Wednesday, May 09, 2007

The science of fruits, nuts, and flakes

Some years ago, not long after reading parts of Carl Sagan's The Demon-Haunted World, a book about the remarkable persistence of superstitions among educated people, I discovered that a highly intelligent woman I know, who has a science degree and has always seemed rationalist in her outlook, believes in astrology. She described the traits of my sign, Aquarius, and they did seem to fit me fairly well. I don't remember her description now, so instead I'll post what Wikipedia says about the sign:
Individuals born under this sign are thought to have a modest, creative, challenging, inquisitive, entertaining, progressive, stimulating, nocturnal, and independent character, but one which is also prone to rebelliousness, coldness, erraticism, indecisiveness, and impracticality.
I have to admit that that does kind of sound like me. I like to think of myself as creative, challenging, and independent. I know I am impractical and indecisive, and unquestionably I am nocturnal. On the other hand, the article also claims that male Aquarians "are often said to tend to be effeminate in appearance." Hey, I may not be the most macho of sorts, but I sure as hell ain't effeminate looking.

But there lies the problem. One of the most obvious features of pseudoscience is allowing subjectivity to influence the testing of predictions. Having people examine their own personalities is notoriously unreliable, since people tend not to have accurate perceptions of themselves. What exactly would constitute a "bad" astrological reading? If you didn't feel the description sounded like you? If your friends didn't think so? And how close does the description have to match your personality in order to prove that its alleged accuracy isn't merely coincidental? (For example, I suspect that there are many creative and independent types not born under Aquarius.) Astrologers never set up any quantifiable boundaries by which their "predictions" can succeed or fail. It's all left up to the whim of the person reading the horoscope.

Still, I can imagine ways in which the claims of astrologers might be tested. Gather a group of people together, and give only some of them their true astrological readings. Give the others a deliberately false reading. For example, the Gemini gets the Scorpio's reading. None of the people know whether they are receiving a correct reading or not. Perhaps none of them know there even are any incorrect readings. Even the person doing the testing doesn't know which ones are correct and which ones aren't. The subjects are then asked to rate their readings, on a scale of one to ten, by how closely they feel it describes them.

Now comes the fun part: compare the reactions of the people who got false readings and the people who got true readings. If the claims of astrology are valid, then the people who got true readings should be substantially more likely to think the readings accurately describe them. If, on the other hand, there isn't much difference between the reactions of the two groups, then the claims of astrology would seem to be bogus.

Has anyone ever tried such an experiment? Do believers in astrology even care? Hey, I know that even this test probably wouldn't pass muster with the scientific community. It still has the problem that people are examining their own personalities, rather than being objectively evaluated by a disinterested outsider.

Of course, many scientists will dismiss astrology out of hand simply because of its premise that celestial bodies many light years away can have a perceptible effect on human behavior. But I'm willing to entertain this premise for the sake of argument, because astrology in principle does make real predictions about observable facts. If those predictions were scientifically confirmed, we'd have to concede that there's something to the system, no matter how absurd its philosophical underpinnings may sound. Skeptics can take comfort from the fact that we're a long way from ever having to confront that possibility.

No comments: