I found an important paper about age related changes in brain size by scholars Susan M. Resnick, Dzung L. Pham, Michael A. Kraut, Alan B. Zonderman, and Christos Davatzikos. Unlike most studies on ageing and brain size, this study actually measured the brains of the same people over half a decade (longitudinal research) instead of just comparing the brains of different age groups at one point in time (cross-sectional research). The problem with the latter is that you can’t tell whether age differences in brain size are caused by age, or by birth-year (secular changes), but longitudinal studies like this one can prove the differences are caused by ageing. The authors write:
Remarkably, this community-dwelling sample of older adults shows an average loss of 5.4 cm3 of brain volume and increase of 1.4 cm3 in ventricular volume each year.
5.4 cm3 a year is equivalent to 54 cm3 a decade. Assuming the standard deviation (SD) for brain volume is 91 in white men and 90 in white women, as military crania data indicates, then that’s a loss of about 0.6 SD a decade.
Assuming brain size correlates 0.24 with IQ as a recent meta-analysis claimed (probably an underestimate), and the correlation is causal, then brain shrinkage alone should subtract 0.6(0.24) = 0.14 SD a decade from measured intelligence (the equivalent of 2.16 points). This is somewhat consistent with the results of a longitudinal study cited by scholar James Flynn in the book The Rising Curve. The decline in performance on fluid IQ tests from age 25 to 65 was the equivalent of about 10 IQ points.
Of course I don’t know if brain shrinkage begins as early as 25 nor do I know if the rate of loss is the same in young adulthood if it does. The Resnick paper only followed people aged 59 to 85 when they entered the study.
These huge amounts of brain shrinkage (if they generalize across the entire adult years) suggest that the recent paper by scholar Michael Woodley et al, might be correct in showing small secular gains in brain size, because if age differences in brain size can be explained by the ageing process itself, there’s no reason to assume large cohort effects.