The Hippocampus Trials

The hippocampus — the memory-filing structure deep in each temporal lobe — was long treated as a one-way street: it shrinks with age, and nothing you do changes that. Then a year-long randomized trial put a +2% figure on the table. Here is what that trial showed, how replication has gone since, and the honest verdict on whether exercise grows a shrinking brain region.

The Structure That Shrinks

The hippocampus converts experience into durable memory — new episodes get written there before filing elsewhere. From about age 50 onward it loses volume steadily, roughly one to two percent per year in later adulthood — one of the quieter engines behind age-related forgetfulness and dementia risk. The full map of which brain systems decline belongs to How the Brain Ages; this page owns one question inside it. Observational studies had long shown fitter older adults have larger hippocampi — an association, not causation, since fitter people differ in diet, education, and decades of habits. Proving the structure responds to training took a randomized trial.

The Trial: One Year, 120 People, Two Groups

Kirk Erickson's group at Pittsburgh and Illinois recruited 120 sedentary adults aged 60–80: half randomized to a walking program built up to roughly 40 minutes of moderate walking three times a week, half to a stretching-and-toning control that met just as often and controlled the same confounds, but stayed non-aerobic. MRI, blood draws, fitness testing, and spatial memory tests ran at baseline and after the year. The aerobic group's anterior hippocampus increased in volume by about 2% — enough, the authors noted, to offset roughly one to two years of typical age-related loss in that region. The stretching controls did not hold steady: they lost about 1.4% over the same year. Volume change also tracked measurable biology — larger gains accompanied rising serum BDNF, the growth factor the BDNF page covers — and spatial memory improved more in the walkers.

Read the Controls, Not Just the Headline

The +2% is the number that traveled, but the stretching group's −1.4% does half the persuasive work: the gap between groups — a swing of more than three percentage points in a year — is the effect, and the active comparison is what makes the trial mean something. Two details discipline the reading. First, the effect was anatomically specific — caudate nucleus and thalamus volumes were unaffected; this was not "exercise swells the brain generally." Second, the BDNF and memory correlations are associations within the trial, not a demonstrated causal chain. What the trial established is narrower and more useful: in previously sedentary older adults, a year of moderate walking changed the trajectory of a memory structure toward youth.

The Replication Ledger, Honestly

One trial is one trial; the fair question is what happened when others tried. The short answer: support, with exceptions, at half the resolution of the headline. Colcombe and colleagues had already shown, in a 6-month trial of 59 adults aged 60–79, that aerobic training increased gray- and white-matter volume in frontal and temporal cortex — brain-wide plasticity, though not hippocampus-specific. Ten Brinke and colleagues found aerobic training increased left, right, and total hippocampal volume over six months in 86 women aged 70–80 with probable mild cognitive impairment. But when Firth and colleagues pooled every controlled trial — 14 studies, 737 participants — the average effect on total hippocampal volume was not statistically significant. The repeatable signal sat in the left hippocampus, driven less by growth than by prevented decline: exercisers held volume that controls lost.

That is the state of the evidence: one strong randomized trial, several supportive smaller ones, and a pooled result that softens "grows the hippocampus" into something more defensible — aerobic training in later life reliably defends left-hippocampal volume against shrinkage, and sometimes produces genuine gains. Reviews from Erickson's group (2014) reach the same tempered conclusion: these regions "remain pliable and responsive to moderate intensity exercise" over six months to a year, with heterogeneity the field still owes an explanation for.

Why Fitness Change Matters More Than Group Labels

Randomization assigns groups; it does not assign adherence. Within these trials, the people whose cardiorespiratory fitness actually improved were the ones whose brain volumes held or grew — the volume effect tracks the physiological dose delivered, not the label on the session. That matters for expectations: someone who attends every session but never raises their aerobic capacity should not expect the MRI result. The mechanism tested is cardiovascular adaptation, which is why VO₂ max is the number to watch. It also reframes nulls: an under-dosed trial is not a disconfirming trial.

The Dose That Produced the Effect

The prescription inside Erickson 2011 was unglamorous: walking built up to about 40 minutes per session, three days a week, at moderate intensity — conversational pace with effort, the working definition of Zone 2 training. That is roughly two hours of weekly aerobic work — a target the Walking pillar treats as a foundation, not an extreme. To install it, the Cardio Conditioning protocol handles progression and the 3-2-1 formula shows where the aerobic days go. One honest note from the ledger: in the MCI trial, resistance training did not match aerobic training's hippocampal effect — weights matter for muscle and bone (see Strength Training After 40), but this volume evidence belongs to aerobic work. Training the brain through skill rather than fitness is a separate lever, owned by Learning New Skills After 50.

Watch-Items and Honest Limits

• 🚨 Memory changes that disrupt daily life are not a training question — forgetting names is aging; forgetting routes home or repeating questions is a medical evaluation. Start with a clinician, not a treadmill.
• ⚠️ Symptoms during exercise outrank any brain benefit — chest pressure, shortness of breath disproportionate to effort, dizziness, or irregular heartbeat during sessions means stop and get evaluated before continuing.
• 📉 Shrinking volume is a risk marker, not a diagnosis — atrophy raises statistical risk, but individual scans overlap heavily; no MRI reads as destiny.
• ⏳ No lifespan or dementia promises — the trials measured volume and memory tests over months, not years of life. Cohort associations with better late-life brain outcomes are not a contract.

Questions, Answered Briefly

• 🚶 "Did the walking group grow new neurons?" Not shown in humans — the trial measured volume, not cell counts. Volume can reflect cell size, branching, and blood supply; the "grows new brain cells" claim outpaces the data, a boundary What Exercise Can't Fix owns.
• ⏱️ "How fast did the effect appear?" The trial measured at baseline and one year; interim timing is unknown. Shorter trials saw left-hippocampal effects by six months — think months, not weeks.
• 🧪 "Does this apply at 45, or at 75?" The trial enrolled 60–80, the MCI trial up to 80. The aging hippocampus has more decline to defend, which may be why the signal appears there; midlife effects are largely untested.
• 🔁 "What about a single hard session?" Acute bouts transiently bump BDNF and attention for hours — real, but separate from structural change; see the acute-session page. Structure follows months of accumulated work.

Related Topics