Imagine that you are out in the bamboo thickets of the Amazonian cloud rainforest, on an especially elaborate bird watching expedition. You hear what you think is another bird song, but then notice that in fact two birds are singing the same song, alternating in such a rapid and temporally precise way that they sound like a single bird. How do they do it? Birdsong has already been shown to be an excellent model for human speech learning. It turns out that the way this rapid exchange of song takes place may have implications for the way other organisms perform learned actions together in a social setting, such as dancing, or duet singing.
Previous work has found that songbirds need to hear their own song in order to learn and maintain their adult songs. To aid them in this task, they have specialized neurons in an area analogous to premotor and prefrontal cortex in humans, called area HVC. In most birds, these neurons fire the most when a bird hears its own song. (Sensibly, these were named the bird’s own song (BOS) neurons.) The plain-tailed wren, the duetting bird from the Amazonian cloud rainforest, is unusual because BOS neurons also responded to a duetting partner’s song. Additionally, the male’s HVC responded more strongly to the female’s vocalizations than to its own, something that has never been seen before in the songbird field.
Drs. Eric Fortune and Melissa Coleman set about trying to find out exactly how these birds were capable of such an extraordinary vocal feat . They started by recording over 150 hours of song from the birds in the wild, and looked at the changes that occurred in the song when the bird was singing by itself, or as a part of a duet. Here, they found that the gaps between the units of the song, called syllables, were much larger when the birds were singing by themselves, as if they were waiting for the other to join in.
Next, they captured some of the birds, and recorded from area HVC with extracellular electrodes. This involves placing a small electrode into the brain, but not directly inside of a specific cell, and so a single electrode may be picking up signals from multiple neurons. Once the electrode was in place, Eric and Melissa played recorded songs to the birds: male solo songs, female solo songs, and of the combined duet. Although there was some variability, the majority of neurons in both birds preferred duet song, followed by female song, and finally male song, regardless of the sex of the bird. Getting rid of the gaps between the syllables resulted in decreased responses, although the order of preference was still the same. Clearly, the timing of the events is particularly meaningful, as might be expected from an organism that is involved in such a temporally precise behavior.
There are many ways that a coordinated behavior, such as a plain-tailed wren duet, could be encoded by neurons in the brain. It could be that one bird learns primarily from its own performance and cares less about the performance of a partner. As this study has shown, this is not the case, and changes happen in real time, as the female widens the gaps between syllables, waiting for the male to respond and continue the duet. Additionally, both birds are the most sensitive to the combined output of both songs together, not their songs individually. Alternatively, one might imagine that both birds wait for a start signal from the other, and once it is received, execute a pattern that has been learned in the context of the other’s production. This study conclusively showed that the combined duet is encoded differently than either the male or female song, and differently than the sum of the responses to either male or female song in isolation as well. This means that they have an internal representation of what the combined duet should sound like.
In primates, mirror neurons are active both when the primate performs a task, and when they watch the same task being performed. The Area HVC neurons recorded from in this study are very similar to the mirror neurons found in primates, and are even found in a similar region of the brain called the premotor areas. When normal songbirds and plain-tailed wrens sing, these neurons are deaf to auditory activity, and only reflect motor activity. When anesthetized however, these neurons seem to be responding to sound as well as movement. This suggests that listening to song itself may be a part of this bird’s motor planning.
While it may have started out as simply a bizarre natural behavior, this paper now has the potential to change the way we think about coordinated behaviors, such as singing, dancing, and possibly even more complicated forms of cooperation such as coordinated hunting. Much of its strength rests on the fact that it has isolated some of the neural correlates of the internal representation of another organism, a phenomenon also observed in the mirror neurons of primates. I interviewed Drs. Eric Fortune and Melissa Coleman, and we still don’t know why these birds duet. It seems to be used to display the strength of a bond between a mating pair to chase away single birds, to scare off predators, and possibly to signal to other couples and family groups. We just don’t know. More studies will have to be done, but the idea that both participants must have neurons that are sensitive to the emergent property of their combined behaviors for optimal performance, and don’t simply execute their own part in the context of the other while the other participant executes theirs, is a powerful one, and one that now has a neural basis. Future studies that address the way this coordinated behavior is learned, and record from neurons that project to motor learning circuits, would give us valuable insight into the ways individual learning differs from learning a skill in the context of another performer.
 Science. 2011 Nov 4;334(6056):666-70. doi: 10.1126/science.1209867.
Neural mechanisms for the coordination of duet singing in wrens.
Fortune ES1, Rodríguez C, Li D, Ball GF, Coleman MJ.