Machine learning accurately classifies age of toddlers based on eye tracking

Area of Science:

• Cognitive Neuroscience
• Developmental Psychology
• Computational Neuroscience

Background:

• Visual information extraction from complex scenes is key to understanding cognitive processes.
• Traditional eye-tracking studies use experimenter-defined areas of interest, limiting data-driven insights.
• Variability in looking behavior is influenced by image properties, task demands, and individual differences.

Purpose of the Study:

• To employ a data-driven machine learning approach to investigate age-related variability in infant gaze patterns.
• To identify specific features contributing to differences in visual exploration among infants of varying ages.
• To assess the efficacy of machine learning in characterizing developmental changes in attention allocation.

Main Methods:

• Utilized machine learning models, specifically Support Vector Machine (SVM) and Deep Learning (DL).
• Applied these models to eye-tracking data from infants to classify participants by age.
• Analyzed the models to identify salient features distinguishing different age groups.

Main Results:

• Machine learning models accurately classified infants based on age.
• The models identified meaningful visual features that differentiate age groups.
• Demonstrated a correlation between gaze patterns and developmental stage.

Conclusions:

• Machine learning is a powerful tool for analyzing developmental changes in visual attention.
• This data-driven method provides novel insights into how toddlers explore their environment.
• Highlights the utility of machine learning for characterizing diverse developmental capacities through exploratory gaze behavior.