Figure 8: MTF of eSTRFs.

The MTF (or modulation gain) of each eSTRF was calculated, then averaged across conditions. eSTRFs were estimated separately for temporal-filtered stimuli and for spectral-filtered stimuli. (a) The mean MTF for eSTRFs is shown for the BEFORE (1st column) and AFTER (2nd column), condition. Lines represent 5th, 15th, 85th and 95th percentiles. The difference in eSTRF was calculated, and the MTF of this difference is shown in the 3rd column. The MPS of the actual filtered speech stimuli is shown in the 4th column for comparison. Top row are eSTRFs fit on temporal-filtered stimuli, bottom row are eSTRFs fit on spectral-filtered stimuli. (b) The difference in the MTF for the two filter types (temporal–spectral) was calculated for each electrode. Z-scores for this difference are shown for each condition. The MTF of each eSTRF matches its respective filter type (cluster-based permutation t-test across electrodes, n=41, BEFORE P=0.001, AFTER P=0.003, AFTER–BEFORE P=0.001), suggesting that tuning changes emphasize spectrotemporal modulations that are present in the degraded speech sound and are crucial for speech intelligibility. Note that these features were also present in the BEFORE stimulus (since BEFORE and AFTER stimuli were identical) and in the BEFORE eSTRF (as shown in the MTF for the eSTRF BEFORE, left column). This suggests that neural tuning in the BEFORE condition does match filtered speech features, and that this is accentuated after hearing the unfiltered speech.