The present research addressed the development of specialization in face processing in infancy by examining the roles of race and emotion. An other-race face among own-race faces draws adults’ attention to a greater degree than an own-race face among other-race faces due to the “other-race” feature in other-race faces. This feature underlies race-based differences in adults’ face processing. The current studies investigated the development of this mechanism as well as the influence that this mechanism has on emotion processing in infancy. In Experiment 1, Caucasian 3.5- and 9- month-olds exhibited a preference for a pattern containing an Asian face among seven Caucasian faces over a pattern containing a Caucasian face among seven Asian faces. This preference was not driven by the majority of elements in the images, because a control group of infants failed to exhibit a preference between homogeneous patterns containing eight Caucasian versus eight Asian faces. The asymmetrical attentional engagement by other-race faces indicates that the other-race feature is developed by 3.5 months of age. Like race, emotions elicit asymmetrical attention in adults: an emotional face among neutral faces is more rapidly detected than vice versa. In Experiment 2a, 9-month-olds’ preference for a pattern containing a fearful face among neutral faces over a pattern containing a neutral face among fearful faces was greater than their preference for all neutral over all fearful faces. Thus, 9-month-olds exhibited an asymmetry in the processing of emotions. Moreover, this asymmetry was not affected by the race of the faces depicting the emotion. In Experiment 2B, 3.5-month-olds failed to exhibit a preference when tested with the same procedure. Overall, the data suggest that other-race information is processed as a feature by 3.5- and 9-month-olds, which indicates that infants process other-race information in a different, perhaps categorical, manner than own-race information. Also, other-race information does not disrupt emotion processing by 9-month-olds, which suggests that emotion and race information are processed separately in infancy. Finally, the current results indicate that adult-like asymmetrical attention to emotion develops between 3.5 and 9 months of age.

L'évaluation de la sonie globale d'un son non-stationnaire repose sur la transformation de nos sensations de sonie momentanées en une valeur scalaire unique reflétant notre impression globale a posteriori. Cette thèse approfondit la compréhension de ce processus dynamique dans le cas particulier de stimuli présentant des profils d'intensité croissants ou décroissants de quelques secondes. Partant de l'observation que les sons croissants sont perçus plus forts que les sons décroissants alors qu'ils contiennent la même énergie pendant la même durée, cette thèse aborde la question des origines de cette asymétrie perceptive sous différentes perspectives. Une approche systématique est adoptée afin d'évaluer les bases méthodologiques, contextuelles et physiques du phénomène. Il est montré que l'évaluation globale de la sonie d'un stimulus dynamique ne correspond pas une simple moyenne énergétique. Les jugements des auditeurs se fondent sur une intégration temporelle de la région la plus intense des stimuli, mais d'autres mécanismes sont impliqués; des asymétries dépendant de façon complexe et non-linéaire des caractéristiques du profil temporel et du niveau des stimuli sont observées. Les modèles de sonie actuels ne permettent pas directement d'expliquer ces résultats, ce qui soutient l'idée selon laquelle les mécanismes étudiés pourraient provenir de traitements auditifs ou décisionnels "haut-niveau" n'étant pas encore pris en considération dans ces modèles. De nouvelles perspectives de recherche pour l'étude du codage et du traitement de stimuli non-stationnaires par les systèmes sensoriels à partir d'asymétries perceptives avérées sont ainsi offertes. / Global loudness evaluation of a time-varying sound is based on the transformation of our momentary loudness sensations into a single scalar value that reflects our overall impression a posteriori. This thesis deepens the understanding of this dynamical process in the case of specific stimuli that rise or fall in level over a few seconds. Starting from the observation that rising tones are perceived louder than falling tones while both contain the same energy over the same duration, this thesis scrutinizes the origins of this perceptual asymmetry from various perspectives. A systematic approach is adopted to evaluate the methodological, contextual and physical bases of this phenomenon. It is demonstrated that global loudness evaluation of a dynamic stimulus does not correspond to a simple energetic mean. Listeners’ judgments rely on a temporal integration of the loudest region of the stimuli, but additional mechanisms are involved; asymmetries which depend on temporal profile characteristics and sound level in a complex and non-linear way are observed. Current loudness models cannot directly account for the effects highlighted throughout the different studies of this thesis, which supports the idea that the mechanism(s) under study might take place in high auditory or decisional stages not considered in these models. This research offers new perspectives to investigate the coding and the processing of time-varying stimuli by sensory systems from identified perceptual asymmetries.