Previous Top Next

Spike Dynamics and -Timing

While in neither of the groups spike duration was modulated in response to heat (data not shown), operantly trained Drosophila exhibited a small but highly significant modulation of spike amplitude, even when the heat was switched off during the last two periods (p<0.001, fig. 20A).

Figure 20A
Figure 20B
Fig. 20: Mean amplitude indices for all four experimental groups (N=100 flies each). The flies learn to generate larger torque spikes in quadrants with one of the pattern orientations, if they were reinforced in these quadrants during the training periods (dotted bars). This is the case for the operantly (A) and the classically (B) trained flies. The control groups, which did not receive any reinforcement modulated spike amplitude randomly (hatched bars).

Although the classically trained flies do modulate their spike amplitude in response to the heat as does the operantly trained group (Fig. 20B), the difference fails to reach a level of confidence of p<0.05 at t2. However, if one compares all three test-periods (nos. 5, 8 and 9) with the respective control values the modulation comes to lie at a reliable p<0.03.

The t2 amplitude indices of the classical test group show no significant deviations from those of the operant test group (p=0.35). Furthermore, these amplitude indices are in both groups positively correlated with the respective preference indices at t2 (Table 5) and before as well (data not shown). Most importantly, Drosophila seems also to modulate the spike amplitude spontaneously: even in the control groups is the amplitude index highly correlated with the respective preference index (Table 5).

In addition to spike amplitude, the flies use the timing of spikes to avoid the heat in the standard experiment: as noted above, often a volley of spikes is generated to get the fly out of the heat (Fig. 21). Comparing the spike latency in the heated sectors with that in the non-heated sectors, shows the reaction to the heat: the time until the first spike is significantly reduced in the 'hot' sectors (Fig. 21A).

This behavior is maintained even when the heat is permanently switched off. This is not significant for the last two periods alone, but for all three test periods a p<0.041 renders the effect reliable.

Since the latency indices of the operantly trained flies were already very low at t2, it is not surprising that the latency indices of all three test periods from the classically trained batch rise only slightly above control-level (Fig. 21B). However, they could neither be distinguished from the operant batch with a significant reliability.

Figure 21A
Figure 21B
Fig. 21: The two test groups (dotted bars) show mean latency indices that suggest a modulation of the reaction time until the first spike in response to heat (see text). Most suggestive are the high training bars in the operant group (A). In the classical group (B), only in the last period, the first spikes are produced quick enough in quadrants with the heated pattern orientations to bring the index discernibly above the control group (hatched bars).

Interestingly, both groups show different correlations with the preference indices in the respective periods (Table 5). While there is some correlation in the operant groups - indicating that the latency until the first spike has a certain predictive value for avoidance and learning in these flies - there is no correlation between the latency indices and the preference of the classically trained flies at t2. This seems to be an effect of the open loop pattern presentation, since this is also revealed in the classical control group (Table 5).

Even if the timing of the first spike were not modulated in response to the heat, the fly still has the possibility to make all subsequent spikes quicker to avoid the unpleasant flight direction. In the standard experiment, the interspike intervals seem indeed to be shorter in heated sectors than in 'cold' ones: flies that were able to control the appearance of the reinforcer produce shorter interspike intervals in 'hot' sectors compared to the control flies, even if the heat is switched off for the last two periods (t2, p<0.001; p=0.336 at t1; Fig. 22A).

Figure 22A
Figure 22B
Fig. 22: Mean ISI indices for all four groups (N=100 flies each). Drosophila produces torque spikes more quickly when it is heated in closed loop (A, dotted bars). If the heat is switched off, the flies still generate spikes with shorter interspike intervals in quadrants with the 'hot' pattern orientation. The control group, which did not receive any reinforcement only showed random modulation of the interspike intervals (hatched bars).

In the classical groups (B), however, statistical analysis could not establish a significant difference between the test (dotted bars) and the control group (hatched bars). Nevertheless, the values are qualitatively in the proper range to assume that they perform the same behavior as the operant group, although to a lesser degree.

Even if the plotted ISI indices of the classical test group (Fig. 22B) seem rather well in accordance with the findings in the operant test-group, they fail to rise above the required significance niveau. Nevertheless, both pre-test values being in the negative range and all test-values in the positive and above the control-values suggest a qualitatively similar although quantitatively less strong effect. Moreover, the corrected ISI indices at t2 (mean ISI index of the pre-tests subtracted) cannot be distinguished statistically from the respective values in the operant group (p>0.06).

As the amplitude index, the distance and latency indices of the two test groups are highly correlated with the preference index at t2 (Table 5). This is also the case for the spontaneous behavior measured in the control groups (Table 5).

Having demonstrated that at least the operantly trained flies indeed generate spikes more quickly in those quadrants associated with the heat, one can conclude that the spike frequency (number of spikes per time) is elevated as well, which can be seen in fig. 23.

Figure 23A
Figure 23B
Fig. 23: The mean number indices in the operant groups (A) clearly reflect the behavioral strategy of making many spikes in previously punished flight directions: whereas the control group (hatched bars) fails to produce other than random spike frequency modulations, the test group (dotted bars) shows directed spike frequency modulation in the predicted way.
This is also visible in the classical groups (B), however to a lesser degree (see text).

At first sight (Fig. 23B) the response of the classically conditioned group seems to be the same as in the operant group. However, it fails to surpass control-levels to a sufficient degree (p=0.17 for all three test periods). However, omitting period 9 with the strikingly high control value, the modulation exceeds the control values of periods 5 and 8 (p<0.02).

Paralleling the conditions for the interspike intervals, the number indices of all four groups correlate with the respective preference indices (see Table 5) and the corrected number indices in the classical test group do not differ significantly from the values in the operant batch (p>0.05).

Taking together all the data presented so far, it seems as if the major reason why the effects in the classical group appear to be less significant than in the operant group, is the unusually low preference in the two pre-test periods (t1) of the classical group, together with the unusually high spontaneous preference in the periods 8 and 9 (t2) of the corresponding control group. The finding, however, that the values of the classical group are not significantly different from the operant test group either is taken as evidence for the hypothesis that the effects are indeed of the same nature as in the operant group. Moreover, if one expects the control groups to exhibit symmetrical behavior, i.e. zero values in all indices, then one can test the index values of period 8 with a Wilcoxon Matched Pairs Test against zero. This test yields significant effects for the number index (p<0.01, period 8) and the amplitude index (p<0.03, period 8). The ISI index is very close to significance (p=0.067, period 8).

Previous Top Next