Looking twice at the history of science

Sunday, May 5, 2013

Saving the symmetry principle, IIIb: truth in the history of science

This post continues my effort to understand the symmetry principle by distinguishing different senses of the claim “people do not believe things because they are true.” As you can see, this is not an easy job: this post adds 5 readings to the 6 discussed in my previous post. But nor is it an exercise in hair-splitting or nit-picking. I'm not suggesting that we need to make these distinctions explicit whenever we discuss the symmetry principle, the nature of scientific truth, or the role of evidence in settling scientific debates. But our discussions of all those topics would be improved if we kept these distinctions in mind when we formulate our claims and when we assess the claims of non-historians. (Readers who are pressed for time may want to skip to the end of this post, where I summarise my 11 readings and draw some morals from them.) Expand post.


  1. Wow! I must admit I did start to drown in all the cases and subcases. Thanks for the summary.

    I wonder if one can disentangle the historical issues from the hard-core, ancient philosophical conundrum, What Is Truth, or does Truth even exist? Or are they inextricably entwined?

    I notice that only a few of your cases use the term "truth-value" or "truth". However, "believe" crops up frequently.

    I am, I confess, quite unsympathetic to those suffering a "truth"-allergy. (Or as you put it, the quietism about truth.) It ties the discussion up in knots. Suppose we ask why Galileo believed that the moon's surface was rough. Well, what do we mean when we say Galilo believed that? Just that Galileo thought it was true! (Or "vero", I guess.)

    Granted the philosophical matters are as slippery as a banana-peel soaked in oil lying on a sheet of ice. Steven Weinberg (in the same essay I quoted last time) gives an excellent defence for employing the term:

    I remarked in a recent article ... that for me as a physicist the laws of nature are real in the same sense (whatever that is) as the rocks on the ground....[Philosopher Richard Rorty] accused me of thinking that as a physicist I can easily clear up questions about reality and truth that have engaged philosophers for millennia. But that is not my position. I know that it is terribly hard to say precisely what we mean when we use words like "real" and "true." That is why...I added in parentheses "whatever that is." I respect the efforts of philosophers to clarify these concepts, but I'm sure that even Kuhn and Rorty have used words like "truth" and "reality" in everyday life, and had no trouble with them. I don't see any reason why we cannot also use them in some of our statements about the history of science. Certainly philosophers can do us a great service in their attempts to clarify what we mean by truth and reality. But for Kuhn to say that as a philosopher he has trouble understanding what is meant by truth or reality proves nothing beyond the fact that he has trouble understanding what is meant by truth or reality.

    1. Yes I think we can and should separate many of the historiographical questions from "hard-core" philosophical questions. This is the point of the 2nd and 3rd paragraphs of my conclusion to the above post. I include #11 as a philosophical question that can be separated from the historiographical question of the merit of the symmetry principle--but as I say in the post, I think it is not obvious how the separation can occur in this case (more on this in my next comment). We have to do a bit of philosophy of history in order to show that that separation exists.

      Like you, I tend to be unsympathetic to those with "truth-allergies." But again it is important different cases and degrees. I'm not sure how many philosophers have asked "What is Truth" or "Does Truth Even Exist"? They tend to ask more well-defined questions like: is it possible that our most strongly-held beliefs are actually false? is there a world that exists independently of our minds? What does it mean to say that a proposition is true? What is the true answer to big questions like "what is the meaning of life?"

      The mildest common form of truth-allergy that I know of is the one endorsed by anti-realist philosophers of science, who think that the empirical adequacy of a theory (roughly, its ability to make true predictions) is no guide to its truth-value. This is a respectable position that is currently endorsed by one of the most distinguished living philosophers of science (Bas van Fraassen). It should not be confused with more radical positions.

      I'm only half convinced by Weinberg's argument. The fact that successfully deploy a word shows only that everyone agrees on which objects and events count as referents of that word and which do not. It does not show that the referents have the properties that are implied by our application of the word to them.

      For example, it may be that everyone agrees to call "true" all those theories that make a greater number of true predictions than competing theories, and to withhold that label from theories that do not meet that criterion. But we may be still be thoroughly mistaken in thinking that the predictive success is a good guide to the truth of the theory.

  2. After rereading your posts in this series, your "Will's Picture" posts, and Vanessa Heggie's original post, I was struck by one thing: I don't remember ever seeing a definition of the Symmetry Principle.

    Why does this matter? Well, your point #11 intrigued me, particular this part: "But there is another sense in which the symmetry principle does appear to stand or fall with the pessimistic induction", i.e., the claim that "today's theories are no better than yesterday's".

    This harkens back to Vanessa's statement:

    So our current scientific beliefs are considered more accurate or better than those of the past, not just different. While that might be true, it's a really unhelpful assumption when we're trying to understand debates and arguments - unhelpful not just for historians, but also for scientists, science communicators and science fans today.

    So Vanessa does not plump for pessimistic induction; she just says that the contrary stance, that today's theories are better, is "unhelpful" in certain contexts.

    If the Symmetry Principle says:

    Symm-1: In examining any historical debate, we should assume that both sides were equally right

    (whatever "right" means), then yes, that does walk lock-step with pessimistic induction.

    But if it reads:

    Symm-2: In examining any historical debate, we should ignore present-day views on which side was right, and apply equal standards to the evidence and motivations we ascribe to each side

    then pessimistic induction parts company from the Symmetry Principle.

    This is close to your point in the 4th paragraph of #11. However, it's more general, since we could still award the palm to one side, after hearing both sides impartially.

    We seem to be shifting from metaphysical "truth & reality" questions, to logical and psychological "why did X believe Y?" questions.

    Was that one of Will's points? That the "relativity of truth" debate is played out, and a distraction from the important historical questions?

    1. 1. You are right that, in the passage you quote, Vanessa seems to say that the symmetry principle is independent of the merit of the pessimistic induction (or of any other argument against the superiority of present-day theories over past ones).

      My reply is three-fold. Firstly, there are other passages in the same post where Vanessa seems to argue for the symmetry principle on the grounds that Newton (for example) may turn out to have been wrong about his law of gravitation. She seems to say something similar in her TB post.

      Secondly, whatever Vanessa's view on the matter might be, I have seen other STS scholars link the symmetry principle to the pessimistic induction (although they have not used that label to describe their argument).

      Thirdly, whatever STS scholars might think, there is a plausible argument in favour of the view that the symmetry principle is inconsistent with believing the present-day scientific theories that have the best evidence in their favour; moreover, this argument has a similar structure to (and may actually be identical to) the pessimistic induction.

      To summarise the argument given in #11 above: the symmetry principle says that, in the past, there was no consistent link between the truth-value of a person's belief and the quality of the evidence they had in favour of that belief. Now, (and here comes the pessimistic induction) given that this link was loose in the past, it must be loose in the present. The conclusion is that, in the present, the fact that we have excellent-seeming evidence for a theory is little or no grounds for believing the theory.

      I think this argument is faulty, but my points are a) it is not obvious why it is faulty, b) the symmetry principle is in trouble unless we can show that the argument *is* faulty, and c) in the meantime we should not be too harsh on non-historians who reject the symmetry principle on the grounds that it is inconsistent with belief in present-day scientific theories.

      2. I think the distinction you make between Symm-1 and Symm-2 is crucial. In fact by making it you have anticipated my next post! (I'll write the post anyway, and look forward to any comments you have on it.)

      However this distinction does not solve the problem at hand. We have to ask why we would want to adopt Symm-2. Presumably it would be because we think that there is no consistent link between the rightness of past beliefs and the quality of the past evidence presented in their favour. And that is precisely the assumption that gets us into trouble when we generalise it from past beliefs to present beliefs.

      3. I am not really trying to answer any "metaphysical 'truth and reality' questions" in this post. I am trying to show that the symmetry principle is easily confused with lots of theses that are in fact independent of the principle, and it so happens that some (but not all) of those theses are of the "metaphysical 'truth and reality'"-kind. So the only reason I mention those theses in this post is to distinguish them from each-other and from the symmetry principle.

      Yes, you are right that Will Thomas thinks that the big philosophical questions are a distraction from the important questions about how we should do history (here Michael Weiss and I are referring to a conversation between Will and I that summarised in the comments of the previous post:

      However by this Will does not mean (I think) that *philosophers* should stop discussing those issues. He does not appear to side with those (such as Bruno Latour, in some moods) who think that "epistemology is dead." He just thinks that those debates have little or nothing to contribute to the task of improving our historiography.

    2. Ok, I'll just say one thing and then wait impatiently for your next post in this series.

      My own argument for Symm-2 is simple intellectual curiosity, of the "head-crawling" ilk. (I.e., the desire to crawl around inside the heads of smart historical figures.) To revert to Schaffer's original example, if Leibniz had a different theory of gravitation from Newton's, then I'm probably going to enjoy leaning what it was and why Leibniz believed in it, simply because Leibniz was a pretty sharp cookie.

  3. I'd like to respond, but I too am going to have to go away and digest all this first. Thank goodness this blog doesn't have a "guillotine" like some others do!

    Meanwhile, however, can I make a sort of "procedural" point? You make several references, Michael, to tweets you have received. Now I'm not a big fan of Twitter - I can't see how you can have a serious discussion in 140 characters (but let's not get into a debate about that!) - but is there any way that these tweets can in future be forwarded to the blog, so we can all see them? Otherwise it's like having a meeting where every so often a couple of people at the front start a separate, whispered conversation which no-one else can hear!

    1. I see what you mean about the disembodied tweet-quotes I have given. For each of those quotes I've now added links that go back to the original twitter thread (although I can't promise that the original conversations will be very information, given the limits of that format and the nature of the subject-matter).

      I look forward to any responses you have.

    2. I wonder if having to fight one's way past the captcha discourages people from posting comment, causing them to tweet instead.

  4. Thanks for doing that, Michael. But what I think would be better would be if responders could be encouraged to respond via this blog so that we only have one conversation. They can still do it in <140 characters!

    I think the response "I don't understand that tweet" somewhat proves my point about Twitter not being the right medium for this sort of thing.

  5. I've taken up jimgrozier's suggestion of asking responders to reply to this post here, rather than on Twitter.

    Here is a numbered summary of my views, for anyone who has just arrived and doesn't have time to read the above post. Hopefully this list will also make it easier to identify points of agreement and disagreement.

    1. There are lots of readings of the claim "people believe things because they are true."

    2. The symmetry principle, properly understood, corresponds to just one of those readings

    3. On many other readings of the claim, it is a true claim

    4. If we are unable to distinguish the different readings, we do not really understand the symmetry principle

    5. The failure of non-historians to make these distinctions is one of the main reasons they are hostile to the symmetry principle. They think, for example, that the symmetry principle implies that the role of evidence in science has been exaggerated (see readings #1, #7 and #9); or that it implies that today's theories are no better than yesterday's (#11).


    Here is my follow-up to a short exchange that just occurred on Twitter (

    Both Rebekah and Vanessa suggested in their tweets that although the distinctions I have been making might be interesting for philosophers, they are not really relevant to the history of science, whether academic or popular.

    My reply to this is, firstly, that some of my 11 readings are far more relevant to the history of science than they are to the philosophy of science. An example is #10, "Historians can explain the past development of things that are defined in present-day terms." In these cases the historian has no excuse for not distinguishing the different readings, or for confounding those readings with the symmetry principle.

    Secondly, I admit that a number of those readings are of little relevance to the history of science. However, in those cases, that is just my point. It is important to distinguish between these different readings, precisely because some of them have little to do with the history of science (and in particular they have little to do with the symmetry principle). To respect the boundary between philosophy and history we need to recognise that boundary, and to do that we need to do a little bit of philosophy (of history and of science).

    1. Some new tweets:

      Vanessa says: "I still think you're reifying a historical methodology into a philosophy & so unsurprisingly finding it wanting."

      This seems similar to the concern that I responded to in my previous comment. I'm not sure what "reifying" means here. If the claim is that I have tried to clarify some concepts surrounding the symmetry principle, then I agree with the claim but don't see why that exercise is irrelevant to historians. It could mean that I've tried to make explicit what historians usually assume implicitly in their practice, but that seems like a worthwhile task to me--it is also what Vanessa was trying to do in her original post.

      The claim may be that a historian who recognised my distinctions would not write history any differently from one who did recognise them. I don't think that is the case for all of the distinctions I have made above. But even if it were the case for all of them, there remains the question of how historians communicate their methods to the public. A conceptual confusion that makes no difference to a person's history-writing can make a big difference to the way non-historians respond to their work.

      In the end, the key question is whether or not any of my arguments for claims 1-5 are bad arguments. If I have done some illegitimate reifying, it should be possible to say which of my claims are wrong and why.

      Vanessa also tweeted: "basically, i agree with Michael Weiss's 'defence' of my argument, and beyond that...I got nothing to add."

      In a comment above, Michael gave a definition of the symmetry principle and an argument for it. The definition was:

      Symm-2: In examining any historical debate, we should ignore present-day views on which side was right, and apply equal standards to the evidence and motivations we ascribe to each side

      The argument was that we should endorse Symm-2 out of intellectual curiosity.

      As I mentioned in my reply to Michael above, I think there is something in his distinction between Symm-1 and Symm-2. But Symm-2 is still quite vague and re-opens the usual cans of worms that get non-historians agitated when they read about the symmetry principle. And they are right to get agitated.

      For example, Symm-2 says that "we should ignore present-day views on which side was right." The whole point of this post and the previous one was to show that there are lots of different ways of incorporating "present-day views on which side was right" into historical research, and that many of those ways are perfectly legitimate. So it is way too general to say that we should ignore present-day view on which side is right.


      Maybe it would be easier to think about specific cases. Take #10 above, "Historians can explain the past development of things that are defined in present-day terms." This is a good case because Vanessa criticised a commenter (Wolfbone) for assuming that historians could legitimately make use of present-day notions of TB when writing histories of TB. I say that Wolfbone's proposal is perfectly legitimate, and perfectly consistent with the symmetry principle. To think otherwise is to misunderstand the symmetry principle. Any thoughts?

    2. Well yes, my Symm-1 and Symm-2 were dashed off quickly, mainly in service of the point about pessimistic induction that immediately follows them.

      I agree that "there are lots of different ways of incorporating 'present-day views on which side was right' into historical research, and that many of those ways are perfectly legitimate".

      I might agree with Mat about the Principle of Charity.

      I look forward to discussing all this further in your next post in the series.

  6. I spent rather a long time trying to write a reply but at this stage of the day no one wants to hear what I have to say about instability of semantic reference.

    I agree whole-heartedly with the emphasis on evidence though, and would go further to say that where the philosophy does useful work for history is in encouraging us to think harder about the quality of historical evidence which we use, and its limits.

    I also rather suspect that sustained reflection on the principle of charity (usefully wiki'd here: does at this point rather more historiographical work than more discussion of symmetry.

    1. Hi Matt! Thanks for your comment, and sorry about the lateness of this reply.

      I agree with you about the importance of thinking about the quality of the evidence possessed by historians alongside the quality of evidence possessed by this or that scientist. One thing that fascinates me about past writing on the symmetry principle--and on historiographical method in general--is the question of the match or mismatch between the alleged epistemic situation of the scientists and the alleged epistemic situation of the people who are studying them.

      For example: on the one hand, the symmetry principle was introduced partly to bring the history and sociology of science into line with the perceived "naturalism" of natural science. On the other hand, the too-frequent over-extension of the symmetry principle puts many historians and sociologists in the awkward position of saying that they are pretty sure that scientists can never be sure about anything. The usual responses to this dilemma are to say with a shrug that we (the historians and sociologists) can't be sure of anything either after all; or that scientists can in fact be sure of many things, but that we should ignore this for the sake of analysis. I don't find either of these responses very satisfactory.

      I think you are right that the principle of charity is a good one for historians to adopt. But I would add that we should apply it symmetrically. By this I mean that we should be equally charitable towards--equally ready to find coherence and rationality in--the social/political/economic motivations *and* the evidential motivations that someone has for believing something.

      Finally, I look forward to hearing your thoughts on the instability of semantic reference. This comments thread is getting a bit convoluted, but if you think those thoughts are relevant to any of the other posts in this series then let rip.

  7. Having sort of digested the latest posts I am going to have a go at responding …

    Re the pessimistic induction – I have always thought this to be rather flawed because it cherry-picks the evidence. Induction, in my understanding of the term, involves arguing from a finite number of instances of something to a universal rule – the example often given being the observation of a number of white swans and of no non-white ones, from which it is concluded that “all swans are white”. But the pessimistic induction looks at the number of theories which have in some sense “failed”, and ignores the ones that are not (yet) deemed to have failed. It then argues that therefore all theories fail. This is like saying I’ve seen 1000 white swans and one black one, therefore all swans are white.

    To be honest I can’t really see any connection between this and the symmetry principle. I would like to have a go at defining that myself.

    First of all let’s look at the “ideal” method of theory choice. I would say it goes something like this:

    “People adopt the simplest theory available which most predicts the widest range of observed phenomena within the required accuracy”.

    Let’s call that “the standard criterion”. (OK, there are a lot of possibly incompatible superlatives there but I’m assuming that in the event of a dispute there is some way of deciding which of two or more theories meets this criterion most closely. For more on incompatible superlatives, see my blog post here:

    This is a more generalised form of the version I suggested last time, which effectively said “having made repeated observations and found that they are consistent with one another and with a certain theory, and less consistent with other theories, people adopt that theory”.

    [Incidentally, note also that it allows us to adopt two or more apparently incompatible theories depending on what "the required accuracy" is. If we aim to predict the range of a projectile capable of travelling 10 km to within 10 metres on flat land we can probably use the constant acceleration equations; if a projectile capable of travelling 100 km to within 1m, we'd better use Newton's law of gravitation; if the trajectory of a proton moving at 0.999c, we'll need special relativity. We would not consider any of these theories to be unconditionally "false"]

    Now instead of saying “people don’t believe things just because they are true”, I would want to say “people don’t believe things purely on the basis of the standard criterion”. That lets in all the other factors (philosophical, social, religious, boody-mindedness etc) that we as historians need to allow for.

    But to me that is not a symmetry principle. I see Michael Weiss’s “Symm-2” as more of a symmetry principle.

    What I have called the “standard criterion” above is consistent with the idea that historians should not have regard to views or evidence not available at the time; the words “available” and “observed” take care of that.

    Hence I might say something like “historians should assume that, at any particular time in history, scientists apply the standard criterion together with other relevant external factors”

    1. Sorry Jim for this much-delayed reply. I think your "standard criterion" is a good approximation of the sorts of factors that interest the people I call internal historians of science. Of course, such people might also be interested in the different weightings that different scientists have given to the factors included in your criterion, eg. theoretical simplicity v. empirical accuracy.

      I've been mulling over your suggestion that the "standard assumption" (as I am going to call the instruction in quotes in the final paragraph of your comment) does not count as a symmetry principle whereas Symm-2 does count. I wonder what the difference is. The standard assumption does seem to be symmetric in the sense that it is meant to apply to all past scientists, not just those who turned out to be wrong. Perhaps the distinctive thing about symmetry principles is that they explicitly make room for unexpected asymmetries, ie. cases where the people who turned out to be wrong were more rigorous in following the standard criterion than the people who turned out to be right.

  8. PS In my definition of "the standard criterion", the word "most" should not be there.

    1. Ah, "theory choice" --- a delightful topic. Is this is still considered THE central issue in HoS? And HPS, for that matter.

      Also, did Kuhn coin the phrase, or was it around before his famous essay, "Objectivity, Value Judgment, and Theory Choice" (in _The Essential Tension_).

      Maybe once our host finishes with the Symmetry Principle posts, he'd do some on theory choice (if the topic isn't "too big").

      Or Jim, if you want post something on your blog, I'd love to join in the discussion.