The Evidential Argument from the History of Science, Part 2: Detailed Reply to Randal Rauser


Theists hold that there exists an omnipotent, omniscient, and morally perfect person (God) who created the universe. Metaphysical naturalists, on the other hand, hold that the universe is a closed system, which means that nothing that is not part of the natural world affects it. Metaphysical naturalism (N) denies the existence of all supernatural beings, including God. Therefore, N entails that any true scientific explanations must be naturalistic (i.e., non-supernatural) ones.

In my post, “Evidential Argument from the History of Science,” I appeal to evidence (E) regarding the nature of scientific explanations. E1 states the overwhelming  number of plausible scientific explanations for physical phenomena which do not appeal to supernatural agency. While readers may think of the topics that are standard fare for “science and religion” discussions (such as biological evolution, mind-brain dependence, etc.), the scope of E is much broader than that. To put the point somewhat crudely or simplistically, imagine a library that contains textbooks for all of the sciences, e.g., physics, chemistry, biology, astronomy, medicine, etc. Suppose that the textbooks summarize all currently plausible scientific explanations for those fields. The percentage of such explanations which make no appeal to supernatural agency is extremely high, while the percentage of such explanations which do appeal to supernatural agency is, at best, very small.

Furthermore, E2 states that the history of science contains numerous examples of naturalistic (i.e., non-supernatural) explanations replacing supernatural ones and no examples of supernatural explanations replacing naturalistic (i.e., non-supernatural) ones. Of course, one hears about specific scientific questions which (allegedly) do not have a plausible naturalistic (i.e., non-supernatural) explanation, such as cosmological fine-tuning, the origin of life, and consciousness. But that in no way denies the point that there have been numerous examples of naturalistic explanations replacing supernatural ones and no examples of supernatural explanations replacing naturalistic ones.

The central claim of the evidential argument from the history of science (AHS) is its premise (2), which states E is antecedently much more probable on the assumption that naturalism is true than on the assumption that theism (T) is true. In symbols:

(2) Pr(E | B & N) >! Pr(E | B & T).

And, for convenience, here again is the entire structure of AHS in its logical form.

(1) E is known to be true.
(2) 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.

Randal Rauser does not find AHS in the least bit convincing. In his words, “it provides no evidence that theism is false or that naturalism is true.” As I read his reply, his entire objection to AHS is based upon his denial of (2).

Science, Theology, and “Proper Spheres”

Here is Rauser:

Now let’s consider the opening sentence: “If there is a single theme unifying the history of science, it is that scientific explanations work.” While this is certainly true, we should make sure we are very careful in considering how it is true.

Unfortunately, Rauser’s proposed terminological swap voids my sentence of its intended meaning and instead makes it look like a truism or a tautology. If theism is true, it could have been the case that successful scientific explanations were supernatural explanations, i.e., explanations which directly appealed to supernatural agency. For example, biology could have discovered that all animals are not the relatively modified descendants of a common ancestor; neuroscience could have discovered no correlations at all between human minds and brains, etc. Rauser says nothing about this in his reply.

Instead, he argues that science and theology each have their own “proper spheres.”

Science is to the object of its study (i.e. the physical structure of the universe) as hammers are to nails: a great tool for that specific use. However, we would balk at any person who reasoned that the continued success of hammers at pounding nails provided evidence that hammers could replace the other tools on the belt. It would be equally in error to suppose that the continued success of science at explaining the physical structure of the natural world could replace other explanations in their proper spheres.

It may surprise Rauser to learn that there is a certain sense in which I agree with him. Here is an example from ethics. For example, while science has discovered facts about the biological nature of human beings, facts which can and often do have moral significance, science as such tells us nothing about which normative ethical theories are true. I agree with Rauser that it would be a mistake to argue that the success of science shows that all ethical theories are false or that naturalistic explanations have replaced ethical explanations. But so what? Scientific explanations and normative ethical theories are not rival theories. For example, there is no ‘ethical theory of gravity’ as opposed to the scientific theory of gravity. Or again, there is no ‘scientific theory of the good’ as opposed to the Platonic Form of the Good.

Rauser’s “proper sphere” approach, however, breaks down when we try to apply it to science and theology, in the way he wants to apply it. There are obviously many claims in theology which everyone agrees are completely outside the scope of science (i.e., doctrines about salvation, the afterlife, etc.). But there are other possible theological claims which can conflict scientific explanations. For example, what accounts for biogeography, i.e., the distribution of plant and animal life, both as we find it today and as we find it in the fossil record? Some theists, i.e., young earth creationists, have offered a straightforward, supernatural explanation: God flooded the entire Earth for 40 days and nights as described in Genesis (the “Genesis flood”). I’m assuming that Rauser rejects the Genesis flood explanation, but the fact remains that the Genesis flood explanation is just that: a rival, potential (but false) explanation, one which attempts to explain physical phenomena in terms of supernatural agency. Thus, Rauser’s “proper spheres” approach does nothing to undermine premise (2) of AHS.

God of the Gaps Theology

Rauser next quotes an earlier editorial in which he argued against God-of-the-Gaps theology. His crucial point seems to be this.

But God-of-the-gaps theology is not only mistaken because it tends to shrink the conceptual space in which God can act. In add

ition, I believe it depends on the flawed assumption that a natural explanation for a given phenomenon excludes the need for a supernatural explanation.

Again,  it may surprise Rauser to learn that I agree with him: the fact that some physical phenomenon has a natural explanation does not exclude the possibility of a supernatural explanation. Possibility, however, is the key word here. AHS is an evidential argument, i.e., it grants that E is logically compatible with T. Instead, AHS explicitly appeals to probability. Thus, it is irrelevant to cite mere logical possibilities. Rauser needs to present some reason for doubting that premise (2) is true.  His God-of-the-Gaps editorial doesn’t do this.

Rauser’s Pit Bull

Finally, Rauser quotes from an earlier post where he gives his pit bull illustration. Nothing in that illustration is relevant to AHS, however. In his illustration, Suzy’s misfortune is not evidence for the truth of Ray’s advice and against Randy’s advice because her misfortune is equally antecedently probable on both the assumption that Ray’s advice is true and on the assumption that Randy’s advice is true. (Rauser gets it wrong when he writes, “No, it doesn’t, for the simple reason that both Ray and Randy offered advice fully consistent with this unfortunate outcome.” Talk of “consistency” completely misses the point; the relevant question is whether the data is antecedently more probable on the assumption that one hypothesis is true than on the assumption that a rival, competing hypothesis is true.)

But that is irrelevant because the pit bull illustration is itself irrelevant. Here is Rauser.

By the same token, discovering the natural genesis of lightning does not constitute evidence for MCN because the natural genesis of lightning is fully consistent with the existence of God and his action in the world. (Indeed, to assume that there is such a thing as special divine action entails that there is such a thing as regular action in which natural processes are operative, including the discharge of lightning under the right conditions.)

Again, Rauser seems to be confused about AHS. To borrow jargon from the contemporary literature of the philosophy of religion on the problem of evil, AHS is an evidential argument, not a logical argument. In other words, AHS does not claim that E is logically incompatible with T. Rather, AHS claims that E is antecedently much more probable on the assumption that N is true than on the assumption that T is true. Thus, Rauser’s pit bull illustration provides no reason whatsoever to deny premise (2).

It is not until the very end of Rauser’s post that we find a point which could, at least, be potentially relevant to (2).

For at least fifteen hundred years countless theologians have developed what I call transcendent agent models of divine action in the world which view God as the primary metaphysical agent of all natural events in a way that is completely consistent with the advance of scientific explanation in its proper sphere. So once again, and for good measure, the advance of science in its proper sphere which it has rightfully claimed from other disciplines, provides no evidence that those other disciplines do not have their proper spheres. And thus it provides no evidence that theism is false or that naturalism is true.

I addressed this in the updated version of my last post on AHS. For convenience, I will quote my reply there in its entirety.

On the assumption that theism is true, what reason is there to believe that transcendent agent models of divine action in the world are true? Is there any reason that is independent of the success of  non-supernatural explanations?

Let’s define A as the hypothesis that “God is the primary metaphysical agent of all natural events in a way that is completely consistent with the advance of scientific explanation in its proper sphere.” A is clearly logically compatible with E, but the question is whether A undermines premise (2) of AHS. In order to properly evaluate the evidential impact of A, if any, on AHS, I propose that we treat A as an auxiliary hypothesis (to theism). It follows from the theorem of total probability that:

Pr(A | T) = Pr(A | T) x Pr(E | A & T) + Pr(~A | T) x Pr(E | T & ~A)

In the context of explanatory arguments, Draper calls that theorem the “weighted average principle” (WAP).[6] As Draper points out, this formula is an average because Pr(A | T) + Pr(~A | T) = 1. It is not a simple straight average, however, since those two values may not equal 1/2; that is why it is a weighted average.[7] The higher Pr(A | T), the closer Pr(E | T) will be to  Pr(E | A & T);  similarly, the higher Pr(~A | T), the closer  Pr(E | T) will be to Pr(E | T & ~A).[8]

WAP shows that, in order to be successful, an objection to an evidential argument must do more than simply identify an auxiliary hypothesis which can explain the data. The objection must also provide an antecedent reason for thinking that A is true, i.e., a reason for thinking that A is more probable given the core hypothesis–in this case, theism–than given the negation of the core hypothesis. Without such a reason, the objection reduces to the fact that E is merely logically compatible with the core hypothesis (in this case, T), which is no objection at all to an evidential argument.

For this reason, then, this objection is, at best, incomplete. It successfully identifies a relevant auxiliary hypothesis (A), but does not (yet) provide an antecedent reason for expecting that hypothesis to be true, on the assumption that theism is true.

I hope it is now clear why my original reply focused solely on Rauser’s appeal to A and why the other parts of his reply are not even relevant to AHS.

Addendum: Rauser’s Second and More Critical Look at AHS

Due to the fact that my initial response to Rauser only addressed one of his objections to AHS, he posted a second article criticizing AHS.   I want to comment on that response here.

Theologians in Perpetual Retreat?

In direct response to Rauser’s positing the transcedent agent model, I proposed that we treat that model as an auxiliary hypothesis (A). In order to determine if this model is just an ad hoc proposal invented to accomodate E, I asked what antecedent reason exists on theism to expect that A is true.  Rauser writes:

Theologians had biblical reasons for adopting a TA model (e.g. explaining the relationship between human and divine agency in election).

The evidential argument from the history of science (AHS) is an argument against theism, not Christian theism per se. Now before anyone says, “Aha! Lowder admits AHS doesn’t touch Christian theism,” I say, “Not so fast!” Since Christian theism entails theism, it logically follows from the axioms of the probability calculus that the probability of Christian theism cannot be greater than (generic) theism. (Anyone who doubts this should draw a simple Venn diagram for proof.)

So how does TA affect the antecedent probability

of E in AHS? “Biblical reasons for adopting a TA model” are evidentially relevant if and only if the probability relations specified by the Weighted Average Principle (WAP), which follows from the theorem of total probability, are satisfied. Rauser’s reply says nothing about WAP, however. So Rauser has not yet provided an antecedent reason to think that such Biblical reasons are more probable than not, on the assumption that theism is true. And therefore he has not yet provided any reason to think the “TA model” refutes premise (2) of AHS.

Rauser continues:

And they had countless philosophical and theological reasons as well. For instance, TA models flow naturally from a commitment to divine simplicity and atemporality, two mainstays of classical theism. So theologians prior to the rise of science had many reasons to endorse TA models of divine action.

The claim, “TA models flow naturally from a commitment to divine simplicity and atemporality,” is just that: a claim, an assertion, in need of support. I don’t find that support in what Rauser has written (so far).