A gaze occurred
Someone’s eyes were directed toward you in the world.
Selected Work
Between Strangers
Reconstructing subtle embodied social experience as designable relations before symbolic-form selection.
An underground corridor. A stranger’s gaze lands on you for two seconds. What just happened—and what could a design say about it?
01 · The design problem
A Glance That Leaves No Trace
Public-space gaze is often brief, backgrounded, embedded in walking, and difficult to describe as a stable event.
Yet it can reorganize attention, interpersonal boundaries, and possible interaction. Something may have happened—it simply left no legible trace.
Two ways to fail frame the problem.
One is to draw the meaning: a force striking a body, a spotlight on a person—images that turn an interpretation into a verdict.
The other is to stop at findings that describe the experience but never say what to do with it.
Both fail at the same point: they skip the relations.
The challenge was not to invent a symbol for “being looked at.” It was to find out which parts of this subtle experience could be reconstructed as relations a designer could work with, before deciding how those relations should look or behave.
Four states a design must not conflate.
A gaze was noticed
It entered your awareness as a cue, distinct from the background.
An exchange formed
You looked back, and two gazes met in a shared relation.
A meaning was construed
It came to mean something to someone—invitation, scrutiny, threat, or nothing at all.
Keeping these states distinct preserves room for interpretation. Collapsing them risks turning inference into judgment.
The early direction crossed all four; the proposition stops at the third.
02 · An early direction
I Drew the Answer Too Early
The project began outside the research: standing in a transfer corridor in Shanghai at rush hour—surrounded by people, feeling completely alone.
In an atomized city, the social-emotional support that once came from stable relationships is fragmenting. Could the small, non-verbal encounters of public space become a real source of warmth?
That ambition was the reason to start—and it was too big to design.
The project soon ran into a more basic problem: I did not yet have a designable account of the experience itself.
I could not yet say what entered awareness, what sustained an episode, what closed it, or which relations a design could legitimately work with.
Before I had that account, I did what felt natural: I drew the answer.
I designed a first XR concept around three symbols: a large transparent bubble around each person, a small bubble materialized where two gazes met, and ground ripples that carried the echo of every interaction.
The story wrote itself: the boundary could feel protective without being closed; a glance could become a shared object; a passing encounter could leave the space a little warmer.
The ceiling
The intuition had warmth, but the design was symbol-first.
Each symbol was a verdict: this boundary is permeable, this gaze is a connection, this interaction charges you up.
A heartbeat pulse when eyes met, and energy absorbed on parting, turned subtle relational states into emotional judgments the system had no evidence—and no right—to make.
Nothing in the design could be traced back to what the experience actually supported.
The form survived. The reasoning behind it did not.
The membrane-and-bubble family returns in the final proposition—but only because it was re-derived, element by element, from what the experience itself could support. What follows is that re-derivation.
03 · Entering the experience
A Walk, Then a Conversation
16native Mandarin-speaking participants
13:02hours of interviews
160,162Chinese characters transcribed
40–60minutes per session
Participants watched a first-person, five-minute walk through an underground pedestrian passage in urban Shanghai.
They returned to moments when another person, a face, or a gaze entered attention—and to comparable episodes from everyday life.
The interview followed what was noticed, what remained in the background, what changed after noticing, and how the episode ended.
Every evidence unit received a unique ID, so later design decisions could be followed back to the account that supported them.
The gaze direction of people in the video was not measured. The material consists of situated descriptions, not recordings of real-time cognition.
Two Different Jobs in the Design Process
Perceptual analysis
What allowed gaze to enter—or leave—experience?
Walking and collision avoidance often stayed primary. A person could become noticeable before their gaze was localized.
Participants associated noticeability with field of view, environmental distraction, passage pressure, duration, and persistence. Looking away or passing could close the episode and return attention to walking.
These findings were translated into conditional requirements: make a gaze event available without demanding constant monitoring; distinguish a visible person from a gaze directed toward self; show whether the relation is brief, continuing, or over; and make withdrawal and return to baseline perceptible.Conceptual-metaphor analysis
What relations organized participants’ descriptions?
Across different accounts, gaze was described as something that travelled, arrived, stayed, or met another gaze.
It also carried content, exerted pressure, crossed a boundary, illuminated a person, or exposed a watcher.
These expressions were not treated as ready-made design ideas. Each metaphorical mapping was reconstructed into its underlying spatial and event relations, then checked against its evidence record.04 · The critical turn
The Interviews Did Not Tell Me What to Draw
The two analyses defined functions and relations, not a visual form.
A participant might describe a prolonged gaze as pressure or intrusion. Drawing pressure striking a body would preserve that particular construal—but it would turn a situated interpretation into a judgment made by the system.
That is exactly the failure the early direction had already made—now visible from the other side.
The interviews gave the design relations, not images. That was the contribution.
The question became what to do with those relations—which to keep, which to move, which to refuse.
05 · The evidence
What the Evidence Could Support
For every audited metaphorical mapping, I separated the participant’s situated meaning from the more basic spatial or event relation that organized it.
Compatible relations were then synthesized across records.
The result describes what the evidence supports—not what I wanted the final imagery to look like.
This supports a directed path with an origin, movement, arrival, persistence, withdrawal, and restoration.
This supports contact formation, maintenance, separation, and possible recontact between two directed gaze relations.
This supports a conditional bounded-access branch—not a claim that every sustained gaze is intrusion.
These become attached qualities or conditional modules. They modify the backbone and do not form a universal event sequence.
Output of the metaphor branchEvidence-Side Image-Schematic SystemA cross-record synthesis of evidence-supported relations. Every component remains traceable to its source mapping, and each individual record instantiates only part of the system.
06 · The design decisions
Keep, Move, Refuse
The evidence gave me relations, not instructions. Selective projection decided what to do with them—what to keep, what to move, what to constrain, and what to refuse.
Four inputs met in the projection: the evidence-side relations, the perceptual functions, the design value, and the reference situation.
Keeping meaning with the people is not a slogan here—it became a set of concrete decisions about what the system would never show.
Let gaze emerge from the walking background; locate its source and direction; distinguish brief, continuing, and ended relations; preserve withdrawal and return to baseline.
Path, contact, boundedness, content, pressure, illumination, and detection were available as evidence-supported relational resources—not mandatory visual features.
Support gaze exchange as a recognizable, optional entry point into social interaction. Recognition must not become a demand to respond.
The imagery must coexist with walking, route maintenance, passing strangers, short encounters, and each participant’s first-person viewpoint.
Gaze as directed movement, contact, and withdrawal
The object shows source, direction, arrival, persistence, and withdrawal.
Arrival moves to the space between, not the body
Endpoints leave the person; the object stops at a common relation position between the two people.
Pressure and intrusion become limits, not states
Objects stay outside both domains; sharpness stays low; no force acts on the person.
Interior without content
Objects keep a container interior; the system never fills, labels, or classifies it.
No spotlight
Illumination and exposure are excluded; event changes appear in the relation, not on the person.
Geometry without judgment
“Catching” is reduced to contact geometry; reciprocity shows without accountability or blame.
07 · One relation system, three carriers
One Relation System, Three Visual Carriers
Only after the relations had been selected and reorganized could the design-side system be specified. It fixes the sequence below while leaving shape, material, color, and animation open.
Output of selective projectionDesign-Side Image-Schematic SystemA carrier-independent behavior specification derived from the evidence-side relations, conditional perceptual requirements, design value, and reference situation.
The design-side system represents observable gaze relations and state changes, while specific meaning construal remains participant-level.
Three Possible Carriers
Membrane–Bubble
Transparent domains and bubble-like objects make generation, movement, merge, split, and reabsorption physically continuous.
Wireframe–Ring
Low-occlusion outlines reduce visual mass, but their orientation and state changes invite competing spatial readings.
Field–Condensation
Particles make appearance and disappearance natural, but the boundary and object roles become harder to distinguish.
The same participant domains, gaze objects, common relation position, and merge/split behavior were expressed through three different visual languages.
Five external designers with graduate-level interaction-design education and experience in XR imagery compared the candidates.
Their review supported Membrane–Bubble for further development—it did not validate the method, behavior script, or experiential effect.
The Proposition: Membrane–Bubble
These images specify a visual proposition, not a finished XR experience.
The complete transition sequence, sensing implementation, visual load while walking, situated interpretation, and social consequences still require prototyping and evaluation.
08 · What this produced
What the Project Actually Produced
Produced
A traceable way to move from first-person accounts to design requirements; an evidence-side reconstruction; a documented set of projection decisions; a design-side behavior script; three visual carriers; and a preferred proposition for further development.
Still open
Whether gaze can be sensed reliably, whether the imagery remains legible without disturbing walking, how participants interpret and act on it, and what consequences it creates for people who did not choose to use the system.
The early direction visualized meanings. The proposition visualizes relations—and leaves meaning with the people.
The contribution is not a finished visualization of gaze. It is a method for reconstructing subtle embodied social experience as designable relations before symbolic form is chosen: a way to make an invisible relation legible without deciding in advance what it means.
Demonstrated here through a proposition that can now be built, contested, and revised.