Science means knowledge, and the object of which the sciences are oriented is the natural world. Science therefore aims at knowledge of the natural world. In order to make sense of the world, especially of things observed, scientists develop theories constituting “a network of propositions” involving “theoretical concepts”. Whether or not one thinks these theoretical concepts truly describe the way the world is depends on whether or not one is a scientific realist. Are scientific theories “mere scaffolding which can be dispensed with once they have outlived their usefulness,” which is to say that science merely aims to manipulate the world and make predictions within it – this is the essential anti-realist position – or is there something more enduring to theories? The realist opts for a more enduring perspective.
In what follows is a case for scientific realism, but a realism that is less dogmatic in its epistemic convictions regarding the nature of theoretical entities, and more dogmatic on theoretical structures. The realism argued in this paper is also cognizant of our natural epistemic limitations and the major shifts in theories during the history of science, and therefore recognizes a limited role for the anti-realist perspective, albeit not without qualification. J. P. Moreland refers to this as an integrative model that approaches realism/anti-realism on a “case-by-case basis, guided by some useful principles to aid in such decision making.” In other words, I am arguing for an eclectic structural realism. Eclectic structural realism is a tempered realism that is most consistent with our natural epistemic limitations and the theoretical shifts in the history of science. It is therefore the most defensible version of scientific realism.
Defining Scientific Realism & Anti-Realism
J. P. Moreland lists 5 core tenets of scientific realism. A definition based on these tenets can be written as follows: Scientific realism holds that mature scientific theories in mature sciences are true or approximately true, meaning they refer to real entities in the world, and it is possible to discern the relative truthfulness of competing theories based on abductive reasoning and epistemic virtues (e.g. simplicity, predictive success, scope of relevance), which constitutes the aim of science as producing true conceptions of the theory-independent external world.
In this view, “it is claimed that science has made progress towards this aim, insofar as it has arrived at theories that are at least approximately true and discovered at least some of what there is.” This notion of progress in understanding the world as it is, at least approximately, is a key aspect of realism. We can think of this as understanding the world in degrees. Another definition of scientific realism makes this more explicit: “The realist believes that in principle theories are to be taken literally to some degree, that to some degree they provide us with actual descriptions of the underlying structure of nature or with actual truth.” Here, “to some degree” is used in place of approximate truth.
The previous definition also hints at a particular view of realism known as structural realism. According to this realist view, we should “epistemically commit ourselves only to the mathematical or structural content of our theories.” The continuity of theories, as later theories replace earlier theories, is in their form or structure, not their content. In other words, the natures of unobservable entities are viewed skeptically. Structural realism is viewed as being the best of both worlds (a sort of middle road, but self-consciously realist) and “the most defensible form of scientific realism.” While this paper emphasizes a structural realist perspective, much of what I articulate applies to scientific realism in general.
Although a detailed definition of anti-realism is not necessary for what follows, a simple definition will nonetheless serve to clarify the dividing line, though we have already seen a brief definition by Chalmers above. Okasha provides further insight by remarking as follows: “Anti-realists argue that empirical adequacy, not truth, is the real aim of scientific theorizing.” This is the main concern of the anti-realist, not the truth simpliciter of underlying theoretical entities. Of course, if theories can simply be discarded after serving some instrumental purpose, as we saw above, then theories obviously are not thought to describe the world as it is in itself, lest one dispense with the world.
Arguments for Scientific Realism
We can think of the arguments for scientific realism as generally establishing the view that scientific progress through workable theories – empirically and predictively successful – could not have come about apart from some version of realism. That is, theoretical entities are, for the most part, true or approximately true. The notion of approximate truth is a fundamental tenet of scientific realism. For this reason, it will serve us well to briefly consider an illustration of this fundamental notion.
As an illustration of approximate truth, let’s consider two color-blind men, of differing color-blindness types, who look at a group of swans. In this illustration, the color-blindness represents the limitations of science. One of the men says the swans are gray, while the other says they are black. Knowing that swans (typically) are white, does it not make sense to say that the first man’s assessment is approximately true, while the latter is not? After all, gray is closer to white than is black. Even though black swans do exist, while gray swans do not, the rarity of black swans make the assessment of swans being gray more approximately true to reality, as the vast majority of swans are white. This illustration is particularly illustrative of the notion of approximate truth, as we are here dealing with a color scale (grayscale) which corresponds to the definition of realism as pertaining to degrees of understanding theoretical entities.
Let’s consider a similar illustration. Let’s say there are two, non-color-blind men who live in different parts of the world. One man comes across a bevy of white swans. He therefore concludes that all swans are white. The other man comes across a couple of black swans. He therefore concludes that all swans are black. Knowing the rarity of black swans, does it not make sense to say that the first man’s conclusion is more approximately true, while the latter man’s conclusion is not? Even if we switched out the word “all” with “most” or “some,” the point would be the same.
Now, some may object that we possess the requisite knowledge to test the claims of these two groups of men. This is granted. However, the point of the illustrations is that approximate truth is a rational concept, readily perceived from that which can be observed. If approximate truth is readily perceived in observables, it should be granted in unobservables, while the strength of our theoretical assessments – which theories are more approximately true – pertains to the reliability of the indicators.
Lastly, anti-realists often object that approximate truth has not been clearly defined. While a clear definition of approximate truth may be lacking, it nonetheless remains something we all readily perceive. After all, biologists still debate the definition of life, yet no one denies the reality of life, and all recognize at least some of its contingencies. Do we not all recognize, and regularly communicate in like manner, degrees of closeness to the reality of things, both quantitatively and qualitatively? We therefore recognize at least some of the contingencies of approximate truth.
The No-Miracles Argument & Mature Sciences
This argument relies on abductive reasoning – a plausibility argument or inference to the best explanation. It is said that this is the most formidable of the arguments for scientific realism, as the successes of theories which posit unobservable entities seem to require that they at least be approximately true. In other words, the successes of science are not due to the miraculous but to scientific methodologies resulting in a true or approximately true apprehension of the nature of things, both observable and unobservable.
As is so often the case with the anti-realist rebuttal, appeal is made to the history of science, with many empirically successful theories turning up false. Worrall admits the principle reality of the anti-realist objection at this point. His response is to point to the no-miracles argument as the “independent criterion” of the mature sciences, with the mature sciences essentially falling outside of the anti-realist examples of false theories. This is because the theories worthy of this no-miracles label are those which have experienced “genuine predictive success,” which is a stronger criterion than mere empirical success. So, a mature science is one that has “theories within it which are…predictive of general types of phenomena, without these phenomena having been ‘written into’ the theory [ad hoc or of an already known kind].”
It is important to note that predictive success does not necessarily indicate the truthfulness of a theory. This is important, because Chalmers, who is a structural realist of sorts, slips up on this point when he rhetorically asks, in reference to Einstein’s theory replacing Newton’s predictively successful theory, “doesn’t history force the realist to admit that predictive success is not a necessary indication of truth?” Chalmers seems to be confusing necessity for sufficiency. While the realist would say, and I think anti-realists would agree, that predictive success is a necessary indicator of truth – that is, not having predictive success would indicate a false theory – he would not say that it is a sufficient indicator of truth – that is, predictive success alone does not indicate a true theory. It is not enough to have predictive success, but one should expect predictive success for a true or approximately true theory.
The no-miracles argument is closely tied to theoretical progression and mature sciences, as we have seen from Worrall. However, later theories don’t always appear to be essentially related to earlier theories. Worrall uses the example of a tadpole eventually developing into a frog to illustrate what would be a cumulative development of scientific theories. While there is little resemblance of a tadpole in the frog, one would nonetheless see the essential relationship (approximation) of any immediately preceding developmental stage.
On a similar note, Boyd rightly asserts that “the realist’s account of the methods of science predicts that there will be early stages in the history of any currently mature science in which the relevant background theories will have been too far from the truth to ensure the sort of reliability of methods that is characteristic of mature sciences.” This is to simply recognize that science is a process of trial and error, and that the early stages of any scientific field are more likely than not to contain error. This, however, is a necessary process for a science to become mature. This maturation is what brings us ever closer to truth, what he refers to as “the growth of approximate knowledge.” This growth is a cumulation of theoretical essentials. A limitation of Boyd’s argument, however, is that essential cumulativity is convincing only if a clear and consistent definition of maturity can be achieved, which he fails to produce, according to Worrall. So, while Worrall and Boyd essentially agree on the notion of mature sciences for establishing a cumulative progression of theories, Worrall goes a step further by setting up the no-miracles argument as the independent criterion for a mature science.
Lastly, the idea that a completely overthrown theory, which nonetheless had empirical and predictive success (to some extent), brings into question the realist perspective is a misguided contention. I say this because the fact that the theory has been replaced by another is an example of epistemic scientific progress. It means science is getting closer to discerning the reality of theoretical structures, if not the entities themselves. The more developed a science becomes, which naturally involves a large degree of trial and error early on, the closer theories will conform to the reality of things.
The argument goes as follows: If one and the same thing is apparently detected by various and distinct means, it is safe to conclude that that one thing in fact exists. The fundamental principle undergirding this argument is the principle of multiple witnesses. The detection of a theoretical object via a single mode could be an anomaly; but multiple distinct modes of detection reduces the likelihood of anomalous results. While this does not necessarily support a strong realism, which posits the epistemic reality of the nature of theoretical entities, it does call attention to the structural realist claim of making a commitment to the mathematical or structural content of theories.
Anti-realists argue that “the empirical data…can in principle be explained by many different, mutually incompatible, theories.” That is, the data do not necessarily determine any given theory – it is underdetermined. This leads anti-realists to be skeptical of the reality of theoretical entities.
While there is a trivial sense in which this is true, theory choice goes beyond mere compatibility with the data. One must take into consideration the epistemic virtues. Is one theory simpler in its explanation of the evidence, less ambiguous, more probable, corroborated by theories from other scientific areas (see corroboration above), more predictively successful, etc.? The better of the theories may not be immediately perceivable – more predictions and experimentation may be required, but in due time it will be discerned. Finally, the history of science has seen relatively few actual cases of underdetermination, which is not what would be expected if the underdetermination of the data was more than a triviality.
Eclectic Realism: A Case-by-Case Approach
Now, even though a brief case for scientific structural realism has been made, this is not to say that anti-realism has no place in scientific theories, particularly with regard to immature sciences or theories. This, however, should not be seen as establishing the -ism of anti-realism. To say otherwise is to promote immaturity in the sciences.
It is important to stress that the recognition of a place for anti-realism does not apply to all anti-realist perspectives. As Moreland notes, any anti-realist arguments that contradict the correspondence theory of truth, the adequacy of sensory and cognitive faculties for arriving at justified true beliefs, or the embracing of rational beliefs should be rejected outright, both for rational and theological reasons.
Moreland advocates for a realist/anti-realist decision on a case-by-case basis, noting that “one should start with particular cases in science and not with general criteria for choosing between realism and antirealism.” He is quick to add, and rightly so, that scientific realism should be the assumed position due to its rational strengths, placing “the burden of proof on the antirealist.” For example, a theory should be interpreted in an anti-realist fashion if a particular scientific area has exhibited a replacement of theories as opposed to theoretical refinements to a large degree, or if it has maintained acceptable status through numerous ad hoc adjustments.
This eclectic realism is consistent with the following definition:
The italicized phrases containing “can” and “at least” exhibit a tempered realism that seeks to conform to our natural limitations and the history of scientific theory changes. These things can be so but are not necessarily so. Even Richard Boyd, a strong proponent of scientific realism, remarks that a realist who offers arguments like the one’s provided above “is not committed to the view that rationally applied scientific method will always lead to progress towards the truth, still less to the view that such progress would have the exact truth as an asymptotic limit”. Realism, after all, endeavors to assess situations according to the nature of actual circumstances – limitations and all – as opposed to idealism which endeavors to assess situations according to a preconceived ideal. It would seem, then, that realism demands an eclectic approach to the sciences, as science is a highly nuanced, complex, and ever-changing enterprise.
Theories are like pieces of a puzzle, and the puzzle is the cosmos. What is more, each theory is a puzzle in itself. The only problem is that we weren’t given the puzzle box with the picture on it, at least not in full, though we do have clues. Nonetheless, there is a picture (reality), and the trial-and-error of scientific theories continues to bring us closer to visualizing (understanding) it.
The eclectic structural realist position avoids the rigidity of other perspectives, both within realism and anti-realism. It is realist in that it holds to fundamental tenets of realism, such as approximate truth and the no-miracles argument; it is structural in that its dogmatism lies mainly in theoretical structures, not the nature of entities; it is tempered in that it asserts that such things can be known; and it is eclectic in that it seeks to approach issues on a case-by-case basis, utilizing the strengths of anti-realism when necessary. Such an eclectic epistemology of scientific theories is required due to the diverse, complex, and ever-changing nature of the sciences.
Boyd, Richard. “Realism, Approximate Truth, and Philosophical Method.” In The Philosophy of Science, edited by David Papineau. New York: Oxford University Press, 1996.
Chakravartty, Anjan, “Scientific Realism”, The Stanford Encyclopedia of Philosophy (Summer 2017 Edition), Edward N. Zalta (ed.), https://plato.stanford.edu/entries/scientific-realism/
Chalmers, A. F. What Is This Thing Called Science?. 4th edition. Indiana: Hackett Publishing Company, Inc., 2013.
Craig, William Lane, J. P. Moreland. Philosophical Foundations for a Christian Worldview. 2nd edition. Illinois: InterVarsity Press, 2017.
Ladyman, James, “Structural Realism”, The Stanford Encyclopedia of Philosophy (Winter 2020 Edition), Edward N. Zalta (ed.), https://plato.stanford.edu/entries/structural-realism/
Okasha, Samir. Philosophy of Science: A Very Short Introduction. United Kingdom: Oxford University Press, 2016.
Ratzsch, Del. Science & Its Limits: The Natural Sciences in Christian Perspective. Illinois: InterVarsity Press, 2000.
Worrall, John. “Structural Realism: The Best of Both Worlds?.” In The Philosophy of Science, edited by David Papineau. New York: Oxford University Press, 1996.
 Del Ratzsch, Science & It’s Limits: The Natural Sciences in Christian Perspective (IL: InterVarsity Press, 2000), 74.
 Samir Okasha, Philosophy of Science: A Very Short Introduction, 2nd ed. (UK: Oxford University Press, 2016), 55.
 Chalmers, What Is This Thing Called Science?, 210.
 J. P. Moreland and William Lane Craig, Philosophical Foundations for a Christian Worldview, 2nd ed. (IL: InterVarsity Press, 2017), 369.
 Ibid., 354.
 A. F. Chalmers, What Is This Thing Called Science?, 4th ed. (IN: Hackett Publishing Company Inc., 2013), 219.
 Ratszch, Science & Its Limits, 73.
 James Ladyman, “Structural Realism,” The Stanford Encyclopedia of Philosophy (Winter 2020 Edition), Edward N. Zalta (ed.), https://plato.stanford.edu/entries/structural-realism/. Last accessed on July 26, 2021.
 Anjan Chakravartty, “Scientific Realism,” The Stanford Encyclopedia of Philosophy (Summer 2017 Edition), Edward N. Zalta (ed.), https://plato.stanford.edu/entries/scientific-realism/. Last accessed on July 26, 2021.
 John Worrall, “Structural Realism: The Best of Both Worlds?”, in The Philosophy of Science, edited by David Papineau (UK: Oxford University Press, 1996); Chalmers, 224-226.
 Ladyman, “Structural Realism”.
 Okasha, Philosophy of Science, 55.
 Please note that this is simply for illustration, and the exact nature of their color-blind types is therefore not important.
 Okasha, Philosophy of Science, 59.
 Ladyman, “Structural Realism”.
 Okasha, Philosophy of Science, 60.
 Worrall, “Structural Realism: The Best of Both Worlds?,” 153-154.
 Chalmers, What Is This Thing Called Science?, 217.
 Worrall, “Structural Realism: The Best of Both Worlds?,” 146.
 Richard Boyd, “Realism, Approximate Truth, and Philosophical Method,” in The Philosophy of Science, edited by David Papineau (UK: Oxford University Press, 1996), 240.
 Ibid. 242.
 Worrall, “Structural Realism: The Best of Both Worlds?,” 153.
 Chakravartty, “Scientific Realism”.
 Okasha, Philosophy of Science, 67.
 Ibid., 67-68.
 Ibid., 68.
 Moreland and Craig, Philosophical Foundations, 366-367.
 Ibid., 367.
 Ibid., 367.
 Ibid., 367-368.
 Ratzsch, Science & Its Limits, 82. Emphasis added.
 See Ratzsch’s discussion of limited realism and metaphoric realism, Science & Its Limits, 83-85.
 Richard Boyd, “Realism, Approximate Truth, and Philosophical Method,” in The Philosophy of Science, edited by David Papineau (UK: Oxford University Press, 1996), 215.