With odor means removed, parent glomerular signals reduced test error by 3.4% in a fixed linear test of 269 traced cells from two mice

We tested whether a traced parent glomerulus helps predict changes between repeated odor trials. We used 269 strict cell matches in 49 parent groups from two anesthetized mice. The target was the 0 to 2 second calcium mean, with the pre-odor level and each odor mean removed. A fixed ridge rule used other delivery modules for fitting and tested a complete excluded module. The baseline used nonsister activity in the cell plane and a global glomerular mean. Adding the parent signal reduced mean test error by 3.32% and 3.44% in the two mice. The equal-mouse mean was 3.38%. Other glomeruli from the same recorded plane sets increased error by 0.80% on average. The parent advantage was 4.17 percentage points of baseline error. The mean parent score exceeded all 9,999 source label exchanges within plane sets. A descriptive group resampling range was 1.46% to 5.16%. This range holds the two mice fixed. Planned odor-period checks kept positive gains. Removing module repeat offsets kept a mean gain of 2.84%. The pre-odor gain was positive in one mouse and negative in the other. The source glomerular signal pools projection-cell dendrites, including processes from the tested cells. Thus, this test links slow somatic and dendritic trial changes within traced groups. It does not identify common receptor input, causal direction, spike coupling, or a mechanism. Few repeats, two mice, shared state, and fluorescence effects limit the result. The package preserves source checks, compact inputs, code, all primary fits and predictions, reference draws, and figures. A fresh run reproduced all numerical results.

olfactory bulb · structure and function · glomerulus · calcium imaging · trial variability · anatomical tracing · prediction · reproducible analysis