Figure 3.

Encoding models trained on the target-onset envelopes. A, Preparation of stimulus envelopes: stimulus vectors were computed by inserting the target tone into a vector of zeros at the time of its occurrence, extracting the envelope and subsequently computing the halfwave rectified first derivative (black line). B, Temporal response functions to the target tone for all three experimental conditions. The inset shows the topography at the negative peak (0.18–0.20 s). C, Source localization for the negative peak of the temporal response function (averaged over conditions, 0.18–2.0 s), suggesting auditory generators.