People generally admit that attractive faces are those which have particular physical features such as blonde hair or big eyes. However, we still have individual preferences and standards of beauty. Researchers wanted to investigate if computers could be used to identify people’s criteria for attractiveness, so they conducted an experiment to test this. The principles were similar to those of dating apps, in the sense that people looked at portraits to find an attractive one, but in this case, their brain response was recorded.
Preferences detected at the brain level
Researchers were interested in studying the brain of people exposed to attractive faces, since attractiveness is rather a complex concept, as it depends on factors of psychological and cultural nature of the individual. For this reason, physical attraction depends both on the looks and the preferences of a person.
They discovered that computers can identify physical features that people find attractive and, based on their personal preferences, create a new image that matched their taste. They used artificial intelligence to detect the participants’ brain responses who looked at different artificial faces. In the end, the computer produced images of attractive faces customized for each individual.
Experiment with attractive faces
The researchers used artificial intelligence to create artificial faces that were shown to 30 participants. The subjects paid more attention to those faces that were appealing to them, and their brain activity was recorded via electroencephalography (EEG). Then, each individual’s EEG data, namely the artificial faces they liked, was interpreted by a computer interface, and this interface together with a generative neural network produced a new artificial face based on those features that the participant liked. The new images of attractive faces were presented to the participants and matched their taste with over 80% accuracy.
Further development of the study
The study proves that artificial intelligence is capable of producing customized images of artificial faces. The researchers claim that further developments of the study could investigate other individual preferences such as decision-making to identify the different principles of each individual making decisions.
If you are interested in reading more about the study, you can find the information here.