A fixed ridge model gave correlations near -0.15 between input and motion similarity for 11 wing muscles in three fly maps

We tested whether shared synaptic input marks muscles with similar associations to wing motion. We used all recorded direct input to 22 motor neurons in MANC v1.2.1, Male CNS v1.0 and BANC v888. A fresh scan of 170,782,903 original connection rows recovered every selected edge. The functional panel had 72,219 wingbeats from 372 movies and 31 eligible dates. A fixed ridge model fitted 12 muscle calcium signals and wingbeat frequency to four left-wing motion angles. We compared input cosine with the cosine of the 11 matched muscle coefficient profiles. Correlations were -0.1495, -0.1562 and -0.1496. Full muscle-label upper fractions were 0.86 to 0.88. Exact attachment-group fractions were 0.60 to 0.71. The model reduced angular error by 7.28% against frequency alone with each date withheld. Some muscle profiles had low repeatability across disjoint date sets. Marginal association profiles instead gave positive correlations of 0.15 to 0.21. Thus the functional definition changes the result. The evidence gives no positive match for the fixed conditional profiles. It does not establish a population anticorrelation or a lack of shared motor control. Calcium kinetics, selected movies, linear mechanics, unknown animal grouping and cross-animal anatomy limit the result. The package preserves exact movie moments, input edges, source hashes, all fits and draws, figures, code and repeat instructions.

connectome · Drosophila · wing steering · motor neurons · muscle calcium · wing kinematics · ridge regression · sufficient statistics · structure and function