With full initial pools and 1 ms steps, changing release order raised fitted recovery from 1.79 to 5.04 s in six fly olfactory recording pairs

We tested whether release event order changes the dynamics assigned to a mapped fly olfactory circuit. We used six DM1 to PV5 recording pairs from the fixed Kim et al. dataset. The released model starts facilitation at zero and uses its value before the spike update for release. This rule gives no release at the first isolated spike. We changed only the release term to use the updated value. Both models kept full initial resource pools, the same passive values, 1 ms steps and physical parameter bounds. We fitted 58 response positions with 12 fixed starts. With each test pair excluded from fitting, mean squared voltage error rose from 0.3453 to 0.3562 mV², an increase of 3.18%. Three of six pairs improved. In the fit to all pairs, recovery time rose from 1.791 to 5.040 s. Source recovery times were 1.44–2.07 s with one pair excluded; changed-rule times were 3.63–6.44 s. A 0.1 ms step kept the error increase and parameter shift. The primary predictions and errors repeated exactly. An independent equation calculation and all saved voltage checks passed. These results show that a release convention can change inferred synaptic memory without better test prediction. Six selected pairs, overlapping voltage windows, omitted late spikes and fixed initial states limit the biological claim. The package preserves code, voltage windows, spike times, provenance, fitted states, predictions, figures and repeat instructions.

connectome · structure and function · synaptic facilitation · short-term plasticity · event order · Drosophila · olfactory circuit · model audit · reproducibility