Lightness, brightness, and anchoring

Color constancy mechanisms in virtual reality environments

Prior research has demonstrated high levels of color constancy in real-world scenarios featuring single light sources, extensive fields of view, and prolonged adaptation periods. However, exploring the specific cues humans rely on becomes challenging, if not unfeasible, with actual objects and lighting conditions. To circumvent these obstacles, we employed virtual reality technology to craft immersive, realistic settings that can be manipulated in real time. We designed forest and office scenes illuminated by five colors. Participants selected a test object most resembling a previously shown achromatic reference. To study color constancy mechanisms, we modified scenes to neutralize three contributors: local surround (placing a uniform-colored leaf under test objects), maximum flux (keeping the brightest object constant), and spatial mean (maintaining a neutral average light reflectance), employing two methods for the latter: changing object reflectances or introducing new elements. We found that color constancy was high in conditions with all cues present, aligning with past research. However, removing individual cues led to varied impacts on constancy. Local surrounds significantly reduced performance, especially under green illumination, showing strong interaction between greenish light and rose-colored contexts. In contrast, the maximum flux mechanism barely affected performance, challenging assumptions used in white balancing algorithms. The spatial mean experiment showed disparate effects: Adding objects slightly impacted performance, while changing reflectances nearly eliminated constancy, suggesting human color constancy relies more on scene interpretation than pixel-based calculations.

Dynamic decorrelation as a unifying principle for explaining a broad range of brightness phenomena

The visual system is highly sensitive to spatial context for encoding luminance patterns. Context sensitivity inspired the proposal of many neural mechanisms for explaining the perception of luminance (brightness). Here we propose a novel computational model for estimating the brightness of many visual illusions. We hypothesize that many aspects of brightness can be explained by a dynamic filtering process that reduces the redundancy in edge representations on the one hand, while non-redundant activity is enhanced on the other. The dynamic filter is learned for each input image and implements context sensitivity. Dynamic filtering is applied to the responses of (model) complex cells in order to build a gain control map. The gain control map then acts on simple cell responses before they are used to create a brightness map via activity propagation. Our approach is successful in predicting many challenging visual illusions, including contrast effects, assimilation, and reverse contrast with the same set of model parameters.

Seeing lightness in the dark

From intense sunlight in bright snow down to a moonless night in a dark forest, we can use light to recognize objects and guide our actions. This remarkable range mainly rests on having two different types of photoreceptors, the rods and the cones. The cones are active under daylight conditions, allowing high acuity and color vision. Rods are mainly active under very dim illumination conditions and have an exquisite sensitivity to light [1]. There are obvious detriments to visual perception in near darkness, such as a central scotoma, reduced motion perception [2], and most of all a lack of color [3]. There is only one type of rod, and thus intensity and wavelength differences cannot be disentangled when only the rods are active. This is captured well by the old saying "at night all cats are gray", meaning that different colors inevitably get mapped onto different shades of gray. Here we show that the perception of lightness is also different for night vision: our results indicate that surfaces that appear to be white under daylight conditions, at best, appear medium gray under night vision, suggesting that activation of the cones is necessary for the perception of white.

Theoretical approaches to lightness and perception

Theories of lightness, like theories of perception in general, can be categorized as high-level, low-level, and mid-level. However, I will argue that in practice there are only two categories: one-stage mid-level theories, and two-stage low-high theories. Low-level theories usually include a high-level component and high-level theories include a low-level component, the distinction being mainly one of emphasis. Two-stage theories are the modern incarnation of the persistent sensation/perception dichotomy according to which an early experience of raw sensations, faithful to the proximal stimulus, is followed by a process of cognitive interpretation, typically based on past experience. Like phlogiston or the ether, raw sensations seem like they must exist, but there is no clear evidence for them. Proximal stimulus matches are postperceptual, not read off an early sensory stage. Visual angle matches are achieved by a cognitive process of flattening the visual world. Likewise, brightness (luminance) matches depend on a cognitive process of flattening the illumination. Brightness is not the input to lightness; brightness is slower than lightness. Evidence for an early (< 200 ms) mosaic stage is shaky. As for cognitive influences on perception, the many claims tend to fall apart upon close inspection of the evidence. Much of the evidence for the current revival of the 'new look' is probably better explained by (1) a natural desire of (some) subjects to please the experimenter, and (2) the ease of intuiting an experimental hypothesis. High-level theories of lightness are overkill. The visual system does not need to know the amount of illumination, merely which surfaces share the same illumination. This leaves mid-level theories derived from the gestalt school. Here the debate seems to revolve around layer models and framework models. Layer models fit our visual experience of a pattern of illumination projected onto a pattern of reflectance, while framework models provide a better account of illusions and failures of constancy. Evidence for and against these approaches is reviewed.

Lightness perception in simple images: testing the anchoring rules

One approach toward understanding how vision computes surface lightness is to first determine what principles govern lightness in simple stimuli and then test whether these hold for more complex stimuli. Gilchrist (2006) proposed that in the simplest images that produce the experience of a surface (two surfaces differing in luminance that fill the entire visual field) lightness can be predicted based on two anchoring rules: the highest luminance rule and the area rule, plus a scale normalization. To test whether these anchoring rules hold when critical features of the stimuli are varied, we probed lightness in simple stimuli, painted onto the inside of hemispheric domes viewed under diffuse lighting. We find that although the highest luminance surface appears nearly white across a large variation in illumination (as predicted by the highest luminance rule), its lightness tends to increase as its luminance increases. This effect is small relative to the size of the overall luminance change. Further, we find that when the darker region fills more than half of the visual field, it appears to lighten with further increases in area but only if it is a single surface. Splitting the dark region into smaller sectors that cover an equal cumulative area diminishes or eliminates the area effect.