GraFIX: A semiautomatic approach for parsing low- and high-quality eye-tracking data

Abstract

Fixation durations (FD) have been used widely as a measurement of information processing and attention. However, issues like data quality can seriously influence the accuracy of the fixation detection methods and, thus, affect the validity of our results (Holmqvist, Nyström, & Mulvey, 2012). This is crucial when studying special populations such as infants, where common issues with testing (e.g., high degree of movement, unreliable eye detection, low spatial precision) result in highly variable data quality and render existing FD detection approaches highly time consuming (hand-coding) or imprecise (automatic detection). To address this problem, we present GraFIX, a novel semiautomatic method consisting of a two-step process in which eye-tracking data is initially parsed by using velocity-based algorithms whose input parameters are adapted by the user and then manipulated using the graphical interface, allowing accurate and rapid adjustments of the algorithms’ outcome. The present algorithms (1) smooth the raw data, (2) interpolate missing data points, and (3) apply a number of criteria to automatically evaluate and remove artifactual fixations. The input parameters (e.g., velocity threshold, interpolation latency) can be easily manually adapted to fit each participant. Furthermore, the present application includes visualization tools that facilitate the manual coding of fixations. We assessed this method by performing an intercoder reliability analysis in two groups of infants presenting low- and high-quality data and compared it with previous methods. Results revealed that our two-step approach with adaptable FD detection criteria gives rise to more reliable and stable measures in low- and high-quality data.

Disentangling the mechanisms underlying infant fixation durations in scene perception: A computational account

“…Eye-tracking data from infants may contain considerably higher levels of noise than data from more compliant participants such as adults due to various factors including their high degree of movement, lack of compliance to the task, poor calibration and corneal reflection disturbances due to the underdeveloped cornea and iris (Hessels, Andersson, Hooge, Nyström, & Kemner, 2015;Saez de Urabain, Johnson, & Smith, 2015;Wass, Smith, & Johnson, 2013). To account for this potential quality/age confound, dedicated in-house software for parsing and cleaning eye tracking data has been developed (GraFix, Saez de Urabain et al, 2015).…”

“…To account for this potential quality/age confound, dedicated in-house software for parsing and cleaning eye tracking data has been developed (GraFix, Saez de Urabain et al, 2015). This software allows valid fixations to be salvaged from low-quality datasets whilst also removing spurious invalid fixations.…”

“…Due to the correlations between disengagement latencies and mean fixation durations (especially in the infants) participant mean fixation durations could have also been used to perform the same stratification but the decision was taken to use the disengagement latency as this is a more established developmental marker (Matsuzawa & Shimojo, 1997). Gathering a large amount of data from the same infant entails some technical and practical difficulties (Saez de Urabain et al, 2015). Due to the low number of fixations recorded from each infant, it was not possible to directly compare the two infants with the most extreme disengagement latencies (i.e.…”

“…The quality of the raw data generated by the eye-tracker may vary depending on many different factors such as the eye-tracker model and manufacturer, the eye physiology, the calibration procedure, the position of the participant relative to the eye-tracker, the degree of head, or even ethnicity (Blignaut & Wium, 2014;Holmqvist et al, 2011;Holmqvist, Nyström, & Mulvey, 2012;Nyström, Andersson, Holmqvist, & van de Weijer, 2013;Saez de Urabain et al, 2015). Eye-tracking data coming from populations such as infants may contain considerably higher levels of noise than data from more compliant participants.…”

“…Fixations were detected using GraFIX, a semi-automatic approach for parsing low-and high-quality eye-tracking data (Saez de Urabain et al, 2015). The input is initially parsed by using velocity-based algorithms whose input parameters are adapted by the user, and then manipulated using the graphical interface, allowing accurate and rapid adjustments of the algorithms' outcome.…”