In their analysis of the acoustical makeup of sounds, the researchers discovered that screams ranked highly on "roughness," which they defined as "big swings in loudness and a very unpleasant perception." When they smoothed out the rough signals in the screams, people said they were not as scary. The rough ones triggered increased activity in the amygdala, "a brain structure that helps a person detect threats," according to functional MRI brain scans. In a physiological context, that could explain why people find screams so alarming.
Human screams have "a unique acoustic property found to activate not just the auditory brain but also the brain's fear circuitry," according to the researchers. According to Poeppel, "If you ask a person on the street why screams have a particular effect, the person will say that scream are loud or have a higher pitch, but other sounds are loud and high pitched, so you'd want a scream to be genuinely useful in a communicative context."
Poeppel's post doc Luc Arnal, who is now at the University of Geneva, led a series of studies to analyze the properties of screams, using recordings from YouTube videos, popular films and volunteer screamers. The research team plotted the sound waves in a way that reflects the firing of auditory neurons and saw that screams activate a range of acoustic information that scientists had not thought important for communication (http://phys.org/news/2015-07-occupy-privileged-acoustic-niche-biological.html#jCp).
Screams and alarms differ from other sounds because of a property called roughness, which is the speed at which a sound changes in loudness. While normal speech patterns only have slight differences in loudness, screams can modulate very fast. Screams with the most roughness are considered to be the most terrifying, which activates the fear response in the amygdala.
The researchers hope to use their findings to help acoustical engineers to design better alarm signals, to analyze the roughness of particular human screams and to apply their work to animal screams to see how the trait works in other species. Poeppel concludes, "Screaming really works. It is one of the earliest sounds that everyone makes, so we thought maybe this is a way to gain some interesting insights as to what brains have in common with respect to vocalization."