NeuroTerroir Research Ledger

What this project will not claimA fly connectome cannot judge wine, reproduce human flavor perception, establish quality, or reveal terroir by itself. NeuroTerroir will use documented fly olfactory pathways as an inspectable teaching scaffold. Human sensory language and wine chemistry remain separate layers.

Review protocol

How evidence enters the build

A source can inform the lab only when its role and limits are visible. Code review is for architecture, controls, and failure lessons—not visual or functional cloning.

Primary first

Papers, official datasets, and maintained documentation outrank summaries and demos.

Map cautiously

Odorant → receptor → glomerulus claims need direct evidence; uncertain mappings stay uncertain.

Keep controls

Biological wiring must be compared with silenced, input-free, and rewired conditions.

No borrowed theater

The lab will not copy another project’s visual identity, game loop, or unsupported claims.

Code provenance

GitHub projects under review

The list is intentionally finite. Each entry records why the code matters and the line NeuroTerroir will not cross.

awesome-fly

Reviewed

Curated catalog of contemporary fly-connectome experiments and their stated limitations.

Used for: Prior-art inventory, vocabulary, and links to source repositories.
Will not copy: Any listed project’s mechanic, visual presentation, or claim framing.

fly-connectome-template

Reviewed

React and Three.js workbench built around MaleCNS and Flybody assets.

Used for: Architectural comparison and its unusually clear separation between anatomy, model output, and display.
Will not copy: Its source-available code, interface styling, or assets. NeuroTerroir’s first milestone is an independent static implementation with no template-derived code.

fly-odor-onoff

Reviewed

Reproducible odor ON/OFF interventions using a filtered MaleCNS LIF graph.

Used for: Input-pathway audits, intervention structure, and honest fixed-circuit limitations.
Will not copy: Its binary odor experiment or DM1/VA2 readout as NeuroTerroir’s primary experience.

The Fly’s Table

Planned

Includes a mushroom-body model and separate odor-conditioning experiments with dopamine-cut controls.

Used for: Measured PN → Kenyon cell → MBON organization and intervention design.
Will not copy: Card play, shared persistent brain, reward mechanic, or project-defined learning rule.

boltzmann-fly

Planned

Connectome-masked model using a right-hemisphere mushroom-body circuit and rewiring controls.

Used for: Topology-control design and transparent comparison with dense and randomized models.
Will not copy: Consumer-behavior prediction, energy-model framing, or its trained couplings.

Drosophila brain model

Planned

Research implementation of the Shiu et al. connectome-based leaky integrate-and-fire model.

Used for: Scientific simulation assumptions, activation/silencing patterns, and reproducible experiments.
Will not copy: Whole-brain fidelity claims unsupported by NeuroTerroir’s reduced teaching circuit.

Eon fly-brain

Planned

FlyWire whole-brain LIF implementation with several simulation backends.

Used for: Performance comparison and separation of neural runtime from embodiment.
Will not copy: A full-graph engine where a smaller auditable pathway is sufficient.

neuprint-python

Planned

Official Python client for querying neuPrint connectome datasets.

Used for: Reproducible neuron and connection queries rather than hand-entered graph claims.
Will not copy: No interface concept; this is a data-access dependency candidate.

HouseFly

Boundary

Home Assistant integration connecting household sensor states to a virtual fly.

Used for: Duplication check and lessons about explicit sensor-to-neuron mappings.
Will not copy: Home Assistant entities, a persistent household fly, looming-object logic, or room preference.

FlyBrain

Boundary

Browser FlyWire simulation with food, touch, light, and temperature controls.

Used for: Duplication check for general-purpose sensory dashboards.
Will not copy: Generic room simulation, body controls, or its multisensory panel.

Evidence queue

Scientific and domain sources

“Queued” means the source has been identified but its exact claim has not yet passed the project’s extraction and citation check.

Origin article

The fly-connectome project article

Prompting context and overview of the current connectome-building wave; not a source for biological claims.

Reviewed

Dataset

MaleCNS

Official adult male central-nervous-system connectome and release history.

Primary

Dataset

FlyWire

Adult female brain connectome; a distinct dataset that must not be blended casually with MaleCNS.

Primary

Data access

neuPrint

Graph queries and provenance for named neurons and measured connectivity.

Primary

Odor response

DoOR 2.0 — Comprehensive Mapping of Drosophila melanogaster Odorant Responses

Consensus odorant-response resource for the Drosophila olfactory system.

Primary

Connectome paper

A connectome and analysis of the adult Drosophila central brain

Hemibrain connectome methods, named neuron types, and network-analysis context.

Primary

Dedicated pathway

A conserved dedicated olfactory circuit for detecting harmful microbes in Drosophila

Primary evidence for the dedicated geosmin-sensitive Or56a pathway used as the lab’s highest-confidence input.

Primary

Wine chemistry

Australian Wine Research Institute sensory resources

Wine sensory language and aroma context, kept separate from fly receptor claims.

Queued

Compound records

PubChem

Compound identifiers and chemical properties for aroma ingredients used by the interface.

Primary

Claim register

Candidate aroma mappings

This is a research queue, not a finished receptor table. Only high-confidence pathways will enter the first simulation.

Compound	Wine context	Fly pathway candidate	Current status
Geosmin	Earthy odor; often a fault or environmental taint context	Or56a → DA2	High-confidence candidate; primary-paper check pending
Ethyl acetate	Fruity at low levels; solvent-like at higher levels	Broad response; Or42b/DM1 among candidates	Needs concentration-aware evidence review
Isoamyl acetate	Banana/pear-drop ester	Or22a/ab3A → DM2 candidate	Needs direct response-source confirmation
Acetic acid	Volatile acidity	Ionotropic acid-sensing pathways, including Ir64a	Needs pathway and concentration disambiguation
Linalool	Floral/terpenic aroma	Potentially distributed receptor response	No single-pathway claim until reviewed
Ethanol	Matrix component and fermentation volatile	Multisensory/broad physiological effects	Will not be reduced to one receptor without evidence

Build history

Research log

Dated changes keep later conclusions traceable to the evidence available when they were made.

September 19, 2026

First interactive milestone implemented

Built a reduced deterministic circuit with six relative inputs, five glomerular channels, A/B comparison, blind profiles, single-pathway silencing, and a seeded rewiring control. Downstream weights and readouts are visibly labeled as project-defined. Automated desktop and mobile tests passed preset recalculation, intervention rendering, comparison distance, console-error, and overflow checks.

September 19, 2026

Ledger opened before implementation

Recorded scope, non-claims, initial code-review set, source queue, and tentative odor mappings before interactive code was written.