Fixation classification: how to merge and select fixation candidates

Area of Science:

• Cognitive Science
• Human-Computer Interaction
• Neuroscience

Background:

• Eye-tracking technology is widely used across disciplines like cognitive science, medicine, and marketing.
• Numerous algorithms exist for classifying eye-tracking data into behaviors such as fixations and saccades.
• The influence of selection rules on these classifications remains under-explored.

Purpose of the Study:

• To investigate the impact of fixation duration and saccade amplitude selection rules on fixation duration distributions.
• To evaluate how different classification algorithms interact with these selection rules.

Main Methods:

• Utilized high and low-quality eye-tracking datasets.
• Applied seven distinct eye-tracking data classification algorithms.
• Analyzed the effect of varying minimal saccade amplitude and minimal fixation duration parameters.

Main Results:

• Selection rules critically influence the merging and selection of potential fixations.
• For data with good precision (RMSD < 0.5°), the choice of classification algorithm is less critical.
• Effective classification relies on selecting saccades > 1.0° and fixations > 60 ms.

Conclusions:

• Eye-tracking data selection rules are crucial for accurate behavioral analysis.
• Researchers should consistently report the selection rules and parameter values used in their studies.
• Standardized reporting of selection parameters enhances the reproducibility and comparability of eye-tracking research.