The complexity continuum in the radex and hierarchical models of intelligence

doi:10.1016/0160-2896(83)90023-5

Intelligence

Volume 7, Issue 2, April–June 1983, Pages 107-127

https://doi.org/10.1016/0160-2896(83)90023-5 Get rights and content

Abstract

The hierarchical and radex models of ability organization are shown to be parallel. Both models suggest a complexity continuum along which cognitive performance tasks can be arrayed. In our revised radex model, the complexity continuum from the center to the periphery is shown to correspond to the general-to-specific dimension in factor analyses, or to test correlations with the general factor; complexity is redefined as apparent processing complexity. Examination of the theoretical and empirical bases for this continuum indicates its central importance for theories of intelligence.

References (46)

R.E. Snow
Aptitude processes
L. Terman
Concept mastery test
(1950)
R.B. Cattell
Abilities: Their structure and growth
(1971)
R.W. Coan
Facts, factors, and artifacts: The quest for psychological meaning
Psychological Review
(1964)
L.J. Cronbach
Essentials of psychological testing
(1970)
L.J. Cronbach et al.
Aptitudes and instructional methods
(1977)
J.W. French et al.
Kit of reference tests for cognitive factors
(1963)
J.P. Guilford
The nature of human intelligence
(1967)
J.P. Guilford
Cognitive psychology's ambiguities: Some suggested remedies
Psychological Review
(1982)
L. Guttman
A new approach to factor analysis: The radex

L. Guttman

The structure of the interrelations among intelligence tests

L. Guttman

Integration of test design and analysis

J.R. Haynes

Hierarchical analysis of factors in cognition

American Educational Research Journal

(1970)

J.L. Horn

Human abilities: A review of research and theory in the early 1970s

Annual Review of Psychology

(1976)

L.G. Humphreys

The organization of human abilities

American Psychologist

(1962)

L.G. Humphreys

The primary mental ability

A.R. Jensen

Can we boost intelligence and academic achievement?

Harvard Educational Review

(1969)

A.R. Jensen

Hierarchical theories of mental ability

A.R. Jensen

Test validity: g versus the specificity doctrine

S.C. Johnson

Hierarchical clustering schemes

Psychometrika

(1967)

J.B. Kruskal

Multidimensional scaling by optimizing goodness of fit to a non-metric hypothesis

Psychometrika

(1964)

J.B. Kruskal

Nonmetric multidimensional scaling: A numerical method

Psychometrika

(1964)

J.B. Kruskal et al.

How to use KYST, a very flexible program to do multidimensional scaling and unfolding

(1973)

Cited by (261)

How useful are specific cognitive ability scores? An investigation of their stability and incremental validity beyond general intelligence
2024, Intelligence
Many intelligence tests measure multiple specific cognitive abilities. Practitioners use these specific ability scores, which encompass both specific ability and general intelligence variance, and the resulting intelligence profiles to make counseling and intervention decisions. In the present study, we investigated the temporal stability of eight specific abilities and their profiles over one school year, as well as their incremental validity in the prediction of school grades with German grade 7 to 9 students (N = 326 at T1; N = 311 at T2; N = 257 with IQ data at both times of measurement). The mean rank-order stability was 0.80 and ranged from 0.71 to 0.85. Intelligence profiles replicated significantly above chance levels (Mdn_κ = 0.31). The incremental validity coefficients were mostly small, but the Reasoning score substantially contributed to the prediction of math grades (ΔR² = 0.07–0.09), the Verbal Ability score to the prediction of German grades (ΔR² = 0.05–0.09), and the Crystallized Intelligence score to the prediction of geography grades (ΔR² = 0.03–0.08) beyond the general intelligence score. Our study of specific ability scores indicated moderate to high rank-order stability, fair to moderate profile stability, and substantial incremental validity for some specific ability scores.
Using a multi-strategy eye-tracking psychometric model to measure intelligence and identify cognitive strategy in Raven's advanced progressive matrices
2023, Intelligence
Previous studies have found that participants use two cognitive strategies—constructive matching and response elimination—in responding to items in the Raven's Advanced Progressive Matrices (APM). This study proposed a multi-strategy psychometric model that builds on item responses and also incorporates eye-tracking measures, including but not limited to the proportional time on matrix area (PTM), the rate of toggling (ROT), and the rate of latency to first toggle (RLT). By jointly analyzing item responses and eye-tracking measures, this model can measure each participant's intelligence and identify the cognitive strategy used by each participant for each item in the APM. Several main findings were revealed from an eye-tracking-based APM study using the proposed model: (1) The effects of PTM and RLT on the constructive matching strategy selection probability were positive and higher for the former than the latter, while the effect of ROT was negligible. (2) The average intelligence of participants who used the constructive matching strategy was higher than that of participants who used the response elimination strategy, and participants with higher intelligence were more likely to use the constructive matching strategy. (3) High-intelligence participants increased their use of the constructive matching strategy as item difficulty increased, whereas low-intelligence participants decreased their use as item difficulty increased. (4) Participants took significantly less time using the constructive matching strategy than the response elimination strategy. Overall, the proposed model follows the theory-driven modeling logic and provides a new way of studying cognitive strategy in the APM by presenting quantitative results.
What's so hard about hierarchical control? Pinpointing processing constraints within cue-based and serial-order control structures
2023, Cognitive Psychology
Most task spaces require a hierarchical structure, where decisions on one level are contingent on previous decisions made on one or more higher levels. While it is a truism that increasing the number of hierarchical levels makes it harder to solve a given task, the exact nature of this “number-of-levels” effect is not clear. On the one hand, processing costs might be strictly “local,” incurred only when higher-level settings need to be updated, while otherwise lower-level decisions are insulated from the presence of higher-level settings (local updating costs with ballistic control). On the other hand, maintaining and integrating more complex hierarchical structures could require overall greater representational resources, negatively affecting each individual decision within the represented task space (global integration/maintenance costs). Further, navigation through hierarchical structures can be guided either through prompts in the environment (cue-based), or through sequential plans (serial-order based), with potentially distinct maintenance and updating demands. In two experiments, we assessed performance as a function of hierarchical level and format (serial-order vs. cue-based). Model comparisons showed that the pattern of costs in the serial-order format was consistent with a global maintenance/integration account. In contrast, in the cue-based format, costs arose at updating points and when one additional relevant level beyond the current decision was relevant, while additional levels produced no further costs. This overall constellation of costs can be explained by assuming that each decision requires checking the immediately relevant higher-level context for that decision. For cue-based control, this context involves the “next-level-up” rule, whereas in the serial-order format, each trial requires updating of the current position within the sequence, which in turn requires integration across all relevant hierarchical levels.
To predict the future, consider the past: Revisiting Carroll (1993) as a guide to the future of intelligence research
2021, Intelligence
There is a widely held consensus in the field of intelligence research that the broad factors identified by Cattell, Horn, and Carroll are an adequate summary of individual differences in human cognitive abilities. Most researchers would agree that the redundancy among these factors is best accounted for by an overarching general factor. We think the best way to acknowledge major accomplishments is to build upon them with the goal to challenge the status quo. Here we want to do so by discussing six broad ability factors that are either considered in Carroll's epochal book or could be candidates for future inclusions to the list of established cognitive ability factors: fluid intelligence, crystallized intelligence, cognitive speed, creativity, social and emotional intelligence, and collaborative problem solving. We conclude with four pleas: reunite correlational and experimental research, enrich construct interpretations, reunite educational and psychological measurement of maximal cognitive effort, and reconsider the sampling of indicators and content validity.
Intelligence matters for stochastic feedback processing during sequence learning in adolescents and young adults
2021, Intelligence
The ability to effectively handle partly false feedback is an important skill because it can disrupt an already mastered ability. An instant and drastic deterioration occurred in 8- to 11-year-old children's sequence learning of four left-right button presses from close to ceiling performance with 100% deterministic feedback to about 30% accuracy when they encountered stochastic feedback that was only 85% correct and 15% randomly false (Lange-Küttner et al., 2012). However, children's performance recovered in repeated trials, but only after positive feedback even if it was false. The present study investigates whether this deterioration and recovery from partly false feedback still occurs in adolescents and young adults. We tested whether coping with partly false feedback is dependent on intelligence with the Raven Standard Progressive Matrices Test, or rather occurs because of better developed coping and defending in the social domain with the Cognitive Emotional Regulation Questionnaire (CERQ). In a sample of N = 75 adolescents from 13 years onwards until young adulthood, cognitive performance was low when encountering stochastic feedback, at 55.5% accuracy on average, but improved during repetitions. Intelligence was more important than emotional coping for successful processing of stochastic, partly false feedback. Also in adolescence and young adulthood, positive feedback appeared to support perseverance as it yielded significantly more recovery than negative feedback. Coping was not needed for recovery from partly false or negative feedback, but there were some rare cases of high intelligence and low positive coping skills with exceptionally low sequencing performance.
Is there a “g-neuron”? Establishing a systematic link between general intelligence (g) and the von Economo neuron
2021, Intelligence
The construct of general intelligence (g) is one of psychology’s most replicated and predictively useful constructs. Although research indicates that g is a highly heritable trait, deeply rooted in brain physiology, to date neither a strong biological correlate nor a comprehensive explanatory model involving neuronal mechanisms have been established. In this article I aim to do so by hypothesising that the von Economo neuron (VEN), a unique nerve cell thus far implicated in social cognition and interoception, may in fact represent a central biological constituent of g. After presenting supportive evidence from neuroscience, psychiatry/neurology, clinical gerontology, and comparative psychology, an integrated reductionist framework is outlined, which reaches from the level of cognitive theory to the level of single neurons. Based thereon, it is concluded that the VENs might contribute to individual differences in g by rapidly inducing the coherence of neuronal oscillations within a functionally invariant parieto-frontal network underlying higher-order cognition, thereby facilitating mental efficiency.

View all citing articles on Scopus

^☆: This article is based on a paper presented by the first author in the symposium, “Research on Aptitude Process,” at the meeting of the American Psychological Association, San Francisco, August 1977. The work reported herein was sponsored by the Personnel and Training Programs, Office of Naval Research and the Advanced Research Projects Agency under Contract Nos. N00014-75-C-0882 and N00014-79-C-0171. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the Office of Naval Research, the Advanced Research Projects Agency, or the U.S. Government.

¹: We gratefully acknowledge the contribution of Patrick Kyllonen who commented on an earlier version of this paper. David F. Lohman is now at the University of Iowa.

View full text

The complexity continuum in the radex and hierarchical models of intelligence☆

Abstract

Abilities: Their structure and growth

Facts, factors, and artifacts: The quest for psychological meaning

Psychological Review

Essentials of psychological testing

Aptitudes and instructional methods

Kit of reference tests for cognitive factors

The nature of human intelligence

Cognitive psychology's ambiguities: Some suggested remedies

Psychological Review

A new approach to factor analysis: The radex

The structure of the interrelations among intelligence tests

Integration of test design and analysis

Hierarchical analysis of factors in cognition

American Educational Research Journal

Human abilities: A review of research and theory in the early 1970s

Annual Review of Psychology

The organization of human abilities

American Psychologist

The primary mental ability

Can we boost intelligence and academic achievement?

Harvard Educational Review

Hierarchical theories of mental ability

Test validity: g versus the specificity doctrine

Hierarchical clustering schemes

Psychometrika

Multidimensional scaling by optimizing goodness of fit to a non-metric hypothesis

Psychometrika

Nonmetric multidimensional scaling: A numerical method

Psychometrika

How to use KYST, a very flexible program to do multidimensional scaling and unfolding