How much nature is enough

The lower bound is the clearest, and it is 12%. Qinghua Lei had 15 participants work at green coverage ratios of 0, 0.2, 5, 12 and 20% of a workspace in 2021. At 0.2 and 5% nothing changed psychologically. Only 12 and 20% moved brain activity, and 20 was overwhelming. 12% came out as the working optimum.

Window size has a threshold too, and a price. The architecture researcher Won Hee Ko put 60 people through 80 office scenes in virtual reality in 2026. A window-to-wall ratio of 25% is enough for a direct view when the content is good. Poorer content needs about 40%, and a side view with poor content can need up to 80. Ko also costs it out: across three American climate zones the difference runs to roughly 8% of a building’s energy use, and beyond a ratio of 50% the shading has to be designed for.

What makes a natural pattern restorative has been narrowed down twice. In 2016 the environmental psychologist Agnes van den Berg had 40 people view natural and built scenes at original size and at 400 and 1,600% magnification. Natural scenes were looked at longer and rated more restorative. Controlling for the perceived complexity of the magnified parts cut that difference by 47% for viewing time and by 20% for restorative quality. Recursive complexity accounts for a fifth of the restorative difference.

The forestry researcher Amari Davison narrowed it again in 2025 with 235 people making 60 forced choices between 12 leaf silhouettes. Leaves of moderate fractal complexity, a dimension of 1.4 to 1.5, won on excitement, interest, naturalness and appeal. Leaves around 1.3 won on relaxation. Ginkgo biloba ranked highest for relaxation, plane and horse chestnut for excitement.

The psychologist Kelly Robles asked people to rate fractal designs already built into real rooms rather than shown on a screen, across a series of experiments with 39 to 81 undergraduates each and one drawn from 130 adults. Preference, complexity and engagement all rose with fractal complexity while refreshment and relaxation stayed flat or fell. Robles found no complexity value that raised preference and relaxation together.

Andrew Thatcher reproduced a plant effect in a laboratory and could not reproduce it in two call centres over six and fourteen weeks. Yiming Shao found a living wall slowed the build-up of negative states during desk work and left people less alert after resting beside it.