It depends on how you look at it: scanpath comparison in multiple dimensions with MultiMatch, a vector-based approach

Area of Science:

• Cognitive Science
• Human-Computer Interaction
• Neuroscience

Background:

• Eye movement sequences, or scanpaths, are crucial for understanding visual attention and cognitive processes.
• Existing scanpath comparison methods often fail to capture both spatial and temporal scanpath properties effectively.

Purpose of the Study:

• To validate a novel geometric vector-based method, MultiMatch, for comparing eye movement scanpaths.
• To demonstrate MultiMatch's superiority over existing methods like ScanMatch in analyzing scanpath variations.

Main Methods:

• Developed and tested the MultiMatch algorithm, which uses geometric vectors for multidimensional scanpath comparison.
• Validated MultiMatch using synthetic data (Experiment 1) and real eye movement recordings (Experiment 2).
• Compared MultiMatch's performance against ScanMatch, a popular Levenshtein-based method.

Main Results:

• MultiMatch showed greater sensitivity to spatial position variations compared to ScanMatch using synthetic data.
• Experiments with real data revealed MultiMatch's ability to differentiate scanpaths based on direction, locus shifts, and scaling.
• The multidimensional approach of MultiMatch effectively captures nuanced differences in scanpath characteristics.

Conclusions:

• MultiMatch provides a more comprehensive and sensitive method for comparing eye movement scanpaths.
• This technique has broad applicability in research areas requiring analysis of eye movement consistency, such as learning and mental imagery.
• MultiMatch addresses limitations in current algorithms, particularly for complex tasks like "eye movements to nothing" research.