AN EPISTEMOLOGICAL CLAIM

for believing that the theory is true or that it is empirically ad­equate (in one or more of the senses I have noted).

Can this be defended whether or not nature itself is simple? Since the epistemic claim is so important from a scien­tific perspective, I will spend the most time with it.

My general response to the Epistemological Claim will be to reject it. The fact that a theory is simple may be very desirable, but it is no reason whatever to believe it is true. An appeal to it may be made in the course of an argument to a conclusion, but that appeal does no epistemic work. In chapter 3, I will argue for this claim in my discussion of Newton's argument for gravity. But before getting to that point, in the remainder of this chapter I will look at ways the Epistemological Claim might be defended.

William Whewell makes this historical observation: “we have to notice a distinction which is found to prevail in the progress of true and of false theories. In the former class all additional suppositions tend to simplicity and harmony.................................................................. In false theories the contrary is the case.”^[76] Whewell seems to be saying here that true theories, as they evolve over time, tend to be simple and false ones, as they evolve, tend to be complex.

We start with the idea that if we look at the histor­ical facts, simple theories have been much more “empir­ically successful” than complex ones. Their predictions, explanations, calculations, etc., have been better experimen­tally confirmed than those of complex theories. To isolate the power of simplicity, let us suppose that both the simple and the complex theories we are considering are roughly equal with respect to other virtues that might conceivably be regarded as epistemic, but that among these, the simple theories have been much more experimentally successful than the complex ones. And to avoid the problem of different formulations of the same theory, some simple, some more complex, we can idealize by supposing that if there are such formulations, there is some widely accepted canonical for­mulation with respect to which questions about simplicity and complexity will be considered.

Also, we need to avoid “selection” bias. Suppose, e.g., that simpler theories are easier to test and, if needed, to modify in light of new experiments; but when scientists testing com­plex theories find an experimental problem, they give up more easily and abandon the search for a solution. So let us suppose that the theories we select in both the simple class and the complex one have been thoroughly tested and (to simplify even further) either generally confirmed or disconfirmed.

Now, we need to ask the question: Confirmed when? If we are to judge the empirical success of theories and use this as a basis for an epistemic claim about simplicity, then we must use experimental data available to us, not just to those during the time the theory was proposed and favored. On this basis, many of the theories once considered highly confirmed are no longer considered so. If we were to make an induction from just the fact that the “temporary” success rate of simple theories has been higher than the “temporary” success rate of complex theories, that induction could at best be only to the conclusion that simpler theories in the future will prob­ably have a better temporary success rate than complex ones. But champions of the Epistemological Claim want a much stronger conclusion than that. Let's see what we can do.

6.