1. Introduction

Knowledge workers increasingly operate in environments where sustained, uninterrupted focus is essential. Programmers, designers, and researchers depend on states of deep work that are easily disrupted by intrusive visual notifications. Contemporary graphical interfaces compound this problem: notification badges, popups, and directional cues such as arrows occlude the very content they reference. This occlusion problem in gaze guidance remains unsolved by current methods.

I propose a fundamentally different approach: a peripheral visual notification that is inherently self-erasing. My candidate mechanism is the scintillating grid illusion, a geometric optical phenomenon first described by Schrauf et al. In this illusion, illusory dark spots flicker at grid intersections in peripheral vision and disappear automatically when the viewer fixates directly on them. VanRullen and Dong demonstrated that this suppression is driven by attentional focus—meaning the brain itself acts as the switch, eliminating the need for eye-tracking hardware.

Existing psychophysical research on the scintillating grid illusion has been conducted exclusively on 2D LCD displays. The behaviour of the illusion in physical 3D objects is entirely unexplored. This research gap motivates the present study: how does the transition from 2D digital to 3D physical materialise the scintillating grid illusion, and what physical parameters govern its strength?

This paper presents the progression of a two-stage design research programme. Stage 1 established a scientific baseline from 2D literature and conducted exploratory 3D prototyping. Stage 2 (reported here) defines a rigorous experimental framework and conducts a standardised qualitative pilot study. The work adopts Material-Driven Design (MDD) as its methodological anchor, prioritising the definition of physical boundaries before committing to application scenarios.

2. Background and Related Work

2.1 The Scintillating Grid Illusion

The scintillating grid illusion is a variant of the Hermann grid illusion, whose illusory spots at intersections were classically attributed to lateral inhibition in concentric receptive fields of retinal ganglion cells—an account later shown to be insufficient, with cortical mechanisms playing a necessary role. It consists of white circular discs at the intersections of a grey grid on a black background (Figure 1). Illusory dark spots are perceived within the discs in peripheral vision and disappear at fixated locations.

For clarity, the perceptual sequence can be described step-by-step: (1) when viewers look at the panel without directly fixating a specific disc, dark scintillating spots tend to appear at multiple peripheral intersections; (2) when attention and gaze move to one target intersection, the spot at that fixated location weakens or disappears; and (3) when gaze shifts away again, comparable spots re-emerge in other peripheral intersections. This repeatable peripheral-appearance / foveal-disappearance cycle is the operational basis of the “self-erasing” effect discussed in this paper.

Ratio dependency. Matsuno and Sato demonstrated that illusion strength is maximised when the bar width : disc diameter ratio (bar w : disc w) is approximately 1:1.4. This finding, derived from experiments on a 27-inch TFT monitor (viewing distance 68 cm), constitutes the primary 2D baseline for this research.

Attentional suppression. VanRullen and Dong showed that covert spatial attention suppresses the scintillation surrounding intersections. This mechanism—attention as switch—is the theoretical foundation of the self-erasing property.

Depth and orientation effects. Schrauf and Spillmann found that stereoscopic depth separation between discs and grid progressively weakens the illusion. Matsuno and Sato showed that rotating the stimulus from 0° to 45° monotonically decreases perceived strength.

2.2 Research Gap: Physical 3D Materialisation

All prior psychophysical research has been conducted on 2D digital displays. No published study has examined the illusion in physically fabricated 3D objects. Schrauf and Spillmann manipulated stereoscopic disparity on a 2D display—a fundamentally different phenomenon from physical Z-axis height. This gap is the primary focus of the present work.

2.3 Material-Driven Design

Material-Driven Design (MDD) is a methodology in which material properties serve as the primary driver of design decisions. Rather than starting from a predetermined application, MDD requires designers to characterise material behaviour—its “material experiences”—before mapping to design possibilities. I adopt MDD to avoid premature commitment to use cases and ensure that design guidelines are grounded in empirical material behaviour.

2.4 Peripheral Interaction and Calm Technology

Bakker et al. characterise peripheral interaction as a design approach that allows awareness without disrupting primary task engagement. Managing this transition in software has been addressed through peripheral display toolkits that modulate notification salience across attention levels. Weiser and Brown’s concept of calm technology proposes that the most humane technologies “inform but don’t demand attention.” Ambient displays operationalise this ideal and have been evaluated through structured heuristic methods. These displays are peripheral and aesthetically driven visualisations of non-critical information. The self-erasing scintillating grid illusion aligns precisely with these principles.

3. Research Questions

Primary Research Question: How can the scintillating grid illusion be physically materialised in 3D-printed artefacts as a peripheral visual notification that self-erases upon direct fixation, and what physical parameters govern the strength of this illusion?

Secondary Research Questions:

1. How does physical Z-axis height affect illusion strength, and is there a minimum threshold?
2. How does surface finish (Non-Ironing vs. Ironing) modulate illusion strength?
3. What standardised experimental conditions are necessary for reliable measurement in physical artefacts?

4. Methodology

4.1 Overall Approach

I adopt MDD as the overarching framework. The research proceeds in three phases: (1) establishing a 2D theoretical baseline from literature, (2) exploratory 3D prototyping (Stage 1), and (3) a parametric pilot study with standardised conditions (Stage 2).

4.2 Parameter Structure

A central contribution is the explicit definition of the parameter space as three distinct layers (Table 1; Figure 2). The XY-plane ratio of 1:1.4 is adopted directly from Matsuno and Sato. The Z-axis ratio of 1:1 is a design premise that prevents variable contamination by ensuring that observed height effects reflect volumetric relief as a whole. The Z-axis height is the independent variable of the present study.

4.3 Stage 1: Exploratory Prototyping

3D-printed substrates with precisely specified surface geometry have been validated as stimuli for psychophysical perception studies; I extend this approach from the tactile to the visual domain by fabricating raised relief patterns whose parameters are varied systematically.

Stage 1 produced five prototypes (P1–P5) varying height from 0.05 to 1.0 mm, and additionally tested two surface finishes (Non-Ironing and Ironing) on a 100×100 mm Black PLA panel (2 mm base). The Stage 2 sample matrix and finish-by-thickness structure are shown in Figure 3. Two preliminary observations emerged:

Observation 1 (Thickness). Physical height is a critical parameter. A height of 0.5 mm produced a noticeably stronger illusion than 0.1 mm, suggesting that volumetric cues (cast shadows, edge highlights) enhance luminance contrast.

Observation 2 (Surface finish). In the current filament set, Ironing visibly weakened the illusion. Specular reflection reduced effective luminance contrast between discs and background, reconfirming Matsuno and Sato’s emphasis on luminance contrast as the primary driver.

Two methodological problems were identified by the RPC committee during Stage 1 review: (a) Ironing confounds gloss with surface geometry change (nozzle scratching potentially distorts disc geometry); (b) comparing 3D prints to 2D displays is an “apple-to-orange” comparison, since screens emit light while prints reflect it. Stage 2 addresses both.

4.4 Stage 2: Standardised Parametric Pilot Study

Stage 2 crosses two independent variables: thickness parameter t (3 levels: 0.1, 0.3, and 0.5 mm, selected to bracket the threshold range identified in Stage 1) and surface finish (2 levels: Non-Ironing, Ironing), yielding 6 conditions labelled A–F (Table 2). Here, t is applied equally to bar and disc thickness (Z ratio = 1:1), so the total object thickness becomes 2 mm + 2t (base thickness = 2 mm).

4.4.1 Constants and Rationale. The viewing distance of 40 cm was selected to approximate the visual angle of Matsuno and Sato’s stimulus for the 100 mm sample size. No prior literature establishes a viewing distance for physical samples of this size; 40 cm is an empirically motivated baseline, explicitly acknowledged as such.

Color temperature was set to 5500 K. Natural daylight is commonly approximated in the 5300–6500 K range, so 5500 K provides a spectrally neutral, daylight-like illuminant for viewing reflective samples. It also offers a practical link to the 2D baseline: Matsuno and Sato’s 27-inch TFT study used an emissive display, typically in the ∼6000–6500 K class; 5500 K is a neutral high-CCT choice for comparing physical reflection to that literature without biasing the scene toward warm (low-CCT) tungsten sources. Work on CCT and spatial brightness further motivates using a well-specified, higher-CCT light source for tasks where perceived luminance contrast matters.

The lighting protocol addresses the variable contamination issue identified by the committee. Keeping light-source angle, color temperature, and illuminance fixed (40°, 5500 K, 575 lux) ensures identical photometric conditions across surface conditions. This constraint itself constitutes an empirical finding: Ironing fundamentally alters the optical constraints of the observing environment.

4.4.2 Apparatus. Samples are mounted vertically on a ball-head tripod. A fixed chin rest positions participants at 40 cm from the sample surface, ensuring consistent viewing distance and angle. A smartphone inclinometer verifies 90° alignment. During sessions, sample orientation was kept consistent by the experimenter across presentations. Light-source angle (40°), color temperature (5500 K), and illuminance (575 lux) are fixed and verified for all sessions. Angle and illuminance are checked before each session using a smartphone inclinometer and the smartphone lux-meter application LM-3000. Filament products used in Stage 2 are WOL3D PLA Pro+ Midnight Black (base), eSUN PLA+ Gray (bars), and Creality Hyper PLA White (discs).

Samples are presented as finish pairs at each thickness, in fixed order: A+B (0.1 mm Non-Ironing vs. Ironing) → C+D (0.3 mm Non-Ironing vs. Ironing) → E+F (0.5 mm Non-Ironing vs. Ironing). After pairwise evaluation, all six samples are displayed simultaneously for the Review Round (Phase 3).

4.4.3 Study Design and Participants. This is an exploratory qualitative pilot (N = 3). The goal is not statistical validation but hypothesis generation: identifying directional trends in height threshold and the modulating role of surface finish. Participants are recruited from IDC School of Design (students and faculty) who are familiar with optical illusions but naive to the specific sample set.

4.4.4 Data Collection. Data are collected through semi-structured interviews recorded with participant consent. Consent for recording and photograph use in documentation was obtained from participants in advance. The interviewer uses a structured question protocol (Table 4). Verbal descriptions of perceptual experience are the primary data; no rating scales are used.

The interviewer avoids the words “scintillation,” “illusion,” or “flickering” throughout Phases 1 and 2 to prevent priming effects. Questions B (Phase 2) and F (Phase 4) are specifically designed to elicit spontaneous reports of the self-erasing mechanism without researcher suggestion. The Review Round (Phase 3) is a methodological innovation: displaying all six samples simultaneously allows direct visual comparison across height levels that sequential presentation cannot support. R3 is the primary question for threshold identification.

5. Stage 2 Pilot Study Results

Thematic analysis of three pilot interview transcripts (N = 3; participants are IDC School of Design students and faculty, naive to the sample set) yielded four cross-cutting themes: height threshold, surface finish effects, self-erasing mechanism, and diagonal dominance. Figure 7 presents the full thematic map with representative participant quotes. Coding was conducted at the utterance level: initial codes and higher-order categories were generated directly from participant statements, and iterative cross-checking was performed throughout analysis.

5.1 Height Effect: Optimal at 0.3 mm

All three participants independently ranked Sample C (0.3 mm Non-Ironing) as the strongest panel, producing the counter-intuitive ordering: 0.1 mm < 0.5 mm < 0.3 mm (strongest).

P2: “the second panel was the most visible, frequent”; P3 explicitly ranked: “second one is the most strongest…the third panel is the most least.” P1 similarly selected the 0.3 mm panel as the overall strongest. This finding revises the Stage 1 single-observer impression that 0.5 mm was optimal. A proposed mechanism is that at 0.5 mm, the physical height of the raised discs casts strong lateral shadows onto surrounding grid bars, reducing the luminance contrast between disc and grid and paradoxically weakening the illusion—a non-monotonic height–strength relationship.

5.2 Surface Finish: Non-Ironing Stronger, Ironing Confounded

All participants perceived Non-Ironing panels as producing more prominent illusory dots than the corresponding Ironing panels. However, P2 identified a printing artefact in two of the Ironing samples—a diagonal misalignment of the ironing pass leaving an unironed strip—and reported: “there is a diagonal line where the ironing is missing…it is affecting the observation strength.” This observation provides in-session empirical confirmation of the committee’s variable contamination concern: Ironing introduces two simultaneous confounds, (a) specular gloss reducing effective luminance contrast, and (b) surface geometry distortion due to nozzle contact, which cannot be disentangled in the current design.

5.3 Self-Erasing Mechanism

P3 produced an explicit, unprompted description of foveal suppression precisely consistent with VanRullen and Dong: “if I focus on some point and I see a black dot, I try to look at the black dot and it suddenly disappears—then if I focus on that part where the black dot was, the surrounding points again start to bring black dots.” P1 and P2 did not produce analogous verbal descriptions. This 1-of-3 result is not interpreted as absence of the mechanism: the foveal suppression effect may be present for all participants at a sub-reportable level, with the bottleneck lying in verbalization rather than perception.

5.4 Diagonal Dominance

All three participants independently reported that illusory dots were perceived predominantly in diagonal zones of the panel rather than uniformly distributed. P1: “black dots are on the diagonals—the bottom-left corner and the top-right corner, or around those areas more”; P2: “along this diagonal belt I see more”; P3: “the corner of my eyes seeing black dots—immediately if I try to look it disappears.” This consistent spatial anisotropy is not predicted by the 2D psychophysical literature and likely reflects an interaction between the directional geometry of the raised relief and the fixed incident lighting angle.

6. Discussion

6.1 Theoretical Implications

The pilot data provide partial empirical support for the operability of the self-erasing mechanism in 3D reflective artefacts. P3’s description—dots appearing in the periphery and vanishing upon direct fixation—is phenomenologically identical to VanRullen and Dong’s account of attentional suppression. If replicated in a larger sample, this would constitute the first empirical evidence that foveal suppression of the scintillating grid illusion is robust to the transition from emissive (screen) to reflective (physical) display contexts.

The most theoretically significant finding is the non-monotonic height–strength relationship (0.1 mm < 0.5 mm < 0.3 mm). In a 2D display, increasing luminance contrast monotonically increases illusion strength. In a 3D physical object, the relief geometry introduces a competing factor: excessive height causes lateral edge shadows that fall on the white disc surfaces, reducing their apparent luminance and counteracting the contrast benefit of the raised structure. The optimal height of approximately 0.3 mm represents a balance point at which shadow-cast contrast enhancement is maximised without shadowing the disc surface itself. This non-monotonic behaviour is absent from the 2D literature and constitutes a novel empirical contribution.

Diagonal dominance—the consistent perception of illusory dots along diagonal axes rather than uniformly—is similarly unpredicted by 2D models. It plausibly reflects the directional geometry of incident lighting in the experimental setup. The fixed light-source angle may produce asymmetric shadow profiles across differently-oriented grid segments, modulating local luminance contrast in a direction-dependent manner. This finding has practical implications for the orientation of physical illusion-based notification panels relative to ambient light sources.

6.2 Design Implications for Peripheral Notifications

From an interaction-design perspective, the current pilot suggests three actionable constraints for physical self-erasing notification artefacts. First, parameter tuning should prioritise the “contrast balance point” around t = 0.3 mm rather than maximising relief height, because excessive height can reduce effective disc contrast through self-shadowing. Second, Non-Ironing surface behaviour should be treated as the default material target unless gloss can be isolated from geometry deformation. Third, installation geometry matters: because perception appears direction-dependent, panel orientation and light-source direction should be co-designed as part of the interface specification rather than treated as environmental noise.

These implications also refine the scope of where such a mechanism is likely to work in practice. The approach appears most suitable for low-urgency, awareness-level signals that benefit from peripheral salience but should disappear under direct attention (e.g., ambient task-state reminders). It is less suitable for critical alerts requiring immediate, invariant visibility under diverse lighting conditions. Framed this way, the contribution is not only that the illusion can be materialised physically, but that its reliable operation depends on a controllable “material + lighting + viewpoint” envelope.

6.3 Methodological Contribution

A key contribution is the explicit methodology for studying the scintillating grid illusion in physical artefacts: (1) the three-layer parameter structure (Table 1); (2) the standardised observational apparatus; and (3) the interview protocol (Table 4) that elicits perceptual reports without priming effects.

6.4 Limitations

The pilot study’s small sample size (N = 3) does not permit statistical inference; findings are directional and hypothesis-generative. The Non-Ironing vs. Ironing comparison was further compromised by printing artefacts identified in-session by P2—a diagonal misalignment in the ironing pass that confounds gloss with surface geometry change; future work must isolate these factors using a glossing spray applied to standard Non-Ironing prints. In addition, the apparent optical effect of Ironing can be filament-dependent (e.g., some filaments may appear more matte after Ironing), so filament manufacturer and product line should be explicitly reported as controlled metadata in subsequent replications. The viewing distance of 40 cm is not validated by prior literature for physical samples, and the current setup used a single controlled lighting configuration (40°, 5500 K, 575 lux), limiting claims about robustness across real-world illumination conditions.

Additional methodological constraints should be noted. First, sample presentation order was fixed (A→F), so potential order effects (adaptation, fatigue, expectation drift) cannot be ruled out. Second, outcome data are primarily verbal reports from semi-structured interviews. This is appropriate for exploratory phenomenology, but it does not provide continuous psychometric estimates of threshold or effect size. Third, the current analysis does not include objective luminance or reflectance measurements at the disc and bar surfaces, which limits mechanistic precision when attributing changes to contrast modulation or shadow geometry. Fourth, all participants were recruited from a single institutional context and were familiar with optical illusions in general, constraining external validity across broader user populations and task contexts.

Post-pilot committee feedback also identified a documentation-clarity limitation in the current Stage 2 write-up: although orientation was controlled by the experimenter, participant-facing orientation instructions were not explicit enough for third-party comparison.

Finally, the present study focuses on a single material system (PLA with fixed base geometry and fixed color assignment) and a single panel scale. The transferability of the observed parameter relationships to other materials, fabrication methods, panel dimensions, filament chemistries, and ambient environments remains untested. Accordingly, the current claims should be interpreted as establishing a controlled baseline and a set of falsifiable hypotheses, rather than a generalised design rule for all physical implementations.

7. Future Work

The present study is the first in a planned sequence:

EXP 1 has established the current baseline through a qualitative pilot (N = 3), indicating a non-monotonic height–strength relationship with an apparent optimum around 0.3 mm. Building on this, EXP 2 will perform precision height mapping (0.2, 0.3, 0.4 mm; Non-Ironing only) to locate the local peak more reliably and quantify the shape of the curve around the current optimum. EXP 3 will then isolate the surface-finish mechanism by comparing Non-Ironing, gloss-sprayed Non-Ironing, and Ironing at the same nominal height (0.3 mm), separating specular-gloss effects from geometry deformation introduced by nozzle contact.

After these controls are stabilised, EXP 4 will examine Z-axis ratio effects by varying bar thickness : disc thickness ({1:0.5, 1:1, 1:1.4, 1:2}) while holding the previously optimised height constant. EXP 5 will test the directional-lighting account of diagonal dominance by systematically varying panel tilt (90°, 75°, 60°, 45°). Finally, EXP 6 will scale the protocol to a full quantitative user study (N ≥ 20) to validate the parameter map and estimate effect robustness across participants.

To directly address the current limitations, the next protocol revision will also implement counterbalanced sample order to reduce adaptation and expectation effects, add repeated-measures psychometric ratings alongside interviews, and include objective photometric measurements (disc/bar luminance and surface reflectance) for mechanism-level analysis. External-validity constraints will be addressed by recruiting participants beyond a single institutional cohort and by replicating the study across multiple material systems, fabrication settings, and panel scales. Robustness to environmental variation will be tested by extending the current single lighting setup to multiple controlled illumination conditions around the present baseline.

In response to post-pilot committee feedback, the next revision will additionally add explicit orientation markings and orientation-matching instructions for all side-by-side comparisons.

8. Conclusion

This paper reports an ongoing design research programme investigating the scintillating grid illusion as a physical, sensor-free, self-erasing peripheral notification mechanism. I defined a three-layer parameter structure, established a standardised experimental apparatus, and conducted a qualitative pilot study (N = 3) that yielded four thematic findings. First, illusion strength is non-monotonic with respect to height: 0.3 mm outperforms both 0.1 mm and 0.5 mm, overturning the Stage 1 assumption that greater height monotonically increases strength. Second, Non-Ironing finish consistently produces stronger illusion than Ironing, with in-session identification of a printing artefact confirming that Ironing introduces dual confounds. Third, the self-erasing foveal suppression mechanism was explicitly reported by one of three participants, consistent with VanRullen and Dong. Fourth, illusory dots are perceived predominantly along diagonal peripheral zones, an empirically novel finding without precedent in the 2D literature. These findings motivate a revised experimental roadmap centred on precision height mapping around 0.3 mm and surface-finish isolation. The work contributes an experimental methodology and the first empirical data on the scintillating grid illusion in 3D-printed reflective artefacts.

In relation to the stated research questions, the present pilot provides partial rather than complete answers. The primary research question is addressed at proof-of-concept level: the illusion can be physically materialised in reflective 3D prints under controlled conditions, but robustness is not yet established. Secondary RQ1 (height effect) is partially resolved by the observed non-monotonic trend with a provisional optimum near 0.3 mm; however, quantitative threshold estimation remains open. Secondary RQ2 (surface finish) is directionally resolved in favour of Non-Ironing, but causal attribution is not yet clean because Ironing is confounded by geometry artefacts. Secondary RQ3 (standardised conditions) is resolved at baseline-protocol level through the current apparatus, yet generalisability across broader lighting, material, and user contexts remains an open question.

Acknowledgments

I thank Prof. Sugandh Malhotra for supervision and guidance, and Prof. Kanika Jolly and Prof. Yohan Engineer for their critical feedback, which substantially improved the experimental methodology. This work was conducted at IDC School of Design, IIT Bombay.

References

1. Alcon Lighting. 2023. Color Temperature Guide.
2. Saskia Bakker, Elise van den Hoven, and Berry Eggen. 2015. Peripheral interaction: Characteristics and considerations. Personal and Ubiquitous Computing 19, 1 (2015), 239–254.
3. Ludimar Hermann. 1870. Eine Erscheinung simultanen Contrastes. Pflüger Archiv für die gesamte Physiologie des Menschen und der Tiere 3, 1 (1870), 13–15.
4. Jiaqi Ju, Dahua Chen, and Yandan Lin. 2012. Effects of corrected color temperature on spatial brightness perception. Color Research & Application 37, 6 (2012), 450–454.
5. Elvin Karana, Bahar Barati, Valentina Rognoli, and Anouk Zeeuw van der Laan. 2015. Material Driven Design (MDD): A method to design for material experiences. International Journal of Design 9, 2 (Aug. 2015), 35–54.
6. Jennifer Mankoff, Anind K. Dey, Gary Hsieh, Julie Kientz, Scott Lederer, and Morgan Ames. 2003. Heuristic evaluation of ambient displays. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI ’03). ACM, 169–176.
7. Toyomi Matsuno and Yasuichi Sato. 2019. Dissociation of perceived size and perceived strength in the scintillating grid illusion. Journal of Vision 19, 13 (Nov. 2019), 15.
8. Tara Matthews, Anind K. Dey, Jennifer Mankoff, Scott Carter, and Tye Rattenbury. 2004. A toolkit for managing user attention in peripheral displays. In Proceedings of the 17th Annual ACM Symposium on User Interface Software and Technology (UIST ’04). ACM, 247–256.
9. Kun Qian, Yuki Yamada, Takahiro Kawabe, and Kayo Miura. 2009. The scintillating grid illusion: Influence of size, shape, and orientation of the luminance patches. Perception 38, 8 (2009), 1172–1182.
10. Peter H. Schiller and Christina E. Carvey. 2005. The Hermann grid illusion revisited. Perception 34, 10 (2005), 1375–1397.
11. Michael Schrauf, Bernd Lingelbach, and Ernst R. Wist. 1997. The scintillating grid illusion. Vision Research 37, 8 (1997), 1033–1038.
12. Michael Schrauf and Lothar Spillmann. 2000. The scintillating grid illusion in stereo-depth. Vision Research 40, 7 (2000), 717–721.
13. Chelsea Tymms, Denis Zorin, and Esther P. Gardner. 2018. Tactile perception of the roughness of 3D-printed textures. Journal of Neurophysiology 119, 3 (March 2018), 862–876.
14. Rufin VanRullen and Tao Dong. 2003. Attention and scintillation. Vision Research 43, 21 (Sept. 2003), 2191–2196.
15. Mark Weiser and John Seely Brown. 1996. Designing calm technology. PowerGrid Journal 1, 1 (1996), 75–85.