Unstable resistance training: read each cognitive test
In short
In 162 older adults with probable MCI, 24 weeks of unstable resistance training beat a low-intensity control on letter-A fluency (+1.2) and forward digit span (+0.7). Other cognitive outcomes did not differ significantly between groups.
After 24 weeks, letter-A verbal fluency and forward digit span favored unstable resistance training. Those are specific cognitive test results. They cannot stand in for every aspect of memory, day-to-day thinking, or prevention of dementia. Read the outcome name before deciding what the exercise achieved.
Who trained, and what was the comparison?
The assessor-blinded randomized trial enrolled 162 physically independent adults aged 65 or older with probable mild cognitive impairment, or MCI. Mean age was 70.4 years and mean MoCA score was 20.1. This population is different from healthy recreational lifters seeking a stronger squat or an extra cognitive edge.
| Condition | Exercise over 24 weeks |
|---|---|
| Unstable resistance training, 80 people | Twice weekly; seven exercises on surfaces such as balance pads and Bosu balls; three sets targeting 10–15 repetitions |
| Active control, 82 people | Low-intensity functional exercise, stretching and relaxation without progression in intensity or complexity |
What do the score differences establish?
At week 24, the between-group advantage was +1.2 on letter-A fluency, with a 95% confidence interval of 0.2–2.2, and +0.7 points on forward digit span, with a 95% confidence interval of 0.1–1.3. The remaining cognitive outcomes showed no significant between-group differences. That leaves a narrower finding than a general claim that brain function improved. It also does not establish that every other outcome has exactly zero effect.
The control program did not progress its intensity or complexity. Consequently, this comparison cannot isolate instability itself from the other differences between the programs. It does not tell us whether unstable resistance training beats stable resistance training matched for progression and training demands. Replacing a stable heavy lift on that basis would answer a question the trial never tested.
How should a lifter separate these results in a log?
- Name the goal: strength, a balance task, or a cognitive assessment.
- Record the surface, equipment and repetitions so that loads performed under different conditions stay distinguishable.
- If cognitive testing is part of professional care, record the test and assessment date separately rather than interpreting a Big 3 personal record as a cognitive result.
The fairlift Muscle Index calculator compares Big 3 strength. A higher score does not measure cognitive change. Our discussion of where unstable-surface training fits addresses the separate question of strength and functional training outcomes. A training log becomes more useful when each result remains attached to the task that produced it.
This trial is not a treatment prescription or a reason to put heavy barbell lifts on unstable surfaces. Exercise choices for a person with suspected cognitive impairment require individual assessment by the relevant clinical and exercise professionals.
Frequently asked questions
Did unstable resistance training improve cognition overall?
Two cognitive measures favored the intervention at 24 weeks. The other cognitive outcomes did not differ significantly between groups, so the trial does not establish broad cognitive improvement or dementia prevention.
Was it better than strength training on a stable surface?
The control used low-intensity functional, stretching and relaxation exercise without progression. This does not establish superiority over stable resistance training matched for intensity and progression.
Should healthy lifters make barbell training unstable?
The participants were adults aged 65 or older with probable mild cognitive impairment. The results do not test cognitive benefits in healthy lifters or the value of unstable heavy barbell exercise.
Source: PubMed — The Journal of Nutrition, Health & Aging (published )