El-Boustani et al. reply

replying to S.-H. Lee, A. C. Kwan & Y. Dan Nature508,http://dx.doi.org/10.1038/nature13128(2014) Several recent studies have examined the function of parvalbumin-expressing (PV + ) and somatostatin-expressing (SST + ) inhibitory neurons in V1 (refs 1 , 2 , 3 ). Although it is commonly agreed that these cell types alter the responses of pyramidal neurons in distinct ways—via divisive or subtractive inhibition—their specific roles remain a matter of debate. The Comment by Lee et al. 4 presents new data suggesting that the differences between the results of Lee et al . 2 compared to Atallah et al . 3 and Wilson et al . 1 could be explained by the strength and duration of laser stimulation used to optogenetically activate these two classes of inhibitory neuron. The data presented by Lee et al . 4 now clarify that PV + neurons, when probed with small amounts of optogenetic activation, do not significantly change the tuning of their target cells, confirming Atallah et al. 3 and Wilson et al . 1 . The new SST + results presented in the Comment 4 show that SST + neurons can subtract responses, consistent with Wilson et al. 1 , but we suggest that the switch of function of SST + neurons in their data between short (1 s) and long (4–5 s) stimulation reveals a core principle of inhibition in cortical networks rather than simply being a peculiarity of stimulation protocols. The fundamental difference between these two conditions resides in the temporal overlap between inhibitory neuron activation and target-cell responses: when these overlap, inhibition is divisive (causing no change in tuning width of target neurons), but when they do not overlap, inhibition is subtractive (and reduces tuning width).