Human discrimination and modeling of high-frequency complex tones shed light on the neural codes for pitch
Fig 7
Ideal-observer predictions for pure-tone frequency discrimination.
Results of the frequency discrimination simulations. (A) Simulated FDLs versus frequency for a pure tone in each auditory-nerve model. Simulations in this panel include no parameter roving. Points indicate the simulated FDLs at a particular frequency while lines indicate a locally estimated scatterplot smoothing (LOESS) fit to the simulated FDLs. The solid black line indicates the predicted FDLs from Micheyl et al. [58] scaled by a factor of 0.002 (to roughly match the low-frequency side of the curve to the best-performing model predictions from the present study). The axis on the right-hand side corresponds to the unscaled FDLs predictions from Micheyl et al. [58]. (B) Simulated all-information FDLs and vector strength (top row) and simulated rate-place FDLs and Q10 (bottom row) versus frequency with a double y-axis. To choose the warping on the y-axis for vector strength and Q10, linear models were fit to predict log-transformed FDLs as a function of log-transformed reciprocals of vector strength (Q10) for the all-information FDLs (rate-place FDLs). The fitted regression equations were then used to warp the y-axes. In other words, we warped the y-axes for vector strength and Q10 to maximize overlap with the FDL predictions (across all three models) in order to visually demonstrate the relationship between vector strength and Q10 and the simulated FDLs. (C) Ratio of simulated FDLs at 8.5 kHz and 2.0 kHz in the non-roved simulation at 30 dB re: threshold for each model (left) and ratio of behavioral estimates of FDLs at 8.5 kHz and 2.0 kHz from various studies (right). Simulated FDLs were interpolated using LOESS while behavioral FDLs were linearly interpolated on log-log coordinates.