Stimulus-Oriented Approach to Detection Re-Examined
Lloyd A. Jeffress · The Journal of the Acoustical Society of America · 1967
The present paper is a reexamination of some of the conclusions of an earlier one. It is motivated by some new insights resulting from attempts to replicate experiments with human observers, through the use of an electrical model of the auditory system. It is concerned primarily with the effect of signal duration on detection in the presence of a continuous masking noise. The model, of those tried, that best fits human performance consisted of a bandpass filter obtained by subtracting the output of a 500-Hz sharp-cutoff, low-pass filter from another having a cutoff of 525 Hz. The filter was followed by a linear half-wave rectifier and it, in turn, by an integrator having a 100-msec decay time. The integrator can be thought of as a device that takes a running average of its input. The probability density distributions for N and SN yielded by the model lie between the Rayleigh Rice distributions on the one hand and a pair of normal distributions of unequal variance on the other. The exact shape of the two distributions depends upon both the bandwidth of the filter employed and the time constant of the averager.