New York Times

March 1, 1998

Tests Prove That Nobody's Smart About Intelligence

By GEORGE JOHNSON

After decades of agonizing over the fairness of SAT scores, the differences between male and female mathematical skills and the gaps in IQ between various races and ethnic groups, the notion of intelligence and how to measure it remains more political than scientific, and as maddeningly elusive as ever.

Confronted with a seemingly irreconcilable mix of good and bad news about the state of America's intellectual skills, one is tempted to conclude that the country's dismal aptitude for mathematics carries over to the very people designing and administering the tests. Are Americans getting dumber or smarter? The numbers don't add up to a convincing answer. Last week, results from the Third International Mathematics and Science Study showed American high school seniors near the bottom, compared with some 20 other countries. The outcome would surely have been worse had the overachieving Asian nations chosen to compete.

Even more jarring were the results of the competition between the world's most precocious seniors, those taking physics and advanced math classes. No other country's best and brightest performed more poorly than the Americans.

These results are puzzling when one considers this country's dominance in technological innovation and scientific research. And the news was hard to reconcile with recent reports showing that scores on the math portion of the Scholastic Assessment Tests reached a 26-year high last year, continuing to pull out of the "Great Decline" that began in the early 1960s.

For a variety of reasons -- possibilities include grade inflation, the freewheeling style of the '60s and the increasing economic and ethnic diversity of college-bound students -- scores peaked and then went into a free fall lasting until the early '80s.

SAT results are notoriously difficult to interpret. Other tests, however, also show signs of improving mathematical abilities. Just a year ago, the federally mandated National Assessment of Educational Progress test found that the percentage of 12th graders with at least average math scores had increased to 69 percent from 64 percent the year before. This test is given not only to college-bound seniors but to a broad range of students.

Maybe things are getting better -- at least with the third of the three R's. Scores on the verbal portion of both the SAT and the NAEP remain flat, showing little signs of recovering from the Great Decline.

This roller-coaster ride of encouraging and discouraging reports has taken place against the backdrop of a perplexing phenomenon called "the Flynn effect": all over the world IQ scores are steadily rising. Since at least the late 19th century, right on through the SAT decline, people have been getting smarter, or at least better at taking IQ tests.

This strange brew of statistics is not necessarily inconsistent. "The SAT is an achievement test rather than an intelligence test," said Dr. Ulric Neisser, a Cornell University psychologist and head of a recent American Psychological Association task force on human intelligence. "It is aimed at what kids learned in school. It's perfectly possible for kids to get smarter in an IQ sense -- especially in terms of logical reasoning and analysis -- and yet learn less and less of the substantive curriculum.

"The U.S. ranks near the bottom on schoolchildren's knowledge of math and science," he said, "because our society is not committed to the importance of learning these subjects."

Decade of the Brain

Part of the problem in gauging the gross national mental product is a sheer lack of scientific understanding. This is the next-to-last year of the much-vaunted Decade of the Brain and what is there to show for it? All kinds of data on which parts of the head light up when a person balances a checkbook or listens to classical rock, but very little about how all the chemicals sloshing around in the head combine to produce the neuronal buzz called thinking.

A hundred years ago, in science's blissful ignorance, intelligence seemed naively simple. One can measure kidney or heart function, so why not brain function? And so came the IQ tests, attempts to quantify some kind of general intelligence, which the psychometricians called "g."

The idea is not completely crazy. The brain is presumably some kind of information processor. Why can't the cerebral engineers measure how well the neurons are manipulating data in much the same way computer engineers benchmark the blazing speed of an Apple G3 chip against that of an Intel Pentium 2?

Studies indeed show that the value of g measured on IQ tests correlates somewhat with something called "nerve conduction velocity" and other purported gauges of neurological processing speed.

But the meaning of these tests is as open to interpretation as the SAT, and the idea of a general intellectual skill has been eclipsed over the years by the view that intelligence is a many-faceted thing. Trying to capture it with a single number is like gauging the abilities of a supercomputer by how much heat it emits.

According to the psychologist Robert Sternberg's triarchic model, there are three fundamental kinds of intelligence: analytic (what the IQ tests measure), creative and practical. Kenyan children who are great at identifying medicinal herbs (high in practical intelligence) score low on IQ tests, and vice versa.

In his more extravagant theory of "multiple intelligences," the psychologist Howard Gardner adds even more flavors: linguistic, musical, logical/mathematical, spatial, bodily/kinesthetic and intrapersonal and interpersonal abilities. If you're not a candidate for Mensa, you might be a genius at the tango.

In his best seller, "Emotional Intelligence" (Bantam, 1995), Daniel Goleman argued that things like empathy, self-control, persistence and other affective skills are more important to success in life than IQ.

But the importance of g may be making a comeback. The mysterious Flynn effect, described in "The Rising Curve" (American Psychological Association, 1998), a forthcoming book of essays edited by Neisser, applies not to how well you get along with others or figure-skate but to the old-fashioned definition of intelligence -- how well you do on tests of abstract reasoning. IQ, according to the Flynn effect, is rising by about three points a decade.

Escalating even faster are results of Raven's Progressive Matrices, often billed as the best gauge of g. Takers of the test are shown a 3-by-3 grid. Eight of the nine cells contain geometrical patterns. The challenge is to discern the rules by which the patterns are arranged and then fill in the missing one.

There is nothing trivial about the task, and it is somewhat astounding that people are getting better at it all the time, even while studies show a profound level of ignorance over matters like whether dinosaurs lived at the same time as cavemen or whether the earth orbits around the sun. Can people really be getting sharper and more ignorant at the same time?

Flynn himself (James R., a political scientist in New Zealand) has infuriated many of his colleagues by refusing to believe in his own effect, dismissing it as an artifact. It seems implausible, he argues, that we are so much smarter than our grandparents.

Other psychologists have embraced the unexpected good news, offering a variety of possible explanations. Most compelling is that the increasing complexity of the world is challenging the brain to become more and more agile, stretching the synapses further every year. Video games, programming the VCR -- even the willfully ignorant are exposed to ever more intricate visual patterns and forced to learn more complicated tasks.

Why then did students in the early 1960's do so much better on SATs -- especially on the verbal scores? Maybe they compensated for their slower brains by studying harder, reading books instead of watching black-and-white TV.

While the brain's natural ability to grasp visual patterns has probably been getting more exercise, there is another kind of thinking that doesn't come so easily -- the plodding, step-by-step labor of following through a chain of logical deductions, or composing a coherent paragraph and combining it with others into a compelling argument.

This deliberate, serial thinking, which many consider the hallmark of humanity, doesn't get much of a workout from playing Tetris or watching MTV. And it's the hardest of all to quantify.