I am a reading researcher, and I started out my career in eye-tracking in the Autumn of 1999 on a cloudy day. I do not actually remember the weather that day, but it was in Belgium, so cloudy is a pretty solid guess. I had just started my internship in the final year of my Masters (called ‘licentiaat’ at the time) under the supervision of Marc Brysbaert at Ghent University. Marc took me to a room that contained several boxes, and we started unboxing what was going to be the focus of my academic life for quite a while: The EyeLink 1 eye-tracker.
Back in those days, the software support for the EyeLink eye-trackers was not yet what it is today, and the first months of my internship were therefore spent programming software in by-now ancient computer languages such as Borland C and Windows Visual C++. I programmed code for running reading experiments and analyzing the results. Nowadays you can download software that does this for you and the EyeLink eye-trackers have become particularly user friendly, especially for reading researchers.
However, the current user friendliness does come somewhat with a cost. I learned so many basic facts about eye movements from having to work with data that was so much more “raw” than what my students work with nowadays (Note that I do not wish for my students to have to do what I did. After a few months of programming, I discovered that I was maybe not too bad at it but also that I most definitely did not want to become a full-time computer programmer!)
The authors of the paper I am writing this blog about, are part of a strong methodological tradition to examine just about any assumption ever made about eye tracking. And this research line is even more important now that we have excellent software packages available for processing eye movements but that inadvertently increase the distance between the researcher and the raw signal from the eye-tracker. Packages that quite often obscure some of the choices that have already been made on how to process the signal that comes from the eye-tracker.
From that perspective, I was very happy to read the recommendation of the authors to show your students the eye tracking record in its most pure version. I always show my students the output of the EyeLink which just contains a computer time stamp, an x and a y coordinate and the pupil size. As a reading researcher, we work predominantly with fixations, and it is always brilliant getting the students to understand how the reported fixation location is actually made up of hundreds of samples that are surprisingly messy. This little exercise also shows them how much pre-processing goes on before they start working with the data in the form they typically get.
The current paper is part of that tradition of questioning assumptions in eye tracking by examining operationalizations, the process of quantifying a phenomenon that is not directly measurable. Multiple phenomena are described such as workload and mind wandering and the various ways in which these phenomena have been interpreted from the presence of certain eye movement behaviors such as pupil diameter or how deeply the processing of a specific word shows up in the fixation time on it.
I will go back in time once more to illustrate another reason why I like this paper. It was in the first year of my PhD when my supervisor left to take up a position at Royal Holloway, University of London. I fortunately had just acquired an independent fellowship from the Fund for Scientific Research Flanders allowing me to continue my PhD on the topic of eye movements during reading. Marc remained the unofficial supervisor and was still a very good supervisor albeit only via email. I was the only one working on eye movements in Ghent at the time, so the focus of this series of articles on individuals who are the first to start eye-tracking research feels familiar to me. As I was a bit isolated, I started applying for travel grants so I could spend time in labs that possessed more technical and theoretical expertise. So, it came about that in 2004 I spent the first of many research trips at the University of Massachusetts (UMass) in the States working with the great Keith Rayner and Sandy Pollatsek, who unfortunately both passed away.
During my time at UMass, one of the activities I enjoyed most was the reading group we had every two weeks in which we dissected and discussed a research paper. I was particularly fond of the one we always had after the Psychonomics conference in which everyone had to nominate a talk they saw for the best and worst (!) talk of the conference. But most of the sessions were on individual papers and those sessions taught me so much that came in handy when for instance I started reviewing and editing papers myself for journals.
I will be looking forward to talking about this specific paper on operationalizations in future reading groups. It will be brilliant for my students to question their assumptions and together with the other papers in the series bring them so-to-speak closer to the raw output signal from the eye-tracker. I also really like that as a reading researcher, this paper shows how researchers from different fields grapple with the same issues, such as drawing appropriate interest areas to analyze eye movement behavior. And as a result, feel part of that much bigger community of eye movement researchers.
The saying goes that science is standing on the shoulder of giants. As an eye movement in reading researcher, I stand on the fundamental methodological research that underpins eye tracking. The researchers who created this paper are such giants (literally, because some of them are quite tall).
Featured Psychonomic Society article
Hooge, I.T.C., Nuthmann, A., Nyström, M., Niehorster, D.C., Holleman, G.A., Andersson, R., & Hessels, R.S. (2025). The fundamentals of eye tracking part 2: From research question to operationalization. Behavior Research Methods. https://doi.org/10.3758/s13428-024-02590-2
