Comparison of absolute latency, first inter spike interval, rate and relative latency. (A) The tuning curve for absolute latencies of the whole population over all stimuli showed clear velocity dependence. The median of the fastest responses was 22ms at 6m/s for both directions. (B) In the tuning curves of the whole population for the fist inter spike intervals and (C) spike rate no velocity dependence was observed. The response variability to up- and downward movements, however, differed. The upward movements showed higher variability both for first ISI and rate. (D) The median of the relative latencies remained rather constant at around 10ms for all velocities. (E) – (H) Velocity estimation with the maximum a posteriori (P) estimator on basis of absolute latencies (E), first ISI (F), spike rate (G) and relative latencies (H) with the combined data from all experiments. Perfect estimation would result in a diagonal line of pixels with 300 correct estimations each. The grey scale indicates the number of estimations. Completely random estimation would result in a grey area with a grey value of 13.64. Red crosses indicate the calculated centres of mass of the estimated velocities that were calculated for all presentations of each velocity. (E) Velocity reconstruction on basis of absolute latencies was close to perfect with a maximum of 205 correct estimations for a stimulus velocity of 1m/s. Centres of mass show that high velocities were slightly underestimated. (F) Velocity reconstruction based on the first ISI was not possible. Upward movement was overrepresented in all cases. (G) In the case of spike rate at least the correct direction could be reconstructed in many cases. Velocity reconstruction, however, was not possible. With relative latencies (H) reconstruction was inferior to the estimation on basis of absolute latencies (E) but clearly superior to the reconstruction based on first spike interval (F) and rate (G). Again the calculated centres of mass (red crosses) deviated mainly for high velocities. In (I), (J) (K) and (L) the reconstruction results from the best (red lines) and the worst (blue lines) trials are compared. Best results of 100% correct estimation were only observed in velocity reconstruction based on absolute latencies (I) followed by estimation with the relative latencies (L) with 18 correct estimations out of 22 velocities. Note that axes for all graphs are not linear, as the velocity steps of the stimulus differ. Data are based on 5 experiments from 5 archerfish with a total of n = 109 recorded ganglion cells.