Previous Top Next

Spike Detection

Fig. 8 gives a notion how highly variable the flight modes in different Drosophila individuals can be. As was shown by Heisenberg and Wolf (1984), size and shape of the torque spikes are under reafferent control. As will be shown below (3.5), size, timing and polarity of the spikes are subjected to conditioning dependent modulations. Because of this high intra- and interindividual variability in general spike appearance, it is impossible to construct an error proof spike detector. Even the human eye is sometimes not capable of unambiguously discerning spikes from the torque baseline. In the end, only the fly 'knows' whether it has produced a spike or not. Fig. 11 shows this difficulty: the human subject is obviously following different rules to detect spikes according to the general appearance of the torque trace. The volunteers detected less spikes than the computerized spike detector if the task was judged to be 'easy' and spikes were clearly distinct from the baseline (Fig. 11, first and second fly). However, when the baseline was oscillating very much the number the spike detector produced was lower than the number of spikes the volunteers counted (Fig.11, third fly). The same was true if the baseline was very low and the spikes very small.

Figure 11
Fig. 11: Comparison of the spikenumber from four flies during the first three periods of flight between five volunteers and the spike detector used in this study. See text for details.

Apparently, there is quite an amount of subjectivity in the detection of spikes. This is of course also reflected in the programmed spike detector. In this case the number counted by the programmer was closer to the spike detector than the numbers of the five volunteers (not shown). It is important, that throughout the study one rule is followed in order to always compare the same turning maneuvers and that there are not too many turning maneuvers between the counted spikes which may corrupt the results (see below, 3.3).

Previous Top Next