To save the symmetry principle it is not enough to separate that principle from its false companions, as I have tried to do so far in this series. It is also necessary to show that adherents of the principle are likely to write better histories of science than those who flout it. In the previous post I defined the Symmetry Principle as the view that we should not reason from the truth or falsity of a belief to the goodness or badness of the believer's reasons for holding the belief. The best defense of this principle is simply to observe that sometimes past scientists have been right for the wrong reasons and wrong for the right reasons. But there's a problem with this defense: it seems to lead to radical skepticism about present-day science. In this post I want to show how we can accept the Symmetry Principle without abandoning present-day science or erecting artificial barriers between scientists and historians. The fallacy is not always applicable The Symmetry Principle states that it is a fallacy to go from truth-value to goodness-of-reasons. Before showing that this is indeed a fallacy, let me point out that it is not always a useful fallacy. To use the fallacy for a given belief, we would need to identify that belief as either true or false. The problem is that many past beliefs are neither strictly true nor strictly false, either because they are vague or because they are precise but only partly true. To make matters worse, some beliefs are still matters of mainstream debate. Perhaps we could solve the problem of partially true beliefs by resolving them into truth and false components. Then we would infer that the true component was due to good reasons and the false component to bad reasons. This will not usually work, however. It is implausible to believe, for example, that Aristotle was guided by reason when he saw efficient causes in nature but driven by superstition or self-interest when he saw final causes. Surely those two ideas were too neatly joined in his mind to have such disparate origins. Two unsound defenses of the Symmetry Principle Nevertheless there are a fair number of historical disputes that have clear winners and losers. Kepler was right that the planets have elliptical orbits, or at least much more right than Galileo, who believed their orbits to be circular. The speed of light is finite, as Christiaan Huygens believed and as René Descartes denied. Heat is not a fluid, rock crystal is not made of frozen water, coral is not a mineral, and the moon is not translucent: most of us believe these claims even though each was hotly disputed at some point in the past. Why should we apply the Symmetry Principle to these beliefs? A common answer is that they may turn out to be false. Maybe we will one day discover that the speed of light is infinite, and if that happens we would have to revise our history books to assign good reasons to Descartes and bad reasons to Huygens. But, so the argument goes, such post-hoc revisions would be absurd. The problem with this argument is that it can be levelled against any procedure for generating beliefs about the past. Maybe we will one day discover that Descartes was the victim of a clever and mischievous type-setter who inserted his own ideas in key places in the philosopher's major works. Such a discovery would lead to a retrospective overhaul of Descartes scholarship. But the possibility of such revisions does not undermine our faith in textual analysis as a way of finding out about the past. Another common answer is that it is simply not possible for the good reasons to fall on one side of a dispute and the bad reasons to fall on the other side. On this view, every party in every scientific debate is motivated by a roughly equal mixture of sound argument and corrupting self-interest. As per my last post in this series, I do not think this a good answer. All we can say with confidence is that there is a non-zero amount of both in the causal history of most beliefs. Deciding whether or not one kind of cause dominates in any given case is a matter for empirical investigation. (I suspect that the only reason people assert otherwise is that they conflate this assertion with the Symmetry Principle). A better defense, and two diagnoses So much for the wrong answers—what about the right ones? In my view the best answer to the symmetry sceptic is simply to list cases in which a famous past scientist held a true belief for a bad reason or a false belief for a good reason. As examples of the latter one need only think of Galileo, whose belief in the motion of the earth was partly based on an explanation of the tides that seems incoherent to us and that convinced very few people at the time. As examples of false beliefs backed by good reasons, one need only think of all errors that were not due to methodological sloppiness but to the unavoidable absence of key pieces of evidence. Ptolemy did not have the benefit of Galileo's telescopic observations when he built his world system; William Thomson had no knowledge of radioactive decay when he gave his first estimates of the age of the earth; etc. Many errors are due to bad luck, not bad method. Given such examples, why would anyone want to deny the Symmetry Principle? And why do people sometimes violate the principle even though they should know better? I suspect that there are two factors at work. One is that some past beliefs agree so closely with present-day claims that the agreement cannot be due to coincidence. To continue with the speed of light example, the eighteenth-century British astronomer James Bradley (the subject of this recent blog post) gave a figure that was within 2% of the currently accepted value. Assuming that the current value is correct, it is absurd to think that Bradley made a lucky guess; he must had had good reasons for deciding on this value. The problem is that many scientific theories, including some of the historically most interesting ones, are not of this kind. Often there are only a small number of theories to choose from, and in such cases there is plenty of room for lucky guesses. Way back in the fifth century BC, the Greek philosopher Empedocles maintained that the speed of light is finite. The fact that he was right about this does not tell us anything about the quality of his reasons for believing so, since even with the flimsiest reasons he would still have had a 50/50 chance of getting the right answer. Overconfidence in the scientific method is the second factor that may lead to violations of the Symmetry Principle. Anyone who believes in a fail-safe method is bound to deny that false beliefs can be explained by that method. False beliefs must instead be explained by deviations from the true method, deviations brought about either by sheer carelessness or the action of self-interest, prejudice, religious convictions, etc. A serious objection to the Symmetry Principle, and two bad replies The argument from scientific method is one that historians of science are trained to dismiss out of hand. However there is no room for complacency here. There is a serious objection behind the appeal to method, an objection that we need to disarm if we want people to take the Symmetry Principle seriously. The objection goes like this:
Commonsense realism tells us that present-day scientists are a pretty good guide to the truth: if a large majority of the relevant experts say that a given theory is true, or that the theory is false, they are probably right. But the only way the experts can make these judgements is by examining the arguments for and against the theory. In other words, goodness-of-reason is a reliable guide to truth-value. It follows, as a matter of elementary logic, that the reverse is true, ie. truth-value is a reliable guide to goodness-of-reason. So far this argument applies only to present-day science. But there is nothing special about 2014, so it must hold for the past as well. Therefore historians are entitled to infer the goodness of a scientist’s reasons for a belief from the truth-value of that belief.It would not be a good idea for historians to respond to this objection with a blanket denial of commonsense realism. Maybe commonsense realism is a misguided doctrine. But the fact is that the realism debate is still a live one among the relevant experts, namely philosophers of science, and it would not be wise for historians to ground their discipline on a form of skepticism that may be false and that many people reject. This would be an especially foolish move if historians intend to use their case studies to attack commonsense realism; to do so would be to assume what they are trying to prove. Another false move is to say that commonsense realism is OK for scientists but not for historians of science (this move is sometimes called "compartmentalism" or "meta-relativism"). On this view, scientists are right to be realists and historians are right to reject realism. This will not do. If the claim is that scientific realism is true for scientists but false for historians, then the claim is guilty of an incoherent form of relativism. Perhaps the claim is instead that scientific realism is true tout court, but that it is methodologically suitable for historians to sideline this truth when they study past science. To say this is to restate the problem, not solve it. The challenge is precisely to reconcile realism with the methodological value of sidelining realism. Some better replies to the serious objection Here are some better replies to the objection that I raised in the previous section. Firstly, the objection only applies to beliefs that enjoy a wide consensus among the relevant experts. If at time t there was no consensus about a given theory, then the objection does not apply to that theory at time t. The reason the objection does not apply to disputed past theories is that commonsense realism does not apply to disputed present theories (it makes little sense to be a realist about a given theory and its rival theories). Since the objection assumes commonsense realism, it has no purchase on theories that are under dispute. This restriction is important because disputed theories are usually the ones that are of most interest to historians and sociologists of science. Consider again the question of whether the speed of light is finite or infinite. It would be no great loss to historians if they could only apply the Symmetry Principle to this question as it was answered prior to 1750, by which time most informed people were finitists about the speed of light. The second reply is that empirical study is a better guide than philosophical inference when it comes to learning about a particular past event. In other words: however good the objection might be, it is no substitute for a careful study of the documents relating to the case that interests us, whether this is the disagreement between Huygens and Descartes over the speed of light, or nineteenth-century debates about natural selection, or whatever. Here's a World Cup analogy to illustrate the point: historians can ignore scientific realism when they study a particular case for the same reason that soccer fans can ignore the predictions of the pundits when they watch a particular match. It’s always worth checking what the result actually is, even if the predictions are right most of the time. These seem to me to be the most straightforward replies. There are other replies that, though promising, are likely to be more controversial, either among philosophers or historians or both. One is to accept that the objection holds for recent science but deny that it holds for earlier periods, on the grounds that evidential standards were lower in the past than they are now. In the early modern period, even the best reasons were not that good, and as a result they did not track the truth as reliably as today’s precision instruments and randomised controlled trials. Another promising reply is to accept commonsense realism about today’s empirical laws—such as the rules of geometric optics—but reserve judgement about today’s high-level theories—such as string theory. This is a perfectly respectable move within the philosophy of science, and it would restrict the scope of the objection in question so that it only applied to past empirical laws and not to high-level past theories. That way we would be free to apply the Symmetry Principle to the eighteenth-century debate over the nature of vis viva, the nineteenth century debate over the existence of the ether, and so on. If you can think of other replies, controversial or otherwise, please feel free to post them in the comments below. For the moment, I hope to have shown that the objection is strong enough to take seriously but not strong enough to scupper the Symmetry Principle. That principle is compatible with confidence in present-day science. There remains the question of whether anti-realism is worth retaining as a heuristic device when writing about past science. In my next post I will argue that it is not worth retaining. In fact, if you need this device then you have not understood the Symmetry Principle. Expand post.