1977. “Health as a Theoretical Concept.” Philosophy of Science 44 (4):542–73.Google Scholar
Bozic, Ivana, and Wu, Catherine J.. 2020. “Delineating the Evolutionary Dynamics of Cancer From Theory to Reality.” Nature Cancer 1 (6):580–88.Google Scholar
Buller, David J. 1998. “Etiological Theories of Function: A Geographical Survey.” Biology and Philosophy 13 (4):505–27.10.1023/A:1006559512367CrossRefGoogle Scholar
Garson, Justin. 2016. A Critical Overview of Biological Functions. Dordrecht: Springer.Google Scholar
Garson, Justin. 2017a. “Against Organizational Functions.” Philosophy of Science 84 (5):1093–1103.10.1086/694009CrossRefGoogle Scholar
Garson, Justin. 2017b. “A Generalized Selected Effects Theory of Function.” Philosophy of Science 84 (3):523–43.Google Scholar
Germain, Pierre-Luc. 2012. “Cancer Cells and Adaptive Explanations.” Biology and Philosophy 27 (6):785–810.Google Scholar
Germain, Pierre-Luc, and Laplane, Lucie 2017. “Metastasis as Supra-Cellular Selection? A Reply to Lean and Plutynski.” Biology and Philosophy 32 (2):281–87.10.1007/s10539-016-9555-xCrossRefGoogle Scholar
Godfrey-Smith, Peter. 2009. Darwinian Populations and Natural Selection. Oxford: Oxford University Press.Google Scholar
Goldwasser, Seth. 2023. “Standard Aberration: Cancer Biology and the Modeling Account of Normal Function.” Biology and Philosophy 38 (4):1–33.10.1007/s10539-023-09894-1CrossRefGoogle Scholar
Gonzalez, Laura L., Garrie, Karin, and Turner, Mark D.. 2020. “Role of S100 Proteins in Health and Disease.” Biochimica et Biophysica Acta (BBA): Molecular Cell Research 1867 (6):118677.Google Scholar
Greaves, Mel, and Maley, Carlo C.. 2012. “Clonal Evolution in Cancer.” Nature 481 (7381):306–13.Google Scholar
Griffiths, Paul E. 1994. “Cladistic Classification and Functional Explanation.” Philosophy of Science 61 (2):206–27.Google Scholar
Griffiths, Paul E. 2006. “Function, Homology and Character Individuation.” Philosophy of Science 73 (1):1–25.Google Scholar
Hausman, Daniel M. 2012. “Health, Naturalism, and Functional Efficiency.” Philosophy of Science 79 (4):519–41.Google Scholar
Hausser, Jean, and Alon, Uri. 2020. “Tumour Heterogeneity and the Evolutionary Trade-Offs of Cancer.” Nature Reviews Cancer 20 (4):247–57.10.1038/s41568-020-0241-6CrossRefGoogle ScholarPubMed
Lean, Christopher, and Plutynski, Anya. 2016. “The Evolution of Failure: Explaining Cancer as an Evolutionary Process.” Biology and Philosophy 31 (1):39–57.Google Scholar
Mashouri, Ladan, Hassan Yousefi, Amir R. Aref, Ali M. Ahadi, Fatemah Molaei, and Alahari, Suresh K.. 2019. “Exosomes: Composition, Biogenesis, and Mechanisms in Cancer Metastasis and Drug Resistance.” Molecular Cancer 18(1):75.Google Scholar
Matsumoto, Kunio, Masataka Umitsu, Dinuka M. De Silva, , Roy, Arpita, and Bottaro, Donald P.. 2017. “Hepatocyte Growth Factor/Met in Cancer Progression and Biomarker Discovery.” Cancer Science 108 (3):296–307.Google Scholar
Millikan, Ruth G. 1984. Language, Thought, and Other Biological Categories: New Foundations for Realism. Cambridge, MA: MIT Press.Google Scholar
Millikan, Ruth G. 1989. “In Defense of Proper Functions.” Philosophy of Science 56 (June): 288–302.10.1086/289488CrossRefGoogle Scholar
Neander, Karen. 1991. “Functions as Selected Effects: The Conceptual Analyst’s Defense.” Philosophy of Science 58 (2):168–84.Google Scholar
Neander, Karen. 2017. A Mark of the Mental: A Defence of Informational Teleosemantics. Cambridge, MA: MIT Press.Google Scholar
Organ, Shawna L., and Tsao, Ming-Sound. 2011. “An Overview of the C-Met Signaling Pathway.” Therapeutic Advances in Medical Oncology 3 (1):S7–S19.Google Scholar
Peinado, Héctor, Maša Alečković, Simon Lavotshkin, Irina Matei, Bruno Costa-Silva, Moreno-Bueno, Gema, Hergueta-Redondo, Marta et al. 2012. “Melanoma Exosomes Educate Bone Marrow Progenitor Cells toward a Pro-metastatic Phenotype Through Met.” Nat Med 18 (6):883–91.Google Scholar
Piccinini, Gualtiero and Garson, Justin. 2014. “Functions Must Be Performed at Appropriate Rates in Appropriate Situations.” British Journal for the Philosophy of Science 65 (1):1–20.Google Scholar
Plutynski, Anya. 2017. “Evolutionary Perspectives on Molecular Medicine: Cancer From an Evolutionary Perspective.” In Philosophy of Molecular Medicine: Foundational Issues in Research and Practice, edited by Boniolo, G. and Nathan, M. J., 122–145. New York, NY: Routledge.Google Scholar
Plutynski, Anya. 2018. Explaining Cancer: Finding Order in Disorder. New York: Oxford University Press.Google Scholar
Plutynski, Anya. 2019. “Cancer Modeling: The Advantages and Limitations of Multiple Perspectives.” In Understanding Perspectivism: Scientific Challenges and Methodological Prospects, edited by Massimi, M. and McCoy, C., 160–177. New York, NY: Routledge.10.4324/9781315145198-10CrossRefGoogle Scholar
Quail, Daniela F., and Joyce, Johanna A.. 2013. “Microenvironmental Regulation of Tumor Progression and Metastasis.” Nature Medicine 19 (11):1423–37.Google Scholar
Roux, Etienne. 2014. “The Concept of Function in Modern Physiology.” The Journal of Physiology 592 (11):2245–49.Google Scholar
Weber, Marcel. 2017. “How Objective Are Biological Functions?” Synthese 194 (12):4741–55.Google Scholar
Wright, Larry. 1973. “Functions.” Philosophical Review 82 (2):139–68.Google Scholar
Zebrowska, Aneta, Widlak, Piotr, Whiteside, Theresa, and Pietrowska, Monika. 2020. “Signaling of Tumor-Derived sEV Impacts Melanoma Progression.” International Journal of Molecular Sciences 21 (14):5066.Google Scholar
Zhang, Lin, and Yu, Dihua. 2019. “Exosomes in Cancer Development, Metastasis, and Immunity.” Biochim Biophys Acta Rev Cancer 1871 (2):455–68.Google Scholar
Figure 0

Figure 1.(a) Melanoma-derived small extracellular vesicles (sEV) (here labeled “exosomes”) carry mesenchymal-epithelial transition factor (here labeled “MET”) to bone marrow progenitor cells (b) as well as sites of metastasis (here represented by the lungs) (d). The function ascribed to melanoma-derived sEV in Peinado et al. (2012) is the delivery of Met to bone marrow progenitor cells (b), which mobilizes those cells (c) to inflame sites of metastasis, induce vascular leakiness (here labeled “extravasation”), and promote vascular growth (here labeled “proangiogenic”) altogether facilitating tumor growth and metastasis (d). Adapted from Matsumoto et al. (2017).