New research suggests that the human appreciation for beauty may be rooted in biological frugality. A study published in PNAS Nexus indicates that images requiring less energy for the brain to process are perceived as more aesthetically pleasing. These findings imply that our visual preferences may serve as an evolutionary mechanism to conserve metabolic resources.
The human brain is an exceptionally expensive organ to maintain. It accounts for only a small fraction of total body mass. Yet, it consumes roughly 20 percent of the body’s daily energy reserves. This metabolic demand places a constant pressure on the organism to manage its resources.
A massive portion of this fuel is dedicated to a single sense. The visual system alone is responsible for approximately 44 percent of the brain’s energy expenditure. Consequently, simply looking at the world is a metabolically heavy task. Every scene we view triggers cascades of electrical and chemical activity that require glucose and oxygen to sustain.
Evolutionary theory suggests that organisms adapt to avoid wasting energy. It is logical that biological systems would develop strategies to encourage efficiency. One proposed solution is the use of affective heuristics. These are mental shortcuts that use feelings, such as pleasure, to guide decision-making.
If a specific action saves energy, the brain might reward the individual with a positive feeling. This reward encourages the organism to repeat the efficient behavior. Researchers have hypothesized that this principle applies to vision. They proposed that we might find certain images beautiful simply because they are easy to process.
Yikai Tang, William A. Cunningham, and Dirk B. Walther from the University of Toronto sought to test this idea. They investigated whether the idiom “easy on the eyes” reflects a physiological reality. Their central hypothesis was that aesthetic liking is inversely related to the metabolic cost of neural representations. In other words, if an image costs less to see, we should like it more.
To test this hypothesis, the authors employed a two-pronged approach. They first utilized a computational model to simulate the mechanics of the visual system. Computer simulations allow researchers to isolate specific processes without the noise of human psychology. They utilized a deep convolutional neural network known as VGG-19.
This artificial intelligence program was trained to categorize objects and scenes. Its architecture is designed to mimic the hierarchical layers of the human visual cortex. The researchers presented the model with nearly 5,000 distinct images of real-world objects and scenes. These images varied widely in content.
For every image, the team calculated a specific “metabolic cost” for the computer model. They defined this cost based on the activity of the artificial neurons. A unit within the network was considered active if its output value was greater than zero.
The researchers counted the total number of active neurons required to represent each image. They also calculated the sum of the activation intensities. An image that triggered a large number of neurons was deemed expensive. An image that triggered fewer units was considered efficient.
The authors then compared these machine-generated cost estimates to human opinions. They utilized a dataset of aesthetic ratings collected from over 1,000 participants. These individuals had rated how much they enjoyed looking at the same images used in the computer simulation.
A distinct pattern emerged from the comparison. The researchers found a negative correlation between the model’s energy cost and human enjoyment. Images that required fewer active neurons in the artificial network received higher aesthetic ratings from people. As the processing cost decreased, the reported pleasure increased.
To ensure this result was not a coincidence, the team ran the same test on untrained neural networks. These were models with the same structure but no experience in recognizing objects. These random networks did not show the same pattern. This difference suggests that the preference for efficiency arises from the specific way a system learns to structure visual information.
While the computer model provided a strong proof of concept, it remained a simulation. The authors needed to verify that biological brains operate under the same principles. To do this, they analyzed brain imaging data from four human participants. These individuals viewed the large set of images while inside an MRI scanner.
The researchers utilized functional magnetic resonance imaging. This technology tracks changes in blood oxygen levels in the brain. When neurons in a specific region fire, they deplete local oxygen supplies. Fresh, oxygenated blood then rushes in to replenish the area.
This distinct signal, known as the BOLD signal, serves as a proxy for metabolic energy use. A stronger signal indicates that the brain is working harder and burning more fuel. The researchers focused their analysis on specific regions of the brain dedicated to visual processing.
They examined the early visual cortex, which handles basic features like edges and contrast. They also looked at higher-level areas responsible for recognizing complete scenes, faces, and objects. The results from the human scans aligned with the findings from the artificial intelligence model.
In nearly all the visual areas examined, higher metabolic activity correlated with lower aesthetic ratings. When the visual cortex had to work harder to encode an image, the participant reported enjoying it less. This relationship was notably stronger in the high-level regions that process complex scene information.
The authors interpret these findings as evidence of an energy-conservation heuristic. The visual system seems to aim for a “sweet spot.” It requires enough stimulation to be interesting, but it penalizes excessive cost. The images we perceive as most attractive appear to be those that provide rich information without demanding a heavy metabolic toll.
This concept aligns with the “processing fluency” theory in psychology. This theory posits that information that is processed faster and more easily elicits a positive emotional response. The current study grounds this psychological concept in the physical reality of energy consumption.
The researchers also noted a distinction in other brain areas. While visual processing areas preferred efficiency, the Default Mode Network showed a different pattern. This network is often associated with self-reflection and daydreaming. In these regions, higher activity sometimes correlated with higher enjoyment.
This suggests that while the visual machinery prefers ease, the cognitive parts of the brain may enjoy engagement. However, the study emphasizes that the primary visual intake favors low-energy states. The “beauty” detected by our eyes is largely a signal of efficiency.
There are limitations to how broadly these findings can be applied. The study focused on rapid visual processing. Participants gave quick ratings of their initial impressions. This setup does not capture the deeper appreciation that comes from contemplating complex art.
Looking at a blank white wall is extremely energy-efficient. Yet, most people would find it boring rather than beautiful. The authors acknowledge that a stimulus must meet a baseline level of interest. The efficiency principle likely operates only after that baseline of necessary stimulation is met.
Furthermore, the “boredom” factor was not the primary variable manipulated in this study. The researchers focused on the positive-habituation component of aesthetic preference. Future work would need to disentangle the effects of simplicity versus boredom.
The authors also point out that context matters. In a museum, a viewer might expend effort to understand a difficult painting. In that specific context, the cognitive reward of understanding might outweigh the visual cost. The current study captured a more automatic, default mode of viewing.
Directions for future research include testing for causality. The current study establishes a correlation. The researchers suggest manipulating images to specifically alter their metabolic cost. Observing if this manipulation directly changes aesthetic ratings would strengthen the causal link.
They also propose investigating other sensory modalities. It is possible that the auditory system operates on a similar budget. If so, “easy listening” might be the auditory equivalent of the visual efficiency observed here.
In summary, this research provides physical evidence for a biological basis of beauty. It suggests that our aesthetic sense is not merely a product of culture or personal taste. It is also a practical mechanism for energy management. We are drawn to visual experiences that are kind to our metabolic budget.
The study, “Less is more: Aesthetic liking is inversely related to metabolic expense by the visual system,” was authored by Yikai Tang, William A. Cunningham, and Dirk B. Walther.
