Humans are notoriously bad at detecting whether a potential romantic partner is dropping subtle hints of interest. Men overestimate the odds that women are flirting with them, while women underestimate the odds that a man is flirting with them. A person’s perception and interpretation of another’s behavior is often colored by their desires and their fear of rejection. Maybe an objective observer could do better.
In a study from Stanford University, researchers were able to more accurately determine if a heterosexual pair was flirting with a “flirtation detection system,” rather than the people who actually experienced the conversation. The subjects went on a series of speed dates, and then rated whether or not they were flirting and if they thought the other person was flirting with them. After that, the researchers pored through the audio recorded during the date and created an automated system that divided up the linguistic elements of the conversation — all the vocabulary, the pauses, the inflections.
In the end, the system beat the humans. The system could tell with 71.5 percent accuracy if the women were flirting with men, the men were only correct 56.2 percent of the time. Women were closer to the course — they were 62.2 percent accurate, while the system was 69 percent accurate.
So what do we do? An article at Inverse goes on to give us specific examples of the “flirting words” the system looked for, which was different for men and women. Maybe an automated system can teach us something about how to flirt and how to detect flirtation in others. -via Pajiba
(Image credit: Ion Chibzii)