As the luminance of screens will differ by model and settings, you would need to measure the luminance of the screen. Most people in visual psychophysics use light meters to do this, but they aren't cheap. You could probably borrow one from a physics/psychophysics lab.

Hey Kathrin!

I agree with Ben - and now comes the obligatory BUT: You'll need a close angle lens to measure the whole screen from a distance, since you probably have different kind of non-uniform stimuli (for which the luminance is not constant). If you have CONSTANT surrounding conditions (same lighting for every TP, e.g. no daylight changes etc.) than you could in principle compute the luminance of the screen with your favorite image processing program (e.g. just average all pixel values). This will not give you a correct luminance, but a fair estimate, which you could correlate with pupil diameter (or at least use it to normalize your data somehow).

Greetings, David

I have been dealing with this same issue in my own research. Quantifying pupillary light reflex within video scenes is a very tough job. There are video light meters, but they are neither accurate or precise, especially over short measurement time periods and small spatial areas.  And then there is the task of linking each subject's pupil measurements at gaze locations in time with luminance measurements in time and space.   One approach to this problem is to measure average pupil size across the experimental (emotional) video segment, and average pupil size in a control condition (e.g., a blank screen with a constant luminance level).  Then compute the ratio of experimental to control condition, a measure of the pupillary reaction to the emotional stimuli due only to the evocative content of the experimental stimuli. 

You can take a look at our paper. We use a simple linear model to factor out light reflex. Still a work in progress, but the method works!

Pallavi Raiturkar, Andrea Kleinsmith, Andreas Keil, Arunava Banerjee, and Eakta Jain. 2016. Decoupling light reflex from pupillary dilation to measure emotional arousal in videos. In Proceedings of the ACM Symposium on Applied Perception (SAP '16). ACM, New York, NY, USA, 89-96. DOI: https://doi.org/10.1145/2931002.2931009

It is a hard task, but it can be achieved by several means (e.g., light meter, calculating the mean pixel values (for each color channel) per frame, etc.). In our case, we controlled the effects of low-level features on pupil diameter (i.e, luminance, contrast) through a mosaic-scrambled method. This mosaic-scrambled approach preserve the low-level features while removing the video's semantic content. So you present 2 versions (the original one vs one scrambled). The difference in pupil size (corrected for baseline) between the two video versions is due to semantic content. The scrambling procedure was made on a 5 by 4 matrix through rotation along the 3 orthogonal axes (x, y, and z) using MAGIX Movie Pro X5 software (MAGIX Software GmbH, Berlin, Germany). You can find everything here:

Carvalho, J. & Rosa, P. J. (2020). Gender Differences in the Emotional Response and Subjective Sexual Arousal Toward Non-Consensual Sexual Intercourse: A Pupillometric Study 17(10), 1865-1874 https://doi.org/10.1016/j.jsxm.2020.06.018