Nancy Felipe and Robert Sommer wanted to know what people actually do when their space is invaded, so they went to a university library and did it. An experimenter would select a woman sitting alone at an otherwise empty table, and sit down next to her — not opposite, not a seat away, but in the immediately adjacent chair, sometimes moving his own chair closer still.
Within ten minutes, about a third had left. Within thirty, more than half. In the control condition, where nobody sat near them, almost nobody left.
The finding that makes the study worth reading is not the departure rate. It is what the subjects said. Almost none of them protested. Almost none asked the experimenter to move. Very few even mentioned the intrusion afterwards, and when asked why they had left they produced reasons: they were finished, they had a class, they wanted a different section. The discomfort was severe enough to drive them out of the building and not accessible enough to be named.
Felipe and Sommer had also run a version in a mental hospital, sitting next to patients on benches, with similar results. The behavior does not depend on politeness norms in the way one might assume; it appears to be pre-verbal.
Seven years later, Middlemist, Knowles and Matter demonstrated that the arousal is physiological rather than merely social, in a study whose method has been debated on ethical grounds ever since. They observed men in a public lavatory and timed the delay before urination and its duration, with a confederate standing either immediately adjacent or one position away. Close proximity increased onset delay and shortened duration — measures of sympathetic arousal that are not under voluntary control and cannot be explained as social awkwardness.
Put these together and you have the mechanism behind a set of coercive practices that are rarely described as coercion because nobody touches anybody.
Interrogation seating is designed on this basis: the subject in a chair with the interviewer's chair closer than a stranger's chair should be, no table, the door behind the interviewer. Crowd control uses density deliberately — kettling produces sustained proximity stress in people who cannot leave, and the resulting agitation is then cited as evidence of the crowd's character. High-pressure sales floors put the closer beside the customer rather than across from them. In each case the pressure is real, measurable, and deniable, because the only thing that happened is where somebody was standing.
Two practical points follow from the library study specifically.
The first is that your own response to an engineered space will be flight, rationalized. If you find yourself wanting to leave a room and generating reasons, check the geometry before you accept the reasons.
The second is that naming the intrusion is unusually effective precisely because so few people do it. The technique depends on the target's reluctance to make an issue of proximity, and a mild, direct sentence — could I have a bit more room, or moving your own chair without explanation — costs almost nothing and ends the pressure immediately. If it produces an outsized reaction, you have learned something worth knowing about what the geometry was for.
See Defense: exiting an engineered space