The neuronal mechanisms that serve to distinguish between light-emitting and light reflecting objects are largely unknown. It has been suggested that luminosity perception implements a separate pathway in the visual system, such that luminosity constitutes an independent perceptual feature. Recently, a psychophysical study was conducted to address the question whether luminosity has a feature status or not. However, the results of this study lend support to the hypothesis that luminance gradients are instead a perceptual feature. Here, I show how the perception of luminosity can emerge from a previously proposed neuronal architecture for generating representations of luminance gradients.
Under daylight illumination conditions, looking at a television or computer screen rarely produces the sensation that displayed items are light-emitting, although each pixel of the screen emits light ( [41], with references). But to perceive objects as being luminous, it is not necessary to have a physically source of light emission. Halos were used by artists since a long time as a means to create luminosity effects in their paintings ( [41], with references). When a region is painted with a halo surrounding it, then one perceives this region with enhanced brightness, or even as glowing, without physical light emission being present. Thus, the perception of glow can be evoked on (light reflecting) paper or canvas, and text or pictures being displayed on a (light emitting) computer screen are not necessarily being perceived as luminous. In other situations perception and physics are not divergent. For example, the sun is always perceived as light emitting, and so are stars at night. In such situations, the strong contrast between light sources and background may provide the key factor to the perception of luminosity [3,4]. A recent fMRI study has identified a region in the brain which seems to be associated with the perception of luminosity [27]. In this study, different configurations of the glare effect display ( [5,23,40]; figure 5, top row) were presented to human observers. The results of the study were indicative to that luminosity might constitute a perceptual feature much like contrast, orientation, motion, or faces. The question about whether luminosity is a perceptual feature or not motivated a corresponding psychophysical study [7]. The study was based on the idea that perceptual features are distinguished from other object properties by being processed in a more efficient way. This means that visual features consume less * Electronic address: matsATcvc.uab.es; threequarksATyahoo.com; AT=@ attentional resources than non-features [18], what is reflected in, for example, "pop out" effects. A visual search paradigm such as the one used in the study of [7], therefore can serve to distinguish features from non-features. Unexpectedly, the results of Correani at al. are compatible with that luminance gradients instead of luminosity are a visual feature. Several authors have already formulated the hypothesis that luminance gradients are involved in the perception of luminosity [23,40,41,42], as there is evidence that luminance gradients can influence lightness perception under certain circumstances. I therefore asked whether a recently proposed theory for the perception of luminance gradients ("gradient system") could account for the just-described observations. The gradient system has been successful in quantitatively predicting available data on Mach bands [22]. It furthermore provided an account for Chevreul's illusion in terms of luminance gradients [20], and in addition is capable of real-world image processing. In this work I will show how spatial configurations of luminance gradients can interact to produce the perception of luminosity in the absence of physical illuminants. The results presented here also contribute to the further understanding of how luminance gradients interact with lightness computations and brightness perception, respectively. Specifically, representations of luminance gradients provide a straightforward explanation of "selfluminous grays" [41,42], and why it is that perception of luminosity is independent from lightness anchoring.
This section provides an overview over important characteristics of the gradient system. A more detailed description of it, as well as its formal definition, can be found in [20] and [22]. A notched square wave grating (or briefly “notch grating”) is used for illustration of the processing stages. A notch grating is a square wave with notches being centered at each luminance step, and luminance decays (for the bright stairs) and increases linearly (for the dark stairs), respectively, to a common luminance level (the luminance profile is shown in figure 2). This means that the faint lines centered at each step have the same intensity value, yet they are perceived with different brightness. See section II B for a detailed explanation of the processing stages.
The original motivation for proposing representations of luminance gradients was that they are of different utility for object recognition. It is known, for example, that they may aid to (i) recover three-dimensional information to compute surface shape (shape from shading, e.g. [29,33]), (ii) to resolve the three-dimensional layout of visual scenes (e.g. [2,24]), (iii) to identify material properties of object surfaces (e.g., mat versus glossy), and are therefore complementary to lightness computations (lightness is associated with surface representations). In situations, however, it may happen that luminance gradients rather would interfere with the goal of generating invariant surface repre
This content is AI-processed based on open access ArXiv data.