Chapter 4


We have been talking in passing about causes, but have mainly been referring to "the solution" of an effect. It is time to make some more definitions:

To explain is to state as a possible fact something which makes some other set of facts not a real contradiction. Another verb that means the same thing as "to explain" is "to account for."

That is, explanations deal with effects, and "make sense" out of them. The set of facts which is the effect-plus-its-explanation is now not a contradictory set of facts--provided the explanation actually is a fact.

Before getting into the implications of this "possibleness," let me refute the silliness that has come from people like Auguste Comte, who wanted to distinguish science from "metaphysics" and "speculation" by saying that the "metaphysical" way of thinking looks for "causes" behind things, whereas science is only looking for "laws," or invariant ways in which things act. Science, in this view, is not interested in why things happen, but in how they happen; it simply describes and does not explain.

In spite of the fact that this view is still held by the vast majority of scientists, it is complete rubbish. Science, among other things, is predictive, and the fact that things have happened this way so far gives you no grounds for expecting that they will happen this way ever again. This was one of Hume's points about his view of "cause," in fact.

That is, does the mere fact that the sun has always risen in the morning allow you to say that it will keep rising in future mornings? Does the fact that so far you have waked up every morning allow you to say that you will keep waking up every morning forever?

Obviously, the two are different. But if "laws" are simple descriptions of what has always happened up to now (or even what has usually happened up to now), then the two "laws" above are on an equal footing, and we could formulate all sorts of "laws" that we know have no predictive value. No human being has blown up the planet in all the millennia of our existence; therefore, we have nothing to worry about from nuclear war. And here's one that has nothing to do with human beings: The earth has never yet been hit by an asteroid big enough to destroy it; and so it never will be.

Now then, what gives laws of the scientific type their predictive value, and makes events that have always happened in a certain way so far have no or little predictive value? Obviously, it is because in the first case, you know something about what has "always happened this way so far" that allows you to say, "Well, it hasn't just always happened this way so far, it always happens this way, because it has to happen this way; and so it will keep happening this way."

But the only possible way you could say this is on the grounds that if it doesn't "keep happening this way," whatever it is you know about the thing in question is contradicted. Or in other words, the prediction is on the basis of the fact that the occurrence of the event in the future is the effect whose explanation lies somewhere in the thing that's acting.

That is, we can predict that the sun will rise in the future, because we know that the explanation of its rising in the past is the rotation of the earth, and there aren't any brakes out there that would stop the earth from rotating. We can't predict that we will keep waking up forever because we know that our body is such that eventually it will run down.

So "laws" are "laws" only if, lurking somewhere in the "description" is an explanation showing why this is a law and not an accident: why it makes sense to say that what's always happened this way will continue to happen this way in this case (given that not everything that has "always happened this way" will continue to happen in this way).

Besides, in point of fact, science is always talking about unobservable "entities" of one sort of another, like electrons, photons, gravity, the libido, the gene, and so on; and as Carl Hempel showed in the 1950s in The Theoretician's Dilemma, attempts to replace them with "sets of observations" got you into impossibilities.

So yes, science deals with explanations, not with "descriptions." It has to. If it's solving problems (as every scientist says he is doing) then by definition it's giving explanations, not simply "describing the way things are."

Having got that out of the way, we can formulate the First Rule of Explanation:

An explanation cannot leave any aspect of the effect unaccounted for.

The reason for this, of course, is that if it does, it is not the explanation of that effect, because some of the effect remains a contradiction (and therefore by definition doesn't make sense); and the explanation is supposed (also by definition) to make sense of the effect.

That sounds trivial, but if you add to it that you might not have noticed all the aspects of the effect you are trying to explain, it is quite possible that your explanation might "fit" all the aspects that you were aware of and still not explain the effect you thought you were explaining.

For instance, Newton's Theory of Universal Gravitation explains not only falling bodies but why things like planets don't fall and move in orbits. Since it is mathematical, it also predicts what those orbits have to look like. Unfortunately, at the beginning of this century, very accurate observations of the planet Mercury showed that the orbit of Mercury was not what Newton's explanation of planetary motion said it was (another way of saying this is, as I mentioned, that Newton's "prediction" of what Mercury's orbit would be was off)(1). That is, there was a tiny aspect of Mercury's orbit that didn't make sense on Newton's explanation of orbital motion; and so his explanation of orbital motion was false. Relativity theory, which explains all that Newton's gravitation theory explained plus this aspect of Mercury's orbit, has since supplanted Newton's theory.

Let me now define another term: Speculation is thinking of an explanation for an effect.

The "step" that scientific method calls "hypothesis" is actually speculation. Science likes to think of this step as "finding" a possible "hypothesis" that implies the "set of events" in question (i.e. a statement of the form "If <hypothesis> then <effect>.")

But you can't "find" explanations as if they were "there" someplace waiting for you. If you could, then obviously the explanations you found would have to be facts. But to take Newton's Gravitation Theory as an example, if he had "found" the force of gravity as an explanation of falling bodies and orbits, it would not be the case (as we now know) that there is no force of gravity, but a warping of space-time. That is, his explanation wasn't something he discovered; he made it up.

Scientific method is actually a way of checking to see whether this explanation you have dreamed up is actually a fact or not; you have to "experiment" to see if it "fits the facts observed" and then "predict new facts" from it and check to see if they are facts; and so on.

It is not our purpose here to follow this checking process, because our method is going to be rather different. I simply want to point out that science in practice recognizes that "forming a hypothesis" is speculation, whatever it chooses to call it.

The problem with speculation is that, since it just expresses what is possibly a fact, which logically would imply the effect in question, then (because of the logic involved, which I don't want to go into in any detail here) there are an infinity of explanations for any effect.(2)

Pure speculation satisfies itself when it has a possible fact that does account for the effect in question. But science, though it uses speculation, is not satisfied with pure speculation. The reason is obvious: The effect in question by itself is a contradiction; but there are no real contradictions; therefore, there is a fact which explains it. That is, one of the infinity of explanations is the true one, and there has to be a true explanation (one that is actually the case) or the effect is a real contradiction which only could have but doesn't actually have anything that makes sense out of it.

So science is not interested in speculation as what could account for the effect in question, but in speculation as a means of getting at whatever in fact does account for the effect. And this is why it uses all these devices like experiments and verifying predictions to check to see how likely it is that the hypothesis speculated about is the fact that actually does do the job of making sense out of the effect that was noticed.

In ordinary life, we are not satisfied with pure speculation either. To go back to the example of the keys, John would not be satisfied with, "You probably dropped them somewhere," and might answer, "Well, I might have, but were they stolen?" Either of these explains how the keys got to be missing; but which of them actually occurred can make a big difference. For one thing, if John dropped them, he can retrace his steps with a hope of finding them; if they were stolen, his house might be in danger of burglary.

So now let me formulate the Second Rule of Explanation:

Any serious explanation must devise some way of ruling out alternative explanations as not actually facts.

How successfully you can do this, of course, lifts what you say down from the ivory tower of speculation into the real world of facts.

Note that it does not follow that the only way to go about this ruling-out process is that of the empirical sciences, however sound that process might be (and remember, it fooled people for over a century with Newton's Theory of Gravitation). But now that we know what scientific method is trying to do by its checking process, we need not treat it as Sacred Dogma. What we need (if we are going to be sensible) is some way of assuring ourselves that the explanation we come up with is the fact, and not one of the infinity of possible facts that might be the fact but isn't.



1. If you want to know what it is, it is the "advance in the perihelion of the planet's orbit." That is, as Mercury goes around in its elliptical orbit, there is one point in that orbit that is closest to the sun (the perihelion). The orbit, however, itself "precesses," or turns around the sun, as when a spinning top as it tips not only spins but begins to move in a circle. The amount of this precession was predicted by Newton's equations, and it turns out that observations proved that the perihelion was not advancing at the rate predicted--even though the error was so small as not be observed until the twentieth century.

2. For those who simply need to know, then briefly, statements of the form "If X then Y" can argue validly from X to Y, or from not Y to not X, but it is a logical fallacy to argue from Y to X. That is, it logically follows that if Chicago is in Illinois, and I am in Chicago, then I am in Illinois; or alternatively if I am not in Illinois, then I am not in Chicago either. But it obviously does not follow that if I am in Illinois, then I am (or am not) in Chicago. The attempt to "argue" to the cause, then, from the effect is a case of arguing from being in Illinois to being in Chicago, and is logically invalid.