Bat and Bird Songs for Systems Neuroscience

This week’s Science features an article about the songs of bats - not the ultrasonic echolocating calls most often associated with these wonderful creatures but complex vocalizations that bats use socially.

The most well-studied “songsters,” as the author, Virginia Morell, calls them, are songbirds. In species like canaries, zebra finches and starlings, the male sings a courtship song that he learns from his father to woo females.

Male zebra finches, for example, learn just one song, which typically consists of several syllables, and repeat the same song throughout their lives (hundreds or thousands of times per day). Each rendition is virtually indistinguishable from another, and yet finches can learn to modify the pitch of any given syllable on a millisecond timescale. Canaries also sing to woo females, but their songs are long and complex, with relationships among different syllables that imply nontrivial syntax. In systems neuroscience, such birds are used to ask how the brain learns and produces complex motor movements and sequences  Some popular articles and even some research papers hype up birdsong to be a model of human language. While some singers produce songs with apparent syntactic structure, the handful of papers that have tried testing birds’ comprehension of syntactic rules have been questionable in their experimental design (see e.g. Dan Margoliash’s paper on starlings learning recursive grammar or Dai Watanabe’s paper on Bengalese finches learning syntactic rules and a general criticism here). One challenge in asking if animals possess human-like cognitive abilities is that experimental tasks that scientists design must only be solvable using the cognitive ability in question - it is very hard to design a task that cannot be solved through simpler rules (the most famous and extreme case of this problem is the German horse Clever Hans  who was claimed to be able to solve arithmetic problems, but was actually using his trainer’s body language to arrive at correct answers).

In contrast to birds, vocal communication in mammals is limited mostly to humans and cetaceans - chimps don’t really communicate vocally and rodents do so to a limited extent  Furthermore, birds’ brains don’t have cortex, the six-layer structure that allows mammals to do cognitive acrobatics, which fact on the one hand makes it intriguing how cognition has evolved in birds, and on the other hand means that the neural circuits for vocal learning are not quite the same in birds as they are in mammals.

And but so this is why the story of singing bats is interesting: they are mammals that communicate vocally. It’s possible that bats could become a model system for serious neuroscientific study of vocal communication. For this to happen, someone would need to show that (1) bat songs are produced and controlled by the central nervous system, especially the neocortex, which controls language and vocal production in humans and (2) that their songs are learned and somewhat malleable.