Figure 2. ITD statistics predict human ITD-change detection thresholds.

(A) Hypothesis (top) and null hypothesis (bottom) of an adapted neural code underlying human ITD discrimination. (B) Classic study by Mills, 1958 estimated the minimum azimuth change detection across frequency and locations for sounds in free-field averaged across subjects; these measures were converted to threshold dITD as a function of reference ITD (left). Scatter plots on the middle and right show free-field dITD thresholds as a function of ITDrc and √FI_ITD. (C) Test conducted in the present study to specifically assess dITD thresholds for tonal sounds delivered through headphones (dichotic stimulation). Left, mean dichotic dITD thresholds over subjects as a function of reference ITD across frequency. Middle, dichotic dITD thresholds as a function of ITDrc. Right, dichotic dITD thresholds as a function of √FI_ITD. Bars indicate 50% confidence intervals of mean dITD thresholds. Black lines represent power functions fit to all the analyzed frequencies (solid) and excluding 250 Hz frequency from the analysis (dotted).