PSD weights raised uniform-loss reversal risk in 155 pooled ferret spine records under a fixed projection model

We computed probability bounds for loss of a positive orientation component in 155 public ferret spine records with matched structure and visual tuning. Every subset of a fixed loss count was assumed equally likely. An integer-grid subset-count method gave deterministic lower and upper bounds. At 36 losses, PSD area weights gave a reversal probability from 1.407e-8 to 2.066e-8. Equal weights gave 1.689e-25 to 2.039e-23. Starting projections were similar, at 0.215 and 0.222. PSD weighting therefore raised model risk by at least 6.9e14. Positive lower bounds also resolved rare events left unseen in an earlier Monte Carlo study. Exhaustive small cases and independent resolution checks supported the calculation. These are results for a fixed pooled table and an assumed loss process. Missing cell labels, unknown efficacy, preference errors, and absent circuit dynamics prevent neuron-level risk or transfer claims.

structure and function · dendritic spines · uniform loss · rare events · probability bounds · ferret V1