TEDxUCLA 2012: Open

The sound of fear


About Daniel

Daniel T. Blumstein is a Professor and Chair of the Department of Ecology & Evolutionary Biology, at UCLA, and a Professor in UCLA’s Institute of the Environment & Sustainability. He has studied animal behavior and conservation biology in Australia, Canada, the Caribbean, Germany, Kenya, New Zealand, Pakistan, Russia, and the United States.


Have you ever been afraid of a sound? Has a sound ever scared you? What about that shower scene in Psycho?

I’m interested in studying and understanding what fear sounds like, and I sort of stumbled onto that accidentally.

You see, I study marmots, large ground squirrels found throughout the northern hemisphere. And I spend an inordinate amount of time going out in meadows and listening to what they say, and while I’m there I listen to what other animals say as well. And one day when I was holding a baby that emerged from its natal burrow a few days before gently holding it, I heard this horrible sound and I almost dropped it. And that was really striking because I, normally I’m not so emotionally aroused by sounds I hear that animals produce. And I really wondered, “What’s going on with this? Why did I have that response?”

And I started reading around and looking and I started finding that if you look at screams and really sort of high-urgency vocalizations that other animals make they can, they’re very remarkably convergent. They contain things that are common and shared among all of them.

One of those things is noise. All this dark stuff on the spectrogram is noise. There are other what are called nonlinearities in these sounds as well. And I started reading more and learning more about how these are produced and they’re produced when vocal systems are overblown, kind of like when you turn up your stereo, it gets louder and louder and louder, and then at some point it gets distorted. Same thing when you blow a horn. I mean, I can’t blow a horn, but if I could, I could blow it and it would get louder and louder and louder and then at some point it would break, it would become noisy, there would be rapid frequency changes. And these nonlinearities, this noise, is a sign that a system, a system is being overblown and that same thing happens when you scream.

So I started reading around and looking for other cases where maybe this was associated with urgency. In meerkats, as meerkats get more and more scared, as the urgency increases in a situation where they’re vocalizing, these sounds getting noisier. And we can see that in two types of alarm calls, but also in recruitment calls.

A variety of other animals have the crescendo of what they’re trying to communicate. Chimpanzee pant-hoots, overblowing that vocalization system, making it noisy, raspy. We all know dogs sound like when they get unhappy, and we really know dogs sound like when they’re really unhappy because those barks sound different, they sound raspier to us and rougher. Macaques screams, piglet squeals and screams, all chock full of noise and nonlinearities.

And I wondered, “Well, what happens if you add noise to a normal vocalization? An alarm vocalization, say? Can it be more evocative?” And I found that when you play back alarm calls to marmots, they look and they stop foraging. But if you add noise to those alarm calls, they suppress their foraging even more.

Now, this isn’t just a story about marmots, although I have an inordinate fondness for marmots. This is something that I think is much more universal. So some students and I did an experiment with Caribbean grackles and we synthesized sounds and we put we played back pure tones to these animals and then we added specific forms of these nonlinearities and noise to these pure tones. And what we found was the addition of noise decreased relaxed behavior in Caribbean grackles.

So this raises a question: “Is the sound of fear noisy? Is the sound of fear or non-linear?”

I was giving a public talk a number of years ago and I said, “You know, I bet this, if this is true, I bet it applies to movies and music.” And at a break, a now colleague and friend Peter Kay came up to me, he’s a film score composer and he was a musician. And he was studying how emotions are communicated in music and film. And he said, “I bet you’re right.” I said, “You want to collaborate?” He’s like, “Yeah.” So I said, “Let’s get a UCLA undergraduate to help us with this.”

So we went out and got a UCLA undergraduate and we looked at a whole bunch of movies that were reported or ranked to be, you know, the best this: the best adventure, the best horror, the best sad dramatic, the best war. And our hypothesis was, will we see or what we wanted to know was, will we see different sorts of noise and nonlinearities specifically associated with these different genres?

I think you’re all familiar with this one. That first scream is noisy. You can see it in the spectrogram, you hear it and it got your attention.

When we went out and tabulated and looked at short little sort of iconographic scenes from iconographic films, we found that sad films suppressed noisy sound effects. There were fewer noisy sound effects in sad films than would be expected by chance. And horror films use noisy female screams (duh) while sad films suppressed them.

So this is great. This is correlative. And it suggests that people who make a lot of money in this area do so by sort of tapping into our inner marmot. But it suggests that. It doesn’t experimentally test it.

So I was talking with a colleague at UCLA, Greg Bryant, who’s in the Department of Communication Studies, and Greg studies humans, and he’s also a musician, and he also studies how emotions are communicated in voice. And I said, “Greg, I wonder if we can make an experiment where we test this experimentally.” He’s like, “Yeah.”

So Greg and Peter made music. And we basically made 10-second clips of music. This is an example of one of them, and it’s sort of Muzak-y music. And this was sort of our control treatment. We had a lot of these little clips. But the test treatments started with five seconds of the muzak-y Muzak. That distortion is noise. That noise is distortion. And we wanted to know how people felt about these various sounds.

And what we found was when we broadcast these to people and ask them what they thought about them, we asked them how arousing were these and how, what was the valence. How positive or negative was this sound?

And we found that arousal was enhanced by the addition of noise and that valence was reduced by the addition of noise. In other words, these sounds became more negative when noise was put in.

So I’d like to suggest that by listening to my inner marmot, I’ve gained some appreciation of why we respond emotionally the way we do to certain sorts of sounds. I think evolution is a lens through which we can understand our own behavior, and I think that by looking left — far left — to what other species do, we can gain some universal insights into things like why we’re afraid at movies and why certain films and soundtracks are particularly evocative.

And I’d like to leave you with a question. What does the sound of joy sound like? What does the sound of sadness sound like? Thank you very much. (applause)