Related Experiment Videos
Testing the Perrone and Stone (1994) model of heading estimation
1School of Optometry, University of California at Berkeley 94720, USA.
Vision Research
|June 1, 1997
Summary
Human heading perception is inaccurate with simulated gaze rotations. A model using optic flow alone fails to predict human errors, suggesting limitations in its assumptions about target motion.
Area of Science:
- Visual perception
- Human psychophysics
- Computational neuroscience
Background:
- Human observers exhibit heading judgment errors under simulated gaze rotations.
- Two hypotheses, extra-retinal-signal and retinal-image, attempt to explain these errors.
- Existing models using optic flow alone have limitations.
Purpose of the Study:
- To evaluate a specific optic flow model (Perrone & Stone, 1994) against human data.
- To test the model's gaze-stabilization constraint assumption.
- To determine if the model aligns with human heading perception accuracy.
Main Methods:
- Comparison of model predictions with human data from Royden et al. (1994) experiments.
- Analysis of heading errors in gaze-stabilized and gaze-unstabilized conditions.
- Evaluation of the gaze-stabilization constraint's impact on model performance.
Main Results:
- The model's gaze-stabilization constraint predicts an error asymmetry not seen in human data for the gaze-unstabilized experiment.
- Human heading judgment errors are proportional to rotation rate.
- The model's predictions diverge from human performance in specific conditions.
Conclusions:
- The Perrone and Stone (1994) model, as formulated, is inconsistent with human heading perception data.
- The gaze-stabilization constraint is a key factor contributing to the model's discrepancies.
- Further research may involve post-heading estimation processing to reconcile model and human data.