Exercise and the Brain: Hippocampal Growth vs Preservation

Most “brain boosters” come with a mechanism, a mouse study and a bottle. Exercise comes with something rarer: a randomised trial in which the structure that handles memory measurably changed size, plus a decade of meta-analyses arguing about how much. The short version: the effect is real, clearest in older adults who start out less fit, and smaller than the headlines.

What BDNF is and how much exercise raises it

Brain-derived neurotrophic factor is a signalling protein that neurons rely on to grow and survive; the Erickson trial below calls it a mediator of neurogenesis, the birth of new neurons, in the hippocampus’s dentate gyrus. Animal models consistently show hippocampal growth with aerobic exercise (Firth 2018); the human question is whether exercise moves BDNF enough to matter.

Human studies measure BDNF in blood, a proxy for what happens inside the skull rather than a direct reading. With that caveat, two meta-analyses put a size on the effect:

• Szuhany 2015 pooled 29 studies (1,111 participants). A single exercise session raised BDNF with a moderate effect (Hedges’ g = 0.46). Regular training intensified the response to a single session (g = 0.59). Resting BDNF after a programme of regular exercise rose only slightly (g = 0.27). The more women a study included, the smaller the BDNF change it found.
• Dinoff 2016 pooled 29 studies of exercise programmes lasting at least two weeks. Resting BDNF was higher after training (standardised mean difference 0.39), but the subgroup analysis put the effect in aerobic programmes (SMD 0.66); resistance training produced no significant change (SMD 0.07). It found no male-female or serum-plasma difference.

The BDNF response is reliable but modest: a clear bump after each session, a small upward drift in baseline with aerobic training, and little from lifting alone. Whether blood BDNF mirrors hippocampal BDNF is not something these studies can measure.

The trial that grew the hippocampus

The anchor study is Erickson 2011, published in PNAS. It randomised 120 older adults to a year of aerobic walking or a non-aerobic control programme, with hippocampal volume measured before and after.

The walkers’ anterior hippocampus grew by about 2 percent. The control group’s shrank over the same year, as the hippocampus normally does in late adulthood. The authors described the difference as reversing one to two years of age-related volume loss. The gain was specific: caudate nucleus and thalamus volumes did not change. Spatial memory improved, and larger hippocampal volume went together with higher serum BDNF. Higher fitness at entry also partially protected the control group from losing volume.

Two points of honesty. First, a 2 percent change is real but small, and it took a year of sessions to get it. Second, a companion paper from the same one-year walking programme (Voss 2013, 65 participants, mean age 66.4) found no group-level rise in BDNF, IGF-1 or VEGF from the intervention; walkers whose growth factors rose more simply showed larger gains in temporal-lobe connectivity, a link absent in the control group. The BDNF story here is a correlation within the trial, not a group effect.

What the meta-analyses say about volume

Pooled trials since have found something weaker than Erickson’s 2 percent. Firth 2018 combined 14 controlled trials (737 participants) and found no significant change in total hippocampal volume with aerobic exercise. A significant effect on the left hippocampus came, on post-hoc analysis, from exercise heading off the shrinkage seen in control groups rather than adding tissue — the authors’ phrase was “volumetric retention”.

Wilckens 2021, a larger analysis of 23 interventions from 22 studies, did find total hippocampal volume significantly higher with exercise, but with the same texture: the effect was significant in samples aged 65 or older, in programmes lasting longer than 24 weeks, and in programmes with less than 150 minutes of exercise per week — moderate amounts sustained for more than six months, in populations vulnerable to hippocampal atrophy.

Put together, the fair summary is that exercise preserves hippocampal volume against age-related loss, and may add a little in older adults if kept up for six months or more. “Grows your brain” overstates it; “slows the shrinkage” is what the pooled data support.

Memory: one session versus months of training

Structure is one thing; memory is what people care about, and here there is a twist.

Winter 2007 put 27 healthy volunteers through a randomised crossover: short all-out sprints, a gentle run, or rest, each followed by learning a novel vocabulary. The sprints sped vocabulary learning by 20 percent relative to either other condition. They also produced the strongest rise in BDNF and catecholamines; volunteers whose BDNF stayed elevated through learning did better in the short term, while longer-term retention tracked dopamine and epinephrine.

Roig 2013 found that this acute effect is where the memory benefit is largest. A single bout of aerobic exercise had a moderate-to-large pooled effect on long-term memory (SMD 0.52) and a moderate effect on short-term memory (SMD 0.26). Long-term training programmes, by contrast, had a small effect on short-term memory (SMD 0.15) and no significant effect on long-term memory (SMD 0.07). The authors’ reading is that a single session readies the molecular machinery that encodes and consolidates new information, while long-term training tunes that machinery without a measurable memory gain of its own.

Practically, the memory effect that is easiest to demonstrate is timing exercise close to learning. The caveats: Winter’s was a 27-person study, and Roig’s pooled estimate for long-term memory came with a wide confidence interval (0.28 to 0.75).

Which exercise works after 50

For cognition more broadly — attention, processing speed, executive function — the strongest evidence is in adults over 50, and it is not restricted to aerobic work.

Northey 2018 pooled 36 randomised trials in adults over 50 living in the community and found a pooled effect of 0.29 (95% CI 0.17 to 0.41) on cognitive function. Aerobic exercise, resistance training, multicomponent programmes and tai chi each showed a significant effect on their own. Sessions of 45 to 60 minutes at moderate intensity or higher were associated with benefit, regardless of cognitive domain or participants’ cognitive status.

Resistance training deserves its own line because it is the one exercise type that does not raise resting BDNF (Dinoff 2016) yet still improves cognition. Landrigan 2020 pooled 24 studies of resistance exercise and found gains on composite cognition scores (SMD 0.71) and executive function (SMD 0.39), but not working memory (SMD 0.15, not significant), with high heterogeneity across studies. A Bayesian network meta-analysis of 44 trials (4,793 participants) in adults aged 50 and over (Gallardo-Gómez 2022) found resistance exercise outperformed other modalities, no minimum threshold below which exercise did nothing, and clinically relevant change at about 724 MET-minutes a week — a lower dose than WHO guidelines, the authors note — with less clear gains above 1,200.

In healthy individuals more broadly the picture is more modest. Ludyga 2020 ran a meta-regression of 80 RCTs and found a small overall effect that did not differ by cognitive domain. Coordinative exercise did better than other types, the gain from longer programmes grew when sessions were also longer, and the effect was weaker in women, echoing the BDNF finding from Szuhany 2015.

Why older and less-fit adults show the clearest effect

Nearly every strong result above comes from older adults, and there is a plain reason: the hippocampus loses volume in late adulthood, and the trials measure exercise against that decline. In a younger adult there is far less loss for exercise to offset. Erickson’s control group lost volume; the walkers did not. Firth’s pooled effect came from preventing decreases over time. Wilckens found the effect significant in samples aged 65 and over. The benefit is largest where the baseline trajectory is worst.

The same logic applies to fitness: higher starting fitness partially protected Erickson’s control group on its own, so those starting least fit have the most to gain, and someone already training regularly may have banked much of it. The hippocampus is also central to chronic stress and the brain, a topic that overlaps with this one.

Next to the supplement aisle the contrast is stark. Creatine’s cognitive effects show up mainly under sleep loss and in vegetarians, and lion’s mane has NGF data from cell cultures and a handful of small human trials. Exercise has a 120-person trial with the hippocampus measured, pooled RCTs in thousands of adults over 50, and an acute learning effect seen in a crossover trial and a meta-analysis. The trade is 45 to 60 minutes a session, on as many days as feasible, for months, for effects that are small to moderate rather than transformative.

Bottom line

Exercise raises blood BDNF reliably after each session and slightly at rest with aerobic training. One year-long RCT of walking increased anterior hippocampal volume by about 2 percent in older adults; pooled trials show the more dependable effect is preserving volume against age-related shrinkage, mainly in people aged 65 and over who keep it up for more than six months. Cognition improves modestly across aerobic, resistance, multicomponent and tai chi programmes after 50, with 45 to 60 minute sessions at moderate intensity or higher, and resistance training came out ahead of other modalities in one network meta-analysis. For memory specifically, one bout of intense exercise close to learning has better evidence than long-term training. Anyone with a heart condition, a recent injury or a long sedentary stretch behind them should check with a clinician before taking up intense exercise.