With full initial pools and linear readout, 197 of 608 initially depressing mouse V1 triplets at 30 Hz failed a fixed release and exponential recovery test at an error factor of 1.10

Connectome models often add resource depletion to fixed connection weights. We tested a model with a fixed release probability, exponential pool recovery, full initial pools, equal pulse intervals, and positive linear voltage readout. For any number of independent pools, the three mean responses must satisfy A2 squared / A1 <= A3 <= A2 <= A1. We give an exact error test and code that requires no parameter fit. We applied it to stored mouse V1 voltages from the Chou optomapping data at 30 Hz. Of 991 links in an earlier checked cohort, 608 had a positive third response and an initial decrease. Among these, 367 fell outside the exact bound. For 197 of 608 links (32.4%), no fit was possible when each amplitude could change by a factor of 1.10. The equal-date mean was 37.4%, with a 29.8-45.1% date resample range. For triplets with all amplitudes at least 0.1 mV, 99 of 454 (21.8%) still failed this test. More contacts or varied fixed pool parameters cannot remove a failure under the stated model. The result concerns processed mean voltages. Trial errors, pulse-specific spike times, and the causes of the failures remain unknown.

connectome · short-term synaptic plasticity · vesicle depletion · mouse visual cortex · model feasibility · optomapping · three-pulse responses · synaptic dynamics