A model of human heading judgement in forward motion

Area of Science:

• Computational Neuroscience
• Human Perception
• Visual Neuroscience

Background:

• Human heading judgment is crucial for navigation and is primarily based on retinal flow.
• Existing models often struggle to fully capture the complexities of human performance in diverse visual scenarios.

Purpose of the Study:

• To develop and validate a novel computational model for human heading judgment using retinal flow.
• To investigate the model's consistency with human performance across various simulated self-motion conditions.

Main Methods:

• A computational model was developed based on assumptions of numerous sampling points and a symmetric sampling region.
• The model estimates self-rotation parameters and recovers heading direction after removing rotational components from retinal flow.
• Model performance was evaluated against human data in three psychophysical experiments with varied stimuli and perturbations.

Main Results:

• The model's simulation results were consistent with human performance when simulating self-motion towards ground, cloud, or frontoparallel planes.
• In experiments with perturbed vertical velocity, the model accurately predicted the observed increase in the perceived versus simulated heading function slope.
• The model successfully explained the characteristics of human heading judgment observed across all experimental conditions.

Conclusions:

• The proposed computational model provides a robust framework for understanding human heading judgment from retinal flow.
• The model's ability to replicate human performance highlights the importance of sampling density and symmetry in visual flow processing.
• This research offers valuable insights into the neural mechanisms underlying visual self-motion perception and navigation.