Today’s Older Adults Are Cognitively Fitter Than Older Adults Were 20 Years Ago, but When and How They Decline Is No Different Than in the Past

Today’s Older Adults Are Cognitively Fitter Than Older Adults Were 20 Years Ago, but When and How They Decline Is No Different Than in the Past. Denis Gerstorf et al. Psychological Science, October 25, 2022. https://doi.org/10.1177/09567976221118541

Abstract: History-graded increases in older adults’ levels of cognitive performance are well documented, but little is known about historical shifts in within-person change: cognitive decline and onset of decline. We combined harmonized perceptual-motor speed data from independent samples recruited in 1990 and 2010 to obtain 2,008 age-matched longitudinal observations (M = 78 years, 50% women) from 228 participants in the Berlin Aging Study (BASE) and 583 participants in the Berlin Aging Study II (BASE-II). We used nonlinear growth models that orthogonalized within- and between-person age effects and controlled for retest effects. At age 78, the later-born BASE-II cohort substantially outperformed the earlier-born BASE cohort (d = 1.20; 25 years of age difference). Age trajectories, however, were parallel, and there was no evidence of cohort differences in the amount or rate of decline and the onset of decline. Cognitive functioning has shifted to higher levels, but cognitive decline in old age appears to proceed similarly as it did two decades ago.

Discussion

Our findings indicate that later-born older Berliners tested in the 2010s outperformed their earlier-born age peers tested in the 1990s. Contrary to our hypotheses, results showed that later-born older adults did not exhibit shallower declines on perceptual-motor speed or a later onset of decline. The later born cohort’s cognitive performance was shifted upward from the earlier-born cohort’s, but trajectories of cognitive aging were parallel.

Historical change in cognitive performance

Consistent with the Flynn effect (Pietschnig & Voracek, 2015), results from our carefully matched longitudinal data obtained from same-aged older adults tested two decades apart provide more evidence of historical change in levels of performance. The effect size (d = 1.20) is striking and even larger than that obtained in our earlier time-lagged cross-sectional analysis of a subset of participants (Gerstorf et al., 2015: d = 0.85). This constitutes one more set of evidence that cultural changes over the last 30 years, including better access to individual resources (e.g., quantity and quality of education) and innovations in science and technology (e.g., advances in medicine and nutrition; Drewelies et al., 2019), have contributed to improved cognitive performance in old age. Future work should detail how the many different mechanisms that drive improvements in unique and specific resource constellations can further improve cognitive functioning (and productivity) of older adults.

Are old-age cognitive declines today shallower or postponed to later ages?

Our results parallel those of studies that did not find history-graded improvements in cognitive aging trajectories (e.g., Brailean et al., 2018) but differ from studies that had found such improvements (e.g., Dodge et al., 2014). Beyond similarities in the calendar years participants were born and tested, the discrepant findings may result from country-level differences in health care and differences in the studies’ measurement and analysis procedures.

In the 1990s, studies had documented that elevated blood pressure in midlife (rather than old age) is predictive of steeper cognitive declines in old age (Launer et al., 1995). Since then, widespread prescription and use of effective anti-hypertensive medication may have weakened those links. However, implementation in Germany occurred about a decade later than in the United States and other nations (Wolf-Maier et al., 2003). Consequently, our later-born older Berliners may have already been too old to have benefitted from widespread changes in delivery of health care. Back when this generation of older adults was in midlife, blood pressure treatment had not yet improved (Koenig et al., 2018). Going forward, cross-national studies can be used to test hypotheses about long-latency treatment effects of midlife blood pressure for cognitive decline in old age.

Interestingly, studies that reported cohort differences in rates of cognitive decline either did not include perceptual speed measures (Dodge et al., 2017 and Gerstorf et al., 2011: reasoning, verbal meaning, and memory), did not find cohort effects on perceptual speed measures (but on verbal fluency and working memory; Grasset et al., 2018), or found that cohort differences in perceptual speed measures were smaller than for other measures (executive functions; Dodge et al., 2014). Although measures of perceptual speed capture age-related declines well, they may not be very sensitive to history-graded changes in decline. More systematic charting of how cohort differences manifest across a wider set of aging-sensitive (e.g., memory) and more aging-resilient (e.g., crystallized) abilities is needed.

Our analytic approach also differed from approaches used in other studies. The nonlinear growth-modeling framework allowed us to account for a variety of potential confounds. First, the observation-level age matching between BASE and BASE-II samples drawn from the same underlying population provided a strong foundation for testing differences between same-age observations obtained from different cohorts. Second, we modeled and accounted for retest effects that often emerge with repeated test taking. Third, our model explicitly separated between-person from within-person age effects (age gradients vs. intraindividual change), allowing for more precise testing of hypotheses about history-graded shifts in cognitive aging—a distinctly intraindividual process.

To our knowledge, this is the first study to directly test cohort differences in the age of onset of cognitive decline. Contrary to the cognitive-reserve hypothesis, results showed no evidence for a shift in the onset of decline. However, this finding is consistent with both the preserved-differentiation perspective (Salthouse, 2006), by which level differences established in early life are maintained and carried forward into old age, and recent meta-analyses showing that differences in education have substantial effects on levels of cognitive functioning but null effects on rates of cognitive aging (Lövdén et al., 2020) or brain aging (Nyberg et al., 2021). It seems that history-graded improvements resulting from early-life education, cognitive stimulation, and health care persist into old age, but not because aging processes have been any kinder.

Limitations and future directions

Several limitations in our design and sample must also be noted. A time window of two decades may suffice to identify historical change in levels of perceptual-motor speed but may not be long enough to identify historical change in key features of cognitive aging trajectories. Further, because our assessments were obtained only in old age, we were unable to disentangle late-life processes from those unfolding during early life and mid-life. With the Flynn effect reversing among young men (Bratsberg & Rogeberg, 2018), future research should systematically examine how history-graded changes may proceed differentially throughout life.

Participants were drawn from one geographical region and represent a positively selected population segment. One key question is whether our findings apply to resource-poor population segments. Conceptual perspectives on manufactured survival (Olshansky & Carnes, 2019) suggest that some older adults today carry disease burdens longer than did older adults in the past. Future research should carefully examine whether cohort differences in decline emerge in more diverse samples and are moderated by access to resources.