Study: Cognitive training, not mild aerobic exercise, drove cognitive gains in healthy older adults

What the study tested

The trial, conducted by Evelyn Shatil and published in Frontiers in Aging Neuroscience in 2013, randomized healthy residents of the Lakeview retirement community in Lenexa, Kansas, into four groups. One group (n = 33, mean age 80, range 66–90) did computerized cognitive training. A second (n = 29, mean age 81, range 65–89) did mild aerobic exercise. A third (n = 29, mean age 79, range 70–93) did both. A fourth (n = 31, mean age 79, range 71–92) served as an active control, meeting weekly as a book club. To enter, participants needed a Mini-Mental State Examination score of at least 24 and no stroke or heart attack in the previous 5 years. 59% of all participants were 80 or older (range 65–93).^[1]

The cognitive training was the CogniFit program: 32 hours delivered as 48 forty-minute sessions, three times a week for 16 weeks, personalized to each participant. The aerobic program (a senior exercise video) ran as three 45-minute sessions a week for 16 weeks, deliberately kept mild given the participants' age. The combined group did both, attending twice as many sessions. Of 180 randomized participants, 122 (67.8%) completed the study; only 3 dropped out after training had actually begun, all for health reasons.^[1]

The outcome measure was the CogniFit neuropsychological evaluation, three 15-minute sessions, 15 tasks scored on response time and accuracy, reduced to 17 cognitive abilities, of which 13 were analyzed for group differences. The instrument was validated on a healthy reference sample of 861 people (517 women, 344 men; mean age 65.7), and the paper reports internal consistency of 0.70 and test–retest reliability of 0.80.

What it found

When the two cognitive-training groups (cognitive alone plus combined) were compared with the two non-cognitive groups (aerobic alone plus control), the cognitive-training participants improved significantly on five cognitive abilities: Hand-Eye Coordination, Naming, Speed of Visual-Spatial Information Processing, Visual Scanning, and Global Visual Memory (which combines working memory and long-term memory). The effect sizes were medium to large, Hand-Eye Coordination d = 0.80 (F = 57.60, p < 0.001) and Naming d = 0.82 (F = 8.12, p = 0.005) were large, while Speed of Processing d = 0.71, Visual Scanning d = 0.77 (p = 0.021), and Global Visual Memory d = 0.64 (p = 0.003) were medium. The differences held after education and health were added as covariates.^[1]

Within the cognitive-training group alone, improvements from baseline reached significance on seven abilities even under a strict Bonferroni correction (alpha of 0.05 divided across 13 tests): Divided Attention (p = 0.001), Avoiding Distractions (p = 0.001), Hand-Eye Coordination (t = −10.84), Naming, Speed of Visual-Spatial Processing, Visual Scanning (p = 0.002), and Global Visual Memory. The combined group improved on three abilities (Hand-Eye Coordination, Naming, and Speed of Processing), fewer, not more, than cognitive training alone.^[1]

Crucially, neither the mild-aerobic group nor the book-club control improved on any of the 13 abilities, and the three-way TIME × cognitive × physical interaction was not significant for any measure. In other words, combining exercise with brain training did not beat brain training by itself.

Why exercise alone showed no benefit

The null result for aerobic exercise was the surprise, since it runs against a broad consensus that aerobic fitness supports cognition. The paper offers two reasons. First, the aerobic program was purposely mild because more than half the sample was over 80, so it may not have been vigorous enough to move cognitive scores. Second, the intervention lasted only 4 months; the paper points to evidence that aerobic training may need to be sustained for at least a year before cognitive benefits emerge. The reading is not that exercise is useless for the aging brain, the paper stresses the broader health necessity of physical activity, but that, at this intensity and duration, it produced no measurable cognitive gains while cognitive training did.

Limitations: what this does not show

The paper states its own limits plainly: “The limitations of the study include a possibly insufficient duration of aerobic training, the lack of health, functional, and quality-of-life endpoints to evaluate any associations with benefits in daily living or quality of life impact, and no post-intervention follow-up to assess the maintenance of observed improvements.” In plain terms, the study did not measure whether the gains transferred to everyday tasks, whether they lasted beyond the training period, or whether they changed quality of life.

Several other caveats matter. The combined group was somewhat less educated than the others (a difference that reached significance only against the aerobic group), so education was treated as a covariate throughout. Non-completers were on average about 4 years older and roughly 1.2 years less educated than completers, and non-completion was highest in the combined group (39.6%) and lowest in the cognitive-training group (26.7%), differences that could bias which participants remained. Finally, the author is an employee of CogniFit, which supplied the cognitive training program; this is a conflict the paper itself discloses.^[1]

Where this fits in the broader evidence

This trial replicates and extends earlier CogniFit work. A companion randomized double-blind trial by Peretz and colleagues found the same personalized program outperformed off-the-shelf computer games on visuospatial learning and working memory in healthy older adults, and the evaluation used here as the outcome measure was validated in prior studies, including work by Horowitz-Kraus and Breznitz. On the exercise side, the paper sets its mild-aerobic null against trials such as Voss and colleagues, which indicate that longer or more intense regimens may be needed before cognitive change appears. The takeaway is narrow: in a sample this old, a personalized cognitive-training program moved cognitive scores where four months of mild exercise did not. For more CogniFit research, see our research index.