Natural spatial hearing relies on binaural spectral disparity to cancel source-spectrum ambiguities
Abstract How do we localize natural sounds? Interaural time and level differences convey little more than left versus right; elevation, front–back position and externalisation depend on spectral cues. These cues are thought to be read out monaurally, yet the spectrum at each ear confounds the unknown source spectrum with the head’s directional filter. Because natural sounds typically carry their own spectral structure, monaural localization is an ill-posed inverse problem, one that laboratory studies using flat noise have largely avoided. Here we show that monaural elevation localization of unfamiliar, spectrally structured sounds is close to chance, whereas binaural listening improves it far beyond a weighted combination of two monaural estimates. A generative model explains why: the interaural spectral ratio cancels the source spectrum, analogous to binocular disparity. Spectral priors help only modestly, and head motion adds little, likely because natural spectra fluctuate rapidly. Natural spatial hearing therefore relies on binaural spectral disparity rather than monaural template matching.
Authors
- Timo Oess (ORCID: https://orcid.org/0000-0002-4803-837X)
- Heiko Neumann (ORCID: https://orcid.org/0000-0001-7687-5792)
- Charlotte Roy (ORCID: https://orcid.org/0000-0002-7667-3135)
- Marc Ernst
Institutions
- Universität Ulm (DE)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-17
- DOI
- https://doi.org/10.21203/rs.3.rs-11441293/v1
- Primary Topic
- Neuroscience and Music Perception
- Type
- preprint
Funders
- Baden-Württemberg Stiftung
- Deutsche Forschungsgemeinschaft