The results of experiment 1 suggested that, at least with the conditions described, it is likely that CDD is mainly due to within-channel masking. Narrow band noise maskers that are correlated with the signal appear to be more effective maskers than when they are uncorrelated. It is still possible that an across-channel grouping effect plays a role in raising the correlated threshold. The maskers and signal used in the experiments described so far have all had a simultaneous onset. It has been reported that simultaneous onset has at least as large an effect as correlated amplitude changes in the perceptual fusion of sounds (Rasch, 1978; Darwin, 1984). Also, asynchronous onset promotes segregation (Darwin and Carlyon, 1995; Bregman, 1978). Introducing a sufficiently large onset asynchrony between the signal and masker bands should therefore provide a good test of whether perceptual fusion plays a role in the CDD experiments described here. Similar work was performed by McFadden and Wright (1990) on CDD and McFadden and Wright (1992) on CMR.
The task in experiment 2 was to detect a 20-Hz wide band of noise centred at 1500 Hz. The signal was presented with a masker consisting of either 2 or 6 20-Hz wide bands of noise. The masker bands were always correlated with each other. The signal band could be either correlated or uncorrelated with the maskers. If 2 bands were presented, these were centred at 900 Hz and 2100 Hz (similar to conditions 5 and 6 in experiment 1, but with the lower band moved from 1071 Hz to 900 Hz). A low-pass noise was added to mask possible combination bands. The low-pass noise was produced by filtering a white noise (Hewlett Packard 3722A) using four 48 dB/Octave low pass filters (Kemo VBF/8) in series. The -3 dB point of each filter was set to 400 Hz. The level of the low-pass noise was then adjusted so that it had a spectrum level of 38 dB SPL at 300 Hz (2x900-1500) (i.e. the frequency at which a cubic-difference band between the lower band and signal band may be present). The noise was presented continuously.
If 6 bands were presented, these were centred at 547, 765, 1071, 2100, 2940 and 4116 Hz (These conditions corresponded to conditions 1 and 2 in experiment 1). No low pass noise was needed with 6 bands as the masker bands would effectively mask any combination bands.
The onset asynchrony between the masker and signal was either 0 or 200 ms (the masker being presented first). The signal and masker offsets were simultaneous. The signal always had a steady state duration of 300 ms with 50ms raised cosine ramps. The masker was therefore presented for either 300 or 500 ms per trial. The inter-stimulus interval was 200 ms. All combinations of the number of bands, modulation correlation and onset asynchrony were tested independently. The level of the masker bands was 65 dB SPL.
The stimuli were derived from those used in experiment 1. The signal was presented at random in one of three observation intervals which were marked by lights on the response box. The intervals were separated by 200 ms. Signal timing was controlled by a Texas Instruments 990/4 computer. Each gate consisted of two analogue multipliers (AD534L) connected in series. The gating voltages were derived from 12-bit digital-to-analogue converters. The signal level was controlled by a Charybdis model D programmable attenuator.
The same procedure was used as in experiment 1.
Three subjects participated, all with absolute thresholds less than 10 dB HL at all audiometric frequencies. Subjects SB and GEM had extensive practice, having previously completed experiment 1. Subject DM had a number of hours of practice on the stimuli used in experiment 1 until his results stabilised.
The results for all three subjects are shown in figure 3.1. Table 3.i shows the average thresholds across the subjects.
Figure 3.1. The effects of introducing an onset asynchrony between signal and maskers. The error bars show the standard deviation of the thresholds.
Table 3.i. The effects of introducing an onset asynchrony between signal and maskers. Results are averaged across all three subjects.
It can be seen that, independently of the number of masking bands or the correlation of the masker and signal, introducing a 200-ms onset asynchrony has no significant effect on the threshold. This is not consistent with a mechanism for CDD based on perceptual grouping. Even if one argued that when presented with long duration signals (such as the 300-ms signals used here), the onset was less important as a cue for grouping, one would still expect to see some effect. None is seen. It is not possible directly to compare the thresholds with those of experiment 1 as the two-band conditions do not have the same frequency spacings as in the earlier experiment. However, the magnitude of the CDDs can be compared between the two experiments as CDD is a relative measure and does not change greatly with small changes in spacings of the bands (see chapter 4). The CDD is only slightly larger for six bands than for two bands in the current experiment. This differs from the results of experiment 1 in which the CDD was found to be much larger in the six-band condition than in the two-band condition. The difference between the two-band and six-band conditions in experiment 1 was attributed, at least partly, to combination bands being detected when two bands were used and the masker and signal bands were correlated. A low-pass noise was used in the current experiment to mask any combination bands. The similarity of the CDDs with two bands when a low-pass noise is present (as in the current experiment) and with six bands (in experiment 1) is consistent with the inner bands being mainly responsible for the masking and the CDD.
McFadden and Wright (1990) investigated the effect of introducing an onset asynchrony in CDD. They discussed their results in terms of the basis behind the temporal decline of masking and did not discuss the results with respect to grouping or within-channel cues. They used both long and short signal durations. For the short signal durations (steady state of 30 or 40 ms), the CDDs measured were very small and will not be discussed further here. The longer duration signal was closer to the signals presented in the previous experiments. The steady state duration was 220 ms with 10 ms ramps. This compares to the 300 ms steady state with 50 ms ramps. McFadden and Wright found that the larger signal duration produced CDDs of about 8 dB. There was little decline in the CDD as the onset of the signal was delayed. However, when the masked thresholds themselves were considered, the subjects split into two groups (the relative proportion of people in each group is unimportant as subjects were subject to screening). Some subjects had low masked thresholds which declined only very slightly as the onset asynchrony increased. The other group of subjects had very high masked thresholds (59 dB SPL and 68 dB SPL for the uncorrelated and correlated signals respectively with the minimum asynchrony (5 ms)). The thresholds for the second group declined by roughly -8 dB/decade of onset asynchrony. Now, given that McFadden and Wright used four maskers with a maximum individual overall level of 70 dB SPL and that the closest maskers were separated by 400 Hz (or roughly 2 ERBs at the signal frequency of 1250 Hz), it is perhaps surprising that the second group of subjects should have such high thresholds. Their discussion of the individual differences did not suggest any reasons for the difference, merely that one ought to be aware that differences may exist. The results of the present experiment are consistent with the results of the more sensitive group used by McFadden and Wright. It is possible that the difference between the two groups used by McFadden and Wright was due to one group taking advantage of cues that were present at all lengths of onset asynchrony, for example within-channel cues e.g. phase-locking. The second group did not take advantage of such cues and thus had to rely on different strategies to detect the signal. Such mechanisms could be across-channel grouping processes which would be expected to decline as the onset asynchrony increased or simply relying on excitation patterns which may show a sharpening as masker duration progresses due to processes such as the critical-band narrowing (Green, 1969).
The results of experiment 2 are consistent with subjects using a within-channel cue that is present at all onset asynchronies. It should be noted that if more subjects had been used, then the results may have fallen into two groups in a similar manner to McFadden and Wright (1990).