Sunday, September 26, 2010

Varieties of Evidence Redux

About a year ago, I posted three blog posts here, arguing that scientific evidence serves a more complex and dynamic set of functions in scientific inquiry than simply supporting hypotheses.  I've finally manage to work the idea out in a form that I'm satisfied with:

The Functional Complexity of Scientific Evidence (Draft)

I'm especially indebted to the commenters on this blog for the content of section 6, including Thomas Basbøll, Greg Frost-Arnold, Gabriele Contessa, and Eric Winsberg.  (I hope I've appropriate credit where credit is due there.  I was a bit stymied in how exactly to refer to a conversation we had on the blog, and so made the acknowledgments there fairly general.  Advice on that point is welcome.)

I hope I've managed to present it in a compelling way and answer the objections in a satisfactory way, even though I'm sure many traditionalist won't be convinced.  The goal in this paper is to motivate the need for more complex, functionalist, dynamic model of evidence in contrast with the oversimplification of the traditional-type model, to set out in detail such a model, to illustrate it with an example, and to reply to some basic objections.  I've got a second paper in progress which applies the basic framework to a variety of problems of evidence, from theory-ladenness and the experiment's regress to "evidence for use" and evidence-based public policy.  My central claim there is that this apparently diverse set of problems all share a set of assumptions, and the strongest way to solve them all is to adopt the dynamic evidential functionalism that I've laid out in this first paper.

One reason that I needed to whip this paper into shape is that I'm presenting on the topic of the sequel at the Pitt workshop on scientific experimentation.  Getting this in final form is part of finishing up that paper.  The working title there is "From the Experimenter’s Regress to Evidence-Based Policy: The Functional Complexity of Scientific Evidence."

If anyone gets a chance to look at the paper, I'd appreciate any comments, here or via email. 


  1. I've just had a quick look at it and I noticed that you're maintaining a rhetorical posture that puzzled me when we last talked about this. Back then you referred to the support model of evidence as a "pet notion" too, and you do so again in this paper (23).

    But you also call it the "traditional model", and (rightly) present it as the received view. Indeed, I think there is a very big difference between a "pet" notion and what can rightly be called a "stock" concept. That is, we're not talking about housepets here but livestock.

    On the farm, there is a big difference between losing a favourite dog and losing a prize bull. The reasons to save the latter are not just personal or sentimental, but have to do with the viability of the whole project. My defense of our received views of evidence in the larger project of epistemology/philosophy of science are carried out in that spirit.

    I'll have a look at that Snow example when I have a bit more time.

  2. Thomas,

    I suspect my inclination to use such terms and your puzzlement and exception at them represents our differing evaluations of the prospects for productivity of the notions in question. When the herd's productivity has long since been far less than the expense necessary to maintain it, it is hard to justify keeping it around beyond a certain sentimental fondness.

    Or, perhaps more in the spirit of my own suggestion, when you have a prize bull, but it is never brought into its proper interaction with cows, with the final aim of producing a certain type of result, then all you have is a pet.

    Anyhow, despite my puns, your point is fair and well-taken. Perhaps it would be advisable to soften my rhetoric here. I will consider the matter when revising.

    Thank you for this and all your prior comments.

  3. Yes, we are dealing, if we are, with an aging stud here, not a sick puppy.

  4. This comment has been removed by the author.

  5. Another little detail (I'm still brooding on Snow):

    One page 19, you say:

    "In the model of inquiry I've been discussing, functionalism guarantees that many different types of things count as evidence: not only particular, observed facts, but also historical developments, statistical analyses, general trends, 'phenomenological' laws, and anything that adequately serves some part of the functional roles of evidence and some stage of the inquiry."

    But surely the "support model" would allow historical developments, statistical analyses, and general trends (at the very least) to count as evidence (for appropriate hypotheses), so long as there was some sense in which they were also "facts".

    So, the result of a survey questionaire for example will emerge from a "statistical analysis" of the reponses. "40% of respondents believe the government is hiding important information about UFOs" is presumably the result of "statistical analysis". It is also perfectly good evidence for the hypothesis that there is widespread mistrust of government in the US. More sophisticated analyses could yield evidence for more sophisticated hypothesis.

    But why do we need the analysis itself, and not just its result, to count as "evidence"? And why would we want to rethink the function of evidence in order "guarantee" that we could count it as such?

  6. Yes, that's true. As I point out in the earlier sections of the paper, a subset of the traditional model is those that accept functionalism (so, reject the essentialism of, e.g., Russell), and so accept that evidence is multi-modal, but are still mono-functionalist. I can see how putting that sentence where I did confuses the issue, however. I should either strike it, or take it out from between the two sentences talking about different functional roles.

    I was using "statistical analysis" a bit loosely, to stand for the result of the analysis, rather than just the process. I thought that was standard usage but maybe I should be more clear.

    Thank you for the careful attention to the paper, Thomas.

  7. I wonder if there's a danger here that you're building a(n admittedly very complex) straw man. After all, you seem to be saying that there already is a position out there that would count as evidence "historical developments, statistical analyses, general trends, 'phenomenological' laws, and anything that adequately serves" as support for a hypothesis.

    If the DEF model has something to add beyond that, it won't be to license general trends as evidence (and I'm actually not really sure that Russell's "essentialism" would rule such trends out if they could be established as "facts").

    It is a very narrow (and I think uncharitable) reading, even of hardcore logical positivism, to say that only "sense-data" could count as evidence. Once you leave essentialism and absolutism out (because, as you rightly say, no one really defends those positions today), you get a "support model" that allows everything you want, while, I would argue, distinguishing evidence from other aspects of inquiry that are worth so distinguishing.

    But I owe you a reading of the Snow case to show that a narrower "support"-oriented definition of evidence will shed more light on the history of cholera than the DEF model. More later.

  8. Thomas,

    It may be that essentialism and absolutism are straw-man positions. Perhaps I real the (classical and logical) empiricists differently that you do, but it seems to me that some of them hold both. And I think so do some contemporary epistemologists.

    I think there are still certainly a lot of non-contextualists/absolutists our there in philosophy, though I don't know that there are many serious ones in philosophy of science. On the other hand, I think that arguing that evidence has a complex functional profile rather than a one-dimensional one, and arguing that questions about evidence have to be temporally-index (or indexed to stages of inquiry) are both pretty challenging claims on the present stage. I can't see how that could be construed as straw-mannery.

  9. I'm not saying your DEF is a straw man; I'm saying that your image of the support model is. Maybe I mean this: by even mentioning essentialism and absolutism (in a phil sci context) you are giving the support model a couple of extra straw arms that you can make a show of deftly chopping off. At first pass, it looks like you've wounded the model you're attacking, but on closer inspection you haven't even drawn blood. That's classic straw-manning: attributing unnecessarily weak arguments to your opponent and rebutting those first.

  10. After writing that comment I knew I had to go back and check your paper. Sure enough, on page 3 you address this concern head on. I think the paper would be stronger if it went only at that "weak version of one of these theses". But I also think it would have to be a good deal stronger to defeat that version of the support model.