Comparison of different experiments. The 5 experiments showed differences in reconstruction quality based on absolute latencies (A to E). Experiments 1 - 3 (A - C) showed very good results while in experiments 4 and 5 (D + E) many errors occurred by confusing speeds with similar absolute latencies (compare to N and O) and/or opposite directions. The best trials (red lines) showed nearly perfect reconstruction in experiments 1-3 (A - C) and even the worst trials (blue lines) estimated the correct velocity far above chance level. For experiments 4 and 5 this was not the case. (F - J) Qualitatively similar results were obtained for velocity reconstruction based on relative latencies in experiments 1 - 3. However, the overall reconstruction quality slightly deteriorated. Reconstruction worsened especially for upward movements in experiment 2 (G) and for downward movements in experiments 1 and 3 (F + H). In experiment 4 (I) only direction reconstruction was possible. Velocity reconstruction with data from experiment 5 was not possible at all (J). (K – O) The tuning curves of the latency from stimulus onset to population response onset separated for the single experiments show higher variability in experiments 4 (N) and 5 (O), compared to experiments 1 to 3 (K – L). This indicates that the timing of the population response and therefore the single cell responses are less precise. Chance level is at 2.73 (4.55%) for each pixel in the single experiments. The number of analysed cells in each experiment is indicated on the left side.