Psychometric Approaches to Intelligence
Psychologists interested in the structure of intelligence have relied on factor analysis as an indispensable tool for their research. Factor analysis is a statistical method for separating a construct - intelligence in this case -into a number of hypothetical factors or abilities the researchers believe to form the basis of individual differences in test performance. The specific factors derived, of course, still depend on the specific questions being asked and the tasks being evaluated.
Factor analysis is based on studies of correlation. The idea is that the more highly two tests are correlated the more likely they are to measure the same thing. In research on intelligence, a factor analysis might involve these steps: (1 ) Give a large number of people several different tests of ability. (2) Determine the correlations among all those tests. (3 ) Statistically analyze those correlations to simplify them into a relatively small number of factors that summarize people's performance on the tests. The investigators in this area have generally agreed on and followed this procedure, yet the resulting factorial structures of intelligence have differed among theorists such as Spearman, Thurstone, Guilford, Cattell, Vernon, and Carroll.
spearman: theory of g
Charles Spearman is usually credited with inventing factor analysis (Spearman, 1927). Using factor-analytic studies, Spearman concluded that intelligence can be understood in terms of both a single general factor that pervades performance on all tests of mental ability and a set of specific factors, each of which is involved in performance on only a single type of mental-ability test (e.g., arithmetic computations). In Spearman's view, the specific factors are of only casual interest because of their narrow applicability. To Spearman, the general factor, which he labeled "g," provides the key to understanding intelligence. Spearman believed g to be attributable to "mental energy." Many psychologists still believe Spearman's theory to be essentially correct (e.g., Jensen, 1998; see essays in Sternberg & Grigorenko, 2002). The theory is useful in part because g accounts for a sizable, although not fixed, percentage of variance in school and job performance, usually somewhere between 5 % and 40% (Jensen, 1998). Spearman (1923) provided a cognitive theory of intelligence. He suggested that intelligence comprises apprehension of experience (encoding of stimuli), eduction of relations (inference of relations), and eduction of correlates (application of what is learned). He therefore may have been the earliest serious cognitive theorist of intelligence.
thurstone: primary mental abilities
In contrast to Spearman, Louis Thurstone (1887-1955) concluded (Thurstone, 1938) that the core of intelligence resides not in one single factor but in seven such factors, which he referred to as primary mental abili ties: verbal comprehension, measured by vocabulary tests; verbal fluency, measured by time-limited tests requiring the test-taker to think of as many words as possible that begin with a given letter; inductive reasoning, measured by tests such as analogies and number-series completion tasks; spatial visualization, measured by tests requiring mental rotation of pictures of objects, number, measured by computation and simple mathematical problem-solving tests; memory, measured by picture and word-recall tests; and perceptual speed, measured by tests that require the test-taker to recognize small differences in pictures or to cross out a "each time it appear in a string" of varied letters.
guilford: the structure of intellect
At the opposite extreme from Spearman's single g-factor model is J. P. Guilford's (1967, 1982, 1988) structure-of-intellect model, which includes up to 150 factors of the mind in one version of the theory. According to Guilford, intelligence can be understood in terms of a cube that represents the intersection of three dimensions - operations, contents, and products. Operations are simply mental processes, such as memory and evaluation (making judgments, such as determining whether a particular statement is a fact or opinion). Contents are the kinds of terms that appear in a problem, such as semantic (words) and visual (pictures). Products are the kinds of responses required, such as units (single words, numbers, or pictures), classes (hierarchies), and implications. Thus, Guilford's theory, like Spearman's, had an explicit cognitive component.
CATTELL, vernon, and Carroll: hierarchical models
A more parsimonious way of handling a number of factors of the mind is through a hierarchical model of intelligence. One such model, developed by Raymond Cat-tell (1971), proposed that general intelligence comprises two major subfactors - fluid ability (speed and accuracy of abstract reasoning, especially for novel problems) and crystallized ability (accumulated knowledge and vocabulary). Subsumed within these two major subfactors are other, more specific factors. A similar view was proposed by Philip E. Vernon (1971), who made a general division between practical-mechanical and verbal-educational abilities.
More recently, John B. Carroll (1993 ) proposed a hierarchical model of intelligence based on his analysis of more than 460 data sets obtained between 1927 and 1987. His analysis encompasses more than 130,000 people from diverse walks of life and even countries of origin (although non-English-speaking countries are poorly represented among his data sets). The model Carroll proposed, based on his monumental undertaking, is a hierarchy comprising three strata -Stratum I, which includes many narrow, specific abilities (e.g., spelling ability, speed of reasoning); Stratum II, which includes various broad abilities (e.g., fluid intelligence, crystallized intelligence); and Stratum III, a single general intelligence, much like Spearman's g.
In addition to fluid intelligence and crystallized intelligence, Carroll includes in the middle stratum learning and memory processes, visual perception, auditory perception, facile production of ideas (similar to verbal fluency), and speed (which includes both sheer speed of response and speed of accurate response). Although Carroll does not break new ground in that many of the abilities in his model have been mentioned in other theories, he does masterfully integrate a large and diverse factor-analytic literature, thereby giving great authority to his model. Whereas the factor-analytic approach has tended to emphasize the structures of intelligence, the cognitive approach has tended to emphasize the operations of intelligence.
Cognitive theorists are interested in studying how people (or other organisms; Zentall, 2000) mentally represent and process what they learn and know about the world. The ways in which various cognitive investigators study intelligence differ primarily in terms of the complexity of the processes being studied. Among the advocates of this approach have been Ted Nettelbeck, Arthur Jensen, Earl Hunt, Herbert Simon, and myself. Each of these researchers has considered both the speed and the accuracy of information processing to be important factors in intelligence. In addition to speed and accuracy of processing, Hunt considered verbal versus spatial skill, as well as attentio-nal ability.
Nettelbeck (e.g., 1987; Nettelbeck & Lally, 1976; Nettelbeck & Rabbitt, 1992; see also Deary, 2000, 2002; Deary & Stough, 1996) suggested a speed-related indicator ofintelli-gence involving the encoding of visual information for brief storage in working memory. But what is critical in this view is not speed of response but rather the length of time a stimulus must be presented for the subject to be able to process that stimulus. The shorter the presentation length, the higher the score. The key variable is the length of time for the presentation of the target stimulus, not the speed of responding by pressing the button. Nettelbeck operationally defined inspection time as the length of time for presentation of the target stimulus after which the participant still responds with at least 90% success. Nettelbeck (1987) found that shorter inspection times correlate with higher scores on intelligence tests [e.g., various subscales of the Wechsler Adult Intelligence Scale (WAIS)] among differing populations of participants. Other investigators have confirmed this finding (e.g., Deary & Stough, 1996).
choice reaction time
Arthur Jensen (1979, 1998, 2002) emphasized a different aspect of information-processing speed; specifically, he proposed that intelligence can be understood in terms of speed of neuronal conduction. In other words, the smart person is someone whose neural circuits conduct information rapidly. When Jensen proposed this notion, direct measures of neural-conduction velocity were not readily available, so Jensen primarily studied a proposed proxy for measuring neural-processing speed - choice reaction time, the time it takes to select one answer from among several possibilities. For example, suppose that you are one of Jensen's participants. You might be seated in front of a set of lights on a board. When one of the lights flashed, you would be expected to extinguish it by pressing as rapidly as possible a button beneath the correct light. The experimenter would then measure your speed in performing this task. Jensen (1982) found that participants with higher intelligence quotients (IQs) are faster than participants with lower IQs in their reaction time (RT), the time between when a light comes on and the finger leaves the home (central) button. In some studies, participants with higher IQs also showed a faster movement time, the time between letting the finger leave the home button and hitting the button under the light. Based on such tasks, Reed and Jensen (1991, 1993 ) propose that their findings may be attributable to increased central nerve-conduction velocity, although at present this proposal remains speculative.
More recently, researchers have suggested that various findings regarding choice RT may be influenced by the number of response alternatives and the visual-scanning requirements of Jensen's apparatus rather than being attributable to the speed of RT alone (Bors, MacLeod, & Forrin, 1993). In particular, Bors and colleagues found that manipulating the number ofbuttons and the size of the visual angle of the display could reduce the correlation between IQ and RT. Thus, the relation between reaction time and intelligence is unclear.
lexical access speed and speed of simultaneous processing
Like Jensen, Earl Hunt (1978) suggested that intelligence be measured in terms of speed. However, Hunt has been particularly interested in verbal intelligence and has focused on lexical-access speed - the speed with which we can retrieve informa tion about words (e.g., letter names) stored in our long-term memories. To measure this speed, Hunt proposed a letter-matching RT task (Posner & Mitchell, 1967).
For example, suppose that you are one of Hunt's participants. You would be shown pairs of letters, such as "A A," "A a," or "A b." For each pair, you would be asked to indicate whether the letters constitute a match in name (e.g., "A a" match in name of letter of the alphabet but "A b" do not). You would also be given a simpler task, in which you would be asked to indicate whether the letters match physically (e.g., "A A" are physically identical, whereas "A a" are not). Hunt would be particularly interested in discerning the difference between your speed for the first set of tasks, involving name matching, and your speed for the second set, involving matching of physical characteristics. Hunt would consider the difference in your reaction time for each task to indicate a measure of your speed of lexical access. Thus, he would subtract from his equation the physical-match reaction time. For Hunt, the response time in indicating that "A A" is a physical match is unimportant. What interests him is a more complex reaction time -that for recognizing names of letters. He and his colleagues have found that students with lower verbal ability take longer to gain access to lexical information than do students with higher verbal ability.
Earl Hunt and Marcy Lansman (1982) also studied people's ability to divide their attention as a function of intelligence. For example, suppose that you are asked to solve mathematical problems and simultaneously to listen for a tone and press a button as soon as you hear it. We can expect that you would both solve the math problems effectively and respond quickly to hearing the tone. According to Hunt and Lansman, one thing that makes people more intelligent is that they are better able to timeshare between two tasks and to perform both effectively.
In sum, process timing theories attempt to account for differences in intelligence by appealing to differences in the speed of various forms of information processing; inspection time, choice RT, and lexical access timing all have been found to correlate with measures of intelligence. These findings suggest that higher intelligence may be related to the speed of various information-processing abilities, including encoding information more rapidly into working memory, accessing information in long-term memory more rapidly, and responding more rapidly.
Why would more rapid encoding, retrieval, and responding be associated with higher intelligence test scores? Do rapid information processors learn more? Other research on learning in aged persons investigated whether there is a link between age-related slowing of information processing and (1 ) initial encoding and recall of information and (2 ) long-term retention (Nettelbeck et al., 1996; Bors & Forrin, 1995). The findings suggest that the relation between inspection time and intelligence may not be related to learning. In particular, Nettelbeck et al. found there is a difference between initial recall and actual long-term learning -whereas initial recall performance is mediated by processing speed (older, slower participants showed deficits), longer-term retention of new information (preserved in older participants) is mediated by cognitive processes other than speed of processing, including rehearsal strategies. This implies speed of information processing may influence initial performance on recall and inspection time tasks, but speed is not related to long-term learning. Perhaps faster information processing aids participants in performance aspects of intelligence test tasks, rather than contributing to actual learning and intelligence (see also Salthouse, Chap. 24). Clearly, this area requires more research to determine how information-processing speed relates to intelligence.
Recent work suggests that a critical component of intelligence may be working memory (see Morrison, Chap. 19 for a discussion of working memory in thinking). Indeed, Kyllonen (2002) and Kyllonen and Christal (1990) have argued that intelligence may be little more than working memory! Dane-man and Carpenter (1983) had participants read sets of passages and, after they had read the passages, try to remember the last word of each passage. Recall was highly correlated with verbal ability. Turner and Engle (1989) had participants perform a variety of working-memory tasks. In one task, for example, the participants saw a set of simple arithmetic problems, each of which was followed by a word or a digit. An example would be "Is ((3 x 5) - 6 = 7?" TABLE. The participants saw sets of from two to six such problems and solved each one. After solving the problems in the set, they tried to recall the words that followed the problems. The number of words recalled was highly correlated with measured intelligence. It therefore appears that the ability to store and manipulate information in working memory may be an important aspect of intelligence, although probably not all there is to intelligence (see Morrison, Chap. 19 for discussion of working memory and thinking).
the componential theory and complex problem solving
In my early work on intelligence, I (Sternberg, 1977) began using cognitive approaches to study information processing in more complex tasks, such as analogies, series problems (e.g., completing a numerical or figural series), and syllogisms (Sternberg, 1977, 1983, 1985). The goal was to find out just what made some people more intelligent processors of information than others. The idea was to take the kinds of tasks used on conventional intelligence tests and to isolate the components of intelligence - the mental processes used in performing these tasks, such as translating a sensory input into a mental representation, transforming one conceptual representation into another, or translating a conceptual representation into a motor output (Sternberg, 1982). Since then, many people have elaborated upon and expanded this basic approach (Lohman, 2000).
Componential analysis breaks down people's reaction times and error rates on these tasks in terms of the processes that make up the tasks. This kind of analysis revealed that people may solve analogies and similar tasks by using several component processes including encoding the terms of the problem, inferring relations among at least some of the terms, mapping the inferred relations to other terms that would be presumed to show similar relations, and applying the previously inferred relations to the new situations.
Consider the analogy, LAWYER : CLIENT :: DOCTOR : (a. PATIENT b. MEDICINE). To solve this analogy, you need to encode each term of the problem, which includes perceiving a term and retrieving information about it from memory. You then infer the relationship between lawyer and client - that the former provides professional services to the latter. You then map the relationship in the first half of the analogy to the second half of the analogy, noting that it will involve that same relationship. Finally, you apply that inferred relationship to generate the final term of the analogy, leading to the appropriate response of PATIENT. Studying these components of information processing reveals more than measuring mental speed alone (see Holyoak, Chapter 6, for a detailed discussion of analogical reasoning).
When measuring speed alone, I found significant correlations between speed in executing these processes and performance on other traditional intelligence tests. However, a more intriguing discovery is that participants who score higher on traditional intelligence tests take longer to encode the terms of the problem than do less intelligent participants, but they make up for the extra time by taking less time to perform the remaining components of the task. In general, more intelligent participants take longer during global planning - encoding the problem and formulating a general strategy for attacking the problem (or set of problems) - but they take less time for local planning - forming and implementing strategies for the details of the task (Sternberg, 1981).
The advantage of spending more time on global planning is the increased likelihood that the overall strategy will be cor rect. Thus, brighter people may take longer to do something than will less bright people when taking more time is advantageous. For example, the brighter person might spend more time researching and planning a term paper but less time in actually writing it. This same differential in time allocation has been shown in other tasks as well (e.g., in solving physics problems; Larkin et al., 1980; Sternberg, 1979, 1985); that is, more intelligent people seem to spend more time planning for and encoding the problems they face but less time in the other components of task performance. This may relate to the previously mentioned metacog-nitive attribute many include in their notions of intelligence. The bottom line, then, is that intelligence may reside as much in how people allocate time as it does in the amount of time it takes them to perform cognitive tasks.
In a similarly cognitive approach, Simon studied the information processing of people engaged in complex problem-solving situations, such as when playing chess and performing logical derivations (Newell & Simon, 1972; Simon, 1976). A simple, brief task might require the participant to view an arithmetic or geometric series, figure out the rule underlying the progression, and guess what numeral or geometric figure might come next; for example, more complex tasks might include some problem-solving tasks (e.g., the water jugs problems; see Estes, 1982). These problems were similar or identical to those used on intelligence tests.
Although the human brain is clearly the organ responsible for human intelligence, early studies (e.g., those by Karl Lashley and others) seeking to find biological indices of intelligence and other aspects of mental processes were a resounding failure despite great efforts. As tools for studying the brain have become more sophisticated, however, we are beginning to see the possibility of finding physiological indicators of intelligence. Some investigators (e.g., Matarazzo, 1992) believe that we will have clinically useful psychophysiological indices of intelligence very early in the current millennium, although widely applicable indices will be much longer in coming. In the meantime, the biological studies we now have are largely correlational, showing statistical associations between biological and psychometric or other measures of intelligence. The studies do not establish causal relations (see Goel, Chapter 20, for a description of the neural basis of deductive reasoning).
One line of research looks at the relationship of brain size to intelligence (see Jerison, 2000; Vernon et al., 2000). The evidence suggests that, for humans, there is a modest but significant statistical relationship between brain size and intelligence. It is difficult to know what to make of this relationship, however, because greater brain size may cause greater intelligence, greater intelligence may cause greater brain size, or both may depend on some third factor. Moreover, it probably is more important how efficiently the brain is used than what size it is. On average, for example, men have larger brains than women, but women have better connections of the two hemispheres of the brain through the corpus callosum. So it is not clear which gender, on average, would be at an advantage, and probably neither would be. It is important to note that the relationship between brain size and intelligence does not hold across species (Jerison, 2000). Rather, what holds seems to be a relationship between intelligence and brain size relative to the rough general size of the organism.
speed of neural conduction
Complex patterns of electrical activity in the brain, which are prompted by specific stimuli, appear to correlate with scores on IQ tests (Barrett & Eysenck, 1992). Several studies (e.g., McGarry-Roberts, Stelmack, & Campbell, 1992; Vernon & Mori, 1992) initially suggested that speed of conduction of neural impulses correlates with intelligence as measured by IQ tests. A follow-up study (Wickett & Vernon, 1 994), how ever, failed to find a strong relation between neural-conduction velocity (as measured by neural-conduction speeds in a main nerve of the arm) and intelligence (as measured on the Multidimensional Aptitude Battery). Surprisingly, neural-conduction velocity appears to be a more powerful predictor of IQ scores for men than for women, so gender differences may account for some of the differences in the data (Wickett & Vernon, 1994). Additional studies on both males and females are needed.
positron emission tomography, functional magnetic resonance imaging
An alternative approach to studying the brain suggests that neural efficiency may be related to intelligence; such an approach is based on studies of how the brain metabolizes glucose (simple sugar required for brain activity) during mental activities. Richard Haier and colleagues (Haier et al., 1992) cited several other researchers who support their own findings that higher intelligence correlates with reduced levels of glucose metabolism during problem-solving tasks -that is, smarter brains consume less sugar (and hence expend less effort) than do less smart brains doing the same task. Furthermore, Haier and colleagues found that cerebral efficiency increases as a result of learning on a relatively complex task involving visu-ospatial manipulations (the computer game Tetris). As a result of practice, more intelligent participants show not only lower cerebral glucose metabolism overall but also more specifically localized metabolism of glucose. In most areas of their brains, smarter participants show less glucose metabolism, but in selected areas of their brains (believed to be important to the task at hand), they show higher levels of glucose metabolism. Thus, more intelligent participants may have learned how to use their brains more efficiently to focus their thought processes on a given task.
More recent research by Haier and colleagues suggests that the relationship between glucose metabolism and intelligence may be more complex (Haier et al., 1995; Larson et al., 1995). Whereas Haier's group
(1 995) confirmed the earlier findings of increased glucose metabolism in less smart participants (in this case, mildly retarded participants), the study by Larson et al. (1 995) found, contrary to the earlier findings, that smarter participants had increased glucose metabolism relative to their average comparison group.
One problem with earlier studies is that the tasks used were not matched for difficulty level across groups of smart and average individuals. The Larson et al. study used tasks that were matched to the ability levels of the smarter and average participants and found that the smarter participants used more glucose. Moreover, the glucose metabolism was highest in the right hemisphere of the more intelligent participants performing the hard task - again suggesting selectivity of brain areas. What could be driving the increases in glucose metabolism? Currently, the key factor appears to be subjective task difficulty with smarter participants in earlier studies simply finding the tasks too easy. Matching task difficulty to participants' abilities seems to indicate that smarter participants increase glucose metabolism when the task demands it. The preliminary findings in this area need to be investigated further before any conclusive answers are reached.
Some neuropsychological research (e.g., Dempster, 1 991) suggests that performance on intelligence tests may not indicate a crucial aspect of intelligence - the ability to set goals, to plan how to meet them, and to execute those plans. Specifically, persons with lesions in the frontal lobe of the brain frequently perform quite well on standardized IQ tests, which require responses to questions within a highly structured situation, but do not require much in the way of goal setting or planning. If intelligence involves the ability to learn from experience and to adapt to the surrounding environment, the ability to set goals and to design and implement plans cannot be ignored. An essential aspect of goal setting and planning is the ability to attend appropriately to relevant stimuli and to ignore or discount irrelevant stimuli.
Some theorists have tried to understand intelligence in terms of how it has evolved over the eons (e.g., Bjorklund & Kipp, 2002; Bradshaw, 2002; Byrne, 2002; Calvin, 2002; Corballis, 2002; Cosmides & Tooby, 2002; Flanagan, Hardcastle, & Nah-mias, 2002; Grossman & Kaufman, 2002; Pinker, 1997). The basic idea in these models is that we are intelligent in the ways we are because it was important for our distant ancestors to acquire certain sets of skills. According to Cosmides and Tooby (2002), for example, we are particularly sensitive at detecting cheating because people in the past who were not sensitive to cheaters did not live to have children, or had fewer children. Evolutionary approaches stress the continuity of the nature of intelligence over long stretches of time, and in some theories, across species. However, during evolution, the frontal lobe increased in size, so it is difficult to know whether changes in intelligence are just a manifestation of physiological changes or the other way around.
According to contextualists, intelligence cannot be understood outside its real-world context. The context of intelligence may be viewed at any level of analysis, focusing narrowly, on the home and family environment, or extending broadly, on entire cultures (see Greenfield, Chap. 27). Even cross-community differences have been correlated with differences in performance on intelligence tests; such context-related differences include those of rural versus urban communities, low versus high proportions of teenagers to adults within communities, and low versus high socioeconomic status of communities (see Coon, Carey, & Fulker, 1992). Contextualists are particularly intrigued by the effects of cultural context on intelligence.
In fact, contextualists consider intelligence so inextricably linked to culture that they view intelligence as something that a culture creates to define the nature of adaptive performance in that culture and to account for why some people perform better than others on the tasks that the culture happens to value (Sternberg, 1985). Theorists who endorse this model study just how intelligence relates to the external world in which the model is being applied and evaluated. In general, definitions and theories of intelligence will more effectively encompass cultural diversity by broadening in scope. Before exploring some of the contextual theories of intelligence, we will look at what prompted psychologists to believe that culture might play a role in how we define and assess intelligence.
People in different cultures may have quite different ideas of what it means to be smart. One of the more interesting cross-cultural studies of intelligence was performed by Michael Cole and colleagues (Cole et al., 1971). These investigators asked adult members of the Kpelle tribe in Africa to sort concept terms. In Western culture, when adults are given a sorting task on an intelligence test, more intelligent people typically sort hierarchically. For example, they may sort names of different kinds of fish together, and then the word fish over that, with the name animal over fish and over birds, and so on. Less intelligent people typically sort functionally. They may sort fish with eat, for example, because we eat fish, or clothes with wear, because we wear clothes. The Kpelle sorted functionally - even after investigators unsuccessfully tried to get the Kpelle spontaneously to sort hierarchically. Finally, in desperation, one of the experimenters (Glick) asked a Kpelle to sort as a foolish person would sort. In response, the Kpelle quickly and easily sorted hierarchically. The Kpelle had been able to sort this way all along; they just hadn't done it because they viewed it as foolish - and they probably considered the questioners rather unintelligent for asking such stupid questions.
The Kpelle people are not the only ones who might question Western understandings of intelligence. In the Puluwat culture of the Pacific Ocean, for example, sailors navigate incredibly long distances, using none of the navigational aids that sailors from technologically advanced countries would need to get from one place to another (Gladwin, 1970). Were Puluwat sailors to devise intelligence tests for us and our fellow Americans, we might not seem very intelligent. Similarly, the highly skilled Puluwat sailors might not do well on American-crafted tests of intelligence. These and other observations have prompted quite a few theoreticians to recognize the importance of considering cultural context when assessing intelligence.
The preceding arguments may make it clear why it is so difficult to come up with a test that everyone would consider culture-fair - equally appropriate and fair for members of all cultures. If members of different cultures have different ideas of what it means to be intelligent, then the very behaviors that may be considered intelligent in one culture may be considered unintelligent in another. Take, for example, the concept of mental quickness. In mainstream American culture, quickness is usually associated with intelligence. To say someone is "quick" is to say that the person is intelligent and, indeed, most group tests of intelligence are quite strictly timed. Even on individual tests of intelligence, the test-giver times some responses of the test-taker. Many information-processing theorists and even psychophys-iological theorists focus on the study of intelligence as a function of mental speed.
In many cultures of the world, people believe that more intelligent people do not rush into things. Even in our own culture, no one will view you as brilliant if you decide on a marital partner, a job, or a place to live in the 20 to 30 seconds you might normally have to solve an intelligence-test problem. Thus, given that there exist no perfectly culture-fair tests of intelligence, at least at present, how should we consider context when assessing and understanding intelligence?
Several researchers have suggested that providing culture-relevant tests is possible (e.g., Baltes, Dittmann-Kohli, & Dixon, 1984; Jenkins, 1979; Keating, 1984); that is, tests that employ skills and knowledge that relate to the cultural experiences of the test-takers. Baltes and his colleagues, for example, designed tests measuring skill in dealing with the pragmatic aspects of everyday life. Designing culture-relevant tests requires creativity and effort but probably is not impossible. A study by Daniel Wagner (1978), for example, investigated memory abilities - one aspect of intelligence as our culture defines it - in our culture versus the Moroccan culture. Wagner found that level of recall depended on the content that was being remembered, with culture-relevant content being remembered more effectively than irrelevant content (e.g., compared with Westerners, Moroccan rug merchants were better able to recall complex visual patterns on black-and-white photos of Oriental rugs). Wagner further suggested that when tests are not designed to minimize the effects of cultural differences, the key to culture-specific differences in memory might be the knowledge and use of metamemory strategies, rather than actual structural differences in memory (e.g., memory span and rates of forgetting).
In Kenya, research has shown that rural Kenyan school children have substantial knowledge about natural herbal medicines they believe fight infection; Western children, of course, would not be able to identify any of these medicines (Sternberg et al., 2001; Sternberg & Grigorenko, 1997). In short, making a test culturally relevant appears to involve much more than just removing specific linguistic barriers to understanding.
Stephen Ceci (Ceci & Roazzi, 1 994) found similar context effects in childrens' and adults' performance on a variety of tasks. Ceci suggests that the social context (e.g., whether a task is considered masculine or feminine), the mental context (e.g., whether a visuo-spatial task involves buying a home or burgling it), and the physical context (e.g., whether a task is presented at the beach or in a laboratory) all affect performance. For example, fourteen-year-old boys performed poorly on a task when it was couched as a cupcake-baking task but performed well when it was framed as a battery-charging task (Ceci & Bronfenbrenner, 1985). Brazilian maids had no difficulty with proportional reasoning when hypothetically pur chasing food but had great difficulty with it when hypothetically purchasing medicinal herbs (Schliemann & Magalhiies, 1990). Brazilian children whose poverty had forced them to become street vendors showed no difficulty in performing complex arithmetic computations when selling things but had great difficulty performing similar calculations in a classroom (Carraher, Carraher, & Schliemann, 1985). Thus, test performance may be affected by the context in which the test terms are presented. In this study, the investigators looked at the interaction of cognition and context. Several investigators have proposed theories that seek explicitly to examine this interaction within an integrated model of many aspects of intelligence. Such theories view intelligence as a complex system.
Systems Approaches to Intelligence gardner: multiple intelligences
Howard Gardner (1983, 1993) proposed a theory of multiple intelligences, in which intelligence is not just a single, unitary construct. Instead of speaking of multiple abilities that together constitute intelligence (e.g., Thurstone, 1938), Gardner (1999) speaks of eight distinct intelligences that are relatively independent of each other. Each is a separate system of functioning, although these systems can interact to produce what we see as intelligent performance.
In some respects, Gardner's theory sounds like a factorial one because it specifies several abilities that are construed to reflect intelligence of some sort. However, Gardner views each ability as a separate intelligence, not just as a part of a single whole. Moreover, a crucial difference between Gardner's theory and factorial ones is in the sources of evidence Gardner used for identifying the eight intelligences. Gardner used converging operations, gathering evidence from multiple sources and types of data.
Gardner's view of the mind is modular, Because as a major task of existing and future research on intelligence is to isolate the portions of the brain responsible for each of the intelligences. Gardner has speculated regarding at least some of these locales, but hard evidence for the existence of these separate intelligences has yet to be produced. Furthermore, Nettelbeck and Young (1996) question the strict modularity of Gardner's theory. Specifically, the phenomenon of preserved specific cognitive functioning in autistic savants (persons with severe social and cognitive deficits, but with corresponding high ability in a narrow domain) as evidence for modular intelligences may not be justified. According to Nettelbeck and Young, the narrow long-term memory and specific aptitudes of savants is not really intelligent. As a result, there may be reason to question the intelligence of inflexible modules.
Sternberg: the triarchic theory of successful intelligence
Whereas Gardner emphasizes the separate-ness of the various aspects of intelligence, I tend to emphasize the extent to which they work together in the triarchic theory of successful intelligence (Sternberg, 1985, 1988, 1996, 1999). According to the triarchic (tri-, "three"; -archic, "governed") theory, intelligence comprises three aspects, dealing with the relation of intelligence (1) to the internal world of the person, (2) to experience, and (3) to the external world.
How intelligence relates to the internal world. This part of the theory emphasizes the processing of information, which can be viewed in terms of three different kinds of components: (1 ) metacomponents - executive processes (i.e., metacognition) used to plan, monitor, and evaluate problem solving; (2) performance components - lower order processes used to implement the commands of the metacomponents; and (3) knowledge-acquisition components - the processes used to learn how to solve the problems in the first place. The components are highly interdependent.
How intelligence relates to experience. The theory also considers how prior experience may interact with all three kinds of information-processing components. That is, each of us faces tasks and situations with which we have varying levels of experience, ranging from a completely novel task, with which we have no previous experience, to a completely familiar task, with which we have vast, extensive experience. As a task becomes increasingly familiar, many aspects of the task may become automatic, requiring little conscious effort to determine what step to take next and how to implement that next step. A novel task makes demands on intelligence different from those of a task for which automatic procedures have been developed.
According to the triarchic theory, relatively novel tasks - such as visiting a foreign country, mastering a new subject, or acquiring a foreign language - demand more of a person's intelligence. In fact, a completely unfamiliar task may demand so much of the person as to be overwhelming.
How intelligence relates to the external world. The triarchic theory also proposes that the various components of intelligence are applied to experience to serve three functions in real-world contexts -adapting ourselves to our existing environments, shaping our existing environments to create new environments, and selecting new environments.
According to the triarchic theory, people may apply their intelligence to many different kinds of problems. Some people may be more intelligent in the face of abstract, academic problems, for example, whereas others may be more intelligent in the face of concrete, practical problems. The theory does not define an intelligent person as someone who necessarily excels in all aspects of intelligence. Rather, intelligent persons know their own strengths and weaknesses and find ways in which to capitalize on their strengths and either to compensate for or to correct their weaknesses.
In a recent comprehensive study testing the validity of the triarchic theory and its usefulness in improving performance, we predicted that matching students' instruction and assessment to their abilities would lead to improved performance (Sternberg et al., 1996, 1999). Students were selected for one of five ability patterns: high only in analytical ability, high only in creative ability, high only in practical ability, high in all three abilities, or not high in any of the three abilities. Then students were assigned at random to one of four instructional groups that emphasized memory-based, analytical, creative, or practical learning followed by subsequent assessment. We found that students who were placed in an instructional condition that matched their strength in terms of ability pattern (e.g., a high-analytical student being placed in an instructional condition that emphasized analytical thinking) outperformed students who were mismatched (e.g., a high-analytical student being placed in an instructional condition that emphasized practical thinking).
Teaching all students to use all of their analytic, creative, and practical abilities has resulted in improved school achievement for all students, whatever their ability pattern (Grigorenko, Jarvin, & Sternberg, 2002; Sternberg, Torff, & Grigorenko, 1998). One important consideration in light of such findings is the need for changes in the assessment of intelligence (Sternberg & Kaufman, 1996). Current measures of intelligence are somewhat one-sided, measuring mostly analytic abilities with little or no assessment of creative and practical aspects of intelligence (Sternberg et al., 2000; Wagner, 2000). A well-rounded assessment and instruction system could lead to greater benefits of education for a wider variety of students - a nominal goal of education.
Perkins (1995 ) proposed a theory of what he refers to as true intelligence, which he believes synthesizes classic views as well as new ones. According to Perkins, there are three basic aspects of intelligence - neural, experiential, and reflective.
Neural intelligence concerns what Perkins believes to be the fact that some people's neurological systems function better than do the neurological systems of others, running faster and with more precision. He mentions "more finely tuned voltages" and "more exquisitely adapted chemical catalysts" as well as a "better pattern of connecticity in the labyrinth of neurons" (Perkins, 1995, p. 497), although it is not entirely clear what any of these terms means. Perkins believes this aspect of intelligence to be largely genetically determined and unlearn-able. This kind of intelligence seems to be somewhat similar to Cattell's (1971) idea of fluid intelligence.
The experiential aspect of intelligence is what has been learned from experience. It is the extent and organization of the knowledge base and thus is similar to Cattell's (1971) notion of crystallized intelligence.
The reflective aspect of intelligence refers to the role of strategies in memory and problem solving, and appears to be similar to the construct of metacognition or cognitive monitoring (Brown & DeLoache, 1978; Flavell, 1981).
No empirical test of the theory of true intelligence has been published, so it is difficult to evaluate the theory at this time. Like Gardner's (1983) theory, Perkins's theory is based on literature review, and, as noted previously, such literature reviews often tend to be selective and then interpreted in a way that maximizes the fit of the theory to the available data.
the bioecological model of intelligence
Ceci (1996) proposed a bioecological model of intelligence, according to which multiple cognitive potentials, context, and knowledge all are essential bases of individual differences in performance. Each of the multiple cognitive potentials enables relationships to be discovered, thoughts to be monitored, and knowledge to be acquired within a given domain. Although these potentials are biologically based, their development is closely linked to environmental context, and it is difficult, if not impossible, to cleanly separate biological from environmental contributions to intelligence. Moreover, abilities may express themselves very differently in different contexts. For example, children given essentially the same task in the context of a video game versus a laboratory cognitive task performed much better when the task was presented in the video game context.
The bioecological model appears in many ways more to be a framework than a theory. At some level, the theory must be right. Certainly, both biological and ecological factors contribute to the development and manifestation of intelligence. Perhaps what the theory needs most at this time are specific and clearly falsifiable predictions that would set it apart from other theories.
Was this article helpful?
This Amazing Course Will Blow Your Results Out Of Proportion! Master These Ultimate Goal Setting Techniques And Watch Your Results Soar Sky High In A Fraction Of The Time! Save Hundreds Of Hours Blindly Chasing Results By Tapping Into These Mind-Blowing Goal Setting Secrets Which Will Skyrocket Your Results Quickly!