We have all been reeling from recent news reports suggesting privacy is pretty much dead. But surely there’s one part of our lives we can keep to ourselves if we choose to do so: Our deepest, most personal emotions.
Actually, check that. Researchers from Carnegie Mellon University announced today that they have come up with a technique that can identify the emotions you’re feeling by measuring brain activity.
“Despite manifest differences between people’s psychology, different people tend to neutrally encode emotions in remarkably similar ways.”
“Despite manifest differences between people’s psychology, different people tend to neutrally encode emotions in remarkably similar ways,” Amanda Markey, a co-author of the paper, told the university’s press office. The study has just been posted on the online journal PLOS ONE.
The research team, led by Karim Kassam, scanned the brains of 10 actors borrowed from Carnegie Mellon’s drama department. In a variation on the famous method acting technique, they were asked to repeatedly enter into nine emotional states, including anger, disgust, happiness, and lust. A computer model analyzed the patterns of brain activity produced by each emotion and attempted to identify the feeling they were experiencing.
Afterwards, still in the scanner, the actors looked at a series of photos, including some that elicited feelings of disgust. The idea was to see if the computer model could identify a subsequent strong feeling when it was instantaneously aroused.
The results were, if not perfect, impressive. “Specific emotions were identified on the basis of neural activation reliably,” the researchers write. The computer model did not correctly identified the emotion in question 100 percent of the time, but it did so a rate far better than random guessing.
Analyzing the brain scans of the actors while they viewed the disgusting photos, the model picked “disgust” (one of nine emotions it could choose from) 60 percent of the time. Even when it was in error, it wasn’t far off: “disgust” was one of its first two guesses 80 percent of the time.
The computer model was most accurate at identifying happiness, and least accurate when attempting to identify envy. As the Carnegie Mellon summary puts it:
It rarely confused positive and negative emotions, suggesting that these have distinct neural signatures. And it was least likely to misidentify lust as any other emotion, suggesting that lust produces a pattern of neural activity that is distinct from all other emotional experiences.
The researchers conclude their paper with a statement that, given recent revelations, sounds a bit more ominous than they presumably intended. They see “the possibility of producing a generative model that could predict an individual’s emotional response to an arbitrary stimulus (e.g. a flag, a brand name, or a political candidate),” they write.
While that would be enormously helpful to psychological researchers, the paranoid among us can ruminate about its possible use by marketing firms, or intelligence agencies. Interrogator to subject: “Sure, you say your feelings towards your country are only positive. But let’s see what your brain scan tells us.”