Wednesday, September 7, 2011

Week 4: Verification & Confirmation

Next week, we’ll be switching gears from the “Context of Discovery” to the so-called “Context of Justification”. An apparently motley collection of questions come up here, which will largely occupy us through fall break: How do scientific theories receive support from hypotheses? How do we gauge the strength of this support? Can we offer any justification for treating evidence the way we do? How does all this bear on whether a theory counts as scientific in the first place?

On Tuesday, we’ll think again about the Hypothetico-Deductive account of confirmation that we encountered back in Week 2. Though Hempel’s description of it was rather plausible, it faces a number of problems. One problem we will be rather quick with: it involves using the idea of verification as a way of separating science from other fields (such as metaphysics). What’s metaphysics? French is brief because the question is tricky (true generalizations about metaphysics are scarce), but let me say a bit more than French does.

Despite what bookstores might have you believe, Metaphysics (in philosophy, anyway) does not concern fortune-telling, divination, or astrology. Rather, it concerns a cluster of topics about certain fundamental features of the world (including ourselves) that are not clearly susceptible to empirical treatment from the sciences. For example, many metaphysicians wonder whether we have free will. While results from physics or neuroscience may bear on this question, it is far from obvious that they’d be able to settle it. Conceptual work looks like it’d be necessary. Ditto for other paradigmatically metaphysical questions, such as when some objects compose other objects, how they persist, whether there are any “abstract objects” (such as numbers or qualities), what the nature of time or causation is, and so on (for more, you might check out the Stanford Encyclopedia of Philosophy (“SEP”) entry on Metaphysics). There’s currently a vigorous debate in philosophy about what the proper relationship is between metaphysics and science, but for a long time many philosophers (particularly, the Logical Positivists) got so sick of metaphysical pronouncements that they attempted to construct ways of showing not only what made science special but what made metaphysics worthless: there was no way of verifying metaphysics claims, they said. Perhaps such claims are even meaningless. They’re like “Green ideas sleep furiously”: while they might seem meaningful at first glance, they don’t really mean anything. How would one verify that green ideas do sleep furiously? (“Well, first you get some green ideas. . . .”)

There are a number of reasons why this tempting idea doesn’t work in general. As French explains, as the fortunes of the verification camp waned, an emphasis on confirmation took its place. This is roughly the context in which we find Hempel working: attempting to clarify how confirmation works. One of the interesting dynamics we’ll take an extended look at over the next few meetings is the tension between the feeling that something like his account is on the right track (at least descriptively!) but finding it devilishly difficult to get the details right.

On Thursday, we will discuss one of the most famously difficult problems in philosophy: the problem of induction and its relevance for scientific confirmation. It is in fact what drives Karl Popper to his fascinating views about science as a series of “conjectures and refutations” — a view we will discuss in more detail in weeks five and six. The problem is generated by asking a simple question: how is inductive inference to be justified? How can we show that the methods we use to infer from specific observations the very general facts (that all emeralds are green, that tigers are carnivorous, that material objects obey F=ma, &c.) that science is awash with are good or reliable methods? The inductive skeptic (spelled ‘sceptic’ in Foster’s article) denies that this is possible. She doesn’t deny that our inductive efforts have been successful. That is not at issue. That is obvious. The question is what, if any, reason such past success gives us for thinking that our inductive methods will continue to be successful. The skeptic offers us a compelling argument that there is nothing about our previous experience that should incline us even toward the probability that things will continue as they have. It turns out that the skeptic’s argument is very difficult to rebut. No one (in my view) has yet done so successfully. However, I’m not ready to give up on inductive inference. The stakes are too high: according to many, the rationality of science hangs in the balance!

Tuesday (9/13): Verification & Confirmation
French, pp. 43–49
Hempel, “Criteria of Confirmation and Acceptability”* [PDF]
Note: The Hempel paper here is optional supplementary reading — it’s really interesting, though, and I will discuss it in class, so I recommend you read it. But I won’t count on you having read it, though I think it’s worth reading. In the future, I will merely mark these sorts of “further reading” assignments with an asterisk.

Questions: (respond to two)
  1. Does “verifying” a theory mean showing that the theory is definitely true? Why or why not?
  2. Try to state the argument against using the verification principle as a demarcation criterion as clearly as you can (and in your own words, of course). 
  3. French mentions “a more plausible” version of the verification stance on p.47: that the “greater then number and variety of verifications the greater the support for the theory and the higher the probability of its being true”. Can you think of ways in which this simple statement is pretty clearly too simple?
  4. Explain in your own words the Quine-Duhem problem (as conveyed by French).

Thursday (9/15): The Problem of Induction
Foster, "The Problem of Induction" [PDF] — note: you can safely ignore for now the bit about Goodman on p. 5 (we’ll get to Goodman properly soon enough)
Popper, selections from The Logic of Scientific Discovery [PDF]*
You might wish to look over pp. 17–23 of French again before diving in to this week’s reading.

Questions: (respond to one)
  1. Why can’t one validly deduce from the premises that unsupported coins have always fallen that this coin will fall if I release it? Explain as clearly as you can.
  2. Why does it not help to rely on a principle about the uniformity of nature? Can you think of reasons other than those mentioned by Foster for being skeptical about such a principle?
  3. Can the inductive skeptic be reasonably interpreted as urging us to be modest about drawing general conclusions from particular matters of fact?
  4. Choose one of the three rebuttal strategies that Foster discusses in §IV: rephrase the debate in your own way, adding clarifications or raising questions you deem appropriate.

No comments:

Post a Comment