Psychologists are finding, to their great surprise, that tests given to infants in the first six months of life can predict their IQs at school age.

This means, say the researchers, that it may be possible to pick out babies who could benefit from special help early in life to forestall learning problems when they start school. It also means that psychologists may have new clues about what constitutes intelligence in the first place.

At a recent meeting sponsored by the National Institute of Child Health and Human Development in Bethesda, psychologists and geneticists stressed that the early predictors of IQ are by no means 100 percent accurate. There is still a lot of room for upbringing and other environmental factors to affect intelligence scores and, in fact, there is every reason to believe that IQ is determined by both heredity and environment.

"This enterprise is not pinning a number on an individual that the individual will carry for the rest of his or her life," said Marc Bornstein of New York University. Instead, such a test is a way of estimating a person's inherited ability to organize and remember information.

Nor do these new findings shed light on a key question concerning IQ tests, namely, whether such tests predict how a person will fare in life. In fact, said David Feldman of Tufts University, intelligence test scores are unrelated to success or happiness. IQ scores predict school performance very well. But, said Feldman, "how well kids do in school is only poorly related to how well they do in life. If your criterion is success or fulfillment or how well you do in a job or marriage, success in school is only poorly correlated."

Bornstein agrees. "A child is more than IQ," he said.

Nonetheless, IQ test scores are important to educators and specialists in child development who want to forestall learning problems. And for the psychologists, the fact that they can make predictions of IQ at all is tremendously exciting. "If we'd had this conference five years ago, we would have written off the first year of life," said Robert Ploman of Pennsylvania State University. "I think this is a dramatic turnaround."

IQ started out to be a true "intelligence quotient" -- a fraction consisting of a child's mental age, as defined by the test, divided by chronological age. Psychologists multiplied that quotient by 100 to get IQ scores. A child whose mental age and chronological age were the same would have an IQ of 100.

Today, no one uses the actual quotient. Instead, psychologists take data from tests of large numbers of people and deem the average score to be 100. They set other scores according to how much they deviate from this average. Persons with IQs of 115, for example, score higher on the test than 84 percent of the population. Those with IQs of 85 score higher than only 16 percent of the population.

The tests that now seem to have predictive value were developed by Joseph Fagan of Case Western Reserve University, who devised them not to predict intelligence, or problem-solving ability, but to determine how much a baby knows.

The test is very simple. A researcher shows a baby a photograph of woman's face, for example. Then, the researcher shows the baby the photo again but puts it beside another photograph of the face of a different woman. The researcher measures how long the baby spends looking at the old photo as compared to the new one.

What the test is really measuring is the baby's visual memory. Babies tend to spend more time looking at things that are new to them. So if a baby remembers that it has seen a face before, he or she will spend less time staring at it. The babies that were most adept at recognizing photos or pictures they had already seen did best on IQ tests when they reached school age.

The relation between the baby memory test and intelligence tests has been independently reported recently by several groups of psychologists, including Bornstein and Susan Rose of Albert Einstein College of Medicine in New York. The investigators think the results may be telling them something about how the human mind processes information. At school age, for example, one of the best predictors of IQ is visual memory. These results with babies are consistent with the school age finding.

Visual memory, Ploman explains, is one measure of "how you take in information. Do you remember what you've seen? Do you use it?"

Rose remarks, "We doubt that what we're seeing is about the continuity of visual memory per se. Clearly, we think we're into information processing, into capturing how the organism deals with information. Dealing with novelty is what life is all about. If we had found that toe length was correlated with IQ, I would not be so interested."

Intelligence tests were developed nearly 100 years ago by Albert Binet, who was asked by the Parisian school system to find a way to identify children who were likely to fall behind in their academic work. Binet's test was successful, and, around the turn of the century, it was translated into English. During World War I, the U.S. government began using it to identify men who were candidates to become military officers, and suddenly, according to Bornstein "the term IQ came into everyone's home."

Psychologists, including Arnold Gesel of Yale University and Nancy Bayley of the University of California at Berkeley, scaled down Binet's intelligence test so it could be given to children and even made an unsuccessful version of it for infants. But psychologists began to ask whether the infant tests tell anything at all about the child's future IQ scores.

Bayley herself did a long-term study of the relationship between IQ test scores of babies and children and reported, in 1949, that the baby tests had no predictive value. Her conclusions, said Bornstein, were in accord with the philosophy of our democratic society that "if you manipulate the environment in the right way, anyone can become anything."

Gradually, starting in the 1960s, psychologists improved their methods of assessing infants and began to learn that babies know and can do much more than was expected. Still, it was assumed that there was so much normal variation in infancy that no tests of babies would have much predictive value. A baby that seemed to be a slow learner at age 4 months, for example, could be a genius at age 6 -- and vice versa. The only way to test that assumption was to follow groups of infants for several years, looking for characteristics that hold constant. It is these results that led to the surprising finds that are now being reported.

The next step is to determine what techniques will help babies with poor visual memories do better in school. Psychologists have some ideas -- more verbal interaction, for example, and generally stimulating all the senses -- but the only way to learn if they are useful is to do long-term studies and follow the babies who receive special help.

At the same time, said Bornstein, psychologists "are trying to be clever in trying to figure out how much a baby knows. Now that we know that they can tell things apart, what else can they do? And what difference does it make to later development?"

Gina Kolata is a reporter for Science magazine.