Illumination Effect on Psychophysiological Measurements in the Automatic Assessment of Multimedia Exposure ("Advertisement Engagement" Dimension)
Literature  shows that increases in task demands generally lead to increases in pupil dilation across the cognitive control domains of updating, switching, and inhibition. Based on these findings alone it cannot be decided whether this reflects an increase in task demands or an increase in actual effort exertion since it at best reflects a psychophysical marker . Researchers emphasize the effect that external stimuli can have on pupil dilatation, and well- controlled lab conditions are always required to prevent alternative explanations . The goal of our research was to determine if measures of predicting exposure can be found by measuring signals of pupil dilatation at rest and under positive load. We have hypothesized that based on the standardized pupil signals, we are able to find statistically significant correlation between pupil signals and answers from questionnaire of four exposers. The purpose of this study was to determine if pupil signals could replace questionnaires in the future researches.
Two user studies were designed and conducted in the LUCAMI lab at UL FE. Answers on advertisement engagement in the form of questionnaire and psychophysiological signals such as pupil size and heart rate were collected from 60 participants while watching specific multimedia content. We got data of normal pupil signals, pupil signals under positive load, and results from four exposures (engagement, psychological reactance, awareness and attitude (AA), and purchase intention). Then we calculated correlations between exposures and pupil signals (standardized and unstandardized) with the aim to detect any statistically significant relations between signal and exposure. We estimated Kendall tau (due to non-normal distribution of data) and the p-value was set at 0.05 %.
We made altogether 12 scatter plots (31 participants provided useful results) and measurements of correlation. The highest correlation among all of them was between the average normal signal and AA (Kendall tau: 0.238 and p-value: 0.063). Based on the results, we rejected the hypothesis because we did not find any statistically significant correlation (with standardized or unstandardized data) between pupil signals and exposures.
Our further research will be directed in search of pupil dilation features that can better correlate with exposures and are therefore better candidates for prediction models. Researchers should also include some other physical signals and make combined models, which will be able to better predict exposures.
 P. van der Wel, H. van Steenbergen, “Pupil dilation as an index of effort in cognitive control tasks: A review” in Psychonomic Bulletin & Review, vol. 25, no. 6, p. 2005-2015. February 2018.