Home » Library » Modern Library » Michael Martin Naturalism

Michael Martin Naturalism

Justifying Methodological Naturalism (2002)

Michael Martin


In his recent book Tower of Babel Robert T. Pennock argues against the New Creationists for failing to realize that science is committed to Methodological Naturalism (MN) and not Ontological Naturalism (ON).[1] Roughly speaking ON is the view that only natural processes or events exist. It maintains that insofar as God, angels, the Devil, ghosts, and other such entities are supernatural, they do not exist. MN is a much weaker position. It does not deny the existence of supernatural entities per se. It simply assumes for the purpose of inquiry that they do not exist. It goes on the assumption that in the context of inquiry only natural processes and events exist.

According to Pennock, MN has the advantage of ON without its drawbacks.[2] MN would prevent Creationists of all stripes from appealing to divine intervention as an alternative to Darwinian evolution. It would also prevent religious believers from appealing to divine intervention in other areas of science besides the evolution of life. Pennock maintains that a few Christian fundamentalists who accept a literal interpretation of the Biblical story of the Tower of Babel have made the same sort of argument for the evolution of language that Creationists make with respect to biological evolution. The use of MN would rule out such explanations without forbidding religious belief in nonscientific contexts. As Pennock has said, religious believers have nothing to fear from MN since, in contrast to ON, it is compatible with belief in God so long as science is God-free.

Although I am an Ontological Naturalist I will not defend this position in this paper. I am interested in considering the following question: If you reject ON and yet believe that MN is an appropriate stance in the context of science, how can MN be justified? I will not be concerned about whether the rejection of ON is justified. Lots of people reject ON and accept MN. I will consider if they have any case for accepting MN. I will consider five rationales for MN: testability, the use of laws in explanations, fruitfulness, the promotion of agreement and cooperation, and the avoidance of blocked inquiry. Other rationales are no doubt possible but I will not consider them here. I will argue that the first three rationales are unacceptable, and that the fourth has serious problems that may also make it unacceptable. The last rationale offers promise, however, and should be adopted.

I will proceed by formulating MN as the methodological rule:

MN = In the context of science only use natural explanations!

As such it is neither true nor false. It can, however, be evaluated by pragmatic criteria. Thus I will ask here what the consequences would be of following MN in contrast to an alternative rule of Methodological Supernaturalism such as:

MS = In the context of science assume that God is the direct or indirect explanation of everything!

Now, it should be noted that although MN excludes explanations such as “Mary recovered because God intervened,” “This is the result of God’s punishment,” and “This happened because of God’s will,” MN does not say that all phenomena must be explained in naturalistic terms. Rather, it says that in the context of science anything that is explainable must be explained naturalistically. Although historians who forego supernatural explanations of both particular historical events and the general course of history can be considered Methodological Naturalists, as can Biblical scholars who reject supernatural explanations and strive instead to interpret Scripture in naturalistic terms, MN allows nonnaturalistic explanations to be used in nonscientific contexts. It also allows that some phenomena may be explainable in naturalistic terms, yet are not worthy of such explanations, and that others are incapable of any explanation. Whether there are in fact such “brute”–i.e., unexplainable–facts is an open question. Whether it is a good policy to suppose that there are will be considered later.


One plausible rationale for MN is testability. This requirement is often derived from a definition of science that maintains that scientific claims are testable whereas religious claims are not. Since naturalistic entities are capable of scientific investigation but supernatural entities are not, naturalistic explanations are testable. So, following MN fulfills one of the defining attributes of doing science: using testable explanations.

Stated negatively this idea can be formulated by the following methodological rule:

MR1 = Do not allow empirically untestable claims in science!

But MR1 can be objected to on the grounds that the concept of empirical testability is unclear, and that when philosophers have given it a precise definition this has either been so broad as to allow everything to be testable or so narrow as to exclude parts of science.

One of the best-known examples of this problem is Popper’s attempt to demarcate science from nonscience in terms of falsifiability.[3] Rejecting inductive confirmability as a criterion of science, Popper argues that a set of statements is scientific when it is falsifiable; that is, when it contradicts a potential falsifier. For example, the set of statements “All copper conducts electricity” and “Object X is a piece of copper” is falsifiable since this set is inconsistent with the potential falsifier “Object X does not conduct electricity.”

Two problems have plagued Popper’s attempt at demarcation, however.[4] One is that if Theory T is falsifiable, so is the conjunction of T with any statement at all including “God exists.” But then the conjunctive statement consisting of T and “God exists” becomes part of science. Moreover, unrestricted existential statements such as “There is a comet” are not falsifiable and are not scientific.

Now one might suppose that a looser formulation of Popper’s falsifiability criterion might capture its spirit and have fewer problems. One such candidate is that a statement is testable if and only if there is at least some observational statement that could count against it. In this formulation what exactly is meant by “count against it” is not formally explicated but is understood intuitively. Presumably this intuition would be enough to rule out the problems with Popper’s more restricted view. For example, one could imagine evidence that could count against “There is a comet,” for example, the failure of astronomers to observe any comet over a 2000-year period. Moreover, if evidence counts against some theory T, it would not follow that this evidence would count against any conjunction formed by T and some arbitrary statement. However, this looser formulation of the falsifiability criterion does not exclude all theological statements from science, for some theological claims are also testable in this intuitive sense.

Consider the claim that Mr. Smith’s cancer was cured by a miracle; that is, by a direct intervention of the supernatural in the natural course of events. Surely, evidence that Mr. Smith’s cure was brought about by a new medicine he received would tend to count against this miracle explanation. Moreover, even the hypothesis that God exists–at least on some conceptions of God–is testable in this intuitive sense. After all, the existence of evil has been thought by some to be evidence against the existence of an all-good, all-powerful, and all knowing God; indeed, Ted Drange has argued that the existence of many nonbelievers counts at least against the hypothesis of the existence of an evangelical Christian God.[5] So it would seem that the criterion of testability–at least when formulated along these lines–cannot keep theological theories out of science and cannot provide a rationale for MN.

Explanation by Laws

Another rationale for MN is explanation in terms of laws. This requirement is also often derived from a definition of science. It is argued by definition: scientific explanations are in terms of laws while religious explanations are not. Thus, Michael Ruse writes that science deals with “the natural, repeatable, which is governed by law.”[6] Since naturalistic explanations are closely tied to science while supernatural theories are not, naturalistic explanations are explanations by laws. So, following MN fulfills one of the defining attributes of doing science: explanation by law.

Stated negatively this idea can be formulated in terms of the following methodological rule:

MR2 = Do not allow explanations in science that are not at least implicitly based on laws!

However, once again there are objections that can be raised regarding this criterion.[7]

First, in recent years the philosopher of science Bas van Fraassen has maintained in an extended argument that there are no natural laws.[8] To be sure, he agrees that there are regularities. But on most accounts laws are not just regularities but have at least two other attributes: they are supposed to ground and explain regularities, and they are supposed to be necessary, albeit in a weaker sense than logical necessity. Directing his arguments to the deep problems inherent in both of these claims, Van Fraassen has attempted to show that no adequate account of the concept of law has been given but that science can proceed without recourse to laws. I will take no position here on whether his critique is valid except to say that, unless his argument is rebutted, one should hesitate to accept the use of laws as a defining property of science.

The second objection to MR2 is that laws are normally tied to the repeatable. But, arguably, the Big Bang is or at least could be nonrepeatable, and this hardly seems grounds for excluding its explanation from science.[9]

A third objection to MR2 is that, in the social sciences, the idea that laws are needed to explain social phenomena is extremely controversial. Indeed, whether social scientific explanations are based on laws was one of the main points of contention within the philosophy of social sciences in the 1960s and 1970s. Lining up on the side favoring nomological explanations were such philosophers as Carl Hempel and Ernest Nagel. Denying the need for laws were such equally distinguished philosophers as Michael Scriven and William Dray. And this debate continues today although in a muted form.[10]

Finally, even if one waives these problems, the possibility remains of understanding miracles in terms of laws. In Atheism: A Philosophical Justification I provisionally defined a miracle as an event directly brought about by the exercise of a supernatural power.[11] This definition is compatible with a miracle’s violating no law. A supernatural power is one that is possessed by supernatural beings (e.g., gods, angels, devils) and is markedly superior to those powers possessed by humans. If such entities exist, one could imagine these powers being governed by laws although not those investigated by science. Construed in this way science could not be defined in terms of the use of laws per se.


Another possible rationale is the comparative fruitfulness of MN over MS. It could be argued that comparatively speaking following MN has produced more confirmed theories than following MS. This claim would be supported by appeal to the history of science. For example, in the context of the evolution of biological life it might be argued that MN has produced the well-confirmed edifice of Darwinian evolution while MS has produced the factually dubious theories of Creationist Science.

However, this is to attribute too much responsibility to MN per se for scientific success. After all, MN is compatible with naturalistic theories that are wildly mistaken. Suppose, for example, that the scientists that following MN in the context of biological evolution do not assume that evolution is guided by natural selection. Although no supernatural explanations are tolerated, they suppose that evolution is directed by super intelligent aliens. On the other hand, let us imagine that scientists that follow MS are committed to Darwinian evolution except in this respect. They believe that unless science is able to explain the origin of life naturalistically after 500 years of study and research then it should be explained by a postulating the direct action of God. With respect to everything else they believe God indirectly works through natural processes. So strictly speaking these scientists are followers of MS.

Now it seems clear that following MS would be more fruitful than following MN given these circumstances. The moral to draw is that the fruitfulness of MN and MS is partly a function of the substantive theories utilized by the people who follow them and not merely a function of MN and MS per se.[12]

Furthering Agreement

Another possible rationale for MN is that following MN would further agreement and cooperation in science and for this reason should be used. For it could be argued that allowing supernatural explanations would cause great controversy and lack of cooperation within science and that for this reason they should be excluded.[13] This rationale harks back to an argument used by Pierre Duhem to exclude metaphysics from science.[14] According to Duhem, a “physical theory reputed to be satisfactory with the sectarians of one metaphysical theory will be rejected by the partisans of another school.”[15] In order to avoid this impasse Duhem argued that metaphysical doctrines should be excluded from science. A similar argument could be used regarding supernatural theories. In order to avoid this sort of divisive controversy supernatural explanations would be excluded from science. Consider then the following methodological rule:

MR3= In the context of science use only basic ontological assumptions and commitments that are agreed to by all scientists!

However, MR3 excludes not just supernatural assumptions and commitments from science but controversial metaphysical ones as well. Following its theories in terms of atoms as well as angels would be excluded from science. Not all scientists have accepted micro entities. Some have regarded them as useful fictions. Indeed, it is interesting to speculate about what metaphysical assumptions there could be that all scientists would agree to. In any case, such a science would be a metaphysically purified one–so purified that it may have very little relevance to what we now understand as science. So at the very least, MR3 has implications that not just religious believers but many others will find hard to accept.

Nevertheless, it is true that a Duhemian science, which embraced MR3 would be able to achieve greater agreement and cooperation and would have fewer undesirable implications than a science that is committed to admittedly partisan religious and metaphysical assumptions specific to different groups. According to what Alvin Plantinga calls Augustinian science, there could not only be a Christian science, as Plantinga desires, but a Jewish science, an Islamic science, a Buddhist science, and a Mormon science. But why stop there? Why not a Scientology science, an Astrology science, and a Voodoo science? After all, Plantinga argues that “a Christian academic and scientific community ought to pursue science in its own way, starting from and taking for granted what we know as Christians.” But Jews, Moslems, Buddhists, Mormons, Scientologists, Astrologists, and Voodooists could assert something similar. Surely this appalling implication is one good reason for embracing a rule similar to MR3 and a Methodological Naturalism purified of metaphysics.

Prevention of Blocked Inquiry

Another possible reason for adopting Methodological Naturalism is that supernatural explanations are “science stoppers.” Plantinga has stated this point as follows:

[B]ut what we want to know in science are answers to questions like “What is this made out of? What is its structure? How does it work?. . . . Claims to the effect that God has done this or that (created life or created human life) directly are in this sense science stoppers. If this claim is true, then presumably we cannot go on to learn something further about how it is done, how the phenomenon in question works; if God did it directly, there will be nothing further to find out. . . . Ascribing something to the direct action of God tends to cut off further inquiry.[16]

This idea can be stated by the methodological rule:

MR4 = Do not use explanations in science that cut off further inquiry!

Curiously, Plantinga does not believe that because appeal to God’s direct actions cuts off further inquiry one should forsake theistic explanations. First of all, he says that just because the claim that God directly caused life is a science stopper, it does not follow that God did not directly create life. Second, he says that it would be worth knowing, if possible, which things God did do directly and that to know this would be “an important part of a serious and profound knowledge of the universe.”[17] Moreover, he seems to believe that such knowledge can be arrived at through empirical investigations. Thus, he argues that there is no reason why a question like “Did God create life directly?’ cannot be answered empirically. A theistic scientist, he says, could reason as follows: after a great deal of study, we cannot see how God created life indirectly; “thus probably he has created it directly.”[18] Third, in any case Plantinga does not think that all theistic descriptions and explanations cut off further inquiry. He argues that what God does not bring about directly, He brings about indirectly. In the indirect cases we can ask, and in principle answer, questions such as: What is this made out of? What is its structure? How does it work?

Despite what Plantinga says, there are serious objections to his arguments. First, his thesis that the claim that God created something directly can be based on empirical investigation is dubious. He assumes without argument that if Christians are not able to understand, for example, how and why life came about in terms of physics and chemistry “after a great deal of study,” then they can infer it was probably the result of the direct action of God. Yet other hypotheses are at least as equally well supported: for example, that scientists have not studied enough or that they need to revise their theories in basic ways. Perhaps Plantinga is assuming some hidden premise to the effect that if there are two possible types of explanations–X and Y–of phenomena P, and if after a great deal of investigation a large sample of explanations of type X have been shown to be inadequate, then it is probable that an explanation of type Y is true. But the truth of this premise will be a function of several factors, including the representative nature of the sample and the type of investigation. In many natural scientific contexts inferences based on this premise are especially risky. Scientific attempts to explain certain phenomena are in their infancy and are increasing in sophistication and power. If we take as our population all past, present, and future scientific theories, then the sample of past and present day scientific theories that attempt to explain life, for example, is hardly a representative sample of the whole. We can make no reliable inferences based on their failure up to the present time about their success or failure in the future.

Second, in any case, Plantinga’s suggestion that claims that appeal to God’s direct action be based on empirical investigations can hardly be taken seriously. He makes it clear that revelation is at the foundation of Christian science. He says:

First, the thought that there is such a person such as God is not, according to Christian theism, a hypothesis postulated to explain something or other, nor is the main reason for believing there is such a person as God the fact that there are phenomena that elude the best efforts of current science. Rather, our knowledge of God comes by way of general revelation, which involves something like Aquinas’s general knowledge of God or Calvin’s sensus divinitatis and also, and more importantly, by the way of God’s special revelation, in the Scripture and through the church. . . .[19]

This passage makes it clear that Christian scientists can decide what is explained by the direct action of God, not by empirical investigation of the kind outlined by Plantinga, but by revelation. Science could be stopped and inquiry blocked without any preliminary empirical investigation. For example, Creationists reject Darwin’s evolutionary theory primarily because it conflicts with their interpretation of what Plantinga calls special revelation. Of course, even with respect to evolutionary theory what phenomena are considered to be brought about directly by God’s actions will vary in terms of the sort of interpretation one gives of Scripture, and this will generate different points where inquiry is blocked.[20] This in turn will bring about different and conflicting Christian sciences based on different interpretations of Scripture.

Third, Plantinga implies that questions asked in science such as, what this is made of, what its structure is, how it works, take us only so far and eventually lead to God’s direct action and the cutting off of inquiry at some later stage of investigation. For example, even if one believes that science explains how life came about in terms of physical-chemical process Z, inquiry would be closed by saying that God directly created Z. As before, whether this would be said would depend on the interpretation of Scripture and would vary from one Christian sect to another.

In contrast to this view, inquiry guided by MR4 is never cut off and no answer is a science stopper. What is the justification for this rule? One might be inclined to argue for MR4 on inductive grounds as follows: we have inductive reasons to suppose that there are adequate naturalistic explanations waiting to be discovered, even though success has so far been elusive. The inductive grounds would be that in many similar seeming cases success was achieved despite many initial failures. Moreover, the number of naturalistically explained cases is increasing while the number of naturalistically unexplained cases is decreasing. Such evidence gives us some justification for supposing that all cases will eventually be explained by science.

However, the main argument for MR4 is based not on induction but on vindication. Quite simply, one can never rule out the logical possibility that there is a scientific explanation of a phenomenon. Following MR4 would not assure us that such an explanation would be discovered even in the long run. But operating with alternative rules could result in the loss of potential scientific explanations if they exist.

For example, suppose one adopted this alternative rule:

MR5 = Do not use explanations in science that cut off further inquiry unless inquiry-blocking explanations are the only plausible ones that can be thought of after twenty years of deliberation and research!

Following MR5 would be unacceptable. Suppose there were an explanation that could only be discovered after more than twenty years of deliberation and research. Following MR5 would prevent one from finding it. Other restricted methodological rules would have similar implications.

Despite the promise of MN4 proving a rationale for MN there is one remaining problem. We must make the assumption that naturalism in the context of science does not block inquiry. But this is not obvious. Certainly MN does not entail MR4. Indeed, there could be naturalistic explanations that are supposed to be brute facts, ultimate principles and the like. Would not these block inquiry just as much as appealing to God’s direct action? They would if they were allowed. But good scientific methodology should never permit anything to be more than a tentative brute fact, or a provisional ultimate principle since we can never be sure that what we think is an ultimate naturalistic explanation really is not.

Thus MR4 could be understood as an independent constraint on MN or MN could be revised to include this constraint. Thus we could characterize Methodological Naturalism by this methodological rule:

MN’ = In the context of science assume that explanations can be explained naturalistically at least in the long run!

MN’ is compatible with scientists acting for long periods of time as if certain explanations are ultimate relative to their evidence and background theories. But such practice would not block inquiry because in contrast to explanations in terms of God, it would be possible for these explanations themselves to be explained as science progresses.


In conclusion, then, it has been argued that the atheistic implications of ON do not seem to make it a suitable basis for science because not all scientists are atheists. However, MN has the advantage of making science God-free and at the same time permitting belief in God in nonscientific contexts. But on what grounds should belief in God be excluded from science once belief in His existence is allowed elsewhere?

Some have argued that belief in God as an explanation conflicts with the scientific criteria of testability and the necessity of nomological explanations. But these criteria are problematic in certain respects. Others have maintained that following MN is more fruitful than following MS. But this puts too much weight on methodological considerations and not enough on the substantive theories used by advocates of MN and MS. It is more plausible to maintain that allowing explanations in terms of supernatural entities would introduce irreconcilable controversy in science and the blockage of inquiry. On the other hand, a metaphysically refined MN that did not allow any brute facts or ultimate explanations would be free from these problems. Although the prospects of a metaphysically uncontroversial science are not encouraging, I see no reason why science cannot be conducted in a way that does not block inquiry and I suggest that MN should be justified in terms of not blocking inquiry.


[1] Robert T. Pennock, Tower of Babel (Cambridge, MA: The MIT Press, 1999).

[2] If an Ontological Naturalist believes in God at all, this entity would have to be a part or the whole of the natural rather than the supernatural order. Such a naturalistic view of god or gods has been advocated. For example, John Dewey, in A Common Faith (New Haven: Yale University Press, 1934), p. 42 defines “God” as “the unity of all ideal ends arousing us to desire and action.” Such a unity is surely explainable by scientific method and is part of the natural order. Auguste Comte, in turn, developed an organized religion completely within a naturalistic framework in which the object of worship was humanity. Comte sometimes referred to humanity as the Supreme Being. (See William Alston, “Natural Reconstructions of Religion,” Encyclopedia of Philosophy, Vol. 7. pp. 145-146; Christopher Kent, “Positivism”, Encyclopedia of Unbelief, vol. 2, pp. 512-518.) Again, such a notion of God or a Supreme Being is compatible with naturalism. Other philosophers, for example Spinoza, identified God with nature itself. Such pantheistic views may well be compatible with naturalism in so far as nature itself is explainable by scientific method. Thus, the acceptance of naturalism does not exclude belief in god or gods interpreted in a nonsupernatural way. However, belief in a god or gods is usually given a supernatural interpretation. Most of the famous atheists of history could well believe that God exists if God only means what Dewey meant by this term. This suggests that great confusion would result by defining “God” along Deweyian or other naturalistic lines.

[3] See Karl Popper, The Logic of Scientific Discovery (New York: Basic Books, 1959).

[4] Israel Scheffler, The Anatomy of Inquiry, (New York: Knopf, 1963) pp. 137-150. See also Adolf Grünbaum, “Is Falsifability the Touchstone of Scientific Rationality? Karl Popper versus Inductivism,” Essays in Memory of Imre Lakatos, ed. R.S. Cohen, P.K. Feyerabend, and M.W. Wartofsky, (Dordrecht, Holland: D. Reidel Publishing Co., 1976), pp. 213-252.

[5] See Theodore M. Drange, Nonbelief and Evil (Amherst, NY: Prometheus Books, 1998).

[6] Michael Ruse, Darwinism Defended, (Reading: Addison-Wesley, 1982). p. 322.

[7] See Alvin Plantinga, “Methodological Naturalism?” Perspectives on Science and Christian Faith, 49. 1997, pp. 145-146.

[8] Bas van Fraassen, Laws and Symmetry (Oxford: University Press, 1989), Chapters 3-5. See also reviews by Michael Tooley, British Journal for the Philosophy of Science, 46, 1995, pp. 280-283 and Adam Morton, Philosophical Review, 102, 1993, pp 408-410.

[9] See Plantinga, “Methodological Naturalism,” p. 146. See also Quentin Smith, “Atheism, Theism and Big Bang Cosmology,” Australasian Journal of Philosophy, 69, 1991, pp. 51-52.

[10] See Lee C. McIntyre, Laws and Explanations in the Social Sciences, (Boulder, CO: Westview Press, 1996).

[11] I am indebted here to Paul Fitzgerald, “Miracles,” The Philosophical Forum, 18, 1985, pp. 48-64.

[12] A related point is made by Reed Richter in his paper, Naturalism and Creationism,” <URL:http://members.aol.com/glauconii/creationism.html>.

[13] See Plantinga, “Methodological Naturalism,” pp. 150-152.

[14] Pierre Duhem, The Aim and Structure of Physical Theory translated by P.P. Wiener with a forward by Prince Louis de Broglie, (Princeton: Princeton University Press, 1906).

[15] Ibid., p. 10.

[16] Plantinga, “Methodological Naturalism?” p. 152.

[17] Ibid., p. 152.

[18] Ibid., p. 154, n. 31.

[19] Ibid., p. 149.

[20] Pennock distinguishes various positions within the Creationist camp. See Pennock, Tower of Babel, Chapter 1.

all rights reserved