Can a Fly Ride a Bicycle? [and Discussion]

In humans visual flow field information is available to many motor output programs enabling them to guide the whole organism. This basic organization still constitutes a major challenge for any theory of orientation. Studies of the fly Drosphila melanogaster during the past 10 years have shown that in the lower animals sensory-motor control has a similar degree of flexibility. In the flight simulator the tethered fly adjusts the strength of its motor commands to their efficacy. It can stabilize the panorama against rotations not only by yaw torque but also by thrust. It learns to invert its flight manoeuvres in response to positive feedback in order to stabilize a stripe in the frontal visual field. In the present report we demonstrate that Drosophila is able to use the force of its legs to stabilize the panorama irrespective of the polarity of the feedback provided experimentally. All these behavioural performances have a common functional organization with the following properties: (i) the system has a desired state from which the actual state may deviate; (ii) to reach the desired state the system randomly activates a range of motor programs; (iii) the system compares efference copies of the motor programs with those sensory inputs which represent the deviation from the desired state; and (iv) if a significant correlation is detected for a certain motor program, this is used to shift the sensory input into the direction of the desired state. It is proposed that for organisms with more than one motor output this is the basic scheme of sensory-motor coordination.