For thousands of years people have admired the beauty of birdsong. But scientific understanding of this phenomenon is still restricted to studying learning and production mechanisms, as well as ecological function (how songs are used to attract mates and to defend territories). But what is it that makes a birdsong effective in attracting mates? What are the acoustic structures that are designed and able to engage a listener, and affect his or her emotional state? To approach these questions, I would like to test if principles derived from music theory and aesthetics can help us understand the biology of birdsong. For example, I would like to test if songs are delivered in a manner that can hold a listener's attention similarly to the manner in which a street musician does: through fulfilling anticipations, creating suspense by delaying their fulfillment, or surprise by violating them. Once I manage to identify such features, I should be able to perform comparative studies (through analysis and experimentation) across birdsong and human music. I assume that the structure of birdsongs may be better understood by investigating its aesthetics explicitly. I hope that this way, biological mechanisms for inducing engagement and driving emotions can be derived, at least in part, from aesthetic principles.
I suspect that a very basic principle instantiated in many aesthetic artifacts, and perceived as salient by their addressees, lies in dynamically mixing contrast and similarity. Similarity between elements lets us form predictions, and creates a sense of ease of processing where these are fulfilled, while contrasts grab our attention, and surprise us. An effective way to generate both contrasts and similarity is using a repertoire of discrete elements to combine into aesthetic artifacts. In music, for instance, the pitch scales that melodies are based on arise through a discretized use of the continuous frequency space, and rhythms rely on combining discrete - not continuous - time intervals at which notes are placed. Discrete elements have another advantage: they are easy to memorize and perceive, which makes them ideal for communication purposes, especially where communication signals are learned, or culturally transmitted.
Project "Discrete rhythm categories in birdsong"
If the discretization of continuous feature space really is a general property of aesthetic signals, we might also find it implemented in birdsong.
In a joint project with Ofer Tchernichovski and David Poeppel, we are exploring whether a songbird's rhythm (i.e. the relative timing of notes in a sequence) is discretized. Surprisingly, rhythm in birdsong has not been extensively studied. We do not even know for sure that birds "have rhythm" at all, in that they are in active control of the timing of their notes. Alternatively, note timing might be a mere epiphenomenon of stringing together the vocal gestures to produce the desired sounds, instead of being something that the bird controls. This means, the resulting rhythms would depend on peripheral dynamics of the vocal apparatus, not on centrally controlled patterning.
However, our preliminary results seem to point the other way. When we analyzed a class of simple rhythms in thrush nightingale song, alternating (short-long) repetitions of the same note, we indeed found evidence for a discretized use of rhythm space. Thrush nightingales preferably produce time intervals related to each other in small-integer ratios, avoiding the ratios in between. We observed three well-separated rhythm classes: isochronous rhythms, dyadic rhythms with interval ratios centered around 1/6 of the cycle duration, and dyadic rhythms with ratios around 1/3 of the cycle duration (see figure). This invites the interpretation that note timing in thrush nightingale song is not a mere epiphenomenon of stringing up gestures to produce a sequence of vocal elements. Instead, rhythm to the birds seems to be a category containing distinct classes, resembling human musical rhythm.
Project "The rate of birdsong as a window into auditory aesthetic perception"
At the same time as searching for common mechanisms with which birdsong and music entice their listeners, I would like to better understand how principles of auditory processing may set species-specific limits to what kinds of sound sequences are typically appreciated. In other words, why do music and birdsong sound so different? Is it because we process sequences differently than birds? To approach this question, we focus on tempo preferences between species, as tempo differs considerably between music and birdsong (most of birdsong being much faster). By asking human listeners about their preferred tempo in birdsong playbacks, we hope to find out whether the slower pace of music reflects a general preference for slower note rates in any sequence (including birdsong), or whether music is slower for other reasons than rate preferences (such as motor limits when producing music).