Effects of motion and configural complexity on color transparency perception

Peggy Gérardin, Philippe Roud, Sabine Süsstrunk, Kenneth KnoblauchVisual 2006 · DOI

TL;DR

This study investigates how motion and the complexity of visual patterns influence our perception of color transparency. The researchers found that adding motion and increasing pattern complexity can significantly alter whether a visual stimulus appears transparent, and these effects interact with the specific type of color transformation involved. This highlights that factors beyond just color shifts are crucial for understanding transparency perception.

What Problem Does This Paper Address?

Color transparency is a fascinating perceptual phenomenon where a surface is seen as a single entity, even though its color is altered by a transparent filter. While previous research has explored how specific color transformations (like translation and convergence in color space) can evoke transparency, the role of other visual cues, such as motion and the structural complexity of the visual scene, has been less systematically investigated. This paper aims to fill that gap by examining how motion and different levels of "configural complexity" (how the visual elements are arranged) interact with various color transformations to influence perceived transparency. Understanding these interactions is crucial for a comprehensive model of transparency perception, which has implications for areas like object recognition and scene understanding.

Methods

The researchers created a series of visual stimuli designed to elicit transparency perception. These stimuli involved a central overlay region against a background, with controlled color changes between the two.

  • Chromatic Transformations: Five types of color changes were systematically applied: translation, convergence, shear, divergence, and rotation. These transformations were defined in CIE XYZ color space.
  • Configural Complexity: Two levels of complexity were used: a simple bipartite configuration (two large colored regions) and a more complex checkerboard pattern.
  • Motion: The central overlay could be either static or moving.
  • Luminance: Three luminance conditions were tested: equiluminant (no luminance difference), luminance increment (filter-like), and luminance decrement (illuminant-like).
  • Modulation Depth: The magnitude of the color change was varied across three levels.

Participants were asked to judge whether the central overlay appeared transparent or opaque for each stimulus. The data were analyzed using log-linear models, specifically a Generalized Linear Model with a Poisson family and logarithmic link function, to account for the count data. Model selection was guided by Akaike's Information Criterion (AIC).

Datasets

Not discussed in the paper. The stimuli were synthetically generated for the experiment.

Key Insights and Ideas

  • Motion and Complexity are Not Neutral: Adding motion and increasing configural complexity are not simply additive factors that enhance transparency universally. Their effects are highly dependent on the specific color transformation being used.
  • Shears Can Be Transparent: Contrary to some previous findings suggesting shears lead to opacity, this study found that shears can appear transparent, particularly with short vector lengths and when combined with motion. This suggests that previous models might have been too restrictive.
  • Motion Overrides Color Cues: In some cases, motion was powerful enough to override conflicting color information and lead to a strong perception of transparency, even for transformations that might otherwise appear opaque.
  • Interactions are Key: The most significant finding is the complex interplay between chromatic transformations, motion, and configural complexity. For example, convergence and translation benefit greatly from motion and checkerboard patterns, leading to near-unanimous transparency judgments.

Key Results

  • Convergence and Translation: These transformations showed increased transparency with motion and checkerboard patterns, reaching near 100% classification in some conditions.
  • Shear: Shearing transformations were generally less transparent, but showed increased transparency at low modulation depths, especially when combined with motion.
  • Divergence: Divergent transformations were rarely classified as transparent for larger chromatic modulations, but motion increased transparency perception.
  • Rotation: Rotations were more often perceived as transparent with smaller chromatic modulations, with motion and complexity further boosting this perception.
  • Equiluminance: Interestingly, increasing chromatic modulation often disadvantaged transparent responses at equiluminance for shears and rotations.

Limitations and Future Work

  • Limited Color Space: While the study used CIE LUV for controlling color differences, the underlying color transformations were defined in CIE XYZ. Future work could explore these effects in other color spaces.
  • Observer Variability: Although the study found similar performance across observers, the sample size was small (three observers). Future research could include a larger and more diverse participant pool.
  • Generalizability of Models: The authors suggest that their findings challenge the limitations of existing models like the Generalized Convergence Model (GCM), especially concerning transformations like shear. Future work should focus on developing more comprehensive models that account for the interplay of these various cues.
  • Quantitative Investigation of Cue Interactions: The study highlights the need for more quantitative investigations into how different cues (color, motion, complexity) interact in transparency perception. Understanding these interactions is crucial for understanding the mechanisms of surface perception and object segmentation.

Summary updated

Abstract

We tested whether motion and configural complexity affect perceived transparency. A series of five coherent chromatic transformations in color space was applied across a figure: translation, convergence, shear, divergence and rotation. The stimuli consisted of a bipartite or a checkerboard configuration (10 x 10 degrees), with a central static or moving overlay (5 x 5 degrees). Three different luminance conditions (the plane of chromatic transformation oriented toward higher, lower, or equal luminances) were also tested for each of three modulation depths. For each stimulus, the observer judged whether the overlay appeared transparent or not. The main results indicated an interaction between the type of chromatic transformation and stimulus motion and complexity. For example, convergences are judged to appear transparent significantly more often when motion is added for bipartite configurations, or when they are generated in a checkerboard configuration. Surprisingly, shears that have been reported to appear opaque, are more frequently reported to appear transparent with short vector lengths and when combined with motion. Other transformations are also affected by motion, although the effectiveness of figural complexity on transparency seems to depend on both the type of color shifts and the presence of motion. The results indicate that adding motion and stimulus complexity are not necessarily neutral with respect to the chromatic shifts evoking transparency. Thus, studies that have used motion to enhance transparency may yield different results about the color shifts supporting transparency perception from those that did not. The same might be supposed for stimulus complexity under some conditions.
BibTeX
@inproceedings{gerardin2006effects,
  author = {Peggy Gérardin and Philippe Roud and Sabine Süsstrunk and Kenneth Knoblauch},
  title = {{Effects of motion and configural complexity on color transparency perception}},
  booktitle = {Visual},
  year = {2006},
  doi = {10.1017/s0952523806233352},
}
Figure