Understanding Lightness Illusions
Author Information
Author(s): David Corney, R. Beau Lotto
Primary Institution: UCL Institute of Ophthalmology, University College London
Hypothesis
Lightness illusions arise from the ambiguity of natural stimuli and are resolved through past visual experience.
Conclusion
The study suggests that lightness illusions are caused by the inherent ambiguity of visual stimuli and the statistical resolution of this ambiguity by visual systems.
Supporting Evidence
- Artificial neural networks exhibited lightness constancy similar to human perception.
- The networks made systematic errors in behavior that correspond to human lightness illusions.
- Illusions arise from the statistical relationship between images and scenes based on past visual experience.
Takeaway
This study shows that our brains sometimes get confused by light and colors, making us see things that aren't really there, just like how a computer can learn to see things in a similar way.
Methodology
Artificial neural networks were trained to predict surface reflectance in synthetic scenes with varying illumination.
Limitations
The study does not directly model human physiology or cognition, focusing instead on statistical learning from visual stimuli.
Statistical Information
P-Value
p ≈ 0
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website