Robert A. Lutfi - US grants

A new methodology is proposed to investigate information processing of complex, auditory patterns by normal-hearing listeners. In this procedure, each auditory pattern represents a randomly selected sample of values of some acoustic parameter. In a simple case, for example, the patterns might be tone sequences with randomly selected frequencies. The listeners task is to discriminate target from nontarget patterns based on a average difference in the values of the sampled parameter. By definition, the uncertainty associated with the random variation in patterns is a source of information. The listener must process this information in order to perform the sample-discrimination. Sample-discrimination experiments are proposed for a variety of auditory patterns of increasing complexity. Listener performance is expressed relative to that of a theoretical ideal observer to permit comparisons across experiments. Also, trial-by-trial analyses are proposed to provide tests of specific hypotheses regarding how listeners utilize the information in these patterns. The results of these analyses will allow answers to questions regarding how the detection and discrimination of complex auditory patterns is affected by information processing capacity, information weighting, and interference both within and across stimulus dimensions. As a long term goal of the project, experimental results will be used to develop a computational model for determining how best to package the acoustic information in patterns so as to maximize information transmission.

DESCRIPTION (provided by applicant): We rely critically on our ability to identify simple objects and events from sound to function normally in everyday listening. Yet, despite its importance, little is known regarding this ability. A new psychophysical method promises to change this situation. Perturbation analysis has enjoyed recent success in vision as a means of revealing decision processes underlying object identification. Here it is applied to the auditory identification of elementary sound sources and their attributes. The application proceeds in three stages: First, the sounds of simple resonant sources are synthesized according to their equations of motion from theoretical acoustics. Second, listener decision strategy is determined from regression weights relating listener judgments to lawful perturbations in acoustic parameters as dictated by the equations for motion. Third, limits in processing are identified by comparing the obtained weights and residuals to those of a maximum-likelihood observer that bases decisions on acoustic information unique to the source attribute(s) being judged. The approach represents a significant advance over past methods that infer decision strategy from performance accuracy or from the effect of placing acoustic cues in unlawful opposition. Specific aims are: (1) to determine precisely how listeners use multiple sources of acoustic information to identify rudimentary sound sources and their attributes, (2) to isolate the factors that limit identification using training feedback and instruction, and (3) to test a general theoretical framework for predicting results in which less accurate though perceptually more robust acoustic cues offer a viable basis for identification. By advancing our understanding of the normal processes underlying sound source identification the results may prove key in the development of technologies and rehabilitative strategies that deal more effectively with the impact of dysfunctional hearing on everyday listening.

DESCRIPTION (provided by applicant): Our remarkable capacity to process information in sound is demonstrated everyday as we make sense of the continuous pattern of variation in the acoustic signals we encounter. The long-term goal of this project is to better understand this ability in normal-hearing adults through controlled measures of their ability to detect and discriminate variation in acoustic patterns made up of tones. There are three key elements of our approach. First, all efforts are linked by a single theoretical framework where the information in the patterns is given precise meaning and listener performance is evaluated relative to a common performance standard. Second, the manner in which different internal factors influence the listener's response is determined from trial-by:trial analyses of the data. Third, specific hypotheses regarding the outcome of experiments are generated based on known nonlinear transformations performed at the auditory periphery and a decision model that has made accurate predictions for the results of many past studies [R.A. Lutfi, J. Acoust. Soc. Am. 94, 748-758 (1993)]. Present aims are to: (1) measure auditory nonlinearity in the discrimination of pure-tone frequency, intensity and duration (2) determine the specific influence of auditory nonlinearity, listener decision weights, and internal noise on the integration of information across frequency and time, and (3) demonstrate successful achievement of aims by maximizing information rate through stimulus design. By advancing our understanding of the normal adult capacity to process information in sound the results may prove key in the development of technologies and rehabilitative strategies that deal more effectively with the impact of dysfunctional hearing on everyday listening.

DESCRIPTION (provided by applicant): We possess a remarkable ability to 'hear out' (segregate) the individual sound sources that makeup the complex, acoustic environments we encounter in everyday listening. Understanding this ability has been a preoccupation of psychoacoustic research, but it is a tremendous challenge. Perturbation analysis is a methodological approach that has advanced understanding of related problems in vision [Murray, R.F. 2011. J. of Vision 11, 1-25]. Here the approach is adapted to audition. The application proceeds in three stages: First, simple speech and environmental sounds are synthesized according to a generative model of the sound-producing source. Second, listener decision strategy in segregating a target from non-target (noise) source is determined from decision weights (regression coefficients) relating listener judgments regarding the target to lawful perturbations in acoustic parameters, as dictated by the generative model. Third, factors limiting segregation are identified by comparing the obtained weights and residuals to those of a maximum- likelihood (ML) observer that optimizes segregation based on the equations of motion of the generating source. The approach represents a significant advance over traditional methods that infer listener decision strategy from performance accuracy or from the effect of placing acoustic cues in unlawful opposition. Specific aims are to apply this method to (1) test between two major classes of segregation models, target enhancement vs. noise cancellation, (2) test predictions of acoustics-based versus mechanics-based hypotheses regarding source segregation, common fate vs. common source, (3) quantify the relative influence of target-noise uncertainty and similarity, (4) determine the influence of level dominance on sound source segregation, and (5) account for diverse patterns of behavior across tasks by using the decision weights to identify individual listening styles. It is expected that the knowledge gained from these studies will inform research on broader aspects of the problem of sound source segregation, including those relevant to the evaluation and treatment of disordered hearing as it impacts everyday listening in noisy environments.

Project Summary Young adults with normal audiometric thresholds vary widely in their ability to listen in everyday noisy environments. Many perform as poorly in studies as their age-matched, hard-of-hearing counterparts [Kidd et al. (2002). J. Assoc. Res. Otolaryn. 3, 107-119]. The variation challenges the conventional view of hearing loss, which assumes the audiogram to be the gold standard for evaluating hearing; however, the causes of the variation remain unclear. Now, new developments in our lab and others promise progress in understanding. Measures of threshold fine structure have shown that wide individual variation in thresholds are missed by the conventional audiogram [Lee & Long (2012). Hear. Res. 283, 24-32]. Individual differences in decision weights, taken to reflect the reliance listeners place on different frequencies in psychophysical tasks, have been linked to irregularities in cochlear micromechanics unique to individual ears [Lee et al. (2016), Adv. Exp. Med. Biol. 894:457-465]. Computational models have isolated a small number of factors that show promise in predicting individual differences across diverse target-in-noise listening tasks [Lutfi et al. (2013), J. Acoust. Soc. Am. 134:2160-2170]. And, levels of noise exposure, once thought to present no risk of harm, have been shown to produce irreversible loss of synaptic connections to hair cells and subsequent degeneration of afferents in mice. The pathology is not detected by conventional audiometry, leading to speculation that it may be widespread in the population, affecting listening in noise [Kujawa and Liberman (2009), J. Neurosci. 29(45):14077?14085]. Sparked by these developments, this proposal represents a new effort to understand individual differences in the ability of young clinically normal-hearing adults to listen effectively in noise. It differs fundamentally from past efforts in approach. The goal is to isolate the primary sources of variation and model their effects by determining precisely how listeners differ in their use of information distinguishing targets from noise. This is achieved by (1) formulating a general decision model relating the listener's trial-by-trial judgments to this information, (2) estimating the parameters of this model by logistic regression of the trial-by- trial data and (3) comparing the estimated values to those of a maximum-likelihood observer that yields best performance for each task. The approach represents a significant advance over conventional methods that infer underlying processes from measures of performance accuracy. Specific aims are to apply this approach to (1) determine the relative contribution of peripheral and central processes and their interaction to individual differences, (2) account for general patterns of listener behavior from `individual listening styles' and (3) develop a low-parameter computational model for predicting individual differences across diverse listening tasks. It is expected that the knowledge gained from these studies will inform efforts that seek to improve the evaluation, classification and treatment of what is a debilitating problem for many, the challenge of listening in everyday noise.