Anticipated ITD statistics are built into human sound localization

The variability of natural scenes places perceptual processes in the realm of statistical inference. Perceptual tasks may be optimized if the invariant statistical structure of sensory cues is built into the neural processing. We investigated this question in human sound localization. Localizing sounds in the horizontal plane relies on interaural time differences (ITD). We estimated components of ITD statistics from human head-related transfer functions (HRTFs), which can be assumed invariant across contexts. ITD varied with azimuth following a sigmoid relationship, whose slope was steepest at the center in most frequencies. In addition, ITD was more variable over time for sounds located in the periphery compared to the center, in a frequency-dependent manner. We tested the hypothesis that these statistics are anticipated by the human brain, influencing spatial discriminability and novelty detection. Previously reported thresholds for discriminating ITD changes were better predicted by a model relying on both ITD slope and variability than on ITD slope alone. Furthermore, mismatch negativity (MMN) brain signals, a pre-attentive index of deviance detection recorded in subjects undergoing a spatial oddball paradigm, were weighted by ITD slope and variability of the standard location. These results show that spatial discriminability thresholds and novelty detection are consistent with a representation of anticipated ITD statistics in the brain, supporting the hypothesis that high-order statistics are built into human perceptual processes biasing behavior.