Experiments with Gap Detection

Test procedures and signals used by different research groups are described. Results from a comparison of two procedures, three-alternative forced-choice and BMsy, are reported. The characteristics of the test sipal and the masker are discussed. INTRODUCTION Gap Detection is a method for measuring the time resolution of the ear. A noise segment with a gap of a certain length is presented to the listener. The task of the subject is to detect the gap. If the he/she can do this, noise with a shorter gap is presented but if he/she cannot detect the gap, it is lengthened. A Gap Detection threshold is then determined with an up-down method (Levitt, 1971). Gap Detection has been used extensively both in research and clinically. However, there has been no consistency in the choice of stimuli and procedure. There are several different test procedures. The three principal ones are: two-alternative forced-choice, 2AFC (Glasberg & Moore, 1989; Shailer & Moore, 1983), threealternative forced-choice, 3AFC (Tyler, Summerfield, Wood, & Fernandes, 1982) and adjustment of the signal-to-noise ratio to a determined gap length by use of a B6k6sy technique (Fitzgibbons, 1983; 1984; Fitzgibbons & Gordon-Salant, 1987). The differences are greater in the use of stimuli. Some researchers use broad-band noise while those who want to make a frequency-specific measurement filter the noise in some way. Mostly some kind of band-limited noise has been used but highpass and low-pass filtered noise have also been applied. There have been many different centre frequencies and bandwidths. The filter slopes are almost always very steep (96 dB/octave). With this kind of filter it has been necessary to add a masker. The masker is most often notch-filtered with the same filter characteristics as for the signal and presented at 20 dB below it. METHOD Procedure The two procedures we wanted to compare were 3AFC and a B6kesy technique where the length of the gap was adjusted according to the listener's response. The test using 3AFC was a program under DOS and had been used a lot (more than 200 patients). The other test was a completely new Windows program. In the 3AFC test, three noise bursts separated by one second are presented to the listener. One of them contains a gap. The task of the listener is to decide in which segment the gap is located. In the B6k6sy test, a continuous noise is presented and at regular intervals there is a gap. The task is to detect the gap and answer by holding down or releasing a button. The length of the gap is changed in steps of 1 ms.