Edit Module
Edit Module
Edit Module
Edit Module

Can We Use Computers to Make the City Look More Natural?

For centuries, landscape architects have tried to replicate the natural landscape inside our built ones. One local professor is trying to figure out what exactly makes one park look more natural than the next.

The South Lagoon in Lincoln Park   Photo: John J. Kim/Chicago Tribune

Insignificant as it may seem, one of the hardest things about living in Chicago for me is the lack of nature. When people leave for Chicago because of the weather or whatever—as my friends have left for Los Angeles, New York, Albuquerque, San Francisco—the only thing I’m really jealous of is that there’s a 100 percent chance they’ll be closer to something that looks like nature instead of hundreds of miles from the nearest mountain. Taking my daughter to the Garfield Park Conservatory when it gets particularly gray and grim, as it’s been this winter, is nice, but it’s not quite the same.

Chicago has nature; Chicago’s part of nature. But perhaps there’s something to being surrounded by it, just being able to see it. Reading this CityLab piece by Laura Bliss reminded me of not just what I miss, but why.

It’s about a study led by University of Chicago psychology prof Marc Berman, who used algorithms in an attempt to determine city park scenes that look “natural.” On one side of the experiment, subjects said whether the pictures looked natural or not; on the other, the experimenters matched the pictures with “low-level visual features” to distinguish what looked natural.

And the determinants would be familiar to anyone who uses Photoshop. For example:

(A) has less diversity in both hue and color saturation than (B), making it more natural-looking. Below, a look at how the researchers pulled out straight edges (B) from an image (A). Fewer straight edges increased an image’s being perceived as natural. “Edge density” © also mattered—the more densely clustered the edges, the more natural an image looked.

In short, they took things perceived to be natural, despite being designed, and broke them down to their simple constituent parts. What’s the point?

[O]ur purpose was not the classification of scenes, but rather identifying simple, low-level visual features that related to subjective perceptions of naturalness and could be readily manipulated in visual stimuli. Future research could then use such features, which can be easily manipulated, to test and design new environments in ways that may improve psychological functioning.

What’s interesting about this is that it’s a new approach to a very old concept in civic life—imitating nature within the built environment to improve the health and well-being. That’s been constant, though the designers who have brought that to life—whether it’s Frederick Law Olmsted, Jens Jensen, Piet Oudolf, or Gina Ford—have employed different, evolving, and sometimes conflicting methods. Berman’s approach, in that context, is interesting: What if you used the scientific method to determine what “natural” looks like?

Prior research by Berman and others suggests the benefits that might come from employing such a built natural environment. For instance, from a prior study led by Berman, before he came to Chicago:

“Walking in nature may act to supplement or enhance existing treatments for clinical depression, but more research is needed to understand just how effective nature walks can be to help improve psychological functioning,” he said. Dr. Berman’s research is part of a cognitive science field known as Attention Restoration Theory (ART) which proposes that people concentrate better after spending time in nature or looking at scenes of nature. The reason, according to ART, is that people interacting with peaceful nature settings aren’t bombarded with external distractions that relentlessly tax their working memory and attention systems. In nature settings, the brain can relax and enter a state of contemplativeness that helps to restore or refresh those cognitive capacities.

Or a well-known study from the University of Illinois’s Frances Kuo:

The University of Illinois researchers tested the conventional wisdom that, in the inner city, barren spaces are safer than spaces with trees and greenery that could hide illicit activity. The study compared crime rates for inner-city apartment buildings with varying amounts of vegetation and found that the greener the surroundings, the fewer crimes occurred against people and property.

The scientists compared crime rates for apartment buildings with little or no vegetation to buildings with high levels of vegetation. They found that roughly half as many crimes (48 percent fewer property crimes and 56 percent fewer violent crimes) were reported in buildings with high amounts of vegetation. In addition, buildings with medium amounts of vegetation had 42 percent fewer total crimes (40 percent fewer property crimes and 44 percent fewer violent crimes) than did buildings with low levels of vegetation. Far from shielding criminals, nearby vegetation seems to shield against them.

The urban park as moral uplift is a very old idea, flourishing during the Progressive and City Beautiful eras and captured in flowering prose, such as that of Jensen, the legendary Chicago landscape architect (via Robert Grese’s Jens Jensen: Maker of Natural Parks and Gardens):

Gardens create a love for the soil in the minds of the children, that will develop into a desire for better, cleaner, and healthier homes in the mind of an adult. They appeal to the finer feelings of mankind and elevate the depressed in soul and mind to a higher place in the human family and to a greater appreciation of the responsibilities of free-born men and women.

Compare that to Berman et al:

Previous research has shown that interacting with natural environments can have a salubrious effect on cognitive and affective processing compared to interacting with more urban/manmade environments [1], [3], [5]. This suggests that there is something about natural environments that differs from urban environments that could improve psychological functioning.

But it’s essentially the same thing. Jensen, again:

Straight lines spell autocracy, of which most European gardens are an expression, and their course points to intellectual decay, which soon develops a prison from which the mind can never escape.

Berman (emphasis mine):

The LD classifier was able to successfully predict whether an image was perceived as high- vs. low-natural with 79% accuracy. This prediction accuracy is well above chance performance (50%) and suggests that these low-level visual features reliably predict individuals’ perceptions of naturalness. When we examined the features that appear most critical to classification, we found that edge density, the number of straight edges, and the standard deviation of hue were the most critical features.

As a writer I appreciate the spunk of “straight lines spell autocracy,” even if reactionary nationalism is not—the tremendous results of Jensen’s work aside—a reliable guide to successful landscape architecture.

Berman’s approach is 180 degrees from Jensen, actually testing the form of constructed nature versus our perceptions of it. And perhaps unsurprisingly, it comes to similar conclusions as Jensen, and for a similar purpose: determining how to uplift the spirit, even if we now call it cognitive and affective processing.


Edit Module


Edit Module
Submit your comment

Comments are moderated. We review them in an effort to remove foul language, commercial messages, abuse, and irrelevancies.

Edit Module