Tuesday, April 26, 2011

Scepticism and the Philosophy of Science: A Comment on Rowbottom

In the thread on the Synthese affair, Darrell Rowbottom raises a point that deserves wider discussion. In her piece on what's-his-name, Barbara Forrest makes a distinction roughly between faith, which is directed at particular propositions, and science which is a method of inquiry. In response to my defence of Forrest, Rowbottom writes: "You may well 'find it sophistical to equate faith in God the Creator with "faith" in reason [and] experience'. But that is no argument. I was alluding to the issue of epistemological regress, and the tu quoque argument raised by Bartley in The Retreat to Commitment." He then expands his point as follows: "Is there any position, or belief set, that is not underpinned by faith (or that lacks dogmatic elements)? The worry is that if there is not, then relativism beckons; that it would be acceptable to choose whichever poison one likes (or to stick with whichever poison one inherits)."

Rowbottom has put his finger on a crucial pivot. Suppose I question Galilean Relativity on the grounds that it is refuted by the Michelson-Morley experiment. Is it legitimate to respond: "Oh, but I take GR on faith. And don't you go citing M-M against me. You would thereby display unquestioning faith in the veracity of sense-experience and scientific observation. And this is no better than my unquestioning faith in GR." Tu quoque.

A good bit of General Philosophy of Science is influenced by the desire to provide a response to scepticism that is better than mere "dogmatism". And this is a problem. For it is the business of epistemology to respond to scepticism. Epistemologists generally talk about beliefs as such, including those formed in ordinary non-scientific contexts. GPOSers, on the other hand, are constrained by their discipline to address the question in restricted contexts (e.g., laboratory situations, theory construction). And responses to scepticism generally do not make a lot of sense in restricted contexts.

In my view, it is important to differentiate arguments that work against any empirical proposition, or worse against any proposition at all, from arguments for and against a particular proposition. A scientific argument for or against Galilean Relativity cannot be transformed by mere substitution of terms into an argument for or against the Michelson-Morley experiment, or the Brain-in-a-Vat proposition. A sceptical argument against GR can be so transformed. Call the former kind of argument content-restricted and the latter content-general.

Here's a methodological observation. Philosophy of X type areas of philosophy -- Philosophy of Biology, Philosophy of Mathematics, Philosophy of Physics -- deal with content-restricted arguments. Epistemology is (most often) concerned with content-general arguments. Given that General Philosophy of Science is itself content-restricted -- i.e., it is concerned with scientific content -- should it not concern itself with content-restricted arguments? The tu quoque argument raised by Bartley is content-general. Therefore, it should have no place in General Philosophy of Science, and certainly no place in Philosophy of Biology, or any other Philosophy of X.

In the Synthese affair, Forrest is operating in the context of various content-restricted scientific arguments against Intelligent Design. She implies that these content-restricted arguments are different in kind from arguments supporting Intelligent Design. Her distinction is quite different from mine: her claim is that anti-ID arguments stem from a methodology not from an interest in a particular proposition. Very broadly, however, this suggests that ID theory employs content-general arguments, while anti-ID arguments are content-restricted. There is some justification for the latter position, but it would take a lot of hard analysis to provide it. (Forrest herself doesn't provide this, but then, she is not doing epistemology.) My point is much easier to make: Rowbottom's argument against Forrest is completely content-general. His argument is exactly the same as the tu quoque argument above, defending Galilean Relativity against Michelson-Morley.

It is my contention that Rowbottom's argument has no place in the Philosophy of Biology. But actually the point that I am making has wider consequences. A good bit of General Philosophy of Science -- operationalism, anti-realism, relativism, etc. -- is a response to content-general arguments. In my view, this is misplaced.


  1. Mohan, isn't "content-restricted" - in Kuhian terms - working inside the "paradigm"? But the advocates of ID (a view that I think is, scientifically and epistemologically speaking, a non-starter) want to change the paradigm! So, this distinction seems to beg the question.

    "In the Synthese affair, Forrest is operating in the context of various content-restricted scientific arguments against Intelligent Design. She implies that these content-restricted arguments are different in kind from arguments supporting Intelligent Design."

    Right, but isn't that circular? Of course: IF the paradigm (or research program, to use Lakatos's term) is as given by standard evolutionary theory (plus the assumption: all mechanisms of mutation rates, selection, etc., involve the usual physical forces), then ID is ruled out - for it requires extending evolutionary theory with a weird theological force field ("theons") that influences the motion of DNA molecules. But why not change the paradigm? From the point of view of Kuhnian/Feyerabendian philosophy of science, Dawkins, Forrest et al are dogmatic authoritarians, and the "dissent" of the IDers is being suppressed! (Didn't Steve Fuller defend the ID side in the recent course case?)

    If one says, "ID is ruled out by standard evolutionary theory", this merely says they are in conflict, which everyone already knows. So, I do think one needs to provide *general* reasons for preferring standard evolutionary theory over ID. These reasons will involve aspects of scientific method: observation, reason, simplicity, explanatory power, coherence with other parts of science, and so on.

    Suppose theory A is standard ET (without the "no theon" assumption, and admitting that our knowledge of detailed aspects of certain mechanisms is incomplete) and theory B is theory A plus theons filling in these gaps. Then theory B is a consistent extension of theory A, possibly even a conservative extension, as B seems to make no new predictions over A. But theory A is preferable to its extension B, because the theon extension of a theory is, in general, bad science, for general epistemological reasons, plus given what we already know from molecular physics, chemistry and biology. But this evaluation requires that one discuss general epistemogical questions about theory comparison, scientific method, simplicity, explanatory power, etc.

  2. Hi Jeffrey,

    Content-restricted is not the same as "working inside the paradigm." In fact, the notion has absolutely nothing to do with paradigms, premises, or presuppositions.

    An argument against p is content-general if it can be transformed into an argument against any q by mere substitution of terms. The following argument against the Michelson-Morley results:

    (1) We can't rule out that we are brains in a vat.
    (2) If we are brains in a vat, then Michelson-Morley is a bad result.
    (3) Therefore, we can't rule out that Michelson-Morley is a bad result.

    is equally an argument against the Millikan oil-drop experiment by simple substitution of terms. Thus, it is content-general.

    A content-restricted argument is one that is not content-general. An argument against Michelson-Morley on the grounds that the mirrors used in the interferometers were not perfectly flat is content restricted because it cannot be transformed into an argument against Millikan by simple substitution of terms.

    The idea that ID arguments are different from anti-ID arguments may be controversial, but not question-begging in the way you suggest. Forrest suggests that anti-ID arguments are scientific in that they rely on certain methods; she doesn't say they rely on doctrine. Presumably, the ID side argues that certain complexities in biological organisms cannot be explained by natural selection, and require a creator. The anti-ID side points out how natural selection can account for these complexities -- no question begging there. Forrest argues further that the creator-posit is epistemologically unsound: she's not relying on evolutionary theory in this respect.

    You say: "From the point of view of Kuhnian/Feyerabendian philosophy of science, Dawkins, Forrest et al are dogmatic authoritarians." That's true, at least of Feyerabend's philosophy of science. But Feyerabend uses content-general arguments. And this was what I was objecting to.

    Just to be clear: I don't think content-general arguments are bad arguments as such. As I said, they are in the proper domain of epistemology, and I certainly don't reject epistemology. My contention is simply that content-general arguments are out of place in philosophy of science.

  3. Right, yes - I see what you have in mind better now, cheers. But there's a sense in which all deductively-valid arguments are content-general: for if P1,...Pn implies A, then this remains so under uniform substitution of predicates.

    So, maybe you mean there's some killer "general sceptical" premise involved on the content-general arguments - along the lines of "if we can't rule out that we're brains in vats, we can't rule out that ...". I agree; that's no good. Going from specific criticisms of an experiment (e.g., neutrino mass, gravitons, Higgs for modern cases) to considerations based on Descartes's Meditations is way too big a step! And maybe that's also a good response, as you suggest, to Darrell's objection based on Bartley's CCR.

    But the specific content-general sceptical premise you're objecting to can surely be weakened from heavy-duty Cartesian scepticism, to some formulation of methodological scepticism/fallibilism/revisabilism. The thing is: we do want to have that in our epistemology of science. After all, working scientists routinely accept such meta-claims.

    So, there are different kinds of sceptical arguments that might be invoked in debates about theory change, and Cartesian ones are no good; while maybe Humean, Popperian, Kuhnian and others are ok.

    And the advocates of ID, it seems to me, want to invoke this kind of theory-change revisability principle to suggest that the whole ET paradigm or research program is flawed, because it is based on a paradigm one of whose "core" assumptions is physicalistic naturalism. They wish to change the paradigm.

  4. Hi Jeffrey,

    Well, I should have said "sound argument", or even "plausibly sound argument." Uniform substitution of predicates doesn't preserve soundness.

    I agree that the content-general premise I am objecting to can be weakened to something like: "We can't rule out that this experiment is a mistake" (instead of the brain in a vat premise). And what I am saying is that it is a mistake to have such premises figure in an argument that e.g. "the whole ET paradigm or research program is flawed."

    Of course, it is open to the ID advocate to argue that:

    (1) Naturalism is false.
    (2) Evolutionary theory is based on naturalism.
    (3) Therefore, evolutionary theory should be abandoned.

    That's certainly a content-restricted argument, but it's not clear to me that this is what the ID advocate is arguing. Rather, something more along the lines (highly simplified) of:

    (4) Living things are very complex.
    (5) Very complex things can't be explained by natural selection.
    (6) Therefore, evolutionary theory (which includes TNS) doesn't explain living things.
    (7) Therefore, naturalism is false.

    I said it's possible that the invocation of (5) is content-general, though this would be very hard to show. However, my claim was that Darrel's argument was more along the lines of my original one, involving Cartesian scepticism, and that it is content-general.

  5. Hi Mohan, I've not studied Beckwith's views in detail (hadn't heard of him until two weeks ago! - that's how uninterested I am in these debates), but I think Beckwith gives arguments a bit like the first; and Dembski gives arguments a bit like the second. Beckwith's objections seem to be against naturalism.
    On premise (5), Dembski seems to think it can be justified by arguments based on complexity theory. Elsberry & Shallit have a paper in the Synthese issue on Dembski-style arguments, and there's an earlier piece here:

  6. Thanks for this interesting post, Mohan. It's a pleasure to be invited to do philosophy, rather than politics. I am better at the former than I am at the latter, although this is no accolade!

    There is so much I could say about this, I find it very difficult to know where to start and where to stop. My first reaction would have been the same as Jeff's, so it was great to wake up to see his response and the following exchange.

    In this post, I hope to offer some general remarks. My apologies if they are somewhat disorganised. I will also make another post, a bit later (hopefully tomorrow), in which I will address Forrest's paper in detail, and explain why I do think that discussion of the kind of issue I raised is appropriate _in this specific context_. The short story is that Forrest raises the issue of faith, in demarcating ID and science. I will also note the very title of Forrest's piece; it is on the non-epistemology of ID. The suggestion is that science has one. (As I'll also explain in my post on her paper, Forrest's way of discussing this is imprecise and I find it difficult to understand exactly what she means. Clearly, science doesn't have an explicit epistemology, _qua_ theory of knowledge; and even understanding 'epistemology' as 'theory of knowledge' rather than 'study of knowledge' is odd to me. I find it highly implausible that it - science, as a whole - has an implicit epistemology either. And Forrest offers no argument to the contrary. I am therefore left to conclude that Forrest is really just saying that what scientists do has epistemic merit - as I'll explain, it looks to me like she's assuming some kind of internalism - or is defensible/praiseworthy/not-blameworthy on epistemic grounds.)

    OK - you now see why I am worried about saying too much! - to the general response:

    If my objection had been something like "But we might be BIVs!", then I agree it would have been inappropriate. One might call it a kind of inappropriate context shift; and I say this with a view to the contextualist movement in contemporary epistemology. (It would perhaps be a reasonable transcendent criticism, nevertheless. I am not sure where you stand on this, Mohan, but would like to know.)

    Yet I don't see philosophy as falling into neat compartments, and I don't think it would be good to have a general rule about what counts as acceptable criticism of a paper published in a particular area. In short, I submit that context is king. I accept that sometimes BIV-style criticisms are inappropriate. But I don't think that has anything to do with subject areas, which are largely, in my view, administrative units. I'll allow myself to quote Popper here. (I agree with the flavour, but would add that problems have similarity relations... and hence there may be better and worse ways to carve things up. I say 'I'll allow myself' because I have been trying to avoid being labelled as a Popperian; on the contrary, I depart from his views in many significant ways, e.g. in my anti-realism.)

    [M]y subject does not exist because subject matters in general do not exist. There are no subject matters; no branches of learning – or, rather, of inquiry: there are only problems, and the urge to solve them. A science such as botany or chemistry (or say, physical chemistry, or electrochemistry) is, I contend, merely an administrative unit… even serious students are misled by the myth of the subject. (Popper, _Realism and the Aim of Science_, 5)

    So I see general philosophy of science and epistemology (understood very broadly, to go beyond 'knowledge' and to cover issues such as understanding, general method of inquiry, the sources of concepts, and so on) as inextricably intertwined. One could also tell a socio-historical tale, _inter alia_. Think of the problem of the empirical basis. Or of the problem of induction (and/or how we should respond to ampliative inferences).


  7. On this note, I find it surprising that you mention anti-realism. I contend that this is (a) not necessarily based on any kind of global sceptical considerations and (b) not necessarily based on any kind of sceptical considerations. I think (a) is easy to argue; some anti-realists are sceptical about our ability to gain knowledge of unobservable/theoretical entities, but not necessarily, say, physical objects and/or sense impressions. As for (b), I will just point to van Fraassen's work. He does not advocate agnoticism about the unobservable entities discussed in science, for scientists or others. He simply says that belief in the existence of such things is supererogatory.

    In closing, I would also like to ask whether you think that what you say goes equally for, say, metaphysics or metaethics. For example, if one objects to a discussion in a paper on philosophy of physics, by appeal to general metaphysical theories about properties or laws, then is that a mistake of the same kind? (There is generality here in so far as it doesn't matter what properties, or what laws, are being discussed.)

  8. Thanks Darrell.

    Actually, I have been thinking a lot about ways to bracket scepticism. So in a paper of which I have a draft, I define a notion of "empirical certainty" along the following lines: p is empirically certain if the epistemically possible scenarios that falsify p are content-general. (I don't use this terminology.) This notion works against brain-in-vat scenarios, but it works pretty well against much more moderate scenarios as well.

    The unfortunate thing is that there is no easy way to generalize the notion and define say "empirically probable", etc. But still the distinction between content-general and content-restricted is quite rich. In the main post, I was having a go at using it to define acceptable methods in philosophy of science.

    There are other directions I have been exploring, with the help of this and some closely related ideas. I think, for example, that it might be possible to reinstate a limited foundationalism, a distinction between sense-properties and other properties, a new understanding of primary and secondary qualities, and so on.

    In short, then, I am going in a direction exactly opposed to your conviction that there are no "compartments" in philosophy!

  9. Mohan: There is a way of approaching general philosophy of science in which it effectively becomes epistemology and metaphysics. Some of my work has gone that way. For example, my work on underdetermination led me to thinking about scepticism, which led to a paper about Thomas Reid and dogmatism. So perhaps I see more value than you do in general philosophy of science that heads in that direction.

    Nevertheless, I think you are correct to say that there is another way of approaching philosophy of science. The second approach requires (for example) that underdetermination, in order to be a problem for philosophy of science, be something different than scepticism.

    Since these are distinct approaches, there's an ambiguity in many debates about scientific realism. If the debates are really about the deep metaphysics of (for example) electrons, then they quickly open onto questions of general metaphysics. If the debates are just about whether scientific inferences underwrite belief in electrons, then the philosophical big questions can be set aside. I think van Fraassen is sometimes guilty of exploiting this ambiguity.

  10. Hi P.D.,

    You're right: there is a way of thinking about e.g. unobservable entities that impinges on general metaphysics, and a way of thinking about e.g. confirmation that impinges on epistemology. Of course, that way of thinking is not fully exploited if people just talk about electrons or laboratory observation, and not about the full generality of the issue.

    My point is that if you are going to talk about electrons in a fully metaphysical way, what you say should address, or at least be informed by, what metaphysicians say about e.g. unobservable material constitution. But if you are going to talk about current views about the electron, then you don't want to be so informed.

    I think we agree about this. Content-general considerations are out of place in content-restricted domains.

  11. Both my examples and yours were about epistemology, and we agree. One quibble: I am sometimes tempted to be dismissive of metaphysics and so deny that general philosophy of science should ever broaden its questions so much as to require doing deep metaphysics. (Only sometimes.) In any case, I think that the broad epistemic questions are perfectly legitimate.

    The broader point on which we agree, I think, is that philosophy of science done as being about science especially employs a different methodology than philosophy of science done as big-picture philosophy.

  12. It might be worth pointing out to newer readers of the blog that we're revisiting an earlier theme, to some extent, discussed in the following threads (in chronological order):

    Are General Philosophers of Science Becoming An Endangered Species?

    On the decline of general philosophy of science

    Exciting Trends in General Philosophy of Science