Detection and tracking of moving objects by the fly Musca domestica

The experiments and their analysis, presented in this paper, deal with the tracking behaviour of tethered flying flies (Musca domestica) towards moving black objects in front of white or randomly contrasted backgrounds. Experiments were carried out under closed—as well as under open-loop conditions. Under closed-loop the fly is coupled to its environment, whereas under open-loop its motor output does not influence the visual input. The following expeeiments have been carried out: Tracking of a black object in front of a white background (Section 3.1); under closed-loop, a fly follows an object, moved at constant angular speed ω, however, lagging behind it by the tracking angle ψtr. ψtr is linearly related to ω. Stable tracking in Musca is confined to about±20o both sides of the flight direction. Tracking of a black object, imbedded in a randomly contrasted pattern of different contrast (Section 3.2): under these conditions the same tracking rules as in the case of a white background are observed. With decreasing contrast difference between the object and the random pattern, the steepness of the linear relation between ω and ψtr increases. These observations are consistent with earlier experimental results and can be explained in terms of a phenomenological theory describing visual orientation in Musca; Poggio and Reichardt (1973a), Reichardt and Poggio (1975). Detection and tracking of a black object, hidden in a random background pattern of the same contrast (Sections 3.3 and 3.4): it is shown that the necessary condition for detection and tracking of an object under these conditions is the presence of relative motion between the object and the background. When the background is moved and the object is stationary, no detection is observed. The effect strongly depends upon parameters such as the object and the background configurations and their contrast differences. Full independence between the nervous correlates induced by the object and the background is also reached when the object is moved with constant angular speed against the background. In this case the phenomenological theory is easily applicable. Detection is demonstrated under closed—as well as under open-loop conditions. Analysis and interpretation of the detection phenomenon requires a mathematical formalism describing the responsible nonlinear interaction processes in the Central Nervous System of the fly. The Volterra description is such a formalism; Poggio and Reichardt (1973b). In terms of this formalism and in view of more recent experiments by Pick (1974) and by Heimburger et al. (1975), the detection effect can be accounted for by fourth order nonlinear interactions between the signals received from two to four receptors. The latter observations and their interpretations might lead to a better understanding of the discrmination of the optical environment into figure and ground