Many human and animal tasks are thought to be controlled with the tau informational variable. It is widely accepted that controlling the rate of change of tau (tau) during decelerative tasks, such as when braking or landing, is one common perceptual control strategy. However, many tasks require accelerating before decelerating to a goal, such as reaching. An advancement of tau theory shows how a single action formula may be used to control the full action unit from initiation to peak velocity, and to rest at the goal, with the same perceptual tau information as before and accounting for the same decelerative kinematics as before. Here, we test the theory against data from high-speed video of a hummingbird flying to its flower feeder. We find that the theory accounts for 97% of the variance in the data, and thus supports it.