MCAT content library / Psychology & sociology / Intelligence theories and measured ability
Compare ability models using correlation patterns, then keep heritability, test validity, and group-level statistics from becoming claims about an individual’s fixed potential.
About 18 minutes, including a synthetic correlation matrix and measurement checks.
On this page
- Intelligence theories make different predictions about covariation
- Concept: a test score is a measurement, not the person
- Worked application: read a correlation pattern
- Pattern: heritability is a population statistic
- Rule: group differences require measurement and sampling scrutiny
- Check your reasoning
- Continue learning
- Sources
Intelligence theories make different predictions about covariation
Some theories emphasize a general cognitive factor that contributes to performance across many tasks. Other theories divide ability into multiple broad or relatively distinct domains, such as analytical, practical, creative, linguistic, spatial, or other capacities depending on the framework.
Theories should be compared by what they predict. A strong general-factor account predicts substantial positive correlation among diverse cognitive tasks. A model with partially distinct abilities can still predict correlations, but may expect stronger relationships within domains than between them.
Finding correlation among tasks does not prove that one biological mechanism causes all performance. Correlation is a pattern that models must explain.
Concept: a test score is a measurement, not the person
A cognitive test samples performance under defined conditions. Reliability asks whether measurement is consistent; validity asks whether the interpretation is supported for the intended construct and use. A score can be useful without being a complete measure of every intellectual capacity.
Standardized scores are also relative to a norming population. Changing the test, language, testing conditions, or reference group can affect interpretation. Repeated scores can vary because of measurement error, learning, health, motivation, sleep, or genuine change.
Worked application: read a correlation pattern
An original study gives four validated tasks to the same 500-person sample. V1 and V2 are verbal tasks; S1 and S2 are spatial tasks.
| Pair | Correlation r |
|---|---|
| V1–V2 | 0.78 |
| S1–S2 | 0.74 |
| V1–S1 | 0.42 |
| V1–S2 | 0.39 |
| V2–S1 | 0.41 |
| V2–S2 | 0.40 |
All tasks are positively correlated, which is compatible with some shared influence across cognitive tasks. But within-domain correlations are substantially stronger than between-domain correlations, also supporting domain-specific structure.
The data therefore fit a hierarchical interpretation better than an extreme model in which every task is completely independent or every task measures exactly one identical ability. The table alone cannot determine whether the shared factor is genetic, educational, motivational, neurological, or some combination.
Pattern: heritability is a population statistic
Heritability estimates the proportion of observed variation in a trait within a particular population and environment that is associated with genetic variation under the statistical model used. It does not report what percentage of one person’s intelligence “comes from genes.”
A highly heritable trait can still be affected by environment. For example, if environmental conditions become more uniform, genetic differences can account for a larger share of the remaining variation even though the environment remains essential for development.
Heritability also need not be identical across populations or historical periods because the ranges of genetic and environmental variation differ.
Rule: group differences require measurement and sampling scrutiny
If two groups have different average test scores, several questions must come before causal explanation: Was the same construct measured validly in both groups? Were language and testing conditions comparable? Were samples selected similarly? Do socioeconomic, educational, health, or cultural exposures differ?
A mean difference does not imply that every member of one group scores above or below every member of another. Group distributions can overlap extensively. Nor does a group-level association establish a fixed biological cause.
The strongest claim stays at the level of the design: what was measured, in which sample, under what conditions, and what alternative explanations remain.
ACTIVE PRACTICE · EVIDENCE AND TRANSFER
Read a Correlation Pattern Without Treating Scores as Identity
The source’s 500-person study measures two verbal tasks (V1,V2) and two spatial tasks (S1,S2) in the same sample. All reported r values are positive, but within-domain associations exceed between-domain associations.
Data represent the lesson’s bounded original or synthetic teaching design. Associations and readout differences do not alone prove specific biological or psychological causation.
Check your reasoning
Choose an answer before opening its feedback.
Check 1
Several different cognitive tasks show moderate positive correlations. Which conclusion is safest?
A. Some shared source of variation may contribute across tasks.
Correct. Positive covariation is compatible with a shared factor.
B. The tasks must measure exactly the same ability.
Review this choice. Moderate correlation leaves substantial task-specific variation.
C. Genes are proven to be the only cause.
Review this choice. Correlation among tasks does not identify the source of shared variance.
D. Measurement reliability is irrelevant.
Review this choice. Reliability affects observed score relationships.
Check 2
A trait has high heritability in one population. What follows?
A. Genetic variation explains a large share of observed variation in that population under those conditions.
Correct. That is the level at which heritability is defined.
B. Each individual’s trait is mostly fixed by genes and cannot change.
Review this choice. Heritability does not imply immutability or individual causation.
C. Environment has no effect.
Review this choice. High heritability can coexist with strong environmental influence.
D. The same estimate must apply to every population.
Review this choice. Heritability is population- and environment-specific.
Check 3
Two groups differ in average test score, but the test has not been validated equivalently across their languages. What is the most important limitation?
A. Measurement comparability is uncertain.
Correct. The observed difference may partly reflect different measurement properties.
B. The group difference automatically proves biology.
Review this choice. Causal inference is not justified.
C. Sampling no longer matters.
Review this choice. Sampling remains a separate concern.
D. Averages describe every individual exactly.
Review this choice. Group distributions contain individual variation and overlap.
Continue learning
- Psychological Measurement
- Cognitive development across the lifespan
- Browse Psychology & Sociology MCAT lessons