Intelligence And Learning Research Paper

1. Intelligence: A Valid Predictor Of Achievement Differences

Individuals with similar cultural, social, and educational backgrounds differ from one another in the time they need to process certain information, in their ability to understand complex ideas, in the efficiency with which they can deal with novel, transfer-demanding tasks, and in the learning outcome that results from attending certain instructional environments. The construct of psychometric intelligence attempts to clarify what is behind such achievement variations that cannot be explained by differences in learning environments or in amount of practice.

1.1 Measuring Intelligence

About a century ago Alfred Binet constructed problems designed to determine whether children who did not meet certain school requirements suffered from mental retardation or from behavioral disturbances. Since then many psychologists have been quite successful in developing reliable verbal and nonverbal intelligence tests for children and adults. Intelligence tests contain items composed of verbal, numerical, and pictorial material, and they require various mental operations, among them inductive and deductive reasoning, pattern recognition, and memorization. So-called speed tests contain items that are comparably easy for everybody; individual differences in the numbers of correctly solved problems only occur because of time limitations. In power tests the items are ordered according to their difficulty, and limits in intelligence become apparent if people do not solve all problems despite having sufficient time.

The distribution of achievement scores in all intelligence test scales follows the bell curve (normal distribution). This reflects the fact that the majority of people resemble each other quite a lot with respect to their cognitive capabilities, and only a few people show extraordinarily low or high competencies. Normal distribution is the statistical prerequisite for measuring intelligence on interval level by indicating deviations from the mean score. To determine the intelligence quotient (IQ), test scores are converted to a scale in which by convention the mean is 100 and the standard deviation is 15. The reliability of IQ tests, revealed either by correlation coefficients based on repeated measurement or by figuring out internal consistency, is between 0.80 and 0.90, which is higher than for most other psychometric measures. Nonetheless, a reliability lower than 1.0 only allows us to interpret a range rather than a single value. For example, if a person’s tested IQ is 110, and if the reliability of the test is 0.90, the IQ of this person is between 101 and 119 with a probability of 95 percent. A test reliability of 0.80 reveals a range of 97–123.

Despite broad variations in the content and the form of presentation of intelligence test items, all tests have in common that they do not presuppose knowledge that can only be acquired in particular learning environments not accessible to everybody. Moreover, differences in test scores within a group only reflect differences in intelligence between these persons if all of them have had access to similar learning environments. In other words, each member of the group must have had, in principle, the opportunity to acquire the knowledge necessary to solve the problems. Intelligence test scores can be raised considerably by practicing the respective types of items, while individual differences do not disappear but rather remain quite stable at a higher level. The Flynn effect, named for its discoverer, is the observation that mean IQ scores have increased considerably in the past 50 years. This may be at least partly due to training effects caused by widespread exposure to intelligence test items in various media. Interindividual differences in intelligence test scores between persons can only be interpreted as differences in intelligence if similar amounts of practice can be presupposed. Attempts to construct so-called culture-free or culture-fair tests, which were supposed to be unaffected by prior experience, have failed because it turns out that different cultures are not prepared in the same way even for dealing with nonverbal material and mental operations that are not part of institutional schooling. Although some studies reveal ethnic differences in mean IQ, hitherto there is no convincing evidence that these differences are genetically affected.

Within a fairly homogeneous cultural context intelligence can be considered as a personality trait mainly for two reasons. First, performance on intelligence tests that are based on different contents and mental operations show significant correlations. For instance, correlations between tests on inductive reasoning with pictorial material and with verbal material are around 0.50, and correlations between tests that measure basic visuospatial competencies such as mental rotation and tests of verbal fluency are about 0.30. In numerous studies run all over the world, multivariate statistical methods such as factor analysis have revealed that a single factor, called factor g, can account for 30–40 percent of the variance in IQ-test batteries composed of various scales covering a range of content knowledge and mental operations. Second, longitudinal studies have revealed that IQ is a fairly stable measure across the life span. Long-term stability of IQ for adolescents and adults comes close to the reliability of the tests, and even in early childhood long-term correlations are around 0.50 for both verbal and nonverbal tests.

Overall, intelligence test scores predict academic performance fairly well: the correlations between IQ and grades in school and university are about 0.50. The correlations between intelligence test scores and measures of outside-school success such as income or professional status are lower but still significant. That intelligence cannot account for a larger amount of achievement variation is not at all surprising, given the importance of numerous other factors, among them social background, motivation, and effort.

Because verbal and nonverbal intelligence tests are good predictors of how well an individual will succeed in school and university, they are quite helpful for making recommendations for different educational trajectories. They allow educators to identify children who cannot be expected to gain from normal schooling and therefore need special education, or children who might benefit from an advanced learning environment. According to Guthke (1993), the validity of intelligence tests can be increased if they are presented as learning tests, which means that all test-takers practice the items in several trials by getting feedback and thinking hints so that individual differences in familiarity with tests are compensated for. Particularly for children from disadvantageous social environments, learning tests are a more reliable and valid measure of intelligence than conventional IQ tests.

Attempts to extend the construct of intelligence by including social and emotional competencies as well as striving for success are controversial because the theoretical background of these concepts is still vague, and, more importantly, because the tests designed to measure these aspects do not meet the strong diagnostic criteria that have been established in psychometric research.

1.2 Factor G, Specific Abilities, And Knowledge Acquisition As Determinants Of Cognitive Development

What cognitive capabilities are behind factor g is still a controversial question in psychology (Neisser et al. 1996). According to low-level theories, g reflects individual differences in basic brain functions, for example, neuronal speed. These theories are supported by results from studies that reveal 10–25 percent common variance between intelligence test scores and simple reaction time measures. However, the search for a single brain function that determines differences in cognitive capabilities has not, as yet, met with much success.

Anyhow, the heterogeneous pattern of correlations between the range of intelligence subtests leaves no doubt as to the fact that human abilities vary beyond general basic brain functions. Though significant, many correlations between intelligence subtests are low, indicating the involvement of independent mental resources. Even between tests on inductive reasoning which are based on different forms of representation (i.e., pictures, numbers, words), only medium correlations are revealed, suggesting that cognitive processes are to a large extent guided by specific verbal, visualspatial, or numerical abilities, among others.

Among the several attempts that have been made to integrate general and specific abilities into broader theories of intelligence is the hierarchical model developed by Cattell (1971). The author proposes two major subfactors, which are related to one another. Fluid intelligence (Gf), which describes basic information processing and reasoning is measured by content-poor nonverbal tests. Crystallized intelligence (Gc), which is particularly revealed in content-rich verbal tests, represents the accumulation of higher order knowledge over the life span of an individual. Both kinds of intelligence are closely interrelated in childhood because Gf partly guides the development of Gc. From the age of about 25 years on, however, Gf and Gc take different developmental trajectories. While Gf starts to decrease at the age of 25, first slightly and from about 50 years on more quickly, Gc is unaffected by age and even has good chances to increase until the age of around 70. It is worth noticing that because of the large individual differences, the given age information is only a rough estimation.

Developmental changes also occur with respect to the structure of intelligence. The prominence of Gf decreases during childhood while Gc as a function of repeated enlargement and rearrangement of knowledge becomes more useful. When people from their adolescence on, at the latest, start to specialize in different areas, general cognitive resources have to be completed by specific abilities. This process of differentiation is characterized by decreasing correlations between intelligence subtests from childhood over adolescence to adulthood. Only in old age is a process of dedifferentiation observed which is characterized by a growing impact of Gf (Baltes et al. 1999).

After years of controversial debate, researchers now agree that part of within-group differences in IQ is determined by genes. Twin and adoption studies conducted in North America and Europe suggest that about 50 percent of performance variation in intelligence tests is due to genetic differences. In general, the amount of variance in intelligence test scores to be explained by genes is higher the more members in a society have access to school education, health care, and sufficient nutrition. Findings which suggest that heritability increases during the life span and is not lower for Gc than for Gf are contra-intuitive only at first glance. To understand these results one has to realize that societies which provide access to a broad variety of cognitive activities in professional as well as in private life enables adults more than children to actively select special environments which fit their genes. People who have found their niche can perfect their competencies by deliberate learning.

Genes, however, not only indirectly guide learning and knowledge acquisition via general and specific abilities, rather, there is growing evidence that during evolution the human mental architecture has been equipped with quite specific knowledge structures for instance about visual or auditory patterns, physical objects, language use, and social situations. Such knowledge structures, called input systems or modules, allow appropriate cognitive and behavioral functioning beyond learning. Moreover, the representational redescription of innate modules may provide the basis for acquiring higher order domainspecific knowledge which only emerges during cultural development (Karmiloff-Smith 1992). By integrating the differential perspective of psychometric intelligence research and the universal approach of mental modularity, Anderson (1992) presents a new theory of differential cognitive development in his very informative book. According to this approach individual differences in cognitive capabilities start to increase from birth onwards because the speed and efficiency with which modules are redescribed into higher order knowledge structures is determined by general and, to a lower extent, by specific cognitive abilities. Despite its general plausibility, however, several central claims of this theory are in conflict with some well-established and appreciated theories of cognitive development and psychometric intelligence.

2. Prior Knowledge: The Best Predictor Of Learning Outcome But Not A Substitute For Intelligence

In three lines of research domain-specific knowledge has become an important variable that has challenged the value of intelligence tests as a diagnostic instrument.

2.1 Research On Outstanding Professional Expertise

Nobody would challenge that most of the people who have contributed in an outstanding way to an academic domain score in the upper third on intelligence tests. However, one should note that this claim is almost all that intelligence research can contribute to explaining expertise. For outstanding expertise in various fields, among them chess, medicine, and physics, having access to an elaborate domain-specific knowledge base was proven to be necessary and sufficient.

2.2 Comparisons Between Intelligent No Ices And Less Intelligent Experts

When entering learning settings, learners often differ from one another with respect to prior domain-specific knowledge and IQ. Studies that have investigated the learning effects of attending mathematics lessons at school and reading texts about sports have revealed much better outcomes for persons with high prior knowledge levels (experts) and low IQ than for persons with little prior knowledge (novices) and high IQ (Weinert and Helmke 1998). Note that it was novices, rather than laypersons, who were considered. In contrast to laypersons, novices possess the necessary domain-specific knowledge in terms of rules and core concepts but differ from experts in their lack of practice.

2.3 Domain-Specific Determinants Of Cognitive Development In Childhood

Research on cognitive development suggests that variations in prior domain-specific knowledge can often better account for achievement differences between younger and older children than general cognitive capabilities. Moreover, longitudinal studies suggest that within-age-level achievement differences in core elementary school subjects are to a remarkable extent determined by domain-specific prior knowledge obtained in the preschool years. Early numerical competencies could account for achievement variations in mathematics after partialing out general intelligence, and early indicators of letter identification and phonological awareness predicted later performance in reading and writing (Weinert and Schneider 1999).

The reported results are important because they show that at least in complex knowledge domains a high IQ cannot compensate for a lack of prior knowledge, and moreover, that there is no direct connection between intelligence and achievement in content domains based on rich specific knowledge. However, regression analyses based on longitudinal studies reveal that the confounded variance of prior knowledge and intelligence predicts differences in learning outcome better than each single variable. This means that prior knowledge and intelligence are to a certain extent inextricably linked with each other. Intelligence may guide the selection of learning environments and thereby determine the acquisition of prior knowledge. A person with a low IQ will hardly follow courses on theoretical physics even if he or she is credited with extra time. Moreover, intelligence may influence in how many content areas a person is able to acquire a profound amount of prior knowledge.

3. Intelligence And Learning

Since Sternberg (1985) claimed in his pioneering book that research on intelligence has to focus on different aspects of information processing, as there are meta-components, performance components, and knowledge-acquisition components, several attempts have been made to find out what kinds of cognitive activities are affected by intelligence during learning and problem solving. Creative assessment paradigms and stimulating ideas will be discussed in the following section.

3.1 Efficiency In Constructing And Accessing Knowledge: Few Knowns And Many Unknowns

At first glance it sounds plausible that, given that individual differences in prior knowledge can be controlled, intelligence affects the speed of mastering complex systems as well as competencies in dealing with novelty. Intelligence may affect the efficiency with which a knowledge base is constructed and moreover it may determine the means of accessing it. For instance, from an early age intelligent persons may represent certain problem-solving strategies in a more abstract way so that they can be easily transferred to novel situations. In addition, efficient metacognitive strategies for accessing and combining knowledge may increase the lead of the intelligent persons. Such general claims have been somewhat supported by research on complex problem solving (Frensch and Funke 1995). In this paradigm learners are presented with complex computer scenarios that, for example, require them to maximize the profit of a company, increase the supply of water in the desert but at the same time avoid an ecological collapse, or organize a city as a mayor. Tests and interviews make sure that prior knowledge about the content of the scenarios as well as experience with the computer are the same for all learners. To reach a certain goal, learners have to manipulate various variables that show a complex pattern of interaction with each other. Correlations of 0.30–0.50 between IQ and speed of mastering a complex scenario have been obtained in various studies. Moreover, contrary to a long-held scientific opinion, the impact of intelligence on learning outcome does not decrease with practice.

Although such results are somewhat promising, they often encourage the establishment of rash and arbitrary definitions of intelligence, such as speed of information processing, working memory capacity, ability to abstract, ability to deal with novelty, or metacognitive competencies. It is, however, often difficult to distinguish between such concepts, and moreover, they often lead to inconsistent conclusions. For instance, those learners who are thought to acquire information quickly may do so at the expense of profoundness. Given the controversies in cognitive psychology about modeling general working memory capacity, this construct is hardly ready for describing individual differences. Also the plausible claim that intelligent persons represent their knowledge in an abstract way that facilitates transfer does not hold (Detterman and Sternberg 1993). On the one hand, as mentioned above, research on expertise indicated that an elaborate knowledge base is not characterized by abstraction but rather by the efficient organization of specific elements. On the other hand, one of the most consistent results of psychology is that even persons with an IQ above average only rarely show spontaneous knowledge transfer even when they are presented with problems with an isomorphic formal structure. Moreover, because novelty is a combination of hitherto unrelated knowledge, the construction of completely new demands has turned out to be unattainable. The enthusiasm about metacognition, which was appreciated as a synonym for IQ by some researchers in the 1980s, has died down considerably after it has turned out that metacognitive competencies can rarely account for achievement differences in the normal and upper performance range. In sum, despite the many attempts to identify components of information processing as sources of individual differences, hitherto more questions have arisen than have been resolved.

3.2 Educational Implications Of Differences In IQ

Comparisons between schooled and unschooled groups reveal a strong effect of education on intelligence test scores even on nonverbal tests. Only by systematic education can individuals’ intelligence emerge and approach an optimum. However, given that a basic level of education has been encountered, schooling in general and special training programs in particular increase intelligence only very modestly, if at all (Perkins and Grotzer 1997). IQ differences remain quite stable over time in groups who have been attending stimulating learning settings. Education highlights individual differences in intelligence rather than compensating for them.

Broad variance in intelligence is a challenge for designing educational environments. The question arises of whether learners of different intelligence levels gain more if they are assigned to different learning environments. A few studies have revealed so-called Aptitude-Treatment Interactions. Less intelligent students gained more from structured than from open instruction, while for more intelligent learners the reverse pattern occurred. However, most studies have failed to reveal interactions between intelligence and educational treatment, some of them probably because they lacked the statistical power necessary for revealing interaction effects. In general, when assigning learners to different learning treatments according to their intelligence, one must remember that IQ follows the normal distribution. This means that 68 percent of the population do not differ by more than one standard deviation in either direction from the mean—they are quite similar. Therefore, assigning an unbiased group of learners to two different learning environments by median split of IQ scores is not useful at all. Moreover, research has shown that similar educational environments can equip learners with a broad range of IQs with necessary prior knowledge in various areas. Eventual deficiencies in intelligence can be compensated for by effort and deliberate practice to a considerable extent. However, given the relationship between intelligence and efficiency in learning and information processing, a higher IQ facilitates the exploitation of learning environments, leading to the acquisition of knowledge that is broad as well as deep enough to master many of the as yet unknown demands of the future.

Bibliography:

1. Anderson M 1992 Intelligence and Development: A Cognitive Theory. Blackwell, Oxford, UK
2. Baltes P B, Staudinger U, Lindenberger U 1999 Lifespan psychology: Theory and application to intellectual functioning. Annual Review of Psychology 50: 471–507
3. Cattell R B 1971 Abilities: Their Structure, Growth, and Action. Houghton Mifflin, Boston
4. Dettermann D K, Sternberg R J (eds.) 1993 Transfer on Trial: Intelligence, Cognition and Instruction. Ablex Publishing Corporation, Norwood, NJ
5. Frensch P A, Funke J (eds.) 1995 Complex Problem Solving: The European Perspective. Erlbaum, Hillsdale, NJ
6. Guthke J 1993 Development in learning potential assessment. In: Hamers J H M, Sijtsma K, Ruijssenaars A J M (eds.) Learning Potential Assessment. Swets & Zeitlinger, Amsterdam, pp. 43–67
7. Karmiloff-Smith A 1992 Beyond Modularity: A Developmental Perspective on Cognitive Science. MIT Bradford Books, Cambridge, MA
8. Neisser U et al. 1996 Intelligence: Knowns and unknowns. American Psychologist 51(2): 77–101
9. Perkins D N, Grotzer T A 1997 Teaching intelligence. American Psychologist 52(10): 1125–33
10. Sternberg R J 1985 Beyond I.Q.: A Triarchic Theory of Human Intelligence. Cambridge University Press, New York
11. Weinert F E, Helmke A 1998 The neglected role of individual differences in theoretical models of cognitive development. Learning and Instruction 8(4): 309–23 (Special Issue)
12. Weinert F E, Schneider W (eds.) 1999 Individual Development from 3 to 12: Findings from the Munich Longitudinal Study. Cambridge University Press, Cambridge, UK