Lateral Interactions in Brightness Perception: Psychophysics and Modeling

人类的视觉系统非常擅长感知环境中物体外观的不同亮度，即使在不同的照明条件下也是如此。单个表面的亮度感知会受到附近表面亮度的强烈影响，这通常表现为视错觉。“亮度感应”的一个经典例子是，相同的灰色斑块可以根据周围区域的亮度而显得更亮或更暗。这种亮度错觉显示了视觉场景的不同区域如何相互作用，从而产生视觉刺激。尽管一个多世纪以来出现了许多理论，但我们仍然不了解这些感知背后的横向相互作用机制。这项工作涉及理论建模和人类观察者的心理物理测量。一种特殊的新模型已经被开发出来，以解释整个视觉效果组，包括一些新颖的亮度配置。该模型现在结合了视觉反应和组合规则的非线性，并通过感知结果与预测的匹配程度来测试其有效性。这些量化亮度感知变化的结果对视觉神经科学非常重要，因为它们为眼睛和大脑视觉皮层的神经生物学网络如何产生有意识的感知提供了见解。其影响也将从视觉神经科学扩展到感知心理学，并可能在视觉显示的设计和开发人工视觉系统中发挥重要作用。

The visual system of humans is very good at perceiving different levels of brightness in the appearance of objects in the environments, even under varying conditions of lighting. Brightness perception of a single surface can be affected quite strongly by the brightness of nearby surfaces, as often shown by optical illusions. A classic example of 'brightness induction' is when identical gray patches can be made to appear either brighter or darker depending on the brightness of the areas surrounding them. Such brightness illusions show how different areas of the visual scene can interact in creating a visual stimulus. We still do not understand the mechanisms of lateral interactions underlying these perceptions, despite a number of theories over more than a century. This work involves theoretical modeling and psychophysical measurements from human observers. A particular new model has been developed to account for a whole group of visual effects, including some novel brightness configurations. The model now incorporates non-linearities in visual responses and rules for combinations, and is tested for validity by how well the perceptual results fit the predictions. These results quantifying perceptual variations in brightness are very important for visual neuroscience, because they provide insights into how neurobiological networks of the eye and the visual cortex in the brain might produce conscious perceptions. The impact also will extend beyond visual neuroscience to perceptual psychology, and could be important in the design of visual displays and in developing artificial visual systems.