Doing Philosophy Of Science, An Example
    By Massimo Pigliucci | October 6th 2013 08:15 PM | 2 comments | Print | E-mail | Track Comments
    About Massimo

    Massimo Pigliucci is Professor of Philosophy at the City University of New York.

    His research focuses on the structure of evolutionary


    View Massimo's Profile

    I have recently been to the European Philosophy of Science Association meeting, where my colleague Maarten Boudry and I have hosted a symposium on our recently published book on the Philosophy of Pseudoscience.

    I have, of course attended several other sessions and talks, as is customary on these occasions (it is also customary to enjoy the local sights, food and drinks, which I dutifully subjected myself to...).

    One of these talks was entitled "Explanatory fictions and fictional explanations," by Sorin Bangu, of the University of Bergen (Norway). I want to use it as a stimulating example of one way of doing philosophy of science. Before we get into it, however, a couple of crucial caveats. As you probably know, some scientists (Lawrence Krauss immediately comes to mind as a major offender) declare philosophy of science to be useless. By this they mean useless to scientists, as apparently their limited imagination cannot conceive of how something could possibly be interesting if it doesn't contribute to science (Shakespeare, anyone? Jazz?? Soccer???). I have argued for a time now that philosophy of science is interesting in at least three senses:

    1. It is a self-contained exercise in reconstructing and understanding the logic of science. (E.g., discussions of paradigms and scientific revolutions, or what you are about to read below.)

    2. It is useful to science theorizing when it deals with issues at the borderlines between science and philosophy. (E.g., discussions of species concepts in biology, or of interpretations of quantum mechanics in physics.)

    3. It is socially useful either as science criticism or in defense of science, whenever science either makes questionable claims or is under attack by reactionary forces. (E.g., criticism of exaggerated claims by evolutionary psychologists or fMRI enthusiasts, defense against creationism and Intelligent Design so-called "theory.")

    Now, nn. 2 and 3 should be pretty obvious (ok, not to Krauss, but still). The first mode of doing philosophy of science, however, is naturally a bit more obscure to the outsider, as is the case for pretty much any intellectual endeavor (trust me, there is a lot of science being handsomely funded about which you would scratch your head and ask "who cares?"). So what follows is just a taste of philosophy of science done as an intellectual activity in its own right, aiming at reconstructing the logic of how science works. To paraphrase Groucho Marx, this is my example, if you don't like it, I have others...

    The question which got Bangu started is that of how fiction can have explanatory power. And by "fiction" Bangu means pretty much any scientific theory or model, which are by definition human imaginative inventions, i.e., fictions. Scientists, of course, are fine with a positive answer to that question, indeed my bet is that they would scoff at it as a non-question. Traditionally, however, many philosophers have answered in the negative for a variety of reasons. Bangu, however, is cautiously optimistic that one can positively deal with the problem. If you are still with me, let's be clear on what exactly is being attempted here: no philosopher is suggesting that somehow scientists have been wrong all along in using "fictional" accounts in their understanding of the world. The question is logical, not practical: how can a notion that is, strictly speaking, false (a theoretical model, which is always approximate) successfully account for something that is true (the world as it really is)? If this isn't your cup of tea (fair enough), you may want to skip to a more interesting post. If your intellect is even slightly tingled, read on...

    If the way I framed the issue so far still sounds bizarre (and it might), then consider clear cases in which fictions don't, in fact, explain facts. For instance: no, Santa (a fiction) didn't bring the presents (a truth) last Christmas. The general logical point is that fictions cannot explain because falsehoods do not explain. But of course in science we are talking about idealizations and approximations, not outright falsehoods, i.e., fictions "concerned with the truth." Bangu's project, then, is to unpack in what (logical) sense the Santa falsehood differs from the the type of falsehood-concerned-with-truth that scientists traffic in.

    There are several ways of tackling this problem, but the particular starting point considered by Bantu is that in science not just the explanans (i.e., the thing that does the explaining) but also the explanandum (the thing to be explained) has fictional content. But, wait, what does that mean? Are we sliding toward some form of idealism in metaphysics, where reality itself is somebody's (God? The Big Programmer in the Sky?) mental construction? Nothing of the sort (besides, an idealist would simply reply that mental constructions are real, just not physically so!). Instead, Bantu reminded us that data - the raw starting point of any scientific analysis - is immediately shaped by scientists into phenomena, that is, phenomena are constructed from data, they are not "out there," they are posited. To put it into more formal language: fictions in the explananda is what allows the successful use of fictions in the explanans. Bantu refers to this idea as the "Monopoly principle": you can't buy real property with fictional money (well, unless you are Goldman Sachs, of course), but there is no problem in buying fictional property with fictional money...

    Okay, enough with the preliminaries, let's consider an actual example of scientific practice. The one Bantu picked was the answer to the deceptively simple question: why does water boil? The explanandum is water's (or other substances) capacity to undergo "phase transitions." The explanans these days is couched in terms of statistical mechanics. In current practice, a phase transition can be explained by invoking a role for (mathematical) singularities of the function describing the temperature curve of the system transitioning between phases, assuming that the system contains an infinite number of particles. But singularities are "fictional," and of course no real system actually contains an infinite number of particles. Nevertheless, the role of singularities is to represent the phenomenon to be explained, and they do a very good job at it. Moreover, physicists - at least for now - simply do not have a definition of phase transition that doesn't invoke singularities/infinities.

    There are of course a number of further issues raised by Bantu's talk. I have already mentioned that a scientist would immediately point out the difference between idealizations and fictions. It turns out, however, that this only kicks the can a bit further down the road without solving the problem, since now we would have to unpack the (perceived) difference between idealizations and fictions. One could, for instance, think of idealizations as a sub-class of fictions; or maybe one can cash out the idea of idealization in terms of verisimilitude (truth-likeness, which is another philosophically more-difficult-than-you-think idea). And of course there is the broader question of how widely applicable Bantu's account of the relationship between truth and fiction in science actually is: are all scientific theories "fictional" in the philosophical sense of the term?

    However you go about it, two things are important to keep in mind: a) no, this isn't the type of philosophy of science that should concern or worry scientists (who can go on using their tools without having to deal with how those tools logically work); but b) yes, this is an interesting intellectual puzzle in its own right, if your intellectual curiosity happens to be stimulated by logical puzzles and epistemic problems. If not, you can always go back to types 2 and 3 philosophy of science described above.

    Originally on Rationally Speaking, September 4, 2013


    It seems that one of the problems is semantic confusion of using words like "true" or "false" within the scientific context. While true or false may be legitimate elements of a logical argument, when they can actually be established, their use in considering scientific ideas is suspect.

    This can be more readily understood when one considers that scientific information is approximate or accurate (or at least increasingly accurate). To put the aura of true or false around it, completely misconstrues the point. For example, since "pi" is a non-rational number, then there can never be any representation of it that could be considered true. It is always used as an approximation and its usefulness is established precisely by the degree of accuracy one needs for the problem. Just as the concept of infinity can never be determined to be true, since by definition, any "infinity" is perpetually in the process of being created, so therefore it can never have an end to be considered complete or true.

    As a result, it seems that instead of true/false or fiction/non-fiction or even idealized, if the concepts were examined from the perspective of approximation and varying degrees of accuracy the problem can be recognized for what it is.

    But singularities are "fictional," and of course no real system actually contains an infinite number of particles.

    We don't have a problem if we consider singularities and infinities as approximations to render the problems more tractable. This isn't viewing them as "idealized" nor as fictions, but rather as useful approximations that make the mathematics work, which is no different than recognizing that using pi in a calculation doesn't require a supercomputer for the desired accuracy level in every problem.

    I followed this on Rationally Speaking, but I couldn't post because I didn't have a Facebook account or something.

    What puzzles me is this: the starting point of Bantu seems to be applicable on Bantu himself. If his idea has any merit, it's perfectly possible that he's giving a fictional explanation of something that's (at least partly) fictional.

    You write: "Bantu reminded us that data - the raw starting point of any scientific analysis - is immediately shaped by scientists into phenomena, that is, phenomena are constructed from data, they are not "out there," they are posited"

    This looks to me as a "phenomenon", posited by Bantu starting from "raw" data he somehow acquired. (I would like to know how "raw" these data are, by the way.) I get the uncomfortable feeling that we're supposed to accept that scientist create (partially fictional) phenomena and find fictional explanations for it, but that philosophers of science somehow are able to find "real explanations" for "real things".

    There seems to be a strange but fundamental difference between the instruments and techniques of philosophers and the instruments and techniques of scientists. Scientists are making observations, and these are problematic. Philosophers of science observe things - mainly in books or articles they read, I suppose - and these are somehow less problematic.

    But maybe I'm mistaken. Maybe the philosophy of science doesn't claim to be be about "science as it is", but about fictional explanations and descriptions of fictional constructions of phenomena.