Scientists at the University of Zurich have discovered that humans can more efficiently process and categorize screams that evoke positive or nonalarming emotions than those that communicate aggression or fear. 

The findings, published Tuesday in PLOS Biology, contradict prior assumptions that humans are better equipped to identify screams of fear and rage in order to recognize and avoid environmental dangers. The team isolated and classified six types of screams representing different emotional extremes, including overwhelming pleasure, sadness and anger. 

Throughout the experiment, "We always had the Beatles in mind," Sascha Frühholz, a University of Zurich neurologist and the study's lead author, told The Academic Times. "The Beatles always had contact with groupies, and you have all this footage with thousands of people screaming out of joy."

This was a reminder, Frühholz said, that humans have a much more complicated relationship with screams than other animals do. Across most of the animal kingdom — and especially in primates and birds — screams have limited communicative power: They're used to express either fear or aggression. But as the University of Zurich team confirmed, human screams have a more robust sound pattern that can evoke vastly different emotional extremes, both positive and negative.

The team's research built on evidence from the field of affective neuroscience, the study of how the brain decodes facial expressions, voices and other emotional cues. Recent studies have indicated that humans are able to detect fear, more than any other emotion, in other bodily indicators, such as sweat. 

In order to gather samples for the study, researchers asked 12 volunteers to record a total of 420 screams. Each person was asked to scream under different imagined scenarios. To prompt a joyful yelp, for instance, the scientists asked participants to imagine that their favorite sports team had just won a major game. The team avoided recruiting actors or vocal experts in order to capture a more unrefined dataset that represented the general population's screaming tendencies, rather than those of a specialized subset. 

Because screams are often spontaneous — prompted by a startling change in one's surroundings — screaming outside of an appropriate context might feel socially uncomfortable to some individuals. To counteract this problem, the researchers designed a private environment for participants to record their screams without fear of judgment. "We have a sound booth that is completely isolated," Frühholz explained. "We told them, 'Okay, you go inside, we close the door, we cannot hear you, we cannot see you, there's only a microphone, and [you can] scream as much as possible.'"

Then the researchers asked several sets of volunteers to listen to and categorize sets of 84 of the original screams across all six categories, plus the additional neutral category. Participants were able to classify screams of joy faster than all types of alarm-inducing screams. They also detected screams of pleasure faster than screams of pain. 

A final group of participants listened to screams while undergoing functional magnetic resonance imaging scans. The researchers predicted that screams of danger would activate the limbic system — a brain circuit responsible for detecting danger — to a greater extent than the nonalarm screams. Instead, they noticed opposite results, with the amygdala, in particular, showing greater sensitivity to positive screams.

Volunteers were more likely to wrongfully guess that screams were alarming compared with the nonalarming or neutral groups. This trend did somewhat match investigators' expectations, as it would be evolutionarily advantageous to overidentify troubling screams as a safety precaution in uncertain situations. 

All of the screams that the researchers studied, no matter the type of emotion they evoked, tended to fall within two frequency bands: a low range of around 60 hertz and a high level of around 160 hertz. Humans subjectively describe noises that match those frequencies as having a particularly "rough" quality, which aligns with the kinds of shrill, croaky and harsh sounds that are typically associated with shrieking, shouting, squawking or yawping. 

Despite their shared frequency ranges, human hollers evoke vastly different tonal qualities. When a machine-learning algorithm was used to analyze 88 acoustic features in each of the screams, the researchers found they could successfully identify shrieks across six categories, plus an additional control category — an intense vocalization that the researchers described as a "neutral scream" — with nearly 80% accuracy. 

There could be an environmental bias embedded in the findings, the researchers cautioned, since the study was conducted in Switzerland, a relatively safe area, far removed from the predatory dangers that would have frequently plagued early humans. Future studies could include participants who live in environments with more direct exposure to wildlife and weather to assess if their brains are more finely tuned to screams that evoke immediate danger. But while humans might not be as sensitive to alarm-inducing vocal cues in the modern age, their scream detection could instead be calibrated for navigating social interactions.

"We try to avoid many of the threats or dangers that animals would encounter in their environments," Frühholz said. "Most common environments of humans are safe. And my impression is that in these environments, positive emotions have much more importance because we use these emotions much more to regulate our interactions."

The study "Neurocognitive processing efficiency for discriminating human non-alarm rather than alarm scream calls" published April 13 in PLOS Biology, was authored by Joris Dietziker, Matthias Staib and Wiebke Trost, University of Zurich; and Sascha Frühholz, University of Zurich and University of Oslo.