EPPP: PART 1, KNOWLEDGE • DOMAIN 2: COGNITIVE-AFFECTIVE BASES

Intelligence Models — Differentiate major models of intelligence and apply them to assessment scenarios

Understanding the theoretical foundations that shape how clinicians measure and interpret cognitive ability.

Historical Context & Motivation

The scientific study of intelligence arose from a fundamentally practical need: how do we objectively evaluate an individual's cognitive capacity and predict future performance? Before the late nineteenth century, judgments about intellectual ability were informal, subjective, and often tainted by cultural bias. The emergence of standardized testing and factor-analytic methods transformed intelligence into a measurable construct, though debates about its underlying structure have persisted for over a century. For clinicians preparing for the EPPP, understanding these theoretical models is not merely academic—it directly informs how you select, administer, and interpret cognitive assessment instruments in practice.

The evolution of intelligence theory can be understood as an ongoing tension between lumpers—those who emphasize a single, general cognitive ability—and splitters—those who argue that intelligence is better conceived as multiple, relatively independent faculties. This dialectic has produced an increasingly nuanced set of models, each with distinct implications for psychological assessment.

1904
Spearman's g Factor
Charles Spearman used early factor analysis to propose a single general intelligence factor (g) that accounts for the positive correlations observed across diverse cognitive tests.
1938
Thurstone's Primary Mental Abilities
L.L. Thurstone challenged Spearman by identifying seven primary mental abilities (e.g., verbal comprehension, spatial visualization), arguing that no single g adequately explains cognitive performance.
1963
Cattell's Fluid & Crystallized Intelligence
Raymond Cattell proposed a hierarchical resolution, distinguishing fluid intelligence (Gf)—novel problem-solving capacity—from crystallized intelligence (Gc)—accumulated knowledge.
1983
Gardner's Multiple Intelligences
Howard Gardner proposed at least eight distinct intelligences (linguistic, logical-mathematical, musical, etc.), broadening the definition of intelligence well beyond conventional psychometric boundaries.
1993–2012
CHC Theory Integration
The Cattell-Horn-Carroll (CHC) theory synthesized decades of factor-analytic research into a comprehensive three-stratum model that now serves as the dominant framework underlying modern intelligence batteries like the WISC-V and WJ IV.

The central question that drives these evolving models remains clinically urgent: Is intelligence best understood as one thing, several related things, or many independent things? Your answer determines which instruments you select, how you interpret index scores, and what you communicate to clients and referral sources.

Core Principles & Foundational Models

Every model of intelligence rests on assumptions about the latent structure of cognitive abilities—specifically, how many distinct factors exist and how they relate to one another. These assumptions are tested through factor analysis, a statistical technique that identifies clusters of intercorrelated variables. When a person who scores highly on vocabulary also tends to score highly on reading comprehension, factor analysis detects this shared variance and posits an underlying factor (e.g., verbal ability). The number, breadth, and hierarchical arrangement of these factors differ across competing models, and each arrangement carries direct implications for test construction and clinical interpretation.

1

Spearman's Two-Factor Theory

Intelligence consists of a general factor (g) common to all cognitive tasks, plus task-specific factors (s). The g factor explains the positive manifold—the finding that performance on any cognitive test correlates positively with performance on every other test.
2

Cattell-Horn Gf-Gc Model

Fluid intelligence (Gf) represents the capacity to reason with novel stimuli, while crystallized intelligence (Gc) captures acculturated knowledge. Gf peaks in early adulthood and declines; Gc remains stable or increases across the lifespan.
3

Carroll's Three-Stratum Theory

John Carroll's 1993 re-analysis of over 460 data sets yielded a three-stratum hierarchy: Stratum III (g), Stratum II (broad abilities like Gf, Gc, Gv, Gs), and Stratum I (narrow abilities such as induction, lexical knowledge, and perceptual speed).
4

Sternberg's Triarchic Theory

Robert Sternberg proposed three interacting subtheories: analytical (componential), creative (experiential), and practical (contextual) intelligence, emphasizing real-world adaptive competence beyond traditional test performance.
5

Gardner's Multiple Intelligences

Gardner defined intelligence as the ability to solve problems or create products valued in a cultural setting. He identified at least eight independent intelligences (linguistic, logical-mathematical, spatial, musical, bodily-kinesthetic, interpersonal, intrapersonal, naturalistic), arguing that standard IQ tests capture only a narrow slice.
KEY TAKEAWAY
Think of the debate about intelligence models like the question of whether music is one thing or many. A music theorist might say 'musicality' is a single talent—analogous to Spearman's g. A different scholar might separate rhythm, melody, harmony, and timbre as independent skills—analogous to Thurstone or Gardner. CHC theory offers a middle path: musicality exists as a general capacity, but it is composed of distinct broad abilities, each of which can be independently measured and trained. In clinical practice, the model you adopt shapes whether you report a Full-Scale IQ alone, emphasize index discrepancies, or recommend domain-specific interventions.

Visual Explanation — The Hierarchical Structure of Intelligence

The following diagram illustrates the Cattell-Horn-Carroll (CHC) three-stratum model, which is the most empirically supported and clinically influential framework for understanding intelligence today. The model is hierarchical: a single general factor (g) sits at the apex (Stratum III), broad ability factors occupy the middle tier (Stratum II), and narrow abilities form the base (Stratum I). Modern intelligence batteries such as the Wechsler scales, the Stanford-Binet 5, and the Woodcock-Johnson IV are organized to measure constructs at each stratum level.

The CHC model arranges cognitive abilities into three strata. Stratum III is the general factor g; Stratum II includes broad abilities such as Gf, Gc, Gv, Gs, and Gsm; Stratum I consists of narrow abilities like induction, vocabulary, and working memory span. Most contemporary intelligence tests are designed to yield scores at all three levels.

In the diagram above, note that every narrow ability (Stratum I) loads onto a specific broad factor (Stratum II), and all broad factors load onto g (Stratum III). This hierarchical structure explains why a composite Full-Scale IQ can be meaningful (it approximates g) while simultaneously justifying the interpretation of index-level scores (which approximate Stratum II factors). When a clinician observes a statistically significant discrepancy between, say, the Verbal Comprehension Index and the Working Memory Index on a Wechsler battery, the CHC model provides the theoretical rationale for treating those indices as measuring genuinely distinct constructs.

Psychometric Foundations — Factor Analysis and Score Interpretation

The mathematical engine behind every structural model of intelligence is factor analysis. In its exploratory form (EFA), the technique examines a correlation matrix of test scores to identify the smallest number of latent variables (factors) that can account for the observed pattern of intercorrelations. Confirmatory factor analysis (CFA) then tests whether a hypothesized factor structure provides a good fit to the data. Both procedures rely on the concept of factor loadings—the correlation between an observed variable and the latent factor it is presumed to measure.

FACTOR MODEL (SIMPLIFIED)
Xᵢ = λᵢ₁F₁ + λᵢ₂F₂ + … + λᵢₖFₖ + eᵢ
Where Xᵢ = observed score on test i, λᵢₖ = factor loading of test i on factor k, Fₖ = latent factor score, and eᵢ = unique/error variance. High loadings indicate that a test is a strong indicator of the factor.

In practice, clinicians rarely compute factor loadings themselves. Instead, they rely on the standardized scoring frameworks built into published instruments. Modern IQ batteries report scores at three levels that map directly onto the CHC strata. Understanding this mapping is essential for appropriate test interpretation.

STANDARD SCORE CONVERSION
Standard Score = (X − M) / SD × 15 + 100
Where X = raw score, M = normative mean, and SD = normative standard deviation. This yields a distribution with mean = 100 and SD = 15, the standard metric for Wechsler-family instruments.
CONFIDENCE INTERVAL
CI = Obtained Score ± z × SEM
Where SEM = standard error of measurement = SD × √(1 − rₓₓ), and rₓₓ = reliability coefficient. Clinicians use the 95% CI (z = 1.96) to determine whether observed score differences are statistically meaningful.
📊 Clinical Application Note
On the EPPP, you may be asked to determine whether a discrepancy between two index scores is statistically significant. The key is to compare the observed difference against the critical value provided in the test manual (which itself is derived from the SEMs of both indices and their intercorrelation). A statistically significant difference does not automatically imply clinical significance—base rate tables indicate how common such discrepancies are in the normative population.

Detailed Model Comparison — From Spearman to Contemporary Frameworks

To prepare for EPPP questions, you must be able to quickly discriminate between models based on their structural features, the number of factors they propose, and the instruments with which they are most closely associated. The diagram below contrasts the major models visually, and the subsequent table provides a side-by-side comparison of their most clinically relevant features.

Side-by-side structural comparison of four major intelligence models. Spearman's two-factor model places g at the center; Thurstone distributes intelligence across seven independent primary abilities; CHC theory integrates both views hierarchically; and Sternberg's triarchic model emphasizes functional aspects of intelligence. Gardner's framework (bottom) treats each intelligence as a fully separate domain with no overarching g.
Summary of Major Intelligence Models and Their Clinical Relevance
ModelStructureKey ConstructsAssociated InstrumentsEPPP Relevance
Spearman (1904)Two-factor: g + s factorsGeneral intelligence (g); specific factors (s)Raven's Progressive Matrices (strong g loading)Foundation for understanding FSIQ; positive manifold
Thurstone (1938)Multifactorial; no hierarchical gSeven Primary Mental Abilities (V, N, S, W, P, M, R)Primary Mental Abilities Test (PMA)Precursor to index-score interpretation
Cattell-Horn (1963–1966)Gf-Gc dichotomy (originally no g)Fluid (Gf) and Crystallized (Gc) intelligenceCulture Fair Intelligence Test; WJ (earlier editions)Gf-Gc distinction central to lifespan development questions
Carroll (1993)Three-stratum hierarchyStratum III (g), II (broad), I (narrow)Empirical meta-framework; basis for CHCValidates hierarchical test structure
CHC (integrated)Three strata; ~16 broad abilitiesGf, Gc, Gv, Ga, Gs, Gsm, Glr, Grw, Gq, etc.WISC-V, WAIS-IV, WJ IV, SB5, KABC-IIDominant framework; know which CHC factor each index measures
Sternberg (1985)Triarchic: three interacting subtheoriesAnalytical, Creative, Practical intelligenceSternberg Triarchic Abilities Test (STAT)Emphasizes ecological validity; practical intelligence concept
Gardner (1983)Multiple independent intelligences8+ intelligences (linguistic, musical, etc.)No standardized psychometric instrumentKnow critiques: lacks psychometric support; influential in education

Worked Example — Applying Intelligence Models to an Assessment Scenario

Consider the following clinical scenario: A 9-year-old child is referred for evaluation due to academic difficulties in reading and math despite seemingly strong verbal skills in conversation. The psychologist administers the WISC-V and obtains the following index scores: Verbal Comprehension Index (VCI) = 118, Visual Spatial Index (VSI) = 105, Fluid Reasoning Index (FRI) = 94, Working Memory Index (WMI) = 82, and Processing Speed Index (PSI) = 78. The Full-Scale IQ (FSIQ) is 95. How would different intelligence models guide interpretation?

Interpreting a WISC-V Profile Using Intelligence Models
1
Step 1 — Evaluate the Meaningfulness of the Full-Scale IQUsing the CHC framework, we first ask whether the FSIQ of 95 is a valid summary of this child's cognitive abilities. The discrepancy between the highest index (VCI = 118) and the lowest (PSI = 78) is 40 points. A difference this large suggests substantial variability across Stratum II abilities, meaning the FSIQ (a Stratum III estimate) may not be the most informative or interpretable score. In Spearman's model, the FSIQ would be treated as a straightforward estimate of g, but the CHC model and cross-battery assessment approaches would caution against over-relying on it when significant index-level variability exists.
FSIQ of 95 is likely not interpretable as a unitary estimate of g given 40-point index variability.
2
Step 2 — Analyze the Index-Level Profile Through CHC LensesEach WISC-V index corresponds to a CHC Stratum II factor: VCI → Gc (crystallized intelligence), VSI → Gv (visual processing), FRI → Gf (fluid reasoning), WMI → Gsm (short-term memory), and PSI → Gs (processing speed). The profile reveals a clear pattern: this child's crystallized knowledge (Gc = 118, High Average) significantly exceeds their processing speed (Gs = 78, Borderline) and working memory capacity (Gsm = 82, Low Average). From a Cattell-Horn perspective, the Gf-Gc distinction is critical: the child's accumulated knowledge and verbal skills (Gc) are well-developed, but their capacity for effortful, speeded processing (Gs) and memory manipulation (Gsm) are significantly weaker.
Profile shows Gc strength (118) with Gs (78) and Gsm (82) weaknesses—a pattern consistent with processing efficiency deficits.
3
Step 3 — Consider Statistical Significance of DiscrepanciesUsing the test manual's critical value tables (derived from the SEM formula: SEM = SD × √(1 − rₓₓ)), we determine whether the VCI–PSI difference of 40 points exceeds the threshold for statistical significance at the .05 level. For the WISC-V, the critical value for VCI vs. PSI comparison is approximately 16–18 points at the .05 level. Since 40 > 18, this discrepancy is statistically significant. We then consult base rate tables: a discrepancy of ≥ 40 points between VCI and PSI occurs in approximately 5–8% of the normative sample, suggesting this pattern is also relatively uncommon.
VCI–PSI discrepancy of 40 points is statistically significant (p < .05) and relatively uncommon (base rate ≈ 5–8%).
4
Step 4 — Generate Hypotheses Using Multiple ModelsSternberg's triarchic model would prompt us to consider whether this child demonstrates practical or creative strengths that standardized tests do not capture—important for treatment planning and parent feedback. Gardner's theory would encourage attention to the child's strong verbal/linguistic intelligence and potential strengths in domains like interpersonal or musical intelligence that are not measured by the WISC-V at all. However, neither Sternberg's nor Gardner's framework provides the psychometric precision needed for diagnostic decision-making in this scenario. The CHC-aligned interpretation directly supports diagnostic considerations (e.g., specific learning disability with processing speed and working memory as cognitive processing weaknesses) and intervention planning (e.g., extended time accommodations, working memory strategy training).
CHC provides the most actionable clinical interpretation; Sternberg and Gardner add contextual breadth but lack psychometric specificity for diagnosis.

Strengths, Limitations, and Clinical Considerations

No single model of intelligence is without flaws, and the EPPP frequently tests your ability to recognize both the strengths and limitations of each framework. The following table distills the most frequently tested distinctions. Pay particular attention to the contrast between models with strong psychometric support (Spearman, CHC) and those with broader theoretical appeal but weaker empirical foundations (Gardner, Sternberg).

Strengths and Limitations of Major Intelligence Models
ModelStrengthsLimitations
Spearman's gParsimonious; explains positive manifold; strong predictive validity for academic and occupational outcomes; supported by massive factor-analytic literatureOversimplifies cognitive diversity; does not account for specific ability profiles; culturally biased if g is equated with a single test score
ThurstoneAcknowledged multidimensionality; precursor to index scores and profile analysis; empirically grounded in factor analysisSeven PMAs are not fully independent (they intercorrelate, implying g); largely superseded by CHC
Cattell-Horn Gf-GcGf-Gc distinction has strong developmental and neurobiological support; differentiates age-related patterns; guides lifespan assessmentHorn denied g, which conflicts with robust evidence for a general factor; original model was incomplete (fewer broad factors)
CHC (integrated)Most comprehensive empirical framework; integrates g with broad and narrow abilities; underlies modern test construction; supports cross-battery assessmentComplexity can be overwhelming for clinical communication; ongoing debates about exact number of broad abilities; less attention to non-cognitive factors
SternbergEmphasizes real-world adaptive functioning; includes creative and practical domains; promotes ecological validityDifficult to operationalize and measure reliably; limited psychometric evidence; STAT has not gained clinical traction
GardnerBroadens definition of intelligence; validates diverse talents; influential in educational reform and multicultural sensitivityLack of psychometric evidence; intelligences may be better described as talents or aptitudes; no standardized measurement tool; conflates intelligence with skill
KEY TAKEAWAY
For the EPPP, remember this hierarchy of empirical support: CHC theory sits at the top as the most validated and clinically useful framework, followed by the Gf-Gc and Spearman g models. Sternberg and Gardner occupy a different niche—they are theoretically rich and broaden our conceptualization of intelligence, but they lack the psychometric rigor needed for standardized assessment and clinical decision-making. When an EPPP question asks about the 'best-supported' or 'most widely used' model, the answer is almost always CHC. When a question asks about broadening the definition of intelligence beyond traditional tests, think Gardner and Sternberg.

Connections to Advanced Theory — Neuropsychology and Contemporary Developments

Intelligence models do not exist in isolation from neuroscience and contemporary cognitive psychology. An emerging body of research seeks to ground the factor-analytic constructs of CHC theory in neural substrates, and several developments extend the classical models in ways that are increasingly relevant to EPPP content and clinical practice.

Classical Intelligence Concepts and Their Contemporary Extensions
Classical ConceptContemporary ExtensionClinical Implication
Spearman's g factorParieto-Frontal Integration Theory (P-FIT): g is associated with the efficiency of a distributed frontoparietal network, not a single brain regionNeuroimaging can complement cognitive testing; frontal lobe injuries disproportionately affect g-loaded tasks
Cattell-Horn Gf-GcInvestment Theory: Gc is conceptualized as the product of Gf 'invested' in culturally valued learning over timeExplains why Gf declines with age while Gc remains stable; relevant to geropsychology assessments
CHC broad abilitiesCross-Battery Assessment (XBA): systematic method for supplementing a core battery with subtests from other instruments to ensure comprehensive CHC coverageAllows clinicians to test specific hypotheses about Stratum II weaknesses when the primary battery has insufficient coverage
Sternberg's practical intelligenceEmotional Intelligence (EI): Salovey and Mayer's ability model of EI overlaps with Sternberg's contextual subtheory, focusing on perceiving, using, understanding, and managing emotionsEI measures (e.g., MSCEIT) complement traditional IQ batteries in clinical and organizational settings
Gardner's multiple intelligencesNeurodiversity frameworks: Gardner's emphasis on diverse cognitive profiles aligns with neurodiversity-affirming approaches that reject deficit-only modelsEncourages strength-based assessment and reporting alongside deficit identification

Looking forward, the integration of intelligence models with advances in computational cognitive modeling, dynamic assessment (which measures learning potential rather than static performance), and culturally responsive assessment practices will continue to reshape how clinicians conceptualize and measure intelligence. For EPPP preparation, remain alert to questions that bridge classical models with these contemporary trends, particularly the distinction between static and dynamic assessment approaches, and the role of cultural factors in test selection and interpretation.

Practice Problems

PROBLEM 1CONCEPTUAL
A psychologist argues that a single composite IQ score is the most valid and useful summary of a client's cognitive functioning, regardless of the variability among subtest scores. Which theoretical model of intelligence most closely aligns with this position, and what is the primary assumption underlying it?
PROBLEM 2BASIC CALCULATION
A client obtains a Verbal Comprehension Index (VCI) of 112 on the WAIS-IV. The reliability coefficient (rₓₓ) for the VCI is .96, and the index has a standard deviation of 15. Calculate the 95% confidence interval for this score.
PROBLEM 3INTERMEDIATE
A school psychologist is evaluating a bilingual 8-year-old who immigrated to the United States two years ago. On the WISC-V, the child's Fluid Reasoning Index (FRI = 108) is significantly higher than her Verbal Comprehension Index (VCI = 82). Using CHC theory and the Cattell-Horn Gf-Gc distinction, explain the likely meaning of this pattern and identify what additional assessment considerations are warranted.
PROBLEM 4APPLIED
A neuropsychologist is assessing a 72-year-old client with suspected early-stage Alzheimer's disease. Historical records show the client earned a graduate degree and had a successful career as an attorney. Current testing reveals: VCI = 120, FRI = 88, WMI = 80, PSI = 75. Apply your knowledge of the Cattell-Horn Gf-Gc model and the CHC framework to explain why this specific pattern of scores is clinically significant for dementia evaluation.
PROBLEM 5CRITICAL THINKING
A colleague argues that Gardner's theory of multiple intelligences should replace CHC theory as the basis for psychological assessment because it is more inclusive and culturally sensitive. Construct a balanced critique of this argument, addressing (a) the theoretical merits of Gardner's position, (b) the psychometric requirements for clinical assessment instruments, and (c) how the strengths of both frameworks might be integrated in practice.

Summary — Intelligence Models and Assessment Applications

The study of intelligence has evolved from Spearman's single g factor through Thurstone's seven primary mental abilities to the integrative Cattell-Horn-Carroll (CHC) three-stratum model, which is now the dominant framework for intelligence test construction and interpretation. CHC theory organizes cognitive abilities hierarchically: a general factor (g) at Stratum III, approximately sixteen broad abilities (Gf, Gc, Gv, Gs, Gsm, etc.) at Stratum II, and numerous narrow abilities at Stratum I. The Gf-Gc distinction is especially important clinically because fluid intelligence (novel reasoning) and crystallized intelligence (accumulated knowledge) follow different developmental trajectories and are differentially affected by neurological conditions.

Alternative models enrich our conceptualization of intelligence without replacing CHC for formal assessment. Sternberg's triarchic theory emphasizes analytical, creative, and practical dimensions, while Gardner's multiple intelligences broadens the construct to include domains like musical and interpersonal ability. For the EPPP, know that CHC has the strongest empirical support, understand which instruments map onto which model, and be prepared to apply factor-analytic concepts (factor loadings, confidence intervals, index discrepancy analysis) to clinical scenarios. Remember: the model you adopt determines whether you interpret a Full-Scale IQ as definitive or supplementary, and how you communicate cognitive strengths and weaknesses to clients and referral sources.

Varsity Tutors • EPPP: Part 1, Knowledge • Intelligence Models — Differentiate major models of intelligence and apply them to assessment scenarios