The brain is organized on multiple length scales. Individual brain regions harbor local circuits with distinct properties; these brain regions associate into mesoscale circuits with other cortical and subcortical areas via long-range connections. Information is represented by patterns of action potentials in widely distributed ensembles of neurons. We are trying to understand how neural representations of information (i.e. patterns of action potentials) drive behavior, and how these representations in turn are shaped by multi-regional neural circuits. We address both questions in behaving mice in the context of motor planning and short-term memory.

Motor planning plays key roles in motor control. Movements that are preceded by periods of motor planning are faster and more accurate than in the absence of planning. Motor planning is also a prospective form of short-term memory that links past events and future movements. During motor planning, neurons in motor cortex show persistent activity related to specific movements, long before movement onset, in the absence of sensory input. I will present our studies on how multi-regional neural circuits maintain selective persistent activity and how this activity relates to movement.