One way to better understand how brain circuits control behaviour is to take recordings from as many individual brain cells as possible. Making single-neuron recordings in an awake animal as a behaviour unfolds is tricky, and the hard work doesn't necessarily pay off: in animals with complex brain circuits that contain mutual connections, the technique doesn't always give clear results. And for Michale Fee, a neuroscientist at the Massachusetts Institute of Technology in Cambridge, the process is further complicated by his temperamental subject, a 15-gram songbird that won't sing if handled too much.

To get around these obstacles, Fee and his postdoc, Michael Long, devised a simple but innovative contraption to study how a zebra finch's brain controls the timing of its singing.

Male zebra finches sing one song throughout life — a song comprising a repeating motif that sounds a bit like a squeaky wheel and lasts for up to about one second. A song's length depends on several factors, including how many motifs are repeated, the timing between motifs, the duration of the syllables that make up the motif and the timing between each syllable. Data from a study done in 2002 suggested that a brain area called the premotor nucleus HVC (HVC) acts as the clock for at least one aspect of song timing.

But simply observing HVC neurons firing during a song would not be conclusive, says Fee. “You can't distinguish which neurons are generating the timing and which neurons are following the timing just by watching them because these circuits have mutual connections between neurons,” he says.

Long had begun investigating how HVC cells generate their regular, clock-like bursts of activity. Brain areas can be reversibly inactivated by cooling them to about 10 °C, and Long and Fee discussed using temperature to temporarily block the activity of the HVC. It dawned on Fee that less drastic action — cooling the HVC by just a few degrees — could be a good test for the control of song timing.

“Temperature has a very strong effect on the speed at which neurons operate,” explains Fee. “We realized we could use temperature to very subtly control the speed of the circuit and see whether that affected the timing of singing.” The experiment required a device that could precisely cool the HVC but that would still allow finches to move and sing freely. Fortunately, the HVC is near the brain surface and Fee is handy at engineering tiny devices.

He disassembled a thermoelectric device — one that converts an electric current into a temperature difference — and used its components to build two tiny gold-plated probes to place on both the right and left hemisphere HVCs in a finch's brain. He and Long then tested the instrument. “We ran the current through the cooling device, put a female in front of a male, he started singing and it came out more slowly,” recalls Fee. “It was amazing. The effect was very clear.”

All components of the song's timing seem to be controlled by the HVC, and stretch out by about 3% for every 1 °C of cooling. Long and Fee propose that the HVC clock operates similarly to a set of dominoes, with each neuron kicking off the next one to fire, and then resetting itself for the next song motif (see page 189).

Fee says his invention should be a powerful way to study questions about how brain circuits control timing. But, he adds, his gizmo-building is exclusively reserved for the lab. “I'm surprisingly resistant to home repairs.”