LFP and MUA signals were averaged across trials. CSD was calculated from LFPs by numerical differentiations to approximate the second order spatial derivative of the LFP. One channel at the depth of layer 4 was selected for further analyses. Mean amplitudes were estimated during a post onset response period (10 ms) during which MUA increased and CSD and LFP signals deflected downward, and baseline amplitudes (−30~−5 ms from the stimulus onset) were subtracted before derivation of tuning curves. The best frequencies (BFMUA, BFCSD and BFLFP) and the tuning bandwidths (BWMUA, BWCSD and BWLFP) were estimated from tuning curves. To quantify tuning curves across recording sites, curves were normalized by their peaks, and were further shifted on the frequency axis to align the BFMUA to zero. The amplitudes of LFP responses to BBN were measured at 24 ms post onset of sound and baseline subtracted at each recording depth. For each penetration site, the distribution of amplitudes was normalized to the mean of absolute amplitudes across depths. To quantify normalized amplitude distributions, the median values and 95 % confidence intervals (bootstrap) were derived. Band-limited signals were calculated using wavelet transform. See Supplemental Experimental Procedures for details of these analyses.
