Science’s Useful Fallacy

The expression “the science is settled” has been invoked as a way to end numerous discussions of scientific importance. On issues involving evolution, dietary science, or exercise physiology, it is not uncommon for one side to claim that the research has settled the issue. But, however much evidence there may be for any particular scientific theory, is the science of it ever really “settled”?

Although many scientists don’t like to hear it, the nature of scientific reasoning itself prevents any scientific theory from ever being settled. The problem of the level of certainty in scientific judgments goes much deeper than any specific issue. It has to do with the very kind of logic science must employ in order to come to its conclusions.

To put it bluntly, scientific reasoning is based on a logical fallacy, and because of this fact, science is never settled.

How Scientists Reason

Science largely involves the use of hypothetical arguments. A hypothetical argument looks like this:

If all men are mortal, then Socrates is mortal
All men are mortal
Therefore, Socrates is mortal

There are four possible forms of hypothetical arguments, and only two of them are valid. The example above is a modus ponens (“the way of affirmation”):

If P, then Q
Therefore, Q

We affirm the first part of the first statement (“All men are mortal”) and therefore affirm the second part of the first statement in the conclusion (“Socrates is mortal”).

The other valid form of hypothetical reasoning looks like this:

If all birds fly, then an ostrich can fly
But an ostrich cannot fly
Therefore, it is not true that all birds can fly

This is called modus tollens (“the way of negation”):

If P, then Q
Not Q
Therefore, not P

In modus ponens, we say, “If P, then Q,” then we affirm P, and therefore affirm Q. In modus tollens, we say, “If P, then Q,” then “not Q, therefore not P.” The first is the way of affirmation and the second the way of negation.

The Problem With Scientific Reasoning

So how does scientific reasoning work? It is, as we said, hypothetical in nature. It takes a hypothesis about whether one thing results in another. It then conducts an experiment to see whether, when we have that one thing, we get the other.

In 1916, Albert Einstein had published his theory of general relativity. Among the implications of the theory was that light would bend when it encountered the gravity of a large astronomical object. But the theory had not been confirmed. In 1919, there was an eclipse of the sun in the southern hemisphere. It just so happened that, at the same time, the sun would cross the path of the Hyades star cluster. If Einstein was correct, the eclipse would allow scientists to tell whether the light from this star cluster bent as it came around the sun:

If general relativity is true (P), then light of the Hyades star cluster will bend around the sun (Q)

When the results of the experiment came in there were cheers from many scientists. The light indeed had bent around the sun. The theory (at least this predication of the theory) was born out.

If general relativity is true, then light of the Hyades star cluster will bend around the sun
The light of the Hyades star cluster bends around the sun
Therefore, general relativity is true

But hold on. We said that there were only two valid forms of hypothetical reasoning—modus ponens (If P, then Q; P; therefore Q) and modus tollens (If P, then Q; not Q; therefore not P). But this scientific syllogism doesn’t do either. Rather than argue either of these two ways, it argues as follows:

If P, then Q
Therefore, P

Instead of affirming P (modus ponens) or denying Q (modus tollens), it affirms Q. In formal logic this is known as the “Fallacy of Affirming the Consequent.”

Suppose I were to say:

If it rains, then the pavement will be wet
The pavement is wet
Therefore, it must have rained

This is of the very same form as the general relativity argument. But the conclusion does not logically follow. The pavement could be wet for some other reason than rain. Someone might have turned on the sprinklers. It might be wet because some children were playing in water. There are many other reasons the pavement could be wet, which is why this form of reasoning is usually considered suspect.

The Einstein example seems pretty convincing. But it does not eliminate the possibility that some theory other than relativity could explain the bending of light. In fact, relativity theory has still not been reconciled with quantum physics—one explains some things and the other other things. Einstein himself thought that there must be some other theory, not yet discovered, that would explain everything.

Numerous scientific theories have been victimized by this fallacy: The Q that they thought was being caused by the P was really being caused by an R or an S or a T. This is what prompted Karl Popper to say that we can never justify scientific theories; we can only make efforts to refute them. (In other words, you can prove that your P doesn’t cause your Q, but you can never be completely sure, no matter how many experiments you do, whether it is really P and not something else that is actually causing Q.)

Is science ever proven?

It was this problem that was behind Morris Cohen’s quip: “All logic texts are divided into two parts. In the first half, on deductive logic, the fallacies are explained. In the second half, on inductive logic, they are committed.”

For many years scientists put confidence in experiments that suggested something they called phlogiston was the cause of burning (it turned out to be a reaction between fuel and oxygen); they assumed for a while that there was an ether through which light waves had to travel (then light was discovered to be a photon or a “wave-particle”); and the expanding universe hypothesis has gone back and forth numerous times between the static and the dynamic.

In all of these cases the original P that was thought to cause the Q really didn’t cause it at all, despite the fact that, at least for a while, every time they did P, they got Q. Scientific theories are never “proven” in the strictest sense, but only corroborated.

The fact that the chief mode of scientific reasoning is a fallacy is not an excuse for dismissing science. Far from it. But it should be a lesson to us that, though certain theories may be said to be well-established, the findings of science are always to some extent tentative.

Leave a Reply

Skip to content