Tracer connection strength predicts spontaneous inter-areal spike correlations across mouse cortex beyond distance (24 mice, 30 areas), but not for visual to non-visual area pairs
Question: Does the direct axonal connection strength between two mouse cortical areas in the Allen tracer connectome predict how correlated their spiking is, after the effects of distance and area identity are removed? Approach: We used Allen SHIELD Neuropixels recordings (DANDI:001051; 76 sessions, 24 mice, 30 Knox summary areas, 2,111 session x area-pair samples, 317 different pairs). Functional coupling (FC) is the mean unit-pair spike-count correlation between two areas. The primary measure uses 100-ms bins in the grey-screen block, with the global mode of all other units removed. Structural coupling (SC) is log10(W_ab + W_ba) from the Knox 2019 ipsilateral connectome after QC (Nathan et al. 2026). The models include session and area fixed effects and log inter-unit distance. We used mouse-cluster bootstraps, area-label permutation, per-mouse tests and leave-one-mouse-out prediction. Results: The FC change per SD of SC is 0.23 SD (95% CI 0.17-0.29; permutation p = 0.001). With a flexible distance control, it is 0.28. The per-mouse partial Spearman r is 0.16 and is positive in 20 of 24 mice. The effect occurs at 20-ms, 100-ms and 1-s bins, in running and still epochs, in the task pre-change window, without global-mode removal, and for cross-probe unit pairs only. The effect is small: SC increases the held-out R2 from 0.135 to 0.154. The effect occurs among frontal, somatomotor and retrosplenial areas (0.21) but is absent for visual x non-visual pairs (0.02, CI -0.12 to 0.13). Shared-input and shared-output similarity alone do not improve prediction. Pre-QC and post-QC connectomes are almost identical for these pairs, so this dataset cannot test the value of QC. Limitations: The absolute correlations are small. The connectome is a spatially smoothed model from different mice. Visual x visual pairs are too few to test. The still state has only 9 mice. Only the left isocortex was sampled, and the result is correlational.
structure-function coupling · mesoscale connectome · Neuropixels · spike-count correlation · functional connectivity · mouse cortex · Allen SHIELD · Knox connectome · tracer connectivity · spontaneous activity
References
8- usesdoi:10.48324/dandi.001051/0.260922.1827
Source of the SHIELD Neuropixels spike data (CC BY 4.0), used through the shared cache allen_shield_spontaneous_cortex_spikes/v1_2026-11-01.
- doi:10.1016/j.neuron.2024.06.015
SHIELD implant and recording method paper.
- usesdoi:10.5281/zenodo.20276701
Knox 291-region connectomes before and after experimental QC (Nathan et al. 2026).
- doi:10.64898/2026.02.20.707091
Preprint that describes the connectome QC.
- doi:10.1162/netn_a_00066
Knox et al. 2019 regionalized voxel model of the mouse connectome.
- doi:10.1038/nature13186
Oh et al. 2014 Allen Mouse Brain Connectivity Atlas and normalized connection strength.
- doi:10.1016/j.cell.2020.04.007
Allen CCFv3 (Wang et al. 2020) used for area labels.
- doi:10.1073/pnas.0811168106
Honey et al. 2009: structure predicts resting-state functional connectivity in human fMRI; the comparison case for our spiking result.
Cited by
9 publicationsPublic catalog entries on Sciscape.
- Multi-step connectome communication measures add 0.4 to 1.9 percent of explained FC variance beyond direct wiring, distance and node effects in worm, fly, mouse, marmoset and human
Related mouse tracer vs spike-correlation coupling study.
- In the same mice, the Allen tracer connectome predicts cortical resting fMRI beyond distance under medetomidine with isoflurane, but much less under isoflurane alone or awake
Campaign study that links tracer connection strength to inter-areal spike correlations in mouse cortex; our result shows that the size of such links in fMRI depends on brain state.
- usesAnatomical hierarchy did not reliably predict spontaneous spiking timescales across mouse cortical areas in Allen SHIELD and IBL Neuropixels data; laminar sampling explained the SHIELD association
Campaign study that made the SHIELD spontaneous spike cache used here; it linked the Knox connectome to inter-areal spike correlations.
- Tracer-based structure-function coupling does not decrease along the anatomical hierarchy in marmoset and mouse cortex; distance explains the human tractography coupling gradient
Related campaign result on tracer strength and inter-areal spike correlations.
- In 40 freely moving C. elegans, gap junctions predict pairwise co-activity after controls for neurite contact, soma distance and per-neuron effects; chemical synapse weight does not
Related mouse result: tracer connection strength predicts spontaneous inter-areal correlations.
- usesSingle-neuron axon co-innervation predicted spike correlations between 30 mouse cortical areas, but real collateral patterns did not predict better than source-matched shuffles at 100 ms
Source of the SHIELD inter-areal FC table, tracer SC and distance predictors, and of the FC model that we extend.
- In one V1 Deep Dive mouse, population coupling of a neuron predicts its local EM inputs, but pairwise response similarity gives no pair-specific gain after cell-level control
Related campaign result on structure-function links at the inter-areal scale (tracer strength vs spontaneous spike correlations).
- In larval zebrafish, a single-neuron projectome predicts region-level trial-residual coupling beyond distance and region effects, but not stimulus-locked coupling (25 regions, 11 fish)
Related campaign result: mouse tracer strength predicts spontaneous inter-areal spike correlations beyond distance.
- FlyWire wiring predicts neuropil calcium correlations beyond distance in 50 flies; walking and rest coupling agree after global-signal removal; transmitter-matched wiring adds no gain
Related campaign result: tracer connectivity predicts spike correlations beyond distance in mouse cortex.