Comparison of simultaneous measures of pupil dilation, verbal response time and subjective evaluation of listening effort

Listening effort (LE) describes the amount of cognitive and attentional resources required to perform a listening task. Over the last decade, estimates of LE have complemented more traditional measures of task performance accuracy (i.e. speech intelligibility), on account of the need to consider both auditory and cognitive factors when investigating listening processes. Although a direct and comprehensive measure of LE is currently unavailable, several proxy indices have been proposed. Correlations between these quantities are uncommon, suggesting that they may probe different underlying cognitive dimensions. To date it is still unclear how to best quantify LE, and if and how different proxy measures should be combined to provide the most comprehensive description of this complex construct.

This study aimed to identify changes in LE during a speech reception task, using pupil dilation (physiological measure), verbal response time in a single task paradigm (behavioral measure) and self-ratings (subjective measure). Normal hearing adults (N=24) were presented with the matrix-sentence test in the Italian language, using three signal-to-noise ratios (SNR: -3, -6, -9 dB). Direction of selective attention was manipulated across blocks, by making the origin of the speech signal either blocked or random across the two sound sources. In addition, the role of visual cues in detecting the origin of the auditory signal was assessed by visually marking or not the position of the unseen loudspeakers. Data on pupil dilation and verbal response time were acquired simultaneously for each experimental trial.

Although the correlation between subjective and behavioral/physiological measures has been analyzed in recent works, the relation between pupil dilation and verbal response time remained unexplored, due to the different requirements of the two measurement procedures (presence vs. absence of a retention period after the end of the auditory stimulus). No retention period was included in this study, so that a mean pupil dilation was calculated only over the listening phase. Despite this methodological constraint when measuring dilation, the results indicated that all three proxies of LE were sensitive to SNR. In addition, a significant interaction between SNR and vision emerged for response time, and a significant interaction between SNR and attention was found for mean pupil dilation. This indicates that the two quantities were sensitive to the cognitive manipulations included in the study. No significant correlation was found between the three proxy measures, supporting the hypothesis that they analyze different aspects of LE.