Related Experiment Video
Updated: May 10, 2026
07:12
A Gaze-Contingent Display Framework for Perceptual Learning Research with Simulated Central Vision Loss
Published on: April 11, 2025
Heading recovery from optic flow: comparing performance of humans and computational models
Andrew J Foulkes1, Simon K Rushton, Paul A Warren
1School of Psychological Sciences, The University of Manchester Manchester, UK.
Frontiers in Behavioral Neuroscience
|June 27, 2013
Summary
This study evaluated computational models of human heading perception. While two models showed some similarity to human performance, none fully captured all aspects of heading perception data.
Area of Science:
- Visual perception
- Computational neuroscience
- Human psychophysics
Background:
- Human observers exhibit precise heading perception.
- Several computational models exist for heading perception.
- Understanding these models is crucial for advancing visual neuroscience.
Purpose of the Study:
- To compare four computational models of heading perception against human performance.
- To identify which model best replicates human visual processing of movement parameters.
- To assess the impact of stimulus quantity and quality on model and human performance.
Main Methods:
- Generated human performance profiles by manipulating optic flow stimuli (dot quantity, noise).
- Created comparable performance profiles for four candidate computational models.
- Regressed model outputs against human performance data to quantify model-human fit.
Main Results:
- Two models were excluded due to significant divergence from human performance.
- Two remaining models demonstrated partial similarities in performance magnitude and threshold patterns.
- No single model fully accounted for all observed human heading perception data.
Conclusions:
- Current computational models partially explain human heading perception.
- Further refinement of models is needed to capture the full complexity of human visual heading estimation.
- This research highlights limitations in current computational approaches to optic flow processing.
