The stability of cognitive abilities over time has always been treated, in differential psychology, as an essential assumption for the validity of intelligence tests. However, the meta-analysis conducted by Breit, Scherrer, Tucker-Drob and Preckel (2024), by systematizing 1,288 test-retest correlations in 205 longitudinal studies with more than 87 thousand participants, offers a turning point: despite the high overall stability (ρ = .76 for young adults at five-year intervals), there are critical discontinuities in early childhood and practical limits that cannot be ignored in the clinic, at school or in the justice system.
The conception that IQ is stable and predictable is only supported under specific conditions. At ages below seven years, or with intervals greater than five years, the correlation does not reach the minimum cutoff point for individual diagnostic decisions (rtt = .80). This conclusion undermines the automated use of tests in high-consequence contexts, such as early screening for high abilities, court decisions in custody disputes, or the application of educational strategies based on supposedly “fixed potentials.” From a technical perspective, cognitive stability requires not only the repetition of the same test, but also the constancy of the neurobiological substrate, environmental conditions, and psychosocial context—which rarely holds true over time.
The most critical implication of the meta-analysis lies in diagnostic error. By treating IQ as a stable and reified measure, experts may overlook significant fluctuations mediated by factors such as prefrontal cortex maturation, synaptic plasticity in response to the environment, and stressful events that affect executive functions and processing speed. This point is supported by the literature on the neuroarchitecture of intelligence DWRI (Development of Wide Regions of Intellectual Interference), as described in Rodrigues (2023), which emphasizes cross-talk between brain regions involved in social, emotional, and abstract cognition. Conventional tests do not capture these dynamic variables, which creates an illusion of stability.
Therefore, it is necessary to abandon the notion that intelligence is a static data. The meta-analysis by Breit et al. does not invalidate the use of tests, but requires that their interpretation be anchored in complex models that integrate neuroscience, developmental context, and psychosocial factors. Without this, we run the risk of taking statistical correlations as biographical truths, converting average probabilities into individual sentences.
ABNT Reference:
BREIT, Moritz; SCHERRER, Vsevolod; TUCKER-DROB, Elliot M.; PRECKEL, Franzis. The stability of cognitive abilities: A meta-analytic review of longitudinal studies. Psychological Bulletin. Manuscript version. 2024. DOI: 10.1037/bul0000425.
