A peripheral model of gap detection

Gap detection is an important measure of the temporal resolution of the auditory system. We present a model of gap detection involving the statistics of peripheral auditory neural activity: detecting a gap involves the detection of an interval that is different from the usual interspike interval found in the auditory spike train. The model is capable of resolving a number of important observations in empirical gap research including why across-channel (or across-frequency) gap detection is an order of magnitude larger than gap detection performed in channel (i.e., using the same frequency). We achieve this by recognizing that in-channel detection involves a different strategy than across-channel detection. The model also provides an explanation for why the intensity dependence of gap detection is limited to a 40 dB range, and why critical bands are crucial towards defining the notion of a channel in gap detection. Specific predictions can be made about the shapes of the psychometric functions.