I am interested in all aspects of sensorimotor control – the control of motor actions based on sensations and intentions – in humans, animals and artificial systems.
I started my career designing helicopter autopilots, which are computer programs that can fly a helicopter based on sensory inputs and goals set by the pilot. These artificial control systems were fascinating, but I was even more intrigued by the control systems observed in nature. It is finely-tuned sensorimotor control that enables, for example, a gymnast to perform powerful yet elegant movements, a dragonfly to capture almost any prey in the blink of an eye, or a flock of birds to fly collectively as a unit.
How are biological control systems and artificial control systems different?
Driven by this question, in graduate school I created theoretical and computational methods to study collective motion of animal groups and mimic some of its features with teams of robots. However, how an animal implements sensorimotor control is intimately related to how its nervous system functions, something I knew very little about. For this reason, in 2011 I dove into neuroscience by joining Anthony Leonardo’s group to study the neural basis of prey pursuit in dragonflies, an extraordinary example of goal-oriented behavior.
Studying the detailed motions of the dragonfly head and body, my colleagues and I were amazed to find that the dragonfly strongly relies on predictions. As it vigorously steers its body to intercept the prey, the dragonfly simultaneously steers its head to predictively cancel the effect of body steering on the visual position of the prey. In this way, the dragonfly eyes can remain fixated on the prey and detect prey maneuvers that require the dragonfly to correct its course. This predictive control of movement is functionally analogous to that employed by the mammalian brain to perform targeted reaching movements.
In the research group of Adam Hantman, I am now investigating how different regions of the brain contribute to the interplay between predictive model-based control and reactive feedback control, by studying the control of dexterous forelimb movements in mice.