Contrast and assimilation in lightness judgments

Experiments studied how assimilation and contrast varied as a function of the reflectance of figures above and below the reflectance of a gray background. The results showed that for the experimental conditions investigated: (a) contrast always occurs when the reflectance of lines is above the reflectance of the background, (b) assimilation occurs when the reflectance is below that of the background, (c) circles produce the same degree of assimilation and contrast as lines of equal width, and (d) repeated judgments do not affect contrast but reduce assimilation; as line darkness and line width increase assimilation gives way to contrast following repeated judgments. The results are discussed in connection with the hypothesis that assimilation and contrast arise from opponent processes in the visual system,     

Contrast and assimilation in the Baldwin illusion

Four experiments were conducted to investigate the role of "large" squares on the occurrence of assimilation and contrast in the Baldwin illusion. In Exps. 1 and 2, in which subjects reproduced the length of the test line, large squares induced illusions of assimilation. In Exp. 3, reproduction of the entire length of the test line also resulted in illusions of assimilation but bisection of the test line and reproduction of one-half the length of the test line resulted in contrast. Exp. 4 replicated the findings involving judgments of one-half the length of the test line. Over-all, the results indicate that judgments of one-half the length of the test line result in contrast, whereas judgments of the entire extent result in assimilation.     

Contrast and assimilation in categorization and exemplar production

We studied contrast and assimilation in three tasks: an exemplar-production task, a categorization task, and a combined categorization-then-production task. On each trial of the first task, subjects produced a circle when prompted with a category label. In the second task, they classified lines that differed in length into one of four categories. On each trial of the combined task, they classified two lines and then produced a line when prompted by a category label. All three tasks converged on the same conclusion: subjects' representation of the categories (measured in pixels in the production tasks and by the direction of errors in classification) shifted systematically from trial to trial. When successive stimuli were from the same category, the representation of that category was pulled toward the exemplar from the previous trial. When successive stimuli were from different categories, the representations of the neighbouring categories were pushed from the category of the initial stimulus. We conclude that accounts of categorization and identification must accommodate both assimilation and contrast as a function of trial-to-trial shifts in representation.     

Contrast effects in judgments of health hazards

Researchers commonly use 2 models to explain contrast effects (CEs): the standard-of-comparison model and the set-reset model. The 2 models focus on the role of categorization to predict when a CE (instead of an assimilation effect) will happen, while minimizing the role of knowledge accessibility and relevance in determining whether any effect will occur. A 3rd model, the selective-accessibility model (F. Strack & T. Mussweiler, 1997), focuses on knowledge accessibility and relevance, but it is a model of assimilation effects in the anchoring bias. In the present study of CEs, the authors tested 3 predictions implied by the selective-accessibility model. The authors found a CE only when anchor- and target-rating dimensions matched and only in the 1st of multiple targets rated. The CE required a minimum amount of attention to the anchor. These results support the account that selective knowledge accessibility and relevance play an important role in CEs.     

Contrast effects on the auditory evoked response and its relation to psychophysical judgments

An auditory average evoked response (AER) procedure was used to search for a physiological counterpart of the contrast effect-the tendency for a S’s judgment of the intensity of a stimulus to be modified by the intensity of prior stimulation. When a tone was preceded by a softer tone, its AER amplitude increased; conversely, when a tone followed a louder tone, amplitude decreased. Irregular intervals between tones decreased the AER contrast effect, as did increasing the interval between tones. Further, significant correlations were found between individual scores on the AER contrast procedure and performance on a battery of psychophysical tasks.     

Spatial and temporal lightness anchoring

Research in the area of lightness perception has not adequately addressed the influence of previously viewed visual fields on perceived surface reflectance. In the spatial realm, a spot-in-a-void will appear darker when a second surface of higher intensity is placed adjacent to it. The brighter surface takes the role of white, the anchor, and the dimmer is scaled accordingly. We find that when a spot-in-a-void is presented to observers in a light controlled chamber it is influenced by nonretinal temporal relationships mediated by the degree of complexity. We also find that the influence cannot be explained in terms of successive contrast at a high or low level in the visual system, but can be explained by an anchoring model. The present results follow the same rules governing spatial integration and anchoring and thereby support the currently proposed concept of temporal anchoring.     

Impossible shadows and lightness constancy

The intersection between an illumination and a reflectance edge is characterised by the 'ratio-invariant' property, that is the luminance ratio of the regions under different illumination remains the same. In a CRT experiment, we shaped two areas, one surrounding the other, and simulated an illumination edge dividing them in two frames of illumination. The portion of the illumination edge standing on the surrounding area (labelled contextual background) was the contextual edge, while the portion standing on the enclosed area (labelled mediating background) was the mediating edge. On the mediating background, there were two patches, one per illumination frame. Observers were asked to adjust the luminance of the patch in bright illumination to equate the lightness of the other. We compared conditions in which the luminance ratio at the contextual edge could be (i) equal (possible shadow), or (ii) larger (impossible shadow) than that at the mediating edge. In addition, we manipulated the reflectance of the backgrounds. It could be higher for the contextual than for the mediating background; or, vice versa, lower for the contextual than for the mediating background. Results reveal that lightness constancy significantly increases when: (i) the luminance ratio at the contextual edge is larger than that at the mediating edge creating an impossible shadow, and (ii) the reflectance of the contextual background is lower than that of the mediating one. We interpret our results according to the albedo hypothesis, and suggest that the scission process is facilitated when the luminance ratio at the contextual edge is larger than that at the mediating edge and/or the reflectance of the including area is lower than that of the included one. This occurs even if the ratio-invariant property is violated.     

Spatial articulation affects lightness

In a recent paper, Agostini and Bruno (1996) showed that the size of simultaneous lightness contrast increases under Gelb lighting. To extend Agostini and Bruno's work, we applied their methodology to a set of more spatially articulated displays. In four experiments, we investigated the role of spatial articulation on the size of the simultaneous lightness contrast effect. In the first experiment, we found a decrease of the simultaneous lightness contrast effect as the spatial articulation increased. In the second experiment (the control experiment), performed under homogeneous illumination, we found that the effect of spatial articulation is not detectable, even though the data seem to show the same trend as that in the previous experiment. In the third experiment, we found that spatial articulation affects not only the middle reflectance region, but also the lowest one. As the spatial articulation increases, the effect on the lightnesses of both regions decreases. In the last experiment, performed with a reduced range of reflectances, we found a lightening effect for all the reflectances and, again, an effect of spatial articulation. The results of these experiments are interpreted according to the model proposed by Gilchrist et al. (in press).     

Illusory lightness in the Delboeuf figure

A white disk presented inside a small concentric circle (a display like the Delboeuf figure) looks lighter than when inside a larger one. The magnitude of the effect was measured with 20 observers (10 women, 10 men; ages 19 to 26 years). The decrements in luminance of the disk inside the smaller circle, in order to match the lightness of the disk inside the larger circle, are reported for three different grey backgrounds and correspond to 9.13, 10.33, and 10.46%.     

Explaining lightness illusions.

Grey looks darker when set against white than when set against black. In some complex figures this illusion becomes startling, and can be shown to depend on the perceptual organisation of regions within the image. The most widely accepted explanations of such effects are based on the analysis of the junctions formed where the boundaries of nearby regions meet. Even theories where junctions are not the subject of special concern underline their importance as grouping cues. In this paper I present several new families of figures that challenge both views, and conclude that junctions do not play any crucial role in lightness estimation.     

Capturing lightness between contours

Homogeneously coloured bars may exhibit lightness differences at the intersections. A well-known example is the Hermann grid illusion, where crossing white bars on a black background show dark patches at the crossings. Jung (1973, Handbook of Sensory Physiology volume VII/3, pp 1-152) found that the dark patches persist when thin outlines are drawn at the intersections, and are even visible in foveal vision. Recently, it has been shown that making distortions to the contours of a Hermann grid-like configuration results in the disappearance of the illusory dark spots (Geier et al, 2008 Perception 37 651 665). We show that thin outlines at the crossings of the distorted Hermann grid induce lightness differences in the same direction as in the original Hermann grid illusion, even in foveal vision and in displays consisting of two crossing bars. Our experiments reveal that the induced lightness differences are independent of the luminance polarity and shape of the contours at the intersection. We suggest that the effect results from lateral inhibition and an additional spreading and capturing of these differences between luminance contours. A similar capturing between collinear contours may play a role in peripheral vision in the original Hermann grid.     

Local and global processes in surface lightness perception

Various demonstrations show that a target of constant luminance can be made to appear darker in perceived lightness merely by introducing an adjacent region of higher luminance. This has often been interpreted as a manifestation of contrast effects produced by lateral inhibition, a relatively local process. An alternative interpretation holds that the highest luminance in such a display serves as an anchor that defines the white level. This interpretation is global in the sense that the anchor need not be located near any particular target in order to serve as its standard. Edge integration processes have been postulated that would enable such remote comparisons, but there is controversy about the strength of these processes. We report a series of experiments in which local and global processes were assessed. Specifically, we tested whether the introduction of a higher luminance has a greater darkening effect on an adjacent target than on a remote target. We found no difference, suggesting that the darkening effect is a matter of anchoring, not contrast, and that edge integration processes required by anchoring are relatively strong.     

The unbearable lightness of priming

Repetition priming from text to isolated words has been difficult to observe. One explanation for this difficulty is that previous attempts to observe this type of priming have utilised conditions that normally reduce priming. Two experiments were conducted to evaluate this hypothesis. Experiment 1 involved participants being presented with words in isolation and in text passages. The words were then presented again in a lexical decision test. Results indicated that priming occurred as a result of exposure to both isolated words and words in text, although priming was greater in the word-word condition. Experiment 2 investigated whether priming occurred in a lexical decision test on words that had been read prior to the test in Milan Kundera's novel "The Unbearable Lightness of Being." There was some evidence that participants who had read the book recently were faster at lexical decision to words from the book than participants who had not read the book. The two experiments therefore indicate that priming can occur from text to isolated words, although it is smaller in magnitude to that observed from word to word. Reasons for this difference, as suggested by Kirsner and Speelman (J. Exp. Psychol.: Learn. Mem. Cogn. 22 (1996) 563) model of repetition priming, are discussed.     

The psychological relationship between lightness and saturation

Lightness and saturation, though independent in a physical sense, have often been seen to be psychologically related. The first experiment uses a multidimensional scaling technique to again verify this nonindependence for untrained Os. A further experiment, using a matching technique, quantifies this relationship, giving the extent of the error in lightness for varying levels of saturation. This relationship is found to be independent of hue if saturation is measured by the colorimetric purity of the hue. Errors in value (lightness on the Munsell system) were found to be linearly related to colorimetric purity.