Is there a minimum exercise dose below which cognition does not improve, or is the dose-response continuous?

The curve rises from low doses rather than switching on at a threshold

The clearest answer comes from a large synthesis of randomized trials in people aged 50 and over. Pooling 44 trials and nearly 4,800 participants, the analysis found a non-linear dose-response relationship between overall exercise and cognition, and explicitly reported no minimal threshold for a beneficial effect [5]. In other words, the curve does not sit flat and then jump upward once a certain amount of exercise is reached; it bends gradually.

That same analysis did estimate a dose at which changes became clinically relevant: about 724 METs-minutes per week, a MET-minute being a way of combining how hard an activity is with how long you do it. But this is a marker of when benefits become large enough to matter in practice, not a floor below which nothing happens. The authors also found that benefits became less clear beyond roughly 1,200 METs-minutes per week, so more was not simply better [5].

A community-based randomized trial in older adults points the same way: cognitive benefits were evident even at low exercise doses, which directly undercuts the idea of a hard minimum [1]. Together these results suggest the practical message is that some exercise beats none, and the returns are graded rather than all-or-nothing.

The threshold moves with age and baseline cognitive health

The picture is not identical across groups. In older adults with some degree of cognitive impairment, a network meta-analysis of exercise prescriptions did identify a minimum exercise dose linked to clinically significant cognitive change, estimated at 934 METs-minutes per week [3]. That figure is higher than the clinically relevant dose reported for generally healthy older adults [5], which fits the idea that people starting from a lower cognitive baseline may need more before a measurable, meaningful shift appears.

This is a threshold for a clinically significant change, not evidence that smaller amounts do nothing. The two numbers come from different populations and different analytic approaches, so they should be read as group-specific benchmarks rather than a universal cutoff.

The shape of the curve depends on the exercise and the person

Mind-body exercise in older adults with mild cognitive impairment shows a U-shaped pattern for global cognition and processing speed, according to a systematic review and meta-analysis [4]. A U-shape means too little and too much are both associated with weaker outcomes, with a favorable middle range. That is a different curve from the gradual rise seen for overall exercise in broader older-adult samples [5], and it suggests the optimal zone can depend on the modality.

Exercise type matters as well: the dose-response association in older adults was found to be exercise-type dependent, with clinically important effects possible at lower doses for many types of exercise, and resistance training standing out as particularly potent [5]. So the amount needed is not a fixed number but shifts with what kind of exercise is done.

In children, the evidence looks different again. In a within-subject experiment where 104 children completed cognitively challenging exergame sessions of 5, 10, 15, or 20 minutes at constant intensity, reaction times differed by duration, with faster processing speed after 15 minutes than after 10 [2]. Accuracy and other attention measures were unaffected, and the study did not test whether a minimum dose exists below which no benefit occurs [2]. Habitual activity also moderated the duration effect, with more active children showing the clearest benefit at 15 minutes [2]. This is a narrow finding about one session length in one age group, not a general rule about thresholds.

What this means if you are deciding how much to exercise

For a generally healthy older adult, the evidence supports starting low rather than waiting to hit a magic number: benefits appear across a continuous, non-linear range, and clinically meaningful gains cluster around 724 METs-minutes per week, with diminishing clarity above about 1,200 [5]. For someone with existing cognitive impairment, the clinically significant benchmark reported is higher, around 934 METs-minutes per week [3].

The honest summary is that the dose-response is continuous in the populations studied, with no demonstrated floor below which cognition fails to improve, but the amount needed for a meaningful change varies by baseline cognitive health, exercise type, and age group [1][3][5]. The children's data add a caution: a single session's duration can matter for processing speed without affecting other cognitive measures, so short-term and long-term dose questions are not the same [2].

Sources used in this answer

Related research questions