- Lennie analyzes time until cumulative spike count can reliably yield detection. I think this measure (more relevant to behavior on individual trials than any feature Maunsell points out in average PSTH) may yield greater effect of lum on latency, see Fig 5. For a dim light, you have to wait a very long time before integrated response reliably emerges from background noise. In retina anyway, time until detection really much more reflects the low peak spike count, that you have to integrate for longer with dimmer light
- Lennie (1981) makes the point that when he increased background illumination, responses became more transient. Maybe this could decrease the effect of lum on latency because neural resp has to reach threshold quickly or not at all?