**A.** Relationship between amplitude of the early MR and AR_{32}, for individual cells, sorted by cell class. Mean spike rate was measured in a 500 ms window starting 150 ms after stimulus onset (MR) or offset (AR_{32}), thereby excluding the transient on- or off-response. Filled points indicate that there was no significant difference between MR and AR_{32} amplitude. Distributions were neither normally nor log-normally distributed (Shapiro-Wilk test, p < 0.05) in the majority of cell classes. Spearman’s correlation, r values: untuned 0.20, simple non-DS 0.41*, simple DS 0.56*, complex non-DS 0.90**, complex DS 0.88** (*correlation significant, p < 0.05; **correlation significant, p < 0.001; non-starred values not significantly correlated, p > 0.05). **B.** Relationship between decay time-constants of MR and AR_{32} for individual cells, sorted by cell class. Decaying exponential curves were fit to PSTHs (bin size 100 ms) from a start point 200 ms after the stimulus onset or offset, thereby excluding the transient on- or off-response. Only cells with AR_{32} significantly above their spontaneous firing rate, and which were adequately fit by decaying exponential functions are included. Raw data distributions failed the Shapiro-Wilk test for normality (p<0.05) for all but the untuned class, but a logarithmic transform of the data yielded normal distributions throughout, Shapiro-Wilk test, p>0.05 for both AR_{32} and MR for each cell class. Hence we have plotted the data on logarithmic axes. Pearson correlation, r values for logarithmically transformed data points: untuned −0.20, simple non-DS 0.48, simple DS 0.20, complex non-DS 0.73**, complex DS 0.80** (**correlation significant, p < 0.001; non-starred values not significantly correlated, p > 0.05).

## PubMed Commons