The folklore is older than the studies. Aristotle taught while walking. Charles Dickens walked twenty miles a day and credited those walks with the structure of his novels. Steve Jobs took every important conversation outside, on foot. Nietzsche claimed that only thoughts arrived at while walking had any value. All of which is suggestive and none of which is evidence. A systematic review and meta-analysis published in PLOS One in 2026 supplies the part that was missing: a quantitative summary of what walking actually does to creative thinking, drawn from twenty-three studies and 1,036 participants [1].
The headline result is large enough to be worth pausing over. Walking produced a large effect on divergent thinking — the cognitive ability to generate novel and useful ideas — with a standardised mean difference of d = 0.93 (95% confidence interval 0.44 to 1.42). Across the subset of randomised trials only, the effect held: d = 0.82 (95% CI 0.35 to 1.28). In the language of effect sizes, anything above 0.8 is conventionally described as "large", the same category occupied by the most robustly demonstrated interventions in clinical psychology. Walking, on a treadmill or in a corridor or outside in a park, moved the dial on idea generation by an amount that, in any other domain, would make headlines.
What divergent thinking is, and what walking did to it
Divergent thinking is the ability to generate many different responses to an open-ended prompt. The classic laboratory task — the Alternative Uses Test — asks the participant to list as many uses for a brick, a paperclip, or a tin can as they can think of in a fixed time. The score is not whether any single idea is brilliant. The score is fluency (how many ideas), flexibility (how many distinct categories), originality (how unusual relative to the population), and elaboration (how developed each idea is). It is the laboratory’s best proxy for the cognitive move that produces new directions in writing, design, science, and strategy — the move that goes from a blank page to a list of options worth considering.
The meta-analysis pooled twelve randomised experimental studies, nine non-randomised experimental studies, and two observational studies, drawn from sixteen articles. The pooled effect was robust to sensitivity analysis. Walking, compared with sitting or with a control condition, reliably increased the number, variety, and originality of ideas the participant produced. The Grading of Recommendations Assessment, Development and Evaluation framework rated the certainty of the evidence as moderate — meaning the effect is likely real and large, with some residual uncertainty about precision.
The null finding is just as interesting
The meta-analysis tested a second cognitive ability in parallel: convergent thinking. Convergent thinking is the analytical move — given a set of options, pick the best one. The Remote Associates Test is its classic measure: presented with three words ("cottage", "swiss", "cake"), find the single word that connects them ("cheese"). It is the cognitive move that goes from a list of options to a decision.
Walking did not help convergent thinking. The pooled effect was d = 0.16 (95% CI -0.31 to 0.63), and the GRADE certainty was rated very low. The finding is plausibly a null. The interpretation is not that walking dulls analytical thought — the confidence interval crosses zero in both directions — but that walking does not appear to be a tool for analytical sharpening. It is a tool for option generation.
Walking is a tool for generating options, not for picking between them.
The practical reading for a working professional is direct. The walk is the brainstorm. The desk is the decision. A morning walk to think through what an article might be about, what a strategy could include, what shape a difficult conversation could take — that is the activity the new evidence supports. A walk to decide which of three options to commit to — less so. The decision wants the desk and a list.
Why walking, mechanistically
The meta-analysis does not adjudicate mechanism, but the candidate explanations have accumulated. Walking increases cerebral blood flow, particularly to the prefrontal cortex — the region most active in idea-generation tasks. Oppezzo and Schwartz’s 2014 Journal of Experimental Psychology paper, one of the trials pooled into the meta-analysis, ran four experiments comparing walking and sitting and found the divergent-thinking gain held whether participants walked indoors on a treadmill facing a blank wall, walked outdoors, or walked outdoors then sat down [2]. The setting did not matter much. The action of walking did.
A separate strand of the literature points to the brain-derived neurotrophic factor system. Aerobic activity at moderate intensity raises peripheral BDNF, and BDNF appears to support the synaptic plasticity that underlies flexible association — the cognitive move that links a brick to a doorstop, a sound dampener, a planter, an anchor for a sheet pulled across a clothesline. The pathway is not yet definitive but the picture is coherent: a body in motion is a brain in a state more disposed to make unusual connections.
What this is not
The participant population matters. Most of the studies pooled in the meta-analysis used post-secondary students — younger adults, predominantly in education, with above-average baseline cognitive performance. The effect in older adults, in clinical populations, in adults under chronic stress or sleep debt, has not been quantified to the same standard. The pace and intensity of walking varied across the trials and was not consistently reported. The interventions were short — minutes to half an hour — and the outcome measurements were acute, not weeks-later. The finding is not that habitual walkers think more creatively than habitual sitters over years (that is a different and more difficult study to run). The finding is that thirty minutes of walking, before an idea-generation task, produces a large and replicable improvement in the output of that task.
A small protocol for the working professional
The applied version of the finding is unremarkable enough to be useful. Before any work that requires generating options — first draft of a piece, a list of possible angles on a client problem, the next quarter’s priorities, the names for something new — walk for half an hour. Don’t listen to a podcast. Don’t take a call. Carry a notebook or use a voice memo app. Walking with the prompt held loosely in mind is the intervention. The walk doesn’t need to be brisk in the cardiovascular sense; the divergent-thinking effect appears across speeds. It does need to be the activity itself, uninterrupted, with the cognitive prompt available.
The walk after the work, when the day is done and the ideas have been generated and decided, is a different thing. That is the cardiovascular walk — the one that does the body the slow steady favour the longevity literature documents. Both walks are real. They serve different purposes and they happen at different times. The new meta-analysis names the morning one and gives it a number.
