Scientists put eye-tracking cameras on test subjects and showed them photographs of American federal buildings. Then they watched where the eyes went.
Two federal courthouses. Same function. Separated by a design era.
The traditional Howard M. Metzenbaum Courthouse in Cleveland — opened 1910, Beaux-Arts facade, arched windows, carved detail — lit up red in the heat map. Eyes moved from element to element, drawn continuously by the visual complexity, never settling into blankness.
The modern Sandra Day O'Connor Federal Courthouse in Phoenix — glass, steel, flat surfaces — barely generated any strong fixation points. The eyes had nowhere to go.
This was not preference. The subjects were not asked what they liked. The cameras recorded where their eyes actually traveled: the automatic, pre-conscious visual system doing what evolution built it to do.
The finding held across every building pair the researchers tested. The traditional EPA headquarters in Washington drew intense fixation. The modern HUD headquarters nearby — same city, same function — "people barely focused on any of it." A modern federal building in Ogden, Utah, registered the most extreme result: barely a single red dot across the entire facade.
A 2020 Harris Poll of over 2,000 Americans found that 72% preferred traditional architecture for federal courthouses and office buildings. The eye-tracking matched the stated preference exactly. What people say they want and where their eyes actually go turn out to be the same thing.
This is the research of Ann Sussman and Justin Hollander, architects and researchers at Tufts, collected in "Cognitive Architecture" (2015, expanded 2021). It is not comfortable reading for a profession that built those modernist courthouses. It is among the most empirically direct challenges to the postwar architectural consensus we have.
What the Brain Was Built to Find
The human visual system evolved for one fundamental task: extract relevant information from a complex environment fast enough to act. The specific hardware evolution gave us for this is face detection. The brain has dedicated neural circuits for recognizing faces and reading their signals. Friend or foe. Alert or calm. Alive or dead.
This hardware does not switch off when we look at buildings. It is always running.
Traditional architecture, across cultures and centuries, produces facades with face-like proportions. Two upper windows like eyes. A central feature like a nose. An entry door like a mouth. The classical tripartite facade that recurs from ancient Athens to 20th-century civic buildings is not coincidental. It is the spatial arrangement that face-detection hardware processes as legible, social, present.
A glass curtain wall presents no face. The brain reads it as socially null, slightly threatening the way any blank stare is threatening.
The second piece of relevant evolutionary hardware: thigmotaxis. We are a wall-hugging species. Across cultures, people prefer to move along edges rather than crossing open featureless space. Jane Jacobs documented this empirically in her research on how people actually used urban plazas — they moved around the perimeter, clustered where there was something to be near, avoided the center. Sussman applies the scientific term: thigmotaxis. It is a survival orientation inherited from ancestors on the savanna, where edges meant shelter and open ground meant exposure.
Street walls, arcades, colonnades, covered passages — traditional urbanism works partly because it respects this instinct. The glass tower set back on a superblock, surrounded by open plaza, violates it entirely. The plaza empties not because people don't want to be outside, but because the space produces low-level neurological discomfort that compounds with time.
Who Designed for Whom
Sussman, in a 2017 essay co-authored with researcher Katie Chen, offered an unsettling explanation for how this mismatch got institutionalized.
Two men shaped the dominant architectural language of the 20th century: Le Corbusier and Walter Gropius. Both showed markers that multiple biographers have documented without always connecting to their architectural outputs.
Le Corbusier's biographers describe characteristics consistent with what we now associate with autism spectrum patterns: intense sensory sensitivity, strong preference for reduction and sameness, difficulty with the social complexity of crowded urban space. His famous hostility to "the hubbub of crowded Parisian sidewalks" was not an intellectual position. For his neurology, stripping surfaces of visual complexity was a form of survival. He designed for himself.
Gropius carried forward from the trenches of the First World War with hypervigilance and avoidance responses that researchers now recognize as characteristic of PTSD. He designed environments stripped of sensory triggers. He designed for himself too.
Through institutional power — Gropius running Harvard's architecture school, Le Corbusier dominating the Congrès Internationaux d'Architecture Moderne — what began as personal neurological accommodation became the universal standard.
Sussman is careful here: she is not diagnosing architects posthumously. She is connecting documented behavioral observations from their biographies to what neuroscience now shows about how atypical neurologies respond to visual complexity. The argument is not malice. It is that highly personal sensory needs were mistaken for universal design principles, and then taught as such.
What Alexander Already Knew
Christopher Alexander spent decades asking which buildings feel alive and which do not. He identified fifteen properties of living structure from thousands of examples: levels of scale, strong centers, boundaries, alternating repetition, contrast, roughness, the void.
He had no eye-tracking software. He arrived at these properties by phenomenological attention — by looking, feeling, and testing with human response.
Sussman's heat maps converge on the same answer from the opposite direction. What Alexander identified as the properties of living structure are, at the mechanistic level, the visual properties that the evolved human perceptual system is calibrated to find. Levels of scale match what fractal-sensitive visual processing expects. Strong centers are what face-detection hardware seeks. Roughness and contrast are what a sensory system tuned for a complex natural world needs to remain engaged.
Alexander intuited it through practice. Sussman is confirming it with instruments.
The 2020 Salingaros and Sussman pilot study published in Urban Science used decades of aggregated eye-tracking data to evaluate building elevations. The software predicted the result before any human participants were tested: traditional window arrangements are implicitly engaging, contemporary facades are not. The response is not cultural. It is mechanical.
The Cross-Domain Version
The same structure appears in other domains.
Industrial agriculture stripped complexity from soil — monocultures, synthetic inputs, no biological diversity — and then required a pesticide and fertilizer industry to compensate for the instabilities that biological complexity had been managing for free.
Ultra-processed food strips nutritional complexity from ingredients and then a supplement industry sells back what the processing removed.
Open-plan offices strip spatial complexity — no edges, no enclosure, no variation in ceiling height or material — and then companies spend on wellness consultants and ergonomic furniture to compensate for the neurological stress that spatial variation had been managing.
In each case: remove the complexity that evolved systems depend on. Create a market for the deficiency.
The glass curtain wall is the architectural version of this pattern. Strip all visual information from the facade. Then wonder why the plaza is empty and the building feels hostile.
Eye-tracking technology is less than a century old. The visual system it is measuring is three and a half billion years in the making.
Which one had more time to figure out what works?