Skip to main content

Critical Appraisals of Spearman's Theory

Introduction​

Scientific progress thrives on challenge and debate. While Spearman's two-factor theory revolutionized intelligence research, it also sparked vigorous criticism that ultimately enriched our understanding of cognitive abilities. Three prominent psychologists—E. L. Thorndike, G. H. Thomson, and L. L. Thurstone—developed alternative frameworks that fundamentally questioned Spearman's assumptions.

These critiques weren't merely academic disagreements; they represented different philosophical approaches to understanding human cognition. Should we search for a general faculty underlying all thinking, or recognize a collection of independent abilities? Is intelligence one thing or many things? These debates continue to resonate in contemporary cognitive science and educational practice.

Key Question: Does the observed correlation between cognitive tests reflect a single general factor, or could multiple specific abilities produce the same pattern of results?

The Theoretical Landscape​

Before examining each critique, let's understand the fundamental disagreement:

E. L. Thorndike's Multifactorial Theory​

The Challenge to General Intelligence​

Edward L. Thorndike (1926), the influential American psychologist, mounted a frontal assault on Spearman's concept of general intelligence. His critique operated at both empirical and conceptual levels.

Empirical Criticism: The Correlation Problem​

Thorndike's First Objection: The intercorrelations Spearman studied were too small to decisively test whether a common factor existed.

Statistical Argument:

  • When correlations between tests are moderate (r = 0.30-0.60), multiple factor structures could fit the data
  • Small sample sizes in early studies limited statistical power
  • Measurement error could obscure true relationships

Example: Two tests correlating at r = 0.45:

  • Spearman's interpretation: They share 20% variance through g
  • Thorndike's counter: Perhaps they share 20% variance because both require numerical manipulation, not because of general intelligence

Conceptual Criticism: Locating Intelligence​

Thorndike's Core Argument: Intelligence doesn't reside in the individual as a general capacity—it resides in the nature of the tasks themselves.

The Task Commonality Hypothesis:

Tests correlate not because people have varying amounts of general intelligence, but because tests share concrete requirements:

Task FeatureTests Sharing Feature
Verbal knowledgeVocabulary, reading comprehension, analogies
Numerical operationsArithmetic, calculation, quantitative reasoning
Spatial visualizationMental rotation, map reading, geometric reasoning
Processing speedReaction time, perceptual comparison, rapid decision

Individual Differences in Thorndike's Model​

Thorndike proposed two dimensions of ability:

1. Altitude (Level)

  • How difficult a task can an individual perform?
  • Example: Student A can solve simple equations; Student B can solve complex calculus problems

2. Range (Breadth)

  • How many different types of tasks can an individual perform?
  • Example: Student A excels only in mathematics; Student B performs well in mathematics, literature, and science

Critical Insight: People don't vary along a single "general intelligence" dimension. They vary in:

  • Maximum difficulty they can handle in specific domains
  • Number of domains where they show competence

Three Types of Intelligence​

Thorndike classified intellectual activities into three categories:

1. Abstract Intelligence​

Definition: Ability to understand and manipulate ideas, symbols, and relationships

Examples:

  • Solving mathematical problems
  • Understanding philosophical arguments
  • Creating logical proofs
  • Analyzing abstract concepts

2. Mechanical Intelligence​

Definition: Ability to understand and manipulate concrete objects and mechanisms

Examples:

  • Repairing machinery
  • Building structures
  • Understanding how physical systems work
  • Spatial reasoning with real objects

3. Social Intelligence​

Definition: Ability to understand and interact effectively with people

Examples:

  • Reading social cues
  • Predicting others' behavior
  • Managing interpersonal relationships
  • Navigating social situations

Important Note: These categories describe types of tasks, not underlying mental faculties. Thorndike avoided reifying intelligence as an entity.

Evaluation of Thorndike's Theory​

Strengths:

  1. Anticipated Modern Views: Contemporary theories (Gardner's multiple intelligences, Sternberg's triarchic theory) echo Thorndike's pluralistic approach
  2. Practical Relevance: Recognizing different types of competence has educational value
  3. Task Analysis: Focusing on task requirements anticipated cognitive psychology's approach

Limitations:

  1. Overlapping Categories: Abstract, mechanical, and social intelligence aren't completely independent—successful engineers need abstract and mechanical intelligence
  2. Doesn't Explain Correlations: If abilities are truly separate, why do diverse tests correlate positively (the "positive manifold")?
  3. Measurement Challenges: How do we separately assess "altitude" and "range"?

Guilford's (1953) Assessment: "There seems to be agreement that the types that interested Thorndike are essentially the same as the [group factors] which Spearman used in his correlation matrix."

Interpretation: Thorndike's types might represent group factors rather than completely independent abilities, suggesting his theory and Spearman's converge more than Thorndike acknowledged.

G. H. Thomson's Sampling Theory​

The Mathematical Challenge​

Godfrey H. Thomson (1939), a Scottish psychologist and statistician, proposed the most mathematically sophisticated alternative to Spearman. His "sampling theory" demonstrated that patterns appearing to support general intelligence could arise from completely different underlying structures.

The Core Argument: Bonds, Not a Factor​

Thomson's Revolutionary Insight: What if intelligence involves not a general factor, but a vast collection of independent "bonds" or "neural connections"?

The Sampling Hypothesis:

  1. The brain contains innumerable independent neural elements (bonds)
  2. Each cognitive task activates a sample of these elements
  3. Tests correlate when they sample overlapping sets of elements
  4. NO general factor exists—only statistical overlap from sampling

Why Tests Correlate: Tests 1 and 2 share many elements (5, 12, 15, 22), producing high correlation. Tests 1 and 3 share few elements, producing low correlation.

Mathematical Equivalence​

Thomson's Bombshell: Using factor analysis on correlation matrices, you cannot distinguish between:

  • Spearman's model: One general factor + specific factors
  • Thomson's model: Many independent elements sampled differently

Both produce identical correlation patterns!

Implications:

  • Factor analysis reveals structure in correlations, not necessarily structure in the mind
  • The existence of g as a statistical factor doesn't prove g exists as a psychological reality
  • Alternative explanations remain viable even when factor analysis "supports" g

The Hierarchical Approximation​

Thomson acknowledged that when tests sample many common elements:

"General Ability" emerges as:

  • A statistical regularity, not a basic entity
  • The constant combination of frequently co-sampled elements
  • An approximation of underlying complexity

Similarly:

  • Group factors: Combinations of more limited element collections
  • Specific factors: Elements appearing in single tests only

Neurological Plausibility​

Thomson grounded his theory in neural architecture:

Anatomical Support:

  • Cortex contains billions of neurons with trillions of connections
  • Complex tasks recruit distributed neural assemblies
  • Different tasks activate overlapping but non-identical networks
  • No single "intelligence center" exists in the brain

Modern Neuroscience Confirmation: Contemporary brain imaging shows:

  • Intelligence involves distributed networks
  • Different cognitive tasks activate overlapping brain regions
  • Network efficiency (connectivity quality) predicts performance
  • Supports element-sampling over single-factor models

Evaluation of Thomson's Theory​

Strengths:

  1. Parsimony: Explains intelligence without postulating unmeasurable "mental energy"
  2. Biological Plausibility: Consistent with neural network architecture
  3. Methodological Caution: Highlights limits of factor analysis interpretation
  4. Predictive Equivalence: Generates same predictions as Spearman's model

Limitations:

  1. Unfalsifiability: If statistically indistinguishable from g-theory, how can we test it?
  2. Element Identification: Guilford's (1953) critique: "There seems to be little likelihood of demonstrating experimentally the existence of the elements hypothesized"
  3. Group Factors: Doesn't easily explain why certain abilities (verbal, spatial, numerical) consistently cluster

Historical Verdict: Thomson demonstrated that factor-analytic results alone cannot reveal underlying mental architecture. This epistemological insight remains relevant: correlation patterns consistent with multiple theories require additional evidence (behavioral, neural, genetic) to adjudicate.

L. L. Thurstone's Primary Mental Abilities​

The Revolutionary Alternative​

Louis L. Thurstone (1935) offered the most influential challenge to Spearman through sophisticated factor-analytic techniques that revealed multiple independent factors rather than a single general factor.

Methodological Innovations​

Thurstone advanced factor analysis beyond Spearman:

Technical Improvements:

  1. Multiple Factor Analysis: Extracted several factors simultaneously
  2. Rotation Methods: Repositioned factors to achieve "simple structure"
  3. Large-Scale Studies: Used 57 tests on substantial samples
  4. Matrix Algebra: Generalized and formalized Spearman's methods

The Primary Mental Abilities (PMAs)​

Thurstone identified seven primary factors:

AbilitySymbolDescriptionExample Tests
Verbal ComprehensionVUnderstanding word meanings and relationshipsVocabulary, reading comprehension, verbal analogies
Word FluencyWRapid word generationName words starting with 'S', rhyming words, anagrams
Number FacilityNSpeed and accuracy in arithmeticAddition, multiplication, numerical reasoning
Spatial VisualizationSMental manipulation of shapesMental rotation, map reading, assembling puzzles
Associative MemoryMRote memorizationPaired associates, recall of facts, memorizing lists
Perceptual SpeedPRapid identification of similarities/differencesFinding matching symbols, proofreading, pattern recognition
ReasoningRProblem-solving and rule discoveryNumber series, analogies, logical puzzles

Why Thurstone's Results Differed from Spearman's​

Two Critical Factors:

1. Population Sampling​

SpearmanThurstone
Random population samples (especially children)University students only
Wide ability rangeRestricted ability range
Large individual differencesHomogeneous group

Consequence: In homogeneous groups, g becomes less apparent because everyone has relatively similar general ability. Specific abilities show through more clearly.

Analogy:

  • Measuring heights in general population: clear differences emerge
  • Measuring heights in professional basketball players: minimal variation, other factors (speed, shooting) differentiate

2. Test Selection​

SpearmanThurstone
Dissimilar testsGroups of very similar tests
Avoids overlapping specificsIntentionally includes similar items
Minimizes group factorsMaximizes group factors

Spearman's Warning: Very similar tests cause s factors to overlap, creating additional correlations that emerge as separate factors.

Thurstone's Strategy: Deliberately used similar test clusters to bring out group factors more clearly.

The Reconciliation: Hierarchical Structure​

After extended debate, Thurstone acknowledged that his primary abilities themselves correlated, necessitating a higher-order factor—essentially Spearman's g.

The Integrated Model:

Eysenck's (1972) Summary: Thurstone eventually recognized "a hierarchical structure of intellect, with g at the top, and the 'primary abilities' (whose intercorrelations necessitated the postulation of g) at a lower level; the actual tests used would of course be at a lower level still."

Contemporary Applications​

Educational Assessment: Many modern intelligence tests (e.g., WISC-V, WAIS-IV) provide:

  • Full-Scale IQ: Estimate of g
  • Index Scores: Estimates of Thurstone-like primary abilities
  • Subtest Scores: Specific skill assessments

Career Counseling: Primary abilities help identify:

  • Relative cognitive strengths (e.g., high spatial, moderate verbal)
  • Suitable career paths (e.g., engineering vs. law)
  • Training needs (e.g., developing weaker areas)

Evaluation of Thurstone's Theory​

Strengths:

  1. Empirical Rigor: Large-scale studies with advanced methods
  2. Practical Utility: Differential ability profiles guide education and career decisions
  3. Theoretical Refinement: Hierarchical model integrates Spearman's and Thurstone's insights
  4. Predictive Validity: Primary abilities predict domain-specific performance

Limitations:

  1. Initial Overreach: Claiming g didn't exist was contradicted by intercorrelations among PMAs
  2. Factor Instability: Different rotation methods yield somewhat different factors
  3. Number of Factors: Why seven? Other researchers found different numbers
  4. Measurement Reliability: Some PMAs (especially M and W) proved difficult to measure reliably

Modern Synthesis: Carroll's (1993) comprehensive review validated both Spearman's g and Thurstone's broad abilities, positioning them at different levels of a hierarchy.

Comparative Analysis​

Theoretical Comparison​

AspectSpearmanThorndikeThomsonThurstone
Core ConceptGeneral factor + specificsMultiple independent abilitiesSampled neural elementsPrimary mental abilities
Tests Correlate Because...They share gThey share task requirementsThey sample common elementsMeasurement artifact of range restriction
Underlying RealityMental energyTask commonalitiesNeural bondsIndependent cognitive processes
Mathematical ApproachFactor analysisQualitative analysisSampling theoryMultiple factor analysis

Strengths and Weaknesses​

Spearman:

  • ✅ Parsimony (single factor explains much variance)
  • ✅ Predictive power (g predicts broad outcomes)
  • ❌ Doesn't explain specialized talents
  • ❌ Reifies statistical construct

Thorndike:

  • ✅ Recognizes cognitive diversity
  • ✅ Practical educational relevance
  • ❌ Doesn't explain positive manifold
  • ❌ Lacks quantitative precision

Thomson:

  • ✅ Neurologically plausible
  • ✅ Methodological sophistication
  • ❌ Empirically indistinguishable from Spearman
  • ❌ Elements cannot be identified

Thurstone:

  • ✅ Comprehensive factor structure
  • ✅ Differential assessment utility
  • ❌ Initial rejection of g unsustainable
  • ❌ Factor instability across studies

The Modern Consensus​

Contemporary intelligence research has largely integrated rather than chosen among these perspectives:

The CHC Model​

Cattell-Horn-Carroll (CHC) Theory synthesizes:

  • Spearman's g (general intelligence)
  • Thurstone's broad abilities (Stratum II)
  • Specific skills (Stratum I)

This hierarchical framework underlies most modern intelligence tests and reflects the cumulative wisdom from a century of debate.

Neuroscience Contributions​

Modern brain imaging suggests all theories capture partial truths:

  • Spearman was right: Distributed brain networks show general efficiency differences
  • Thorndike was right: Different task types activate different regions
  • Thomson was right: Intelligence involves distributed, overlapping networks
  • Thurstone was right: Broad domains (verbal, spatial, etc.) have distinct neural substrates

Reference: Haier (2017). The Neuroscience of Intelligence

Real-World Implications​

Educational Practice​

Lesson: Both general cognitive development AND domain-specific skills matter

Applications:

  • General training: Critical thinking, problem-solving, metacognitive strategies
  • Specific training: Reading fluency, mathematical procedures, spatial skills
  • Individual differences: Tailor instruction to cognitive profiles

Clinical Assessment​

Integrated Approach:

  • Full-scale scores (g estimate) indicate overall level
  • Index scores (broad abilities) reveal cognitive profile
  • Subtest patterns guide intervention planning

Talent Development​

Recognizing Multiple Pathways:

  • High g with balanced abilities: Academic careers
  • High g with specific strength: Domain specialists (mathematicians, linguists)
  • High specific ability with moderate g: Skilled trades, arts, athletics

Memory Aids​

Critic Mnemonic: "Three T's Challenge Spearman"​

  • Thorndike: Tasks not traits
  • Thomson: Sampling not single factor
  • Thurstone: Primary abilities not general ability

Thurstone's Seven PMAs: "VW NSMPR"​

  • Verbal comprehension
  • Word fluency
  • Number facility
  • Spatial visualization
  • Memory
  • Perceptual speed
  • Reasoning

Resolution Mnemonic: "HIERARCHY"​

Higher order: General factor (g) Intermediate level: Broad abilities (PMAs) Each factor: Has specific manifestations Research confirms: Both levels exist All critics: Contributed insights Reconciliation: Through hierarchical model Contemporary consensus: Integrated framework Historical debate: Enriched understanding Yields today: Sophisticated assessments

Self-Assessment Questions​

Conceptual Understanding​

  1. Explain how Thorndike's concept of "altitude" and "range" differs from Spearman's concept of general intelligence.

  2. Thomson's sampling theory produces the same correlation patterns as Spearman's g-theory. Why does this matter for interpreting factor analysis results?

  3. Why did Thurstone initially find seven primary factors while Spearman found one general factor? Explain the role of population sampling and test selection.

Application​

  1. A student scores very high on verbal and spatial tests but average on numerical tests. How would each theorist (Spearman, Thorndike, Thurstone) explain this pattern?

  2. Design a study that could distinguish between Thomson's sampling theory and Spearman's g-theory. What kind of evidence would support one theory over the other?

Critical Thinking​

  1. Guilford stated that Thorndike's types are "essentially the same" as Spearman's group factors. If true, were Thorndike and Spearman really disagreeing? Analyze this claim.

  2. Thurstone eventually acknowledged g's existence when his primary abilities correlated. Does this mean Spearman "won" the debate? Defend your position.

Synthesis​

  1. Modern neuroscience shows both: (a) general network efficiency differences, and (b) specialized regional activations for different cognitive domains. Which historical theory does this support, or does it require a new framework?

Exam-Style Long Answer Questions​

  1. "Thorndike's multifactorial theory and Spearman's two-factor theory represent fundamentally different conceptions of intelligence." Critically evaluate this statement. (20 marks)

  2. Explain Thomson's sampling theory and evaluate whether his critique successfully undermines Spearman's concept of general intelligence. (15 marks)

  3. Trace the evolution from Thurstone's initial rejection of g to his eventual acceptance of a hierarchical model. What implications does this have for understanding intelligence? (20 marks)

  4. "The historical debate among Spearman, Thorndike, Thomson, and Thurstone was resolved through synthesis rather than victory." Assess this claim with reference to contemporary intelligence theory. (25 marks)

Source PDFs:

Further Reading​

Essential Resources​

Contemporary Integration​

  • McGrew, K. S. (2009). CHC theory and the human cognitive abilities project. Intelligence, 37(1), 1-10
  • Carroll, J. B. (1993). Human Cognitive Abilities: A Survey of Factor-Analytic Studies. Cambridge University Press

Historical Analysis​

  • Jensen, A. R. (1998). The g Factor: The Science of Mental Ability. Praeger Publishers
  • Deary, I. J. (2001). Intelligence: A Very Short Introduction. Oxford University Press

Video Resources​