In 12 mice, preferred stimulus activity gave no clear gain in ridge predictions of spine survival after anatomy and mean trial activity

We tested whether activity at a cell's preferred visual stimulus predicts later spine survival after anatomy and mean trial activity. We used 929 spine records from 20 cells in 12 mice in the released Tsimring data. Each outer test excluded a whole mouse. Inner tests selected the ridge logistic penalty. Adding preferred activity raised mean mouse log loss from 0.630678 to 0.632048, a 0.22% increase. In 399 whole-mouse resamples with all fits repeated, the central 95% range for the loss decrease was -1.34% to +2.37%. This gives no clear added gain in the combined panel. The later interval gave a 4.00% decrease, with wide uncertainty across seven mice. Mean and preferred activity were correlated at 0.740. Thus, a larger preferred response in retained spines did not establish added prediction in this test. The result does not show that stimulus alignment or Hebbian plasticity has no role in survival. It concerns optical spine records, two imaging intervals, and additive logistic models. The package preserves a checked compact dataset, source code, provenance, all predictions, resamples, figures, and repeat instructions.

dendritic spines · structural plasticity · structure and function · spine survival · visual cortex · calcium imaging · mouse exclusion · reproducible prediction