You know a scream when you hear it. Whether a shriek, a howl, or a holler, when that familiar sound reaches your ears, you know something's wrong. Screams can even be distinctive, so much so that films use particularly "good" screams over and over again. But what's in a scream, really? Why do we find them so innately alarming?
According to new research from scientists at New York University, screams occupy a unique place in the audio spectrum when it comes to normal conversation and it has to do with a quality of sound called "roughness." That unique characteristic is why we freak out when someone yells "OH MY GOD" in pain, and not when loudly and sarcastically remarking about what the hell Joe is wearing today.
"In terms of vocal communication, it seems like screams are actually unique," said Luc Arnal, one of the researchers who led the study, and who is now based at the University of Geneva. "You want alarm sounds to be unique and not share frequencies with other signals to trigger false alarms. If you have scream-like frequencies in speech, then people would be afraid all the time."
Watch more from Motherboard: Turning Lasers and Limbs Into Instruments
Roughness is a perceived characteristic of audio signals usually associated with musical dissonance: when two notes played together make you grimace because they sound so at odds. This phenomenon was attributed to the perception of roughness in the 1800s by German physicist Hermann von Hemholtz, and over the decades researchers have established that roughness is largely a result of frequency modulations in the bassy 30-150 Hz range.
What does this have to do with screams, you might be wondering? Even though drunken piano key-mashing might be terrifying in its own way, it's not quite run-inside-and-lock-the-door scary, after all. As it turns out, musical dissonance and screams have something in common: they both result from rapid modulations in that 30-150 Hz range.
The NYU researchers first recorded screams in a sound booth and analyzed their frequencies, which revealed high modulation in the roughness range when compared to normal speech. To test the effect on human subjects, they strapped people into fMRI machines and watched their brain activity while they listened to rough and non-rough sounds (musical instruments and buzzing alarms, spoken words and screams).
"That's a way to improve alarm sounds. These are the kinds of applications we're looking toward"
According to a paper published by the researchers today in Current Biology, this experiment revealed that sounds with a lot of roughness—even artificial alarms—activate a part of the amygdala associated with fear responses. The subjects were also asked to fill out a subjective questionnaire about how alarmed or afraid they felt when presented with different sounds.
The more rough a sound was, the more freaked out the patients reported themselves as feeling. According to Arnal, this information could be used to make alarms more, well, alarming.
"If you take a basic pure tone and modulate it in the roughness range, it's going to sound alarming," Arnal said. "That's a way to improve alarm sounds. These are the kinds of applications we're looking toward."
So, what makes a scream a scream? Or a "good" one that can induce terror in film after film, like the stalwart "Wilhelm Scream"? Instead of the subtle arc of intonation or the perceived emotion behind it, it appears that the real culprit is frequency modulation.