Detection of Auditory Signals in Restricted Sets of Reproducible Noise

Abstract

The detectability of auditory signals in reproducible random noise was studied under two conditions: a single noise used throughout a block of 288 trials, and 12 noises occurring at random but with equal frequency throughout a block of trials. Both 2-interval forced-choice judgments and judgments of the presence or absence of the signal in single-noise samples were obtained in separate blocks of trials. On individual trials of the forced-choice judgments, the same noise appeared in both intervals. The noises were stored numerically in a PB-250 computer and converted to sound during the experiment by means of a digital-to-analog converter. The signal frequency was 312.5 cps. Signal detectability was found to be significantly better when a single noise was present in a block of trials. Performance was poorer for presence-absence judgments than for forced-choice judgments. The results support the importance of memory from trial to trial in the detection process.