Explainable deep learning reveals a behavioral strategy underlying human decisions in a spatial navigation task

This paper uses a set of explainable AI (xAI) methods to study human behavior in a spatial navigation task.First, locomotion and gaze dynamics of human subjects recorded during the task were reproduced in a virtual environment and visual snapshots extracted from this simulation were used as a dataset.Second, the dataset was used to train a deep convolutional network to reproduce human navigation decisions.Third, network strategies used for image classification were analyzed using a combination of three xAI methods.Using this analysis, we discovered a specific oculomotor marker that indicated the behavioral strategy used by human participants in this task.We conclude that xAI is a promising approach to study human behavior in complex real-world tasks.