By Brian Hood

Brian Hood, University of Florida

Answering Some Objections to Scientific Realism[1], Preliminaries

This paper is a defense of Putnam’s explanationist argument for scientific realism. Some have claimed that the explanationist defense has dubious hidden premises, e.g., science has always been “aiming for the truth”. After pointing out problems for metaphysical realism, I consider internal realism since some suggested that to be a scientific realist one must be either a metaphysical realist or an internal realist. Metaphysical realism’s causal theory of reference vitiates metaphysical realism. Internal realism’s metaphysical and epistemological commitments render it unacceptable; hence, it is no alternative to metaphysical realism. Lastly, Arthur Fine argues that the explanationist defense of realism is circular—I argue that his worries need not concern the scientific realist.

The Explanationist Defense of Scientific Realism

Putnam’s metaphysical realism (MR) is composed of the conjunction of three theses:

  1. There is a world existing independently of our minds.
  2. The truth-values of our statements, when their meanings are fixed, are

determined entirely by how the world is.

  1. The extensions of natural kind terms in our language are not determined by the

intensions of those words, but primarily by a direct connection between the words and their references via some sort of causal contact of the word users with the objects in the extensions.[2]

Internal Realism (IR) is a thesis stating, roughly, that science may converge to some ultimately true theory, but that truth amounts to warrantable assertability.

For present purposes, scientific realism (SR) is the thesis that our hypotheses or theories that are well confirmed are at least approximately true and that science is an epistemically progressive enterprise. Whether or not SR entails MR or IR under the formulations given is controversial. Some have argued that proponents of SR must endorse one or the other.[3]Both MR and IR are problematic, the former in virtue of (3) and the latter in virtue of its epistemic notion of truth, and that neither, as stated, are acceptable realisms.

Hilary Putnam has claimed that SR is the only philosophy of science that does not make the success of science a miracle; there would be no way to explain success if the explanatory mechanisms posited by a theory were unrepresentative of the way the world actually operates.[4]Boyd has argued along similar lines.[5]If using a certain experimental method, including the operation of experimental apparatus, depends on a theory that posits unobservables, the best explanation for the success of this method is that those unobservables exist.

Hidden Premises and Problems for Scientific Realism

Some have suggested that for the Explanationist Defense of Realism (EDR) to work, one has to assume that (i) science in general, and at any time of history, is aiming at obtaining truth alone, and (ii)science is not prevented in any way from achieving that aim.[6]Otherwise, the measure of success may be different from truth attaining or what success measures, though understood as truth attaining, may be something else.

According to (i), if at any time science has not been aiming for truth, then success fails to be probative for the truth of scientific theories, since the success of non-truth-seeking science would be measured in terms of utility, psychological satisfaction, or whatever the aim of ideal science during that period of non-truth-seeking scientific activity. In order for the explanationist defense of realism to work, one needs to argue that science has always aimed for truth.

The challenge suggested by (ii) is this: given the vastness of the universe and our access to a relatively miniscule portion of it, how do we know that science is approaching truth rather than diverging from it?[7]In order for the explanationist response to be a cogent defense of realism, one must assume that science has a sense of our position with respect to the rest of the universe so that one could gain a sense of whether the track that science is on is truth-conducive. So, we can imagine ancient Egyptian flat-earthers running the explanationist defense of realism and inferring that their belief system is true or is approaching truth. After all, flat-earth science was probably the most predictively successful theory available to ancient Egyptians.

Others have claimed that there is no stable middle ground between IR and MR.[8]Proponents of IR are skeptical of metaphysical claims regarding the correspondence between true theories and the world. Internal realists are dubious of non-epistemic notions of truth. The worry of the internal realist is that no matter how much our beliefs are justified, there is always a possibility that our best theories are radically unrepresentative of the world. That is to say, according to the internal realist, the mere possibility that our theories fail to represent the structure and mechanics of the universe should make us wary of metaphysical notions of truth. The concern is that it is possible that all claims about theoretical entities may fail to refer, though our theories are empirically adequate and predictively successful, so we should abandon the correspondence (or metaphysical) theory of truth. The locus of this worry is belief in a mind-world lacuna, which cannot be conclusively closed by any inductive procedure, for any such procedure would beg the question against Hume.[9]EitherweembraceMRanddefendtheclaimthatthereisametaphysicalconnection between the world and our scientific theories about it, or we take the metaphysically frugal route and radically divorce truth as correspondence from our assertions about the world, although we still may be justified in believing our best theories.

Fine objects that the explanationist defense of scientific realism is circular and, therefore, unavailing to the realist. This is the last objection to realism addressed.

In Defense of Scientific Realism

In evaluating (i) and its bearing on the explanationist defense of realism (EDR), first it must be ascertained whether (i) is actually necessary for EDR. This is far from clear as EDR is silent on the aims of scientific enterprise by both Putnam’s and Boyd’s lights. The claim that if science ever did not aim for truth, then success couldnever be probative for a theory is a dramatic overstatement. Moreover, it begs the question against EDR.

A better formulation of the above worry is “if science did not aim for truth, the success would not be probative for the theory.” Related to this worry is a concern voiced in (ii): the measure of success may be different from truth attaining or what success measures, though understood as truth attaining, may be something else.

Antirealism assumes that SR is a thesis about aims.[10]Why is this? Realists reject such formulations of their position. The answer to this question not only provides a response to the objection under consideration, but it will reveal an overlooked feature of the realist/antirealist dialectic that is responsible for much of the debate.

Granted, SR must suppose that scientific objectives are epistemic if success is to be evidence for SR. But this is different than what the antirealist is claiming. The antirealist presumes that philosophers must defer to science for success assessments and it is this unwarranted presumption that sustains the antirealist’s mistaken characterization of SR in terms of the “aims of science.” Our hypothetical, purely pragmatic science may be remarkably successful, but this (pragmatic) success is not evidence for SR if scientific methods are not truth tracking. Deferring to science for an assessment of success would lead the realist astray, for the type of success at issue is indifferent to the truth of SR. And presuming that this is what the realist does when evaluating success of science as evidence for SR unfairly portrays SR as naïve and makes SR seem more vulnerable than it actually is.

Realism is inferred from the success of science, not as assessed by science or thereto deferring for such an assessment, but by philosophers with theories about what kind of success is probative for a theory, such as scientific realists (as SR, too, is an empirical hypothesis). One such proposal is novel predictive success as a kind of success warranting the imputation of some measure of truth to a theory.[11]It is in cum bent upon such theories to give an account of success so that it can count as evidence for a theory; once such an account is given, the success of that theory can then be considered evidence for SR. Many of the issues in the realism debate are instances of philosophers talking past one another on issues of success and aims. By exposing the antirealists’ presupposition, I hope to make clear why antirealists repeatedly formulate SR in a way that the scientific realist finds objectionable. Then, the realist may diffuse many antirealist arguments by addressing the dubious presuppositions upon which the antirealists’ aberrant formulation of realism depends.

What if science is not aimed at pursuing truth alone? Is this a challenge to SR? Pragmatic success does not preclude epistemic success. One should expect pragmatic success to be a consequence of epistemic success, i.e., having theories that increase our knowledge of the phenomena under investigation. Again, science could be aiming for X (pick your favorite non-truth- related virtue), but if it saves the phenomena and if it has predictive power and success (in the epistemic sense mentioned above), then it seems that, without recourse to SR, success is a mystery. Furthermore, Putnam’s and Boyd’s arguments do not infer the truth of SR from the claims that our current theories are predictively successful and that science aims for truth; rather, the truth of SR is inferred from its purportedly unique ability to explain the success of science.

An appropriate challenge to EDR must fault abduction (as Fine does) or suggest an alternative explanation of the success of science, thereby demonstrating that SR does not have a monopoly on explaining the success of science.

Hence, it seems that EDR does not presuppose (i). Besides attributing to SR a rather naïve view of science, (i) invites difficulties insofar as it suggests a possible difference between the property in terms of which science is assessed for success and the property actually being measured in a success assessment, a spurious worry.

Moreover, whether science has pursued truth in all its history has no bearing on EDR. One could defend EDR against the present complaint by insisting that any practice that does not aim to discover truths is not scientific, thus making the pursuit of truth a necessary condition for qualifying as a science. Such a response would be unsuccessful. However, another criticism to be made of the idea that EDR presupposes (i) is that science does not aim for truth alone and SR does not purport that it does aim solely for truth. Pragmatic virtues do seem to play some role in science; however, they are likely to be subservient to, and to have their provenance in, epistemic concerns.

Turning to (ii), does EDR presuppose that science is not prevented in any way from attaining the goal of truth? My critique of (i) somewhat absolves me from the responsibility of defending my negative answer to this question. The aims of science are irrelevant to the success of EDR and it is a mistake to couch SR in terms of aims at all. This said, I do think that the possibility that science is departing from truth as science “progresses” warrants response. This is a concern related to (ii), but which makes no appeal to the aims of science.

The possibility that science is diverging from truth is a remarkably skeptical concern, the impetus of which is most likely the finite nature of man and our limited access to the world. Imagine our Egyptian flat-earth theorists confronted with EDR. For the flat-earth theorist, EDR would obtain the result that the flat-earth theory is true, or at least partially so.

However, if this possibility gives grounds on which to doubt SR, then, too, we must take as a serious threat to our justification for believing that there exists an external world the possibility that we are all Berkelean souls with external-world-like experiences. It seems immoderately skeptical to take as a serious threat to SR the possibility that our best extant theory is actually less representative of the world than its distant ancestors merely in virtue of the ability of a false theory to accurately predict.[12]

Must one accept either MR or IR to be a scientific realist? MR as formulated in this paper is objectionable; IR, too, is problematic and unacceptable. That one must endorse either MR or IR to be considered a scientific realist seems questionable. The conclusion contains a modest suggestion that cannot appropriately be said to fall under MR or IR, though it does presuppose that truth is metaphysical. It does seem that one can be a scientific realist without endorsing (3), namely the claim that the extensions of natural kind terms in our language are not determined by the intensions of those words, but primarily by a direct connection between the words and their references via some sort of causal contact of the word users with the objects in the extensions. However, traditionally scientific realists have tended to fall into one of these two camps. Certainly, with respect to their metaphysical and epistemic commitments, MR and IR represent the extremes of the positions that philosophers have held in the realist literature. The essential difference between the two positions is the notions of truth to which they subscribe. MR is committed to a metaphysical correspondence theory of truth whereas IR, motivated by skepticism, eschews metaphysics, opting for an epistemic notion of truth.

The most dubious tenet of MR seems to be (3), that is, the Causal Theory of Reference (CTR).[13]CTRstatesthattermsreferinvirtueofacausalinteractionbetweenthelanguageusersand the objects purportedly in the extension of the terms used. If this causal connection cannot be spelled out, then it seems that we must divorce confirmation from truth, for our natural kind terms may fail to pick out natural kinds.

Examples of meaningful, scientific terms, the referents of which are fixed abductively, i.e., non-causally, would be counterexamples to CTR. Interestingly, we are realists about many entities to which we have no ostensible access. Examples abound; consider Pangea (or any of the other historical geological periods of the earth), the Earth’s molten core, dinosaurs (and other extinct species), and galaxies. Moreover, scientific realist Jarrett Leplin writes,

the most confident postulations at the frontiers of science are explicitly abductive: neutron stars, offered as the only possible source of certain radio signals; the missing mass required for a cyclical universe, offered as the only possible source of forces holding together hydrogen clouds in rapid rotation or of violations of Kepler’s law by fast stars at the periphery of the galaxy; black holes required by a sufficient concentration of cold matter, offered as the necessary companions to such stars as the supergiant of Cygnus, whose motions are otherwise unaccountable.[14]

One, and perhaps the only, line of defense against the objections to CTR proffered in this paper would be to construe the causal connection between terms and their referents such that a causal link does exist between the terms in my counterexamples and the posits purportedly in their extensions. However, I am not optimistic about such an approach. Gerrymandering the causal connection cannot work either, for then CTR stands guilty of being ad hoc.

There are other worries associated with CTR. For one, CTR requires us to know the details of how the initial causal link is formed between the entities and terms. The answer to this question is likely to be that the initial referential use of a term, ‘p,’ fixes the reference of ‘p.’ This is not a satisfactory response. It may be impossible to know the conditions of ‘p’’s initial referential use and, hence, we cannot know if our utterance of ‘p’ is referentially successful.

MR has been shown to be problematic, if not false, in virtue of a flawed theory of reference. According to the challenge being entertained presently, there is no recourse but to IR. But how secure is IR?

IR rejects the correspondence theory of truth, contending that there is no non-question begging way to close the mind-world gap, dissociating confirmation and truth (in any metaphysical sense). IR seems like the natural place for a scientific realist impressed by (ii) to retreat, since granting (ii) is tantamount to rejecting MR. Confirmation warrants only the assertion of scientific claims. We call a theory “true” just in case its assertion is sufficiently warranted, that is, justified. According to IR, our best theories are “true” (in a non-correspondence sense of true). Bas van Fraassen’s constructive empiricism and IR are similar in that they both excise metaphysical truth from our evaluations of our scientific explanations and theories. Internal realists would claim that our best scientific theories are “true” (i.e., they enjoy warrantable assertability) because they are empirically adequate, i.e., they save the phenomena, but this seems to get things backwards. Putnam rejects the notion of metaphysical truth altogether, claiming that it is unintelligible.[15]So, while van Fraassen, out of epistemic caution, invokes empirical adequacy where he believes there to be no grounds for imputations of metaphysical truth to theories, Putnam denies that the notion of truth as correspondence even makes sense. He eliminates truth as correspondence, supplanting it with truth as “warrantable assertability,” thus undercutting the motivation for invoking empirical adequacy in the first place. However, without empirical adequacy, there are no grounds for attributions of the internal realist’s “truth.” The internal realist needs some explanation of the justification for making truth attributions to theories, but without empirical adequacy or metaphysical truth, it is not clear which way to turn.

This criticism also demonstrates that a non-epistemic notion of truth is unavailable to van Fraassen (who endorses a metaphysical notion of truth anyway). However, the scientific realist wants to say that her theory is empirically adequate because it is true in a non-epistemic sense. As for the constructive empiricist who abstains from making attributions of truth to theories, satisfied with the notion of empirical adequacy, suffice it to say empirical adequacy is no surrogate for truth.[16]

It is an undesirable consequence of IR that it makes tables, chairs, and cats (as well as electrons and gluons) mental constructs. For even non-scientific claims are relativized to conceptual schemes under IR. IR’s metaphysically anemic notion of truth implies ontological relativism and idealism. If we relativize truth to conceptual schemes, then claims such as “there are electrons” will be judged true only within the context of a theory. This relativism precludes any rational theory adjudication, since, under IR, there are no theory-independent grounds on which to assess the truth of two competing theories. Hence, we have incommensurability. In sum, truth is not epistemic justification.

Arthur Fine’s Objection

Fine’s argues that EDR should be rejected, as it is question begging.[17]If abduction is a legitimate form of ampliative reasoning, then the debate is thus adjudicated in favor of the realist, for it is the justification of abductively inferred claims concerning theoretical entities that is the crux of the debate.[18]

Fine’s argument against EDR points out a potentially vicious circularity in the realist’s reasoning; however, it does not follow that realism is false and it does not follow that an abductive argument for realism like EDR is corrupt. Fine’s argument simply points out that the legitimacy of abduction is not to be presupposed.

It is perplexing that an advocate of enumerative induction, such as Fine, would find fault with abduction. Induction is subject to well-known paradoxes such as Goodman’s grue paradox, the resolutions of which depend on explanatory reasoning.[19]Mere concomitances of properties do not justify the projection of such co-occurrences. That my tires have not failed since I bought my car does not license the inference that they will continue to operate properly and that I have lived everyday of my life up to now does not justify the belief that I will live to see tomorrow or that I will live forever. Enumerative induction and explanatory reasoning go hand in hand and mutually support one another. Our everyday inferential practices seem to vindicate this claim, as we rarely induce without seeking some explanation for the regularity from which we are inducing. It is hardly clear that enumerative inductive reasoning is more basic or methodologically sound than abductive reasoning. We reject abduction only to be faced again with vexing paradoxes of induction. We should be prepared to induce only so long as we are prepared to abduce, and given Fine’s endorsement of Laudan’s historical argument against realism, it is clear that he is more than willing to induce, hence one may argue that his complaint against EDR is unavailing.[20]


I have attempted to defend SR against various objections while not making any arguments for SR directly. The formulation of SR under consideration may not be a sustainable position. However, I do think that some form of SR is correct. Certainly, absent from the formulation of SR discussed in this paper is an acceptable theory of reference; I take it that recourse to IR is patently otiose; therefore realists should focus their efforts on developing and refining MR. More reasonable would be a minimal epistemic realism claiming no more than that there are empirical conditions such that were they to obtain, we would be in a position to assert with justification of some scientific theory positing unobservable empirical structures that it is true, or approximately so.


[1]I would like to thank two anonymous reviewers for their helpful critical comments.

[2]This is not my formulation of MR, but rather that of the opponent to scientific realism whom I am addressing in section II, Dr. Chuang Liu.

[3]In particular, Chuang Liu has argued this point.

[4]Hilary Putnam, “What is Realism?” Scientific Realism, ed. J. Leplin (Berkeley: U of California P, 1984): 140-153.

[5]Richard Boyd, “Confirmation, Semantics, and the Interpretation of Scientific Theories,”The Philosophy ofScience, eds. R. Boyd, P. Gasper and J.D. Trout (Cambridge: MIT Press, 1991): 3-35.

[6]This criticism is Dr. Chuang Liu’s.

[7]I mean by “approaching the truth” an overall increase in our best theories’ representational accuracy. Simply put, our best theories now are more verisimilar that their predecessors. The thesis according to which science is getting ever better at accurately representing the world has been called “convergent realism”; that is, over time, our best theories can be described as converging on the truth.

[8]I attribute this claim to Chuang Liu.

[9]See Hume, David, A Treatise of Human Nature (London: Penguin Books, 1984). This worry’s impetus is likely Hume’s concerns about induction. To attempt to close the mind-world gap via some inductive procedure would be to presuppose the legitimacy of induction, i.e., the very thing in question. It is an inductive skepticism that gives this objection its force.

[10]Van Fraassen, in particular, is guilty of making this assumption when he gives the following formulation of SR: “Science aims to give us, in its theories, a literally true story of what the world is like; and acceptance of a scientific theory involves the belief that it is true.” See Bas van Fraassen, The Scientific Image (Oxford: Clarendon Press, 1980): 8.

[11]Jarrett Leplin is probably the foremost proponent of this approach. See J. Leplin, A Novel Defense of Scientific Realism (New York: Oxford, 1997).

[12]The inspiration for employing this “Moorean” strategy comes from a response that Jarrett Leplin makes to the objection to EDR that the world is ultimately unintelligible. See J. Leplin, “Truth and Scientific Progress,” Scientific Realism, ed. J. Leplin (Berkeley: U of California P, 1984): 212.


[13]See Saul Kripke, Naming and Necessity (Oxford: Blackwell, 1980).


[14]J. Leplin, “Methodological Realism and Scientific Rationality,” Philosophy of Science 53 (1986): 49.


[15]It should be noted, that this is Putnam’s position in his Reason, Truth and History (Cambridge: Cambridge UP, 1981).


[16]I have argued this elsewhere. See my “Van Fraassen, Explanation, and Truth” (unpublished manuscript). It should be added that it is no victory for realism that internal realists have redefined truth so as to evade the problem of attaining metaphysical truth. Such a move is ad hoc.

[17]A. Fine, “The Natural Ontological Attitude,” Scientific Realism, ed. J. Leplin, (Berkeley: U of California P, 1984): 83-107.

[18]Abduction, or inference to the best explanation, is an ampliative method of reasoning whereby we infer a hypothesis H from a set of data O on the grounds that H best explains O. For example, I observe that dropped objects fall down, toward the center of the Earth, instead of up, so I infer that there is a gravitational force responsible for the trajectory of falling bodies. Reasoning of this type takes the form “If P, then Q, Q, therefore P”—if there is gravitational force, then falling bodies behave in a certain manner, i.e., they fall toward the center of the Earth. No other hypothesis better explains this phenomenon, so we infer that there is a gravitational force.

[19]See Nelson Goodman, Fact, Fiction, and Forecast (Indianapolis: Bobbs-Merrill, 1965).

[20]Larry Laudan argues that success is not evidence for truth. To support this claim, he offers a list of successful scientific theories that current science rejects, or whose central theoretical terms fail to refer in contemporary science. See Larry Laudan, “A Confutation of Convergent Realism,” Philosophy of Science 48 (1981): 19-49.

Works Cited

Boyd, R. “Confirmation, Semantics, and the Interpretation of Scientific Theories.” The Philosophy of Science. Eds. R. Boyd, P. Gasper and J.D. Trout. Cambridge: MIT Press, 1991. 3-35.

Fine, A. “The Natural Ontological Attitude.” Scientific Realism. Ed. J. Leplin Berkeley: U of California P, 1984. 83-107.

Gasper, P. and J.D. Trout, eds. The Philosophy ofScience. Cambridge: MIT Press, 1991. Goodman, N. Fact, Fiction, and Forecast. Indianapolis: Bobbs-Merrill, 1965.
Hood, B. “Van Fraassen, Explanation, and Truth” (unpublished manuscript). Hume, D. A Treatise of Human Nature. London: Penguin Books, 1984.

Kripke, S. Naming and Necessity. Oxford: Blackwell, 1980.

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

Leplin, J., ed. Scientific Realism. Berkeley: U of California P, 1984.

“Methodological Realism and Scientific Rationality.” Philosophy of Science 53 (1986): 31-51.

A Novel Defense ofScientific Realism. Oxford: Oxford UP, 1997.

“Truth and Scientific Progress.” Scientific Realism. Ed. J. Leplin. Berkeley: U of California P, 1984. 193-217.

Putnam, H. Reason, Truth and History. Cambridge: Cambridge UP, 1981.

“What is Realism?” Scientific Realism. Ed. J. Leplin. Berkeley: U of California P, 1984. 140- 153.

Van Fraassen, B. The Scientific Image. Oxford: Clarendon Press, 1980.



Brian Hood

S. Brian Hood received his BA in Philosophy from the University of North Carolina at Greensboro where, under the directions of Jarrett Leplin, he completed his Honors thesis, “The Limitations of Kitcher’s Theory of Explanation as Propositional Systematization.” Following two years of graduate work at the University of Florida with Chuang Liu serving as his principle advisor, Brian moved to Bloomington, Indiana, where he is currently pursuing graduate studies at Indiana University in the Department of History and Philosophy of Science.