The question arises: how would the Bayesian way of choosing between the Lamarckian and Darwinian models of evolution or of reshaping one’s views on the mentally challenged compare with the empiricist way or the rationalist way of dealing with these same problems?

The chief danger of empiricism that Bayesians try to avoid is the insidious slip into dogmatism. Many times in the history of Science, empiricist-minded scientists have worked out and checked a theory so thoroughly that they have slipped into thinking that they have found an unshakeable truth. For example, physicists in the late 1800s were in general agreement that there was little left to do in Physics. They believed that Newton and Maxwell, between them, had articulated all the truths of all levels of the physical world, from the atomic to the cosmic. Einstein’s theory of relativity changed all of that. For many physicists of the old school, relativity was a very rude shock.

                                        

                                                     James Clerk Maxwell (credit: Wikimedia Commons)

Today, Physics is in a constant state of upheaval. A few physicists still show a predilection for dogma, or we could say a longing for certainty, but most modern physicists are tentative and cautious. They’ve been let down so many times in the last hundred years by theories that once had seemed so promising, but that later were shown by experiment to be flawed, that most physicists have become permanently leery of any colleague who claims to have “the truth.”

It is regrettable that a similar caution has not caught hold of more of the physicists’ fellow scientists, especially the biologists. Darwinian evolution is indeed a powerful theory. It explains virtually all aspects of the living world that we currently know about. But it is still only a theory, which means that, like all theories, it should be viewed as tentative, not final or irrevocable. It just happens currently to have vastly more evidence to support it than do any of its competitors.

The larger point for our purposes here, however, is that Bayesians never endorse any one model as the last word on anything, and they never throw out any of the old models or theories entirely. Even those that are clearly wrong have things to teach us, and of the ones that are currently working well, we have to say that, simply, they are currently working well. There are no final answers and no final versions of the truth in any model of reality for a Bayesian. The theory of evolution is only currently working well.