TABLE 2.

Summary of regression models

Except where indicated, the STMI is calculated for best modulation rates up to 32 Hz and all NSIM measures are computed using 29 CF neurograms

STMI spectro-temporal modulation index without lateral inhibitory network, STMI LIN spectro-temporal modulation index with lateral inhibitory network, MR NSIM mean-rate neurogram similarity measure, FT NSIM fine-timing neurogram similarity measure

p < 0.001 for all fits except as noted

^aThe adjusted R ² is the proportion of variation in the response variable accounted for by the model regressors. However, unlike the R ², it only increases when an increase in explained response variation is more likely than chance when additional regressors have been added to the model

^bThe corrected Akaike information criterion ratio (Burnham and Anderson 2002) is adjusted for a finite sample size. The sample size is 35, which corresponds to the number of average RAU-transformed perceptual scores and data points for each neural measure. An AICC ratio smaller than 0.5 indicates that the model is less likely than the “best” model (gray row of table) to minimize information loss, while an AICC ratio larger than 0.5 indicates that the model is more likely than the “best” model to minimize information loss