Flying insects solve a daunting control problem of generating a patterned and precise motor program to stay airborne and generate agile maneuvers. In this motor program, consisting of every action potential controlling wing musculature, each muscle encodes significant information about movement in precise spike timing down to the millisecond scale. While individual muscles share information about movement, we do not yet know if they have separable effects on an animal's motion, or if muscles functionally interact such that the effects of any muscle's timing depend heavily on the state of the entire musculature. To answer these questions, we performed spike-resolution electromyography and precise stimulation of individual spikes in the hawkmoth Manduca sexta during tethered flapping. We specifically explored how the flight power muscles themselves may contribute to pitch control, which is necessary to stabilize flight. Combining a correlational study of visually-induced turns with causal manipulation of spike timing, we discovered likely coordination patterns for pitch turns, investigated whether these correlational patterns can individually drive pitch control, and studied whether the precise spike timing of power muscles can lead to pitch maneuvers. We observed significant timing changes of the main downstroke muscles, the dorsolongitudinal muscles (DLMs), associated with whether a moth was pitching up or down. Causally inducing this timing change in the DLMs with electrical stimulation produced a consistent, mechanically relevant feature in pitch torque, establishing that power muscles in Manduca have a control role in pitch. Because changes were evoked in unconstrained flapping in only the DLMs, however, these pitch torque features left a large unexplained variation. We find this unexplained variation indicates significant functional overlap in pitch control, such that precise timing of one power muscle does not produce a precise turn, demonstrating the importance of coordination across the entire motor program for flight.

Moths attached to a 6-axis load cell tether would freely engage in bouts of flapping flight, during which the DLMs would be stimulated with a specific timing relative to the DVMs. Stimulus delay times were chosen at random from a range of 4-40 ms after DVM spike detection, with each delay time applied for a 20 second recording period with EMG and F/T transducer recorded at 10 kHz. Single 0.25 ms pulses of either constant current or constant voltage were applied to silver wires placed on either end of each DLM, with each stimulation separated by at least 4 seconds to avoid entrainment. Timing of stimulus was achieved by passing either the left or right DVM voltage recording through a custom analog spike-detection circuit and microcontroller. The controller would trigger a stimulator (A-M Systems Model 3800) with a stimulus isolation unit (A-M Systems Model 2200) on a controlled delay time from the onset of a detected DVM spike. 

All stimulus pulses were biphasic, but specific stimulus features such as whether constant current or voltage were used and of what amplitude, were calibrated for each moth individually. Stimulus was repeatedly applied while the moth was in a quiescent state, with stimulation amplitude steadily increased until evoked motor action potentials were observed in both DLMs with no evoked action potentials observed in surrounding muscles. While most individuals had their best response from constant current stimulation, differences in electrode placement and individual anatomy and physiology resulted in better results for some individuals from constant voltage stimulation. After calibration, stimuli ranged from 0.05-0.1 mA for constant current stimulation and 5-10 V for constant voltage. Note that evoked action potentials occur roughly 4 milliseconds after a pulse of current is applied.