I. Preliminaries

 

1. Rosenberg begins (p. 22) by saying that science seeks answers to questions and that “other human enterprises” also seek answers to these same questions. Examples?

            Notice that this complicates a little the story about science and philosophy - there are apparently non-scientific answers to some questions. If some of these questions can’t yet be answered by science, then by the early definition of what science is, they were answered by philosophy. But these remarks seem to leave room for other possibilities, e.g., religion, literature, ... ??

            Anyway, then comes a key idea - the difference between science and non-science is in “the sorts of standards that will count” as a good explanation. Notice that we aren’t doing a survey of what scientists accept as explanations. Instead, we have an idea about what they should accept. And we aren’t looking mainly for examples of scientific explanations. Rather, we are looking for a standard, a criterion, an account of what makes some a scientific explanation. We want something that will enable us to rule things in or out.

 

2. It’s important to understand some general points about philosophical methods in thinking about this. R. discusses this on p. 26 in the discussion of “explications.” In general, these are efforts to say precisely just what something means. The particular goal here will be to do this for the idea of a scientific explanation. This is done by stating necessary and sufficient conditions for something being the thing in question. An example - explicate the idea of “telling a lie”. Consider:

 

            S told a lie = S said something false

 

If this is correct, then saying something false is both nec. and suf. for telling a lie.

            This explication of telling a lie is not right. You can object to it by providing a counterexample: an example showing the stated conditions either aren’t necessary (something can be a lie even though it does not satisfy the conditions) or that they aren’t sufficient (something that does satisfy the conditions isn’t a lie). [Another condition might be that the explication must be informative. It can fail to be informative if of the conditions is too obscure (or unhelpful). Also, it can’t be circular. Dictionary definitions are often like this. It might define an explanation as “that which explains”. We haven’t gotten very far.]

            The assumption here is that we have a pretty good idea which things count as cases of what we are trying to analyze - lies, in this case - and we want to make the idea clearer and more systematic. The question is whether the stated conditions do the job.

            This one clearly does not. You show that the definition fails by giving a counterexample. This is a clear cut case showing either that the condition is not nec. or not suf. Honest mistakes show that saying something false is not sufficient for lying. You need to add something like an intent to deceive. Revised definition:

 

S told a lie = a) S said something false; and b) S intended to deceive someone by saying this.

 

Some odd cases raise a question about whether (a) is really necessary.

 

3. Rosenberg says a lot about logical positivism and related ideas in setting things up. It would be good to have at least some idea of what this is about. It is an extremely influential philosophical movement that began in Europe in the 1920s. It’s defenders wanted to make philosophy more like science, and they rejected as nonsense much of the philosophical talk then common “the Nothing nihilates” (Heidegger). Positivists held that all knowledge about the world must be justified by experience. (p. 23) But the more crucial idea is that every meaningful sentence is verifiable. Everything not verifiable was pure nonsense. In part, this was a dismissive reaction to a lot of wacky stuff philosophers and others were saying at the time. Nothing could either prove or disprove it. It also seems to capture an idea that lots of people find sensible. But there were puzzles, too. It called into question aesthetic, moral, and religious sentences. (Notably, it also called into question the meaningfulness of the assertion “All meaningful sentences can be tested by experience.”)

            It also raised questions about the status of mathematics. We allegedly know some mathematical truths, e.g., 2+2=4. Moreover, it isn’t just that this happens to be true. It’s necessarily true. But there’s a question about how we could know that on the basis of experience alone.

 

II. The D-N Model (Hempel)

 

A. See p. 30-31 of R. Here’s another way to say what’s said there:

 

E is an explanation = E contains an explanandum and an explanans such that

1) the explanandum is a logical consequence of the explanans,

2) the explanans contains at least one general law,

3) the explanans has empirical content, and

4) the sentences constituting the explanans are true.

 

Each of these elements requires some discussion.

 

B. An explanation is an argument. We can display an explanation this way:

 

C1, C2, ... Cn - initial conditions
L1, L2, ... Ln - laws of nature
__________
E - explanandum (thing being explained)

 

The initial conditions and laws together constitute the explanans (the thing doing the explaining). The whole argument is the explanation.

 

Go over simple examples from text.

 

It is clear that Hempel takes a particular perspective that we should make clear. Suppose that a child asks about something and you tell a story that alleviates her curiosity. That’s not the sort of explanation being explicated.

 

C. Logical consequence (in the book, this is formulated in terms of a valid deductive argument): Statement Y is a logical consequence of statement X₁-X_n iff necessarily, if X_i-X_n are true, then Y is true. [There are complications here, but we needn’t worry about them.] Give examples of consequences and non-consequences.

 

So, condition (1) has the result that in an explanation you cite factors that guarantee the occurrence the thing explained. Suppose an explanation failed to satisfy this condition. Then the factors cited would leave open the possibility that something else happen. Hence, it wouldn’t fully explain why the thing that did happen happened.

 

D. Laws are true universal generalizations. They are not mere accidental truths. Just leaving this idea unexplained would make Hempel’s account not very informative. The idea is to rule out purported explanations like this: suppose I give a test and it turns out that all the students in the class whose last name begins with ‘S’ get a grade of “B” or better. You ask why Smith got a “B” and I explain it: Well, Smith is in the class, and everyone in the class whose last name begins with ‘S’ got a B or better. [Formulate as valid argument.] That doesn’t provide any real explanation. This generalization is not a law. If, instead, I had cited a psychological law about studying, intelligence, etc. and test performance, then I’d be approaching an explanation.

            More on this next time. Read section 2.3.

 

E. Empirical content: this too is kind of tricky, but the idea is that the statements in the explanation are not just true by definition. Consider the statement “All bachelors are unmarried”. This is true, but it’s a matter of definition. Suppose we ask why Ben is unmarried. Proposed explanation cites this generalization and the initial condition that Ben is a bachelor. [Formulate as valid argument.] This is no scientific explanation of why he is unmarried. But the generalization is no accident. However, it lacks empirical content. A real explanation would have to cite factors that account for his being unmarried. Condition (3) is supposed to deal with this. [Note: the premise saying that Ben is a bachelor does have empirical content. The generalization in this explanation does not. I think that the condition is supposed to require that it does.]

            See also what R. says on p. 31-2 about this.

 

F. True. He means “really true”, “corresponding to how things actually are”. So it’s not an explanation (or a correct explanation?) unless it actually gets things right, that is, unless the statements in the explanans really are true. Note: things that fail this condition can have the proper form for an explanation - they can meet all the “logical” conditions.

 

G. Many familiar statements of explanations are incomplete. See example on p. 209-10 in Hempel.

 

H. Adding statistical explanations. Hempel came to realize that this account was a bit restrictive, in that some explanations were statistical in character. We will get to this later in the week.

 

We will turn next time to some questions and puzzles about this account. One issue - for next time - will be think more carefully about what laws are and how they fit into the picture. Also, we’ll see that there are various proposed counterexamples to Hempel’s account.

 

For now, consider the passage in Rosenberg, pp. 32-3, about the truth requirement. He says that this is “problematical”. The problem is that we can’t know that the proposed laws in any explanation are right, and, given the history of science, we have reason to think that they are not. I don’t think that this is a problem for the account. Suppose the proposed law isn’t right. Then the proposed explanation isn’t really the explanation for what happened.