How Canary Songs Can Teach Us About Memory

New study from BU research team suggests how songbirds can use the memory of past song sequences to choose what to sing next—creating a complex syntax resembling human behavior.

On June 17, a nine-person research team from BU’s Center for Systems Neuroscience, BU Biology, the BU School of Medicine, and the University of Oregon, published new research in Nature on the neural basis of canary song. The findings offer new insight into how short-term memory mechanisms can carry past information to impact ongoing motor control. This research lays the grounds for future study of working memory mechanisms that adapt to new conditions or fail when brain circuits are damaged—a model that could point to new therapies for speech and comprehension deficits associated with aging and neurodegenerative diseases in humans.

Here Yarden Cohen, lead author and former Postdoctoral Research Fellow at Boston University, reflects on the findings and their exciting implications for future work.

Many human behaviors are structured by a composition of basic elements. We take words and compose sentences that have correct grammar and meaning. From musical notes we compose melodies that have harmony and pleasant improvisation (as in jazz). From gestures, we make dances that have beauty. It is impossible to do any of these without a memory of what was previously said, played, or danced, and in what order.

Canaries compose their songs from basic vocal elements called syllables. Songs contain hundreds of syllables organized in a way that indicates that during song these birds use the memory of past song sequences to choose what to sing next, creating a complex syntax much like many human behaviors.

We investigated neural activity in the canary’s “song circuit,” the part of the bird’s brain driving the production of song. We found that beyond driving the currently-produced song syllables, neurons in the canary’s song circuit encode past song phrases and transitions, carrying information relevant to future choice of phrases as “hidden states” during song.

We refer to these activity patterns as “hidden neural states” because different patterns are active during the production of the same song elements and indicate song variations occurring several seconds and tens of syllables in the past. This finding opens a unique window on short-term memory mechanisms for creating complex animal and human behaviors that proved challenging to observe during unconstrained motion in mammals or in songbirds with simpler syntax rules.

Following this research, we will further dissect the components of the song circuit contributing these hidden neural states. Studying the neural basis of canary song production may make it possible to understand how working memory mechanisms develop as neural circuit mature, adapt to new conditions, or fail when brain circuits are damaged. Developing such a model may point to new therapies for speech and comprehension deficits that come with aging and in neurodegenerative diseases such as Parkinson’s and Alzheimer’s.