A Mixed-Reality Method for Capturing User-Generated Spatial Meaning in Built Environments
Oğuz Emre Bal
Faculty of Architecture · Istanbul Technical University
Mixed Reality Method
Place Attachment
Spatial Computing
Built Environment Analysis
Pilot: n=7 · 56 anchors
realityoverlap.com · Ethics approved · ITU No. 2026-01-822
The Observation
"Architectural space is not exhausted by plan, program, or geometry. Buildings are also constituted through repeated occupation, bodily orientation, memory, interpretation, and social encounter."— Introduction, §1
The meaning of a building emerges at the scale of the micro-location — the specific, bounded sites where users pause, orient, remember, and return.
A threshold where one habitually pauses
A bench where a life-defining decision was made
A tree-shelter where a community gathers after exams
Fig. 1 — Layers of architectural meaning
Lived Meaning
occupation · memory · social encounter
← targeted by Reality Overlap
Program
function · use · brief
Geometry
plan · section · form
↑ The layer existing methods have missed
The Blind Spot
Existing methods excel — but miss one layer
Space Syntax
✓Movement & accessibility logic
✓Topological configuration
✓Socio-spatial structure
✗Why a specific corner carries a life-defining memory
Behaviour Mapping
✓Where people are & what they do
✓Spatial activity distribution
✓Observable behaviour patterns
✗Sees bodies, not stories or biographical depth
Post-Occupancy Eval.
✓Performance & satisfaction data
✓User feedback at scale
✓Evidence-based improvement
✗Cannot locate satisfaction to a specific micro-place
Missing: first-person · in-situ · spatially precise · socially accumulative meaning — for familiar occupants of a building
Target Phenomenon
Cumulative Interpretive Accretion
The spatially explicit build-up of user-generated meanings at micro-locations through repeated acts of marking, describing, sharing, encountering, and responding. Requires familiar occupants — first-time visitors annotate impressions; long-term occupants annotate histories.
①
Meaning Density
Concentration of interpretation acts at a micro-location
②
Semantic Diversity
Difference in type, content, and valence of interpretations
③
Interpretive Overlap
Multiple users converging independently on same location
④
Social Reorientation
Prior traces altering later users' attention and response
Mixed Reality is our instrument — the way a microscope is an instrument. We study what it reveals, not the instrument itself.
The Method
Reality Overlap — Three Integrated Layers
Layer 1
Meta Quest 3
Unity / C# · Passthrough MR
→ Spatial anchor creation
→ Voice + passthrough screenshot
→ Real-time Firebase sync
→
Layer 2
Firebase
Firestore + Storage
→ Persistent anchor storage
→ Multi-session sync
→ Shared-trace access control
→
Layer 3
Three.js + LiDAR Twin
Browser analysis
→ 3D twin overlay
→ Timeline path playback
→ AI anchor classification
Session Origin Drift — two-decimal-place spatial alignment calibration between headset origin and LiDAR model
Not a technical pin — a persistent, spatially registered, user-generated semantic marker. Each anchor carries four data types simultaneously. After collection, an AI model (llama3.1:8b) classifies it into one of six categories.
Fig. 3 — Anchor anatomy
Spatial Datum
Calibrated coordinates in LiDAR twin. Timestamp and viewpoint at creation.
Semantic Datum
Voice annotation or text — what this location means to this user, in their own words.
Media Datum
Passthrough screenshot at moment of creation — visual context of the physical location.
Both stages are required. Stage 1 captures personal meaning-making; Stage 2 captures trace-mediated social reorientation (CIA Construct 4).
01
Private Meaning-Making
Participant moves through ground-floor micro-spaces wearing the headset, guided by researcher prompts — placing anchors independently without access to others' traces.
→Free-roam MR session, 45–60 min
→Voice annotation + passthrough screenshot per anchor
→Visibility decision (private / shared) per anchor
02
Social Encounter & Response
Participant revisits with shared anchors from previous participants visible — encountering and responding to others' interpretations overlaid on the same physical space.
→Shared traces rendered in the headset view
→Responses and interactions recorded
→Post-session debriefing + 3 survey rounds
Sampling: 24–40 participants (full study) · Pilot: n=7 · Purposive maximum-variation · ≥1 academic semester in building
Pilot Study · ITU Taşkışla · April–May 2026
What We Did — and What We Found
7
participants · architecture students
56
usable spatial anchors
83%
daily building presence
6
anchor categories (AI-classified)
Why familiar occupants? — The phenomenon we study (CIA) only exists where people already carry biographical meaning. First-time visitors annotate impressions; long-term occupants annotate histories.
Ground floor + central courtyard · April 7 – May 2, 2026 · 45–60 min per session · 3 survey rounds
Transfer conditions: stable digital twin · in-situ MR · micro-location externalization protocol → applicable to any occupied building
Fig. 4 — Real-time path and anchor tracking in 3D space
Anchor Taxonomy · llama3.1:8b AI Classification
Six Anchor Categories
Developed from prospect-refuge theory (Appleton 1975) and refined through pilot data. Threshold was not in the original framework — it emerged from the pilot itself.
33.9%
Refuge
"Post-exam quiet corner." "Rain-shelter tree." "Benches too cold in winter."
21.4%
Memory
"The bench where I decided to study here." "A COVID memory." "A ritual gathering."
17.9%
Social
"Post-exam critique terrace." "Informal group meeting." "The cat corner."
12.5%
Threshold ★
"Dining-to-courtyard passage." "Building entrance." "Stair landing." (Emerged from data)
12.5%
Prospect
"Spot for watching others." "Open courtyard position." "Visual survey point."
1.8%
Discovery
Unexpected finds. Novel spatial interpretations. Rare in familiar long-term users.
AI anchor classification interface in the browser platform
Key Finding 1
Meaning Does Not Follow Movement
The most important spatial finding: anchor density does not track where people walk most. It tracks where people pause, gather, shelter, and remember.
HIGH-TRAFFIC ZONE — Central Courtyard Pool
Visually prominent · high configurational integration · high foot traffic
2
anchors across 2 participants The most visible feature attracted the fewest meaning annotations.
PERIPHERAL ZONE — North-West Tree-Shelter
No programme · no architectural emphasis · syntactically peripheral
5
anchors from 3 independent participants Rain shelter · post-exam retreat · habitual gathering. A different story from each participant — the same location.
Space syntax correctly models the tree-shelter as spatially marginal. But spatially marginal ≠ meaningfully marginal. For familiar occupants, syntactically peripheral spaces carry the richest biographical content.
Fig. 5 — Anchor density map overlaid on Taşkışla ground floor
Key Finding 2
The Biographical Depth of Micro-Places
Anchor notes reveal that micro-places carry specific, precise, irreplaceable personal content. This is the material that no existing method can surface or locate.
Memory
"Before I was a student here, the first time I sat here was when I decided to study at this school."
A specific courtyard seat carries the memory of a life-defining decision. No architectural analysis, no satisfaction survey would ever surface this — or locate it to that exact bench.
Refuge
"In rainy weather, we used to stand under this tree and chat."
A tree-shelter as social refuge. Encodes season, behaviour, and community simultaneously. The architect did not design this function — the occupants created it.
Social · Cross-user convergence
"This is where people do the long critique of the exam after assessments."
Identified independently by 2 different participants without coordination — one of the strongest instances of interpretive overlap in the pilot.
Refuge · Retrofit signal
"The benches here are very cold in winter and rainwater pools on them."
Material complaint embedded in a meaning annotation. Retrofit-relevant intelligence located to specific coordinates — without a POE survey instrument.
Key Finding 3
Survey Evidence (n=6 · 1–7 scale)
Three survey rounds measured whether Reality Overlap altered participants' relationship to the building. Post-session: 67% said the MR experience changed how they see the ground floor. Cybersickness: 0.5/3 mean — minimal.
5.83
Reflective Annotation
"Leaving an anchor prompted me to reflect deeply on why this place is important to me." — The act of annotation is itself a reflective instrument.
out of 7
5.33
Co-presence Through Traces
"I strongly felt the presence of other participants through their digital traces." — Shared annotation creates a form of social presence even when alone in the space.
out of 7
6.17
Altered Spatial Memory — HIGHEST
"After reading others' notes, I will remember the marked places differently." — Reading others' biographical meanings permanently altered how participants will remember the space. MR annotation changes prospective spatial memory.
out of 7
50% said another participant's note surprised them or changed their thinking · Cybersickness 0.5/3 — confirms technical feasibility
Design & Practice Implications
Four Actionable Outputs from the Pilot Data
Conventional analysis produces none of these. Reality Overlap produces all four simultaneously from a single session.
① Micro-Spatial Retrofit Targets
Refuge anchors contain embedded material complaints: cold bench surfaces, pooling rainwater, insufficient shelter. These are coordinate-located diagnostic signals — a retrofit brief that targets exactly where occupants experience discomfort, without a full POE cycle.
② Architecturally Unmarked Thresholds
Threshold anchors cluster at transition zones with high social significance but zero architectural marking. The dining-to-courtyard passage is described by 4 participants as a habitual threshold — the building provides no spatial cue. The data makes a case for threshold articulation.
③ Informal Social Infrastructure
Social anchors identify the post-exam critique terrace, the shelter tree, and the cat corner as informal social infrastructure. Their spatial conditions — seating capacity, shade, acoustic shelter — determine whether the social practices attributed to them can continue. The data names these conditions precisely.
④ Longitudinal Change Detection
Applied across an academic year, repeated application detects whether seasonal change, physical modification, or social reorganisation alters meaning density and type. A new monitoring capability: tracking not spatial performance, but spatial meaning — over time.
What This Contributes
A missing analytical layer — not a replacement
Established toolkit
Space Syntax — configuration & movement prediction
Behaviour Mapping — observable activity distribution
Post-Occupancy Eval. — performance & satisfaction
??? — spatially precise, in-situ, lived meaning layer (absent)
+
Reality Overlap adds — pilot-validated
Cumulative Interpretive Accretion
✓ First-person in-situ meaning (n=7 pilot, 56 anchors)
✓ Meaning ≠ movement (peripheral > central, empirically)
"We are not claiming Reality Overlap replaces space syntax, behaviour mapping, or POE. We are claiming it captures a different stratum of spatial knowledge — one that existing instruments cannot reach."
Interactive Resources
Explore Further
Platform · Interactive manuscript map · All sections, argument flow, 4 density levels
