Daniel L. Weber

Attention bands in absolute identification

If both the number of one-dimensional signals and their range are sufficiently large (about 7 and 20 dB for loudness), the information transmitted in absolute identification is not much increased by increasing either variable (Miller, 1956; Braida & Durlach, 1972). The data can be represented in terms of Thurstonian discriminal dispersions in which the variance is proportional to the square of the signal range in decibels (Durlach & Braida, 1969; Gravetter & Lockhead, 1973), but it is by no means obvious what sorts of mechanisms would lead to this model. An alternative is proposed, namely, that there is a roving attention band, about 10 to 15 dB wide, such that signals falling within the band are represented by a sensory sample size about an order ot magnitude larger than when the same signal falls outside the band. With reasonable choices for parameters, including the subjective continuum growing as a power function of intensity with an exponent about 3, this nicely accounts for the data. In an attempt to examine the change of performance with range, we replicated the BraidaoDurlach experiment with many additional points. These data are not, however, adequate to decide between the two models.

Effects of practice and distribution of auditory signals on absolute identification

In absolute identification of intensity, signals near the edges of the range being used are usually identified more accurately than those in the midrange. In one account, the extreme signals serve as anchors, and judgments are postulated to deteriorate as the distance from the signal to the nearest anchor increases. Our data suggest that, provided one corrects for the inherent asymmetry of errors for end and interior signals, the edge effect is rather smaller than it might first appear and is largely confined to the more intense edge. Moreover, anchors are not necessarily located at the edges of the range, but rather at the edges of the largest subset among which difficult discriminations are required. Further, this subset is not defined wholly by the signals used in a particular run, but by these together with those previously encountered in that day’s session. Neither practice nor payoffs appear to influence the location of the anchor so long as the discrimination requirements are maintained. Finally, the role of anchors is interpreted in terms of the differential location of an attention band which controls the sample size upon which the representation of the signal is based.

Growth of masking and the auditory filter

The threshold for a sinusoidal signal (1, 2, and 4 kHz) centered in the ’’notch’’ of a broadband masker was determined as a function of notch width for five noise spectrum levels (10, 20, 30, 40, and 50 dB SPL). For narrow notch widths the signal‐to‐noise ratio at threshold remains constant as a function of level, which according to the critical‐ratio hypothesis implies an auditory filter of constant bandwidth. For wide notch widths the signal‐to‐noise ratio at threshold increases as a function of level and this implies an auditory filter of increasing bandwidth. If the estimates of the filter bandwidth obtained for wide notch widths are used to predict thresholds for broadband noise, the corresponding signal‐to‐noise ratios will increase as a function of noise spectrum level. The predicted increase in signal‐to‐noise ratio is very small, however, and provides a good description of most available data. In fact, the predicted increase equals that observed by Reed and Bilger [J. Acoust. Soc. Am. 53, 1039–1044 (1973)]. The increase in filter bandwidth has significant consequences only when the signal and masker are widely separated in frequency; for other conditions, the assumption of a constant filter permits accurate predictions of performance.

Temporal factors and suppression effects in backward and forward masking.

We discuss several experiments examining the influence of temporal parameters on suppression effects in backward and forward masking. The signal is always a brief 10-ms 2-kHz sinusoid; the masker a narrow band of noise of 40-dB spectrum level, 200-Hz wide, centered at the signal frequency. In some conditions, the presence of a second band of noise of the same spectral level in the region of 2300--3700 Hz appears to suppress the effects of the masker. Changes in the amount of suppression are examined as functions of the delay and duration of the suppressor (experiments 1 and 2). Adding the suppressor during the 50-ms interval nearest the signal produces changes in the signal threshold that are similar to those produced by reducing the level of the masker during this interval for both backward and forward masking (experiment 4). The similarity of these results suggests the operation of peripheral processes common to both backward and forward masking. However, if one increases the duration of the suppressor beyond this 50-ms interval there is no effect on forward masking, but large additional reductions in backward masking. This difference, in conjunction with other recent experiments, suggests the operation of additional central processes in backward masking. For some observers, these additional processes appear to operate over fairly long periods of time (250-500 ms). Such long durations are inconsistent with the estimates of integration time (less than or equal to 200 ms) obtained for these same observers (experiment 3).

Detection of temporally uncertain signals

The effects of signal uncertainty on detection performance were measured using a new procedure that allows precise specification of the initial temporal uncertainty. Five different models of the detection process, two assuming a continous representation of the sensory input and three assuming a discrete representation, were compared with the obtained data. The effects of varying signal uncertainty (the number of potential signal intervals was one, five, or ten) had little effect on detection performance. The one-parameter form of the choice model can be rejected without hesitation. The continuous Gaussian model and the symmetric two-state model are significantly different from the data. The high threshold and sophisticated two-state models provide accurate descriptions of the data that cannot be rejected on statistical grounds.

Do off‐frequency simultaneous maskers suppress the signal?

Psychophysical tuning curves obtained in forward masking show greater tuning than those obtained in simultaneous masking. This difference is often attributed to the contribution of suppression to the masking produced by off-frequency simultaneous maskers. In this experiment, simultaneous-masking tuning curves were obtained using a 195-ms, 1-kHz sinusoidal signal presented at 40 dB SPL. If the maskers identified in this procedure reduce signal detectability by suppressing the response to the signal, then it should be possible to demonstrate suppression effects between stimuli with the parameters of the masker and signal. One conventional method for demonstrating suppression is to show a reduction in the amount of forward masking produced by one stimulus when a second stimulus is added to it. When used to test the effect of the masker upon the signal, this procedure does not show the suppression supposedly produced by the off-frequency maskers. These data are consistent with an alternative explanation involving only excitatory interactions between masker and signal.

Detection and recognition of pure tones in noise

We examine the predictions of a new theorem relating signal identification (specifying a signal as a particular member of a set of potential signals) to signal detection (discriminating the presence of a signal). The theorem, derived in the context of signal-detection theory, requires that the signals be equally detectable and orthogonal. Our sinusoidal signals are partially masked by noise and their intensities adjusted to produce equal-signal detectability; we do not examine this assumption of the theorem. The theorem generally provides a reasonably accurate description of recognition performance for two-signal and four-signal conditions and is equally accurate for both the Yes-No and category-rating procedures. In a preliminary investigation of the orthogonality assumption, we varied the frequency separation between two signals. When the frequency separation between two signals is small (20 Hz near 1 kHz), the theorem fails to provide a good description of performance.

Suppression effects in backward and forward masking.

The differences in the suppression effect observed in forward and backward masking are consistent with an interpretation that suppression in forward masking results from a reduction of the effective level of the masker in the aditory periphery, and that the suppression in backward masking is influenced by these peripheral processes, but is dominated by additional, central processes. This conclusion is supported by experiments that show differences in the effect of ipsilateral versus contralateral presentation of the suppressor, and differences in the amount of the suppression observed as a function of the level, duration, and frequency of the suppressor.

Time course of suppression

The threshold for a 10‐msec, 2‐kHz sinusoid was examined in backward and forward masking. The delay between the signal and the 500‐msec “critical‐band” masker (N0 = 40 dB) was fixed at 0 msec. If a “suppression” band of noise occurs simultaneously with the masker, thresholds are considerably reduced (20 dB in backward masking, 8 dB in forward masking), or “suppressed.” Varying the temporal position of the “suppression” band with respect to the “critical‐hand” noise and observing changes in threshold allows one to estimate the time course of suppression. In both backward and forward masking, no suppression can be seen for temporal differences greater than 40 msec, whereas the results of other experiments [ L. L. Elliot, J. Acoust. Soc. Am. 42, 143–153 (1967)] suggest that excitation can be seen for 100–200 msec. The relatively greater suppression observed for backward masking is expected if “suppression” is a more rapid process than “excitation.” [Work supported by NIH.]

Combined effect of two suppressors.

Threshold for a 10-ms sinusoidal signal was measured as a function of signal frequency (0.65 to 1.40 kHz) in several forward-masking conditions. For signal frequencies near 1.0 kHz, the forward masking produced by a 395-ms, 100-Hz-wide noise centered at 1.0 kHz (total power 60 dB) could be reduced by the addition of a sinusoid to the noise. The effects of four sinusoidal suppressors (frequencies of 0.70, 0.85, 1.15, and 1.40 kHz, all at 75 dB SPL) were examined individually and in the six possible pairwise combinations. In general, the threshold reduction produced by two suppressors together was no greater than the larger of the reductions produced by the suppressors individually. It appears that suppression produced by different stimuli does not combine to yield significantly larger effects. Instead, the amount of suppression appears to be restricted to a specific range and it is not possible to exceed this limit.