For the better part of a century, IQ scores did one thing reliably: they went up. The pattern, the Flynn effect, held from 1909 to 2013 across dozens of countries, and a 2015 meta-analysis published in Perspectives on Psychological Science, spanning 271 samples, 31 countries and close to 4 million test-takers, put the scale of the climb beyond doubt. Now, in several wealthy nations, the line has bent the other way. The mechanics here are measurable. The cause is not.
Start with the strongest data set, because it is the one that closes the usual escape hatches. Across the years from 1962 to 1991, more than 700,000 men in Norway sat the same IQ test ahead of military service. Scores rose steadily among men born up to about 1975, then reversed, drifting down roughly 0.2 points a year. Bernt Bratsberg and Ole Rogeberg, two economists, showed in a 2018 paper in PNAS that the entire arc — rise, peak, decline — can be recovered by comparing brothers within the same families. Follow the data to its conclusion: siblings born a few years apart show the same slide, which rules out genes and rules out immigration as the driver. Whatever moved the scores is environmental.
Neighbouring conscript records tell a similar story on their own clocks. In Denmark, Thomas Teasdale and David Owen examined test results from more than 500,000 men covering 1959 to 2004, finding performance peaked in the late 1990s before slipping — nutrition ruled out, with the authors pointing instead to how schooling stressed abstract reasoning. Finland, in work by Edward Dutton and Richard Lynn published in the journal Intelligence, gained about 4.0 points a decade between 1988 and 1997, then fell around 2.0 points a decade through 2009, on roughly 25,000 conscripts. The United States is messier: a 2023 study led by Elizabeth Dworak, William Revelle and David Condon worked through survey data on 394,378 Americans collected from 2006 to 2018, and found verbal and matrix reasoning scores falling while spatial reasoning rose. Dworak of Northwestern’s Feinberg School of Medicine put it in a statement: “It doesn’t mean their mental ability is lower or higher; it’s just a difference in scores.”
So what is taking the points back? On closer reading, the honest answer is that almost every tidy explanation fails somewhere. Education change is a live candidate, and the same meta-analysis found gains shrinking in recent decades. A decline in reading has real support: work by the San Diego State University psychologist Jean Twenge tracked 12th graders from 1976 through 2016 and found the share reporting no pleasure reading nearly tripling, to one in three. The catch is the popular attention-span story: a 2024 meta-analysis, published in Personality and Individual Differences and led by Andrzejewski with Zeilinger and Pietschnig, actually logged moderate gains in adult concentration, not collapse. Blame the smartphone in the abstract and the evidence will not co-sign.
What survives scrutiny is narrow but solid: the rise happened, the reversal is real in specific countries, and the numbers themselves are not in dispute — only the why. James Flynn, the University of Otago political scientist who gave the effect its name and died in 2020, never claimed people had grown innately smarter; his 1987 paper argued the tests capture something whose causal link to intelligence is weak. That caution cuts both ways now. If the environment built the gains, it is presumably dismantling them, and a shift this fast across a generation or two cannot be genetic. The scoreboard changed; the reason is still out.
