Total PSD volume reduced squared log error for bouton volume by 24% beyond spine count in five tests on excluded mice

Does postsynaptic size help predict presynaptic bouton volume beyond spine count? We tested CA3 to CA1 electron microscopy records from five mice. Each prediction test excluded one complete mouse from fitting. Both models used count, condition and axon activity label. Adding total postsynaptic density (PSD) volume reduced mean squared log2 volume error from 0.9865 to 0.7490, or 24.1%, with equal mouse weights. All five mice improved. The test used 2,990 complete boutons from 344 axons. Scoring only 693 boutons with several contacts gave a 24.8% decrease. Equal axon weights, a quadratic count term, and restriction to one or two contacts gave similar gains. A separate within-axon count model gave a volume slope of 0.920 in 3,100 boutons. The equal-mouse mean slope was 0.915, with a five-mouse Student 95% interval of 0.788 to 1.043. These data do not establish sublinear scaling across mice. The result supports postsynaptic size as an extra constraint for structural resource models. It does not measure transmitter release or a causal learning effect. Five selected mice, missing records and one source study limit transfer. The package preserves code, source hashes, record selection, coefficients, aggregate results, figures and repeat instructions. Source measurements remain external.

connectome · hippocampus · multisynaptic bouton · postsynaptic density · structural prediction · mouse exclusion