A major challenge in neuroscience is to understand how the computational properties of single neurons contribute to the properties of neural circuits and, ultimately, behaviour.

Since decades, flight control in flies serves as a fruitful model system for understanding basic mechanisms of neural computation and locomotor control. Especially research on visual motion computation shows rapid advances due to the development of modern genetic and electrophysiological tools. Significantly less is known about the neuronal mechanisms by which visual motion information and other high-level signalling are structuring the underlying motor pattern.

In the motor circuit of flies, single motoneurons control wing steering muscles crucial for flight stability and manoeuvring. Their precisely timed activity depends on sensory feedback from the visual system and proprioceptive mechanosensory cells. The flight motor system, including the mechanosensory feedback loop, is usually considered as a “black box” and treated with control theoretical and descriptive models. While these approaches allow to identify functional features, they do not represent the actual neural circuitry and are of limited use to explain on a cellular level how information is processed and behaviour is generated.  As part of an ongoing effort to uncover the neuronal mechanism underlying sensory integration in insects, we model the ion channel dynamics of the fly’s wing muscle motoneurons. Based on detailed anatomical, electrophysiological and behavioural data, we develop a numerical simulation that describes motoneuron action potential generation by dendritic integration of sensory input.

Using a Hodgkin-Huxley model, a standard model for biological plausible simulation of nerve cell dynamics, we simulate the activity of a motoneuron that is driven by visual interneurons and proprioceptive mechanoreceptors via electrical synapses. We determine simulation parameters by fitting the model’s response to previously measured experimental data. Our simulation shows that motoneuron firing frequency and timing depend on magnitude and temporal structure of both visual and mechanosensory feedback. While mechanosensory feedback provides a temporal set-point for AP initiation, graded potentials from the visual system may shift AP timing within the stroke cycle. 

Our results suggest that the computational power of single motoneurons is sufficient to explain rapid multimodal feedback integration in flies. It further suggest that the integration of visual and mechanosensory feedback on the motoneuronal level is a common principle for motor control.