Optical-camera based approaches (e.g., webcams, smart phone) for recording eye movements have historically suffered
from accuracy issues, especially with darker eyes, as it is not straightforward to distinguish pupil from the rest of the iris
based on optical images in real time. However, recent developments in machine learning seems to offer a breakthrough
in this area. A recent publication by Valliappan et al., (2020) compared a smartphone-based eye tracking approach
(without any additional hardware) that they devised to state-of-the-art commercial eye trackers and reported comparable
or better accuracies in a study with 100 participants in oculomotor tasks (i.e., visual search, prosaccade, smooth pursuit),
natural image viewing and reading comprehension difficulty. Similarly, using a webcam-based eye tracking (Webgazer.js
by Papoutsaki et al., 2016), Semmelmann & Weigelt (2018) report that they were able to obtain comparable results from
an online vs. a lab experiment in tasks that involved fixation, pursuit and free viewing, and suggest that webcam-based
eye tracking is overall suitable for all three tasks and holds promise in cognitive science research. Furthermore, in recent
years many freeware and commercial solutions are popping up that offer reasonably accurate eye tracking
