Mixing Memory

This Just In: Monkeys Not Mozart Fans

Over the years I’d heard that, lurking in the basements of psychology departments at various universities throughout the world, there are psychologists studying music cognition, but until the publication of a special issue of the journal Cognition, I hadn’t really paid any attention to them. That issue (especially Ray Jackendoff’s “The capacity for music: What is it, and what’s special about it?“) got me interested in the topic, though, not so much because it can tell us much about cognition in general, but because the experiments are usually pretty cool. So, I’ve been on the lookout for music cognition papers in the big journals, and last week, I came across a fun one by Josh McDermott and Marc Hauser (of the lawfirm Hauser, Chomsky, and Fitch) that is currently in press at Cognition, and I thought I’d tell you about.

Where to start? OK, so one of the interesting (well, kinda) questions in music cognition is, where does music our ability to produce it, enjoy it, and understand it – come from? Are aspects of our “music capacity” unique to humans, and if so, are they unique to music? The best way to approach these pressing issues is to look to our close relatives, nonhuman primates, to see whether they dig music. If they do, then it’s likely that our music capacity is built up of capacities that are used for things other than music, and if they don’t, then rock ‘n’ roll may be a uniquely human ability, with music-specific cognitive components.

i-0762e10be11d4ccb66f5996ebb0389fb-cotton - top tamarin.jpg i-cafe6b31db05750b9aa2fd6d33c94a5c-marmoset.jpg Of course, you can’t run out and test the uniqueness of every aspect of our musical capacity in one set of expeirments, so McDermott and Hauser focused on one: tempo. Humans tend to find fast tempos more arousing than slow ones (go figure), though there’s no evidence that they generally prefer one over the other. What about nonhuman primates, though? Do they like fast music or slow music? I know I’ve always wondered.

In order to determine the preferences of these cute little primates, McDermott and Hauser conducted several experiments using the same methodology. I’ll let them describe it for you:

The apparatus [a V-shaped maze, Fig. 1] and procedure were identical to that of McDermott and Hauser (2004). We placed concealed speakers (PolkAudio Atrium 45p) at the end of each branch of the maze; each speaker and branch was paired with a different sound. Subjects were initially placed at the entrance to the maze. The experimenter then left the room and raised the door to the maze by means of a pulley, allowing the animal to enter. We placed small pieces of food in both branches of the maze, equidistant from the entrance, to entice them to enter; the animals always ate both pieces of food. When the subject moved into a branch for the first time, the experimenter started the playback. The stimulus for a particular side played continuously as long as the animal was on that side, and switched as soon as it switched sides. The animal’s position in the maze thus determined which sound it heard. Testing continued for 5 minutes. After a certain number of sessions with one sound on the left side and another on the right, generally separated by a full day, the sound-side pairing was reversed, and the animals were run for the same number of additional sessions. Data before and after the reversal were combined to distinguish stimulus preferences from side biases. The number of sessions for a single animal in a single experiment was usually 6 or 8. It was always fixed in advance of the experiment. (p. 3)

As a measure of preference, they measured the amount of time each type of music played during a trial. Since the primates determined how long a type of music played in a particular session by either staying in one branch of the maze or moving into the other, this measure shows how much time they decided to listen to each type of music. If they like one type better than the other, they should tend to spend more time in the branch where that type is playing than in the other branch.


In their first experiment, McDermott and Hauser pitted a slow Russian lullaby played on the flute against a fast German techno song (Alec Empire’s “Nobody Gets Out Alive”). The percentage of time tamarins and marmosets spent in the branches playing the Russian lullaby indicates that they preferred it to the techo song (see their Figure 2 below).


Just showing that the tamarins and marmosets prefer the lullaby doesn’t suffice to show that they prefer slower tempos, however. It could be that they just like flute music, whether it’s fast or slow, better than techno. So, to determine which of the differences between the two musical pieces was responsible for the primates’ preference, they then conducted two experiments that isolated particular features of the pieces: their tempo and their attack velocity (the speed of the onset of each sound or note – slow for the lullaby, fast for the techno song). Both the tamarins and marmosets preferred the slower tempo of the Russian lullaby, but showed no preference for either attack velocity, indicating that the tempo of the pieces likely determined which they preferred.

From these first three experiments, it appears that nonhuman primates prefer slow music to fast music. McDermott and Hauser argue that this could be because the alarm calls of both species are fast-paced. It can’t be pleasant to listen to music that reminds you of the calls you hear when something that wants to eat you is in the neighborhood. but after the first three expeirments, the question still remians, do tamarins and marmosets like music at all? I mean, come on, if I were forced to choose between a Russian lullaby and German techno, I’d pick the lullaby (especially if the techno made me feel like I was about to be somebody’s dinner), but that wouldn’t mean that I’d like the lullaby. So, in their fourth experiment, McDermott and Hauser sought to determine whether tamarins and marmosets liked music by pitting music against silence. In this expeirment, humans were also tested, though apparently because pilot studies showed that humans wouldn’t fit into the maze, the procedure was slightly different for the non-nonhuman primates. They were put in a room divided in half by a strip of tape. On each side of the room, there was a speaker, and music was played from the speaker on the side of the room that the participants were standing. The humans, unlike the nonhuman primates, were not given food as a reward, but instead were simply told that they had to stay in the room for 5 minutes. All three types of participants (marmosets, tamarins, and college students) participated in three conditions, the flute-played Russian lullaby vs. silence, a sung German lullaby vs. silence, and a string concerto by Mozart (K458) vs. silence. In each condition, the humans preferred music to silence, while nonhuman primates preferred silence to music. Tamarins and marmosets, it turns out, don’t like human music.

What does all this mean for the study of the origins of our music capacity? McDermott and Hauser make the following claim:

Taken as a whole, however, the body of work on music perception in nonhuman primates suggests fundamental diverences in the way they respond to musical stimuli compared to humans. Our preferences for certain kinds of music may reflect a unique evolutionary history of selection on perceptual mechanisms linked up with our emotional and motivational systems.

That’s just a fancy way of saying we’ve developed the ability to produce sounds that are pleasing to us. In other words, it’s a fancy way of saying very little. In fact, I’d be surprised if nonhuman primates, with their own sound-emotion associations, liked music designed to elicit human sound-emotion associations. Perhaps if we knew more about the types of sounds that marmosets find pleasing, though, we could produce symphonies that marmosets would love.


  1. #1 Todd
    September 9, 2006

    Welcome back Chris. Although your great blog was sorely missed, you don’t need to apologize for being away. Everyone’s entitled to a vacation–even a cognitive superhero like yourself.

    I don’t know much about these types of studies, but it seems like they started backwards. Wouldn’t it be easier to use a synthesizer to create a beat (with varying tempos) and try different combinations of instruments, than to start with a full composition?

  2. #2 Andy
    September 10, 2006

    I just want to comment on the statement “That issue [...] got me interested in the topic, though, not so much because it can tell us much about cognition in general, but because the experiments are usually pretty cool.”

    This sort of argument gets used a lot: “x doesn’t tell us much about cognition in general“. When I read that I wonder what sorts of things do get at cognition in general. My position is that to get to general descriptions you’ve got fight with lots of specifics. Then generalisation can occur by seeing what the many specifics have in common.

  3. #3 Chris
    September 10, 2006

    Andy, in a field that’s as young as cognitivfe science, or cognitive psychology, there tends to be an emphasis on wider relevance. When I say “cognition in general,” what I really mean is “anything but how we process music.” Sure, processing music is a somewhat interesting topic, but because it doesn’t, as far as I can tell, teach us anything about anything else, I have trouble getting myself to find it interesting. That’s not to say that people shouldn’t study it, of course, just that I’m not going to. Of course, when studies like the one I talked about in this post get published, I am a little confused. I mean, is it important, even for the study of musical cognition, to find out whether nonhuman animals like fast tempos? The authors of the paper make no real argument for the importance of the study, and I can’t really find one.

  4. #4 Tim Byron
    September 10, 2006

    Chris, the interesting thing about music to non-musician psychologists is that it pretty much strikes at the heart of the debate in psychology between modularity and general processing principles. In particular, the reason that people like Hauser, Fitch are interested in music is that it’s becoming rapidly more obvious that people researching language-related things can’t really afford to ignore music. Music and language may or may not be related neurally, as they’re similar in a number of ways. If music ends up being mostly ‘done’ by general processing principles, then it makes it more likely that language is done by general processing principles rather than specialised modules. Whereas if the cognition of music is a separate module to the cognition of language, even though they share many qualities in similar, it says interesting things about modularity in the brain.

    That’s the wider relevance of study in this field, I think. This particular study is part of an attempt to argue that preference for music is evolved, even if particular aspects of music just rely on basic perceptual principles. I’m not entirely sympathetic to the approach, but it’s one of dozens of recent articles on the topic of the evolution of music (see pretty much the rest of that issue of Cognition, or an issue of Music Perception from earlier this year), and in that context is interesting.

    I do agree with Todd to some extent that they have started backwards. The issue, though, is that beats like that are only vaguely musical. The tamarins could respond to simple beats based on environmental factors, and not respond to music in general – probably not the best example, but simple beats might remind them of the footsteps of the experimenter or something. And I suspect such stimuli aren’t as sexy as Mozart for the purposes of comparison – how many humans listen to simple beats of varying tempos for pleasure?


  5. #5 Mark
    September 12, 2006

    I know that in earlier studies with pretty much the same set-up they did use simpler stimuli, but they encountered the issue Tim mentioned… that is if any actual humans would consider the stimuli pleasurable or musical. If I remember correctly, they studied consonance and dissonance using filtered organ-like chord sounds.