This erratum concerns Fig. 4 in the previously published paper (Attia et al., 2021). Due to a programming error, the values of [0.07 vs 0.14 m.d.u.] were reversed between the left and right panels. The corrected figure and legend are shown below.
Predictions of the modulation-filterbank model for the “double-pass” AM detection task. A 2-Hz sinusoidal AM was used as the target to illustrate the model's behavior for this task. The SD of the modulation-noise masker, , was set to 0.07 m.d.u. (A) and 0.14 m.d.u. (B). The model was simulated for several levels (i.e., SD, expressed in model units, m.u.) of the additive internal noise. Each dot was calculated from 200 trials. For each level of internal noise (as indicated by the hue code, dark blue corresponding to the highest level), double-pass consistency (percent agreement, PA) and performance (percent correct detection, PC) were simulated for AM depth varying between −40 dB (1%) and –3 dB (70%) in 1–5 dB discrete steps. For a given PC level, PA decreases with increasing levels of internal noise and it increases with increasing levels of external modulation noise.
Predictions of the modulation-filterbank model for the “double-pass” AM detection task. A 2-Hz sinusoidal AM was used as the target to illustrate the model's behavior for this task. The SD of the modulation-noise masker, , was set to 0.07 m.d.u. (A) and 0.14 m.d.u. (B). The model was simulated for several levels (i.e., SD, expressed in model units, m.u.) of the additive internal noise. Each dot was calculated from 200 trials. For each level of internal noise (as indicated by the hue code, dark blue corresponding to the highest level), double-pass consistency (percent agreement, PA) and performance (percent correct detection, PC) were simulated for AM depth varying between −40 dB (1%) and –3 dB (70%) in 1–5 dB discrete steps. For a given PC level, PA decreases with increasing levels of internal noise and it increases with increasing levels of external modulation noise.