This paper examined the use of eye pupil diameter to help in the evaluation of the effects of different dialogue behaviors on the cognitive load of the driver.
The results for four of six drivers support the authors’ hypothesis that pupil diameter can be used to identify major changes in cognitive load during a dialogue. Figure 3 indicates that for these drivers the pupil contracts by about 0.25 mm in the 4-5 seconds after the end of a Taboo game. Note that this effect size is similar to what authors observed when exploring structured verbal tasks, as well as when the study explored pupil diameter changes during an aural vigilance task. These results are encouraging and indicate that using pupil diameter might be a viable approach to estimating the effects of dialogue behaviors on cognitive load changes. Future efforts on this front should focus on collecting and processing large corpora of human-human dialogues and accompanying pupil diameter measurements; however, before such corpora are to be collected, researchers need to carefully identify potential confounding factors, such as effects resulting from the driving task, from the structure of the verbal task, and other effects on pupil diameter such as those due to changes in lighting. (Publisher abstract provided)