Sound localisation is one of the most representative function of the auditory system and, as such, it has been extensively investigated across species. Spatial hearing can be dramatically altered across the life span, yet research in humans have highlighted the remarkable capacity of the brain to adapt to changes of listening conditions, such as temporary ear plugging or long lasting hearing impairments. Although several investigations have examined accommodation to altered auditory cues (Chapter 1), a common theoretical framework seems to lack and a number of questions remain open. This limits the possibility to translate our current knowledge into concrete clinical applications for individuals who experience spatial hearing difficulties after hearing loss. The current dissertation reflects the attempt to answer specific questions regarding the process of sound localisation. The first study (Chapter 2) aimed to investigate the relation between different reference frames in spatial hearing, namely egocentric and allocentric sound representation. We studies this topic in the context of a learning paradigm, assessing to what extent localisation of single sounds in simulated monaural hearing (unilateral ear plugging) can improve following an audio-visual spatial hearing training focused on egocentric sound processing vs allocentric sound processing. An untrained group was also included in the study. We found that localisation performance in the horizontal plane improved specifically in the side ipsilateral to the ear-plug for all groups. Yet, the trained groups showed a qualitatively different change of performance after four days of multisensory ego/allocentric training compared to the untrained group, providing initial evidence of the possible role of allocentric coding in acoustic space re-learning. These results further highlight the importance of including a test-retest group in paradigms of sound localisation training. The second study (Chapter 3) focused on a specific aspect of the phenomenological experience of spatial hearing, namely the subjective confidence about the perceived sound position. We examined the relation between objective localisation accuracy and subjective certainty while participants localised sounds in two different listening conditions – binaural or simulated monaural hearing. Results showed that overall subjective certainty on sound position decreased in the altered listening condition (unilateral ear-plugging). In simulated monaural hearing, localisation accuracy and spatial confidence dissociated. For instance, there were trials in which participants were accurate, but felt uncertain, and trials in which they were less accurate but expressed higher ratings of spatial confidence on sound position. Furthermore, subjective confidence increased as a function of time within the testing block, and it was related to the spatial distribution of the perceived sound-source position. The third study (Chapter 4) exploited magnetoencephalography (MEG) to study the dynamics of the cortical network implied in active sound localisation. We implemented a novel apparatus to study sound localisation in MEG with real sounds in external space, and collected behavioural and subjective responses (i.e., accuracy and confidence, as in Study 2) during this altered listening condition. Results showed that participants were able to perceive the spatial difference between the positions of stimulation, thus proving the reliability of our novel setting for the study of spatial hearing in MEG. MEG data highlight a distributed bilateral cortical network involved in active sound localisation, which emerged shortly after stimulus presentation (100—125 ms). The network comprise the classical dorsal auditory pathway plus other cortical regions usually underestimated in previous literature – most notably, regions in the central sulcus/precentral gyrus possibly involved in head movements. Connectivity analysis revealed different patterns of neural coupling, as a function of frequency band. In particular, coherence in high gamma revealed significant connections involving the parietal cortex and the posterior superior temporal cortex. In the final chapter (Chapter 5), I summarise the main findings of the three studies, discuss their implications and outline potential future directions.

La perception d’un mouvement biologique correspond à l’aptitude à recueillir des informations (comme par exemple, le type d’activité) issues d’un objet animé en mouvement à partir d’indices visuels restreints. Cette méthode a été élaborée et instaurée par Johansson en 1973, à l’aide de simples points lumineux placés sur des individus, à des endroits stratégiques de leurs articulations. Il a été démontré que la perception, ou reconnaissance, du mouvement biologique joue un rôle déterminant dans des activités cruciales pour la survie et la vie sociale des humains et des primates. Par conséquent, l’étude de l’analyse visuelle de l’action chez l’Homme a retenu l’attention des scientifiques pendant plusieurs décennies. Ces études sont essentiellement axées sur informations cinématiques en provenance de différents mouvements (comme le type d’activité ou les états émotionnels), le rôle moteur dans la perception des actions ainsi que les mécanismes sous-jacents et les substrats neurobiologiques associés. Ces derniers constituent le principal centre d’intérêt de la présente étude, dans laquelle nous proposons un nouveau modèle descriptif de simulation bayésienne avec minimisation du risque. Ce modèle est capable de distinguer la direction d’un ballon à partir d’un mouvement biologique complexe correspondant à un tir de soccer. Ce modèle de simulation est inspiré de précédents modèles, neurophysiologiquement possibles, de la perception du mouvement biologique ainsi que de récentes études. De ce fait, le modèle présenté ici ne s’intéresse qu’à la voie dorsale qui traite les informations visuelles relatives au mouvement, conformément à la théorie des deux voies visuelles. Les stimuli visuels utilisés, quant à eux, proviennent d’une précédente étude psychophysique menée dans notre laboratoire chez des athlètes. En utilisant les données psychophysiques de cette étude antérieure 3 et en ajustant une série de paramètres, le modèle proposé a été capable de simuler la fonction psychométrique ainsi que le temps de réaction moyen mesurés expérimentalement chez les athlètes. Bien qu’il ait été établi que le système visuel intègre de manière optimale l’ensemble des indices visuels pendant le processus de prise de décision, les résultats obtenus sont en lien avec l’hypothèse selon laquelle les indices de mouvement sont plus importants que la forme dynamique dans le traitement des informations relatives au mouvement. Les simulations étant concluantes, le présent modèle permet non seulement de mieux comprendre le sujet en question, mais s’avère également prometteur pour le secteur de l’industrie. Il permettrait, par exemple, de prédire l’impact des distorsions optiques, induites par la conception de verres progressifs, sur la prise de décision chez l’Homme. Mots-clés : Mouvement biologique, Bayésien, Voie dorsale, Modèle de simulation hiérarchique, Fonction psychométrique, Temps de réaction / The ability to recover information (e.g., identity or type of activity) about a moving living object from a sparse input is known as Biological Motion perception. This sparse input has been created and introduced by Johansson in 1973, using only light points placed on an individual's strategic joints. Biological motion perception/recognition proves to play a significant role in activities that are critical to the survival and social life of humans and primates. In this regard, the study of visual analysis of human action had the attention of scientists for decades. These studies are mainly focused on: kinematics information of the different movements (such as type of activity, emotional states), motor role in the perception of actions and underlying mechanisms, and associated neurobiological substrates. The latter being the main focus of the present study, a new descriptive risk-averse Bayesian simulation model, capable of discerning the ball’s direction from a set of complex biological motion soccer-kick stimuli is proposed. Inspired by the previous, neurophysiologically plausible, biological motion perception models and recent studies, the simulation model only represents the dorsal pathway as a motion information processing section of the visual system according to the two-stream theory, while the stimuli used have been obtained from a previous psychophysical study on athletes. Moreover, using the psychophysical data from the same study and tuning a set of parameters, the model could successfully simulate the psychometric function and average reaction time of the athlete participants of the aforementioned study. 5 Although it is established that the visual system optimally integrates all available visual cues in the decision-making process, the results conform to the speculations favouring motion cue importance over dynamic form by only depending on motion information processing. As a functioning simulator, the present simulation model not only introduces some insight into the subject at hand but also shows promise for industry use. For example, predicting the impact of the lens-induced distortions, caused by various lens designs, on human decision-making. Keywords: Biological motion, Bayesian, Dorsal pathway, Hierarchical simulation model, Psychometric function, Reaction time