Friday Grey Matters: The Neuroscience of Birdsong

i-508385da10314f00a0555e6a3fc1dcd4-grey !.bmp

First off, some exciting news for Grey Matters: Dr. Irene Pepperberg has agreed to do an interview for the series, likely in September.

Today's feature on Grey Matters is regarding the neural architecture underlying the learning and memorization of songs and sounds in birds. Hopefully it will help address a wide-spread, but fallacious notion that avian brains are too small and unsophisticated for complex learning and memory. Avian brains are not a primitive version of mammalian brains, but rather evolved in parallel under similar environmental conditions. The brain regions which process and reproduce sounds have mammalian correlates, but are entirely unique to song birds and parrots.

This results in some similarities in function but not neural substrates. Birdsong learning has strong similarities to the acquisition of speech in human infants. Both transition though a period of auditory memorization followed by solidification and then salient production. I will be reviewing the May 2006 Nature paper, "Neural mechanisms of birdsong memory" by Johan Bolhuis and Manfred Gahr, which is both fascinating and accessible.

Songbirds, parrots and hummingbirds share with humans and some other mammals the capacity for vocal learning. For successful song acquisition, young songbirds need to be exposed to the song of an adult conspecific (a 'tutor'), after which they progressively form their own song through a sensorimotor process of matching their own vocal output with the stored memory of the tutor song. Birdsong learning is considered to be the closest animal equivalent to human speech acquisition. Analogous to the acquisition of human speech, songbirds form long-term memories of tutor song.

Song-learning birds possess a network of brain regions (called the "song system") which conveys the ability to learn and reproduce song and sounds. This system of interconnected nuclei is lacking in non-song producing birds, and the size of these nuclei increase in size during times of the year when the songs are produced frequently (often breeding season, for example). In addition, the size of the nuclei corresponds with the number of songs that the bird knows!

(More under the fold.)

Below are schematics of the "song system" in a song-learning bird vs. a non-song learning bird. Obviously there are dramatic difference in this nuclei system, with songbirds having an elaborate network of processing which directly connect to the auditory system and the vocal output regions of the bird's brain.

i-467d9571f26baf4d138e7485c9ab4859-song system 1 copy.jpg

Abbreviations of nuclei: DLM, nucleus dorsolateralis anterior, pars medialis; DM, dorsomedial nucleus of the midbrain nucleus intercollicularis; HVC, a letter based name; lMAN, lateral magnocellular nucleus of the anterior nidopallium; mMAN, medial magnocellular nucleus of the anterior nidopallium; NIF, nucleus interface of the nidopallium; nXIIts, tracheosyringeal portion of the nucleus hypoglossus; RA, robust nucleus of the arcopallium; RAm, nucleus retroambigualis; rVRG, rostro-ventral respiratory group; X, Area X.

i-56ffe8f139d76a83016ac72b7048ee8f-non song copy.jpg

Vocal output is regulated by the syrinx (the bird's vocal organ). Non song-learning birds can still produce sounds, but they do not possess the network of forebrain nuclei like songbirds do. The research done to date points to the NCM nucleus as a very important player in the learning of song. Specifically, a significant correlation was found between neuronal activation of the NCM of adult male zebra finches and the strength of song learning, and has been now considered to be specialized for remembering the calls and songs of many individual conspecifics (other birds of the same species/environment).

Bird song learning happens in stages. Consider the below sonogram. This is a representation of the song of an adult male zebra finch (the tutor) and one of his sons at different stages of development.


The first few vocalizations are called "sub-song," and occur before 40 days post-hatching. At around 60-80 days post-hatching the baby bird produces "plastic song" which resembles the tutor song more than the "sub-song." These phases are often compared to babbling and solidification in human infant speech acquisition. At about 100 days post-hatching, the bird produces "crystallized" song which resembles the tutor song with extremely high accuracy. The further input and correction from the "tutor" bird is crucial to this process, as is the timing of the song exposure. Songbirds raised in absolute silence for their entire lives never develop true song, although they can vocalize much like non song-learning birds. Interestingly, the threshold of sound exposure is low. For example, songbirds raised with traffic noises, etc (obviously not bird song they would hear in the wild) develop song-like vocalizations. Baby songbirds can also learn their song from an audio tape, even a song of a different species; a "real" bird isn't even required. These results seem to suggest that while environmental cues provide the "trigger" for song learning and memory, much about bird song learning is innate.

Activation outside of the "song system" set of nuclei is also hypothesized to play a role in learning, although this research is just beginning. Until recently, it was thought that two forebrain pathways connecting a number of 'song control nuclei' comprise the neural substrate for birdsong. Early evidence for the involvement of the song system in song came from a series of neuroanatomical and lesion studies. The caudal pathway, including the HVC and the RA, is involved in the production of song. Lesions to nuclei in this pathway, or to any of its connections, result in immediate, profound and irreversible deficits in song production in adult birds. The rostral pathway, including the HVC, the lateral part of the lMAN and Area X, was thought to have a role in song learning. This suggestion was supported by the finding that bilateral lesions to lMAN or Area X disrupt song acquisition, but have little effect on crystallized song in adults. However, such lesions might indirectly affect the development of premotor regions of the song system. Recent studies involving the expression of immediate early genes (IEGs) showed that exposure to song does not lead to neuronal activation in nuclei in the song system. Such exposure does, however, lead to neuronal activation in other brain regions, particularly the caudal part of the NCM and the caudal part of the medial mesopallium (CMM). Song production by itself does lead to IEG expression in song system nuclei. Therefore it can be assumed that the song system of nuclei are a vital for song learning and production, but not all-inclusive.

Correlates to Human Language
Here's where it gets really cool.

In humans, regions traditionally associated with speech perception, which are centred around Wernicke's area in the superior temporal lobe, are distinguished from speech motor areas, including Broca's area, in the frontal lobe. However, in humans, it has been suggested that speech perception affects speech production from birth onwards. For instance, there is some evidence to suggest that speech perception modulates the excitability of tongue muscles. Infants in their first months of life acquire sophisticated information about their native language simply by listening before they know the meaning of words. This early experience affects not only their discrimination ability and listening preference but also alters subsequent perception and motor performance. The early experience is language-specific, such that speakers that learn a second language after puberty produce it with an accent typical of the primary language. The results of a functional MRI (fMRI) study revealed that, similar to adults, 3-month-old babies who were exposed to speech had significant activity in brain regions in the left hemisphere, including the superior temporal gyrus. These findings show that precursors of adult cortical language areas are already active in infants long before the onset of speech production. So, there is an interesting analogy between the mechanisms of human speech acquisition and song learning in songbirds. That is, brain regions involved in auditory learning in humans and birds are anatomically separate from those involved in sensorimotor learning, and vocal learning involves continual interactions between them.

More on this to come in future episodes of Grey Matters.


More like this

San Francisco-based trade organization representing neuroscience companies has lined up companies nationally and heavy hitters in the nation's capital to lobby for a $200 million superfund to help bring products to market for brain and nervous system diseases ranging from Alzheimer's to insomnia.

Great post!

On a related matter - while I understand that birds sing to attract mates and demarc territory - has anyone come up with a plausible explanation as to what evolutionary pressure would lead to the development of mimicry at the level that parrots exhibit that you've heard?

Hi DV, thats a great question. In the "wild," parrots don't have what would be traditionally termed bird song, but rather a collection of plastic and meaningful calls. For example, a "danger" call, a "food" call, "mating", etc....with the possibilities being endless on the specificity. I also suspect it has evolved as a defense mechanism. Grey parrots feed on the ground, in a open and vulnerable space. Often a few birds will pose as sentinels to keep watch for their many predators so the rest of the flock can eat. I've heard reports that Greys will also mimic large predator noises (growls, hisses, etc) to confuse and intimidate their smaller, and more common predators. Greys are one of the oldest species of parrot, and one of only a few parrot species in Africa (thought to be where all parrots originated). Although I haven't seen any studies (burgeoning field, this) I might wager a guess that Grey parrot's HVC nucleus is rather larger than other birds/parrots, or that their song system is more sophisticated than that of a zebra finch.

Yes, that would go a long way to explain the 'situational appropriateness' that some of the better talking Greys exhibit.

My best talker, Stormy (a female!) will say "hi!" when someone enters the house, and "bye!" when I am leaving for work in the morning - this without formal training. Also "hello" when we pick up the phone. I'm not suggesting that the bird understands what it is saying, only that it has linked certain events with a 'call' we (the rest of the flock) use at that time. This would fit nicely with your hypothesis.

Baby songbirds can also learn their song from an audio tape, even a song of a different species; a "real" bird isn't even required.

Just for clarification -- do the babies learning this way learn their own species' song, or do they end up singing like a different species?

(First time reader here, nice blog!)

Welcome Chuck! It depends on the species, to some extent. Most of the research on this area has been done on zebra finches. For this bird, they can learn their own species' song perfectly from a tape, and if they hear a tape of another species song, they will create a "hybrid song" somewhere in between the other species and their species' song. This is another reason why song learning is considered to have a component thats innate.

Parrots like Greys are different though. They just learn whatever, and reproduce it perfectly. No "hybrid" song for them. So if you played a tape of a cat meowing, or a horn honking to a baby parrot, thats what you would hear in a few months. Not a hybrid of a parrot call + meow/horn. Why that is, not enough research has been done to determine, but I think its because Grey's have a more sophisticated and adaptable song system (i'm guessing here).

hey just wondering if you could maybe give me some ideas on what i should do for my science fair thats coming up?

By cherrelle tanuvasa (not verified) on 03 Dec 2007 #permalink