What happens in our brains as we learn to walk, swim or talk? How are the underlying learning algorithms implemented, and how are the honed skills encoded and generated by the nervous system? For many years, the Olveczky lab addressed these questions in songbirds, an experimentally tractable model that has given us many valuable insights into how complex motor sequences can be acquired. But unlike songbirds who specialize in learning a single vocal sequence early in life, we mammals have evolved a general capacity to learn seemingly arbitrary and highly complex motor sequences, something that has led to language, tool use, and synchronized swimming. This remarkable learning ability is often attributed to our cortex, but little is known about how the cortex teams up with the rest of the motor system to implement it. What are the learning and control algorithms involved, and how are they instantiated in neural hardware? We have started to explore these questions by using rodents as our subjects and training them – in automated ways – to master different types of motor tasks.

By manipulating neural circuit function using optogenetics, lesions, and pharmacological techniques, we can delineate the circuits involved. We then record activity in these circuits during learning and execution of motor skills to arrive at a mechanistic description of how the mammalian brains acquires and executes complex learned motor behaviors.