2

2 Faulty Causal Generalization
People ordinarily speak of one event causing another when the two events belong to two classes related according to the following rule: every event of the first class is closely followed in time by an event of the second class. Imagine two classes of events, say (1) the class of lightning flashes, and (2) the class of thunder claps. The rule implies that if lightning causes thunder, then every flash of lightning will be followed by a peal of thunder. It will be seen that the classes of events are constructed so that there is a paired relationship between members of the first class and members of the second. We know how to construct the first class, relative to the second, since we know how to put each member in a one-to-one correspondence with members of the second class. Do we know how to construct the second class relative to the first? Well, we know that there will be a member of the second class for every member of the first, but we don't know from the rule of construction whether or not there will be some members of the second class left over. In terms of lightning-followed-by-thunder, we know that every instance of lightning will cause thunder, but the rule doesn't tell us whether or not every instance of thunder has been caused by lightning; perhaps some other way of violently disturbing air masses can also cause thunder. Still, accepting the generalization "lightning causes thunder," we should not regard any atmospheric electrical display (say northern lights) as lightning unless it is followed by thunder.
Causal generalizations are not theoretically different from other sorts. Thus they are subject to the ordinary ills of generalization; they can be hasty and selective. Yet, the complexity of the causal relationship is such that errors here are more common and harder to detect.
Some of the complexity arises from the fact that very often there is a class of events when there is no evident class of causes. People develop high fevers, and fever seems a definite phenomenon. Automobile tires go flat, and flat tires are remarkably different in a recognizable way from serviceable, inflated tires. Since we start with a noticeable class of effects, it is a class of causes we wish to discover. In such cases a class of causes is constructed according to the pairing-off rule. It is well constructed if and only if every instance in it is followed in time by an instance of the class of effects. If every instance of a puncture is followed by an instance of flat tire, we can say that punctures cause flat tires. But the class of flat tires, though pretty evident to motorists, is still not so well constructed, considered as effects (of course, in its turn it is a cause of other effects). We don't know that every instance of a flat tire is paired off with a puncture. Maybe this present instance is paired off with a leaky valve.
Since most drivers have wide experience with flat tires, and tires are relatively simple man-made things, few are apt to make a false causal generalization about them. In more complex cases, the causal relationship may be obscured; though the class of effects is unmistakable, it is not well constructed relative to a class of causes. Fever, for instance, is caused by a wide range of infectious diseases; each instance of infection is neatly paired with an instance of fever. Thus the presence of fever is small help to a physician in diagnosis. As "diagnosis" means "ascertaining the cause of an illness," the physician needs a wider range of effects before he can distinguish between, say, a common cold and the onslaught of polio.
From these considerations it can be seen how the most important false generalizations about causes occur by taking the class of effects to be well constructed relative to causes. If by far the most frequent pairing-off of flat tires is with punctures, but still not all, generalizing from this factual base to the statement "Every flat tire is caused by a puncture," is obviously a false generalization. In this form, it would be recognized as one, inviting as it does reference to counter instances: "What about leaky valves?" The mistake is more likely to occur in a concealed form: "That tire must have a puncture." In such a case the false generalization is in a suppressed premise: "Every flat tire is caused by a puncture. That tire is a flat tire. Therefore, that tire must have a puncture." We have just seen that the class of effects is not necessarily well constructed, just because a class of causes is -- in this instance punctures relative to flat tires.
It may be well constructed. The advance of physical science is marked by a progressive narrowing of the classes of effects so that after the pairing-off with the classes of causes, there will be no instances left over. If something comparable were feasible in medicine, doctors would not talk about fever, but about x-fever, y-fever, etc., so that the wide class could be replaced by smaller classes determined in such a way that each had its own unique class of causes, where each instance of one is paired with an instance of the other, with no remainder.
In the ordinary case, speakers are able to say that one thing is the cause of another only as a sort of convenience. They mean it is the cause with a high degree of probability. If someone should say that public ownership of electrical power in Southern California is the cause of the rapid growth of industry there, he presumably would accept the emendation "everything else being equal, such as available transportation, good labor market, etc., between Southern California and some other area." "The cause" here means "a cause selected because it is the important difference." Persons lacking scientific training quite naturally tend to understand "the cause" to mean what it says. When they do not know any other cause, they suppose that what usually happens always happens and argue with irrefutable logic from false premises to what may be a false conclusion.
EXAMPLE COMMENT
Peter develops hay fever whenever exposed to golden-rod. He takes his family for a drive in the country but soon begins to sneeze and cough. "The goldenrod must have bloomed early this year," he exclaims, and turns the car towards home. Peter's statement is an enthymeme. Expanded, his reason for cutting short the drive becomes "The cause of my hay fever is the presence of goldenrod. I now have hay fever. Therefore I am in the presence of goldenrod. Therefore, goldenrod must have bloomed early this year." Peter may be right. But other things may cause the symptoms of hay fever besides goldenrod, as people allergic to one thing are commonly allergic to several. It is wise to turn the car around; the only thing his family might question, if they feel like splitting hairs, is Peter's reasoning, not his action. The example, nevertheless, has its interest, as it shows the common belief that the class of effects is well constructed, simply because the class of causes is, so that one is entitled to make a strict conversion (see #42) from "Every instance of x is a cause of y" to "Every instance of y is an effect of x." This could be so only on the interpretation of "the cause of y is x" as "The only cause of y is x." In the case of hay fever and similar phenomena, this interpretation is almost certainly false.
It is sometimes argued that the ultimate cause of war is human aggressiveness, and thus, as long as human beings are aggressive, there will be wars. Hence world peace organizations and disarmament programs are doomed to failure. The word "ultimate" presumably means a cause behind the proximate causes. The cause of a murder victim's death may be injuries in the organs in the head, in turn caused by contact with the blade of an axe. But "the cause" here refers to the intent to kill on the part of the murderer. Similarly, World War I was touched off by the assassination of an archduke, but historians speak of economic, social, political factors as being causes, and the incident at Sarajevo as being the "occasion" of the ultimatums and mobilizations. Those who propose the above argument are saying that behind these economic and other causes there lies still another cause, which is the cause, namely, the aggressiveness of man. Western man is culturally aggressive, but there is no evidence for asserting this aggressiveness to be incurable. At least theoretically, world peace organizations, like national and civil institutions, could make it possible for communities to live in world harmony as they now so often do in neighborhood harmony. This is incidental. The real trouble with the argument is that it goes too far back in the causal chain. It is as if one should say "The cause of the axe murder is the frustration of the middle class." Many things are interrelated in a spreading causal nexus, but in law and common sense speakers limit their causal generalizations to as few classes as seem sufficient to explain the phenomena and thus to guide their conduct. One says, not that the cause of wars is human aggressiveness, but that a cause of economic and political conflict, which may or may not lead to war, is human aggressiveness. And sometimes, over-aggressiveness will lead to war even where there is no real opposition of interests. This pushing back of causes can go on indefinitely: the cause of war is the nursing habits of mothers -- there is evidence that some aggressiveness may be explained as a result of putting small babies on a rigid feeding schedule!

EXAMPLE	COMMENT
Peter develops hay fever whenever exposed to golden-rod. He takes his family for a drive in the country but soon begins to sneeze and cough. "The goldenrod must have bloomed early this year," he exclaims, and turns the car towards home.	Peter's statement is an enthymeme. Expanded, his reason for cutting short the drive becomes "The cause of my hay fever is the presence of goldenrod. I now have hay fever. Therefore I am in the presence of goldenrod. Therefore, goldenrod must have bloomed early this year." Peter may be right. But other things may cause the symptoms of hay fever besides goldenrod, as people allergic to one thing are commonly allergic to several. It is wise to turn the car around; the only thing his family might question, if they feel like splitting hairs, is Peter's reasoning, not his action. The example, nevertheless, has its interest, as it shows the common belief that the class of effects is well constructed, simply because the class of causes is, so that one is entitled to make a strict conversion (see #42) from "Every instance of x is a cause of y" to "Every instance of y is an effect of x." This could be so only on the interpretation of "the cause of y is x" as "The only cause of y is x." In the case of hay fever and similar phenomena, this interpretation is almost certainly false.
It is sometimes argued that the ultimate cause of war is human aggressiveness, and thus, as long as human beings are aggressive, there will be wars. Hence world peace organizations and disarmament programs are doomed to failure.	The word "ultimate" presumably means a cause behind the proximate causes. The cause of a murder victim's death may be injuries in the organs in the head, in turn caused by contact with the blade of an axe. But "the cause" here refers to the intent to kill on the part of the murderer. Similarly, World War I was touched off by the assassination of an archduke, but historians speak of economic, social, political factors as being causes, and the incident at Sarajevo as being the "occasion" of the ultimatums and mobilizations. Those who propose the above argument are saying that behind these economic and other causes there lies still another cause, which is the cause, namely, the aggressiveness of man. Western man is culturally aggressive, but there is no evidence for asserting this aggressiveness to be incurable. At least theoretically, world peace organizations, like national and civil institutions, could make it possible for communities to live in world harmony as they now so often do in neighborhood harmony. This is incidental. The real trouble with the argument is that it goes too far back in the causal chain. It is as if one should say "The cause of the axe murder is the frustration of the middle class." Many things are interrelated in a spreading causal nexus, but in law and common sense speakers limit their causal generalizations to as few classes as seem sufficient to explain the phenomena and thus to guide their conduct. One says, not that the cause of wars is human aggressiveness, but that a cause of economic and political conflict, which may or may not lead to war, is human aggressiveness. And sometimes, over-aggressiveness will lead to war even where there is no real opposition of interests. This pushing back of causes can go on indefinitely: the cause of war is the nursing habits of mothers -- there is evidence that some aggressiveness may be explained as a result of putting small babies on a rigid feeding schedule!