Auditory model-based selection of the most informative experimental conditions

Identifying the causes of a person's hearing impairment is a challenging task. Even though a broad range of measurement techniques exist, links between the results of one or several listening tests and possible pathologies need to be found. Drawing conclusions from measurement results that were influenced by pathologies in this highly non-linear auditory system remains very difficult. In addition, measurement time is restricted, especially in clinical settings. A central but difficult goal is to maximize the diagnostic information that is collectable within a certain time frame. Computer models simulating auditory processing and possible impairments could be employed to assist in such diagnostics. By using the model-based experiment-steering approach introduced in Hermann and Dietz (2021, Acta Acustica, 5:51, doi: 10.1051/aacus/2021043), the current study demonstrates its applicability using five young, normal-hearing subjects. In the model-based selection procedure, those stimuli providing the most information about the model parameters were identified in parallel to the measurement, and subsequently presented to the participant. The same binaural tone-in-noise detection task was conducted with two measurement procedures: A standard adaptive staircase procedure and the model-based selection procedure. For this proof of concept, an existing auditory processing model was adopted. Its four free parameters enabled the characterization of the subjects' 250 Hz channel. The model parameters best predicting the subject's sensitivity to a diotic and various dichotic conditions, were obtained using a maximum-likelihood approach. On average, the same accuracy of model parameter estimation was reached 2.5 times faster with the model-steered procedure compared to the standard adaptive procedure. Difficulties regarding the choice of a reliable model and issues to be considered when deciding on reasonable discretization steps of the model parameters are discussed. Although the physiological causes of an individual's results cannot be diagnosed with this procedure, a characterization in terms of functional parameters is possible.