The movement of the human eye can determine its characteristics

• According to movement of the human eye can determine its features

There are many different features that are claimed, can tell about the peculiarities of the human person: the zodiac sign, handwriting, favorite flavor of ice cream and so on. But only some of them are based on scientific knowledge.

Gold standard personality test known as a model of the "Big Five". Researchers recently have brought artificial intelligence one step closer to the ability to predict four of the five personality traits of people, and for this there is no need to pass the test. A sufficient amount of information for analysis and ... the movement of the test eye.

"Big Five" personality traits includes openness, conscientiousness, extraversion, neuroticism and complaisance. In the study, the results of which were recently published in the journal FrontiersinHumanNeuroscience, TobiasLoucher from the University of South Australia, and his colleagues showed that their method of tracking eye movements to successfully predict four of these five traits, except for the opening. Instead of opening the eyes tell about one more quality, which is called "perceptual curiosity," or the tendency of man to explore new visual, sound and other sensory stimuli.

It may seem incredible, but if you think about it, the use of eye movements in order to judge a person's character traits, not so surprising method. After all, in the end, anyone who understands intuitively that the stranger, who is constantly flashing, most likely, a little neurotic, and his friend, who loves long eye contact - an extrovert. Previous studies have confirmed this. So, one of them showed that optimists tend to spend less time on unpleasant viewing medical images than pessimists, and the other - the people with the open nature tend to be considered the longest images with abstract animation. The problem, according Louchera and his colleagues, is that this research simply shows that a certain character trait is correlated with certain eye movements. However, if the eye movement can actually talk about the traits of human nature? This opens up a wide range of opportunities, including in the field of robotics. When robots learn to understand what kind of person in front of them, just by looking into his eyes, they can adjust their behavior on the fly and create for him, as for the user a more comfortable and pleasant atmosphere.

For their study, researchers offered 42 students to complete an individual questionnaire to assess their "Big Five" personality traits. Then they were asked to wear special glasses to track eye movements while they performed a task designed to simulate the real scenario of daily life: walking across campus or shopped in the store. The researchers then download the data into a machine learning algorithm, which analyzed the numbers and found a correlation between personality traits and characteristics of participants in the experiment the motion of their eyes.

Points track and record a wide range of eye movements, including the direction of gaze, pupil size, number and type of skakkad, rapid eye movement between the points of fixation, which we unwittingly commit when entering the eye contact with another person. The algorithm has shown that certain types of personalities may be characterized by various types of eye movements.

As for neuroticism, are all a matter of flashing frequency. Integrity is characterized by speed of saccades. Complaisance could be predicted on the basis of how long the person was looking in the lower right corner of the image, while extraversion was manifested in pupil size and the number of small saccades. Perceptual curiosity, as it turned out, due to the size of the pupils of the person. Of course, all this information is simplified - the algorithm to work with a complex combination of many factors, to make their forecasts. The accuracy of these predictions can hardly be called sensational: only 7-15 percent better random. But it was only a small study, and Louchera there is hope for the future. "For machine learning typically require thousands or even millions of data sets on the basis of which it is possible to forecast with reasonable accuracy, - he said in an interview with NewScientist edition. - We hope in the future to achieve much better results. "

However, when success is achieved, it comes time to think about the ethical side of research. Given the recent sad experience, when disgraced the whole world CambridgeAnalytica analyst firm to use the data in order to attack people by political advertisements, based on their personal characteristics, the exact nature of the forecast made by artificial intelligence can discover even more dangerous Pandora's box.