Experiments · E7

Is the fly's own learning circuit present, complete, in the imported brain?

Yes. Every pathway is there, with 61,210 connections at the learning site, and smell receptors correctly never connect to it directly.

In the log: The measured mushroom-body pathway

recordedDate 2026-09-12, as written in the logunclassified0 predictions · 0 result paragraphsEXPERIMENTS.md lines 208–230

Checked · 2 discrepancies

Files checked (2)

data/malecns_mushroom.json
data/malecns_mushroom.npz

Numbers confirmed (5)

claim: KC->MBON 61,210 edges, 463,640 contacts; source: mushroom asset, KC/MBON blocks; ok: true
claim: PN->KC 22,598 / 390,947; APL->KC 4,633 / 196,200, APL GABAergic; source: mushroom asset; ok: true
claim: PAM->KC/MBON 110,450/2,693 (173,789/28,546); PPL 18,663/467 (51,338/11,165); source: mushroom asset; ok: true
claim: MBON->MBON 56% and MBON->PAM 66% inhibitory; MBON->PAM 2,119 / 6,585; source: mushroom asset, consensus transmitter of the presynaptic cell; ok: true
claim: 17.3 contacts per PN->KC edge against 1.8 per KC->KC edge; source: 390,947/22,598 and 1,153,844/642,932; ok: true

Discrepancies

KC->KC (642,932 / 1,153,844) and MBON->MBON (1,605 / 26,258) are each one edge lower on disk than recorded; the current asset removes self-edges.
ORN rows (ORN->PN, ORN->KC = 0) cannot be checked here: the mushroom asset holds no receptor neurons.

Status the catalogued status fits the text

The pathway table recomputes from the stored mushroom-body asset to the edge, apart from two self-edges the newer extraction drops.

From runs/site_data/exp_checks/E7.json; the log itself is not edited.

Results

No result paragraph for this entry was found in the log.

The full record

EXPERIMENTS.md · lines 208–230

E7 — The measured mushroom-body pathway

Date 2026-09-12 · Question Is the fly's own learning circuit present in the extraction, and at what strength? · Provenance full-asset extraction

pathway	edges	contacts	note
ORN -> PN	33,152	500,252	2,635 receptor neurons onto 808 projection neurons
PN -> KC	22,598	390,947	~5.6 PN inputs per Kenyon cell
KC -> MBON	61,210	463,640	the only plastic site
KC -> KC	642,933	1,153,845
APL -> KC	4,633	196,200	100% inhibitory
PAM -> KC / MBON	110,450 / 2,693	173,789 / 28,546	appetitive teaching
PPL -> KC / MBON	18,663 / 467	51,338 / 11,165	aversive teaching
MBON -> MBON	1,606	26,259	56% inhibitory
MBON -> PAM	2,119	6,585	66% inhibitory
ORN -> KC	0	0	receptors never contact KCs directly

Interpretation. The complete circuit is present. ORN -> KC = 0 is correct fly anatomy reproduced from raw tables without being asked for — an unaided check on the extraction. Contact counts say PN->KC synapses carry 17.3 contacts per edge against 1.8 for KC->KC, so the measured anatomy makes the odour pathway an order of magnitude stronger than the recurrent background.