What the window is doing

Between 1972 and 1981, at a two-hundred-bed hospital in Paoli, Pennsylvania, forty-six people had their gallbladders removed and recovered in the same three-story wing of the same building.

Twenty-three of them had a window that looked out on a small stand of deciduous trees. Twenty-three had a window that faced a brown brick wall. Nobody chose. Patients were put in rooms as the rooms came free.

Years later the records were read back in matched pairs. Each tree-view patient was set against a wall-view patient of the same sex, within five years of age, the same smoking status, the same rough weight category, a similar history of previous hospitalization, surgery within six years, and the same floor. On the second floor the pairs were matched on the color of the room as well, blue or green. The rooms were nearly identical in dimensions, window size, and furniture. Every window was 1.83 m high and 1.22 m wide.

Only operations performed between 1 May and 20 October were used, because the trees had to be in leaf.

The tree-view patients went home sooner. Not by much: 7.96 days after surgery against 8.70, so about three quarters of a day. Their nurses wrote fewer negative things about them, and that gap was not small at all. Notes like "upset and crying" or "needs much encouragement" appeared 3.96 times per wall-view patient and 1.13 times per tree-view patient.

Then there is the painkiller result, which is the one almost everybody gets wrong.

What the drug chart actually said

From the second day after surgery to the fifth, the tree-view patients were given fewer of the moderate and strong doses, and more of the weak ones. The wall-view group got many more doses of potent narcotics. The tree-view group more often got aspirin and acetaminophen.

So it is not that they needed less pain relief. It is that the pain relief they were given moved down a shelf.

The paper does not mention morphine. The strong category was hydromorphone, sold as Dilaudid, and large doses of meperidine, sold as Demerol. The weak category was acetaminophen, sometimes with a little codeine.

And the effect has edges. On the day of surgery and the first recovery day there was no significant difference between the groups, which the author expected, because patients that fresh out of an operation are too medicated or in too much pain to look at anything. On days six and seven there was no difference either, mostly because only 45 percent of patients were taking any painkillers by then. The whole result lives in a four-day window in the middle.

Who did this, and what he refused to claim

Roger Ulrich, then in the geography department at the University of Delaware, published it in Science on 27 April 1984, under a title that is careful in a way the internet has never been: "View through a window may influence recovery from surgery." Two pages. A nurse who did not know which view belonged to which room pulled the data out of the files.

It has been cited more than five thousand times since, which is a strange amount of weight for two pages about forty-six people (Semantic Scholar, September 2026).

Ulrich printed his own warning in the last paragraph, and it is the part that never gets quoted. The wall in his study, he wrote, was a comparatively monotonous one, largely featureless. The conclusions cannot be extended to all built views, nor to other kinds of patient. He went further: for a patient who is chronically understimulated rather than anxious, he suggested a built view such as a lively city street might be more stimulating, and therefore more therapeutic, than many natural views.

He also printed the detail that works hardest against him. The nurses' stations on both floors sat somewhat closer to the wall-view rooms.

Why nobody has simply run it again

Because the operation stopped existing.

Ulrich's patients had open surgery, as everyone did then, and stayed in hospital about eight days. Then keyhole surgery arrived. The share of elective gallbladder removals in the United States done laparoscopically went from 3.5 percent in 1989 to 73.7 percent in 1992 (Kemp, Zuckerman and Finlayson, Journal of the American College of Surgeons, 2008, 206(1):28-32). Within eight years of the paper, the eight-day recovery it measured had mostly been designed away.

The study is a fossil. It captured a kind of hospital stay that is largely gone, using a matching method that only works when people stay in one room for a week.

The theory that was already there

The framework people reach for to explain this is Attention Restoration Theory, and it is usually described as if it grew out of Ulrich's paper. It did not. It was already running.

Rachel Kaplan published on the psychological benefits of gardening in Environment and Behavior in 1973, eleven years earlier. She and Stephen Kaplan, both at the University of Michigan, laid out the full account in their 1989 book The Experience of Nature, and Stephen Kaplan gave it its clearest statement in the Journal of Environmental Psychology in 1995.

The argument runs like this. Directed attention, the deliberate kind you use to read a contract or hold a decision in your head, is limited and gets tired. Natural settings pull at a different kind of attention, which the Kaplans called soft fascination: the low-demand, involuntary pull of water moving, or leaves shifting, the sort of thing that holds you without asking anything. It leaves enough of your mind free to think.

It is a theory, not a measurement. It is what people use to explain results like Ulrich's, and it was written before them.

Near is not the same as in

Thirty-one years after Ulrich, Gregory Bratman and colleagues, working out of Stanford, took thirty-eight healthy people who lived in cities, sent nineteen of them on a 90-minute walk through grassland with scattered oaks near the Stanford campus, and sent nineteen on a walk of exactly the same length, 5.3 km, down El Camino Real in Palo Alto, three to four lanes of traffic in each direction. They scanned everyone before and after.

The nature walkers came back with lower scores on a rumination questionnaire and less blood flowing to the subgenual prefrontal cortex, a region tied to the kind of looping self-referential thought that rumination is made of. The road walkers came back unchanged on both (PNAS, 2015, 112(28):8567-8572).

The authors add their own honest caveat: scattered trees line El Camino Real too, so their urban condition was not free of nature, and their effect is probably an underestimate.

That is a walk, though. Ninety minutes, outside, on your feet. It is not a window.

So what does the window entitle you to say

Less than the internet says, and something more interesting.

You cannot use Ulrich to argue that your desk facing a parking lot is worse than a desk facing a garden. He wrote, in the paper, that you cannot. What he showed is narrower and stranger: that for people recovering from abdominal surgery in one Pennsylvania hospital across ten summers, in a four-day window in the middle of recovery, something as passive as which way the glass faced showed up in a drug chart and in what the nurses wrote down about them.

Nobody was asked to do anything. Nobody knew they were in a study. Nobody went outside.

Rewyld runs its practices outdoors, because outdoors is where the signal is clearly strongest, and Bratman's walkers are the better evidence for that. But the study that started the conversation happened to people who never left the room. If all you have today is a window, five minutes of real attention on whatever is alive on the other side of it is not nothing. It is just less than a walk.