The Evidential Argument from the History of Science, Part 5: Reply to RD Miksa

In the combox on Victor Reppert’s Dangerous Idea blog, RD Miksa posted a thoughtful, four part assessment of the evidential argument from history of science (AHS). In this post, I want to reply to Miksa.

Miksa’s Reply to the Informal Statement of the Argument

The Definition of “Science”

Miksa correctly points out that my post did not provide an explicit definition of the crucial term “science.” Following Paul Draper, let’s define two kinds of sciences.

nomological or inductive science: the attempt to “determine how nature normally operates or functions–to discover, classify or explain unchanging laws or properties of nature.”[1]

historical science: an activity, the main goal of which is “to reconstruct sequences of historical events and to explain particular features of nature by reference to the past”.[1]

Unless otherwise specified, I will use “science” as a catch-all phrase which includes both kinds of sciences.

The Theme of the History of Science

I wrote: “If there is a single theme unifying the history of science, it is that naturalistic explanations work.” Miksa’s reply:

Next, it is arguable that the unifying theme of science is that naturalistic explanations work. For example, I could argue that the unifying theme of the history of science is that the more discoveries science makes, the less it seems to know and explain (dark matter, origin of life, language, consciousness, etc.). In addition, another theme would be that the history of science shows that science is still unable to explain, to any degree of adequacy, fundamental problems that have existed since its beginning: as mentioned above, the origin of life, the origin of biological diversity, human language, human consciousness, etc. Given these facts, why put any substantial weigh into the fact that the so far, naturalistic examples seem to work for an enterprise that creates more questions than it provides answers, and which still seems so unable to answer fundamental questions that it arguably should have answered some time ago.

He also asks:

Furthermore, what is your definition of “work”? For you, does work mean “provide the truth”?

To say that an explanation “works” is just to say that an explanation is inductively justified: the inductive probability of the explanation is greater than fifty percent.

If so, then again, this point is arguable given the fact that any scientific theory, given its necessarily “always-open-to-falsification” stance, means that it is indeed questionable whether scientific discoveries can ever truly provide us with knowledge of that which is true. At best, they can only provide a best explanation given the current state of evidence. That might make a scientific theory rational to believe, but not necessarily true.

I agree with most of this, but I obviously disagree with Miksa’s suggestion that the “always-open-to-falsification” stance undermines my point about the history of science. The fact that an explanation may be falsified in the future says nothing about its inductive probability today, and so does nothing to undermine the point that “naturalistic explanations work.”

But What about “Methodological” Naturalism?

Next, Miksa asks, what about the role of methodological naturalism in science? Doesn’t science’s commitment to an absolute methodological naturalism force it to automatically exclude supernatural explanations, even if those explanations are the best available explanations? In Miksa’s words, “So, given methodological naturalism, is it any wonder that science only seems to discover naturalistic explanations?”

As I’ve defined it, science, including nomological science and historical science, is not committed to an absolute methodological naturalism. Rather, science is committed to a modest methodological naturalism: scientific explanations may appeal to the supernatural, but only as a last resort.  To avoid a potential misunderstanding, it’s important to point out that even this modest methodological naturalism does not justify god-of-the-gaps arguments: just because no naturalistic explanation has been found for some fact F, it doesn’t follow that a supernatural explanation of F is the best explanation. Before we are justified in concluding that a supernatural explanation is the best scientific explanation, we would need good reason to believe there is no unknown naturalistic explanation as well. As Draper puts it, “In other words, the search for natural causes should continue until the best explanation of the failure to find one is that there is none.”[1]

Elsewhere, Miksa writes:

You said:

“Indeed, naturalistic explanations have been so successful that even most scientific theists concede that supernatural explanations are, in general, implausible, even on the assumption that theism is true.”

Not much to say here, except: If you mean the very same scientific theists that have been trained in the principle of methodological naturalism and have had this principle enforced upon them on pain of being called “unscientific”, then is it any surprise that even such scientists only offer naturalistic explanations when they do science.

This is a red herring. As Draper writes:

As Philip Clayton (1997, 172) points out, the presumption that natural events have natural causes existed long before the rise of modern science. Indeed, even in the Bible, naturalistic explanations of unusual events are typically the first choice, with appeals to the supernatural being adopted only after the most likely naturalistic explanations are ruled out. For example, according to 1 Sam. 3, the prophet Samuel, upon hearing a voice in the temple, considered the possibility that it was Yahweh only after repeated attempts to find a human cause failed.

Since it is unlikely that the authors of the Bible are guilty of some anti-religious metaphysical bias or that they believe that a faithful or generous God would never act directly in the world, what is the source of this pre-scientific presumption in favor of naturalistic explanations? No doubt it is a simple induction from past experiences. In very many cases, a little investigation reveals natural causes for natural events, even unusual ones. Thus, it follows inductively that, prior to investigation, the probability that the immediate cause of any given natural event is itself natural is high. We did not need science to teach us this.[1]

What About Scientific Discoveries Which Support Theism?

Miksa writes:

Finally, to end off Part One on a tangential but relevant note, does your argument take into account the fact that the history of science also seems to point to the fact that many “theistically-supportive” discoveries seem to be made in opposition to the expected and conventional scientific wisdom. Take, for example, the Big Bang and Fine-Tuning as two examples of this.

Unlike many atheists, I don’t rule out the possibility that cosmological fine-tuning could provide evidence for theism and against naturalism. Indeed, I think one closely related argument for God’s existence, the “argument from moral agency,” is the best argument for theism. Even so, that does not undermine my point here, which is that the success of naturalistic explanations is prima facie evidence against theism. In other words, all I am claiming is that the success of naturalistic explanations is one piece of evidence favoring naturalism over theism. I’m not claiming there is no evidence favoring theism over naturalism. If specific 20th-century scientific discoveries are evidence favoring theism over naturalism, then so be it. And it is still the case that the overwhelming success of naturalistic explanations is evidence favoring naturalism. It’s possible, at the same time, for both the overwhelming success of naturalistic explanations to be evidence favoring naturalism over theism and specific 20th-century scientific discoveries to be evidence favoring theism over naturalism. Indeed, it’s possible that the items of theistic evidence (individually or combined) are much stronger than the evidence from the success of naturalistic explanations. But again, all of that is compatible with the claim that the success of naturalistic explanations, by itself, favors naturalism over theism.

Miksa’s Reply to the Formal Statement of the Argument

Is the “Evidence to be Explained” Genuine  Evidence?

In my original post, I cited two items of evidence. Here is the second item.

E2. The history of science contains numerous examples of naturalistic explanations replacing supernatural ones and no examples of supernatural explanations replacing naturalistic ones.

Miksa questions the accuracy of E2.

First, this does not take into account my earlier point that I made in a previous post (see “Point 2” in one of my other posts), which calls into question whether the naturalist would ever accept a supernaturalistic explanation for an empirical fact or rather whether the naturalist wouldn’t simply create some far-fetched naturalistic explanation for the fact in order to allow him to maintain his naturalism. I would argue not only that naturalists would do this, but that they do do it, thus making it unsurprising and not very worthy from an evidentiary perspective to find out that naturalists always seem to claim that there is “some” naturalistic explanation for any empirical event, no matter how strange such a naturalistic explanation might be.

Although I believe that open-minded naturalists would accept a supernatural explanation when justified, it really doesn’t matter one way or the other. Miksa has shifted the topic from “the history of science” to “the history of naturalists’ attitudes towards supernatural explanations.” But the attitudes of naturalists isn’t the topic. Even if Miksa were correct that naturalists have offered (and continue to offer) “far-fetched” naturalistic explanations so as to avoid supernatural conclusions, that merely biographical fact would be of little philosophical or evidential significance. All that matters is whether item of evidence E2 is epistemically probable.  Miksa’s claim about naturalists’ attitudes towards supernatural explanations does nothing to show that E2 is improbable.

Miksa then proposes rewording E2 because it is “arbitrary” and open to naturalists’ “confirmation bias.” I disagree that Miksa’s proposed rewording is necessary, but I’m going to let that pass. Instead, I want to focus on Miksa’s proposed counterexamples to E2.

… your statement about “NO supernatural explanations replacing naturalistic ones” is just false. Two examples will illustrate the falsity of your claim. First, exorcism events that are assumed to be naturalistic, and are initially treated as such, but are then shown to have no naturalistic explanation that is nearly as reasonable as a supernatural explanation. And the same thing for healing events that happen at such places as Lourdes. Even though assumed as naturalistic in origin, and even though assessed by multiple people in order to discover a naturalistic explanation, a few such events simply are best explained as supernatural in origin. And since you posit that there have been NO such events, and since I claim that at least in these cases, any reasonable person would see that there are a few such cases, then your statement is false.

This objection is multiply flawed. First, the fact that various phenomena are “assumed to be naturalistic, and are initially treated as such, but are then shown to have no naturalistic explanation that is nearly as reasonable as a supernatural explanation,” is not of obvious relevance to E2.  For E2 refers to explanations, not a presumption about the type of explanation (naturalistic or supernatural). For example, the best scientific explanation for the origin of species used to be special creation, but that explanation has been replaced by the theory of common descent. Miksa’s objection, however, fails to make this distinction. I suspect that Miksa is probably correct that both exorcism events and the Lourdes healings are presumed to be naturalistic events, i.e., events that are presumed to have a naturalistic explanation. But that presumption is not the same thing as the actual explanation itself. In order for either of those events to be a counter-example to E2, Miksa needs to:

(1) identify some naturalistic explanation X ;

(2) show that X used to be widely accepted by the scientific community;

(3) identify some supernatural explanation Y; and

(4) show that the scientific community later replaced X with Y as the best explanation.

Miksa hasn’t yet done this, however. Therefore, the objection is, at best, incomplete.

Second, let’s assume, but only for the sake of argument, that there is no good, known naturalistic explanation for exorcism events or Lourdes healings. Just because no good naturalistic explanation has been found for some fact F, it doesn’t follow that a supernatural explanation of F is the best explanation. What’s needed is an argument which shows that the best explanation of the failure to find a natural cause of F is that there is none. I do not find such an argument in Miksa’s comments.

Third, Miksa is correct that E2, as worded, makes a universal generalization: it says that there have been no examples of supernatural explanations replacing naturalistic ones.  Let us suppose, but only for the sake of argument, that that generalization is false and there have been examples of supernatural explanations replacing naturalistic explanations. Even so, it still wouldn’t follow that the history of sciences provides no evidence favoring naturalism over theism.  To put the point somewhat crudely or simplistically, imagine a library that contains textbooks for all of the sciences, e.g., physics, chemistry, biology, geology, astronomy, medicine, etc. Suppose that the textbooks summarize all currently plausible scientific explanations for those fields. The percentage of such explanations which make no explicit appeal to supernatural agency is extremely high, while the percentage of such explanations which do explicitly appeal to supernatural agency is, at best, very small.

Is AHS an Argument for Naturalism?

Miksa also argues that, even if AHS succeeds as an argument against theism, it fails as an argument for naturalism.

You said:

“Such explanatory success is antecedently more likely on naturalism–which entails that all supernaturalistic explanations are false–than it is on theism. Thus the history of science is some evidence for naturalism and against theism.”

Even if we admit—for the sake of argument—that your argument is sound and valid (and this is a pretty big “if”), your conclusion still does not follow. At best, this is an argument against theism. But it does nothing to support naturalism. Why? Because the so-called “success of naturalism” in the sciences is completely and utterly compatible with deism; the deist would be perfectly comfortable admitting the success of naturalism in the sciences all the while noting that this fact would do absolutely nothing against the deistic position.

So, really, what you have is maybe an argument that weakens theism. You do not, however, have a positive argument for naturalism.

This is false. Let’s consider two versions of supernaturalism.

Theistic supernaturalism (theism or T): there exists a supernatural person who (timelessly or temporally) creates and sustains the natural world, acts in it, and is omnipotent, omniscient, morally perfect, and worthy of worship.

Deistic supernaturalism (deism or D): there exists a supernatural person who created the natural world but does not act in it and is not worthy of worship.

The fact that some item of evidence (E) is “utterly compatible” with deism (D) does nothing to support Miksa’s claim that AHS “does nothing to support naturalism. ” The objection ignores the prior probabilities of naturalism and deism. Once we take this prior probabilities into account, however, it becomes clear that deism’s compatibility with the evidence doesn’t undermine AHS as a prima facie argument for naturalism. We can extend the formulation of AHS as follows.

(1) E is known to be true.

(2) E is much more probable on the assumption that naturalism is true than on the assumption that theism is true, i.e., Pr(E | B & N) >! Pr(E | B & T).

(3) T is not much more probable intrinsically than N.


(4) Therefore, other evidence held equal, T is probably false. [from (1), (2), and (3)]

(5) N and supernaturalism (S) have equal intrinsic probabilities.

(6) On the assumption that S is true, D is not certain. [from the definitions of S and D]

(7) Therefore, D is intrinsically less probable than S. [from (6)]

(8) Therefore, N is intrinsically more probable than D. [from (5) and (6)]


(9) Therefore, other evidence held equal, N is probably true. [from (4)-(8)]


[1] Paul Draper, “God, Science, and Naturalism.”