Whatever intelligence is, it is more than one thing. Spearman, in the early 1900s, suggested that intelligence is composed of a "g-factor" (general intelligence) and various "s-factors" (specific intelligences). This approach allows variety in the nature of intelligence displayed by different people. An autistic savant, for example, would have a low g-factor but would have some very high s-factors (e.g., Rainman). Thurstone (1938) suggested 7 different components of intelligence (verbal, mathematical, spatial, etc.), again reflecting the fact that intelligence has many facets. Guilford (1967) suggested a 4 x 5 x 6 array of factors comprising intelligence (i.e., 120 different components). More recent theorists have suggested a number of facets closer to 5 - 10. Until a better definition of intelligence is agreed upon there will be little agreement about the number of factors that contribute to intelligence, but all agree that it is not a unitary thing.

Tests of Intelligence. Although psychologists can not define intelligence very well, we can certainly measure it (a good thing, because the operational definition depends on our tests). The first intelligence test was devised by the French psychologist Binet in 1905 for a very specific purpose: to predict the performance in school of young French children (thus from the beginning intelligence tests focussed on academic ability). Binet's test established a child's mental age by having the child complete a series of problems of increasing difficulty, each of which can be solved by an "average" child of a particular age. The most advanced problems that the child can solve represent her mental age - if the child can solve problems that the average 8-yr old can solve, but can not do better than that, then her mental age is 8. Binet divided the mental age by the chronological age, then multiplied this quotient by 100 to eliminate fractions, and the Intelligence Quotient (IQ) was born. Binet's test was very successful; it was translated into English by psychologists at Stanford, becoming the Stanford-Binet, in 1917, and remains a commonly used test of intelligence in children.

There are two clear problems with Binet's test. First, it provides a single score, despite the general consensus that intelligence is composed of more than one thing. Second, it is very time-consuming (i.e., expensive) because it is a one-on-one test: a psychologist spends about 1 ½ to 2 hours administering the test to a single child. The Wechsler Intelligence Scale for Children (WISC) addressed the first of these problems. The WISC consists of 6 verbal and 5 performance subtests, and yields separate scores for both types (i.e., a child gets a verbal IQ and a performance IQ, as well as on overall IQ). The Wechslers devised a comparable test for adults (the Wechsler Adult Intelligence Scale - WAIS). However, both the WISC and the WAIS are expensive, one-on-one tests.

The first pencil and paper, group intelligence test was devised by the U.S. Army, to deal with the large number of recruits being processed in World War I. This Army Alpha Test is notable solely for this historical reason - as a test of intelligence it was an abysmal failure (although this was not recognized at the time). One would expect the distribution of intelligence scores resulting from any good test to be "normal," that is to have most people scoring around the mean (100) with fewer and fewer scores occurring as you move further and further away from the mean. In fact, in the general population one should expect to see about 2/3 of all IQ scores between 85 and 115, about 95% between 70 and 130, and more than 99.5% of all scores between 55 and 145. Notice that this means about 2.5% of the population should fall below 70 - the arbitrary cut-off point for mental retardation. What the Army found was a very large proportion of recruits scoring in the mentally retarded range - clear indication that the test did not reflect reality.  Fortunately, since the Army Alpha Test was introduced numerous good group IQ tests have been developed.

Intelligence Across the Lifespan. There is only a slight correlation between one's IQ measured as a child and adult IQ, at least within the normal range. This fact should raise serious concern about the practice of assigning IQ-based labels to children for the purpose of determining how to educate them. The changes that one sees in IQ as an adult ages depend very much on the method used to measure IQ at various ages. If one does a cross-sectional study (i.e., measures at the same time the IQs of several groups of varied age), one finds that IQ reaches a peak in the 20s, remains constant until the 60s, then declines sharply. A longitudinal study (in which the IQs of the same people are measured repeatedly as they age) reveals a different pattern: IQ increases gradually into the 50s, then declines gradually. Most agree that the longitudinal study provides the more accurate picture.

Sex Differences in Intelligence. Although males and females do not differ in overall IQ in any particular way, there are differences in specific abilities. Females tend to test higher than males (on average) on tests of verbal ability and fine motor skills, and males tend to test higher than females (on average) on tests of mathematical and spatial reasoning. Again, these are averages - there is lots of overlap. There are males who are better than many females on verbal ability, and there are females who are better than many males at math. The question of interest to psychologists is the nature-nurture issue: are these differences innately determined or are they the result of experience? Although intelligence does have both genetic and experiential determinants, we do not yet know which are responsible for the sex differences.

The Nature-Nurture Question and Intelligence. Is intelligence genetically-determined, or does it result from experience? The answer appears to be that both play a role. Studies of individuals who share varying degrees of genetic relatedness, or varying degrees of experiential similarity, have shown that the more similar two people are in both realms the more similar their IQ scores are likely to be. Cyril Burt, a British psychologist, reported numerous studies demonstrating this, and was knighted for his efforts. Unfortunately Burt is now believed to have made up all of his data, as well as making up the colleagues who supposedly assisted him with the research (this is not the way to do science). Nonetheless, many other studies (real studies) have demonstrated that both genes and experience affect intelligence. 

As an example: studies of identical and fraternal twins, raised either together or apart, are especially informative. Identical twins share 100% of their genes; fraternal twins share 50%. Twins raised together share far more experiences than twins raised apart. On average identical twins are more similar in intelligence than are fraternal twins, regardless of whether they are raised together or apart. On average twins raised together are more similar in intelligence than are twins raised apart, regardless of whether they are identical or fraternal. These results demonstrate that both genes and experience play a role and that neither can be discounted.