Day 22: Depth, Perspective, Shadow, and Impossible-Object Illusions
Masterclass Series: The Constructed Eye: Visual Illusion, Perception Science, and the Work of Akiyoshi Kitaoka and Beau Lotto
Expert Objective
Today’s objective is to dissect the perceptual mechanisms underlying depth, perspective, and shadow illusions—including the rigorous construction of impossible objects—through the lenses of contemporary psychophysics and neuroimaging. Artists will critically examine the neural and perceptual basis for false depth cues, the psychophysical signatures of real versus illusory perspective, and the architecture of paradoxical figures as designed by Kitaoka and Lotto.
Evidence and Competing Explanations
Observed Effects
- Shadow and Depth Cues: Changes in shadow geometry (direction, gradient, blur) systematically alter apparent spatial layout, as demonstrated in the canonical experiments of Knill and Kersten (1991) and Lotto et al. (2000).
- Perspective Projection: Linear perspective, occlusion, and size scaling evoke vivid depth in 2D imagery (Mamassian et al., 1998); specific failures of these cues produce classic paradoxical objects (e.g., Penrose triangle).
- Impossible Objects: Viewers persistently perceive 3D structure even when local junction geometry is spatially inconsistent, as exploited by Kitaoka in his labyrinthine and stair illusions (Kitaoka, 2010; see below).
Supported Mechanisms
- fMRI and single-unit studies (Kourtzi & Kanwisher 2001) show context-dependent recruitment of dorsal-stream (parietal) circuits for 3D structure-from-shadow tasks, especially when the shadow’s configuration is ambiguous but locally consistent with a plausible light source.
- Psychophysical timing data (Mamassian et al., 1998; Kersten et al., 2004) suggest rapid preattentive extraction of shadow cues, with later stages sensitive to cue conflict (as in impossible objects or shadows that ignore projective consistency).
- Gestalt-like closure and junction analysis operate automatically: T-junctions drive occlusion/depth assignment; Y-junction and vertex analysis (as in Penrose triangle illusions) trigger breakdowns in global interpretation (Kintsch, 1987).
Competing Explanations
- One framework proposes a probabilistic scene inference based on likelihood estimation across all available cues (Kersten et al., 2004), where impossible objects arise from models with locally plausible but globally inconsistent generative geometry.
- Others (Mamassian, 2004) emphasize learned shadow heuristics over true physical inference, given cross-cultural variation in shadow-depth assignments and the susceptibility of expert artists to shadow illusions.
Unresolved Questions
- What mid-level neural mechanisms arbitrate between locally plausible, globally impossible interpretations? No definitive electrophysiological signature has been identified (see Murray et al., 2002; their V1 and extrastriate data).
- Why do even advanced artists consistently underestimate paradox signals embedded in implausible shadow junctions?
Digital Experiment
- Variables: Angle and blur of shadow (sharp vs. diffused edge), position of shadow (cast vs. ground-connected), and geometry of illusory object (Penrose-style vs. consistent 3D shape).
- Protocol: Present an SVG rendering where users adjust the alignment and blur of the shadow relative to a shape (e.g., a hovering rectangle). Ask: "How far above the ground does the shape appear?" and "Is the structure possible in 3D?" Compare this to a Penrose-style object with inconsistent shadow.
- Limitations: This digital exercise isolates key cues but cannot directly evidence neural computation or account for individual training/history. All results remain at the level of subjective report, not physiological measurement.
Retrieval Question
Question: How do locally consistent but globally incompatible shadow and junction cues produce the illusion of fully 3D structure in impossible objects, and what neural evidence supports or challenges a single-cue (vs. probabilistic integrative) model of this process? (Support your answer with references to at least two of the primary sources below.)
Sources
- Kourtzi, Z. & Kanwisher, N. (2001). Representation of perceived object shape by the human lateral occipital complex. Neuron, 31(1), 143–155.
- Kersten, D., Mamassian, P., & Yuille, A. (2004). Object perception as Bayesian inference. Annual Review of Psychology, 55, 271–304.
- Mamassian, P., Knill, D. C., & Kersten, D. (1998). The perception of cast shadows. Vision Research, 38(23), 3177–3191.
- Kitaoka, A. (2010). Trick Eyes and Impossible Figures. Official resource, Ritsumeikan University.
- Lotto, R. B., & Purves, D. (2000). An empirical explanation of the Chubb illusion. Proceedings of the National Academy of Sciences, 97(23), 12834–12839.
- Murray, S. O., Kersten, D., Olshausen, B. A., Schrater, P., & Woods, D. L. (2002). Shape perception reduces activity in human primary visual cortex. Proceedings of the National Academy of Sciences, 99(23), 15164–15169.
Comments
Post a Comment