To understand the origin of the difference in signal-to-noise ratio (computed as a z-score in the current study) between LF and HF electrophysiological signals, we decomposed the z-score into its ‘signal’ and ‘noise’ constituents. (A) We addressed the lower z-score for HF compared to LF (left plot) by computing face-selective signal and noise collected over all recorded VOTC contacts (N=7374,, excluding contacts in the white matter). We used all recorded contacts to avoid any potential bias resulting from the use of only contacts with a significant response (i.e. having z-score >3.1). To directly compare signal and noise across signals we computed a ‘signal index’ (middle plot) and a ‘noise index’ (right plot) for each recording contact that accounts for the 1/f relationship between EEG amplitude and frequency and differences in the unit used for LF and HF (µV vs. percent signal change). Specifically, the signal and noise indices were computed as follows: (1) fft spectrum was segmented in 4 segments of 51 bins centered around the first 4 harmonics of the face-selective frequency (1.2–4.8 Hz, similarly as for the computation of the z-score); (2) we summed the 4 segments; (3) based on the summed segments we compute (a) the ‘raw face-selective amplitude’ as the amplitude at the face-frequency (i.e. at the center of the summed segment), (b) the ‘amplitude of the baseline’ as the mean amplitudes in the 48 bins surrounding the face-frequency bin (i.e. excluding the 2 bins immediately adjacent to the face frequency bin), (c) ‘the baseline-subtracted amplitude’ as the difference between the raw face-selective amplitude and the amplitude of the baseline, (d) ‘the standard deviation of the noise’ as the standard deviation of the amplitude in the 48 bins surrounding the face frequency; (4) from these values we compute the signal index as the baseline-subtracted amplitude divided by the amplitude of the baseline, and the noise index as the standard deviation of the noise divided by the amplitude of the baseline. This revealed that the face-selective signal amplitude was on average almost 5 times larger for LF compared to HF (middle plot: mean +/- std: 0.83+/-1.34 for LF vs. 0.15+/-0.45 for HF) while the noise was much more similar across the two types of signals (right plot: 0.28+/-0.03 for LF vs. 0.26+/-0.03 for HF). This indicates that the lower z-score for HF compared to LF is driven by a smaller face-selective signal, and not by a higher noise for HF. (B) Similarly, we also compare the face-selective HF signal amplitude (i.e. HF baseline-subtracted amplitude as described for panel A) and HF noise (i.e. HF standard deviation of the noise as described for panel A) across LF+ contacts with or without significant HF face-selective response. These contacts differ in terms signal amplitude (0.40+/-0.87 for LF+ HF vs. 9.32+/-9.13 for LF+ HF) rather than in terms of noise (0.77+/-0.28 for LF+ HF vs. 0.77+/-0.22 for LF+ HF-). (C) To investigate the overall lower z-score in the ATL compared to more posterior VOTC region (left plot, top row: LF signal; bottom row: HF signal), we computed LF and HF signal (i.e. baseline-subtracted amplitude as described for panel A) and noise (i.e. standard deviation of the noise as described for panel A) in 3 main regions of the VOTC (OCC, PTL, ATL), again using all recorded VOTC contacts (excluding contacts in the MTL). This revealed that the mean face-selective signal amplitude within the ATL was 72% (LF) and 92% (HF) smaller than in the PTL (middle plot). In contrast, the noise in the ATL was only 10% larger (LF) or of equal magnitude (HF) than in the PTL (right plot). This indicates that the lower z-scores in ATL are mostly driven by a smaller face-selective signal amplitude in this region compared to the posterior VOTC.