Science is undeniably great as a predictive tool. But it’s also full of idealizations – false claims in the form of simplification, exaggeration, and outright distortion. That would seem to rule out scientific realism, the idea that science manages to uncover the fundamental structure of reality. But Elay Shech argues that by contextualizing scientific claims relative to scale and internal standards of a theory, scientific realism can still be defended.
Science is our best epistemic enterprise; it is our best source of knowledge, explanation, and understanding, and it would seem irrational not to believe in its claims. Science also tells us amazing things about the world. It tells us that all matter is made up of fundamental entities too small for the naked eye to see. That there can be correlations between these unobservable entities, particles like electrons, even if they are light years apart. We learn about the properties and behaviors of, or patterns associated with, such unobservables. For instance, electrons have a particular mass, spin, and negative charge, they are deflected in magnetic and electric fields in specific ways, they repulse each other in accordance with Coulomb’s inverse-square law, and so on.
Yet, an integral part of what fuels science’s ability for learning about the world are idealizations. In fact, idealizations are ubiquitous in science. They are distortions or false claims that enter into our best theories, laws, models, and scientific representations. Examples include infinite populations, perfectly rational agents, non-viscous fluid flow, the Bohr model of the atom, Schelling's model of segregation, etc. Science, ostensibly, discovers truth by appealing to lies (in the form of idealizations).
This leads to a puzzle: On the one hand, if science is our best epistemic enterprise, so it would seem we should be scientific realists. That is, we should believe that our best theories are approximately true and that postulated unobservable entities exist and behave more or less according to how our theories say they do. On the other hand, if science always appeals to idealizations and simplifies the phenomena it seeks to explain, it seems we ought to adopt scientific antirealism or instrumentalism. This is the view that science just offers us good instruments for prediction, at least concerning what is unobservable. So, what’s reasonable to believe about the nature of science, given the pervasive use of idealizations?
As I argue in my recent book, Idealizations in Physics, the problem of idealization is a real problem, and it isn’t clear to me that we have an adequate solution. And though it certainly applies to the scientific realism debate, the problem is more fundamental, stemming from the fact that knowledge, explanation, and understanding afforded by science are produced and transmitted inferentially. Inferences, in turn, are not generally warranted or justified when based on false premises. In other words, insofar as idealizations are a problem for scientific knowledge and rationality, then the problem is much bigger than the realism-antirealism debate.
Even if we assume that the problem of idealization can be solved—say, it turns out we can extract truth from lies, because our lies are approximately true—we would still need to enquire into what type of scientific realism is tenable in light of the pervasiveness of idealizations. In what follows, I consider two possibilities.
The suggestion is that we ought to be realists about causal patterns, instead of, say, the entities and mechanisms underlying or participating in those patterns
Indispensable Idealization Realism
Traditionally, it is believed that science is successful in making predictions, providing explanations, guiding action, etc., because its theories are approximately true. Scientific realists further believe that approximate truth applies also to the unobservable content of theories, for instance, claims about entities like electrons as noted above. One positive argument for the realist position concerns the so-called “enhanced indispensability argument”  and it goes as follows:
(1) We ought rationally to believe in the existence of any entity that plays an
indispensable explanatory role in our best scientific theories.
(2) X plays an indispensable explanatory role in science.
(3) Hence, we ought rationally to believe in the existence of X.
The claim then is that unobservable entities like electrons can be substituted for the “X” in the indispensability argument so that scientific realism is secured. But what about idealizations?
It turns out that some idealizations may play an indispensable explanatory role in explanations of physical phenomena. It is beyond our scope to get into the details but cases that I discuss in my book include infinite systems in the explanation of first-order thermodynamic phase transitions and the renormalization group explanation of the universality of critical phenomena, infinitely long and absolutely impenetrable solenoids in the Aharonov-Bohm effect, and truly two-dimensional system in the fractional quantum Hall effect. Do such indispensable idealizations reduce the indispensability argument to absurdity since they show that one must also be ontologically committed to falsities vis-à-vis idealization? Consequently, is scientific realism based on such arguments untenable? Perhaps, but I do not think this is the case.
Instead, what may follow is that indispensable idealizations in science are akin to other abstracta, paradigmatic examples of which include mathematical objects. Some mathematical realists (or “Platonists”) have appealed to the indispensability argument in order to extend the ontological commitment of scientific realist to include the existence of mathematical objects like numbers. If we are satisfied in being ontologically committed to concrete-physical unobservable entities (like electrons) in addition to abstract mathematical objects (like numbers), then it does not seem far-fetched to include some abstract indispensable idealizations in our ontology (like infinite or two-dimensional systems). Such proposed realism accords well with the standard scientific realist position, but it carries further metaphysical baggage. Namely, in addition to unobservables we also commit to the existence of indispensable idealizations, denizens of a Platonic realm akin to other abstracta such as mathematical entities, universals, possible worlds, and so on.
Causal Pattern Realism
An alternative approach to the tension between scientific realism and idealizations has recently been debated by Angela Potochnik and Chris Pincock . In particular, Potochnick argues that the widespread appeal to idealization ought to motivate a version of scientific realism that is not truth-centric in the usual way. The idea is that in studying phenomenon in the world, say, the behavior of a gas in the laboratory, scientists use idealization to focus in on causal patterns of interest that are embodied in the phenomenon. For example, at relatively low pressures and high temperatures, said gas manifests behavior consistent with the ideal gas law, and such behavior is a type of causal pattern. The suggestion is that we ought to be realists about causal patterns, instead of, say, the entities and mechanisms underlying or participating in those patterns. The problem, though, is that a causal pattern such as the ideal gas law, which concerns a dependence or functional relationship of pressure, volume, and temperature, is very much observable. Thus, as Pincock highlights, for the view to count as realism worthy of the name, “causal pattern realism” needs to be about unobservable patterns.
Still, one may worry that by moving from truth about the phenomenon to truth about the unobservable causal pattern embodied in the phenomenon, we have only shifted the concern of idealization to a different level of analysis. After all, our representations of unobservable causal patterns are highly idealized. For instance, consider the case of the unobservable causal pattern that is Coulomb’s law and is embodied in electrostatic phenomenon. One can use a torsion balance with pith balls in order to show that the electrostatic repulsive force is doubled when the distance between charged pith balls is halved. That is to say, there is an observable causal pattern that we can perceive in the laboratory relating to Coulomb’s law. However, the law itself represents an unobservable causal pattern between electrons, the kind of thing that causal pattern realists can be a realists about. Unfortunately, Coulomb’s law itself is an idealization of sorts, not to mention the classical electromagnetic picture of electrons as point particles interacting via an electric force. It isn’t clear then in what sense a theory or model is “true” about unobservable causal patterns embodied in the phenomenon such that, on the one hand, we can have justified beliefs about them as scientific realists and, on the other hand, they are not reducible to observable causal patterns that many anti-realists accept. Call this the “pattern realist dilemma.”
How can one escape the pattern realist dilemma? I would like to gesture at a possible amendment to causal pattern realism that may help evade the dilemma and that consists in two parts, one ontological and one semantic.
Ontologically speaking, relative to a scale of interest there exist an unobservable causal pattern that we can be realists about, patterns like Coulomb’s law. Semantically speaking, relative to the standards of classical electrodynamics, Coulomb’s law is a true description of the behavior of electrons, so worries about appeals to idealizations evaporate
It has been suggested that in matters of ontological commitment one must take seriously the level or scale at which phenomena are explained. The idea is that to exist is to exist at a scale so that ontology is “scale-relative” , . Scale-relative ontology can do justice to the fact that we want to be ontologically committed to non-fundamental entities like house finches, viruses, and quasi-particles, in addition to fundamental entities like electrons.
Next, it has been argued that in attributing truth or falsity to claims, one needs to attend to the relevant standards and context at play . So, a statement such Coulomb’s law or the ideal gas law can only be true or false relative to a standard that is decided by the context, namely, the theory appealed to and the scale at which patterns of interest manifest. Putting the two together then, my amendment is as follows: Ontologically speaking, relative to a scale of interest there exist an unobservable causal pattern that we can be realists about, patterns like Coulomb’s law. Semantically speaking, relative to the standards of classical electrodynamics, Coulomb’s law is a true description of the behavior of electrons, so worries about appeals to idealizations evaporate.
To what extent, then, does such proposed realism depart from a more standard scientific realist position, if the latter is the view that science reveals the truth about the way the world is? As I have articulated it, the view does not depart from standard realism since the claim remains that science reveals truth about the world. However, there will be disagreement about particular ostensible truths. For instance, many scientific realists would likely hold that although electrons exist, they aren’t really point particles and Coulomb’s law is only approximately true. Instead, the view suggested here holds that, relative to an appropriately chosen standard and scale, electrons really are point particles and Coulomb’s law is true.
Of course, as I discuss in my book, there are difficulties with this amendment. For example, it isn’t clear how a scale and theory determine a unique standard for the assessment of scientific claims, or how we can know which of various possibly conflicting standards to use in certain scenarios. For instance, the Aharonov-Bohm effect is standardly articulated in the context of semiclassical physics where a quantized electron is subjected an external classical electromagnetic field. But the standards of quantum mechanics and classical electromagnetism are different so the scenario does not seem to determine a unique standard for assessment. Similarly, as is well known, quantum mechanics itself admits of various interpretations and these correspond to large clusters of conflicting standards (e.g., on the orthodox interpretation an object’s position may be indeterminate while on the Bohmian interpretation it is determinate). In my view, such problems are entangled with the problem of idealization that we set aside early on in this essay. Nevertheless, it seems to me that we can make headway on problems posed by idealization for scientific realism by developing the notions of scale-relative ontology and contextual semantics.
 Baker, A. (2009), “Mathematical explanation in science.” British Journal for the Philosophy of Science, 60, 611–633
 Lawler, I., Khalifa, K., and E. Shech (Eds.) Scientific Understanding and Representation Modeling in the Physical Sciences, New York, NY: Routledge.
 Ladyman, J., & Ross, D. (2007) Every thing must go: Metaphysics naturalised. Oxford: Oxford University Press
 Shech, E. & P. McGivern (2021) “Fundamentality, Scale, and the Fractional Quantum Hall Effect.” Erkenntnis 86:1411–1430
 (Davey, K. (2011). “Idealizations and Contextualism in Physics.” Philosophy of Science. 78:16-38 and Liu, C. (2019). “Infinite idealization and contextual realism.” Synthese. 196:1885–1918).