One can spend hours looking at a calm sunset or a clear night sky. These scenes are not only effortless on the eyes — they may also be easy on the brain. People tend to like visual stimuli that require little cognitive effort to process, researchers report in the December PNAS Nexus.
The brain is the most energy-guzzling organ in the body, and visual processing alone accounts for nearly half of its energy use. Researchers have long studied how the visual system conserves energy. But the new study addresses the question from a different perspective. “Not only is the visual system optimized for efficiency, but we might have aesthetic preferences for stimuli that are efficient to process,” says Mick Bonner, a neuroscientist at Johns Hopkins University who was not involved in the study.
Neuroscientist Dirk Bernhardt-Walther of the University of Toronto and his colleagues suspected that such preferences could have evolved as cognitive shortcuts, helping organisms avoid excessive effort as they navigate their environment.
To probe the energy consumed in visual processing, the researchers turned to an existing functional MRI dataset, in which four individuals viewed 5,000 images while their brain activity was monitored. Measurements of oxygen consumption in different parts of the brain provided an indicator of metabolic activity. The team also ran these images through an artificial neural network trained on object and scene recognition, using the proportion of activated “neurons” as a proxy for metabolic expense.
The researchers then compared these metabolic cost estimates — both human and artificial — to the images’ aesthetic ratings, gathered from more than 1,000 online survey respondents who scored each picture on a five-point scale. In both cases, the metabolic effort required to process the images was inversely proportional to their aesthetic ratings.
This negative correlation was stronger in high-level visual regions of the brain, like the fusiform face area responsible for recognizing faces, and the corresponding layers in the artificial neural network. This suggests, Bernhardt-Walther says, that most of the energy savings occur during advanced stages of visual processing such as object recognition rather than low-level functions such as edge or contrast detection.
Previous studies have shown that people tend to like faces, and even cars, that look closer to the average than those that look different. We prefer the platonic version, Bernhardt-Walther says, presumably because outliers force the brain to spend energy updating its internal models of what a face or a car looks like.
Metabolic cost may also explain pleasurable experiences beyond sight. Think of the joy at solving a puzzle after having juggled and weighed multiple solutions. “The ‘aha!’ experience is profoundly enjoyable because there’s a sharp decrease in metabolic needs all of a sudden,” Bernhardt-Walther says.
Bonner say that future work should look at whether metabolic costs directly cause aesthetic preferences, or if both stem from a shared feature such as familiarity. Also, it’s not understood which properties make some stimuli more pleasing and more efficient for the brain to process than others. “What precisely makes an image easier for the visual system to process remains a huge open question,” he says.