Skip to navigation – Site map

HomeIssuesXVI-1SymposiaPragmatism and/on Science and Sci...A Pragmatist Interpretation and D...

Pragmatism and/on Science and Scientism

A Pragmatist Interpretation and Defense of Entity Realism

Maja Sidzińska


This paper offers a pragmatist interpretation of Ian Hacking’s version of entity realism, and shows that such an interpretation enables the view to withstand a number of objections. Specifically, the paper shows Hacking’s rejection of a representationalist epistemology, which realist critics unjustifiably attribute to him, and shows his endorsement of a Deweyan pragmatist epistemology instead. If the interpretation is correct, the objections (a) that entity manipulation is theory-laden, (b) that the concept of home truths cannot do the work Hacking envisioned, (c) that entity realism is a form of inference to the best explanation (IBE), (d) that some real entities aren’t manipulable, and (e) that some non-real entities are manipulable, lose much of their force. Objections (a), (b), and (c) rest on a metaepistemological misunderstanding, while objections (d) and (e) are addressed with (pragmatist) clarifications. This paper also offers a novel pragmatist argument for entity realism: the argument from technology. The argument from technology maintains that the reality of the entities that constitute our technologies is a precondition for those technologies’ meaningful use. In this case, entity realism remains a more viable perspective on science than is commonly acknowledged.

Top of page

Full text

1. Introduction

1Scientific realism (henceforth, SR) can be generally characterized as consisting of three core commitments: (1) a metaphysical commitment to the existence of a mind-independent world, (2) a semantic commitment to literal interpretations of scientific claims about the world, and (3) an epistemic commitment to the truth or approximate truth of the claims made by scientific theories (Chakravartty 2017). A compelling argument for SR is the no miracles argument (henceforth, NMA), famously articulated by Hilary Putnam as follows: SR is “the only philosophy that doesn’t make the success of science a miracle” (1975: 73). But Thomas Kuhn’s argument in The Structure of Scientific Revolutions (1962) as well as other antirealist arguments (e.g., Laudan 1981), cast serious doubt on the kind of realism suggested by the NMA by providing a great number of historical counterexamples to science’s alleged success: a great number of “truths” claimed by science have turned out later to be false! And allegedly true scientific theories are incommensurable with one another over time.

  • 1 Entity realism is sometimes also known in the literature as experimental realism or instrumental re (...)
  • 2 Various understandings of ER are discussed below.

2In response to such challenges, as well as in response to other antirealist arguments (e.g., van Fraassen 1980), Ian Hacking and Nancy Cartwright developed a middle-ground position known as entity realism1 (Hacking 1982/1998; Hacking 1983; Cartwright 1983). Entity realism (henceforth, ER) accepts the reality of select entities posited by scientific theories, but rejects its literal semantics and its representationalist epistemology, where representationalist epistemology is any epistemology that understands knowledge as correct or true representation of reality. In other words, ER keeps SR’s commitment (1), at least under certain conditions, while rejecting SR’s commitments (2) and (3).2 ER’s criterion for the reality of an entity is the entity’s manipulability. If a scientist is able to use or manipulate an entity to produce or affect other phenomena, then the scientist accepts that entity as being real. Thus, there is a sense in which ER is a distinctly pragmatist position. In response to Kuhnian charges, entity realists may argue that, for instance, since phlogiston or Darwin’s gemmules were not used or manipulated to produce or affect other phenomena, then they were not of the sort that should have been accepted as real.

3But ER faces a number of objections:

a) that entity manipulation is theory-laden, and therefore ER cannot in principle pull apart SR’s commitment (1) from commitments (2) or (3) (Gross 1990; Elsamahi 1994; Resnik 1994/1998; Musgrave 1996; Psillos 1999; Massimi 2004; Chakravartty 2007);

b) that Hacking’s concept of “home truths,” which is supposed to stave off objection (a), cannot do the work because home truths, which are rather basic and general background truths, are not in principle distinguishable from theoretical truths (Morrison 1990; Psillos 1999), and thus ER is implicitly committed to SR’s commitments (2) and (3);

c) that ER is a form of inference to the best explanation (henceforth, IBE), with IBE’s attendant problems (Resnik 1994/1998; Reiner & Pierson 1995; Sankey 2012);

d) that some real entities aren’t manipulable (Morrison 1990; Shapere 1993); and

e) that some non-real, i.e., theoretical, entities are nevertheless manipulable (Morrison 1990; Elsamahi 1994; Gelfert 2003; Chang 2022).

4Can ER meet these objections? Here I argue that it can, given its pragmatist roots. Although a number of recent defenses of ER have focused on Cartwright’s variety of ER, which emphasizes the fact that science often trades in causes and causal explanations (see Clarke 2001; Suárez 2008; Egg 2018; Nanay 2019), I primarily address Hacking’s variety of ER (detailed below), which emphasizes entity manipulability and experimental practice. Drawing on John Dewey’s influence on Hacking, I argue that ER should be understood as a pragmatist philosophy of science. If it is understood this way, then objections (a) through (c) are not problems for ER per se, but rather they reflect metaepistemological disagreements. I cite rebuttals that deal effectively with objection (d) by noting that manipulability is only a sufficient condition for ER. And I develop a pragmatist counterargument to objection (e), which I take to be the most consequential for ER, by suggesting that manipulability, understood broadly, doesn’t necessarily require a great deal of theoretical precision. This argument draws on the work of Peter Galison (1987), Don Ihde (1991), Robert Nola (2002), and Boaz Miller (2016), and it allows me to develop a positive argument for ER: technology use, not just scientific experimentation or practice, exemplifies entity manipulation, and is itself a form of everyday “experimental practice.” Notably, however, while my defense has affinities with Hasok Chang’s pragmatist realism (2022), I understand metaphysics, including ER’s metaphysics, differently. Indeed, Chang notes that he parts company with Hacking on some metaphysical questions (2022: 121ff.). My pragmatist defense of ER proceeds on different grounds than Chang’s pragmatist defense of realism.

5The paper is organized as follows. In section 2, I explain ER, and in section 3, I defend a pragmatist interpretation of ER. In section 4, I survey the main criticisms of ER, while in section 5, I show that understanding ER as a pragmatist thesis renders objections (a) through (c) moot with regard to ER per se; I also note a response in the literature to objection (d). In section 6, I develop a pragmatist argument that addresses objection (e), and in section 7, I make the argument from technology for ER. I briefly conclude in section 8.

2. What Is Entity Realism?

6ER is the view that select entities posited by scientists are real, while scientific theories may not literally describe reality and may not be, strictly speaking, true or even approximately true. ER thus maintains SR’s commitment (1) – the metaphysical commitment to the existence of a mind independent world, while rejecting SR’s commitment (2) – a commitment to literal linguistic interpretation of claims about the world, and rejecting SR’s commitment (3) – an epistemic commitment to the truth or approximate truth of the claims made by scientific theories. The criterion for an entity’s reality is its experimental use or manipulability. As will be discussed shortly, this criterion should be understood as a sufficient but not necessary condition for realism about an entity. If a scientist is able to use or manipulate an entity whose reality is in question to produce effects, then the entity realist accepts that entity as real.

7Admittedly, this is but one understanding of ER. Suárez calls this version of ER metaphysical experimental realism (MER), and defines it as follows: “manipulation is a sufficient condition on reality: x is real if x can be manipulated” (2008: 140). This version reflects Margaret Morrison’s (1990: 1), Robert Nola’s (2002: 5), and Ruey-Lin Chen’s (2023: 279-80) understanding of ER. It is motivated in part by Cartwright’s arguments in The Dappled World (1999), but mainly reflects Hacking’s views. Neither Suárez nor others such as Egg (2014), Eronen (2015, 2017), Nanay (2013, 2019), and Khalili (2023) appear to find MER to be the most defensible version of ER, however. Suárez focuses his argument on defending epistemic experimental realism (EER), which is the view that “Manipulation is a necessary and sufficient condition on causal warrant: Our belief that x exists acquires this special kind of warrant if and only if we believe that we manipulate x” (2008: 141; emphasis mine). EER clearly focuses on the epistemological question of what provides a warrant for belief. Egg (2014), Eronen (2015, 2017), Nanay (2013, 2019), and Khalili (2023) likewise develop and defend ER on largely epistemological grounds. Their work has the virtue of accommodating certain comments made by Hacking to the effect that manipulation provides the best evidence for the reality of an entity, rather than constituting a sufficient condition for its reality (see Hacking 1982/1998: 1153, 1167). However, such claims are offset by Hacking’s other theses and claims, in particular by his point about time order (discussed below). Furthermore, MER better preserves the distinction between ER and SR; a defense of realism about the epistemology of science likely amounts to a defense of scientific realism. Thus, MER also strengthens ER’s ability to withstand counterarguments from the history of science and other antirealist counterarguments, for it better accommodates theory change. Even if it remains unclear whether MER is Hacking’s final and settled position, it is at least a plausible understanding of ER – one which has recently been defended by Chen (2023), which can withstand powerful objections, and which yields a novel argument for ER, as will be shown.

8A discussion of Hacking’s central theses is in order. Hacking claims the entities posited by scientists are tools not for thinking but for doing (1998: 1153). He argues that in the first stages of science entities are hypothetical, but if we learn about their causal powers and use them “to build devices that achieve well-understood effects in other parts of nature,” then the entities are “experimental,” and assume a real status (ibid.). He therefore draws on Cartwright’s view, which “advocates causalism” (1983: 36). In Cartwright’s view, “we manipulate the cause” in experimental practice (ibid.: 97-8). The insight of the causalist is that it’s not the set of “true” (physical) laws that is “making things happen.” Rather, it’s the entity – the electron, for example – qua what obtains that is real and producing effects (ibid.: 38).

9Hacking points out that the realist-antirealist debates are usually adjudicated at the level of theory, explanation, and prediction. But he asserts that the results of debates that trade in such epistemological considerations must be inconclusive, and that only at the level of experimental practice is realism unavoidable (1998: 1153). He continues: “this realism is not about theories and truth. The experimentalist need only be a realist about the entities used as tools” (ibid.). He discusses the experimental use of electrons to manipulate weak neutral currents and asks, “why else are we (non-sceptics) sure of the reality of even macroscopic objects, but because of what we do with them, what we do to them, and what they do to us?” (ibid.: 1157). I will return to the issue of macroscopic objects later. For now, the point is that Hacking sees debates about what there is or about what obtains as occurring in representational epistemological terms that trade in what the scientific truth is (about what obtains) or what we are justified in believing (obtains). Yet we cannot get to the heart of what obtains so long as we conceive of knowledge as correct representation.

10Hacking makes two important points about experimental practice – about time-order and about home truths – which will become important to my argument in later sections:

The argument – it could be called the experimental argument for realism – is not that we infer the reality of electrons from our success. We do not make the instruments and then infer the reality of the electrons, as when we test a hypothesis, and then believe it because it passed the test. That gets the time-order wrong. By now we design apparatus relying on a modest number of home truths about electrons to produce some other phenomenon that we wish to investigate. (Ibid.: 1158)

11With regard to time-order, Hacking notes that in the making of PEGGY II – the polarizing electron gun used for electron scattering experiments – no one understood in advance the polarizing properties of gallium arsenide (GaAs); this was found by chance during an unrelated experiment (ibid.: 1163). Thus, he claims that theory does not always precede experimental practice, and that the relationship between theory and experiment are different in different stages of science, as well as in different cases (Hacking 1983: 154-5).

12He also suggests that experimental observation, which is often theory-laden, isn’t necessarily so, and gives several historical examples of theory-free observation: the double refraction of Iceland Spar, diffraction (as studied by Grimaldi and Hooke), Newton’s observation of the dispersion of light – not explained until a century later (ibid.: 155-6), as well as Faraday’s experiments on the polarization of light – also explained later by a different physicist (ibid.: 211). But, he says, “I make no claim that experimental work could exist independently of theory. That would be the blind work of those whom Bacon mocked as ‘mere empirics.’ It remains the case, however, that much truly fundamental research precedes any relevant theory whatsoever” (ibid.: 158). So, he acknowledges the iterative relationship between experimental practice and theorizing (ibid.: 162). While he criticizes philosophy’s overemphasis on theory, he retains a place for both: “One road to new technology is the elaboration of theory and experiment which is then applied to practical problems. But there is another road, in which the inventions proceed at their own practical pace and theory spins off on the side. The most obvious example is the best one: the steam engine” (ibid.: 163). The history of the development of the steam engine is a story of technological development with practical aims that later gave rise to the theory of thermodynamic efficiency (ibid.: 162-5). The development of the microwave oven presents another, more recent example. Microwave technology was originally pursued for use in radar systems, but Percy Spencer accidentally realized its heating properties by experiencing a chocolate bar melting in his pocket (Bih 2003). Manipulations, in these cases, preceded theory.

13Hacking argues, “The experimentalist does not believe in electrons because, in the words retrieved from mediaeval science by Duhem, they ‘save the phenomena.’ On the contrary, we believe in them because we use them to create new phenomena, such as the phenomenon of parity violation in weak neutral current interactions” (1998: 1164-5). He writes, in 1982, that “even if present quantum electrodynamics turns out to need radical revision, the machine called PEGGY II will still work” (ibid.: 1158). In that case, “engineering, not theorizing, is the proof of scientific realism about entities” (ibid.: 1167).

14The claim about manipulability is rooted in a perspective on science that takes experimentation at least as seriously as theorizing and focuses on experimental practice. It shows that scientists need not have knowledge qua correct representations to proceed in all cases. This is a relatively uncommon stance in analytic philosophy of science, which tends to privilege analyses of theory, leading to its preoccupation with the semantic and epistemic aspects of science – a preoccupation Hacking sought to rebalance (Ihde 1991). In effect then, ER is a perspective on science that adopts a realist metaphysics for select entities combined with a pragmatist epistemology. And yet ER’s metaphysics may itself be read as pragmatist since the reality of an entity is required for its manipulability. It’s to the pragmatist interpretation of ER that I now turn.

3. Entity Realism is a Pragmatist Thesis

3a. Hacking’s Deweyan Epistemology

15Hacking draws deeply on pragmatist philosopher John Dewey (1859-1952) (1983: 130-46), and his pragmatist perspective is implicit in ER’s focus on use, manipulability, and experimental practice. He attacks the focus in philosophy of science on theory, which he takes to bias the outcomes of the realism-antirealism debates. Hacking’s epistemology requires a closer look, since it’s primarily on epistemological grounds that he rejects SR. At the end of his 1982 paper, Hacking discusses the historical shift from antirealism to realism about atoms. He notes that in 1870, it made sense to adopt an antirealist stance toward atoms, since “the minute structure of matter could not be proved then” and could only be inferred (1998: 1166). Such inferences provided explanations and helped make predictions. But they “need never produce conviction in the philosopher inclined to instrumentalism or some other brand of idealism” (ibid.). Hacking argues that the epistemic situation is now quite similar to that of the nineteenth century, when knowledge was thought of as correct representation (ibid.). In that case, however,

[O]ne could never get outside the representations to be sure that they corresponded to the world. Every test of a representation is just another representation. “Nothing is so much like an idea as an idea,” as Bishop Berkeley had it. To attempt to argue for scientific realism at the level of theory, testing, explanation, predictive success, convergence of theories and so forth is to be locked into a world of representations. No wonder that scientific anti-realism is so permanently in the race. It is a variant on “The Spectator Theory of Knowledge.” (Ibid.)

16As Hacking notes, Dewey criticized the spectator theory of knowledge as the mistake “pervasive among Western Philosophers from the time of Plato, of thinking that knowledge involves passively representing the world” (ibid.). Under a representationalist epistemology, knowledge involves producing representations which are to be evaluated based on their correspondence with facts, rather than involving the construction of conceptual frameworks that are “judged by their instrumental value in predicting experience and guiding our actions” (ibid.). Hacking argues that “[b]y attending only to knowledge as representation of nature, we wonder how we can ever escape from representations and hook-up with the world” (1983: 130). Without considering experimental practice, there can be no court of appeal about what obtains internal to a representational epistemology (ibid.). Any such internal criterion would be a criterion of logical coherence only. He writes,

I agree with Dewey. I follow him in rejecting the false dichotomy between acting and thinking from which such idealism arises. Perhaps all the philosophies of science that I have described are part of a larger spectator theory of knowledge. Yet I do not think that the idea of knowledge as representation of the world is in itself the source of that evil. The harm comes from a single-minded obsession with representation and thinking and theory, at the expense of intervention and action and experiment. (Ibid.: 130-1)

17Chang’s critique of contemporary analytic understandings of knowledge bear this out as well in the context of his pragmatist defense of (a kind of) scientific realism (2022: 12-67). Similarly to Hacking, Chang, drawing on William James and Dewey, understands the pragmatist conception of knowledge as rooted in activities (ibid.: 18-20).

  • 3 Notably, Chang, in his defense of a kind of pragmatist realism, disagrees with such a framing. He f (...)

18Hacking’s chapter on pragmatism in Representing and Intervening provides more grounds to interpret ER as a pragmatist view. He notes that Dewey’s criterion for truth was warranted acceptability (1983: 61), and this criterion is distinctly not beholden to ideas of reference or correspondence which pervade representationalist epistemologies. Dewey claimed that “[w]hat measures [knowledge’s] value, its correctness and truth, is the degree of its availability for conducting to a successful issue the activities of living beings” (1978: 180). Dewey furthermore claimed in Logic: The Theory of Inquiry, for instance, that “[t]he farmer won’t get grain unless he plants and tills; the general will win or lose the battle according to the way he conducts it […] Neither the grain nor the tilling, neither the outcome of the battle nor the conduct of it, are ‘mental’ events” (1938: 107). David Hildebrand explains that Dewey replaced Kant’s mind-centered, representational system, which was a synthesis of rationalist and empiricist epistemologies, with one that was based in “experience-nature transactions” (2018). It could be argued that Dewey’s epistemology linked Kant’s phenomena with the noumena, by showing that in our scientific (and other) practices, we sometimes manipulate parts of the noumena – the real entities. Hence, from a pragmatist view, we are warranted in accepting the reality of parts of the noumena.3 In other words, we are warranted in accepting SR’s commitment (a). Experience-nature transactions are precisely what experimental practice entails.

19Truth is an empty concept for Dewey, except insofar as it is a name for “the product of competent inquiries” (1985: 16), where the inquiries themselves are enacted, experimental. PEGGY II’s scattering of the electrons is one such product of competent “inquiry.” Dewey wrote: “In scientific inquiry the criterion of what is taken to be settled, or to be knowledge, is being so settled that it is available as a resource in further inquiry; not being settled in such a way as not to be subject to revision in further inquiry” (ibid.). Again, inquiry here is understood as involving practical activity, rather than mere armchair cogitation or abstract reasoning. Importantly for the purposes of my argument, Hacking claims that his view that “realism is more a matter of intervention in the world, than of representing it in words and thought, surely owes much to Dewey” (Hacking 1983: 62). For Dewey and Hacking there is no method of rationality external to practices (ibid.: 63). Thus, Hacking adopts a pragmatist approach to what is real which centers the practical outcomes of experimentation as revealing parts of the real world.

3b. Hacking’s Metaphysics

20Hacking, however, has more sympathy for metaphysical “fixity” than do many pragmatists, pointing out that theoretical change won’t make PEGGY II stop working. This means that we can expect electrons to continue to serve as material means to further inquiry, barring a change in the physical world. Hacking relates a now-famous conversation with a friend: “Now how does one alter the charge on the niobium ball? ‘Well, at that stage,’ said my friend, ‘we spray it with positrons to increase the charge or with electrons to decrease the charge.’ From that day forth I’ve been a scientific realist. So far as I’m concerned, if you can spray them then they are real” (Hacking 1983: 23). Scientific realism is not usually considered a pragmatist perspective, so how can Hacking’s realism be understood as pragmatist? It can be when the “mark of the real” (Nola 2002: 5) is usefulness or manipulability; then, “real” is whatever we use to affect an end.

21Three things must be said about this criterion of reality. First, this is primarily a metaphysical, not an epistemic, criterion; in other words, the “proof” (of reality) is in the pudding, not in its recipe. Hacking’s ER does not provide a justification for ontological commitment; rather, ER holds that real entities are a precondition for the conduct of manipulations, or as Dewey had it, “inquiries.” Again, this contrasts several recent arguments for ER which aim to offer epistemological criteria for the reality of select entities (discussed in Section 3b). It also contrasts Chen’s (2023) approach which aims to justify certain ontological commitments in service of ER. Second, this criterion of reality produces claims to reality that are not generally “objective” or broad enough to satisfy typical scientific realists since the criterion doesn’t involve SR’s commitments (2) and (3). Some realists may thus object that ER is too relativistic. Third, this criterion of reality seems to apply to many domains squarely outside of the context of physics or material reality more generally from which nearly all of Hacking’s examples are drawn in the works being considered here. For instance, if we can manipulate a group’s language to affect changes in its behavior, then we may be warranted in taking certain “linguistic” entities as real. This is not often appealing to realists. (But not coincidentally, language is what Hacking’s later work focuses on and it addresses such issues.)

  • 4 Nature is itself understood as mediated by various practices from within the pragmatist framework, (...)

22What may comfort realists and support Hacking’s claim to being any kind of realist, is that the pragmatist framework can do justice to the way that physical reality causally impacts behaviors and actions, including concept use. Its focus on experimental practice requires attention to the joints, or lack thereof, of nature4 when we engage in engineering projects – including conceptual ones. It’s precisely in the doing – in the success or failure of engineering of various types – when we implicitly learn the locations of nature’s joints. In the domain of physics, the joints appear less plastic than they do in other domains, so at least in that domain the upshot of Hacking’s view is relative metaphysical fixity – something scientific realists should appreciate. What Hacking suggests is that if philosophers of science want to continue to endorse some form of realism in the face of Kuhnian charges, then “they would be better off seeing that experimentalists become pragmatic realists” by becoming familiar with entities that are the “furniture of a scientific world” (Ihde 1991: 104).

4. Objections to Entity Realism

23As noted in the introduction, there are five major objections to entity realism, as follows.

4a. Objection (a): Entity Use or Manipulation is Theory-laden

24A number of scholars have responded to Hacking by arguing that to use or manipulate entities is already to have a theoretical conception of them. Mohamed Elsamahi argued that the experiments involving PEGGY II were more theory-dependent than Hacking suggested (1994). Resnik wrote that the term use has a non-epistemic (i.e., experimental) sense and an epistemic sense, that “experimentation is, in part, an epistemic activity” and that “one cannot use a theoretical entity in an experiment without claiming to know something about that entity, and claiming to have some justification for believing that it has specific causal properties” (1994/1998: 1179). Alan Musgrave puts the point thus:

To believe in an entity, while believing nothing further about that entity, is to believe nothing. I tell you that I believe in hobgoblins […]. So, you reply, you think there are little people who creep into houses at night to do the housework. Oh no, say I, I do not believe that hobgoblins do that. Actually, I have no beliefs at all about what hobgoblins do or what they are like. I just believe in them. (1996: 20)

25Stathis Psillos argues that “the distinction between being realist about entities and being realist about theories is misconceived” (1999: 256) on the same grounds as Musgrave – we must attribute some properties to the entities we posit, and we do this using theory. Michela Massimi argues that the entities that experimental practice requires are underdetermined by data, so a phenomenon – an experimental effect – may be thought to be caused by one of several empirically equivalent causal entities (2004: 38). Thus, scientists cannot choose between empirically equivalent entities unless they’ve already chosen from between rival scientific theories (ibid.). Anjan Chakravartty thinks that ER asks for too much when it asks us to distinguish (unobservable) entities and their uses since the entities cannot stand in isolation from theory when they are to be used experimentally (2007: 31).

26According to these scholars, ER entails that one must be a realist about theories also, and therefore is forced into SR.

4b. Objection (b): Home Truths are not Distinguishable from Theoretical Truths

27In anticipation of objection (a), Hacking deployed the concept of home truths (1998: 1158). Home truths are not the specific theoretical truths under (a particular) investigation. Rather, they are more general background truths that allows scientists to build experimental equipment without subscribing to a more specific theory of the entity under investigation. Home truths are supposed to explain why physicists build experimental equipment sans agreement about what the “correct” theory (or model) of the electron is. But Margaret Morrison pointed out that it wasn’t clear that some allegedly theory-neutral general background knowledge could be isolated from theory proper in the way Hacking suggested (1990). Morrison also argued that home truths are supposed to spare the entity realist from the Kuhnian problem of the historical incommensurability of theories (ibid.: 3). That is, Hacking’s concept of home truths is supposed to preserve the idea that physicists J.J. Thomson, Lorentz, Bohr, and Millikan, were all talking about the same thing – the electron – despite having different theoretical descriptions of it over time (ibid.). This would vindicate Hacking’s view that all these scientists needed were home truths to experiment with one kind of entity – electrons. But Morrison argued that home truths are generally embedded in larger theories or models, and not in principle prior to them (ibid.: 6-7). Thus, she concludes, home truths cannot do the work that Hacking sets out for them. And if home truths cannot do such work, the entity realist is once again forced into SR.

4c. Objection (c): Entity Realism Relies on Inference to the Best Explanation

28Critics have also interpreted Hacking as employing IBE. From an antirealist perspective, IBE is a circular form of reasoning, since it assumes what it’s supposed to explain – the success, that is – truthfulness, of scientific theories. According to Laudan (1981) and Fine (1984/1998), IBE is illegitimate when what is inferred is unobservable. How can we explain the truth of scientific explanation by inferring that its truth is the best explanation of what we don’t see? Explaining truth by inferring truth is circular. Resnik acknowledges that Hacking doesn’t see ER as an argument in the form of IBE, but he argues that nevertheless it is (1994/1998: 1169). Since ER is supposed to explain the success of experimental science, Resnik claims that it’s analogous to other arguments of this type – for instance, J.J.C. Smart’s (1968) version and Richard Boyd’s (1984) version (1994/1998: 1177). Resnik simply sees the subject of Hacking’s IBE as being the success of experimental science – the reliability of the workings of scientific instruments, rather than theoretical success or methodological success (1994/1998: 1178-9). Richard Reiner and Robert Pierson (1995) and Howard Sankey (2012) concur that Hacking’s argument is a form of IBE. (Notably, however, Morrison (1990) does not think Hacking is employing IBE.)

4d. Objection (d): Some Real Entities aren’t Manipulable

29Morrison worries that there are entities we presume are real but that are nevertheless not manipulable – astrophysical entities are her case in point (1990: 13-4). The fact that we do not manipulate astrophysical entities suggests this throws manipulability as a criterion for the reality of an entity into doubt. We cannot hope to manipulate, for example, distant stars (ibid.: 13). Since manipulability, not visual experience, was Hacking’s criterion in the context of microscopy (see Hacking 1983: 186ff.), Morrison claims it should be mutatis mutandis in the context of astrophysics (1990: 13-4). Her objection is simply that, assuming we should be realists about astrophysical entities, Hacking’s view cannot explain why.

30Dudley Shapere (1993) develops this line of argument in response to Hacking’s “Extragalactic Reality: The Case of Gravitational Lensing” (1989), where Hacking suggests that astronomy is not a science in the robust sense since we cannot experiment on astrophysical phenomena. Hacking argues we should adopt antirealism about the various entities posited by astronomy (1989). Shapere objects to this conclusion by suggesting that that Hacking misconceived the meaning and role of experiment in science (1993: 135). He argues that we are warranted in our realism about gravitational lenses, for instance, because although we cannot manipulate them, they have observable effects on light (ibid.: 136-42). And despite a multiplicity of models in use in astrophysics, part of what has occurred is, essentially, progress: the models are converging on the truth as the science matures (ibid.: 145). Both Morrison and Shapere suggest that the criterion of manipulability is too stringent since it leads to the conclusion that a number of entities we intuitively think are real, are not.

4e. Objection (e): Some Non-real Entities are Manipulable

31Morrison also argues that the investigation of entities such as the alpha particle through the use of ions took place at a time when ions were merely hypothetical entities (1990: 9). She notes that there were no particular theoretical commitments about the nature of ions at the time of Charles Wilson’s cloud chamber experiments which nevertheless depended on ions (ibid.: 8-9). She also points out that charmed quarks were manipulated when they were rather theoretical – hypothetical – entities. As with the ions, the existence of quarks was used to investigate charm even at a time when the scientific jury was out with regard to their nature (ibid.: 12). She concludes that manipulation can occur in contexts where the status of the manipulated entity is unspecified, and thus where the allegedly manipulated entity may be merely theoretical.

32In a similar vein, Elsamahi suggests that contrary to Hacking’s account, the experimental results of the PEGGY II experiments may have been identical even if the electron was merely a theoretical entity (ibid.: 176). Elsamahi claims that the experimental results rely on no particular ontological notion, but rather on mathematized and abstract models (ibid.). He contends that it’s possible in principle to use theoretical entities in experimental practice, and therefore the existence of an entity is not necessary for manipulative success (ibid.: 176-7).

33Gelfert offers a more pointed argument to the same effect: ER may warrant belief in entities that don’t exist (2003: 245). Gelfert claims that the entities ER takes to be real must meet determinate identity conditions before they can be said to be manipulated, and therefore real, on ER’s own criterion (ibid.). He claims that “instances of token entities must be identifiable as causal relata in order to serve as a starting point for any experimental generalization to entity types” on pain of the failure of claims about what is manipulated to refer (ibid.). (Otávio Bueno’s argument to the effect that determining identity remains an empirical matter echoes this worry (2018).)

34Gelfert then provides a number of examples where such failure occurred, one of which is the case of manipulating quasi-particles such as unoccupied electron states (“holes” in the Fermi sphere) in semiconductors (2003: 256-7). Since holes are non-entities, it’s not clear in what sense they can be manipulated, and yet they cause effects (ibid.: 251-60). With regard to quasi-particles then, he concludes, “you can spray them […] but they ain’t real” (ibid.: 257). In response to Hacking’s challenge to the effect that the test for the reality of an entity is whether we can put a spin on it, polarize it, and make it scatter (1983: 274), Gelfert argues that quasi-particles pass the test, for “[i]t is possible to selectively ‘put a spin on’ quasi-particles, as experiments with epitaxial thin-film heterostructures confirm. This gives rise to a quasi-particle current that can be injected – sprayed, if you like – across barriers between films of different materials” (Gelfert 2003: 259).

35Chang likewise argues that the history of science shows that entities such as phlogiston had been manipulated (2022: 150-5). Georg Ernst Stahl apparently did add phlogiston to, and remove phlogiston from, different substances, thereby transforming them (i.e., producing effects). And Benjamin Thompson, Count Rumford had manipulated “frigorific rays” by using mirrors and cones (ibid.: 151). Chang thus challenges manipulability as the criterion of an entity’s reality, given that we do not presently claim that phlogiston and frigorific rays are real.

5. Responses to Objections (a) through (d) to Entity Realism

36In this section, I offer responses to objections (a) through (d).

5a. Combined Response to Objections (a), (b), and (c)

37As I argued above, Hacking’s ER should be understood primarily as a pragmatist thesis. If ER is understood this way, then objections (a) – that entity manipulation is theory-laden, (b) – that home truths are indistinguishable from theoretical truths, and (c) – that ER is a form of IBE, lose their force. That’s because these objections make different metaepistemological assumptions – ones which are not defended by the critics, and which pragmatists don’t share.

38Before responding to this set of objections in earnest, it must be noted that Hacking did not object to the idea that experimental practice and some cases of entity manipulation are theory-laden. He objected to the severe overemphasis on theory to the exclusion of experimental practice in analytic philosophy of science. He sees this overemphasis as leading to an epistemological bias which often reduces metaphysical questions to epistemological ones. By focusing exclusively on issues that regard knowledge, including how we know certain entities are or aren’t being manipulated or causing effects, we miss a great deal about how science is actually conducted. Hacking’s examples show that there are cases of theory-free entity manipulation, and scholars must be able to account for these cases. Furthermore, the scholars launching the set of objections (a) through (c) ignore the examples of theory-free entity manipulation that Hacking raises (an additional example is offered by Chen 2023). Instead, they provide historical examples of theory-laden experimental practice. But this only shows that entity manipulation sometimes is and sometimes is not theory-laden, which is compatible with Hacking’s claims. (How often experimental practice is theory-laden is an open empirical question.) Furthermore, the upshot of Hacking’s view is that we should remain antirealist about entities until the entities are manipulated, and critics neither consider the possibility of taking an antirealist stance toward theoretical entities until such a time nor provide any responses to Kuhnian challenges from theory incommensurability. Thus, while their claims are not incorrect per se, they offer no genuine alternatives to ER that account for the diversity of experimental practices in science.

39These points by themselves don’t show that objections (a) through (c) amount to metaepistemological disagreements between Hacking and this set of critics, so it’s time to make that argument. Egg points out that Hacking sees ER as metaphysical thesis, bracketing off epistemological questions having to do with what we know or what we should believe (2018: 121, citing Hacking 1983: 1-2). Hacking understands his own argument from manipulability to be about what obtains rather than being about claims about what obtains, for when we manipulate electrons we manipulate parts of the world, rather than manipulating claims about parts of the world. Theory may be interwoven with our manipulations, but theory is not the subject of manipulation in experiments. And yet, Egg notices that Hacking also writes: “I run knowledge and reality together because the whole issue [about what’s real] would be idle if we did not now have some entities that some of us think really do exist” (2018: 121, citing Hacking 1983: 28). Egg interprets the apparent contradiction of bracketing off epistemology while simultaneously rolling it into metaphysics as suggesting that while Hacking aims to set aside questions of the rationality of science, he is concerned with the representationalist epistemology of SR after all (2018: 121). But given a pragmatist understanding of ER, the more plausible interpretation is simply that Hacking does not think of knowledge as representational. Instead, for Hacking knowledge is embodied in our practices, demonstrated through our manipulations, or directly evident when it has instrumental value in guiding actions. So, Hacking brackets off representationalist epistemology, not epistemology as such.

  • 5 Hacking notes that instrumentalism was later understood as a more thoroughgoing antirealism, but he (...)

40Authors of objections (a) through (c) project a representationalist epistemology onto Hacking’s thesis. They trade in an understanding of knowledge whose truth inheres in correct representations, and one which stands outside of scientific practices, allegedly mirroring them. With regard to the difference Hacking highlights between thinking and doing, Egg notes that “[a]s long as only representation is concerned, even the most useful concepts (‘tools for thinking’) can be interpreted non-realistically” (ibid.). The critics miss this possibility. Objection (a), that experimental practice is theory-laden, is a non-issue for ER, for as Egg implies, theory can have instrumental value in guiding actions just as well whether it is understood realistically or non-realistically. Indeed, Hacking notes that Dewey’s instrumentalism about theory involved an antireal understanding of it as something human beings made – a tool of thought (1983: 63).5 Following Dewey, Hacking asserts that “[t]heories and laws have no truth in themselves. They are only instruments, not to be understood as literal assertions. Terms that seemingly denote invisible entities do not function as referential terms at all” (ibid.). Once knowledge is understood in a pragmatist way, it doesn’t make a difference to manipulations of reality whether theory is true (in the representationalist sense) or not. So, ER is not committed to SR’s theses (2) and (3). The epistemology that gets rolled into Hacking’s metaphysics is a pragmatist, instrumental epistemology.

41On this picture, objection (b) loses force for the same reason as objection (a): because home truths can themselves be understood instrumentally. Home truths are just those “truths” that scientists take for granted and use in experimental practices across a wider variety of disciplinary or subdisciplinary contexts. Thus, home truths can be understood as more foundational, general truths that are instrumental across wider domains of science. While critics assume Hacking must be a realist about home truths, which would be inconsistent with the variety of ER defended here, this is not the case. There is no reason why home truths, on his account, shouldn’t be understood in a pragmatist way. The home truths that are relevant in a particular scientific context are not themselves under test in a given experiment; rather they constitute the preconditions for some new experiment(s) to be conducted. In other words, the priority of home truths cannot be justified in a normative sense; rather, it is implicit in experimental practice. The distinction between theoretical truths and home truths is simply the distinction between what, in a given context, is under test versus what must be the case for the test to be possible. (Here, what “must be the case” can itself be understood within the framework of ER: what entities must be real (for manipulations to cause effects), and what semantic and epistemic commitments are useful.)

  • 6 In light of Suárez’s 2008 and Egg’s 2014 discussions, I note that it is also not necessarily an inf (...)

42With regard to objection (c), that Hacking is employing IBE, ER does not entail an inference from the success of experimental practices to the truth, understood representationally, of the theory of the entity involved in such success.6 What Hacking takes away from experimental success is the foregone scientific acceptance of the reality of some entities qua causal: after all, they were used in the process of inquiry. In line with a broadly Deweyan epistemology, ER does not necessarily make any further inference to any separate, transcendent truth. Indeed, Dewey claimed that complex organisms – including nonhuman ones – modify their environments (1938: 28). (I interpret modification as a form of manipulation.) Dewey furthermore claimed that “[a] person, or, more generically, an organism, becomes a knowing subject in virtue of engaging in operations of controlled inquiry” (ibid.). This view is supported by Chang, whose action-based view of knowledge centers operational coherence, which consists in the teleological coordination of activities, and thereby applies to nonhuman organisms (2022: 40-50). Yet nonhuman organisms make no further inferences despite sometimes conducting successful inquiries, so such inferences cannot be necessary.

43Even if such inferences are made implicitly – something Hacking seems to entertain (1983: 162), this does not imply that the inferred truth must be understood as a correct representation of reality. Here, it is telling that critics are silent on the implications of failed experiments; what representation of reality would they infer then? The post-experimental inference of a truth understood in a realist way is a presumed step that critics unjustifiably tack on to Hacking’s view. Under ER, no further inference necessarily occurs, or false inferences may occur after experimental success. ER easily accommodates this given that its realism is about entities only. A pragmatist understanding of ER’s epistemology renders it immune to the objection that it’s a circular form of IBE since a transcendent “truth” is not necessarily derived from experimental success.

44There’s an important sense, therefore, in which these objections don’t land, for these critics are talking past ER’s epistemology qua pragmatist.

5b. Response to Objection (d)

  • 7 “Local” ER is also what Chen defends (2023).
  • 8 Here, I am setting aside any questions with regard to whether science makes progress or what progre (...)

45Objection (d) to ER is concerned with the fact that some presumably real entities aren’t manipulable; this objection is primarily due to Morrison (1990) and Shapere (1993). They argue that many astrophysical phenomena are presumably real yet can’t be manipulated. Those with sympathy for SR are therefore concerned that manipulability as a criterion doesn’t capture important cases of real entities. But this is not a particular problem for Hacking’s ER. As Resnik points out, Hacking’s realism about entities is on a case-by-case basis (1994/1998: 1175). Hacking endorses what he calls “realism-in-particular” rather than “realism-in-general,” where realism-in-particular is realism that is local to select entities only (Hacking 1983: 30-1).7 When it comes to astrophysical phenomena, Hacking thinks we can do no better than van Fraassen’s empirical adequacy (1989: 578). Both Morrison and Shapere launch objection (d) from the presumption that SR is desirable in general, but it’s one they don’t defend and one that Hacking doesn’t share. Shapere’s argument in particular, which is at pains to show the progress that’s been made in astrophysical science, seems to assume that SR is required for such progress, but this is not established by his argument. It’s a live possibility that empirical adequacy, an antirealist position, would have facilitated the same progress.8

46Furthermore, Hacking aimed to establish manipulability as a sufficient condition for ER only; manipulability is not a necessary condition. (See Section 2 for a discussion of this reading.) As Nola argues, “At best manipulability is a mark of the real and not a criterion for the real. Manipulation is a success term (one cannot manipulate the unreal), and is a sufficient condition for realism; but it is not a necessary condition” (2002: 5). So, even if ER does not provide a standard for determining all and only real entities, it does provide a standard for determining only real entities. This implies nothing in particular for unmanipulated entities, such as astrophysical ones. It is therefore a feature, not a bug, of ER that it remains agnostic about unmanipulated entities. The very reason ER doesn’t pronounce unmanipulated entities real is that it is sensitive to Kuhnian challenges. And absent an alternative realist response to Kuhnian challenges, the objection that ER doesn’t count intuitively real phenomena as real is rather a strength, for those very intuitions have led to many false scientific claims in the history of science.

6. Response to Objection (e)

47Nola’s (as well as Morrison’s and Chen’s) interpretation that manipulability is a sufficient condition notwithstanding, Morrison (1990), Elsamahi (1994), Gelfert (2003), and Chang (2022) challenge the idea that manipulability is a sufficient condition for realism about select entities. If this objection to ER is compelling, it would mean that ER fails on its own terms. Morrison’s, Gelfert’s, and Chang’s counterexamples to theory-free experimental manipulation seem to suggest that non-real entities can be manipulated.

48But there are problems with their line of argument. First, there is no sense in which theoretical entities can have the kinds of causal powers that are required to produce effects. If effects, in the context of physical experiments, are produced, they must be the results of physical causes. It is here that Cartwright’s insight can offer help. By focusing on material, causal interactions, her account shows why it is that theoretical entities aren’t being manipulated: because such “fictional” entities, as Elsamahi had it (1994: 176), can have no physical causal powers! And even if the entities that produced effects were misunderstood, misdescribed, or misconceived, it does not mean they were not real qua effect-producing.

49But, as Massimi suggests (2004), this point raises the question about precisely what entity is causing effects in the examples that Morrison, Gelfert, and Chang, not to mention Hacking, cited, since that is underdetermined by the data. If it wasn’t ions in the case of manipulating the alpha particle, holes in the Fermi sphere in the case of manipulating electrons, or phlogiston in the case of Stahl’s experiments, then on the causal line of reasoning, it must have been another entity or set of entities that produced effects. And yet, if this is correct, it implies that while we may know that some entity or entities are producing effects, we may not know what entity or entities are doing so.

  • 9 See Polanyi on tacit knowledge (1966).

50I propose, however, that ER resists this point rather well for two reasons. The first reason is that ER has no quarrel with theoretical entities as such. This is true given ER’s rejection of SR’s commitments (2) and (3). Theoretical entities may be instrumental for experimental hypothesis testing wherein the experimental attempt is made despite manipulative success not being guaranteed. Theory, combined with the tacit knowledge of scientists, may suggest to scientists the experiments to attempt.9 From the point of view of ER, it is wholly natural to perform exploratory experiments based on entities which we merely suspect exist – entities which we are (rightly) antirealists about, pending further investigation. Then, if these experiments are (repeatedly) successful, scientists have already implicitly accepted the manipulated entity as real, for its reality is embodied in the successful experiments. Again, it’s not that if the experiments are successful, we are justified in inferring the existence of the manipulated entity – that gets the time-order wrong. Rather, we have demonstrated our acceptance of the entity as real by designing effective experimental apparatus which relies upon it.

  • 10 Otávio Bueno (2018) may also be instructive here.

51But again: ER only tells us that some entity or entities are being manipulated, rather than what entity or entities are being manipulated, thus the view warrants realism about some unspecified entity (or entities), rather than realism about any particular entity. As Gelfert pressed: what are the determinate identity conditions of the manipulated entity? Chen provides a response to this issue by arguing that entities have basic, non-theoretical ontic properties, such as individuality, which are not theoretical, and that such properties were sufficient for experimental manipulations carried out with optical tweezers (2023). I am not sure whether properties such as individuality are basic and non-theoretical, however, given the long – and continuing – history of philosophical debates about individuality (e.g., about biological individuality).10 Still, the kinds of non-theoretical ontic properties suggested by Chen (manipulability, tangibility, observability, individuality, countability, etc.) may be understood to fall into the domain of Hacking’s home truths, and therefore not be part of the theories with which the relevant manipulations are allegedly laden.

52But this brings me to the second reason why ER can resist the objection that non-real entities may be manipulable: when we manipulate an entity, we are in effect gripping nature at its joints, even if we are gripping loosely, crudely, or gripping more than one joint at a time. By gripping, I mean wielding in the course of manipulation; gripping an entity means interacting with it in such a way as to “hook up” with its physical reality. This is the active, material analog of Plato’s carving of “carv[ing] nature at its joints” fame (Phaedrus 265e). Gripping is necessary for manipulating. While carving nature at its joints is an activity aimed at discovering the representational truth of the kinds to be found in nature, which are supposed to guide theory, I suggest that gripping nature at its joints is an activity aimed at causing changes in nature’s phenomena, which is supposed to serve human ends, and through which we learn. Such gripping can, in Dewey’s words, lead to the construction of conceptual frameworks that have instrumental value in predicting experience and guiding actions – including experimental ones in science.

53That we can grip electrons via PEGGY II suggests that electrons are real, which means we can likely grip and manipulate them in the future. Even if it turns out later that the electron is in fact a set of closely related phenomena, or that other experimental apparatuses can better manipulate certain entities, PEGGY II will continue working, thereby showing that “the electron” retains its instrumental value and guides future actions because it retains enough unity to be gripped. As Egg explained, while we may not always be able to identify entities qua particular individuals, there’s no reason we may not identify them as such by their causal powers (2018: 124).

  • 11 However, the reality of entities is historically contingent in the sense that the world itself may (...)

54Furthermore, as Nola argued, we need not have full knowledge of what we manipulate to bring about a desired end; we must simply have enough knowledge of (enough of) its causal powers to manipulate it (2002: 6). Gripping an entity does not require its exact delineation. He reminds us that in the nineteenth century, Semmelweis was able to reduce the incidence of maternal death as a result of childbed fever in a particular hospital, sans any theory of what Semmelweis called “cadaveric particles,” by getting doctors to wash their hands (ibid.: 5-6). This was based on Semmelweis’ presupposition that cadaveric particles were remaining on doctors’ hands after they examined cadavers and were being transferred to laboring women during examinations. By implementing hand-washing, Semmelweis changed the frequency of death by childbed fever by manipulating cadaveric particles. Because his preventative technique was so effective, people believed Semmelweis discovered a new kind of infectious particle (ibid.) – in other words, a new entity. This occurred in the absence of any theory of infection and in the absence of any exact delineation of “cadaveric particles.” Nevertheless, cadaveric particles – now known as bacteria – were manipulated. Thus, we need only limited theoretical precision to manipulate some entities. Thus, a way to account for the apparent manipulations of phlogiston or of frigorific rays is to historicize the practice of target system isolation, with the understanding that accuracy and precision with regard to isolating the target entities that one is manipulating in experimental systems develops over time. This should not suggest, however, that the reality of whatever entity caused effects is historically contingent: the effects were either produced or not, although this may be a matter of degree.11 In the case of less well-isolated entities, effects may occur despite a lack of accuracy or precision with regard to what is gripped. Perhaps this is why Chang suggests that it is possible to carry out “coherent activities” with phlogiston theory, and the theory is true to that extent (2022: 194-5), although the truth of theories is not ER’s concern.

55This narrative is a far cry from Gelfert’s desired “determinate identity conditions,” but such conditions are unsustainable anyway under the pressure of Kuhnian arguments. As Peter Galison said, “[e]xperimentalists’ demonstrations of the reality – or artificiality – of an effect or particle will never have the closed form of a deductive argument” (1987: 2). Galison argues that there’s no way to exhaustively ensure that no background entities or phenomena caused the experimental effects, for the world is simply too complex, and for this reason there is no “strictly logical” termination point in scientific experimentation (ibid.: 3). That means there is no particular moment at which an entity’s identity conditions become determinate. That science is ongoing, and its results disputed and revised, illustrates this point. In pragmatist terms: inquiry continues.

56The scientific realist’s desire for determinate identity conditions which can justify the representational truth (or falsity) of scientific claims is wrong-headed. As Ihde suggests, a philosophy of science “which is limited solely to linguistic, logical, or propositional methods will not be able to adequately account for large sectors of scientific activity which entails more than such ‘rational’ procedures alone” (1991: 100). Nevertheless, our experimental practices betray that we are in fact realists about many entities, and rightly: they causally, materially – not just theoretically – serve our ends when we grip them. And many are stable enough over time to warrant claims about them that can be understood as (rather) fixed laws of nature – at least fixed relative to the span of time in which human beings have conducted science, broadly understood. Galison reminds us, “[s]omewhere, something has to hold firm or no experimentalist could ever rely on any set of results” (1987: 4). This should comfort the scientific realist. Not coincidentally, it is the very basis of technology, a topic to which I now turn.

7. The Argument from Technology

57Let’s return to something Hacking asked: why are we sure of the reality of macroscopic objects if not because of what we do to them and with them, and what they do to us? He noted that little or no attention is paid by analytic philosophers to technology (1983: 149). An important argument for ER, not considered by its analytic critics nor by its defenders (with a potential exception, discussed below), is what I will call the argument from technology (henceforth, AFT). The argument, in brief, is this: we manipulate macroscopic entities frequently, often daily or even hourly, in the course of using various technologies (broadly understood). Therefore, as our actions demonstrate, we are implicit realists about the entities embodied by our technologies and involved in making them work reliably. The “success of science,” on this view, can be understood as embodied in technologies that generally work. Manipulating our physical environment through technology is quite like manipulating electrons via PEGGY II. Just because some of our technological manipulations trade in observable rather than unobservable entities makes no difference, for our actions toward our ends in both cases betray our foregone acceptance of the reality of some entities in those cases in which we physically manipulate them. This is the sense in which we engage in what Hildebrand called experience-nature transactions (see Section 3a).

58For example, we engage in experience-nature transactions when we take ibuprofen tablets to relieve headaches. We are thus realists about ibuprofen, for when we consume it in the form of a tablet, we manipulate its causal power in service of the human end of pain relief. (And, indeed, we don’t see the chemical form of ibuprofen when we consume it, and we don’t need an understanding of the underlying chemical theory for effective treatment.) We likewise engage in experience-nature transactions when we use mobile phones. We are implicitly realists about lithium and cobalt, for instance, for when we make phone calls, we manipulate their causal powers in service of reaching the phone of the person we are calling. Our very actions divulge the reality of the entities on which our technologies, and the patterns of our behaviors, depend. Both global scientific realists and global antirealists behave in ways that demonstrate their acceptance of the often theory-free reality of many entities: for instance, they eat food, often cooking it, nutritional and chemical theory or no. Indeed, many of us manipulate ibuprofen and mobile phones, and many other entities besides, in theory-free ways. Were it not the case that we accepted the reality of the entities on which our technologies depend, our actions would be deeply irrational. Their regularity would call for explanations from both scientific realists and antirealists alike!

59This is not an entirely new argument. Ihde addresses what he calls the science-technology interface and contends that science is embodied in its technology (1991). He argues that entity realism emerged when a handful of rather materialist philosophers from both the analytic and continental-phenomenological traditions began to address their philosophies of science to its praxis and instrumentation (1991). While Ihde is mostly concerned about technology in the sense that it mediates what kinds of entities we create or take to exist, his argument can be understood more broadly to suggest that technology use is a form of ongoing experimentation given that we use technology to manipulate the world in daily life. Again, were we not manipulating real entities that, in the words of Dewey, had “availability for conducting to a successful issue the activities of living beings,” we’d be systematically doing things in vain.

60Nola likewise notes that ER can account for our uses of technology (2002). Just as we enter into what Nola calls a “causal nexus” (ibid.: 8) with electrons via PEGGY II, we enter into a causal nexus with ibuprofen and lithium and cobalt; we manipulate these entities in our efforts to cause desired effects. Thus, we are manifestly realists about these and other entities, for it is not only the chemical theory explaining ibuprofen nor only the metallurgical theory explaining the function of lithium and cobalt in mobile phones that allows us to manifest our ends of pain relief and long-distance communication. Nola presents an example from biochemistry: genetic technology has increased crop yields and allowed for greater disease management (ibid.: 4). This is because we have entered the causal nexus of genes and manipulated them; it’s not an accident this is called genetic engineering. Echoing Ihde’s and Dewey’s concerns, Nola highlights that unlike with electron manipulation, genetic manipulation “reaches outside pure science” when it becomes part of the technologies which serve the human ends of food production and healthcare (ibid.). Dewey claimed that inquiry had the same structure across everyday life and science (1938: 101). Farmers plant certain genetically engineered crops – not because they know genetic theories to be true (in the scientific realist sense), but because it is profitable to do so in certain cases. Yet it is not an accident that it is profitable to do so (in certain cases): were the genetically engineered crops – the entities in question – not actually different from their non-engineered cousins, they would be no more profitable to plant.

  • 12 Other interpretations include that Hacking’s argument for ER is a NMA, an indispensability argument (...)

61But the AFT may not seem like much of an argument, and likewise for the argument from manipulability in general, because reasons qua justifications aren’t being given for ER. Rather, a narrative showing our implicit acceptance of the reality of select entities is being provided, one rooted in a description of material activity, not in normative criteria. It’s no surprise Miller suggests that one plausible interpretation of Hacking’s defense of ER is that it is a non-argument (2016).12 What Miller means is that Hacking’s support for ER may be based in the fact that “experimental practice directly illustrates [electrons’] reality” (ibid.: 1004). Miller suggests that the epistemic warrant involved in this interpretation is the direct perception of laboratory practice, rather than any argument for it (ibid.).

62I suggest, however, that it’s not the perception of laboratory practice that provides the epistemic warrant for ER, for the terminology of perception of a fact places ER squarely at the mercy of the kind of representationalist epistemology that Hacking rejects. Rather, the epistemic warrant must be understood pragmatically: it is the experimental – and, in the case of technology – the experiential, practice of successfully manipulating our world that shows what we “know.” While Miller is not optimistic that Hacking’s argument for ER should be understood as a non-argument, I am, given my pragmatist interpretation of ER.

63The AFT can thus be contrasted with other defenses of ER – those mentioned in Section 3b, as well as with Chen’s (2023) defense. This is despite the fact that unlike most other scholars who defend some variety of EER, Chen defends MER, or a thesis very close to MER. The AFT is a novel defense specifically of entity realism. It preserves ER’s middle position between scientific realism and scientific antirealism, thereby preserving its ability to accommodate powerful counterarguments from the history of science as well as accommodating other antirealist arguments.

64Furthermore, the daily practice of theory-free technology use can itself be well-understood as a non-argument. There is no transcendent reason that can be pointed to which justifies realism about certain entities; we simply live – embody, materialize, demonstrate – our acceptance of the reality of select entities by going about our daily technology-filled lives and manipulating them. Our lives contain not just electrons, and not just ibuprofen, lithium, and cobalt, but countless entities which, were they merely theoretical, would render the regularity of our daily interactions with mid-sized objects a massive coincidence. The non-argument interpretation of the case for ER rightly preserves ER’s emphasis on praxis and frees it from dependence on the kind of representational epistemology that Hacking rejects. From a pragmatist perspective, the “success of science” can be understood as embodied in technologies that generally work. That they sometimes fail to work, and that new technologies continue to be developed, is consistent with Dewey’s account of inquiry, as well as with his famous idea that “we live forward” (1917).

8. Conclusion

65I claim that entity realism, understood as metaphysical experimental realism, is a more compelling perspective on science than previously thought, one which, when understood as a pragmatist view, offers a genuine alternative to scientific realism as well as to varieties of antirealism. A pragmatist understanding of entity realism helps dispel a number of objections to it. The case for entity realism is also strengthened by the argument from technology, which shows our implicit commitment to the reality of entities across the everyday and scientific contexts. The argument from technology, however, calls out for further development.

Top of page


Bhaskar Roy, (1978), A Realist Theory of Science, Atlantic Highlands, Humanities Press.

Bih Joseph Z., (2003), “The Microwave Technology,” 13th International Crimean Conference Microwave and Telecommunication Technology (CriMiCo 2003), Sevastopol, Ukraine: 15-21. Doi: 10.1109/CRMICO.2003.158731.

Boyd Richard, (1984), “The Current Status of Scientific Realism,” in Jarrett Leplin (ed.), Scientific Realism, Berkeley, University of California Press.

Bueno Otávio, (2018), “Can Quantum Objects Be Tracked?,” in Otávio Bueno, Ruey-Lin Chen, & Melinda Bonnie Fagan (eds), Individualtion, Process, and Scientific Practices, Oxford, Oxford University Press.

Cartwright Nancy, (1983), How the Laws of Physics Lie, Oxford, Oxford University Press.

Cartwright Nancy, (1999), The Dappled World: A Study of the Boundaries of Science, Cambridge, Cambridge University Press.

Chakravartty Anjan, (2007), A Metaphysics for Scientific Realism: Knowing the Unobservable, Cambridge, Cambridge University Press.

Chakravartty Anjan, (2017), “Scientific Realism,” in Edward Zalta (ed.), Standard Encyclopedia of Philosophy. Online: (Accessed May 22, 2021).

Chang Hasok, (2022), Realism for Realistic People: A New Pragmatist Philosophy of Science, Cambridge, Cambridge University Press.

Chen Ruey-Lin, (2023), “Are There Experimental Arguments Independent of Theories? In Defense of a Hackingian Approach to the Scientific Realism Debate,” International Studies in the Philosophy of Science, 36(4), 279-97.

Clarke Steve, (2001), “Defensible Territory for Entity Realism,” British Journal for the Philosophy of Science, 52, 701-22.

Dewey John, (1917), “The Need for a Recovery of Philosophy,” in John Dewey et al. (eds), Creative Intelligence: Essays in the Pragmatic Attitude, New York, Henry Holt and Company.

Dewey John, (1925), Experience and Nature, La Salle, Open Court.

Dewey John, (1938), Logic: The Theory of Inquiry, New York, Holt, Reinhardt & Winston.

Dewey John, (1978), “The Bearings of Pragmatism Upon Education,” in The Middle Works of John Dewey, 1899-1924, vol. 4, ed. by Jo Ann Boydston, Carbondale and Edwardsville, Southern Illinois University Press.

Dewey John, (1985), The Later Works of John Dewey, 1925-1953, vol. 12, ed. by Jo Ann Boydston, Carbondale and Edwardsville, Southern Illinois University Press.

Egg Matthias, (2014), “Causal vs. Theoretical Warrant,” in Matthias Egg, Scientific Realism in Particle Physics: A Causal Approach, Berlin, De Gruyter.

Egg Matthias, (2018), “Entity Realism,” in Juha Saatsi (ed.), The Routledge Handbook of Scientific Realism, London, Routledge.

Elsamahi Mohamed, (1994), “Could Theoretical Entities Save Realism?,” PSA 1994: Proceedings of the Biennial Meetings of the Philosophy of Science Association, 1, 173-80.

Eronen Markus I., (2015), “Robustness and Reality,” Synthese, 192(12), 3961-77.

Eronen Markus I., (2017), “Robust Realism for the Life Sciences,” Synthese, 196(6), 2341-54.

Fesmire Steven, (2015), Dewey, New York, Routledge.

Fine Arthur, (1998 [1984]), “The Natural Ontological Attitude,” in Martin Curd & Jan A. Cover (eds), Philosophy of Science: The Central Issues, New York, W. W. Norton & Co.

Galison Peter, (1987), How Experiments End, Chicago, The University of Chicago Press.

Gelfert Axel, (2003), “Manipulative Success and the Unreal,” International Studies in the Philosophy of Science, 17(3), 245-63.

Gross Alan G., (1990), “Reinventing Certainty: The Significance of Ian Hacking’s Realism,” PSA 1990: Proceedings of the Biennial Meetings of the Philosophy of Science Association, 1, 421-31.

Hacking Ian, (1983), Representing and Intervening: Introductory Topics in the Philosophy of Natural Science, New York, Cambridge University Press.

Hacking Ian, (1989), “Extragalactic Reality: The Case of Gravitational Lensing,” Philosophy of Science, 56, 555-81.

Hacking Ian, (1998 [1982]), “Experimentation and Scientific Realism,” in Martin Curd & Jan A. Cover (eds), Philosophy of Science: The Central Issues, First Edition, New York, W. W. Norton & Co, 1153-68.

Hildebrand David, (2018), “John Dewey,” in Edward Zalta (ed.), Stanford Encyclopedia of Philosophy. Online: (Accessed June 24, 2023).

Ihde Don, (1991), Instrumental Realism: The Interface between Philosophy of Science and Philosophy of Technology, Bloomington, Indiana University Press.

Khalili Mahdi, (2023), “A Dialogue Among Recent Views of Entity Realism,” Philosophy of Science, 90, 901-21.

Kuhn Thomas, (1962), The Structure of Scientific Revolutions, Chicago, The University of Chicago Press.

Laudan Larry, (1981), “A Confutation of Convergent Realism,” Philosophy of Science, 48(1), 19-49.

Massimi Michela, (2004), “Non‐defensible Middle Ground for Experimental Realism: Why We Are Justified to Believe in Colored Quarks,” Philosophy of Science, 71(1), 36-60.

Miller Boaz, (2016), “What is Hacking’s Argument for Entity Realism?,” Synthese, 993, 991-1006.

Morrison Margaret, (1990), “Theory, Intervention and Realism,” Synthese, 82(1), 1-22.

Musgrave Alan, (1996), “Realism, Truth, and Objectivity,” in Robert Cohen, Risto Hilpinen & Qiu Renzong (eds), Realism and Anti-realism in the Philosophy of Science: Beijing International Conference, Dordrecht, Kluwer Academic Publishers.

Nanay Bence, (2013), “Singularist Semirealism,” British Journal for the Philosophy of Science, 64(2), 371-94.

Nanay Bence, (2019), “Entity Realism and Singularist Semirealism,” Synthese, 196, 499-517.

Nola Robert, (2002), “Realism through Manipulation, and by Hypothesis,” in Timothy D. Lyons & Steve Clark (eds), Recent Themes in the Philosophy of Science: Scientific Realism and Commonsense, Dordrecht, Kluwer Academic Publishers.

Plato, (1952), Plato’s Phaedrus, Cambridge, Cambridge University Press.

Polanyi Michael, (1966), The Tacit Dimension, Chicago, The University of Chicago Press.

Psillos Stathis, (1999), “In Defence of Scientific Realism,” in Stathis Psillos, Scientific Realism: How Science Tracks Truth, London, Routledge.

Putnam Hilary, (1975), Mathematics, Matter and Method, Cambridge, Cambridge University Press.

Reiner Richard & Robert Pierson, (1995), “Hacking’s Experimental Realism: An Untenable Middle Ground,” Philosophy of Science, 62(1), 60-9.

Resnik David, (1998 [1994]), “Hacking’s Experimental Realism,” in Martin Curd & Jan A. Cover (eds), Philosophy of Science: The Central Issues, New York, W. W. Norton & Co.

Sankey Howard, (2012), “Reference, Success, and Entity Realism,” Kairos – Revista de Filosofia & Ciência, 5, 31-42.

Shapere Dudley, (1993), “Astronomy and Antirealism,” Philosophy of Science, 60(1), 134-50.

Smart John Jamieson Carswell, (1968), Between Science and Philosophy, New York, Random House.

Suárez Mauricio, (2008), “Experimental Realism Reconsidered: How Inference to the Most Likely Cause Might be Sound,” in Stephan Hartmann, Carl Hoefer & Luc Bovens (eds), Nancy Cartwright’s Philosophy of Science, New York, Routledge.

van Fraassen Bas, (1980), The Scientific Image, Oxford, Oxford University Press.

Top of page


1 Entity realism is sometimes also known in the literature as experimental realism or instrumental realism.

2 Various understandings of ER are discussed below.

3 Notably, Chang, in his defense of a kind of pragmatist realism, disagrees with such a framing. He formulates the fallacy of pre-figuration, which he defines as the erroneous assumption that “reality has well-defined parts and properties that exist independently of all conceptualization” (2022: 74). Chang writes: “Pre-figuration is a fallacy because any entity that we can speak or think about has to be specified in terms of some concepts. This is a fundamental point that was perhaps first articulated by Kant: there is no escape from mind-framing (to put it in my own terms). All that we can think or speak about are phenomena in the Kantian sense” (ibid.). But part of the point of focusing on experiments in the argument for ER is to show that scientists do things other than talk and think about entities: they manipulate them in the course of experiments. And they sometimes do so in rather theory-free ways (as will be argued in later sections). While Dewey himself railed against “hypostatization,” which is the post hoc reification of conceptualized entities (Dewey 1925; see also Steven Fesmire (2015: 64ff.)), this cannot necessarily be read as a globally mind-dependent metaphysical stance given Dewey’s understanding of inquiry, which sometimes depends on interactions with a real, material environment about which not everything is known. Furthermore, such hypostatization cannot be what occurs in cases of theory-free experimental manipulation by definition (because they are theory-free, and because, whether theory-free or not, they may fail). Chang argues that “if the presumably existent entities are noumenal and therefore not specified in terms of any concepts, then there can be no articulable, factual propositions concerning them” (2022: 76). I return to this point in Section 4. I note for now that Chang’s own account of knowledge does not center “factual propositions,” however, nor do I presume their centrality in my arguments here. I suggest, though I cannot defend here, that a reduction of what exists to what is known (i.e. conceptualized) removes the rationale for any further inquiry, and makes it difficult to understand cases of genuine scientific discovery (e.g., of new organisms that had not been previously observed). Roy Bhaskar called such a reduction the epistemic fallacy (1978: 36).

4 Nature is itself understood as mediated by various practices from within the pragmatist framework, but here I simply use the term to refer, roughly, to whatever is not supernatural.

5 Hacking notes that instrumentalism was later understood as a more thoroughgoing antirealism, but he distinguishes this from Dewey’s variety of instrumentalism, which is the version that is relevant here (1983: 63).

6 In light of Suárez’s 2008 and Egg’s 2014 discussions, I note that it is also not necessarily an inference to the most likely cause (IMLC) because it’s not necessarily an inference of any kind.

7 “Local” ER is also what Chen defends (2023).

8 Here, I am setting aside any questions with regard to whether science makes progress or what progress is.

9 See Polanyi on tacit knowledge (1966).

10 Otávio Bueno (2018) may also be instructive here.

11 However, the reality of entities is historically contingent in the sense that the world itself may (materially) change over time.

12 Other interpretations include that Hacking’s argument for ER is a NMA, an indispensability argument, a transcendental argument, or a Vichian argument. See Miller 2016 for details.

Top of page


Electronic reference

Maja Sidzińska, “A Pragmatist Interpretation and Defense of Entity Realism”European Journal of Pragmatism and American Philosophy [Online], XVI-1 | 2024, Online since 17 May 2024, connection on 11 July 2024. URL:; DOI:

Top of page

About the author

Maja Sidzińska

University of Pennsylvania

Top of page



The text only may be used under licence CC BY-NC-ND 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search