Marjorie R. Leek

Adaptive procedures in psychophysical research.

As research on sensation and perception has grown more sophisticated during the last century, new adaptive methodologies have been developed to increase efficiency and reliability of measurement. An experimental procedure is said to be adaptive if the physical characteristics of the stimuli on each trial are determined by the stimuli and responses that occurred in the previous trial or sequence of trials. In this paper, the general development of adaptive procedures is described, and three commonly used methods are reviewed. Typically, a threshold value is measured using these methods, and, in some cases, other characteristics of the psychometric function underlying perceptual performance, such as slope, may be developed. Results of simulations and experiments with human subjects are reviewed to evaluate the utility of these adaptive procedures and the special circumstances under which one might be superior to another.

Minimum spectral contrast for vowel identification by normal‐hearing and hearing‐impaired listeners

To determine the minimum difference in amplitude between spectral peaks and troughs sufficient for vowel identification by normal‐hearing and hearing‐impaired listeners, four vowel‐like complex sounds were created by summing the first 30 harmonics of a 100‐Hz tone. The amplitudes of all harmonics were equal, except for two consecutive harmonics located at each of three ‘‘formant’’ locations. The amplitudes of these harmonics were equal and ranged from 1–8 dB more than the remaining components. Normal‐hearing listeners achieved greater than 75% accuracy when peak‐to‐trough differences were 1–2 dB. Normal‐hearing listeners who were tested in a noise background sufficient to raise their thresholds to the level of a flat, moderate hearing loss needed a 4‐dB difference for identification. Listeners with a moderate, flat hearing loss required a 6‐ to 7‐dB difference for identification. The results suggest, for normal‐hearing listeners, that the peak‐to‐trough amplitude difference required for identification of this set of vowels is very near the threshold for detection of a change in the amplitude spectrum of a complex signal. Hearing‐impaired listeners may have difficulty using closely spaced formants for vowel identification due to abnormal smoothing of the internal representation of the spectrum by broadened auditory filters.

Auditory temporal resolution in birds: Discrimination of harmonic complexes

The ability of three species of birds to discriminate among selected harmonic complexes with fundamental frequencies varying from 50 to 1000 Hz was examined in behavioral experiments. The stimuli were synthetic harmonic complexes with waveform shapes altered by component phase selection, holding spectral and intensive information constant. Birds were able to discriminate between waveforms with randomly selected component phases and those with all components in cosine phase, as well as between positive and negative Schroeder-phase waveforms with harmonic periods as short as 1-2 ms. By contrast, human listeners are unable to make these discriminations at periods less than about 3-4 ms. Electrophysiological measures, including cochlear microphonic and compound action potential measurements to the same stimuli used in behavioral tests, showed differences between birds and gerbils paralleling, but not completely accounting for, the psychophysical differences observed between birds and humans. It appears from these data that birds can hear the fine temporal structure in complex waveforms over very short periods. These data show birds are capable of more precise temporal resolution for complex sounds than is observed in humans and perhaps other mammals. Physiological data further show that at least part of the mechanisms underlying this high temporal resolving power resides at the peripheral level of the avian auditory system.

Estimation of psychometric functions from adaptive tracking procedures

Because adaptive tracking procedures are designed to avoid stimulus levels far from a target threshold value, the psychometric function constructed from the trial-by-trial data in the track may be accurate near the target level but a poor reflection of performance at levels far removed from the target. A series of computer simulations was undertaken to assess the reliability and accuracy of psychometric functions generated from data collected in up-down adaptive tracking procedures. Estimates of psychometric function slopes were obtained from-trial-by-trial data in simulated adaptive tracks and compared with the true characteristics of the functions used to generate the tracks. Simulations were carried out for three psychophysical procedures and two target performance levels, with tracks generated by psychometric functions with three different slopes. The functions reconstructed from the tracking data were, for the most part, accurate reflections of the true generating functions when at least 200 trials were included in the tracks. However, for 50- and 100-trial tracks, slope estimates were biased high for all simulated experimental conditions. Correction factors for slope estimates from these tracks are presented. There was no difference in the accuracy and reliability of slope estimation due to -target-level-for the adaptive track, and only minor differences due to psychophysical procedure. It is recommended that, if both threshold and slope of psychometric functions are to be estimated-from the trial-by-trial tracking data, at least 100 trials should be included in the tracks, and a three- or four-alternative forced-choice procedure should be used. However, good estimates can also be obtained using the two-alternative forced-choice procedure or less than 100 trials if appropriate corrections for bias are applied.

The internal representation of spectral contrast in hearing-impaired listeners.

Abnormal frequency resolution associated with sensorineural hearing impairment produces a smearing of spectral detail in the internal representation of complex acoustic stimuli. As a result, listeners with hearing loss may have difficulty locating spectral peaks (e.g., vowel formants) within stimuli which cue their identity. This study examined the relationship between frequency separation of peaks in a complex spectrum and the degree of spectral contrast preserved in the internal representations in normal and impaired auditory systems. Hearing-impaired and normal-hearing subjects discriminated a flat-spectrum bandpass stimulus from a stimulus containing a sinusoidal ripple across its frequency range. The peak-to-valley amplitude (in dB) necessary for detection of the ripple was measured for ripple frequencies ranging from 1 to 9 cycles/oct. Auditory filter characteristics were also measured at 1 and 3 kHz in order to examine the internal representations of the stimuli after cochlear processing. There were clear differences between groups in both auditory filter characteristics and spectral contrast detection. However, the amount of contrast in the internal representations predicted from these measurements was nearly the same for all subjects, suggesting that the reduced frequency resolution of the hearing-impaired group was largely responsible for differences in required peak-to-valley amplitude in the input spectra. Further, for all subjects, there was a trade-off between the absolute level of internal contrast necessary for ripple detection and the number of samples of this contrast available to the listener.

Informational masking and auditory attention.

Informational masking is broadly defined as a degradation of auditory detection or discrimination of a signal embedded ina context of other similar sounds; it is not related to energetic masking caused by physical interactions between signal and masker. In this paper, we report a systematic release from informational masking of a target tone in anine-tone rapid auditory sequence as the target is increasingly isolated in frequency or intensity from the remaieining sequence components. Improved target-tone frequency difference limens as isolation increases are interpreted as a reflection of increasingly focused auditory attention. The change from diffuse to highly focused attention is gradual over the frequency and intensity ranges examined, with each 1-dB increment in target intensity relative to the remaining components producing performance improvements equivalent to those produced by a 2% increase in frequency isolation. The results are modeled as bands of attention in the frequency and intensity domains. For attention directed by frequency isolation, there is a strong correspondence with auditory filters predicted by the power spectrum model of masking. These data also support the existence of an attention band of intensity, with a bandwidth of about 5–7 dB at the moderate levels used in this experiment.

Masking of tones and speech by Schroeder-phase harmonic complexes in normally hearing and hearing-impaired listeners

Tone detection and sentence recognition were measured for normally hearing and hearing-impaired listeners using maskers consisting of harmonic series with components summed in positive or negative Schroeder phase. Each task was carried out with the signal set at 60, 70, or 80 dB SPL. For listeners with normal hearing, positive Schroeder-phase complexes masked tones and sentences less than negative Schroeder-phase maskers. In the two experimental tasks, to achieve the same amount of masking, positive Schroeder-phase complexes had to be set as much as 12-15 dB higher than negative Schroeder-phase complexes. Large phase effects were observed on both tasks at all three test levels. The two maskers were more nearly equal in effectiveness in the presence of cochlear damage. The findings support an interpretation that involves differences in the shape of the basilar-membrane waveform generated by each masker and active cochlear processing which enhances the internal signal-to-masker ratio for signals presented in the positive Schroeder masker. This spectral enhancement appears to require nonlinear active gain that is characteristic of normal auditory processing at moderate presentation levels. The results of the sentence recognition task suggest that group differences observed in tone detection transfer fairly directly to speech perception under masking.

Auditory filter shapes of normal-hearing and hearing-impaired listeners in continuous broadband noise

Listeners with sensorineural hearing impairment typically exhibit auditory processing deficits such as reduced frequency and/or temporal resolution. Such deficits may represent separate sequela of auditory pathology or may result directly from the sensitivity loss and the requirement to listen at high levels. To assess the impact of increased thresholds on frequency resolution, auditory filter characteristics were determined for hearing-impaired and normal-hearing listeners at 500 and 2000 Hz in the presence of continuous broadband noise meant as a rough simulation of hearing loss. In the fitting procedure, the low-frequency skirt of the derived auditory filter was allowed to vary as a function of signal level, permitting different filter shapes to be estimated at high versus low signal levels. Listeners with moderate hearing losses at 2000 Hz demonstrated near-normal auditory filter shapes for lower signal levels, but increasingly broad and asymmetric filters as signal level was raised. At 500 Hz, where hearing losses were mild, filter bandwidths increased little at the higher signal levels. The presence of broadband noise had essentially no effect on filter shapes of either listener group. The filter shape abnormalities demonstrated by listeners with moderate hearing loss, which were not observed in normal-hearing listeners at the same signal levels, indicate that poor frequency resolution in these patients for high-intensity stimuli does not follow directly from decreased sensitivity, but instead reflects an independent pathology.

Experience with a yes–no single-interval maximum-likelihood procedure

The report in 1993 by Green [J. Acoust. Soc. Am. 93, 2096-2105 (1993)] describing the application of a new psychophysical method requiring few trials and little time to measure auditory thresholds has generated considerable interest among experimentalists. The procedure uses a single-interval stimulus presentation, requests a yes-no decision by subjects, and implements a maximum-likelihood calculation to determine the next trial stimulus level within an adaptive track, as well as the final threshold estimate. Data are presented here describing separate experiences with this procedure in two laboratories in both detection and discrimination tasks. Issues addressed include comparisons with more traditional psychophysical methods, variability in threshold estimates, experimental time required, and possible minor modifications to improve the basic procedure. Results using this procedure are comparable in terms of variability of estimates to those emerging from more lengthy procedures. However, because it may be difficult for some listeners to maintain a consistent criterion and because attentional lapses may be costly, experimenters must be willing to monitor performance closely and repeat some tracks in cases where excessively high variability is noted. Further, this procedure may not be suitable for tasks for which the form of the psychometric function is not well-established. Modifications allowing a variable slope parameter in the maximum-likelihood evaluations of psychometric functions may be of benefit.

Psychophysical Estimates of Cochlear Phase Response: Masking by Harmonic Complexes

Harmonic complexes with identical component frequencies and amplitudes but different phase spectra may be differentially effective as maskers. Such harmonic waveforms, constructed with positive or negative Schroeder phases, have similar envelopes and identical long-term power spectra, but the positive Schroeder-phase waveform is typically a less effective masker than the negative Schroeder-phase waveform. These masking differences have been attributed to an interaction between the masker phase spectrum and the phase characteristic of the basilar membrane. To explore this relationship, the gradient of stimulus phase change across masker bandwidth was varied by systematically altering the Schroeder-phase algorithm. Observers detected a signal tone added in-phase to a single component of a masker whose frequencies ranged from 200 to 5000 Hz, with a fundamental frequency of 100 Hz. For signal frequencies of 1000-4000 Hz, differences in masking across the harmonic complexes could be as large as 5-10 dB for phase gradients changing by only 10%. The phase gradient that resulted in a minimum amount of masking varied with signal frequency, with low frequencies masked least effectively by stimuli with rapidly changing component phases and high frequencies masked by stimuli with more shallow phase gradients. A gammachirp filter was implemented to model these results, predicting the qualitative changes in curvature of the phase-by-frequency function estimated from the empirical data. In some cases, small modifications to the gammachirp filter produced better quantitative predictions of curvature changes across frequency, but this filter, as implemented here, was unable to accurately represent all the data.