Elementary Cognitive Tasks Research Articles

Task Complexity and the Speed and Efficiency of Elemental Information Processing: Another Look at the Nature of Intellectual Giftedness

This study examined the speed and efficiency of elemental processing among the intellectually gifted. Groups of gifted and nongifted junior high school students were compared on several elementary cognitive tasks (ECTs) with no symbolic content and different degrees of requisite processing complexity. After controlling for the potentially confounding effect of knowledge base on the ECTs, results of this study further substantiated the significant relationship between elemental processing, task complexity, and intellectual giftedness. Differences between the gifted and nongifted groups on the ECT parameters increased monotonically with task complexity. Moreover, despite the fact that the ECTs used in this study have no information content and require no higher-order or metaprocesses for successful task completion, discriminant function analyses including the various elemental processing speed and efficiency measures correctly classified approximately 80% of all subjects. Implications of these results for theory relating giftedness to the speed and efficiency of elemental cognitive processes are discussed.

Application of the techniques of mental chronometry to the study of learning disabilities

This investigation applied the techniques of mental chronometry to the study of learning disabilities (LDs). Twenty-five children with reading disability (RD) were compared with 25 nondisabled children (matched for gender and age) on elementary cognitive tasks (ECTs) with no symbolic content and different degrees of requisite processing complexity. After adjustment for nonverbal intelligence, no significant differences were found between groups on any of the ECT parameters. A significantly greater proportion of children with RD were unable to resolve an inspection time (IT), however. In addition, the correlation between intelligence and IT for children with RD was +0.44, compared with −0.44 for nondisabled children. Implications of these results for theory relating LDs to deficiencies in the speed and efficiency of elemental cognitive processes are discussed.

Intelligence and speed of information processing: Further results and questions on Hick's paradigm and beyond

In the last 10 years, there have been an increasing number of results indicating a relation between parameters of processing speed in elementary cognitive tasks as in the Hick paradigm and basic aspects of psychometric intelligence (usually “general intelligence” or g). Unfortunately the pattern of findings is somewhat inconsistent and there is little theoretical basis for explaining and integrating the various results. Relations of reaction time (RT)-parameters with other aspects of intelligence (e.g. structural components) have largely been neglected. To replicate main results concerning the RT x intelligence relations and to analyze these relations in a broader context of intelligence variables, the Hick paradigm, the Berlin Model of Intelligence Structure (BIS), and Raven's Advanced Progressive Matrices (APM) were employed. The intelligence tests were administered on the first day to 73 subjects, most of them university students, and the Hick paradigm was administered twice on the second day of the study. On a basic level of analysis two of three “classical” assumptions on the Hick paradigm are replicated; significant negative correlations between general intelligence (APM, BIS-AI) on the one hand and intercept and intraindividual variances in RT on the other hand. In contradiction with the “classical” assumptions, the negative correlation between general intelligence and the slope parameter could not be replicated. Surprisingly, RT-medians and -variances per bit level proved to have as least as much prognostic power as the classical Hick parameters directly based on Hick's law (slope and intercept). Significant correlations between RT-parameters and BIS-components, especially processing speed (B), should induce more thorough investigation on the intelligence side of the RT x intelligence relation. The results indicate again that main findings pertaining to the Hick paradigm represent relatively stable relations. But the findings are not consistent and partially give reasons for doubts about some central until now available positive findings. A substantiating theory for explaining and integrating positive results, anomalies and prevailing ad-hoc assumptions and for integrating them into broader theories of intelligence meanwhile seems overdue.

Should we take the shape of reaction time distributions into account when studying the relationship between RT and psychometric intelligence?

The negative correlation between psychometric intelligence and variability of reaction time (RT) in elementary cognitive tasks (ECTs) is now well established. However, there has been relatively littlw research aimed at describing distributions of RT trials other than with mean, median and standard deviation. The present study was designed to examine the relationship between measured intelligence and parameters yielded by exGaussian analysis of RT distributions. Fifty two subjects were administered a computerized discrimination task and five intelligence tests. Data collected indicated that the skewness of RT was more correlated with psychometric intelligence than intraindividual variability in RT was. In addition, and though the exGaussian analysis provided a good description of RT distributions, the results showed that the parameters of the analysis did not correlate significantly with measured intelligence.

The correlation between reaction time and the ponderal index.

Reaction time (RT) and movement time (MT), and intraindividual variability in RT and MT, measured in elementary cognitive tasks at three levels of complexity of information processing, administered to 213 college males, show nonsignificant correlations with body build, specifically the ponderal index, hence not replicating the results of a study by Smith and Boyarsky (1943).

Major genes of general intelligence

Evidence in favour of the major gene theory of intelligence is stated in summary form. Empirical distributions from studies on giftedness by Terman and Weiss and data of social mobility can be explained by the existence of a major gene that in the homozygous state is the prerequisite to having an IQ of 130 or higher. Under the assumption of about 10% misclassification of genotypes, family data are in agreement with Mendelian segregation at such a major gene locus. Elementary cognitive tasks, highly correlated with IQ, are not distributed normally. On the absolute scale of short-term memory capacity (measured in bits), defined as the product of memory span and mental speed, the heterozygotes are intermediate between the homozygotes. Where there are major genes, there must be an underlying biochemical code, which can be detected. To this aim enzymes, responsible for the regulation of brain energy metabolism and correlated with IQ and social status, should be the target of further research. From the point of view of evolution, social stratification and the frequency of major genes of intelligence depend upon each other.

The skewness of the distribution of RT trials does not correlate with psychometric g

Among the parameters of reaction time (RT) on elementary cognitive tasks (ECTs), the intraindividual variability of reaction time (SDRT) (measured as the SD of a subject's RT trials) is the best correlate of psychometric g. One aspect of RT performance that has not been investigated is the relationship between the skewness of the distribution of RT trials and g. Results of this study indicate that RT skewness is not importantly related to g.

A test of Larson and Alderton's (1990) worst performance rule of reaction time variability

The validity of Larson and Alderton's (1990; Intelligence, 14, 309–325) ‘Worst Performance’ rule of reaction time (RT) variability, which states that the worst RT trials correlate more highly with intelligence than the fastest RT trials, was tested by examining the potentially confounding effect of range restriction across the RT bands of elementary cognitive tasks (ECTs) of various degrees of information processing complexity. Results of this study indicate that the intersubject variability within RT bands is not systematically related to the pattern of correlations of RT with intelligence. In addition, results of an analysis of the movement time (MT) bands provide evidence of the divergent validity of the worst performance rule. These data also support Jensen's (1985; Methodological and statistical advances in the study of individual differences. New York: Plenum. 1987; Speed of information processing and intelligence. Norwood, NJ: Ablex) argument for the separate measurement of RT and MT in all speed of information processing research. In sum, results of this study sunstantiate the Worst Performance rule for individual differences in RT variability and intelligence.

Reaction Time, Response Errors, and G

Humphreys (1989) hypothesized that psychometric g is more highly correlated with the number of response errors on elementary cognitive tasks than with the reaction time (RT) measures themselves. Although Humphreys' hypothesis may well hold for relatively complex RT tasks, results of a recent study do not substantiate his hypothesis with regard to some of the simplest tasks. In this study, the average correlation ( r) between g and number of response errors was .05, compared to -.20 for median RT and -.23 for intraindividual differences in RT (measured as the standard deviation of RTs over trials). Even after correction for attenuation, the average r between g and errors is lower (.02) than the average correlations between g and the RT measures.

Reaction Time and Intelligence

James McKeen Cattell's contributions to the development of psychology have been well recognized. His earliest efforts in experimental psychology, however, were regarded by himself and others as unsuccessful. Following in the tradition of his mentors, Wundt and Galton, Cattell tried to demonstrate that psychophysiological measures could be related to general intelligence and achievement, but unsatisfactory results from Clark Wissler's (1901) dissertation combined with the wide acceptance of Alfred Binet's model of measuring intelligence led him to abandon his research and devote his considerable energies to other professional endeavors. Contemporary researchers have recently revived the basic research paradigm that Cattell was using, and considerable support has been found for the hypothesis that individual differences in elementary cognitive tasks are related to general intelligence. It is possible that intelligence theories of the future may owe as much to the Galtonian model of intelligence and its adherents like Cattell as they owe to Binet.

Still no demonstration that g is not unitary: Further comment on Kranzler and Jensen

In their reply to my critique (Carroll, 1991), Kranzler and Jensen (1991b) have still not demonstrated that a factor of general intelligence, g, depends on or “contains” a number of independent processes. All that they have demonstrated is that an estimate of g is predictable from a number of independent components statistically derived from a matrix of correlations among various elementary cognitive tasks (ECTs). But on the basis of correlations between ECTs and psychometric tests, the estimate of g derived from the psychometric tests is likely to contain variance identifiable with variance from ECTs that is independent of g, and to that extent, the estimate of g contains “impurities” which are predictable from ECTs. This does not necessarily imply that g itself contains a number of independent components. I point out sources of Kranzler and Jensen's misunderstanding of this argument, and make suggestions as to possible approaches to assessing whether g is unitary or not.

The nature of psychometric g: Unitary process or a number of independent processes?

This study investigates whether a unitary elemental process or a number of independent elemental processes, as measured by elementary cognitive tasks (ECTs), underlie psychometric g. A sample of 101 university students was administered two intelligence tests (Raven's Advanced Progressive Matrices and the Multidimensional Aptitude Battery) and a large battery of ECTs. The results of this study reject the theory that some single or unitary process underlies psychometric g. Rather, it appears that individual differences in psychometric g may reflect as many as four independent components of variance. These findings support the theory that various complex mental tests correlate highly with each other, giving rise to a psychometric g factor, because they require some of the same elemental processes. Further research will be needed to determine precisely the number and nature of these components. It is also important to note that the multiple correlation of g regressed on these four components derived from elementary cognitive variables is .542. The maximum correlation possible between the psychometric variables and the battery of ECTs in this study is nearly as high as correlations among various standardized IQ tests themselves (canonical r=.603). After correction for the considerable restriction of range on IQ in the sample, the r is increased to .722. Hence, this battery of ECTs accounts for approximately half of the phenotypic variance in g and probably as much as 70% of the genotypic variance. Moreover, the finding that individual differences in conceptually distinct processes (such as speed of visual search and speed of memory search) are highly correlated indicates the presence of individual differences in some neurological level of processing common to both tasks.

No demonstration that g is not unitary, but there's more to the story: Comment on Kranzler and Jensen

Because they employed a statistical procedure that was inappropriate for the purpose, Kranzler and Jensen (1991) have not demonstrated that a factor of general intelligence, g, depends on a number of independent processes. It is more parsimonious to interpret their results in terms of a g that is unitary. A factorial reanalysis of their data, including some that they did not present (cross-correlations between psychometric variables and variables from elementary cognitive tasks, or ECTs), shows that many ECT variables, especially those reflecting decision time rather than movement time, have substantial loadings on a second-order general factor. This factor is orthogonal to a second-order factor representing general speed of movement from a “home button” to the response button in ECTs. Possible implications for interpreting g in terms of information-processing efficiency are discussed.

Unitary g: Unquestioned postulate or Empirical fact?

Carroll (1991) has argued that our empirical test of the hypothesis that psychometric g is a unitary factor fails methodologically to prove that g is not unitary, and that our finding could have resulted from some impurity in the g extracted from 11 psychometric tests. The gist of this argument is that our multiple regression method for testing the unity of g and the outcome of this test would be valid only if it were certain that we had a “perfect estimate” of g as the dependent variable. We argue that the hypothetical ideal of a perfectly pure g is empirically unattainable, but that such purity is an unnecessary condition for testing the hypothesis by the method we used. Our analyses suggest that one would have to assume an improbably large amount of “impure” variance in our g factor to make Carroll's argument compelling. Finally, we are most grateful for Carroll's elegant hierarchical factor analysis of our psychometric and chronometric variables. The unity of g cannot be proved or disproved by factor analytic methods per se and the unitary g hypothesis has only the status of a parsimonious assumption within that framework. But Carroll's factor analysis of our data indeed beautifully represents the relationship between conventional psychometric tests and elementary cognitive tasks based on chronometric techniques and further highlights the central role of efficiency (= speed and consistency) of information processing in g.

Simple reaction time as a suppressor variable in the chronometric study of intelligence

Individual differences in reaction time (RT) to various elementary cognitive tasks (ECTs) reflect variance in both peripheral (sensorimotor) and central components of information processing. Minimizing the variance associated with peripheral processes by controlling simple RT in chronometric studies of more complex ECTs involving choice, discrimination, memory scanning, or other central processes, can increase the correlation between RTs and scores on complex psychometric tests of ability, thereby identifying more clearly the elementary processes involved in psychometric g and other abilities. Simple RT can be controlled by subtraction, partial correlation, and multiple correlation. The relative effectiveness of the different methods depends on various features of the chronometric data. The methods are explicated, with examples from a study of simple, choice, and discriminative RTs (the odd man out paradigm) in 213 male students from three colleges.

Speed of information processing in a calculating prodigy

Shakuntala Devi, one of the world's most prodigious mental calculators on record, past or present, is especially remarkable for the incredible speed with which she performs mental calculations on very large numbers. This rare phenomenon prompted the question of whether such exceptional performance dependes on the speed of elementary information processes. Devi's rather unexceptional reaction times on a battery of elementary cognitive tasks, which were compared with the mean RTs of college students and older adults on the same tasks, contrasts so markedly with her amazing speed of performing huge arithmetic calculations as to indicate that her skill with numbers must depend largely on the automatic encoding and retrieval of a wealth of declarative and procedural information in longterm memory rather than on any unusual basic capacities. Some kind of motivational factor that sustains enormous and prolonged interest and practice in a particular skill probably plays a larger part in extremely exceptional performance than does psychometric g or the speed of elementary information processes.

Effect of Instructions on Elementary Cognitive Tasks Sensitive to Individual Differences

The effect of instructions on basic cognitive tasks was investigated. In the first study, 60 college students completed both a choice reaction time and a modified match-to-sample task. Students were given either written, non-verbal, or no instructions. Mean level of performance changed across type of instruction for each task. The correlations between the task parameters and a measure of general intelligence also appeared to change across instructional conditions. A second study was conducted with 464 Air Force and National Guard recruits who completed choice reaction time, match-to-sample, tachistoscopic threshold, and probed recall tasks. Approximately half of these subjects received standard written instructions while the others received no instructions. Mean group differences were substantial but decreased with practice. Patterns of individual differences between the instruction and no-instruction conditions varied more with increased task complexity, with greater change occurring in more complex tasks such as probed recall. In sum, instructions have a significant impact on mean performance on basic cognitive tasks and a lesser effect on individual differences.

Process differences and individual differences in some cognitive tasks

This study is based on three distinct elementary cognitive tasks (ECTs) using chronometric techniques: (a) the S. Sternberg memory scan task, (b) a visual scan task which is perfectly analogous to the memory scan, except that the target digit is presented first and the subject must then scan a set of digits and indicate the presence or absence of the target digit in the set, and (c) the Hick paradigm, which involves responding to a visual stimulus (a light going “on”) when the stimulus is one among sets of either 1, 2, 4, or 8 equally probable alternatives. Certain parameters of all three tasks, such as intercept and slope of RT as a function of set size, from which different cognitive processes are inferred, are compared experimentally and correlationally. Subjects were 48 university students, tested and retested on the three tasks in a counterbalanced design on two separate days to obtain the retest reliabilities needed to correct all correlations for attenuation. Subjects were also given Raven's Advanced Progressive Matrices as a measure of psychometric g. Parameters of the ECTs are significantly and, in some cases, quite substantially correlated with g. Virtually all of this correlation is due to the general factor of the various ECTs, rather than to specific processing components (independent of the general factor). The results also indicate that different ECT paradigms (e.g., visual search and memory search tasks) may yield markedly different values in terms of the group means of analogous parameters, indicating different processes, and yet not show independence of the parameters in terms of individual differences; that is, measurements of the different parameters in individuals are perfectly correlated (i.e., disattenuated correlations do not differ significantly from r = 1). This is found for the intercept parameter of visual and memory search. The reverse condition is also found for the other parameters (e.g., slope); that is, their mean values are nearly identical, suggesting the same processes, yet disattenuated correlations between individual differences are relatively low, or even negative, indicating different processes. Although the general factor clearly predominates, it does not completely overwhelm individual differences in various component processes that are distinct from the general factor.

G: Artifact or reality?

The highest common factor in any large and diverse collection of mental tests is measured by means of factor analysis, and is conventionally labeled psychometric g (for general ability). The g factor, which is highly correlated across even quite different batteries of tests, provided the tests are fairly numerous and varied, reflects the empirical fact of positive manifold, that is, positive correlations between all mental tests. After briefly explicating the general psychometric conditions and factor analytic methods for the measurement of g, this article addresses the theoretically important question of whether g is merely an artifact of the method of constructing psychometric tests and the mathematical operations of factor analysis or whether it has an authentic claim to represent some natural phenomenon that exists independently of psychometrics and factor analysis. Several lines of evidence which refute the argument that g is a methodological artifact are presented. The g factor, far more than any other linearly independent sources of variance in psychometric tests, is correlated with various phenomena that are wholly independent of both psychometrics and factor analysis, such as the heritability of test scores, familial correlations, the effects of inbreeding depression and of hybrid vigor, evoked electrical potentials of the brain, and reaction times to elementary cognitive tasks which have virtually no intellectual content. This evidence of biological correlates of g supports the theory that g is not a methodological artifact but is, indeed, a fact of nature. However, the causal nature of g itself is not yet scientifically established. That goal must await further advances in neuroscience.

CORRELATIONS AMONG ELEMENTARY COGNITIVE TASKS

Summary. Eight elementary cognitive tasks were selected because of their potential application to the study of flexible thinking. The intercorrelations between these tasks were computed to investigate whether individual differences in performance could be found. These correlations are discussed in terms of the cognitive processes underlying the tasks, and it is concluded that seven of the eight tasks hold promise for future evaluation.