And every time it does this, the "success" of the action, i.e., the 'nonoccurrence' of the shock, further convinces it that lifting its hoof is the "right" reaction. It never learns that the bell is no longer followed by a shock. For all practical purposes, it has acquired a neurotic symptom, persisting in an action that once was appropriate but no longer is. And, it need hardly be said, this kind of problem is by no means limited to animals.

Before long the rat, with its PRACTICAL mind manages "to put two and two together." Since it takes the rat only two seconds to run directly to the food tray, the extra time has to be spent in a way that is basically alien to the rat's normal inclination to head straight for food. Under these circumstances the delay acquires a pseudocausal significance; whatever the rat does during these eight seconds, becomes, in the rat's eyes, the "necessary" action that "produces", or is "rewarded" by, the appearance of food.

These behavior patterns, of course, vary from rat to rat, which gives them a particularly capricious aspect: back-and-forth movements, a certain number of pirouettes to the right or the left, jumps (which the rat may have done purely accidentally at first), are faithfully repeated time after time. And every time the rat finds food in the tray, its belief is confirmed that this particular behavior is what produces the food.

These types of behavior are the obvious equivalent of COMPULSIVE human superstitions, which are often based on the vague belief that they are required by some "divine experimenter."

Professor Alex Bavelas, a noted expert in small-group interaction, has shown in several experiments that this kind of disinformation has a powerful influence on a human being's sense of reality.

In one experiment, two subjects, A and B, are seated facing a projection screen. There is a partition between them so that they cannot see each other, and they are requested not to communicate. They are then shown medical slides of healthy and sick cells and told that they must learn to recognize which is which by trial and error. In front of each of them are two buttons marked "Healthy" and "Sick," respectively, and two signal lights marked "Right" and "Wrong." Every time a slide is projected they have to press one of the buttons, whereupon one of the two signal lights flashes on.

"A" gets true feedback; that is, the lights tell him whether his guess was indeed right or wrong. His situation is one of simple discrimination, and in the course of the experiment, most "A" subjects learn to distinguish healthy from sick cells with a fair degree of correctness (i.e., about 80 percent of the time).

"B's" situation is different.
His feedback is based not on his own guesses, but on A's. Therefore it does not matter what he decides about a particular slide; he is told "right" if "A" guessed right, "wrong" if "A" guessed wrong. B does not know this; he has been led to believe there is an order, that he has to discover this order, and that he can do so by making guesses and finding out if he is right or wrong. But as he asks the "sphinx" he gets very confusing answers because he does not know that the sphinx is not talking to HIM.

In other words, there is no way in which he can discover that the answers he gets are noncontingent -- that is, have NOTHING to do with his questions -- and that therefore he is not learning anything about his guesses. So he is searching for an ORDER where there is none that HE could discover.

A and B are eventually asked to discuss what they have come to consider the rules for distinguishing between healthy and sick cells. "A"'s explanations are simple and concrete; "B"'s are of necessity subtle and complex -- after all, he had to form his hypothesis on the basis of very tenuous and contradictory hunches.

The amazing thing is that A does not simply shrug off B's explanations as unnecessarily complicated or even absurd, but is impressed by their sophisticated "brilliance." "A" tends to feel inferior and vulnerable because of the pedestrian simplicity of his assumption, and the more complicated "B"'s "delusions", the more likely they are to convince A.

(This CONTAGIOUSNESS of DELUSIONS is only too well known outside the communication researcher's laboratory, and we shall later consider some glaring examples.)

**
Before they take a second, identical test (but with new slides), A and B are asked to guess who will now do better than in his first test. All B's and most As say that B will. In actual fact, B shows hardly any improvement, but comparatively speaking, seems to be doing better because A, who now shares at least some of B's obtuse ideas, performs significantly more POORLY than the first time.

***
What Bavelas' ingenious experiment teaches us has far-reaching consequences: it shows that once a tentative explanation has taken hold of our minds, information to the contrary may produce not corrections but ELABORATIONS of the explanation.

This means that the explanation becomes "self-sealing"; it is a conjecture that cannot be refuted. But as Popper has shown [Sir Karl Raimund Popper, "Conjectures and Refutations: The Growth of Scientific Knowledge", Basic Books, 1962], refutability is the condition sine qua non of scientific explanation. CONJECTURES of the kind we are considering here are thus pseudoscientific, superstitious, and, ultimately, in a very real sense, PSYCHOTIC.

As we look at world history, we find that similarly "irrefutable" conjectures have been responsible for the worst atrocities. The Inquisition, ideas of racial superiority, the claim of totalitarian ideologies to have found the ultimate answer, immediately come to mind as examples.