Head-Centred Sound Localisation Behaviour Reveals Head and World Centered Reference Frames in Auditory Cortex

Abstract The auditory system relies on head-centered localisation cues to estimate sound source position. Yet listeners and single neurons can encode sound in world-centred spatial reference frames. However, most studies of sound localisation constrain the subject’s orientation within a fixed speaker ring, making it difficult to distinguish which reference frame underlies spatial coding. We utilised a sound localisation task that required ferrets localise sound location, relative to the head, across rotations in the starting platform within the testing arena. Two ferrets were trained to discriminate front from back sounds. Once trained, the orientation of both the central platform and the target speakers were rotated between testing sessions, retaining the task contingency in head-centered coordinates, but breaking the alignment of head and world centered reference frames. Probe trials were presented from the other speakers so that spatial receptive fields could be constructed and compared across central platform rotations. Neural activity was recorded from auditory cortex using chronically implanted 32-channel arrays. Across 1,777 recorded units, 596 were stimulus driven. Spatially tuned neurons were predominantly head-centred, with 208 and 232 units classified as head-centred in the two animals, compared with 48 and 49 world-centred units and 33 and 26 un-tuned units, respectively. Temporal analysis showed that most units retained the same reference-frame classification from onset to offset, but the magnitude of their head-versus-world preference diminished over time, with offset responses shifting towards more ambiguous and, in some cases, world-centred coding. Together, these results demonstrate that auditory cortex maintains parallel head centered and allocentric representations that are head dominated at onset but incorporate increased world-centred information at offset, supporting perceptual constancy during dynamic listening. Competing Interest Statement The authors have declared no competing interest.