Figure 3

Pupillometric estimate of harmonic tone discrimination thresholds. (a) Schematic spectrograms of standard (blue) and deviant (gray) stimuli. Color intensity corresponds to amplitude of each frequency component. (b) Schematic of the oddball paradigm used to probe harmonic tone discrimination. (c) PD changes evoked by deviant stimuli in one subject. Color gradient corresponds to the F0 difference between the deviant and standard F0s. Lines correspond to mean PD and shading corresponds to ± 1 s.e.m. Red line corresponds to stimulus onset. (d) Same as (c) but pooled and averaged over 7 subjects. Green line and shading corresponds to pupil trace at threshold (ΔF = 0.75 semitones). Teal dashed lines correspond to the GCA window. (e) Average PD changes computed in a 1-s window beginning 1 s after stimulus onset as a function of deviant F0 (whiskers correspond to ± 1 s.e.m.). Significant changes in pupil diameter were observed for deviants with F0s 0.75 semitones or greater than the standard F0 (green; ***: p < 0.0001, ANOVA followed by Bonferroni-corrected post-hoc tests; exact p-values in Supplementary Table 1). Line is linear fit. (f) The percent of trials (circles) in each deviant condition that showed a significant pupil dilation compared to standard stimuli. Line is linear fit. (g) GCA fit to the rising phase of PD changes (teal dashed lines in (d)). Dots are mean pupil diameter in 100 ms time bins, whiskers correspond to ± 1 s.e.m. Solid lines correspond to mixed-effects model fits. Colors as in (d). (h) GCA weight estimates. Colors correspond to the weights of the intercept (blue), slope (red), and acceleration (purple) terms used to fit the PD traces. Asterisks denote statistically significant regression weights (exact p-values in Supplementary Table 2). Significant differences in pupil traces were observed for deviants with F0s 0.75 semitones or greater compared to standard F0s.