Detection of an Octave Band of Noise as a Function of Stimulus Presentation.

Two experimental methods were developed to examine the effects of time uncertainty and other variables on the detection of a 500 Hz centered octave band of noise presented in noise. In one method, denoted the fixed presentation method, the signal was presented at a fixed SNR, following a specified interval of time after the onset of an ambient noise stimulus. The other method, the modified threshold-forced response method, presented the signal with the noise at the start of the trial at a low SNR. During the trial, the SNR increased at a constant rate of one-half dB per two seconds. Upon attaining a specified SNR, the stimulus was gated. Six levels of ending SNR were examined with each method. Four subjects responded on a six-point confidence rating scale. Results indicated that with the variable SNR method, performance was much worse than in the fixed SNR method. Not only were confidences lower, but the probability correct was likewise lower. Results also indicated that subjects could maintain a fairly consistent set of criteria throughout the experiment, as rank ordered correlations of responses to identical tapes were generally high. Consistency was found to increase with SNR.