In our preliminary theory of intensity resolution [e.g., see N. I. Durlach and L. D. Braida, J. Acoust. Soc. Am. 46, 372-383 (1969)], two modes of memory operation are postulated: the trace mode and the context-coding mode. In this paper, we present a revised model of the context-coding mode which describes explicitly a process by which sensations are coded relative to the context and which predicts a resolution edge effect [L. D. Braida and N. I. Durlach, J. Acoust. Soc. Am. 51, 483-502 (1972); J. E. Berliner, L. D. Braida, and N. I. Durlach, J. Acoust. Soc. Am. 61, 1256-1267 (1977)]. The sensation arising from a given stimulus presentation is coded by determining its distance from internal references or perceptual anchors. The noise in this process, combined with the sensation noise, constitutes the limitation on resolution in the model. In the revised model the probability density functions of the decision variable are not precisely Gaussian (and cannot be expressed analytically in closed form). This paper outlines the predictions of the model for one-interval paradigms and for fixed-level two-interval paradigms and derives estimates of the values of model parameters.

This paper reports the results of some preliminary two-interval roving-level discrimination experiments designed to test certain predictions of the oreliminary theory of intensity resolution presented in the first paper of this series [N. I. Durlach and L. D. Braida, J. Acoust. Soc. Am. 46, 372-383 (1969)]. Discrimination performance was measured for roving-level ranges in the interval 0 to 54 dB and intersignal durations in the interval 0 to 9 sec. Although the results, when averaged over levels, are roughly consistent with the theory, there is a substantial dependence of sensitivity on level that is inconsistent with the theory. Much of this dependence appears to be similar to that found previously in one-interval experiments [L. D. Braida and N. I. Durlach, J. Acoust. Soc. Am. 51,483-502 (1972)], and to be interpretable in terms of deviations from Weber's law and the edge effect (caused by the use of the extreme intensities in the stimulus set as perceptual anchors).

This paper reports the results of further experiments on two-interval roving-level intensity discrimination and on the dependence of resolution and bias on level within the range of roving levels (the resolution and bias edge effects). The experimental parameters considered are the intensity range R, the intersignal duration T, and the pulse duration Tp. In order in obtain a picture of the natural bias, no feedback was presented. The results on the dependence of resolution on level are compared to previous results of the same type obtained from one-interval identification experiments. Discrepancies between the experimental results and the preliminary theory of intensity resolution [ Durlach and Braida, J. Acoust. Soc. Am. 46, 372–383, (1969)] are identified and implications of the data for an improved theory aree discussed.

No abstract

This note reports some measurements of the changes in identification performance that result form presenting a fixed standard before each trial. These measurements indicate that the presence of the standard improves performance for test stimuli in the vicinity of the standard, except when the standard is near the extremes of the stimulus range.