Art that reflects us is more enjoyable

What makes certain visual artworks more appealing to us has been debated for centuries. When considering paintings, particular features certainly play a role; for example, people tend to prefer more natural scenes, and more ordered images, while levels of brightness and colour choice can also affect aesthetic appeal. But given the wide array of responses a single artwork can provoke, it's clear that what viewers bring to the equation is a vital part of the puzzle. 

Recent investigations from Edward A. Vessel at the Max Planck Institute for Empirical Aesthetics and colleagues suggest that the personal relevance of an artwork's subject matter is important to shaping viewer responses. The findings suggest that that whether or not someone finds a work of art beautiful and moving has more to do with how well the subject matter resonates with their sense of who they are and their world-view than any particular features of the image itself.  

In an initial experiment, 33 participants viewed 148 images of paintings. These paintings were chosen to reflect a variety of time periods, styles, and genres from the Americas, Europe, and Asia. They were also paintings that had not been widely reproduced, minimising the chance that the participants would have seen them before. Each painting was rated for naturalness and disorder by an independent set of raters.

The participants gave each artwork two ratings: one for how much it 'moved' them (this was the measure of aesthetic appeal in this experiment), and another for 'self-relevance'. 'Self-relevance' was defined for participants as the extent to which something in the image related to them, their experiences or their identity – the things and events that define them as a person.

The results showed that the participants had widely varying views on the self-relevance and aesthetic appeal of each painting. However, rating a painting as highly self-relevant was strongly linked to finding it more aesthetically appealing. In fact, self-relevance predicted about 28% of the variance in scores for aesthetic appeal, while levels of naturalness and disorder accounted for only 8% of this variance. 

The number of participants in this initial study was small, and it explored just one aspect of aesthetic appeal. So, in order to extend their findings, the team then recruited 208 participants to view and rate a subset of 42 artworks for beauty, as well as how moving they were, and self-relevance. As with the first study, more self-relevant artworks were rated as more moving. Participants also tended to rate highly self-relevant images as more beautiful, suggesting that the more personally relevant a participant considered the painting to be, the more aesthetically appealing they found it.

In a final experiment, the researchers aimed to find out whether greater self-relevance caused, not just correlated with, higher aestheticism ratings. They gave a fresh group of 45 participants a questionnaire that asked about their life experiences, aspects of their identity, preferences, and common activities. They then used this information to create personally relevant synthetic artworks for each participant. 

The team did this using a neural network-based technique called 'style transfer', in which photos are integrated and modified to look like an artwork painted in a specific style. For example, for a participant who had reported a memorable holiday that started in Helsinki, the team fed the neural network photos of a landmark Helsinki building, plus an image of a colourful oil painting by Bob Thompson. The neural network then generated an artwork of that building and its surroundings in the style of that painting. 

Each participant viewed four sets of 20 images: self-relevant generated artworks, created on the basis of their questionnaire responses; artworks that had been generated to be highly relevant to one other participant; a control set of generated artworks (not related to anyone's questionnaire responses); and real paintings, taken from the set used in the first studies. The participants rated each image for how moving they found it and for self-relevance.

The team's analysis showed that participants found their self-relevant, artificially-generated artworks more aesthetically pleasing than those generated for another participant. On average, they also slightly preferred their personally generated artworks to genuine paintings. Further analyses revealed that content that related to an individual's 'self-construct' — their past experiences and their identity — was mostly strongly linked to higher aestheticism ratings. 

Real artworks, while getting relatively low ratings for self-relevance, received higher aesthetic appeal ratings as a group than the generated artworks. As the team notes, this indicates that there was something about genuine paintings that the style transfer technique did not capture. Quite what that is was unclear. 

The study itself does have a few limitations. For example, only one of the experiments asked for ratings of beauty, which is commonly viewed as central to the aesthetic experience. Overall, though, the new work does suggest that when it comes to artwork, the old adage 'Beauty is in the eye of the beholder' is true — and that referencing the beholder is key. 

Read the paper in full: https://doi.org/10.1177/09567976231188107