Vision toolkit part 3. Scanpaths and derived representations for gaze behavior characterization: a review

Area of Science:

• Cognitive Science
• Neuroscience
• Computer Science

Background:

• Scanpath analysis offers insights into visual behavior by integrating spatial and temporal gaze dynamics.
• The field currently faces methodological fragmentation with diverse representations and comparison metrics.

Purpose of the Study:

• To review and unify computational approaches for scanpath characterization and comparison.
• To provide a framework for selecting appropriate methods based on research context.

Main Methods:

• Survey of scanpath representations (e.g., geometric, attention maps, symbolic encodings).
• Review of comparison metrics (e.g., point-mapping, elastic alignment, string-edit distances).
• Analysis of underlying assumptions, interpretability, and practical implications of various methods.

Main Results:

• Identified a wide array of methods for describing individual scanpaths and comparing them across conditions.
• Highlighted the strengths, limitations, and sensitivities of different approaches.
• Emphasized scanpath analysis as a set of complementary tools rather than a single optimal metric.

Conclusions:

• A unified conceptual framework is needed to guide methodological selection in scanpath analysis.
• Standardization and clear understanding of methods will foster reproducibility.
• Meaningful interpretation of gaze dynamics across disciplines can be supported by a cohesive approach.