Entrainment in non-human species?
Snowball (or should I say Snowball™) charmed millions of people with its cool dance to Backstreet Boys’ hit song “Everybody”. Showball’s performance also caught the eye of rhythm researchers, as it had been assumed that such sensorimotor synchronisation (SMS) is a uniquely human trait. (I wrote about Snowball earlier.) Last october, a group of Japanese researchers published a study about teaching budgerigars to tap along an audio-visual metronome (Hasegawa et al., 2011).
There are lots of examples of coordinated and even synchronous behaviours in the animal kingdom, the synchronisation of fireflies flashing being one of the most spectacular (here’s a video example). However, nothing seems to compare to the automatic and flexible human capacity to align motor actions to auditory and visual beats; hear a section of groovy music and it’s practically impossible to not have your foot tap to the music; get an audience applauding and you’ll get spontaneous synchronisation of the clapping; people walking side by side entrain and start walking in step; we synchronise our gestures and even minute body sways when in conversation with each other.
This human capacity to synchronise is pervasive and seems to be universal. The search for people otherwise healthy but truly unable to synchronise with auditory beats has yielded just one example so far, but in his case it seems to be an issue of auditory perception and inability to extract a salient beat out of the complex audio signal that is music (the participant was OK when synchronising with a metronome). It has been shown that even newborns have a sense of rhythm. Scientists are very interested in this phenomenon because it seems to underlie all communication, be that verbal, nonverbal or musical.
Humans are flexibly able to synchronise with beats in a wide range of different tempi, to different sounds, even or uneven rhythm patterns, different metrical levels (e.g. clapping on every beat vs. every other beat) and in various contexts. While we are much better at following auditory metronomes, we can sync with visual ones, as well. Also, we adapt to small changes in tempo and automatically correct for small inaccuracies created by ourselves or our co-performers, as we do in group music performance.
One important feature of human sensorimotor synchronisation is that it is based on prediction, not reaction. Tapping to a piece of music or following a metronome, our taps tend to precede the beats of music by some tens of milliseconds. This predictive behaviour (also called the negative mean asynchrony or NMA) is based on the way our cognition works in general: we create expectations of the future based on what we’ve already perceived. The few first chords of the song will activate these predictions about what comes next and when. Using these predictions to time our motor actions allow us to be in time with the music at a millisecond accuracy. Reaction times are much slower than that: if the beat we are following is not steady, but random as in the control condition in this study, we can not predict what comes next but instead will have to react, first listen to the stimulus sound and then tap as quickly as possible, generally meaning that we tap 100-150 milliseconds after the stimulus, instead of 10-50 before as in the predictive mode.
What’s this study about?
But back to birds. To see if Snowball actually synchronises with music, it was necessary to construct an experiment, where the researchers made different versions of “Everybody” for Snowball to dance/bounce to. This was done to eliminate the possibility that the synchrony is just a coincidence of the bird’s preferred bouncing rate happening to be the same as the tempo of the song. As usual, a single case study was not enough to convince everybody, but luckily Hasegawa and colleagues have now produced a much more convincing case of aviary entrainment.
They trained a group of budgies using classic conditioning, rewarding the birds with seeds after successful performances. First the birds learned to respond to sounds and flashing lights by pecking the wall, and little by little they got to the avian grade 8 after having learned to tap along the metronome for six beats. In the experiment, the beeping-flashing metronome had different tempi, with beat lengths ranging from brisk 450 ms (corresponding to 133 BPM) to sluggish 1800 ms IOI (33 BPM). They also included a control condition where the beat length was random, thus creating a sequence of irregular beats.
To test if there is synchronisation between the bird and the beat, two things have to be analysed. First, you check that the period of the bird matches that of the beats, or in other words, that they have the same tempo. Second, you check if the two also align in phase – that the bird pecks at the onsets of the sounds, not after or before. Especially you’ll need to test that the bird is engaged in prediction, not just a reaction task. If the bird is just reacting, but does so with a constant reaction time, it will have a matching period. To demonstrate predictive action, the phase error would need to differ from the expected reaction time and be either aligned with the onset of the stimuli or preceding it slightly. Some researchers would also require that the birds should show consistency of this synchronisation over a longer time, and that if you’d introduce tempo changes or other perturbations in your beats, they should adapt and follow the beat regardless.
The short segments of beats the Japanese researchers used only allow the first two of these analyses, checking for tempo matching and whether the birds predict the beats or react to them. The results of the statistical tests to check for tempo matching were clear, indicating that the tempo in which the budgies pecked the wall was in fact connected to the tempo of the stimulus. When looking at the phase-locking, the picture gets more unclear. In the fastest tempo, none of the birds in none of the trials was found to be predicting the beat. With slower tempi, more and more examples of prediction were found, in the slow tempi of 50, 40 and 33 BPM, about a quarter of the trials had evidence of prediction.
Is this a good result? Depends on what you compare it to. Humans, especially if rewarded with seeds, would probably be in a predictive mode in nearly 100% of the trials, a few trials would be botched due to lapses in concentration (or when your nose starts to hurt while pecking the wall of your cage), but otherwise we know that prediction is the way we do these tasks. We aren’t a fair comparison, though, and the point is not to show if animals are as good as us in predictive timing, but whether they do it at all. To this end, the researchers simulated some of the other ways in which the birds could have produced steady six-peck patterns that are tempo-matched with the metronome beats. Then they compared the birds’ performances to these simulations. In all cases the birds outperformed the simulations, suggesting that it is more likely that their patterns are produced at least to some degree by a beat-tracking mechanism and not just one that is constant or quick to react to sounds.
The researchers have been very thorough with their statistics. Their evidence looks solid and does indeed suggest that these birds can follow the tempi of the stimulus, for short bursts at a time, at least. Also it suggests that at least some of them might be able to track the beats and predict them at least some of the time. Not as solidly as humans would, but to a degree that has not been demonstrated in many non-human animals before.
This is cool not only because it is sort of “black swan” -kind of data, busting a myth about humans being the only animals able to do this, but also because there’s a cool theory about why the birds might be able to do this. Budgies are a vocal-learning species (alongside some other birds, seals, cetaceans, perhaps also bats and elephants), meaning that they can modify their vocalisations based on environmental sounds – human language and vocal music can be seen as extreme versions of this kind of behaviour. Studying these animals and their behaviour, neural anatomy etc. could help in understanding also how human language works.
These results are interesting in that this beat-following behaviour is mimicking what humans are able to do and thus theories linking the vocal learning ability to beat following get now more fuel.
Hasegawa A, Okanoya K, Hasegawa T, & Seki Y (2011). Rhythmic synchronization tapping to an audio-visual metronome in budgerigars. Scientific reports, 1 PMID: 22355637