When something frightens you, your body is moving before you had time to think. The amygdala flags the danger and signals the hypothalamus, which switches on the sympathetic nervous system, and adrenaline sends your heart rate, blood pressure, and breathing climbing, a chain that, according to Harvard Health, starts “even before the brain’s visual centers have had a chance to fully process what is happening.” That cascade is the same one behind why we gasp and why we freeze. The scream is the part of it aimed at somebody else, and most of what researchers know about it concerns what the sound does to the people who hear it.
The Science of Screams
Ask around, and you’ll hear that a scream is just loud and high-pitched, which NYU’s David Poeppel said in 2015 misses the point: “there’s lots of stuff that’s loud, and there’s lots of stuff that’s high pitched, so you’d want a scream to be genuinely useful in a communicative context.” Poeppel and Luc Arnal of the University of Geneva led a team that pulled screams from YouTube videos, popular films, and volunteer screamers, since, as their publisher Cell Press put it, “there is no repository of human screams.” Their study in Current Biology found the signature in a quality called roughness. Ordinary speech carries its information in slow swells of loudness, under 20 per second, while a scream’s loudness flutters far faster, at roughly 30 to 150 times per second, a band that human communication otherwise leaves almost empty. That’s a rate of fluttering, not a pitch.
The empty band is the point: because normal talk doesn’t live there, a scream can’t be mistaken for conversation, which the authors called “a prerequisite to avoid false alarms.” Listeners rated rougher screams as scarier and reacted to them faster, and they pinpointed screams on their left or right faster and more accurately than ordinary voices. Filtering the roughness out of a scream made it less frightening, and adding roughness to a neutral “ah” made that scarier too, though that was the study’s weakest result. When 16 people listened inside an fMRI scanner, unpleasant sounds drew larger responses in the amygdala, which turned out to be sensitive to fluttering in the roughness range specifically.
Screams weren’t the only sounds in that range. Poeppel said that “alarm signals (car alarms, house alarms, etc.) also activate the range set aside for screams.” Dissonant musical intervals are rough too, and the amygdala result above pooled them with the screams, unpleasant sounds against neutral ones. Arnal saw a lesson for whoever designs alarms: “The same way a bad smell is added to natural gas to make it easily detectable; adding roughness to alarm sounds may improve and accelerate their processing.”
The Many Meanings (and Misunderstandings) of Human Screams
What a scream is for is clearer in other animals. “Animal screams occur almost always in the context of a fight or in response to a predator,” Emory psychologist Harold Gouzoules said in 2019, and in nonhuman primates his lab ties screams to what researchers call “agonistic recruitment”: calling allies into a fight. Gouzoules has also described how the sound gets made: “Screams require a lot of vocal force and cause the vocal folds to vibrate in a chaotic, inconsistent way.”
People are another story. “Human screams happen in a much broader array of contexts, which makes them much more interesting,” Gouzoules said. In a 2021 study at the University of Zurich, Sascha Frühholz’s team had 12 volunteers with no acting experience produce screams of pain, anger, fear, pleasure, sadness and joy. With some exceptions, listeners picked out the alarm screams the worst and responded to them the slowest, and the positive screams drew more amygdala activity than the negative ones. “We were surprised by the fact that listeners responded more quickly and accurately, and with a higher neural sensitivity, to non-alarming and positive scream calls than to alarming screams,” Frühholz said. That doesn’t undo the Geneva result, which compared screams to ordinary voices rather than to each other, but it does suggest fear isn’t a scream’s whole job, and Frühholz has said as much: “It’s highly possible that only humans scream to signal positive emotions like great joy or pleasure.”
The ear doesn’t always sort them correctly either. In a 2021 study from Gouzoules’s lab, screams from happy situations “more often yielded higher ratings of fear.” None of the studies measured what happens inside a frightened person in the second before the scream comes out, only what the scream does to everyone within earshot, and on that point Poeppel was blunt: “Screaming really works.”