Substratism VIZ

A dependency-free, offline visual and sonification laboratory for exploring the eight-item Substratism Scale and the moral, resource-allocation, and advocacy associations reported by Ladak et al. (2026).

Launch the interactive moral substrate observatory →

GitHub READMEs cannot execute JavaScript, so the image above is a deterministic preview. Select it to use the full instrument on GitHub Pages, or run the same files locally with no internet connection.

Start offline

Clone or download the repository and open index.html in a modern browser. There is no installation step, CDN, framework, analytics service, network client, or upload path.

For microphone analysis, serve the directory on local loopback so the browser can grant secure-context-compatible media permission:

python3 -m http.server 8000

Then open http://localhost:8000. Generated sonification, local-file analysis, PNG/JSON/WAV export, and most visual functions also work directly from file://.

Four coupled receivers

Receiver	Research fixture	Authored visual response
Concept map	Definition, validated scale, relation to other prejudices, measured outcomes	Four-branch topology rebuilt from the supplied research map
Scale lattice	Eight 1–7 responses with item 2 reverse-scored	Corrected radial coordinates and active-item traversal
Moral delta	Study 4–5 slopes, standardized coefficients, odds ratios, and sample means	Biological-to-digital outcome markers
Effect field	Substratism plus four comparison predictors across six outcomes	Signed coefficient vectors kept separate by β/OR metric

These views are not separate animations. They receive the same corrected scale state, active item, outcome reconstruction, and local audio analyser bands.

Audio controls the visual

The visual response can be driven by one of three browser-local sources:

1. a generated eight-step scale sonification;
2. microphone analysis;
3. analysis of a local audio file.

The generated receiver traverses the same eight items shown in the scale lattice and concept topology. The corrected scale mean controls tempo, biological/digital voice balance, filter frequency, and interval separation. RMS, bass, mid, and high-frequency energy then drive grid motion, glow, reach, displacement, node traversal, and field tension.

The mapping is deterministic for a given scale state, tuning, duration, and sample rate. It is an authored explanatory receiver—not an intrinsic sound of substratism and not evidence for any moral or ontological claim.

Optional rendering

Live exploration requires no export. When an artifact is useful, the browser can create:

• a PNG snapshot of the current visual;
• canonical JSON containing the current responses, corrected score, receiver state, and reconstructed outcomes;
• a deterministic 48 kHz mono WAV of the generated scale sonification;
• an 8, 16, or 30 second WEBM canvas recording, optionally including the generated audio path.

All rendering happens locally in the browser.

Exact scale implementation

Participants respond from 1 (strongly disagree) to 7 (strongly agree). Item 2—“Artificial intelligences should have rights”—is reverse-scored:

corrected item 2 = 8 − response
substratism score = mean(all eight corrected items)

The workbook supplied with this project groups items 1, 3, 5, and 7 as substrate-explicit. Items 2, 4, 6, and 8 capture rights/treatment, functional limits, and capacity denial. The application exposes both four-item channel means without presenting them as independently validated subscales; the paper reports a single unidimensional construct.

Fixture	Published value
Scale items	8
Response range	1–7
Reverse-scored items	1 (item 2)
Standardized factor loadings in final Study 1 scale	0.69–0.84
Cronbach’s α across studies	0.84–0.91
Preregistered studies	5
Total participants	2,129

Outcome reconstruction

The moral-delta receiver is centred on each study’s published sample mean and outcome mean.

For continuous outcomes:

receiver value = published outcome mean + b × (configured score − study score mean)

For binary advocacy outcomes:

logit(receiver probability)
  = logit(published outcome proportion)
  + ln(OR) × (configured score − study score mean)

This makes the coefficient direction and approximate scale explorable. It does not recreate the authors’ full regression models, confidence intervals, covariate settings, or individual-level predictions.

Published coefficient reference

The repository’s original regression chart is retained as a static reference:

The application centralizes the plotted values in js/substratism-core.js and tests representative fixtures, including β = −0.31 for AI-versus-animal donations, OR = 0.66 for the Study 4 charity link, and OR = 0.74 for the Study 5 petition link.

Claim boundary

Layer	Status
Eight scale statements and reverse-scoring rule	Published research fixture
Reliability, sample sizes, study means, β values, slopes, and odds ratios	Published research fixtures transcribed from the paper and project chart
Four concept-map branches and item classifications	Project-supplied explanatory organization
Corrected overall mean	Exact implementation of the reported scale rule
Explicit and treatment/capacity channel means	Descriptive project receivers, not validated subscales
Outcome values shown for a configured score	Transparent reconstruction, not the authors’ fitted prediction
Colours, topology, motion, synthesis, tuning, and audiovisual coupling	Authored observation receivers

The paper defines substratism descriptively: different moral consideration or treatment based on non-biological substrate. It does not itself settle whether substrate differences are morally justified. This application is educational, not diagnostic.

Project structure

index.html                         Offline application shell
style.css                          Responsive industrial visual system
js/substratism-core.js             Scale, scoring, coefficient, and outcome fixtures
js/sonification.js                 Generated, microphone, file, analyser, and WAV paths
js/visuals.js                      Concept map, lattice, outcome, and coefficient receivers
js/app.js                          Interaction, inspection, state, and render orchestration
assets/substratism-viz.svg         Deterministic README preview
8-Item Substratism Scale…xlsx      Project-supplied classified scale table
Screenshot From 2026-07-26…png     Original project-supplied concept map
scripts/render_preview.py          Standard-library preview renderer
tests/                             Scale, coefficient, determinism, and offline checks
.github/workflows/pages.yml        Static GitHub Pages deployment

The research paper, executive summary, supplied workbook, static coefficient chart, and its chart-generation script remain in the repository as source material.

Verification

Node.js is used only for development checks:

npm test

The suite checks the eight-item instrument, reverse scoring, channel arithmetic, published coefficient fixtures, outcome reconstruction, deterministic WAV bytes, offline packaging, accessibility markers, render paths, and GitHub Pages workflow.

Regenerate the README preview with:

python3 scripts/render_preview.py

Sources

• Ali Ladak, Janet V. T. Pauketat, Jacy Reese Anthis, Steve Loughnan, and Matti Wilks, “Substratism: Conceptualizing and measuring moral bias against AI,” Computers in Human Behavior 185 (2026), 109120.
• Open data, code, and experimental materials
• Study preregistrations

Licence

MIT License. Copyright 2026 QSOL-IMC / Quantum-Sourced Optimization-Logic Integrated Meme Company.