Echolocating insectivorous bats are nocturnal mammals that capture fast, erratically moving insects in flight. Bats emit short ultrasonic pulses that form beams of sound and use the returning echoes to guide behavior. The frequency, duration and timing of the sonar pulses, along with the spatial direction of the sonar beam restrict the information returning to the bat, and can be considered a component of the acoustic gaze of bats. A great deal is known about the time-frequency structure of bat echolocation calls and their relationship to the stages of foraging flight in bats. It is however not known how bats direct their sonar beam in flight or how beam direction is related to flight control.

This is the first study of the sonar beam direction in freely flying bats as they chase and capture insects. An apparatus and method to measure the sonar beam pattern of echolocating bats (<it>Eptesicus fuscus</it>, big brown bats) as they fly in a laboratory flight room is described. It is shown that the bat locks its sonar beam tightly onto a target during pursuit (Chapter 2). The flying bat's sonar beam consists of two lobes directed apart in the vertical plane (Chapter 3). There is a coupling between acoustic gaze (sonar beam axis) direction and flight turn rate that can be expressed as a delayed linear control law. The gain of this law (steepness of the relationship) varies with the bat's behavioral state (Chapter 4). The bat, when pursuing erratically flying insects, adopts a strategy that keeps the absolute direction to the target a constant. This strategy is shown, under some assumptions, to minimize time-to-intercept of erratically maneuvering targets and is similar to parallel navigation implemented in guided missiles (Chapter 5). The bat is not helpless against ultrasound-triggered evasive dives evolved by some hearing insects. The bat adopts flight strategies to counter such dives (Chapter 6).

This work allows us to compare spatial behaviors well studied in visual animals, with similar behaviors in an animal that is guided by hearing and make inferences about common computational strategies employed by nervous systems.