Recent Advances in Eye Tracking – Part 2

Last year we have participated in multiple interesting conferences focusing on research methods in general and eye tracking in particular. In the previous article I have shared some methodological topics that I found interesting from those events. In this article I will delve into a more technical topic, namely the definition of fixations.

Many researchers using eye tracking methodology know what fixations are: behaviorally, a fixation is the maintenance of gaze on a specific object or location in order to keep its reflection on the fovea. To change the object being looked at, we perform a voluntary type of eye movements that is know as saccades.

Operationally, things can get more complicated in order to separate those two types of eye movements. The most common way to identify fixations in eye tracking data is by calculating sample-to-sample velocities and labelling samples that have a velocity below a certain threshold as fixations while those that exceed the threshold are marked as saccades (here you could check an example of how to do this in R). A paper by Salvucci and Goldberg (2000) summarizes this and other three methods for identifying fixations.

At this point, one could say that fixations are the opposite of saccades, or the lack of eye movement, and, for the sake of simplicity, often times that is sufficient. However, this is not very accurate! In her talk about the topic at the Eye Tracking Research and Applications conference in Warsaw (ETRA, 2018), Prof. Martinez-Conde discussed the fact that, during fixations, eyes are not actually stationary. Our eyes are in continuous motion, performing very small eye movements when we are fixating an object, known as micro-saccades, and moving from one object to another more rapidly by means of saccades.

In order to observe the minute eye movements during fixations, one needs to use fast and accurate eye tracking devices. Of course, one option could be a scleral search coil system, but that would take really willing participants! Luckily, nowadays we have much less invasive solutions that rely on corneal (near-)infrared reflection to achieve the same task with very good accuracy and sampling rates that go up to 2000 Hz.

Until a short time ago, the clear choice in the scientific community was the Eyelink 1000 system by SR Research. It is fast, accurate and I have used it to study saccades during my master studies with no problems. A couple of years ago, Tobii has made a remote eye tracker that can reach up to 1200 Hz called Tobii Pro Spectrum. During the Scandinavian Workshop on Applied Eye Tracking (SWAET, 2018), Prof. Nyström presented a study in which they compared this (relatively) new eye tracker with the EyeLink 1000 Plus in the study of microsaccades. Their results showed that both systems had comparable rates and amplitudes of microsaccades and both had similar horizontal precision. They concluded that, in fact, the Tobii Pro Spectrum is a valid instrument for the study of microsaccades.

Finally, another interesting talk during SWAET was about the importance of reporting the different methodological aspects of an eye tracking study. Dr. Roy Hessels showed how researchers define certain concepts, like fixations and saccades, in different ways. The authors stress the point of making “definitions more explicit“ in research reports, describing the task well, how eye movements are affected by the task and how is the data processed.

I will end this article with an observation on fixations; The fact that there is no such thing as “not moving the eyes;” Rather, our eyes are in continuous motion. The difference between “fixations” and “saccades” is basically how fast these movements are.

Iyad Aldaqre

Data Scientist at SR Labs Srl

References:

Hessels, R.S., Niehorster, D.C., Nyström, M., Andersson, R., & Hooge, I.T.C. (2018) Is the eye‐movement field confused about fixations and saccades?
A survey among 124 researchers. Presented at the Scandinavian Workshop on Applied Eye Tracking (SWAET), Copenhagen, Denmark, 23-24 August.

Martinez-Conde, S. (2018) From fixation to exploration: Towards an integrative view of oculomotor function. Keynote presented at the Eye Tracking Research and Applications conference (ETRA ’18), Warsaw, Poland, 14-17 June, ACM, New York, USA.

Nyström, M., Niehorster, D.C., Andersson, R. & Hooge, I.T.C (2018) Is the Tobii Pro Spectrum a useful tool for microsaccade researchers?. Presented at the Scandinavian Workshop on Applied Eye Tracking (SWAET), Copenhagen, Denmark, 23-24 August.

Salvucci, D.D. & Goldberg, J.H. (2000) Identifying fixations and saccades in eye-tracking protocols. In: Proceedings of the 2000 symposium on Eye Tracking Research and Applications (ETRA ‘00), Palm Beach Gardens, FL, USA, 6–8 November, ACM, New York, USA.