Mathilde Fontez, editor-in-chief of the scientific magazine Epsiloon, evokes today a recent discovery in art and science, a discovery in psychology. One might think that our artistic preferences are subjective. But no.
franceinfo: Researchers show that we all have the same taste in painting?
Mathilde Fontez: For the colors in any case, we all agree! This is proven by a study carried out by Japanese psychologists. They simply showed master paintings to volunteers, all kinds of paintings, cubist paintings, figurative paintings. And they changed their color shades, or their layout, before asking their audience to rate them.
And what are the original colors that were preferred?
Yes: the vast majority of volunteers who played the game preferred the original color combinations. And this, even if they did not know the painting. And whatever their culture of origin: the study was carried out on a sample of Japanese and Portuguese.
Conclusion of the researchers: there is therefore a form of universality in our tastes. And that joins the conclusions of another study, quite dizzying, which was carried out at the same time, by a team from Caltech University, in California: these researchers have clearly shown that a computer program is able to predict our tastes in art!
Is it still artificial intelligence?
Not just artificial intelligence. The researchers used two types of programs: a neural network, an artificial intelligence therefore, which trained itself to find criteria by analyzing master paintings to judge them. And a classic program, that is to say in which they entered parameters: contrast, saturation, hue. Even high-level features that require human judgment. For example, whether the painting is “dynamic” or “static”.
Like the Japanese, they put everything in their sample of works of art: Picasso, Monet, and they compared the results of the computer with those of 1500 volunteers, recruited for the occasion. Conclusion here again: the programs were right. They were able to predict the tastes of the volunteers: by analyzing a painting, they were able to tell if it was going to be liked, or not.
In fact, there is nothing subjective about art?
So researchers remain cautious. They don’t want to go into caricature. These are average results, and there remains an element of subjectivity. But yes, these studies show that we evaluate the quality of a work of art on very specific criteria, on which we all agree: as if we had all been trained to say to ourselves: this color with this color, it’s beautiful, or this image structure doesn’t work. Art is universal.