Evaluation of spatial tasks in virtual acoustic environments by means of modeling individual localization performances

Virtual acoustic environments (VAEs) are an excellent tool in hearing research, especially in the context of investigating spatial-hearing abilities. On the one hand, the development of VAEs requires a solid evaluation, which can be simplified by applying auditory models. On the other hand, VAE research provides data, which can support the further improvement of auditory models. Here, we describe how Bayesian inference can predict listeners' behavior when estimating the spatial direction of a static sound source presented in a VAE experiment. We show which components of the behavioral process are reflected in the model structure. Importantly, we highlight which acoustic cues are important to obtain accurate model predictions of listeners' localization performance in VAE. Moreover, we describe the influence of spatial priors and sensorimotor noise on response behavior. To account for inter-individual differences, we further demonstrate the necessity of individual calibration of sensory noise parameters in addition to the individual acoustic properties captured in head-related transfer functions.