For the majority of the 20th century, a significant trend in psychology was the consistent rise of intelligence test scores. Researcher James Flynn from New Zealand documented this phenomenon across various nations: on identical standardized assessments, each generation exceeded the performance of its predecessor by about three IQ points per decade, a trend that came to be identified as the Flynn effect. However, in certain countries where this increase had been meticulously recorded, the trend ceased and, in some cases, even reversed.
The most compelling proof of this reversal emerged from a 2018 study published in the *Proceedings of the National Academy of Sciences*, spearheaded by Bernt Bratsberg and Ole Rogeberg from the Ragnar Frisch Centre for Economic Research in Oslo, and subsequently elaborated upon by the science platform The Science Breaker. Norway conscripts nearly all young males, administering a standardized intelligence assessment upon entry, which provides the nation with a remarkably comprehensive historical record of scores extending over several decades. Bratsberg and Rogeberg analyzed 730,000 of these test outcomes, involving men born between 1962 and 1991 and tested between 1970 and 2009.
It is crucial to clarify what the test actually assesses, as the term “intelligence test” might imply more than what the data substantiates. The Norwegian conscription assessment is a mixture of three components: arithmetic, word similarities, and figure completion; it is intended to gauge general reasoning capabilities rather than any isolated skill. It is not a diagnostic tool, and the study does not purport to assess anything beyond the performance on that particular composite, administered to young men at a specific age and under particular testing conditions throughout five decades.
A pivotal moment arose in 1975, leading to a gradual decline.
Scores increased consistently for cohorts born until approximately 1975, in line with the Flynn effect observed elsewhere. After that birth year, the trend shifted. Males born subsequently scored lower, on average, than those born immediately prior, a decrease that the researchers estimated at about seven points per generation. This study alone does not establish conclusive consensus; however, it is among the largest and most rigorously designed in this domain, with its key finding—a rise followed by a decline within the same national testing system—having been similarly reported in Britain, Denmark, Finland, and France, each utilizing its own compulsory military testing records rather than a shared dataset.
What renders the Norwegian study particularly insightful is not just its scale but also a specific methodological approach addressing a particular alternative explanation. One theory regarding declining intelligence, occasionally referred to as a dysgenic hypothesis, posits that if individuals with lower test scores tend to have more children than those with higher scores, average scores could decline across generations for genetic reasons. Bratsberg and Rogeberg directly tested this notion by examining brothers within the same family instead of comparing the overall population. A genetic, family-level effect of this nature would be reflected as a decline between families rather than within them, as brothers predominantly share the same genetic and familial background.
The data, however, did not support this scenario.
The decline was evident equally within families as it was across the population, which the authors suggest eliminates a genetic explanation and instead points to environmental changes that occurred for everyone born after the mid-1970s, irrespective of their familial background.
A different nation paints a more complex picture.
A separate, more current dataset introduces a valuable complication rather than a straightforward confirmation. In 2023, Elizabeth Dworak, a research assistant professor at Northwestern University’s Feinberg School of Medicine, published an analysis in the journal *Intelligence* utilizing data from the SAPA Project, an online cognitive assessment tool. Her team examined 394,378 individuals who took the test between 2006 and 2018 in the United States. During this period, scores declined in verbal reasoning, matrix reasoning, and letter and number series tasks. However, scores for a fourth metric, three-dimensional spatial rotation, actually increased over the same years.
This aspect is significant because it challenges a single coherent narrative of general decline. Dworak was cautious not to assert a cause. Among the hypotheses she proposed were shifts in educational and workplace focus, increased engagement with visually intricate media that might enhance spatial skills specifically, and a broader possibility: that some contemporary test-takers may simply be less accustomed to or less driven when it comes to the type of assessments that older tests were designed to evaluate, a factor that would manifest as a score change without indicating a genuine alteration in underlying capability.
The two datasets do not accurately measure the same phenomenon, and the disparity between them deserves thoughtful consideration rather than simplification. The Norwegian sample is nearly representative of the entire population: almost every eligible young male in the nation, assessed under uniform conditions by the same authority, from year to year. The American sample, in contrast, is self-selected: individuals who discovered the SAPA Project online and