Measuring ongoing delay with reverse noise

Abstract

Counts of spike coincidences provide a powerful means to compare responses to different stimuli or of different neurons, particularly regarding temporal factors. A drawback is that these methods do not provide an absolute measure of latency, i.e., the temporal interval between stimulus and response. It is desirable to obtain such a measurement within the same analysis framework of coincidence counting. Single neuron responses were obtained from several fiber tracts (nerve, trapezoid body, lateral lemniscus) in two species to a broadband noise and its polarity-inverted version. The spike trains in response to these stimuli are the “forward noise” responses. The same stimuli were also played time-reversed. The resulting spike trains are then again time-reversed: these are the “reverse noise” responses. The forward and reverse responses were then analyzed with the coincidence count methods we have introduced earlier. Latency measurements derived from correlograms between forward and reverse noise responses show maxima at twice the latency value measured with other methods, as expected. Correlograms were often asymmetric but, at low-characteristic frequencies, were well-predicted by crosscorrelation of the reverse-correlation function and its reverse. We conclude that reverse noise provides an easy and reliable means to estimate latency of auditory nerve and brainstem neurons.