The eyes really are a听window to the soul. The way they move can reveal your personality type 鈥� a finding that could help robots better听understand and interact with humans.
Psychologists have long believed that personality influences the way we visually take in the world. Curious people tend to look around more and open-minded people gaze longer at abstract images, for example.
狈辞飞,听听at the University of South Australia and his colleagues have used machine learning 鈥� a type of artificial intelligence 鈥� to study the relationship between eye movements and personality more closely.
Advertisement
They asked 42 students to wear eye-tracking smart glasses while they walked around campus and visited a shop 鈥� an activity designed to mimic everyday life. They also asked them to fill out questionnaires that rated them on the 鈥渂ig five鈥� personality traits: neuroticism, extroversion, agreeableness, conscientiousness, and openness.
Their machine learning algorithm found that certain patterns of eye movements were more common among different personalities. For example, neurotic people tended to blink faster, open-minded people had bigger side-to-side eye movements, and conscientious people had greater fluctuations in their pupil size.
The reasons why are unclear, but this doesn鈥檛 matter if the goal is to teach robots to be more socially aware, says Loetscher. 鈥淭hey need to know which eye movements relate to which personality types, but they don鈥檛 need to know why,鈥� he says.
Future research may find that brain chemistry explains these patterns, says听听at the University of Melbourne. There is already evidence that neurotransmitters like dopamine and noradrenaline influence personality and can affect eye movements like blink frequency and pupil dilation, she says.
More to learn
At this stage, the machine learning algorithm only has modest predictive power. It is 7 to 15 per cent better than random chance at predicting neuroticism, extroversion, agreeableness, and conscientiousness, and no better at predicting openness.
However, this is probably because it is based on eye-tracking data from only 42 people, says Loetscher. 鈥淢achine learning usually requires thousands or millions of datasets to make highly accurate predictions,鈥� he says. 鈥淲e expect it to get much better.鈥�
The predictive power could potentially be enhanced by training the algorithm to interpret speech patterns at the same time, says听Carter. Research has found links between language and personality 鈥� for example, extroverts tend to speak faster, pause less and use more informal language than introverts.
The potential applications are wide-ranging, says听听at the Max Planck Institute for Informatics in Germany, who co-authored the study. 鈥淚magine, for example, that smartphones could know and adapt to your personality,鈥� he says. That would allow them to support you and give you personalised information at a completely new level, he says.
Other possible applications might be in听robot companions for older people, self-driving cars, and video games, says Bulling. 鈥淭hese could be so much more natural and personalised if the personality of the user was known,鈥� he says.
However, the technology will need to be regulated to make sure it isn鈥檛 misused by marketers, says Carter. The recent Cambridge Analytica scandal revealed that AI was being听used to听mine Facebook data听without users鈥� consent and target political ads based on personality. 鈥淚f the same information could be gained from eye recordings or speech frequency then it could easily be recorded and used without people鈥檚 knowledge,鈥� she says.
