How to use this simulation
Choose Mean & SDev to specify response statistics directly, or
Binomial n, p, q to generate each response from the number of release
sites (n), release probability (p), and quantal amplitude (q). The selected model applies
to both A and B.
In binomial mode, a response is q times the number of successful releases, plus optional
Gaussian background noise. Correct CV for background noise subtracts the
known noise variance before calculating CV; uncheck it to see how recording noise can
distort CV analysis. If the estimated release variance is zero or negative, the corrected
CV is undefined.
One response is generated every 10 seconds, with a 5-minute induction gap. Reduce the
response counts to see how limited sampling makes the observed result less reliable.
New seed reruns the experiment. Uncheck Reuse seed
to generate fresh responses whenever a parameter or response count changes.
The normalized coordinates are mean(after)/mean(before) and [CV(before)/CV(after)]².
Theoretical markers use the entered population parameters; observed markers use sample
statistics. A point's location does not uniquely identify a mechanism because the
interpretation depends on the assumptions of the release model.
In binomial mode, changing p alters both mean response and release variability, whereas
changing q scales the mean and SDev together and therefore leaves CV unchanged. This
interpretation assumes uniform p and q, independent release sites, at most one quantum
per site, and stable n. In real experiments, changing background noise, unstable
baselines, measurement errors, and outliers can bias the result. CV analysis estimates
the locus of plasticity and is best supported by independent measurements.
References