The No-Miracles Argument (NMA) for scientific realism, reconstructed as an inference to the best explanation (IBE), faces the charge of vicious circularity, and the standard reliabilist escape inherits the Generality Problem of process reliabilism. Newman (2010, Synthese 177(1):111-138) argued that answering this methodological version of the Generality Problem forces the realist to assume that IBE is a unique natural-kind rule, which itself renders the NMA viciously premise-circular, and posed three questions that jointly delimit the space of prospective non-circular defenses of the NMA: (1) a naturalistic, non-question-begging individuation of inferential process-types that avoids both the single-case and no-distinction aspects of the Generality Problem without the natural-kind assumption; (2) an independent, non-IBE-presupposing framework in which IBE possesses a determinate, time-stable reliability or a principled context-sensitivity; (3) whether a Boyd-style 'philosophical package' can secure realism's commitments without reliabilism and without the motivating regress. These three questions remain unanswered and unrefuted in the current literature; the problem remains open.

Background

Scientific realism is the view that successful scientific theories are approximately true and that their unobservable terms (electrons, fields, genes) genuinely refer. Its most influential argument, the No-Miracles Argument (NMA), goes back to Putnam (1975): the empirical success of mature scientific theories would be a miracle if those theories were not approximately true, so their success is best explained by their approximate truth. Under the now-standard reconstruction, the NMA is an instance of inference to the best explanation (IBE) — the rule that, among rival hypotheses, one should accept the one that would best explain the available evidence. Critics reply that, so reconstructed, the argument is circular: it uses IBE to conclude that IBE, or the methods embedding it, is truth-conducive. Realists have answered in two main ways. Psillos (1999) argued that the circularity is benign and that the argument works holistically. The route Newman (2010) examines is reliabilism, an externalist position in epistemology on which a belief-forming process type is justified when it is in fact reliable (tends to produce true beliefs), whether or not the subject can demonstrate this non-circularly; the absence of a non-circular proof of reliability is not supposed to undermine justification.

Process reliabilism, however, has long faced the Generality Problem, pressed by Conee and Feldman (1998): to evaluate reliability one must specify the process type to which a given token inference belongs, but any token belongs to indefinitely many types at different levels of description. The single-case aspect is that the choice among these typings looks arbitrary for each token; the no-distinction aspect is that no principled criterion separates the typings that yield the wanted reliability verdicts from those that do not, so the theory cannot non-arbitrarily distinguish justified from unjustified beliefs. Newman shows that when the realist uses reliabilism to answer the circularity charge against the NMA, the Generality Problem re-arises at the level of scientific methodology — the relevant process is now IBE itself as a scientific method — and that the scientific realist can answer this methodological version only by assuming that IBE is a unique natural-kind rule, that is, a single inferential natural kind (a grouping tracking the real similarity structure of inference rather than our classifications) with a determinate reliability. But granting that assumption as a premise makes the NMA viciously premise-circular, the very defect the reliabilist retreat was meant to avoid. A further background option is the 'philosophical package' of Boyd (1984): realism defended not by a single argument but as part of an integrated package of metaphysical, semantic, and epistemological theses that are jointly and mutually supporting.

In closing, Newman poses three questions that jointly delimit the space of prospective non-circular defenses of the NMA. First, can a naturalistic account individuate inferential process-types (including IBE) in a non-question-begging way that avoids both the single-case and no-distinction aspects of the Generality Problem, without assuming that IBE is a unique natural-kind rule? Second, is there a defensible metaphysical or methodological framework in which IBE possesses a determinate, time-stable reliability (or a principled context-sensitivity), where the warrant for the psychological or methodological natural kinds the framework invokes does not itself presuppose IBE-style inferences? Third, might an alternative package-level defense of scientific realism, such as a Boyd-style 'philosophical package', secure the needed metaphysical and epistemological commitments without invoking reliabilism, and if so, how would it avoid the regress or question-begging that motivated the critique? These are primarily philosophical questions, though the second has an empirical dimension through the historical track records of scientific inference. A resolution of the open problem would accordingly be either constructive — an account, framework, or package of the corresponding form — or negative, showing that every candidate defense of these forms is unavoidably question-begging.

Problem Statement

As formulated in the closing questions of Newman (2010), and subject to their common objective — securing, or proving unobtainable, a non-question-begging defense of the No-Miracles Argument: (1) Develop, or show impossible, a naturalistic, non-question-begging account that individuates inferential process-types, including IBE, so as to avoid both the single-case and no-distinction aspects of the Generality Problem, without relying on the assumption that IBE is a unique natural-kind rule. (2) Exhibit, or show impossible, a defensible metaphysical or methodological framework in which IBE possesses a determinate, time-stable reliability (or a principled context-sensitivity), where the independent warrant for the psychological or methodological natural kinds invoked does not itself presuppose IBE-style inferences — that is, no circular appeal to the very inferential practices whose reliability is in question. (3) Determine whether an alternative package-level defense of scientific realism, for example a Boyd-style 'philosophical package', can secure the needed metaphysical and epistemological commitments without invoking reliabilism, and if so, how it avoids the regress or question-begging that motivated the critique. The three questions jointly delimit the space of prospective non-circular defenses of the NMA: they are independently posed alternatives, so that answering any one of (1)-(3) constructively — exhibiting such an account, framework, or package — would settle the problem, as would a negative answer that rigorously closes all three by showing that every candidate defense of these forms is unavoidably question-begging. A negative result that closes only one of (1)-(3) — for instance, showing that no naturalistic, non-question-begging individuation of inferential process-types exists while leaving (2) and (3) open — does not establish that no non-circular defense of the NMA is obtainable.

References

  1. Newman, Mark (2010). The No-Miracles Argument, reliabilism, and a methodological version of the generality problem. Synthese 177(1):111-138. DOI: 10.1007/s11229-009-9642-5. https://doi.org/10.1007/s11229-009-9642-5
  2. Putnam, Hilary (1975). What is mathematical truth? Historia Mathematica 2(4):529-533. DOI: 10.1016/0315-0860(75)90116-0. https://doi.org/10.1016/0315-0860(75)90116-0
  3. Conee, Earl and Feldman, Richard (1998). The Generality Problem for Reliabilism. Philosophical Studies 89(1):1-29. DOI: 10.1023/A:1004243308503. https://doi.org/10.1023/A:1004243308503
  4. Boyd, Richard N. (1984). The Current Status of Scientific Realism. In J. Leplin (ed.), Scientific Realism, pp. 41-82. University of California Press. DOI: 10.1525/9780520337442-004. https://doi.org/10.1525/9780520337442-004
  5. Psillos, Stathis (1999). Scientific Realism: How Science Tracks Truth. Routledge (the book DOI records the 2005 ebook edition). DOI: 10.4324/9780203979648. https://doi.org/10.4324/9780203979648
  6. Mizrahi, Moti (2012). Why the ultimate argument for scientific realism ultimately fails. Studies in History and Philosophy of Science Part A 43(1):132-138. DOI: 10.1016/j.shpsa.2011.11.001. https://doi.org/10.1016/j.shpsa.2011.11.001
  7. Lyons, Jack C. (2019). Algorithm and Parameters: Solving the Generality Problem for Reliabilism. The Philosophical Review 128(4):463-509. DOI: 10.1215/00318108-7697876. https://doi.org/10.1215/00318108-7697876
  8. Goldman, Alvin (2021). A Different Solution to the Generality Problem for Process Reliabilism. Philosophical Topics 49(2):105-111. DOI: 10.5840/philtopics202149217. https://doi.org/10.5840/philtopics202149217
  9. Weatherson, Brian (2012). The Temporal Generality Problem. Logos & Episteme 3(1):117-122. DOI: 10.5840/logos-episteme20123153. https://doi.org/10.5840/logos-episteme20123153
  10. Rodríguez, Iván (2026). The generality problem for process reliabilism, the sensitivity theory and the safety theory. Synthese 207(5), article 205. DOI: 10.1007/s11229-026-05585-y. https://doi.org/10.1007/s11229-026-05585-y
  11. Rowbottom, Darrell P. (2021). A methodological argument against scientific realism. Synthese 198(3):2153-2167 (first online 2019). DOI: 10.1007/s11229-019-02197-7. https://doi.org/10.1007/s11229-019-02197-7
  12. Golemon, Luke and Graber, Abraham (2023). A deductive variation on the no miracles argument. Synthese 201(3), article 81. DOI: 10.1007/s11229-023-04078-6. https://doi.org/10.1007/s11229-023-04078-6
  13. Climenhaga, Nevin (2017). Inference to the Best Explanation Made Incoherent. Journal of Philosophy 114(5):251-273. DOI: 10.5840/jphil2017114519. https://doi.org/10.5840/jphil2017114519
  14. Gubelmann, Reto (2019). Scientifically Defending Realism I: Psillos' Holistic Approach. In A Science-Based Critique of Epistemological Naturalism in Quine's Tradition, pp. 205-235. Springer. DOI: 10.1007/978-3-030-24524-5_8. https://doi.org/10.1007/978-3-030-24524-5_8
  15. Kahn, Samuel (2023). Chess and Antirealism. Asian Journal of Philosophy 2(2), article 76. DOI: 10.1007/s44204-023-00118-7. https://doi.org/10.1007/s44204-023-00118-7
  16. Erdenk, Emre Arda (2015). Two Tokens of the Inference to the Best Explanation: No-Miracle Argument and the Selectionist Explanation. Beytulhikme: An International Journal of Philosophy 5(1):31. DOI: 10.18491/bijop.59053. https://doi.org/10.18491/bijop.59053