At a 16:9 field scale, two coordinate rules gave no clear added visual axis alignment term for connections among 3,437 neurons in one MICrONS mouse

We tested whether recorded excitatory links align with a target neuron's preferred visual bar axis in one MICrONS mouse. The main panel had 3,437 neurons, 1,905,276 candidate directions and 20,421 recorded links. A fixed linear model controlled source identity, target and local source groups, and geometry. At a planar field scale of 16:9, the adjusted along-axis minus across-axis contrast was -0.085 percentage points under a top-left origin. It was -0.071 points under a bottom-left origin. The two planned tests had Holm-adjusted normal-reference p values of 0.064 and 0.075. An added leverage correction gave pointwise 95% ranges of -0.209 to +0.039 and -0.197 to +0.055 points. All eleven planned scope checks gave negative point estimates. These results give no clear added alignment term in this model. They do not establish zero effect or the absence of contour integration. Official data sources give opposite vertical coordinate rules. All 8,284 matched field centers agree with the current v2 export, so we tested both rules. A new scan of all 303,164,967 original connection rows recovered the complete reused matched graph. The package preserves that graph, the candidate panel, code, source hashes, all results, figures and repeat steps. Limits include one mouse, model field centers, assumed planar geometry, incomplete axons and approximate conditional uncertainty.

MICrONS · connectome · visual cortex · orientation · contour integration · field center · coordinate convention · neuron fixed effects · dyadic covariance · structure and function