Intelligent typography for neurodivergent readers in Extended Reality
AdapTypeXR is a Unity research platform developed at UHasselt — Digital Future Lab that studies how typography, animated text, and adaptive text presentation affect reading performance and comprehension in neurodivergent users within XR environments.
Using real-time eye-tracking and physiological sensors on the Varjo XR-4, AdapTypeXR measures how font choice, text layout, and presentation dynamics influence:
- Reading speed and efficiency
- Comprehension and recall
- Cognitive load (pupil dilation, fixation duration)
- Oculomotor behaviour (saccades, regressions, fixations)
Primary participant profiles: dyslexia, ADHD, autism spectrum, and neurotypical controls.
| Component | Specification |
|---|---|
| Headset | Varjo XR-4 |
| Eye Tracking | Varjo XR SDK (200 Hz, binocular, pupillometry) |
| Render Pipeline | Unity URP |
| Unity Version | Unity 6 (6000.0.x LTS) |
| XR Framework | OpenXR + XR Interaction Toolkit |
| Text Rendering | TextMeshPro SDF |
Assets/AdapTypeXR/
├── Scripts/
│ ├── Core/
│ │ ├── Interfaces/ — Service contracts (IEyeTrackingService, etc.)
│ │ ├── Models/ — Domain types (GazeDataPoint, TypographyConfig, etc.)
│ │ └── Events/ — ReadingEventBus + typed domain events
│ ├── Controllers/ — ReadingSessionController (session state machine)
│ ├── Presenters/ — BookPresenter (3D book + page management)
│ ├── Services/ — VarjoEyeTrackingService, MockEyeTrackingService
│ ├── Repositories/ — CsvDataCollectionRepository (async CSV output)
│ ├── Typography/ — TextRendererController, TypographyAnimator, FontProfileFactory
│ └── Infrastructure/ — AppBootstrapper (composition root), asmdef
├── Scenes/ — Unity scenes
├── Fonts/ — TMP SDF font assets
├── Materials/
├── Prefabs/
└── Data/Sessions/ — Output data (gitignored)
The system follows SOLID and GRASP principles with clear separation of concerns:
AppBootstrapper (Composition Root)
│
├── ReadingSessionController [Controller — session lifecycle]
│ ├── IEyeTrackingService [Adapter — Varjo SDK / Mock]
│ ├── IDataCollectionRepository [Repository — CSV]
│ └── IBookPresenter [Presenter — 3D book]
│ └── ITextRenderer [TextRendererController + TypographyAnimator]
│ └── ITypographyAnimationStrategy [Strategy — RSVP, WordByWord, etc.]
│
└── ReadingEventBus [Observer — cross-system events]
Key patterns used: Adapter, Strategy, Observer, Factory, Repository, Composite, Façade, Builder.
| ID | Font | Animation | Research Hypothesis |
|---|---|---|---|
| C1 | Arial | None | Neutral baseline |
| C2 | OpenDyslexic | None | Dyslexia-targeted letterforms reduce decoding effort |
| C3 | Atkinson Hyperlegible | None | High disambiguation reduces fixation count |
| C4 | Atkinson Hyperlegible | Word-by-word highlight | Guided attention reduces ADHD lost-place errors |
| C5 | Arial | RSVP 250 WPM | Eliminating saccades improves speed for dyslexic readers |
| C6 | Atkinson Hyperlegible | Bionic Reading | Bold anchors reduce regression rate |
- Unity 6 (6000.0.x LTS)
- Varjo Base installed and running
- Varjo XR-4 connected
- Varjo XR SDK for Unity (download from Varjo Developer Portal)
- Clone this repository
- Open in Unity Hub — select Unity 6
- Import Varjo XR SDK:
Window > Package Manager > + > Add package from disk - Configure OpenXR:
Edit > Project Settings > XR Plug-in Management- Enable OpenXR
- Enable Varjo feature set
- Enable Eye Tracking in Varjo settings
- Open
Assets/AdapTypeXR/Scenes/MainReadingScene.unity - Import font assets into
Assets/AdapTypeXR/Fonts/and generate TMP SDF atlases - Press Play — on non-Varjo machines,
AppBootstrapperwill useMockEyeTrackingService
Set Use Varjo Eye Tracking = false on the AppBootstrapper component in the scene.
Enable Auto Start Demo Session to auto-launch a demo reading session on Play.
Sessions are written to:
- macOS/iOS:
~/Library/Application Support/<company>/<app>/Sessions/<sessionId>/ - Windows:
%APPDATA%\..\LocalLow\<company>\<app>\Sessions\<sessionId>\
Each session folder contains:
session.json— participant metadatagaze.csv— raw gaze samples (200 Hz)metrics.csv— per-passage aggregated metrics
This repository includes four AI agent configurations in .claude/agents/:
| Agent | Role |
|---|---|
orchestrator |
Sprint planning, backlog management, coordination |
xr-developer |
Unity/Varjo implementation, architecture, code review |
typography-readability |
Font science, text layout, animation research |
neurodivergent-expert |
Cognitive load, participant profiles, ethics, study design |
See AGENTS.md for full documentation.
- All participants provide informed consent before data collection
- Gaze and physiological data are anonymised from the point of capture
- Participants may withdraw at any time
- Study protocol approved by UHasselt ethics committee before data collection begins
- Maximum continuous XR exposure: 20 minutes per session block
UHasselt — Digital Future Lab Faculty of Sciences and Technology
- Rello, L. & Baeza-Yates, R. (2013). Good fonts for dyslexia. ASSETS '13.
- Rayner, K. (1998). Eye movements in reading. Psychological Bulletin.
- Sweller, J. (1988). Cognitive load during problem solving. Cognitive Science.
- Beatty, J. (1982). Task-evoked pupillary responses. Psychological Bulletin.
- Braille Institute (2019). Atkinson Hyperlegible font design rationale.