Thoughts in a Haystack: Natural Method

Wednesday, October 29, 2008

Natural Method

I've already linked to Dr. Steven Novella's two part response to what appears to be a shift in emphasis, at least, in the Intelligent Design Movement's war against science. As Larry Moran says:

As a general anti-science strategy, it's easy to see why the mind-body problem is resurfacing. The IDiots have lost the battle over evolution so they have to look around for something else to attack. We (scientists) don't understand exactly how the mind works. That's a perfect gap to shove God into, for now.

In the course of discussing that issue, Dr. Novella touched on methodological naturalism and, after a lively discussion in the comments to his post, he decided to expand on that discussion. It's a good exposition on the subject that is, I think, essentially correct. In fact, I've made a similar point to the following in arguments about whether science has disproven the existence of any god that interferes in the world:

This methodoligical approach also deals with the problem of whether or not science can deal with God. The answer is - yes and no. If a supernatural (meaning inaccessible to science) power were meddling with our universe (with stuff science could access), science could detect it, document it, and even describe it. We could say that something was happening.

However (by the premises of this hypothetical situation) if the ultimate cause of these physical effects were beyond scientific methodology, the best science could do would be to describe anomalies. Science comes across anomalies all the time, and the typical approach is to assume (because we really have no choice) that the anomalies are due to either errors in observation, errors in our current theories, or incompleteness in our current theories, meaning there is some new phenomenon to discover.

So far the scientific approach (assuming anomalies will lead to a deeper understanding of reality) has worked out pretty well. This is the best evidence we have that our universe if mostly rational and does not include "supernatural" (by my definition) forces that will remain forever "mysterious." If it did, then we would run across anomalies that we could never explain scientifically. All we could do would be to describe them, but we could never come up with a testable theory of mechanism.

I would quibble that divine action would not necessarily produce anomalies. For example, how could we tell the difference between a random mutation and a miraculous one?

Claiming that we can see no pattern in mutations, or the evolution it powers, does no good because that requires that you make an assertion about what God wants to do and how he, she or it would go about it -- and how could you know that?

This is the problem that Pierre Duhem first raised and which was expanded on by Willard Van Orman Quine. It is, cleverly, now generally known as the Duhem-Quine thesis. Elliot Sober describes it in his excellent new book, Evidence and Evolution, as follows:

Theories rarely make predictions on their own; rather, auxiliary assumptions need to be brought to bear. For example, the general theory of relativity, by itself, does not make predictions about when eclipses will occur or what features they will have. However, if auxiliary information about various celestial bodies is taken into account, the general theory of relativity does make predictions about these matters [leading to Eddington's famous confirmation of the GTR - JP]. Duhem's point holds for most of the hypotheses that the sciences consider, and it also holds when we recognize that prediction rarely involves deduction. Duhem's idea is that the usual pattern in science is that the hypothesis H does not entail whether the observation statement O will be true; rather it is H&A that will have this kind of entailment, for suitably chosen auxiliary assumptions A.

As Sober points out, both the design argument for God and arguments against God based on a supposed lack of design (Stephen Jay Gould's invocation of the panda's thumb being a prominent example) require auxiliary assumptions about the motives and abilities of God that we have no independent justification for.

But, outside of another quibble that "falsifiable" should not be understood as denoting Karl Popper's now discredited "demarcation criteria" but as a synonym for "testable," Dr. Novella is exactly right in his summation:

Saying that science requires methodological naturalism is really just another way of saying that science requires falsifiable hypotheses, which in turn requires the assumption that the universe makes sense - it consistently follows an internal set of rules. ID proponents and others who oppose this view want to inject supernatural explanations into science, by which they mean ideological beliefs that are not testable by science. They try to dress up these beliefs as scientific theories by framing god-of-the-gaps arguments from ignorance (like irreducible complexity) as if they were testable hypotheses - but they are not.

___________________________________

Here is Larry's objection to my point.
.

Labels: Methodological Naturalism