As opposed to action potential thresholds (which are binary yes/no events), electrophysiological thresholds of compound action potentials are arbitrary. Mostly a certain noise level is picked and when the neural signal crosses this 'threshold' it is said to be 'above-threshold'. I have come across various definitions of this threshold, such as SD-based values (2xSD, 4xSD etc), 'visual' thresholds (a bit iffy from my point of view) among other things. Is there a rule of thumb as to define threshold? Alternatively, a review article, book chapter, or a primary research article with considerable impact explicitly dealing with this matter would be very helpful.
Action potentials of single neurons (APs) are the building blocks of compound action potentials (CAPs). While each of the individual APs is a yes-or-no response with a clear-cut defined threshold, CAPs are not. CAPs are build up of hundreds or thousands of neuronal contributions in, e.g., the auditory nerve. Decreased responsiveness as well as desynchronization of the neuronal responses can decrease the amplitude of the CAP.
It all depends on the amount of background noise and stimulus artifacts versus the amplification factor (the gain) whether a CAP can be recorded and identified yes or no. Typically, gross-potential recordings do not allow for single APs to be measured. CAPs are therefore stochastic phenomena and hence the CAP threshold should either be defined by visual examination (subjective), or through noise-dependent threshold criteria (objective). Therefore, unlike AP thresholds, CAP thresholds are arbitrary and heavily dependent on the noise level.