There’s a rumor around campus at Emory University in Atlanta that visiting professors and new students are given a warning before going down a particular hallway in the psychology building. The purpose is to prevent them from dropping their belongings and ducking underneath a table after hearing what is unmistakably a human scream, explains Emory psychologist Harold Gouzoules.
It’s not that anyone is conducting sinister experiments, he adds; it’s simply that in this hallway, he and his team of “screamologists” are using recorded screams to investigate these well-known but poorly understood vocalizations.
Gouzoules has been interested in screams since 1980, when he began studying nonhuman animal vocalizations. Loosely defined as loud, high-pitched vocalizations often associated with distress, screams are found widely across the animal kingdom: in addition to humans and other primates, many species including rabbits and even caterpillars produce some sort of scream. Little is known about the evolution and function of these acoustically complex sounds. In humans and other mammals, air is forced through the vocal folds, making them vibrate in unpredictable and chaotic ways, such that researchers have struggled to pin down whether the resulting vocalizations contain any particular information or are simply loud noises intended to startle or draw attention.
It’s a foundational feature of human behavior so it’s surprising that so many really elementary questions are not yet well understood.—David Poeppel, New York University &
Max Planck Institute for Empirical Aesthetics
During Gouzoules’s time as a postdoc at Rockefeller University several decades ago, he worked with groups of monkeys. By recording their calls and playing them back to the primates, he discovered that they didn’t simply scream out of fear. Instead, their screams, which usually occurred during fights, broadcast an individual’s identity and details about the social rank of its attacker, which served to recruit help from kin and from friends.
The monkeys were communicating about the external state of the world, Gouzoules says. “The screams were acting like simple words.”
Few researchers had ever considered studying humans in this way, so he decided to take up the challenge. “There came a point when it was clear that looking in the literature, nobody had really looked at human screams.”
David Poeppel, a neuroscientist holding positions at New York University and at the Max Planck Institute for Empirical Aesthetics in Frankfurt who has studied human screams, agrees that the subject has traditionally been overlooked. “Since it’s essentially a universal property in humans, it’s worth studying,” he says. “It’s a foundational feature of human behavior so it’s surprising that so many really elementary questions are not yet well understood.”
Through their research, Gouzoules and a few other scientists are now beginning to gain some insight regarding basic questions about the acoustics of human screams, the information they contain, and how they are perceived.
It seems clear that at least one function is to get noticed, says Sascha Frühholz, a neuroscientist who has also studied screams and other nonlinguistic vocalizations. “If you want to express yourself intensely with your voice, you do it with screams,” he says. “They have a clear intention of attracting attention.” Gouzoules agrees, and conjectures that this attention-seeking characteristic is actually the basis for the phenomenon of fans who scream in the presence of megastars, at Beatles concerts in the 1960s, for example, or more recently, at a Justin Bieber show.
Still, it can be challenging to study such an intense and emotionally charged sound experimentally. “When I was studying monkeys, we’d be armed with a tape recorder and literally be amongst the monkeys, and when they’d be involved in a fight we’d record their screams,” Gouzoules says. “You can’t do that with people though, of course. I could stand at a street corner with a tape recorder in the middle of the night and maybe every year or so, I’d get a scream, maybe not.”
Gouzoules’s PhD student Jonathan Engelberg notes that recording genuine screams could present a challenge in terms of ethics board approval. “There’s little recourse to get ethically approved, high-quality versions to use,” he says. “You can imagine studying laughter in the natural context but it would be very difficult to do that with screams.”
Instead, the team have created a “scream library,” a collection of human screams recorded both from natural settings, as captured in YouTube videos, for example, and from scripted contexts such as films and television. “There was some concern that if you used acted screams that they were not legitimate, not real,” Gouzoules says. So “we did a study in which we tested the ability of participants to tell if a scream was acted or natural. And we found that it was shocking just how bad people were”—reassuring the researchers they could use both types of source.
They’ve also asked whether human screams can convey different emotions. That is, could a fearful scream from, say, an unfortunate victim of a jack-in-the-box prank be distinguished from a scream of frustration made by an athlete?
Armed with their extensive library, the researchers recently tested 30 different screams from different emotional situations on listeners who didn’t know those contexts upon hearing the screams. For each scream, participants gave a rating on a scale of 1 to 5 for six emotions: fear, anger, frustration, pain, happiness, and surprise. The scientists also measured various acoustic parameters, such as the frequency and duration of each scream, to determine if these features could help predict the participants’ ratings.
Using data from 182 participants, collected over a two-year period, the study found that people were often able to identify the emotional context of a given scream. Participants typically perceived screams recorded in the negative contexts of anger, frustration, and pain as being similar to one another and noticeably different from screams recorded in the positive contexts of surprise and happiness—a happy scream, for example, tended to score highly on measures of surprise and happiness but low on pain, frustration, or anger.
Fearful screams, on the other hand, were perceived as a category of their own, with participants typically rating them high on fear and not any other emotion. Acoustically, participants were more likely to rate screams with higher frequencies as fearful, the researchers found.
Participants didn’t always guess the screamer’s emotion correctly, however. Screams from happy contexts, for example, were frequently mistaken for shrieks of fear, which raises the question of why screams of happiness exist in the first place. Gouzoules speculates that there’s an evolutionary explanation involving children at play; a child’s playful shrieks might familiarize their parents with the unique acoustic aspects of their screams, so that parents can recognize those sounds in other situations where there could be danger. When people are unaware of the context, as in the experiment, they may default to the fear interpretation to err on the safe side, he says.
Frühholz, who has similarly investigated the connection between screams and emotion, also found in his own research that people can categorize screams according to different emotional contexts. “The important finding of their study and of our study is that these screams are really different and people can tell the difference between them,” he says. “This indicates that these screams have different kinds of communication value.”
Gouzoules speculates that the communication of a range of emotions through screams might be tied to the human capacity for language. His conjecture is that as the cognitive foundations of language evolved, humans slowly modified their use of screams to go beyond a simple startling function or the broadcasting of information about attackers achieved by nonhuman primates, he says. “They’re often discussed in the literature as separate systems. I think it’s probably not wise to separate them so dichotomously.”
Correction (July 9): An earlier version of this story stated that Gouzoules had become involved in scream research in the 1970s. In fact, he began the work in 1980. The Scientist regrets the error.