Fig. 11. (A) Regardless of the approach interval, the mean time of peak DCMD firing occurred earlier after repeated approaches of a `locust' from ±45 azimuth and was relatively insensitive to repeated approaches of either object from 0° azimuth or a `bird' from ±45° azimuth (N=11). For trajectories and intervals in which the time of the peak was invariant, it occurred 28±27 ms(mean ± S.D.) before collision of a `locust' and 50±40 ms before collision of a `bird'. For clarity the S.D. is shown in only one direction. r² values as in Fig. 5. (B) The S.D. of time of peak firing was not affected by repeated presentations at 34 s intervals of a `locust' approaching along either trajectory or a `bird' approaching from 0° azimuth (left) whereas the S.D. increased during repeated approaches of a `bird' from ±45° azimuth (r=0.429, P=0.019). (Right) For approaches at 4 s intervals the S.D. was not affected by repeated presentations of a `locust' or `bird' from 0° azimuth whereas it decreased upon repeated approaches of a `locust' (r=–0.698, P=0.004) or `bird' (r=–0.551, P=0.033) from ±45° azimuth. For approaches from ±45° (4 s intervals) the rate of decrease of the S.D. (b from the single exponential decay function) was greater during repeated approaches of a `locust (b=0.390) compared to a `bird' (b=0.200).