Explicit cue weighting in a Bayesian model of auditory localization: Quantifying the relative contributions of binaural and spectral cues

Auditory localization depends on the integration of binaural and spectral cues, yet their relative contributions to localization judgments remain difficult to quantify within a unified computational framework. This study introduces a Bayesian model of static auditory localization with an explicit cue-weighting (ECW) mechanism that estimates cue-specific weighting parameters by fitting behavioral responses. By embedding ECW into the likelihood function, the model provides a probabilistic framework for characterizing how weighted acoustic cues, spatial priors, and motor noise jointly shape localization responses. The estimated weighting patterns were broadly consistent with classic psychoacoustic findings, showing stronger weighting of binaural cues together with selective high-frequency weighting of spectral cues. The model further accounted for major behavioral trends across broadband stimulation and several challenging acoustic conditions, including non-individualized head-related transfer functions and reduced spectral resolution via vocoders. Together, these results indicate that the proposed ECW model extends Bayesian models of auditory localization by making the relative contributions of established binaural and spectral cues behaviorally estimable while providing competitive predictive performance.