Neural correletes of masked threshold

Abstract

The detectability of tonal and noise signals was measured in the goldfish using classical respiratory conditioning at various levels of a continuous broadband noise masker, and as a function of signal duration. At threshold for an 800-Hz tone, the signal-to-noise ratio (E/N0 = 22 dB and is generally independent of noise level and signal duration. The response of single eighth nerve fibers of the goldfish was measured under identical acoustic conditions and at the same N0 levels used in the psychophysical studies. For E/N0 below 22 dB, phase-locking to the signal occurs, but spike rate is not different from that evoked by noise alone. At E/N0 = 22 dB, the coefficient of synchronization tends to equal 0.5, and spike rate begins to rise above that evoked by noise alone. For E/N0 above 22 dB, synchronization begins to saturate and log spike rate grows linearly with signal level in dB. This function reaches a maximum slope of a doubling of driven rate per 3 dB increase in signal level under optimum conditions. Near E/N0 = 22 dB, the number of spikes evoked by a tone burst signal is linearly related to signal energy. While we are not able to determine whether masked signal detection is based upon synchronization or spike rate criteria, it is clear that behavioral detection is not based on synchronization values below 0.5.