What question did this study set out to answer?

To examine the influence of processing speed on WAIS-IV index scores and the implications for understanding intelligence in relation to AI.

March 21, 2026Open Access

Processing Speed Contamination in WAIS-IV and the AI Mirror: Why IQ Tests Measure the Cognitive Dimension Being Automated

Key Points

To examine the influence of processing speed on WAIS-IV index scores and the implications for understanding intelligence in relation to AI.
Conducted bifactor analyses to evaluate the psychometric structure of WAIS-IV scores.
Analyzed the contamination of processing speed across various index scores.
Explored the implications of human-cognition frameworks in relation to AI systems.
The general factor (g) accounts for 67-69% of common variance in WAIS-IV scores.
Processing speed significantly contaminates the Processing Speed Index and affects Perceptual Reasoning and Working Memory.
73.5% of healthy adults show notable discrepancies in index scores, highlighting issues in IQ interpretation.

Abstract

The Wechsler Adult Intelligence Scale—Fourth Edition (WAIS-IV) is the most widely usedindividual intelligence test in clinical practice. Its four index scores—Verbal Comprehension,Perceptual Reasoning, Working Memory, and Processing Speed—are routinely interpreted asreflecting four distinct cognitive abilities, and discrepancies among them guide diagnosticdecisions for millions of individuals annually. This paper argues that this interpretive practice islargely unsupported by the instrument’s own psychometric structure. Bifactor analysesconsistently show that a general factor (g) accounts for 67–69% of common variance, whileindividual group factors contribute unique reliable variance (ωs) as low as .08—belowthresholds for clinical interpretation. Processing speed contaminates not only the ProcessingSpeed Index but also Perceptual Reasoning (through time bonuses and limits) and WorkingMemory (through timed subtests and rehearsal-speed dependence), creating a speed gradientthat produces the commonly observed VCI/PRI > WMI/PSI profile pattern as a structuralartifact: 73.5% of healthy adults show significant index score discrepancies. Morefundamentally, drawing on the Permanent Boundary framework for human-AI cognition(Sophia, 2025a), the paper demonstrates that every WAIS-IV subtest assesses operations withinexisting descriptive frameworks—pattern matching, rule application, knowledge retrieval,procedural execution—while systematically excluding the cognitive operations that remainstructurally unavailable to language-based AI: reframing (questioning the problem frame itself),grounded verification (checking descriptions against non-linguistic reality), and open semanticgeneration (creating new categories from encounters with novel phenomena). This convergenceis not coincidental: both the test and the AI are confined to the interior of descriptive systems.The paper concludes that what IQ tests call “intelligence” is substantially the cognitivedimension being automated, while the distinctively human cognitive contribution—theirreplaceable part—remains unmeasured.

Processing Speed Contamination in WAIS-IV and the AI Mirror: Why IQ Tests Measure the Cognitive Dimension Being Automated

Key Points

Abstract

Cite This Study

Also Consider

Also Consider