Model structure. Gray blocks: Model’s processing pipeline consisting of 1) the feature extraction stage to compute spatial features from the binaural sound; 2) the Bayesian inference stage integrating the sensory evidence obtained by comparison with feature templates and prior beliefs to estimate the most probable direction; and 3) the response stage transforming the internal estimate to the final localization response. White blocks: Elements required to fit the model to an individual subject consisting of listener performances in estimating sound direction and individual HRTF dataset.