How We Know Bad Architecture is Bad

Eye tracking simulations show how the brain recoils from terrible buildings

I’ve written a lot about architecture on this Substack. Generally, I don’t like contemporary architecture, and neither do you (unless you’re an architect or an architecture prize judge). But why? What’s the matter with the hulking concrete blocks or blank impassive facades that infest fill our cities? It’s simple: the human brain prefers organic shapes and fractal patterns that evoke natural forms. Trees, for instance, are fractals: the same branching pattern recurs at different scales, from trunk to limbs to branches to twigs. Organic patterns tap into our species’ natural biophilia, or love of surroundings that burst with life.

That’s why we’d rather stay in a hotel room with a forest view than one that looks over a parking lot. It’s also why a medieval church or a baroque library is preferable to a brutalist cube or anything that wins a Pritzker Prize. Traditional ornamentation looks like life.

Most normal humans understand this at a gut level. But in the perversely scientistic postwar era, experts frequently use(d) technical jargon to convince each other, and eventually the rest of us, that our gut instincts about things like beauty were wrong, or at least embarrassing and silly.

They got away with this because some scientific facts do fly in the face of common sense. Heavy things turn out to fall at the same rate as light things. Pumping our arms full of sickness-causing germs somehow protects us from getting those same sicknesses (don’t tell RFK). The universe is made of something called “quarks.”

After centuries of such counterintuitive findings, scientists and other experts fell into the habit of assuming that regular people are morons. Sighing with impatience, they approach the public with a default “gotcha!” attitude that relishes overturning false intuitions, bursting mistaken bubbles, revealing the erroneousness of folky assumptions.

The problem comes when this “gotcha!” attitude blunders into domains where scientistic or progressive prejudices are destructive and lack evidence. The public suddenly finds all the Smart People agreeing that, say, Communism is great, or ornamentation on buildings is morally depraved.

So we’ve now been lectured at for decades by architecture experts who insist that, because Progress™, we should love the flat, hulking, featureless, weird, often visually confusing monstrosities they keep building. As Tom Wolfe described in Bauhaus to Our House, they’ve pumped out jargon-filled reviews and convinced wealthy people to endow a gazillion university professorships, which they use to spread the gospel of modern and postmodern architecture through the educated and influential social circles that set the intellectual tone for society.

Yet everyone else still dislikes the buildings they build and prefers traditional architecture. We know this because people (and real estate markets) say so. The British public preferred traditional over modern architecture by a ratio of 3 to 1 in a 2009 study. This ratio was about the same for a similar 2020 study in the US. In another survey, Americans’ most beloved buildings were practically all traditional or Art Deco (a style that modernized but preserved many traditional elements). If you scroll down that list, you won’t find a modern non-monument building until #19, the original World Trade Center buildings. The vast majority of America’s most beloved buildings are Beaux-Arts, neoclassical, neo-Gothic, and other styles rich in the patterned forms that universally make humans everywhere feel good.

But responding to surveys won’t convince experts. Probably no other kind of scientific evidence will either, because the funny thing about technocratic scientism is that it isn’t scientific. It’s not about evidence. It’s about finding ways to make a narrow set of progressive (and usually minimalist) prejudices look like the objective truth. Nevertheless, scientific evidence that contemporary architecture is unpleasant and even unhealthy does keep stacking up.

So what the heck?

Let’s look at some of that evidence.

How We Gaze (Or Shudder in Avoidance) at Buildings

In the past few years, cognitive scientists and architecture critics have stumbled on a new way to quantify exactly how desperately our eyes and brains recoil from looking at the modernist catastrophes that lumber like rectilinear lizards through our once-beautiful cities: eye tracking. When we look at a scene, our eyes move unconsciously and rapidly over the different parts of the image, returning to some places, avoiding others. Eye tracking studies use sensors and software to quantify how this happens before conscious awareness kicks in, a period of “pre-attentive processing” that gives scientists a window into our default preferences and aversions.

In one eye-tracking study, scientist took real photos from German cities and showed them to 100 volunteers. Each photo contained both a traditional building and a contemporary structure. Subjects wore eye-tracking equipment as they looked at the different photos, then they answered some questions about the buildings.

The researchers found that subjects looked more quickly at the traditional buildings, their gaze rested longer on those buildings, and it took longer for them to fixate on any part of the contemporary structures. Heat maps showing how much subjects’ attention lingered on different parts of the photos showed much more color around the traditional facades, and comparatively little around the modern ones.

(Keep reading below to see what I’m talking about.)

But were the subjects really appreciating the traditional buildings, or just staring at them? Sometimes we can’t tear our eyes away from horrible scenes, like car accidents or unfortunate celebrity plastic surgeries. Is that what was going on here? No, because in the surveys, subjects preferred the traditional buildings by ratios of 4 to 1 or more. They also reported finding the traditional buildings far more interesting to look at than their contemporary counterparts.

In other words, the self-report responses dovetailed with the results from the eye-tracking software. People looked longest and most frequently at the parts of the photos they said they liked.

Other eye-tracking studies have turned up similar results. In one, people looked longer at images of New York City neighborhoods that used traditional designs like complex facades than at images of contemporary (glassy, linear, flat) neighborhoods. In another, people shown images of Boston’s brutalist City Hall (one of the most hated buildings in the world) let their gaze rest almost anywhere except the building itself. By contrast, their gazes dwelled easily on the colonial facade of the Old State House, only a couple of blocks away.

So people really do avoid looking at contemporary architecture and relish looking at the old stuff — even before they’re consciously aware of it.

Simulated Eye Tracking

But you don’t actually have to strap expensive eye trackers onto paid subjects to see this preference for yourself. Software that was trained on lots and lots of data from real human eye-tracking studies can simulate visual processing of a scene. Show the software an image, and it spits out a heat map and probable trajectory of gaze patterns.

Mostly this software is used in marketing. Graphic designers generate, say, a full-page magazine advertisement for perfume. The campaign managers want readers of the magazine to look at the ad, grock what it’s selling, and then rush out to buy the perfume. So they feed the mocked-up ad through the VAS (visual attention simulation) software, and voilà, it shows how normal humans will likely respond to the ad.

If the heat spots are all clustered around the close-up photo of the perfume bottle or the sexy model who’s holding it, then the advertisement is probably fine (for advertising purposes). If the simulated eye is drawn instead toward some weird background feature or the strange font of the copy, then the ad people rework the design until the VAS gives them a better result.

If the simulated visual sequence (which is different from the heat map) goes first to the sexy model, then to the perfume, then finally to the name of the perfume in bold font, that’s a good order. You get a strong emotional impression before taking in key information. That’s what advertisers want. But if the gaze goes first to the cap on the perfume bottle, then to a background feature, then to the model’s elbow, then back to the bottle cap, something is probably off. The bottle cap is drawing too much attention.

The point is not to bore you with details about how graphic designers manipulate your last remaining shreds of internet-ravaged attention for profit. The point is that VAS works well enough that corporate types are happy to use it to capture your attention.

VAS for Architecture, or: Fancy Imagery Demonstrates the Obvious

Now let’s look at how VAS processes image of architecture.

Below, you’ll find a photo of the new(ish) Boston University “Jenga” building, also known as the Duan Family Center for Data Sciences and Computing. I went to BU, and I hate this building. It looms like a drunken robot over the cowed and shrunken pillars of the once-august School of Theology and the College of Arts and Sciences. It haunts the sky like the maniacal ghost of an autistic savant. Everything in the world will look like this building when the left brain finally crushes the right brain forever, imposing its thousand-year regime of algorithmic order where love, beauty, and joy are left twitching like smashed insects and the only remaining sounds are the humming of cooling fans and the wind whipping eternally through server farms.

Now, here’s how the VAS system looks at this monstrosity of a building. Or, actually, doesn’t look at it:

Note that the Jenga building, which cost $305 million United States dollars to build and is supposed to be the new heart of campus, does not exactly capture the gaze. In fact, it repels it. Except one blip of blue color near the top left of its upper stories, the Jenga tower, in all its LEED-certified glory, is mostly a black wasteland. Instead, the simulated gaze is drawn to the Art Deco west facade of the School of Theology, part of a background building, and a cluster of street signs in the bottom righthand corner (the orange spots).

I think it’s interesting that the simulated gaze moves up both the traditional spires of the School of Theology and the College of Arts and Sciences, but not the massive bulk of the data sciences tower. Instead, the gaze mostly just follows the tops of the older buildings and the street trees lining Commonwealth Avenue.

If you were a graphic designer and the Jenga tower was your product, you would be very disappointed.

You might get fired.

Now, let’s compare how the VAS processes a photo of a much nicer building:

This is the Peace Palace in the Hague, a UN building for the International Court of Justice. It was built in 1913 in Neo-Renaissance style, which means it has lots of nice pretty elements that humans typically enjoy seeing on buildings. Just as in trees and other fractal forms, shapes are repeated at different scales, like the Roman arches in the entryways and smaller windows and the spires of the different towers. The design is harmonious overall, not too symmetrical, not too unbalanced.

At least, that’s my opinion. What does the VAS say?

Wow! A graphic designer would be very pleased. So would the long-dead architects for the Peace Palace. Those geniuses seem to have designed a building that draws the eye to just the right elements: the clocktower, the entryway arches (it’s always nice to be able to find your way into a building), some detailing on the roof, and a grand arched window with tracery. That is, the building itself draws most of the attention, not the background or foreground or some random spot way over in the corner of the sky. It provides a visual scene the gaze can rest on.

It’s a world of difference from the scaly, inhuman Jenga tower.

Now let’s return to Boston for a third set of images.

Here, we’re standing in Copley Square looking out over one of Boston’s most beloved buildings, the Henry Richard Hobson–designed Trinity Church. To the right is the city’s tallest building, the minimalist John Hancock Tower.

I love Trinity Church. It’s a gloriously authentic American take on a traditional European style, in this case Romanesque. It anchors the surrounding square with its pious gravitas and soothing beauty. It’s also full of fractals. The Roman arch repeats on many different scales, from the entryway to the narthex to the upper windows, and so do the sharp points of the tower spires.

Perhaps surprisingly, I don’t hate the John Hancock Tower. The blue glass color is appealing, and the way it reflects the clouds can be downright mesmerizing on an autumn afternoon. But it isn’t Trinity Church.

What does the VAS say?

Boom. The eye is drawn straight to the central tower of the church, with all its interesting repeated shapes and pleasing proportions. It also rests lovingly on the rows of arches on the second level above the entryway and on a street tree branch overhead. But the simulated gaze avoids the profile of the John Hancock Tower like a contractor on an overdue project refusing your calls. It seems to look everywhere except the glassy surface of the city’s tallest, most prominent building.

In terms of perceptual processing, it’s as if the Hancock Tower simply weren’t there.

Finally, here’s the sequence analysis for the Copley Square scene:

The numbers represent the most probable path an average eye will take when first encountering this image. As you can see, every step in the sequence remains anchored on or near Hobson’s inimitable church. The gaze is drawn first to the high tower’s clay roof, then the second-story arches, then a bit of overhanging tree in the foreground (we love those organic fractals!), and finally to part of the entry arcade. Not once does the simulated gaze sequence bop over to the office tower. That’s because the blank reflective glass is just repeated angles. It’s like an anti-fractal. To the human brain, it’s meaningless.

Science Rehabilitates Common Sense

None of this should really be surprising. But it is illuminating.

We’re so used to experts telling us at every turn that our default preferences are wrong, or that human nature is a blank slate to be reworked at will, that many of us unconsciously believe it. Especially if we’re overeducated. If graduate degree holders see a shapeless concrete mass that’s won a Pritzker Prize, we assume that the problem must be with us. We must not have enough sophistication to appreciate the subtlety of today’s architectural genius.

No, we’re just being sold a bill of goods.

The eye-tracking training data for the VAS software is essentially invariant across cultures. The recurring preference for fractals and organic, repeating shapes and patterns isn’t a societal bias, it’s built into the human brain.

The buildings we consciously dislike don’t attract our attention even unconsciously. Our brains don’t process their facades very easily, and they don’t remind us of living things. As more and more of these faceless buildings flop upwards like toxic mushrooms in the forests of steel that are our cities, we as a species move farther and father away from the environments we naturally crave. The vague discomfort so many of us feel most of the time isn’t just misfiring brain chemicals. It’s a signal that we’re trying to live in places that aren’t designed for us.

But it doesn’t have to be this way. Just now before finishing this essay, I was walking down a side street in Harvard Square on my way to daily Mass. The buildings around me were beautiful Victorian and Queen Anne mansions with ornate porches and mansard roofs draped over with lush mature street trees, brick prewar apartment buildings with bas reliefs and real keystones above the windows. I don’t mean to sound like a fop or aesthete, I just mean that I really liked the environment I was in. Underneath all the layers of interpretation and critique I could make, it made me feel good.

This is more relevant and true information than all the theory-laden opinions of all the architecture prize judges in the world.

They can tell you that you should love their dystopian projects til they’re blue in the face, but if you just pay attention to how your body and brain feel when you’re walking down the street, it will quickly become obvious that they are full of it.

But if trusting your own embodied experience isn’t enough for you, you now have VAS and eye-tracking data to support your gut intuitions, too. As in other areas where experts spent the 20th century telling us to ignore our human sensibilities, scientific evidence might be what finally nudges us to take the first unsteady steps toward reclaiming common sense.