Handbook of psychology volume 7 educational psychology

34

Contemporary Theories of Intelligence

larger left hemisphere better predicted WAIS-R verbal than it

predicted nonverbal ability, whereas in women a larger left

hemisphere predicted nonverbal ability better than it pre-

dicted verbal ability (Willerman, Schultz, Rutledge, & Bigler,

1992). These brain-size correlations are suggestive, but it is

difficult to say what they mean at this point.

Yet another approach that is at least partially bio-

logically based is that of behavior genetics. A fairly complete

review of this extensive literature is found in Sternberg and

Grigorenko (1997). The basic idea is that it should be possible

to disentangle genetic from environmental sources of varia-

tion in intelligence. Ultimately, one would hope to locate the

genes responsible for intelligence (Plomin, McClearn, &

Smith, 1994, 1995; Plomin & Neiderhiser, 1992; Plomin &

Petrill, 1997). The literature is complex, but it appears that

about half the total variance in IQ scores is accounted for by

genetic factors (Loehlin, 1989; Plomin, 1997). This figure

may be an underestimate because the variance includes error

variance and because most studies of heritability have been

with children, but we know that heritability of IQ is higher for

adults than for children (Plomin, 1997). Also, some studies,

such as the Texas Adoption Project (Loehlin, Horn, &

Willerman, 1997), suggest higher estimates: .78 in the Texas

Adoption Project, .75 in the Minnesota Study of Twins

Reared Apart (Bouchard, 1997; Bouchard, Lykken, McGue,

Segal, & Tellegen, 1990), and .78 in the Swedish Adop-

tion Study of Aging (Pedersen, Plomin, Nesselroade, &

McClearn, 1992).

At the same time, some researchers argue that effects of

heredity and environment cannot be clearly and validly sepa-

rated (Bronfenbrenner & Ceci, 1994; Wahlsten & Gottlieb,

1997). Perhaps, the direction of future research should be to

figure out how heredity and environment work together to

produce phenotypic intelligence (Scarr, 1997), concentrating

especially on within-family environmental variation, which

appears to be more important than between-family variation

(Jensen, 1997). Such research requires, at the very least, very

carefully prepared tests of intelligence, perhaps some of the

newer tests described in the next section.

Systems Theories

Many contemporary theories of intelligence can be viewed

as systems theories because they are more complex, in many

respects, than past theories, and attempt to deal with intelli-

gence as a complex system.

The Theory of Multiple Intelligences.

Gardner

(1983, 1993, 1999) proposed that there is no single, unified

intelligence, but rather a set of relatively distinct, indepen-

dent, and modular multiple intelligences. His theory of

multiple intelligences (MI theory) originally proposed seven

multiple intelligences: (a) linguistic, as used in reading a

book or writing a poem; (b) logical-mathematical, as used in

deriving a logical proof or solving a mathematical problem;

(c) spatial, as used in fitting suitcases into the trunk of a car;

(d) musical, as used in singing a song or composing a sym-

phony; (e) bodily-kinesthetic, as used in dancing or playing

football; (f) interpersonal, as used in understanding and inter-

acting with other people; and (g) intrapersonal, as used in

understanding oneself. 

Recently, Gardner (1999) has proposed an additional in-

telligence as a confirmed part of his theory: naturalist intelli-

gence, the kind shown by people who are able to discern

patterns in nature. Charles Darwin would be a notable exam-

ple. Gardner has also suggested that there may be two other

intelligences: spiritual intelligence and existential intelli-

gence. Spiritual intelligence involves a concern with cosmic

or existential issues and the recognition of the spiritual as the

achievement of a state of being. Existential intelligence in-

volves a concern with ultimate issues. Gardner believes that

the evidence for these latter two intelligences is less power-

ful than the evidence for the other eight intelligences. What-

ever the evidence may be for the other eight, we agree that

the evidence for these two new intelligences is speculative at

this point.

Most activities will involve some combination of these

different intelligences. For example, dancing might involve

both musical and bodily-kinesthetic intelligences. Reading

a mathematical textbook might require both linguistic and

logical-mathematical intelligences. Often it will be hard to

separate these intelligences in task performance.

In the past, factor analysis served as the major criterion for

identifying abilities. Gardner (1983, 1999) proposed a new

set of criteria, including but not limited to factor analysis, for

identifying the existence of a discrete kind of intelligence:

(a) potential isolation by brain damage, in that the destruc-

tion or sparing of a discrete area of the brain may destroy or

spare a particular kind of intelligent behavior; (b) the exis-

tence of exceptional individuals who demonstrate extraordi-

nary ability (or deficit) in a particular kind of intelligent

behavior; (c) an identifiable core operation or set of opera-

tions that are essential to performance of a particular kind of

intelligent behavior; (d) a distinctive developmental history

leading from novice to master, along with disparate levels of

expert performance; (e) a distinctive evolutionary history, in

which increases in intelligence may be plausibly associated

with enhanced adaptation to the environment; (f) supportive

Contemporary Theories of Intelligence

35

evidence from cognitive-experimental research; (g) support-

ive evidence from psychometric tests; and (h) susceptibility

to encoding in a symbol system.

Gardner (1993, 1995, 1997) has suggested that the multi-

ple intelligences can be understood as bases not only for

understanding intelligence, but for understanding other kinds

of constructs as well, such as creativity and leadership. For

example, Gardner has analyzed some of the great creative

thinkers of the twentieth century in terms of their multiple

intelligences, arguing that many of them were extraordinarily

creative by virtue of extremely high levels of one of the intel-

ligences. For example, Martha Graham was very high in

bodily-kinesthetic intelligence, T. S. Eliot in linguistic intelli-

gence, and so forth.

The theory of multiple intelligences has proved to be enor-

mously successful in capturing the attention both of the psy-

chological public and of the public in general. Nevertheless,

some caution must be observed before accepting the theory.

First, since the theory was proposed in 1983, there have

been no published empirical tests of the theory as a whole.

Given that a major goal of science is empirically to test theo-

ries, this fact is something of a disappointment, but it cer-

tainly suggests the need for such testing. 

Second, the theory has been justified by Gardner on the

basis of post hoc reviews of various literatures. Although

these reviews are persuasive, they are also highly selective.

For example, there is virtually no overlap between the lit-

eratures reviewed by Gardner in his various books and the lit-

eratures reviewed by Carroll (1993) or Jensen (1998). This is

not to say that his literature is wrong or that theirs is right.

Rather, all literature reviews are selective and probably tend

more to dwell on studies that support the proposed point of

view. A difference between the literature reviewed by Gardner

and that reviewed by Carroll and Jensen is that the literature

Gardner reviews was not intended to test his theory of intelli-

gence or anything like it. In contrast, the literatures reviewed

by Carroll and Jensen largely comprise studies designed

specifically to test psychometric theories of intelligence.

Third, even if one accepts Gardner’s criteria for defining

an intelligence, it is not clear whether the eight or ten intelli-

gences proposed by Gardner are the only ones that would fit.

For example, might there be a sexual intelligence? And are

these intelligences really intelligences, per se, or are some of

them better labeled talents? Obviously, the answer to this

question is definitional, and hence there may be no ultimate

answer at all.

Finally, there is a real need for psychometrically strong as-

sessments of the various intelligences, because without such

assessments it will be difficult ever to validate the theory.

Assessments exist (Gardner, Feldman, & Krechevsky, 1998),

but they seem not to be psychometrically strong. Without

strong assessments, the theory is likely to survive without or

because of the lack of serious attempts at disconfirmation.

Since the theory was first proposed, a large number of

educational interventions have arisen that are based on the

theory, sometimes closely and other times less so (Gardner,

1993). Many of the programs are unevaluated, and evalua-

tions of other programs seem still to be ongoing, so it is diffi-

cult to say at this point what the results will be. In one

particularly careful evaluation of a well-conceived program

in a large southern city, there were no significant gains in stu-

dent achievement or changes in student self-concept as a re-

sult of an intervention program based on Gardner’s (1983,

1999) theory (Callahan, Tomlinson, & Plucker, 1997). There

is no way of knowing whether these results are representative

of such intervention programs, however.

Successful Intelligence.

Sternberg (1997, 1999c, 1999d)

has suggested that we may wish to pay less attention to con-

ventional notions of intelligence and more to what he terms

successful intelligence, or the ability to adapt to, shape, and se-

lect environments to accomplish one’s goals and those of one’s

society and culture. A successfully intelligent person balances

adaptation, shaping, and selection, doing each as necessary.

The theory is motivated in part by repeated findings that con-

ventional tests of intelligence and related tests do not predict

meaningful criteria of success as well as they predict scores on

other similar tests and school grades (e.g., Sternberg &

Williams, 1997).

Successful intelligence involves an individual’s discern-

ing his or her pattern of strengths and weaknesses and then

figuring out ways to capitalize on the strengths and at the

same time compensate for or correct the weaknesses. People

attain success, in part, in idiosyncratic ways that involve their

finding how best to exploit their own patterns of strengths

and weaknesses.

According to the proposed theory of human intelligence

and its development (Sternberg, 1980, 1984, 1985, 1990,

1997, 1999a, 1999b), a common set of processes underlies all

aspects of intelligence. These processes are hypothesized to

be universal. For example, although the solutions to prob-

lems that are considered intelligent in one culture may be dif-

ferent from the solutions considered to be intelligent in

another culture, the need to define problems and translate

strategies to solve these problems exists in any culture.

Metacomponents, or executive processes, plan what to

do, monitor things as they are being done, and evaluate

things after they are done. Examples of metacomponents are

36

Contemporary Theories of Intelligence

recognizing the existence of a problem, defining the nature of

the problem, deciding on a strategy for solving the problem,

monitoring the solution of the problem, and evaluating the

solution after the problem is solved. 

Performance components execute the instructions of the

metacomponents. For example, inference is used to decide

how two stimuli are related, and application is used to apply

what one has inferred (Sternberg, 1977). Other examples of

performance components are comparison of stimuli, justifi-

cation of a given response as adequate although not ideal, and

actually making the response.

Knowledge-acquisition components are used to learn how

to solve problems or simply to acquire declarative knowledge

in the first place (Sternberg, 1985). Selective encoding is

used to decide what information is relevant in the context of

one’s learning. Selective comparison is used to bring old in-

formation to bear on new problems. Selective combination is

used to put together the selectively encoded and compared in-

formation into a single and sometimes insightful solution to a

problem.

Although the same processes are used for all three aspects

of intelligence universally, these processes are applied to dif-

ferent kinds of tasks and situations depending on whether a

given problem requires analytical thinking, creative thinking,

practical thinking, or a combination of these kinds of think-

ing. Data supporting the theory cannot be presented fully

here but are summarized elsewhere (Sternberg, 1977, 1985;

Sternberg et al., 2000).

Three broad abilities are important to successful intelli-

gence: analytical, creative, and practical abilities.

Analytical abilities are required to analyze and evaluate

the options available to oneself in life. They include things

such as identifying the existence of a problem, defining the

nature of the problem, setting up a strategy for solving the

problem, and monitoring one’s solution processes. 

Creative abilities are required to generate problem-solving

options in the first place. Creative individuals typically “buy

low and sell high” in the world of ideas (Sternberg & Lubart,

1995, 1996): They are willing to generate ideas that, like

stocks with low price-earnings ratios, are unpopular and per-

haps even deprecated. Having convinced at least some people

of the value of these ideas, they then sell high, meaning that

they move on to the next unpopular idea. Research shows that

these abilities are at least partially distinct from conventional

IQ and that they are moderately domain specific, meaning

that creativity in one domain (such as art) does not necessar-

ily imply creativity in another (such as writing; Sternberg &

Lubart, 1995). Not all creative work is crowd defying, of

course. Some work is creative by virtue of extending existing

paradigms (see Sternberg, 1999b). 

Practical abilities are required to implement options and

to make them work. Practical abilities are involved when

intelligence is applied to real-world contexts. A key aspect

of practical intelligence is the acquisition and use of tacit

knowledge, which is knowledge of what one needs to know

to succeed in a given environment that is not explicitly

taught and that usually is not verbalized. Research shows

several generalizations about tacit knowledge. First, it is ac-

quired through mindful utilization of experience. What

matters, however, is not the experience, per se, but how

much one profits from it. Second, tacit knowledge is rela-

tively domain specific, although people who are likely to

acquire it in one domain are likely to acquire it in another

domain. Third, acquisition and utilization are relatively in-

dependent of conventional abilities. Fourth, tacit knowl-

edge predicts criteria of job success about as well as and

sometimes better than does IQ. Fifth, tacit knowledge pre-

dicts these criteria incrementally over IQ and other kinds of

measures, such as of personality and of styles of learning

and thinking (McClelland, 1973; Sternberg et al., 2000;

Sternberg & Wagner, 1993; Sternberg, Wagner, Williams, &

Horvath, 1995).

The separation of practical intelligence from IQ has been

shown in a number of different ways in a number of different

studies (see Sternberg et al., 2000, for a review). Scribner

(1984, 1986) showed that experienced assemblers in a milk-

processing plant used complex strategies for combining par-

tially filled cases in a manner that minimized the number of

moves required to complete an order. Although the assem-

blers were the least educated workers in the plant, they were

able to calculate in their heads quantities expressed in dif-

ferent base number systems, and they routinely outper-

formed the more highly educated white-collar workers who

substituted when the assemblers were absent. Scribner found

that the order-filling performance of the assemblers was un-

related to measures of academic skills, including intelligence

test scores, arithmetic test scores, and grades.

Ceci and Liker (1986) carried out a study of expert race-

track handicappers and found that expert handicappers used a

highly complex algorithm for predicting post time odds that

involved interactions among seven kinds of information. Use

of a complex interaction term in their implicit equation was

unrelated to the handicappers’ IQs.

A series of studies showed that shoppers in California gro-

cery stores were able to choose which of several products

represented the best buy for them (Lave, Murtaugh, & de la

Roche, 1984; Murtaugh, 1985). They were able to do so

even though they did very poorly on the same kinds of

problems when the problems were presented in the form of

a paper-and-pencil arithmetic computation test. The same

Contemporary Theories of Intelligence

37

principle that applies to adults appears to apply to children as

well: Carraher, Carraher, and Schliemann (1985) found that

Brazilian street children who could apply sophisticated math-

ematical strategies in their street vending were unable to do

the same in a classroom setting (see also Ceci & Roazzi,

1994; Nuñes, 1994).

One more example of a study of practical intelligence was

provided by individuals asked to play the role of city managers

for the computer-simulated city of Lohhausen (Dörner &

Kreuzig, 1983; Dörner, Kreuzig, Reither, & Staudel, 1983). A

variety of problems were presented to these individuals, such

as how best to raise revenue to build roads. The simulation in-

volved more than one thousand variables. No relation was

found between IQ and complexity of strategies used.

There is also evidence that practical intelligence can be

taught (Gardner, Krechevsky, Sternberg, & Okagaki, 1994;

Sternberg, Okagaki, & Jackson, 1990), at least in some de-

gree. For example, middle-school children given a program

for developing their practical intelligence for school (strate-

gies for effective reading, writing, execution of homework,

and taking of tests) improved more from pretest to posttest

than did control students who received an alternative but

irrelevant treatment.

None of these studies suggest that IQ is unimportant for

school or job performance or other kinds of performance; in-

deed, the evidence suggests the contrary (Barrett & Depinet,

1991; Gottfredson, 1986, 1997; Hunt, 1995; Hunter &

Hunter, 1984; Schmidt & Hunter, 1981, 1993, 1998; Wigdor

& Garner, 1982). What the studies do suggest, however, is

that there are other aspects of intelligence that are relatively

independent of IQ, and that are important as well. A multiple-

abilities prediction model of school or job performance would

probably be most satisfactory.

According to the theory of successful intelligence, chil-

dren’s multiple abilities are underutilized in educational insti-

tutions because teaching tends to value analytical (as well as

memory) abilities at the expense of creative and practical

abilities. Sternberg, Ferrari, Clinkenbeard, and Grigorenko

(1996; Sternberg, Grigorenko, Ferrari, & Clinkenbeard,

1999) designed an experiment in order to illustrate this point.

They identified 199 high school students from around the

United States who were strong in either analytical, creative,

or practical abilities, or all three kinds of abilities, or none of

the kinds of abilities. Students were then brought to Yale

University to take a college-level psychology course that was

taught in a way that emphasized either memory, analytical,

creative, or practical abilities. Some students were matched,

and others mismatched, to their own strengths. All students

were evaluated for memory-based, analytical, creative, and

practical achievements. 

Sternberg and his colleagues found that students whose in-

struction matched their pattern of abilities performed signifi-

cantly better than did students who were mismatched. They

also found that prediction of course performance was im-

proved by taking into account creative and practical as well

as analytical abilities.

In subsequent studies (Grigorenko, Jarvin, & Sternberg,

2002; Sternberg, Torff, & Grigorenko, 1998), students were

taught a subject matter in a variety of ways in order to com-

pare instruction based on the theory of successful intelligence

with other forms of instruction. For example, one set of stud-

ies compared such instruction with instruction based on

critical thinking and instruction based on traditional, mem-

ory-based learning in social studies and science (Sternberg

et al., 1998). Another study compared instruction based on

successful intelligence to traditional instruction in reading

(Grigorenko et al., 2002). Participants in these experiments

ranged from middle-school to high-school levels and covered

the range of socioeconomic levels from very low to very

high. In general, instruction based on the theory of successful

intelligence was superior to the other forms of instruction,

even if tests of achievement measured only memory-based

learning.

At a theoretical level, why should instruction based on the

theory of successful intelligence be more effective than con-

ventional or other forms of instruction? Five reasons have

been proffered. First, instruction based on the theory of suc-

cessful intelligence encourages students to capitalize on

strengths. Second, it encourages them to correct or to compen-

sate for weaknesses. Third, it enables them to encode material

in three different ways, which, by increasing the number of re-

trieval routes to the information, facilitates memory retrieval

later on. Fourth, it encourages elaborative rather than mainte-

nance rehearsal, which results in more elaborated memory

traces for the material. Fifth, it is more motivating to students

because it typically renders the material more interesting than

do conventional forms of presentation.

The theory of successful intelligence has been tested more

extensively than many other contemporary theories of intelli-

gence. Nevertheless, questions remain. For example, even

some who might accept the existence of distinctive creative

and practical abilities might argue that they represent psycho-

logical attributes distinct from intelligence. Second, the

pervasiveness of the general factor in psychological investi-

gations must make one wary of Type I errors in accepting the

notion that the general factor is not truly general, but rather

applies primarily to academic kinds of tasks. Third, there is as

yet no published test that measures the triarchic abilities, and

the research-based tests clearly need further development.

Without published tests, it will be difficult for laboratories