Asymmetric hidden relays stay within 5% transfer error in 31 passive Allen-based models with 1 pF hidden capacitance

We tested whether a 1 pF hidden capacitance limit excludes indirect electrical routes in passive reference models based on Allen mouse V1 slice pairs. A relay with unequal edges has the same DC port matrix as each direct-edge model. With recorded capacitors fixed at 100 pF, a continuous-band bound certifies less than 5% complex cross-transfer error on 0-100 Hz in all 31 steady-state references and all 31 peak-resistance checks. Primary grid errors range from 1.021% to 3.726%. A prior symmetric construction exceeds 5% in all 31 primary models at the same capacitance. The new construction needs a larger edge of 17.14-155.16 nS. This is a conditional circuit result, not measured transient agreement or evidence for hidden neurons. It shows that capacitance assumptions must be tested together with conductance and leak constraints. Separate recordings, reciprocal adjustment, assumed capacitances, strong edges and lumped passive nodes limit biological interpretation.

Allen Institute · electrical coupling · hidden relay · capacitance · structure inference · passive circuits