Cut-back frozen noise characterization of auditory sensory memory

Abstract

Frozen noise stimuli were comprised of repeated concatenated segments of random noise, with the segment duration determining the periodicity or stimulus onset asynchrony (SOA). Seven conditions were constructed with SOAs ranging from 100–1000 ms. For each of the seven conditions, a fraction of the repeated segments were ‘‘cut-back’’ with nonrepeated noise. Subjects were presented with either a frozen noise stimulus or a nonrepeated noise stimulus of equal duration, and told to indicate if they detected periodicity. The thresholds for this two-alternative forced-choice infrapitch periodicity detection task were obtained by an adaptive tracking procedure. The results show that classification performance varies linearly with SOA over the entire 100–1000 ms. range. When normalized by SOA, it becomes evident that detection is characterized by a signal-to-noise ratio (SNR) where the repeated noise is the signal and the nonrepeated noise, the noise. The SNR for all seven conditions was approximately 0.7. It is proposed that cut-back frozen noise stimuli may be used to establish the upper storage limit of auditory sensory memory by extending this technique to SOAs greater than 1000 ms and looking for deviation from linearity in the plot of repeated noise segment vs SOA. [Work supported by NIDCD Grant No. DC 0463801.]