1. Background Method Results Objectives Results Discussion References
Raw Data Analyses Overcome Fixation Filter Constraints in Eye-tracking
Alina Nazareth, Shannon M. Pruden and Michael Riedel
Florida International University, Miami, FL
Background
Method
Results
A total of 78 adult university students, between the ages of 18-25, were presented with two 3D figures of assembled cubes based on the classic Shepard and Metzler (1971) study (see figure below)
Participants were asked to decide if the two 3D figures were a match (with one just rotated) or a non-match (i.e. one was a mirror image of the other). There were forty such trials.
A Tobii X60 eye-tracker was used to record participant eye-gaze patterns during the task.
Time (micro seconds) – Figure 1A
x-position of eye-gaze
Eye-tracking research uses eye fixations and saccades as measures of the underlying cognitive processes (Just & Carpenter, 1976).
Eye-tracking software uses event detection algorithms, which take raw data samples, and detects events within them. These event detection algorithms generate a fixation scanpath, which looks cleaner than the raw data; thus making eye-tracking data both manageable as well as ready for statistical analyses (Holmqvist et al., ).
However, during this process, the number of fixations initially recorded in the raw data is lost. Depending upon a researcher’s filter choice, a particular eye movement instance may or may not be classified as a fixation by the filter’s specific algorithm. The loss of fixations during event detection could lead to the misinterpretation of the cognitive process implicated.
Another drawback of event detection is the lack of meaning provided by the fixation duration, number of saccades, etc. with regard to the cognitive process.
Time (micro seconds) – Figure 1B
Time (micro seconds) – Figure 2A
x-position of eye-gaze
Time (micro seconds) – Figure 2B
Time (micro seconds) – Figure 3
x-position of eye-gaze
Figure 1 (A & B): Raw eye-gaze plots for high performers on a non-match stimuli.
Figure 2 (A and B): Raw eye-gaze plots for high performers on a match stimuli.
Note: A= Female and B=Male
Figure 3: Raw eye-gaze plots for low performers on a non-match stimuli.
Figure: Sample Item from the Shepard and Metzler (1971) MRT
Objectives
Results
Discussion
To resolve the contradictory predictions of cognitive processing using eye- movement, we plotted raw eye-gaze data using MATLAB to compare the time series representation across participants.
Although there were no significant differences in number of fixations, etc., the gaze plots reflected differences in strategy
In the present methodological study, we propose plotting raw eye tracking data as an alternative to standard eye- tracking metrics like eye fixations, etc. To demonstrate the usefulness of raw data plotting, we use cognitive strategy selection in mental rotation as an example.
In the present study we demonstrate the difference in raw eye-gaze patterns representing differences in cognitive strategy based on participant sex, type of stimuli and the interaction of the two.
Thus, while conventional eye-tracking metrics provide an excellent source of information regarding image salience, preference for an image and overall responsiveness to a web page layout, in research examining cognitive processes through eye movement, fixation metrics alone may not be insightful.
References
Holmqvist, K., Nyström, M., Andersson, R., Dewhurst, R., Jarodzka, H., & van de Weijer, J. (2011). Eye Tracking: A comprehensive guide to methods and measures. Oxford University Press.
Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science, 171(3972), Retrieved from
Just, M. A., & Carpenter, P. A. (1976). Eye fixations and cognitive processes. Cognitive Psychology, 8(4), 441–480. doi: / (76)
This research was supported by a generous grant from The Ware Foundation. Presented at the 26th APS Annual Convention, May 21-25, 2014, San Francisco, CA. Contact Information: