Shared input identities reduced estimated orientation prediction error by 2.23% after position controls in a fixed test of 3,461 neurons from one MICrONS mouse
We tested whether common input sources add orientation information beyond the positions of incoming synapses in one MICrONS mouse. We scanned all 337,312,429 records in the v1507 filtered synapse table. The fixed sample retained all 11,615,820 eligible input contacts on 3,461 selected neurons, including sources without a nucleus. Source identity and six spatial contact features used exactly the same contacts. Five spatial test folds and three inner folds selected kernel settings. An identity matrix controlled changes in the ridge penalty. After the position controls, shared input identities reduced squared error for model estimates of orientation from 0.895503 to 0.875513, a 2.23% decrease. The conditional 95% cube range was 1.63% to 2.82%. The gain was 1.90% in a second spatial split, 1.84% in V1, and 2.08% with each imaging session excluded. The adjusted source score exceeded all 199 source-label shuffles within 10 micrometre cubes. Checks at 5 and 20 micrometres gave the same rank direction. The raw orientation score was lower than every shuffle. Thus position adjustment changed the source-score comparison. These results support information beyond the tested geometry model. They do not show a causal effect or transfer to another animal. The outcome is a digital twin estimate, and source roots can be fragments. The package preserves code, exact feature counts, source provenance, all fits and reference scores, figures, audits and repeat instructions. A full main repeat reproduced all 19 numerical files exactly.
connectome · MICrONS · visual cortex · orientation · shared input · synapse position · spatial controls · prediction · data provenance
References
9- usesInput segments without a nucleus reduced orientation prediction error in fixed tests on 3,461 MICrONS neurons sci:sha256:294fd60eacb9f05af75fd1a19e307678b2e734a906199bc08d0b9803ba295fcb
We use its fixed neuron cohort and compare the complete selected input count matrix with its earlier extract. We add input positions and local source label controls.
- usesA penalty control reduced the shared input gain from 1.49% to 0.69% in a fixed test of estimated motion polarity in 1,721 MICrONS neurons sci:sha256:db5c2f1d8feb18400a9782ae8f9a980b600e332779f4c2132fb377b5e5266e1f
We use its identity matrix control to test whether an added source kernel only changes the penalty.
- Separate penalties reduced the added inhibitory input gain from 1.41% to 0.18% in a model-derived orientation test of 5,193 MICrONS neurons sci:sha256:66588b94767d58b8bc4d385a08080d9abd1f0d7d4943566a8a2f4ead93f1620d
This related orientation test showed that penalty choice can change the apparent gain from an added input feature.
- usesdoi:10.1038/s41586-025-08790-w
This paper describes the MICrONS data used for the full source scan and matched target cohort.
- usesdoi:10.1038/s41586-025-08840-3
This paper describes functional connectomics and estimated response properties in the source data.
- usesdoi:10.1038/s41586-025-08829-y
We use exported orientation estimates from the response model described in this paper.
- doi:10.1038/s41586-026-10457-z
This correction clarifies the response model architecture. We use fixed exported estimates and do not retrain that model.
- usesdoi:10.1038/s41586-024-07765-7
We use cell annotations supplied with the MICrONS tables.
- doi:10.1016/j.cub.2022.02.048
This experiment linked dendritic shape and spatial input patterns to visual tuning. It motivates the input position control.
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesWith fixed settings, cleft size weights reduced squared orientation vector error by 1.30% in 3,152 cells from one MICrONS mouse
The inherited geometry and source count controls and compact graph encoding originate here. We adapt its sparse normalization and kernel equations.
- usesRequiring valid axon review marks reduced prediction gains for estimated orientation from 2.23% to 0.18% in a fixed kernel test of 3,461 neurons from one MICrONS mouse
We reuse its complete input contact extract, position features, source records, and MIT code. We add a source partition by valid axon review marks.
- usesShared output targets gave no clear extra gain in a fixed ridge test of estimated orientation in 3,461 neurons from one MICrONS mouse
We reuse its exact checked incoming count and position features. We adapt its compact encoding, kernel formulas, nested prediction method, and resampling method. The new test adds all eligible output contacts and their positions.
- usesCleft size weights reduced model orientation prediction error by 1.36% after count and position controls in 3,461 selected neurons from one MICrONS mouse
We reuse its exact count and geometry features, fixed target records, kernel helpers and extraction code. We add cleft size profiles and local size assignment tests.
- With relay and partner controls, input orientation profiles gave no clear output prediction gain in a ridge test of 1,314 neurons from one MICrONS mouse
It tests shared input identity against a target neuron's own estimated orientation. Our test concerns input and output partner profiles at a relay.