Fig. 4. Dishabituation of DCMD responses between presentations of approach sequences. (A) The number of DCMD spikes for each first approach according to the order of presentation of a sequence (i.e. the order of which the randomized sequences were actually presented, see Table 1). The number of spikes (mean ± S.D.) was lower for approaches of a `locust' than for a `bird'. A Kruskal–Wallis ANOVA on ranks showed that there were no significant differences in the number of spikes within an object type (i.e. `locust' or `bird'). In (B) the data were normalized to the first presentation of a particular object and trajectory. For example, the first approach for each sequence of a simulated locust approaching from 0° azimuth was normalized to the first time that approach was presented. If there were no confounding effects of incomplete dishabituation then the normalized values should not be different from 1. A Kruskal–Wallis ANOVA on ranks showed that there were no significant differences in the normalized number of spikes within an object type. These data show that the DCMDs were fully dishabituated prior to each presentation sequence. (See text for statistical parameters; bars with the same letters were not statistically different, N=11.)