FIGURE 9.
Simulation 7: model fits of a dataset with perturbations applied to fundamental frequency in both shift-up and down directions [normalized by an unshifted control condition; data from Heller Murray (2019)]. Shift-up and down data are fit simultaneously. Measurement of fundamental frequency was taken early in the trial (20–120 ms after voicing onset). (Left) Mean and standard error of experimental data in blue; model fit in red. (Right) Fit quality and optimized parameter values (r = correlation coefficient; αA = auditory feedback control gain, αs = somatosensory feedback control gain, and λFF = learning rate).