Philosophy of Science 101: Realism and Anti-realism 'Compromise'

Foreword

The Philosophy of Science series explores both general questions about the nature of science and specific foundational issues related to the individual sciences. When applied to such subject areas, philosophy is particularly good at illuminating our general understanding of the sciences. This 101 series will investigate what kinds of serious—often unanswered—questions a philosophical approach to science exposes through its heuristic lens. This series, more specifically, will look at the ‘Scientific Realism’ debate throughout, which questions the very content of our best scientific theories and models.

Philosophy of Science 101 will be divided into the following chapters of content:

1. Philosophy of Science 101: The Relationship Between Philosophy and Science

2. Philosophy of Science 101: Scientific Realism

3. Philosophy of Science 101: Anti-Realism

4. Philosophy of Science 101: Realism and Anti-Realism ‘Compromise’

5. Philosophy of Science 101: Causation

6. Philosophy of Science 101: Scientific Models

7. Philosophy of Science 101: Models of Explanation

8. Philosophy of Science 101: Laws of Nature

9. Philosophy of Science 101: Science and Social Context

Philosophy of Science 101: Realism and Anti-Realism ‘Compromise’

Previously, this 101 series has explored both sides of the scientific realism debate: first, one which offers a positive epistemic attitude toward belief in (successful—the "best") scientific theories which postulate the existence of unobservable entities (scientific realism), and second the position which denies various premises of the former (known as anti-realism). The foregoing articles, more specifically, have considered Hilary Putnam’s (1975) notorious "no miracles" argument in favour of scientific realism, including his use of inference to the best explanation ("IBE") and various objections to this such as underdetermination of theory by evidence and empiricism. Other opposing views—as discussed last time—include Bas van Fraassen’s (1980) constructive empiricism and the not-so-widely-accepted reductive empiricism, both examples of anti-realist stances. Why adopt anti-realism in any of its forms, though? Is there good reason to deny some realist theses and not others? A popular response taken by the anti-realist community is that the history of science provides very good reason to reject (various parts of) realism, as this article will discuss. Science has a history, and the theories deemed successful are only "successful" subject to time (says pessimistic meta-induction). It may also be argued, however, that the history of science does not solely suggest that anti-realism is conclusive. This article will therefore investigate a kind of "compromise" position between realism and anti-realism bearing in mind the history of science.

Figure 1: "March of Intellect" print by William Heath, c.1828 (Secord, 2021).

Pessimistic Induction

As philosopher of science Stathis Psillos (2022) puts it, scientific theories seem to have an expiry date. Science is in fact full of theories that were once empirically successful and yet turned out to be false. As such, the pessimistic induction argument goes, scientific realism is highly implausible because of the history of science. Such a history has indeed played a major role in the scientific realism debate and seemingly presents a direct challenge to scientific realism (Psillos, 2022), since if one takes historical evidence seriously, one finds that current theories too will be abandoned (i.e., discovered false). Known as "the pessimistic induction" (or "meta-induction"), this anti-realist line of argument is as follows:

Fact: The history of science is a graveyard of theories that were empirically successful at some point, but which are now regarded as false.

Meta-induction: Currently successful theories will very probably turn out to be false too.

Conclusion: The empirical success of science does not warrant the claim that theories are (approximately) true.

(Doppelt, 2007)

Hence pessimistic induction aims to rebut scientific realism. Please recall the realist "no miracles" argument by Hilary Putnam (1975) which pessimistic induction specifically objects to:

Premise 1: Realism suggests that mature scientific theories have to be successful, and it affords us an explanation of why this is so.

Premise 2: Realism explains the success of scientific theories better than any other philosophical account.

Premise 3: There is evidence that mature scientific theories are successful.

Conclusion (by IBE): Realism is therefore true.

Figure 2: Luminiferous Aether Painting by Neil Mitchell (2017). Per 18th century physics, the universe contained an aether allowing light to travel through space. This has since been debunked by later "revolutionising" science.

As mentioned in the previous 101 article, one might criticise Putnam’s "no miracles" argument on the grounds that IBE is not an acceptable form of inference. As is the focus of this article, however, one might also criticise Putnam’s realist argument on the grounds that premises 1 and 2 specifically are not acceptable or even true (given the history of science). To explore this further, it is useful to think of two "versions" of realism—namely the observational (O) and the theoretical (T) where terms in T refer (i.e., there are things in the world for which they stand) and T is approximately true. With there being two "versions" of realism, there are also therefore two "versions" of each premise in the "no miracles" argument. Either (a) the reference of terms in T suggests that mature scientific theories have to be successful and it affords us an explanation for why this is so, or (b) the fact that T is approximately true implies that mature scientific theories have to be successful and it affords us an explanation for why this is so. Put differently, the "no miracles" argument (Putnam, 1975) either starts by saying that (a) the success of scientific theories can be understood via the reference of T terms to the observational, or (b) the success of scientific theories can be understood via the (approximate) truth of T. Indeed, version (b) follows from (a), as becomes apparent later.

Version 1: Reference and Success

Start by considering the idea that reference suggests success (first version). Arguably, this simply cannot be true. As philosopher of science and epistemologist Larry Laudan (1997) contends, version (a) does not hold since one can find a broad range of referring but unsuccessful theories. An example, amongst many, comes from 18th century chemical atomic theory (Laudan, 1997). For what is known today, its terms refer (namely to atoms), but the theory was so unsuccessful that most chemists abandoned it for affinity chemistry. Laudan (1997) concludes that the realist claim that reference guarantees empirical success is false. Just because the central terms of a theory refer, this does not mean that the claims made about entities are true (i.e., truth is not a guarantee when terms refer). Another example of note is Bohr’s theory of the atom which makes claims about atoms that are simply false, even though its terms refer (Laudan, 1997). Laudan (1997) even goes on to assert that one can construct infinitely many unsuccessful but referring theories. Moreover, the reverse applies too. Success cannot suggest reference either. Again, there are many examples of theories that were once successful and well-confirmed but whose central theoretical terms are (in modern light) non-referring. Many are listed on what is become known as Laudan’s list which has proved unambiguously successful (Worrall, 1994). Perhaps approximate truth on version (b) indicates success instead?

Figure 3: Flammarion's Historical Graphic Interpretation of Man's Quest for Knowledge of the Universe 1888 (Bristol, 2014).

Version 2: Approximate Truth

Ideally, the realist motivation is that if a theory is true then it is successful. No current theory can ever do this, however, hence settling for less: approximate truth. On this version, approximate truth suggests success according to scientific realism. Yet there is an immediate problem, namely that the very notion of approximate truth is often left unexplained. This is impacted by the most prominent account of approximate truth given by Karl Popper (1966) called verisimilitude (i.e., the appearance of being true or real). Popper’s account, however, unfortunately does not lend any support to the claim that approximate truth suggests success. According to Popper (1966), a theory is approximately true if and only if ("iff") its truth content is much greater than its falsity content. More specifically:

Let Tt (and Tf) be the set of true (or false) consequences of theory T; then T is approximately true iff:

C (Tt) >> C (Tf),

where C (Tt) is the "size" of the set Tt, or the theory’s "truth content"; and accordingly for C (Tf) (Popper, 1966). On this account, approximate truth does not imply success since it may well be that all predictions one de facto derives belong to Tf, in which case the theory is unsuccessful. Even more unfortunately, the reverse does not apply either. Success does not suggest approximate truth. No matter how approximate truth is analysed, a theory cannot be approximately true if its central theoretical terms fail to refer (Boyd, 1990). It is a necessary condition for a theory to be approximately true that its central terms refer. Yet, as illustrated by Laudan’s (1997) list discussed earlier, the history of science offers an impressive set of theories that once were successful but whose terms failed to refer. Indeed, this then undercuts the connection between success and approximate truth. One must therefore draw the overall conclusion, based on investigation into both version (a) and (b) of premise 1, that premise 1 of Putnam’s (1975) "no miracles" argument is untenable in light of pessimistic induction. The same goes for premise 2 which follows. Per premise 2, realism explains the success of scientific theories better than any other philosophical account (Putnam, 1975) (hence it is no miracle that mature and predictively successful scientific theories are so successful given realism). Somewhat woefully, however, premise 2 becomes obsolete since premise 1 has been denied. In other words, it cannot be the case that realism is better than any other philosophical account to explain the success of science if premise 1—that realism suggests that mature scientific theories have to be successful, and it affords us an explanation of why this is so—cannot be upheld. Here, the realist’s only option is to try to reconcile the historical record with the claim that successful theories are typically approximately true.

Figure 4: Mid-1970s: First precise model of the Earth (Rocheleau, 2020).

Kuhn and The Progress of Science

Before delving into the realist attempt to reconcile the history of science with so-called empirically "successful" and approximately true scientific theories, it is worth considering Thomas Kuhn’s (1970) philosophy. Indeed, Kuhn is one of the most influential philosophers of science of the 20th century, perhaps the most influential. His 1962 book The Structure of Scientific Revolutions (1970) is one of the most cited academic books of all time (Bird, 2022). His famous account of the development of science held that science enjoys periods of stable growth punctuated by revisionary revolutions. Hence the Kuhnian picture is important to consider here, especially since Kuhn—somewhat controversially—added the "incommensurability thesis" to his theory (Bird, 2022). Kuhn proposes that theories from differing periods suffer from certain deep kinds of failure of comparability. Kuhn (1970) notably starts by setting out the "traditional" picture of the history of science before rebutting it.

In the so-called traditional (and historically—at least before Kuhn—widely accepted) picture of the history of science, science proceeds as knowledge is acquired (Kuhn, 1970). Such knowledge is typically built into theories and in turn refines existing theories by correcting errors. Then, if one theory gets superseded by another theory, the new theory incorporates what was "good" while eliminating its mistakes and adding new elements. Hence, Kuhn (1970) claims, the new theory is better and closer to the truth than the old one on this picture. This is a cumulative process of improvement as such, and this particular (traditional) picture of the history of science is a (more or less) linear accumulation of knowledge (Bird, 2022). According to Kuhn, however, this traditional picture is wrong. There is no accumulation of knowledge, but instead a vicious cycle which may represent the history of science (known as The Kuhn Cycle) (Kuhn, 1970).

Figure 5: Diagram of The Kuhn Cycle: Thomas Kuhn's picture of the history of science (Harich, Bangerter, & Durlacher, 2014).

The Kuhn Cycle starts with a pre-paradigmatic phase where there are many competing schools and approaches on the way to a paradigm. If the ideas developed by one particular camp are eventually (a) promising enough to gain a stable group of supporters and (b) flexible enough to formulate and address different problems, then these ideas become paradigms (Kornai, 2000). A paradigm in this sense refers to the beliefs and standards shared by the scientific community. After the pre-paradigmatic phase comes the "normal science" phase (Kuhn, 1970). As the title suggests, this phase involves problem solving on the basis of a paradigm (hence philosophers of science often refer to ‘mature’ science). Broadly speaking, normal science is the day-to-day research that scientists conduct in order to fill in the gaps in scientific knowledge that are found within the dominant paradigm (Ewoldsen, 2017). Kuhn (1970) notably argued that paradigms always have intractable problems that result in research anomalies though. That being so, the next phase is "model crisis" in the cycle. When a problem cannot be solved on the basis of a paradigm, it turns into anomaly. If the anomaly is considered to be serious, then a crisis results (Kuhn, 1970). Post-crisis comes revolutionary science, where the kind of research during crisis is no longer problem solving. The paradigm itself is at stake and new approaches are explored here (Kuhn, 1970). Finally, scientific revolution occurs when one Yeparadigm is replaced by another. One must note that different paradigms (resulting from revolution) are incommensurable. This is to say that different paradigms have absolutely no common measure (Bird, 2022). Much like Laudan’s (1997) list, which he argues may be extended ad nauseum, Kuhn’s account of the history of science does not look good for scientific realism as the very notion of empirical success and approximate truth is so undermined by history.

The Realist Defence: Selective Realism

Laudan (1997) and Kuhn (1970), in notably different ways, propose discontinuous theory change thus presenting a serious challenge to scientific realism since success (at some point in time) does not appear to be an indicator of (approximate) truth. Might realism have a response, however? This article will consider the various lines of defence of the claim that scientific knowledge grows despite theory change hereafter, starting with selective realism. Though this is a realist account, this article will nevertheless consider selective realism as a kind of "compromise" between realism and anti-realism since theory change across the history of science cannot be denied.

Figure 6: Aristotle sclupture. Aristotle is considered by many to be the first scientist, although the term postdates him by more than two millennia (Lo Presti, 2014).

Selective realism offers a response to discontinuous theory change by arguing that seemingly unfavourable cases are in fact benign if one can show that the success of discarded theories were independent of theoretical mistakes. The "project" is thus to show that success is a function only of certain (select) parts of a theory, and that those parts remain stable across theory change (Vickers, 2017). The source of success that remains unchanged is therefore still approximately true on this account, meaning that discontinuity is no longer a problem. The selective realist in this sense limits realist commitments to those parts of theories that are responsible for success (Vickers, 2017). So-called idle constituents are separated from working constituents to which the realists should commit (Kitcher, 1993). The latter are responsible for theories success and show that they remain stable across theory change. This kind of realism may therefore—arguably—maintain the "no miracles" argument whilst avoiding pessimistic induction. Consider Henri Poincaré’s (1952) argument that this is so, using the equations in scientific theories as an example. As Poincaré points out, equations may express relations, and if the equations remain true it is because the relations they express preserve their reality. The true relations between real objects are the only reality one can attain as such. The intuitive idea is that an equation describes the structure of an object without being committed to claims about the object itself. John Worrall (1989) indeed develops this idea and famously calls the position structural realism, thus providing a saving grace for scientific realism against pessimistic induction (often understood as the only way realism can successfully defend itself against pessimistic induction).

The Realist Defence: Structural Realism

Also known as syntactical realism, Worrall’s (1989) structural realism offers an alternative to the ‘standard’ form of scientific realism. Worrall’s (1989) account was in fact largely influenced by Poincaré—whereas Poincaré talks of relations that hold across theory change, Worrall talks about structures. Poincaré’s talk about relations can be seen as tantamount to talk about structure, and again is a rejection of standard scientific realism (Ladyman, 2020). Though structural realism rejects standard or "fully fledged" realism, this is not to say that structural realism is necessarily anti-realist about science. Rather, one should adopt structural realism and epistemically commit to the mathematical or structural content of scientific theories only (Ladyman, 2020). Retention of structure across theory change, argues Worrall, means that structural realism (a) avoids the force of pessimistic induction (since there is no commitment to belief in a theory’s description of the "furniture" of the world) and (b) does not make the success of science seem miraculous (by committing to the belief that the structure of a theory—over and above empirical content—describes the world) (Worrall, 1989). Hence a kind of "compromise" is reached, which Worrall (1989) calls the best of both worlds.

Figure 7: Photograph of philosopher John Worrall (1989), creator of Structural Realism.

On Worrall’s (1989) view, one must not accept the idea that the nature of things is correctly described by the best scientific theories. This full-blown realist position is in fact successfully undermined by pessimistic induction. Hence it is much easier to be realist about the structure a theory ascribes to a physical system since there is retention of structure across theory change. What instantiates a structure might well change, but the structure itself does not. This is to say that science does not necessarily succeed in identifying real things in nature. Instead, science succeeds in discovering the structure of whatever it is in nature that bears these structures, thus reflecting the epistemological tension between philosophy and science concerning what knowledge or (approximate) "truth" might really refer to (i.e., is it knowledge of a structure, or is it knowledge of the "entities" belonging to that structure in nature? One must decide, given that Worrall's (1989) account is an epistemological modification of realism). Since "natures" are not needed to articulate structures, Worrall (1989) argues that this does not prove to be a problem (Ladyman, 2020). Consider philosopher of science, mathematician and logician Bertrand Russell’s (1927) thoughts on this, who famously asserted that the structure of a theory “does not depend upon the particular terms that make up the field [i.e., the domain] of the relation. The field may be changed without changing the structure” as cited by Friedman (1985, p. 630). Hence in this sense, structural realism is rather like an epistemological modification of standard scientific realism to the effect that one only believes what scientific theories postulate about the relations entered into by unobservable objects, and one suspends judgment as to the nature of the latter (Ladyman, 2020).

Conclusion

The pessimistic induction argument is convincing, and this article has considered its rebuttal of the scientific realist’s notion of epistemic optimism. Broadly speaking, if past "successful" and accepted scientific theories were found to be false, it becomes very difficult to accept the scientific realist’s claim that currently successful scientific theories are approximately true (Psillos, 2022). As this article has explored, the eventful history of science is a significant part of the scientific realism debate and crops up a lot, whether it be to (a) argue against realism (pessimistic induction), (b) deny Putnam’s (1975) "no miracles" argument specifically (i.e., Laudan’s (1997) "list"), (c) provide a new picture of the history of science (Kuhn’s (1970) theory of scientific revolutions), or (d) defend realism (i.e., selective and structural realism, as Worrall (1989) argues).

All in all, the history of science makes the scientific realism debate evermore intense and complex. Most interestingly, and as reflected by an account like structural realism, such a history does not necessarily point to full-blown realism or antirealism. Though pessimistic induction certainly undermines standard scientific realism (specifically epistemic optimism), the history of science—and theory change—may arguably coexist with some distant form of "realism". This results in an intriguing kind of ‘compromise’ which is neither wholly realist nor wholly anti-realist.

Bibliographical References

Bird, A. (2022). Thomas Kuhn. In The {Stanford} Encyclopedia of Philosophy (Spring 2022). Metaphysics Research Lab, Stanford University.

Boyd, R. (1990). Realism, approximate truth, and philosophical method. Scientific theories, 14, 355-391.

Doppelt, G. (2007). Reconstructing scientific realism to rebut the pessimistic meta-induction. Philosophy of Science, 74(1), 96-118.

Ewoldsen, D. R. (2017). Normal science and paradigm shift. The international encyclopedia of communication research methods, 1-17.

Friedman M. (1985). ‘Critical Notice: Bertrand Russell’s The Analysis of Matter: Its Historical Context and Contemporary Interest’, Philosophy of Science 52: 621-639.

Kitcher, P. (1993). The Advancement of Science: Science Without Legend, Objectivity without Illusions, Oxford: Oxford University Press.

Kornai, J. (2000). The system paradigm. Paradigms of social change: Modernization, development, transformation, evolution, 111-33.

Kuhn, T. S. (1970). The Structure of Scientific Revolutions/T.-S. Kuhn.–2-nd Ed.

Laudan, L. (1997). Explaining the success of science: Beyond epistemic realism and relativism. Science and the Quest for Reality, 137-161.

Poincaré, H. (1905/1952). Science and Hypothesis, New York: Dover.

Popper, K. R. (1966). Some comments on truth and the growth of knowledge. In Studies in Logic and the Foundations of Mathematics (Vol. 44, pp. 285-292). Elsevier.

Psillos, S. (2022). Realism and Theory Change in Science. In The {Stanford} Encyclopedia of Philosophy (Fall 2022). Metaphysics Research Lab, Stanford University.

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

Russell, B. (1927). The Analysis of Matter, London: Routledge Kegan Paul.

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

Vickers, P. (2017). Understanding the selective realist defence against the PMI. Synthese, 194, 3221-3232.

Worrall, J. (1989). Structural realism: The best of both worlds?. dialectica, 43(1‐2), 99-124.

Worrall, J. (1994). How to remain (reasonably) optimistic: scientific realism and the “luminiferous ether”. In PSA: Proceedings of the biennial meeting of the Philosophy of Science Association (Vol. 1994, No. 1, pp. 334-342). Cambridge University Press.

Visual Sources