[EXPERT: CONSTRUCTED EYE] Day 21 — Figure-Ground, Gestalt Grouping, and Visual Segmentation

The Constructed Eye Masterclass: Day 21

Figure-Ground, Gestalt Grouping, and Visual Segmentation

Intensive daily masterclass in perception science, with focus on the art and vision science of Akiyoshi Kitaoka and Beau Lotto.

Original SVG: Ambiguity in figure-ground assignments. Notice how grouping and enclosure affect which region is perceived as figure versus background, echoing Gestalt principles and influencing subsequent visual segmentation.

Visual artists and perceptual scientists alike confront the foundational problem of figure-ground organization—how surfaces are dynamically assigned as 'objects' (figures) or backgrounds (grounds). Both Akiyoshi Kitaoka and Beau Lotto exploit this ambiguity in their works, pushing the viewer’s visual system to oscillate between possible interpretations. Figure-ground segregation, Gestalt grouping, and visual segmentation form the core of today's session, traversing the boundaries between psychophysical effects, neural mechanisms, and practical manipulations in advanced art and vision research.

Expert Objective

To critically analyze the psychophysical and neuroscientific basis for figure-ground assignment and Gestalt grouping, with a focus on how these mechanisms inform and challenge advanced studio practice. We aim to discriminate between observed perceptual effects—many robust but not mechanistically trivial—and the array of current explanatory models, including local edge assignment, border-ownership neurons, and hierarchical scene interpretation.

Original SVG: Factorial demonstration of Gestalt grouping cues: similarity, dissimilarity, proximity, and isolation. Notice how each factor biases segmentation and figure assignment even when geometric arrangements are kept constant.

Evidence and Competing Explanations

Observed Effects: When visually parsing complex scenes, humans segment regions into 'figures' and 'grounds', guided by cues such as edge assignment, enclosure, contrast, and similarity (Palmer & Rock, 1994). Works by Kitaoka reveal that reversing contrast direction can force immediate perceptual remining, as seen in his 'Impossible Motion' and 'Café Wall' illusions—a controlled deployment of border ambiguity (Kyoto University).

Supported Mechanisms: Neurophysiological work identifies border-ownership cells in area V2 that respond differentially depending on the side of an edge belonging to a 'figure' (Zhou, Friedman & von der Heydt, 2000). These neurons integrate local cues with longer-range context, aligning with the psychophysical dominance of 'convex' regions as perceived figures (Bertamini & Wagemans, 2013).

Competing Explanations: Some models emphasize a bottom-up process—integration of contrast, continuation, and proximity—while others such as the Bayesian framework of Kersten and colleagues (2004) argue scene interpretation is a probabilistic inference, dependent on prior experience and scene statistics. A persistent debate remains: are figure-ground assignments made locally via feedforward channels, or do global expectations and context play the dominant role (Qiu & von der Heydt, 2005)? Current evidence strongly favors interaction between rapid local computations and slower feedback from object-level representations—see also the occlusion and amodal completion literature.

Unresolved Questions: Despite robust behavioral and neurophysiological readouts, the exact algorithms for multi-object segmentation under naturalistic, cluttered conditions remain poorly specified. How do conflicting cues resolve over time? Can artificial networks reproduce the speed and flexibility of biological segmentation—and if not, which architectural features are missing (Kim et al., 2023)?

Original SVG: Overlapping forms evoke figure-ground reversal. This schematic echoes classic studies—such as Rubin's vase—while directly connecting to Kitaoka's manipulations by making both edge and ownership ambiguous.

Digital Experiment

Experiment: Use the below interactive SVG to test subjective figure-ground assignment as edge polarity reverses.

Protocol: Fixate the center and alternately interpret the dark and light zones as figure. Note the spontaneous grouping that occurs as you mentally "flip" polarity. This process simulates, with fixed geometry, the perceptual switching exploited in Kitaoka’s and Lotto’s works. Controlled variables: shape geometry, contour sharpness, viewing distance. Limitations: This experiment cannot isolate specific neural populations; it tests only subjective assignments under strong geometric and contrast cues.

Retrieval Question

Critically compare the roles of local edge assignment (such as border-ownership neurons) and global perceptual organization in resolving figure-ground ambiguity. How do these systems interact under conditions where multiple competing groups could form, as demonstrated by the Digital Experiment?

Sources

• Zhou, H., Friedman, H. S., & von der Heydt, R. (2000). Coding of border ownership in monkey visual cortex. Journal of Neuroscience, 20(17), 6594-6611.
• Palmer, S. E., & Rock, I. (1994). Rethinking perceptual organization: The role of uniform connectedness. Psych. Rev.
• Kersten, D., Mamassian, P., & Yuille, A. (2004). Object perception as Bayesian inference. Annual Review of Psychology, 55, 271-304.
• Qiu, F. T., & von der Heydt, R. (2005). Figure and ground in the visual cortex: V2 combines stereoscopic cues with Gestalt rules. Neuron, 47(1), 155-166.
• Kim, T., Kim, J., Ko, H., & Paik, S. B. (2023). Neural mechanisms of perceptual grouping and segmentation. Vision Research, 206, 108174.
• Bertamini, M., & Wagemans, J. (2013). Processing convexity and concavity along a 2-D contour: Figure–ground, structural shape, and attention. Cognition, 128(2), 214-229.
• Kyoto University - Akiyoshi Kitaoka’s research
• Beau Lotto Lab – Perception research