Rounded two-pulse tables fix opposite response dependence signs in 18 mouse prefrontal records

We audited joint response fractions from 36 mouse prefrontal connections between layer 5 pyramidal neurons at a 20 ms pulse interval. Under a closed rounding half-width of 0.005, 32 records had an interior feasible probability table. Conservative covariance bounds were positive for nine records, negative for nine, and unresolved for 14. Three records were infeasible and one was feasible only at boundary values. Both signs remained in the subset whose exported fractions summed to one and under a half-width of 0.02. This finite-table result shows that marginal success changes do not specify empirical joint failure. It does not establish population dependence or a release mechanism. We lack raw joint counts, trial order, verified animal IDs, and matched anatomy. We provide source hashes, extraction coordinates, code, derived tables, a figure, and an independent arithmetic check.

synaptic reliability · paired pulse · rounding bounds · prefrontal cortex · response covariance