Interestingly, it appears that one of the estimates of an LFP spread of ~250 μm likely arrived at this estimate by a subtraction procedure whose effect was not unlike that of CSD analysis [i.e., subtracting the mean response across all orientations from the response to a single orientation (Katzner et al., 2009)]. In fact, when we performed this same manipulation on our CSD and LFP tuning curves, we obtained “sharpened” tuning curves, with bandwidths equivalent to those of MUA (Fig. 7). Thus, the mean subtraction artificially sharpens the tuning of the LFP, leading to the conclusion that the LFP itself spreads over a much smaller distance than it actually does (i.e., that the “undifferentiated” LFP is extremely local in its extent). A more subtle collateral effect of the subtraction is that weak positive responses to non-preferred orientation stimuli may become negative responses as if they were inhibitory. Ultimately, when understood in proper context, the findings of Katzner et al., (2009), support a central conclusion of this study: differentiation procedures may confound the analysis of LFP spread on one hand, but on the other hand, they are useful in refining the localization of the LFP and in defining its “spatial domain.”
