In recent decades, the United States and several European countries have seen a plateau and reversal of the population-level gains in cognitive test scores that accumulated over the 20th century-the Flynn effect. We examined whether the Flynn effect and its reversal-declining mean scores across cohorts-differ by socioeconomic status, and whether changes in education can account for this. Using compulsory military conscription cognitive tests linked to administrative records with near-complete population coverage, we studied 579,379 Norwegian men from 25 birth cohorts (1967-1991). Trend reversals appeared first among sons of high-income fathers, with low- and middle-income groups peaking later and showing smaller postpeak declines. These socioeconomic differences were consistent with convergence in educational attainment at conscription: across paternal-income ranks, cohort changes in test scores aligned with cohort gains in education at conscription age, with estimated schooling effect sizes in line with quasi-experimental evidence. Results were robust to adjustment for parental education, and birth weight and height at testing as proxies for prenatal conditions and childhood nutrition. Together, the findings suggest an underlying population-level decline in cognitive scores that began earlier than previously assumed but was temporarily offset in lower-income strata by expanding educational attainment. After educational expansion plateaued, scores declined across all groups, most steeply among higher-income strata. Upper-secondary education completion thus emerges as a potential lever for improving cognitive performance and reducing income-related disparities in cognitive ability scores, while highlighting that broader societal pressures on cognitive test performance may have started earlier than recognized.
Keywords: Flynn effect; cognition; development; education; socioeconomic status.