In MICrONS mouse V1 layer 2/3, inhibitory connections spread 85–113 µm, not the 40 µm a published holographic-stimulation model used; the stimulation data fit wide inhibition with narrow excitation

Question: Oldenburg et al. (2024) explained holographic stimulation of 10-cell ensembles in mouse V1 layer 2/3 (activation <30 µm, suppression at 50–150 µm) with a linear neural-field model. Rabies-derived connection widths (excitatory 147 µm, inhibitory 111 µm) did not fit. The fit needed 70 µm and 40 µm and a narrow excitatory term (kappa 0.045). Does synapse-level connectivity support these widths? Approach: In the MICrONS minnie65 EM volume (v1507, one mouse) we measured connection probability and summed synapse size against lateral soma distance for 520 proofread L2/3 pyramidal axons and 80 L2/3 inhibitory axons. We put the EM kernels into the published model and fitted two strengths to the released per-cell data of 160 ensembles from 12 mouse identifiers. We scored the fits by leave-one-mouse-out error. Results: Inhibitory to pyramidal Gaussian width was 107 µm (95% CI 101–114) by probability and 85 µm (80–92) by synapse size; basket cells 109 µm. Excitatory to excitatory width was 104 µm (101–108; complete axons 131 µm), with a short-range excess (kappa 0.030–0.039 at 20 µm). Slice paired recordings gave PV to pyramidal 94 µm (70–132) as a lower bound. In the model, the data preferred narrow monosynaptic excitation (5–20 µm) with any disynaptic inhibitory width of 95–240 µm. This range includes the EM value (96–160 µm), but not the 80 µm of the published widths. EM-only kernels predicted held-out mice 1.8% (CI −1.4 to 3.4) to 9.2% (3.1 to 13.2) worse than the published fit. They under-predicted near activation. Conclusion: the surround suppression agrees with EM inhibitory widths, and the 40 µm inhibitory width is neither needed nor supported. The unexplained part is the narrow excitation. Wide EM excitation explains it only if wide excitation and disynaptic inhibition nearly cancel. Limitations: EM and function come from different mice. Target positions were simulated from recorded ensemble spread. Proofread axons are not a random sample. The model is the published truncated linear model without orientation rules.

connectomics · MICrONS · mouse visual cortex · holographic optogenetics · surround suppression · lateral inhibition · connection probability · neural field model · inhibitory interneurons · basket cells · layer 2/3 · structure-function · electron microscopy