Saving the symmetry principle, I: two fallacies

Vanessa Heggie has posted a clear, visible summary of what she rightly calls a “core principle” for historians of science, namely the “symmetry principle.” So this is a great opportunity for me to explain why I disagree with much that my fellow historians of science have written on this topic. Behind the symmetry principle there is an insight that is true, important and worth keeping. But we need to save this insight from the ideas that are often associated with it, many of which I think we should reject.

The basic insight is that there is a certain inference, which I will call The Fallacy, that is a tempting yet unreliable way of way of explaining the beliefs of past scientists. The point of this post, the first in a series, is to separate The Fallacy from another fallacy that is so unappealing that no-one had even thought of it before historians of science started finding it everywhere.


The Fallacy

Consider Galileo's theory that the uneven shading of the moon is due to the shadows cast by hills and mountains on the moon's surface. Most of us prefer this theory to one of its seventeenth-century rivals, according to which the moon is a perfectly smooth sphere whose visible blotches are due to the uneven density of the crystalline matter of which it is made.

The fallacy is to say that since the theory is true, Galileo must have believed it because of the evidence in its favour, while those who rejected it must have done so out of superstition, prejudice or self-interest. In general:

Theory X is true, therefore everyone who held it did so for good reasons, while everyone who denied it did so for bad reasons.

This is what I will call The Fallacy. As I have stated it, this fallacy applies to two people with conflicting beliefs, but this limitation is one of convenience rather than necessity. The same kind of fallacy could be applied to two beliefs that have opposite truth-values but that do not contradict each-other. Or it could be applied to two beliefs held by the same individual. Or it could be of these cases in one.

For an example of the latter, consider Galileo's theory that comets are exhalations that mount in a straight line from the surface of the earth. The fallacy would be to infer from the truth of Galileo's moon theory that he held it from a combination of accurate observation and sound inference, and from the falsity of his comet theory that he held it out of dogmatism, amour propre, etc.


The Other Fallacy

It may look like I have simply repeated Vanessa's account with a different example (she uses Leibniz v Newton rather than Galileo). But there's a crucial difference. The teaser at the top of Vanessa's post says that “No one believes something simply because it is true”, and the fifth paragraph urges historians to “forego the assumption that [Newton] believed in his law of gravity because it was true.” This suggests that the fallacy that Vanessa has in mind is something like this:

Theory X is true, therefore everyone who held it did so because it was true, while everyone who denied it did so for bad reasons.

Call this the The Other Fallacy. It is the same as The Fallacy except that it replaces “good reasons” with “truth” as an explanation of the true beliefs of past scientists.

Although The Other Fallacy is indeed a fallacy, I think it is a very rare one, even among the most Whiggish of old-fashioned historians. Imagine asking George Sarton (to pick one such historian) why Copernicus believed that the earth goes round the sun. Would Sarton have said “because the earth does indeed go around the sun”, or perhaps “because Copernicus thought 'the earth goes around the sun' was a true statement”?

I doubt it. Instead he would have said something like: “because Copernicus looked at a whole lot of data, did a bunch of calculations, and found that a number of key phenomena—such as the retrograde motions of the planets and the fact that he never observed Mercury and Venus on the opposite side of the earth to the sun—could be explained in very neat manner by supposing that the earth is just another planet.”

To put one of the first two replies in Sarton's mouth is to commit what I have called the constructivist straw man. Another way to commit the same error is to say that Sarton and his ilk denied that “human agency” or “human activity” played any role in the beliefs of scientists.

Both forms of the straw man can be found in this transcript, which Vanessa links to in her post. First comes the suggestion that, some time in the seventies or eighties, historians started to see scientific knowledge has “a human product, something that had to be made and maintained.” In other words, earlier historians had thought that “true knowledge was immaculate, untouched by human hands.” (This is true only on the perverse assumption that calculating, experimenting, and reasoning—the sorts of things that interested Sarton et. al.—do not count as human activities).

Next in the transcript comes the misattribution that lies behind The Other Fallacy. How did these earlier historians explain the beliefs of scientists, if not as the result of human action? Answer: by appeal to the truth of those beliefs. For example, they would “say that Isaac Newton thought that there was an inverse square law of gravity acting instantly at a distance through empty space between the centers of distant bodies because there is [such a law].”

(Disclaimer: the historian featured in this transcript is my thesis advisor. Clarification: yes, I am disagreeing with my thesis advisor on this point, although I agree with him on much else.)

The problem with The Other Fallacy is that it sets the bar too low. A campaign against it is like an anti-smoking campaign that urges smokers to stop committing murders. Since most smokers are not murderers, the campaign is unlikely to have any effect except perhaps to convince non-smokers that many smokers are, in fact, murderers.

Likewise, urging people to avoid The Other Fallacy is unlikely to solve the real problem, which is The Fallacy. Those who commit the latter are likely to continue doing so, happy in the knowledge that they have not committed The Other Fallacy. And those who avoid The Other Fallacy may end up convincing themselves that everyone else is more wrong-headed than they really are.


Truth and evidence are not the same thing, and it matters

Perhaps historians attack The Other Fallacy, rather than The Fallacy, because the two are hard to tell apart. After all, as noted above, the only difference between them is that between explaining a person's belief by its truth, and explaining that belief by the arguments or evidence that the person found in its favour. And aren't these pretty much the same thing?

No! In the context of a debate, whether in the present or in the past, they are completely different beasts.

This is easy to see from the three quotes I put in the mouth of George Sarton. Imagine if we tried putting those quotes into the mouth of Copernicus rather than Sarton. Does the De Revolutionibus contain statements like: “'the earth moves around the sun' is a true statement, therefore the earth moves around the sun”? Or statements like: “the earth moves around the sun, therefore you should believe that the earth moves round the sun”?

Probably not. Or if it does, they were not the sorts of statement that convinced people that the earth went around the sun. And nor are they the sorts of statement that are used today to convince people that vaccinations work, or that climate change is real, or that God does or does not exist. And this is not something that “modern historians” discovered some time in the seventies and eighties. No sane person has ever denied it.

This should be uncontroversial. Granted, there are controversial issues nearby. There are the questions of the extent to which standards of evidence have varied over time and place, whether there is some super-standard that allows us to assess this alleged multiplicity of standards, whether evidence can reliably inform us about unobservable entities like quarks and quasars, and whether evidence can be decisive in resolving scientific disputes. But we do not need to agree on any of these issues in order to agree that giving evidence for a proposition is different from asserting that the proposition is true.

This large, uncontroversial difference does not stop historians of science running the two together as if they were the same thing, usually when passing judgement on dead historians of science. I've already given one example from a prominent historian. Here's another, from Stephen Shapin's 2010 book Never Pure:

Once upon a time, so the story goes, students of science too believed that truth was its own recommendation, or, if not that, something very like it. If one wanted to know, and one rarely did, why it was that true propositions were credible, one was referred back to their truth, to the evidence for them, or to those methodical procedures the unambiguous following of which testified to the truth of the product.

In this passage Shapin at least recognises that the truth of a claim and the evidence for it are different things. But only just: he also says that they are “very like” each-other. He implies, absurdly, that students of science used to be uninterested in how scientists justified their beliefs. And on the preceeding page he attributes to an unnamed group of “modernist methodologists”—presumably people like Hans Reichenbach, Rudolph Carnap, and Karl Popper—the view that “truth shines by its own light.”

The conflation of truth and justification does an injustice not only to old-fashioned historians like Sarton but also to present-day internalists. The latter may be defined as historians of science whose main interest is the mixture of luck, skill and insight by which past scientists—as individuals or as groups, over months or over centuries—developed arguments for their claims the natural world.

Internalists can be as symmetric as anyone, giving as much attention to the errors of “winners” as they do the insights of the “losers.” They do not give complete accounts of their subject (show me a historian who does!). But they do far more than simply put tautologies in the mouths of past scientists.

To sum up, by conflating truth and evidence we make the latter look as unimportant as the former in the resolution of debates, whether in the present or the past. This is an error as bad as The Fallacy. It is silly to appeal to the presumed truth of a claim in order to persuade people that the claim is true. By contrast, it is silly not to appeal to the presumed evidence for the claim in order to persuade people that the claim is true.

Having identified what I think is the good idea behind the symmetry principle, in my next post I hope to explain why it is a good idea. This can be done, but it is harder to do than my colleagues usually make out.



Postscript 1. If the error that I am attributing to my fellow historians is such a bad one, why had no-one spotted it until now? Actually, at least one person had. Ian Hacking wrote the following in a footnote to his 1999 book The Social Construction of What?:

Evidence, or reasonableness, is quite another matter from truth. [The sociologists of science Barry Barnes and David Bloor] are often taken to hold a symmetry thesis about evidence: you cannot invoke the evidence available to a community for a belief p, in order to explain why people in the community believed p... I find this claim (about evidence, not truth) unsatisfactory (232).

Why did Hacking put this point in a footnote rather than in the main text? Because he only mentioned the symmetry principle on his way to talking about something else, and that thing (the distinction between “nominalism” and “inherent-structurism”) was relevant to truth and not to evidence.

I'm still trying to work out why other commentators on the symmetry principle have not made the point that Hacking made in his footnote, and that I made in the above post. I'ld be delighted to know of any examples of people who have made the point in print.


Postscript 2. Maybe I'm being too harsh. Maybe Shapin et. al. want to draw the following contrast. One can say that a scientist believed P because they were convinced, on the basis of the evidence, that it was true. Or one can say that the person believed P because they were convinced, independently of their views of the truth-value of P, that it was in their interests to believe it.

On this reading, “No-one believes things just because they are true” is shorthand for “No-one believes things just because they are convinced, on the basis of the available evidence, that they are true.”

But if this is what is meant, why not just say “No-one believes things just because they have evidence for them”? This would make it clear that evidence is involved. Or what about saying “No-one believes things just because they believe them to be true”? This would at least make it clear that it is not the truth of a person's belief that is under consideration but the person's conviction that it is true.

Both of these clearer expressions would avoid the rhetorical sleight-of-hand that is involved in saying “No-one believes things just because they are true” when what you mean is “No-one believes things just because there is evidence for them.” The former claim is obviously true. The latter claim may be true, but not obviously so. The ambiguity has the effect of making the intended claim (if the intended claim is indeed the latter one) more obvious than it really is.

But all this may be moot, since I doubt that my charitable reading is correct. I suspect that Shapin et. al. really are accusing past historians and philosophers of thinking that the truth of a scientist's belief (and not just the scientist's conviction of its truth) can be a good explanation of that belief.

One reason for my doubt is that, in the passage quoted above, Shapin includes “truth” and “evidence” as separate items in his list of candidate explanations for the beliefs of past scientists. This suggests that he is not implicitly including “evidence” under the rubric of “truth.” If he were, then it would be redundant to include evidence as a separate item.

Another reason to be skeptical is that, as I understand them, even sociologists of science have trouble making sense of the idea that people can believe things without first being convinced of their truth. That is, sociologists would probably not say “Newton believed his inverse square law because, although he withheld judgement about the truth-value of the law, he thought that this belief would protect his reputation as a national hero.” Rather, their explanation would be something like: “Newton believed his inverse square law because he thought the law was true, and an important source of this conviction was (not the evidence but) his desire to protect his reputation.”

At least, that is what I imagine they would say. And if that is right, even sociologists think that people “believe things just because they are convinced of the truth of those things.” So that cannot be the allegedly false view that Shapin et. al. are attributing to past historians and philosophers of science. The view they are attributing must be the one I assumed in the above post, viz. that people believe things just because those things are, in fact, true. Expand post.

Malheureusement je ne traduis plus tous les posts que j'écrit sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.

Holmes, Porter and Hacking on the Experimental Turn

Frederic L. Holmes (1932-2003)

In my previous post I discussed Jonathan Rée's view that many late-twentieth century intellectuals over-reacted to the excesses of their modernist predecessors. In this post I suggest that something similar may have happened with respect to experiment in the history of science. It is said that past historians studied the ideas of science rather than its materials, its software rather than its hardware. But these claims are overblown, and partly because their authors take a small number of recent idealists as representative of the discipline. I want to flesh out these assertions by looking at a book by the philosopher Ian Hacking and at papers by the historians Frederic Holmes and Roy Porter.

I'll start with Holmes, since he says with authority just about everything I want to say. My source is his “Do We Understand Historically How Experimental Knowledge Is Acquired?” History of Science, vol. 30 (1992), 119-136.

Holmes' speciality was experiment, a topic that is central to the “material culture” of science as historians now understand that label. Holmes begins with a roll-call of authors who have proclaimed the novelty of recent studies of experimental practice. Here is a selection of claims, their authors, and the dates on which they were made:

“Despite the slogan that science advances through experiments, virtually the entire literature of the history of science concerns theory” (Peter Galison, 1987)

“Experiment is a respected but neglected activity ... the process of experimentation is taken to be either unproblematic or uninteresting” (David Gooding, Trevor Pinch and Simon Schaffer, 1989)

“...little attention has been directed at how experiments are actually done, and how they come to be regarded as meaningful” (Jan Golinski, 1990)

Holmes then gives a list of counter-examples to these generalisations. For example, Henry Guerlac declared in 1959 that it was “fallacious to make an arbitrary distinction between ideas and experience, between thought and action, and to treat ideas as if they had a totally independent life of their own, divorced from material reality.”

Other historians practised what Guerlac preached. I. Bernard Cohen published a book in 1956 called Franklin and Newton: An Inquiry Into Speculative Newtonian Experimental Science and Franklin's Work in Electricity as an Examble Thereof. Soon afterwards, Guerlac himself wrote a book called Lavoisier--the Crucial Year: The Background and Origin of his First Experiments on Combustion in 1772. The authors of both books made a point of focusing on experiments as much as, or more than, theoretical speculation.

The next generation, partly inspired by Guerlac and Cohen, went one step further by studying the drafts and lab notes of past scientists rather than their published articles. Others studied long-term changes in scientific domains, with an emphasis on instruments and institutions. Holmes discusses the following books, all published before the three quotes at the beginning of this post:

L. Pearce Williams, Michael Faraday (1966)
Mirko Grmek, Raisonnement expérimental et recherches toxicologiques chez Claude Bernard (1973)
Holmes, Claude Bernard and Animal Chemistry (1974)
Gerald Geison, Michael Foster and the Cambridge School of Physiology (1978)
John Heilbron, Electricity in the 17th and 18th Centuries (1979) [1]
Robert Frank, Harvey and the Oxford Physiologists (1980) [1]
John Lesch, Science and Medicine in France (1984)
A number of articles picked more or less at random from the journal Isis between 1957 and 1966.

Holmes could have gone on. He could have discussed books by Mauric Daumas on the cabinets de physique of the seventeenth and eighteenth centuries, or those by William Middleton on the histories of the thermometer and barometer. Nor should we forget the Harvard Case Studies in Experimental Science, which whatever their flaws certainly contain descriptions of experiments. There is also Stillman Drake's work on Galileo's experimental apparatus, Alistair Crombie's Robert Grosseteste and the Origins of Experimental Science, 1100-1700, and Holmes' own Lavoisier and the Chemistry of Life: an Exploration of Scientific Creativity—all of which predate the three quotes at the start of this post. Many more examples can be found under “Scientific Instruments and Special Techniques” in the Isis Critical Bibliographies from 1955 onwards.

But that's not all. If Heilbron's Electricity in the 17th and 18th Centuries counts as a study of experiment rather than theory, then so do a large proportion of the histories of science written before 1900. Take the first book-length history of electricity, Joseph's Priestley's History and Present State of Electricity, the first edition of which appeared in 1767. The contents page of the third edition can be viewed here, and anyone who glances at it will note that for each figure in his history Priestley promises to describe their “experiments and discoveries.” And indeed, in the body of the book he describes both.

As a work of history, Heilbron's book is greatly superior to Priestley's. But the difference between them is certainly not that Priestley focuses on theory whereas Heilbron focuses on experiment. Granted, Heilbron goes into more detail than Priestley for each major eighteenth-century electrical experiment; but Heilbron also goes into more detail than Priestley for each major theory of that period.

Heilbron himself has written that the “experimental turn” is overrated. Here is the first paragraph of his review of two books from 1989, The Uses of Experiment and The Development of the Laboratory [2].

During the last few years, a few historians and sociologists of science have made the discovery that experiment and measurement are important and problematic parts of the scientific enterprise. The dazzle of this illumination carried conviction of its novelty; and the editors of The uses of experiment open their preface with the astonishing claim that “experiment is a respected but neglected activity.” A few lines later they specify not experiment, but “the process of experimentation" as the neglected subject, only to return immediately to their gambit: “the neglect of experiment is symptomatic of a prejudice against practical activity in favour of speech acts.” Being editors, however, they are condemned to commit speech acts. These include calling attention to rhetorical devices by which scientists put forward arguments based on experiments and emphasising the logical and epistemological difficulties of confirmation and corroboration of results of experiments and instrumental tests. These themes are important; they receive incisive treatment; but they do not have the freshness advertised.

***

Holmes suggests in passing two explanations for the false advertising that Heilbron describes. One is that the self-proclaimed innovators are studying new aspects of experiment rather than newly studying experiment itself. This becomes apparent in the second part of Holmes' article where he contrasts his own approach to experiment with the “newer” approach as represented by that of Leviathan and the Air Pump, a book that many subsequent historians of experiment have taken as a guide. Holmes draws the contrast with respect to Robert Boyle, the man whose air pump features in the title of the book in question:

Shapin and Schaffer portray an assertive Boyle, mobilizing literary resources to secure assent for matters of fact he has produced through a material technology [ie. the air pump]. I portray a playful, exploratory Boyle, performing experiments that seek answers to questions he poses, often tentative and uncertain... (133)

In other words, the new historians of experiment focus on how experimental knowledge is disseminated rather than how it is acquired. Holmes is interested in the latter, as per the title of the paper under discussion.

Holmes seems to say that the former approach is genuinely new, and he rightly admires Schaffer and Shapin's book. But he denies that it represents an increased emphasis on experiment. At most it is an increased emphasis on a particular aspect of experiment, an emphasis that complements traditional accounts but in no way replaces them. The new approach has its own dangers, he says, such as focusing on a small sample of a scientist's experiments and thereby tearing them from “a rich tapestry of intermingled experimentation and reasoning” (126).

Another of Holmes' suggestions is that often it seems to be philosophers rather than historians who are accused of neglecting experiment. At the same time, Holmes would not have written the article if he did not think that historians were among the targets of the accusation. Two possibilities arise, although Holmes does not discuss them in this particular article. One is that philosophers really have neglected experiment, but that the so-called innovators have unjustly tarred past historians with the idealistic brush. Another possibility is that the new historians of experiment are writing a kind of meta-history of experiment: one which considers experiment at a more abstract level than historians usually do, and therefore has new things to say about experiment even though it differs little from traditional accounts at the level of basic description.

***

There are grains of truth in all of these suggestions. But another explanation, and perhaps a more far-reaching one, is that much of the talk about a neglect of experiment is directed at a small but influential group of recent historians and philosophers of science. This group can indeed be said to have neglected experiment (although they should also be credited with showing just how much experiment depends on theory). The claim that experiment is a new topic—whether that claim is made in 1983 or 2013—is often an error based on an unwarranted extrapolation from this group of recent idealists to all historians and philosophers of science who preceded them.

A glance at two pieces, one by a historian and one by a philosopher, can help to substantiate this thesis.

The first is an influential chapter by Roy Porter, a wonderful and much-mourned historian of medicine, Britain, and the eighteenth century. His “The Scientific Revolution: A Spoke in the Wheel?” appeared in 1986 in Porter and Teich, eds., Revolution in History (CUP). The title of the chapter says that it is about the idea of the Scientific Revolution in the seventeenth century, but it is also about how that idea shaped the discipline of the history of science between about 1940 and 1970. The Scientific Revolution, Porter writes, was

...initially the brain-child and shibboleth of a specific cluster of scholars emerging during the 1940s, including the Russian émigré Alexandre Koyré, [the Cambridge historian Herbert] Butterfield, whose outline history popularized Koyré's work, Rupert Hall, who was Butterfield's pupil, and, a little later, Marie Boas [Hall]. Their scrupulous scholarship and prolific works of synthesis animated an emerging discipline, and laid down a coherent framework for future research (295).

For these historians, science was essentially thought – profound, bold, logical, abstract...Thus idealism has been pervasive, and its implications run even to the interpretation of detailed episodes [such as Alexander Koyré's claim that Galileo didn't perform many of the experiments he described] (296).

The first passage answers the “who” question clearly enough, and the second gives an idea of what sort of “idealism” they had in common. It is precisely the kind of idealism that could justly be accused of neglecting experiment, as the example of Koyré suggests. Granted, Porter's article does not make any bold claims for a turn towards experiment; and he explicitly acknowledges that the idealism of Koyré and his English counterparts was a recent and local phenomenon. But by focusing on those idealists and saying little about their predecessors he sets them up as the chief foil to revisionist accounts of the Scientific Revolution, and by extension to revisionist histories of science in general.

Compare Porter's targets with those in an influential book in philosophy, Ian Hacking's Representing and Intervening: Introductory Topics in the Philosophy of Science (CUP, 1983). A fair summary of the book is that “representing” is a problematic activity that philosophers have written too much about, and that “intervening” is a promising activity that they have written too little about.

Hacking makes some bold claims at the beginning of his discussion of “intervening”:

History of the natural sciences is almost always written as a history of theory. Philosophy of science has so much become philosophy of theory that the very existence of pre-theoretical observations or experiments has been denied. I hope the following chapters might initiate a Back-to-Bacon movement, in which we attend more seriously to experimental science. Experimentation has a life of its own.

This is one of the most cited passages—by historians and sociologists as well as by philosophers—in the “new” literature on experimentation. For example, the first two sentences are quoted near the beginning of Leviathan and the Air Pump to show that “even philosophers are beginning to admit the anti-practice and pro-theory prejudices of their discipline” (17). Having read a fair chunk of the recent literature on experimentation, I am confident that a citation analysis would reveal many more examples.

However on reading Hacking's ideas about “intervening” it becomes clear that his main targets are not philosophers of science per se but Karl Popper, Thomas Kuhn, Paul Feyerabend, and Imré Lakatos. He criticises Popper for thinking that observations are usually made to confirm or refute pre-conceived hypothesis (155); he criticises Kuhn for preferring a theoretician over an experimenter in a book that Kuhn was editing (151-152); he criticises Feyerabend for saying that all observation is theory-laden (173); and he criticises Lakatos for implying that there is no such thing as a crucial experiment (254-261).

Hacking's chief conclusion—that there is such a thing as “pre-theoretical observations or experiments”—is novel with respect to this sequence of recent idealists. But with respect to many earlier philosophers of science, from Robert Boyle to John Stuart Mill, it is a statement of the obvious. Encouraged by Hacking's polemical passages, like the one quoted above, historians and sociologists of science have supposed that his chief conclusion is a piece of earthquake-like novelty that will totally reform the way we do the history and sociology of science. There are some new ideas in Hacking's book, but fewer than is often supposed (and insofar as they are new they are controversial, such as the idea that active experiment furnishes powerful arguments for scientific realism that passive observation does not) [3].

The broader lesson of this post is that historians should be clear about who they are accusing of what. The three quotes at the beginning of this post make claims about a large number of past historians of science. But the books and articles in which those quotes appear do not attempt anything like a systematic survey of the books and articles that they indict. The closest approximations that I have seen of such a survey are by those who reject the indictment, like Holmes, Heilbron, and Brian Baigrie [4]. Among those who make the indictment, the one who supplies the most evidence seems to be a philosopher (Hacking, in Representing and Intervening) rather than any historian. If politeness is the reason for this reticence, it is laudable but misguided: better to be openly critical than to confuse the issue with unsupported insinuations.


[1] To be fair, these two books are sometimes acknowledged as precursors of the experimental turn (eg. Leviathan and the Air Pump, 15).

[2] Medical History, vol. 34, no. 3 (1990), 335 (paywall).

[3] This last point generalises. I am not claiming that there is nothing new about the experimental turn, just that the novelty has been exaggerated and poorly defined and that claims to novelty have not been backed up with much historiographical evidence.

[4] Baigrie, “Scientific Practice: The View From the Tabletop,” in Jed Buchwald, ed., Theories and Stories of Doing Physics (Chicago: University of Chicago Press), chap. 5. Expand post.

Malheureusement je ne traduis plus tous les posts que j'écrit sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici.

Agrandir ce message.