Towards an Accurate Measure of Emotional Pupil Dilation Responses: A Model for Removing the Effect of Luminosity

Pupil dilation is a fundamental marker of emotional response, and indicates emotional arousal independently of stimulus valence. As such, pupil dilation provides invaluable insight into emotional engagement and arousal levels. One of the challenges faced when studying emotion response through pupil diameter is distinguishing between dilation caused by light and dilation caused by emotion. This is particularly challenging when examining emotional responses in individuals viewing videos, where luminosity changes constantly occur. To solve this problem, we propose a model that accurately predicts how pupil size changes with brightness, taking into account the nonlinearity of the relationship. Since effects of luminosity and emotion are additive without interacting [1], pure emotional effects on pupil size may be isolated by subtracting the model estimate from the total pupil size recorded in response to the visual stimuli. The general structure of the model was developed from findings in the literature and by analysing data collected from seven participants. The model was then tested on 10 subjects and different monitor models. We obtained an average error of 6.69% (SD:1.05%) and a maximum error of 23.29% (SD: 3.66 %) between the actual pupil size and the pupil size predicted by the model. To the best of our knowledge, this is the only nonlinear model that has been validated on a sample of subjects. This research lays the groundwork for the accurate capture of emotional responses from pupil diameter under varying lighting conditions.