I. Preliminary Comments on Causation

We start next on causal arguments and then on to moral arguments. These are probably the most important arguments - what makes things happen and what should we do about it.

Note frequency of causal claims - lots of words are implicitly causal, e.g., “reduces”, “improves”, “kills” ...

As we will see, there are hard questions about exactly how to interpret some important causal claims. When analyzing causal arguments, you have to be extremely careful to identify properly the causal claim made in the conclusion.

 

II. Causal Statements

A. Distinguish singular (particular) causal statements from general causal statements:

 

1. Jones’s smoking 3 packs a day for 30 years caused Jones’s lung cancer. [Singular]
2. Smoking causes lung cancer in people. [General]

 

Singular: refers to individual events that happen at a particular time. General: not about anyone in particular. Relates types of events.

 

B. On the meaning of singular causal claims (not in text)

Consider:

 

3. She got the measles because she was exposed to the measles virus.

4. The plant died because it was not watered for months.

 

In (3), a necessary causal condition is mentioned. In (4), a sufficient causal condition is mentioned. Notice - factor mentioned in (3) is not sufficient, and factor mentioned in (4) is not necessary.

 

In general, E1 causes E2 just in case E1 is a necessary part of a set of conditions sufficient for E2. That is, E1 causes E2 provided there is some set of conditions, K, such that i) E1 is part of K, ii) K is sufficient for E2, iii) K-E1 is not sufficient for E2.

 

Apply this to (3). Identify some of the other elements of the set of conditions that is sufficient. (It’s almost impossible to state the full set of sufficient conditions.)

 

Often, there is some element of the set of conditions that we are most interested in. If the police officer asks you why you were speeding, explaining that you pressed hard on the gas pedal seems not to be the right sort of answer. If the fire inspector is asked why the warehouse burned down, mentioning the presence of oxygen probably misses the point.

 

There are old and hard philosophical questions about our ability to observe causal relations. In a sense, what we directly observe is that one thing happened and then another. I flipped the switch and then the light came on. Did I observe the causal connection? or did I infer it?

 

C. On the meaning of general causal claims

 

There are two ways to interpret general causal claims:

 

(a) You might take (2) to say something like

 

2a. Smoking causes some people to get lung cancer.

 

In effect, (2) has a missing quantifier, according to this interpretation. You might also think that there’s a different quantifier - “lots”. On this interpretation, (2) is true provided some instances of it are true (or quite a few instances, etc.).

 

(b) But there’s another thing to note. Consider this causal statement:

 

5. Wearing a seatbelt causes people to die in auto accidents.

 

On the present interpretation, this seems to be true: in some cases having a seat belt on is an essential part of a set of sufficient conditions. But it is also true that:

 

6. Wearing a seatbelt causes people to survive in auto accidents.

 

Having the seat belt on is also an essential part of a set of sufficient conditions for survival. So, it causes surviving and not surviving!

 

This is not a fact that auto safety experts want to emphasize. When they say “seatbelts save lives” (which is short for (5)), they probably mean to make a comparative claim:

 

5a. Wearing a seatbelt causes survival more than it causes death among people in auto accidents.

 

Similarly, (2), as asserted by the Surgeon General, is intended to make the comparative claim, not just (2a).

 

Key point: It's important to understand the difference. Notice that (a) does not capture the idea people seem to have in mind when they make causal claims. It’s something like (b) that they typically have in mind.

 

Notice that if you read (2) as (2a), then, analogously, “Smoking prevents lung cancer” comes out true (since some smokers die in house fires before they would have gotten lung cancer).

 

III. More on Comparative Causal Claims

 

Suppose that studying from a particular SAT study guide leads to higher grades, but using any other guide works just as well. What should we say about

 

H1) Using the Idiot’s guide to the SATs causes higher scores

 

Some would say that the particular guide is not a cause. But compare: both aspirin and tylenol relieve headaches. Showing that another thing works does not show that the first doesn’t. The content of the study guide might matter, but it could be that a few of them are equally good. I think that H1 might then be true, even if another guide works well also.

 

Still, one might think that (H1) is false if the particular content does not matter - any content will work as well. It is fairly hard to believe that the content of the study guide makes no difference at all. Suppose it contained nothing more than a lot of pictures of birds. Those who reject H1 must mean that all the main available guides work equally well. And, they think, if that’s the case, H1 is false.

 

But that’s puzzling: studying the particular content will make people do better on the tests, you might say. The study guide does work, but any of them will work just as well. So why say that (H1) is false if the particular content does not matter? Compare:

 

(H2) Exercising regularly while wearing pink shorts causes weight reduction.

 

Is this also false? Maybe, since the stated cause is not an essential part of any sufficient condition for the effect. You can drop that out and still get the effect exactly as often. But this is a confusing business. Consider:

 

(H3) Smoking Chesterfield cigarettes causes lung cancer.

 

Is this false too? Here’s a way to understand what’s going on: you are supposed to consider whether the whole set of properties said to be causes are essential parts of sufficient conditions. Cigarette case: {smoking cigarettes, Chesterfield brand, other relevant factors} are sufficient for lung cancer. You can drop the brand out, and it is still a sufficient condition. Hence, smoking Chesterfields is not really a cause. Just smoking is. Similar things happen for (H2) and (H1). But in part this result is a consequence of how you specify the causal properties. Smoking cigarettes is itself a complex matter. Divide it up into, say, smoking tobacco products with nicotine, and then, if it’s the nicotine alone that matters, then, on the this reasoning, smoking cigarettes doesn’t really cause lung cancer either. It’s the nicotine that comes along with it that matters. That is, {getting nicotine, by smoking cigarettes, ...} might be a way to describe the set of causal properties. If so, then smoking isn’t really essential. This line of thought might get the result that using a study guide isn’t really a cause. Rather, it’s thinking about certain kinds of questions that does it. It becomes hard to see whether we will end up saying that any familiar things are really causes. That seems wrong, and it would make things very confusing.

 

I think that it would be better to say that (H1) is true (as are H2 and H3), even if other things can have the same effect. The events stated as causes do regularly lead to the stated effects. But this leads us to an important point about the intended content of causal claims, or at least what people often are trying to convey when they say that one thing causes another. They are making comparative causal claims. The usual idea behind H1 is really something like this:

 

H1': Studying this guide is a more effective cause of doing well on the SAT exam than is studying other guides.

 

This is refuted by results showing that people studying other guides do better (Assuming everything about the comparisons is right.) Similar things are true for the other causal claims. Of course, in saying (H1) you don’t have to be comparing this guide to other guides. The comparison you have in mind may be to using no guide at all. That is:

 

H1": Studying this guide is a more effective cause of doing well on the SAT than is not using any study guide at all.

 

Suggestion: in analyzing causal claims and the arguments for them, assess the causal conclusion and consider what comparative causal claim is at issue.

 

Notice that the kinds of correlation studies we’ve mentioned lend themselves to this very well. E.g., if you are interested in (H1'), then you compare results using various guides. If you are interested in (H1"), then you compare using this guide to using no guide.

 

Of course, getting the desired results won’t necessarily establish the conclusion. Suppose you compare students who used the guide to students who used no guide and the former group does better. If the students got to decide whether to use the guide or not, then it might have been that the more conscientious students chose to use the guide, they would have done better in any case, and (H1') is false even though using the guide is positively correlated with doing well among students who take the SATs.