Interactions between egocentric and allocentric spatial coding of sounds revealed by a multisensory learning paradigm

Abstract

Although sound position is initially head-centred (egocentric coordinates), our brain can also represent sounds relative to one another (allocentric coordinates). Whether reference frames for spatial hearing are independent or interact remained largely unexplored. Here we developed a new allocentric spatial-hearing training and tested whether it can improve egocentric sound-localisation performance in normal-hearing adults listening with one ear plugged. Two groups of participants (N = 15 each) performed an egocentric sound-localisation task (point to a syllable), in monaural listening, before and after 4-days of multisensory training on triplets of white-noise bursts paired with occasional visual feedback. Critically, one group performed an allocentric task (auditory bisection task), whereas the other processed the same stimuli to perform an egocentric task (pointing to a designated sound of the triplet). Unlike most previous works, we tested also a no training group (N = 15). Egocentric sound-localisation abilities in the horizontal plane improved for all groups in the space ipsilateral to the ear-plug. This unexpected finding highlights the importance of including a no training group when studying sound localisation re-learning. Yet, performance changes were qualitatively different in trained compared to untrained participants, providing initial evidence that allocentric and multisensory procedures may prove useful when aiming to promote sound localisation re-learning.

SPHERE: A novel approach to 3D and active sound localization

“…In addition, because physical sound sources are present, these approaches face the problem of controlling the contribution of visual cues to sound localization, as any visual prior will contribute to the interpretation of auditory cues (Jackson & Morton 1984;Makous & Middlebrooks 1990). Participants are therefore blind-folded, sometimes from the moment they enter the experimental room (e.g., Ahrens and al., 2019), or are instructed to close their eyes at specific moments during the task (e.g., , or face speakers hidden behind a fabric panel of some sort (e.g., Rabini et al, 2019).…”

Reaching to sounds in virtual reality: A multisensory-motor approach to re-learn sound localisation

“…Finally, following previous works (Rabini et al, 2019;Steadman et al, 2019;Valzolgher, Campus, Rabini, Gori & Pavani, under review) we examined whether re-learning across blocks relates to the magnitude of the immediate ear-plug effect. To this aim, we correlated across participants the ear-plug effect (measured as the difference in absolute localisation error between B and M1; positive values indicate larger plug-effect) and re-learning (measured as the difference in absolute localisation error between M1 and M4; positive values indicate larger re-learning).…”

“…In addition, they change on a daily basis, due to the diversity of listening contexts to which we are all exposed (Majdak, Goupell & Laback, 2010). To cope with these continuous changes in auditory cues, the brain remains capable of updating sound-space correspondences throughout life (e.g., Carlile, Balachandar & Kelly 2014;Keating & King, 2015;Rabini, Altobelli & Pavani, 2019;Strelnikov, Rosito & Barone, 2011;Van Wanrooij, John & Opstal, 2005). In the present work, we examined reaching to sounds in virtual reality as a multisensory-motor strategy for re-learning sound localisation in adulthood.…”

“…Whenever the other sensory systems (e.g., vision) provide reliable spatial information about auditory events, the brain exploits these additional sensory sources to calibrate and optimize internal models for spatial hearing (Keating & King, 2015). This notion emerged, for instance, from studies that examined re-learning of sound-space correspondences when auditory cues were temporarily altered using monaural ear-plugs (Rabini et al, 2019;Strelninkov et al, 2011;Trapeau & Schönwiesner, 2015), ear molds (Van Wanrooij & Van Opstal, 2005) or non-individualised HRTFs (Head-Related Transfer Functions; Honda, Shibata, Gyoba, Saitou, Iwaya & Suzuki, 2007;Parseihian & Katz, 2012;Steadman, Kim, Lestang, Goodman & Picinali, 2019). In these simulated altered-listening conditions, multisensory training procedures proved effective for re-learning sound-space correspondences (for reviews see: Carlile, 2014;Keating & King, 2015;Knudsen & Knudsen, 1985;Mendonça, 2014;Irving & Moore, 2011).…”

“…Yet, studies that examined spatial hearing re-learning have typically limited the possibilities for head-movements and, in general, the contribution of active listening. Headmovements have been prevented by using a chinrest (Kumpik et al, 2010;Strelnikov et al, 2011) or by instructing participants to refrain from head-movements during listening (Rabini et al, 2019;Van Wanrooij & Van Opstal, 2005;Zahorik et al, 2006).…”