Il existe aujourd’hui un large consensus sur le fait que le système cognitif est capabled’avoir des activités sur lui-même, on parle de métacognition. Si plusieurs travaux se sontintéressés aux mécanismes qui sous-tendent cette métacognition, à notre connaissance,aucun ne l’a fait dans une perspective « sensorimotrice et intégrative » du fonctionne-ment cognitif comme celle que nous proposons. Ainsi, la thèse que nous défendons dansce travail est la suivante : l’information métacognitive, notamment la fluence, possèdestrictement le même statut que l’information cognitive (i.e., sensorielle et motrice). Dansun premier chapitre, nous proposons un modèle de la cognition respectant ce principe.Ensuite, dans les deux chapitres suivants, nous mettons à l’épreuve notre hypothèse parle biais d’expériences et de simulations effectuées à l’aide du modèle mathématique quenous avons élaboré. Ces travaux ont porté plus précisément sur des phénomènes liés à troispossibilités originales prédites par notre hypothèse : la possibilité de méta-métacognition,la possibilité d’intégration entre information sensorielle et information métacognitive, etla possibilité d’abstraction métacognitive. / There is today a broad consensus that the cognitive system is capable of having acti-vities on itself, we are talking about metacognition. Although several studies have focusedon the mechanisms underlying this metacognition, to our knowledge, none has done so ina "sensorimotor and integrative" perspective of cognitive functioning such as the one wepropose. Thus, the thesis we defend in this work is the following : metacognitive infor-mation, especially fluency, has strictly the same status as any cognitive information (i.e.,sensory and motor). In a first chapter, we propose a model of cognition respecting thisprinciple. Then, in the next two chapters, we test our hypothesis through experimentsand simulations using the mathematical model we have developed. This work focusedmore specifically on phenomena related to three original possibilities predicted by ourhypothesis : the possibility of meta-metacognition, the possibility of integration betweensensory information and metacognitive information, and the possibility of metacognitiveabstraction.

Le travail de recherche présenté dans cette thèse considère la mémoire humaine comme un système unique et non abstractif qui reflète l’ensemble de nos expériences sous forme de traces épisodiques multimodales. Les objectifs de cette thèse sont multiples, mais le principal est de montrer qu’un effet robuste de la mémoire, l’effet de distinctivité, peut émerger aussi bien dans les tâches implicites que dans les tâches explicites et que cet effet s’expliquerait en termes de mécanismes spécifiques (activation et intégration multimodales) et non pas en termes de systèmes mnésiques sous-jacents.Trois séries d’expériences ont été élaborées. Dans une première série, nous avons manipulé l’information contextuelle extrinsèque associée à ces concepts. Une tâche de catégorisation nous a permis de démontrer que l’effet de distinctivité pouvait se manifester avec une tâche implicite de mémoire.Dans une deuxième série d’expériences, nous avons manipulé la distance entre des images à catégoriser dans une phase d’encodage. Ainsi, les images apparaissaient soit plus éloignées les unes des autres, soit plus proches. Suite à une récupération implicite, nous avons mis en évidence que notre hypothèse de distinctivité spatiale était validée, c’est-à-dire que les items spatialement plus distinct lors de l’encodage sont associés à des performances supérieures en phase test.Dans la troisième série d’expériences, nous avons mis en évidence que lesperformances liées à l’effet de distinctivité dans les tâches implicites et explicites de mémoire variaient selon différents niveaux d’intégration des dimensions sensorielles. Cette idée a été testée en rappel libre, en décision lexicale et en reconnaissance.Au final, nos résultats expérimentaux suggèrent que les performances issuesdes tâches implicites et explicites peuvent être expliquées au sein du même système mnésique unique. Ainsi, les processus d’intégration seraient à l’origine de ce phénomène. / The series of research presented in this thesis considers human memory as a single and non-abstractive system which reflects all of our experiments in an episodical multimodal traces form. Objectives of this thesis are multiple, but the main issue is to show that the distinctiveness effect, a well known phenomenon of memory, can emerge in implicit memory tasks as well as in explicit memory tasks and this effect can be explained in specific terms of mechanisms (activation and multimodal integration) and not in terms of subjacent memory systems.Three series of experiments were elaborated. In the first series, we used an extrinsic contextual information associated to concepts. A categorization task permitted us to show that distinctiveness effect can appear within an implicit memory task.In the second series of experiments, we manipulated the distance between images to be categorized in an encoding phase. Thus, images presented were either more distant to each other, or more closer. Following an implicit retrieval, we highlighted that spatial distinctiveness hypothesis was validated. This means that spatially more distinct items during the encoding phase are associated to higher retrieval performances in the test phase.In the third series of experiments, we showed that performances related to distinctiveness effect in implicit and explicit memory tasks varied according to various levels of sensory dimension integration. This postulate was tested with free recall, lexical decision and recognition tasks.Finally, our experimental results suggest that performances from implicit and explicit memory tasks can be explained within the same single memory system. Thus, the integration process would be at the origin of this phenomenon.

The cognitive processing of a stimulus does not only depend on the physical properties of the stimulus itself but also on the larger context in which it occurs. In this thesis I will present a number of studies that investigate this context-dependency at different levels of cognition. In particular these levels include (1) sensory processing within a modality, (2) sensory integration across modalities and (3) the relation between sensory perception and motor behavior. Accordingly the chapters in this thesis are partitioned into three larger parts, each of which relates to one of these levels. The first study in Part 1 investigates the role of neural oscillations during perceptual grouping. By measuring EEG during contour integration we were not only able to identify the neural sources involved in this process but also to demonstrate local and long-range synchronization of oscillatory activity within frontoparietal networks. This study is then followed by a more general discussion about the properties of oscillatory activity and how they might relate to event-related potentials. The focus of Part 2 will then be on cross-modal interactions and their possible utilization for real-life applications. First we show that simultaneously presented auditory and tactile cues lead to interactions on both a behavioral and neural level. Subsequently we demonstrate how the observed perceptual effects can be used to optimize auditory and tactile localization performance. Finally we propose a setup for utilizing tactile information to enhance the perceptual interpretation of 360° visual scenes. The third and last part of this thesis is dedicated to problems and applications of measuring EEG in the presence of eye movements. Therefore we use eye tracking to investigate and characterize EEG artifacts resulting from ocular activity. Subsequently we develop an algorithm that allows objectively and reliably identifying these artifacts and removing them from the data without affecting the signal from neural sources. Employing this algorithm we then demonstrate that combined EEG and eye tracking can be used for monitoring and shaping both the gaze behavior and the related brain activity in ASD patients. Next to studying cognition with regard to perceptual and behavioral context, this thesis also focuses on the question how the context-relevant signal components can be identified and extracted from the EEG. In the studies presented here we applied a variety of different strategies to approach this problem. These range from resorting to prior knowledge and analyzing only activity from predetermined cortical sources on the one hand, to purely data driven approaches based on logistic regression or eye tracking information on the other hand.

Cette thèse décrit l'intégration multimodale voix-visage pour la perception du genre à l'aide de méthodes comportementales et d'imagerie cérébrale. Dans une première étude psychophysique, les observateurs ont départagé des paires de stimuli voix-visage selon le genre du visage, de la voix ou du stimulus (sans instruction particulière). Une seconde étude a reproduit ce paradigme en ajoutant du bruit visuel et/ou auditif. Conformément à nos résultats théoriques, tâche et bruit peuvent tous deux être modélisés comme des facteurs de pondération. Les deux effets pourraient refléter des changements similaires de hiérarchie fonctionnelle avec la communication par cohérence comme implémentation potentielle de ce mécanisme en termes de modulation sélective de l'information par synchronisation des rythmes d'oscillation neuronaux. Une asymétrie en faveur de la modalité auditive a été trouvée dans les deux études comportementales ainsi que deux interactions : un effet multiplicatif du genre significatif lorsqu'on juge le visage et le stimulus ; un effet de cohérence significatif lorsqu'on juge le visage ou la voix. Une troisième étude en IRMf s'est intéressée aux modulations de connectivité effective entre l'aire fusiforme du visage et l'aire temporale de la voix durant la présentation de stimuli voix-visage en prêtant attention au genre du visage, de la voix ou du stimulus. Une telle modulation a été trouvée dans les tâches du visage et du stimulus en réponse au genre, et dans les tâches du visage et du stimulus en réponse à l'incohérence, deux modulations indépendantes qui pourraient être supportées par l’architecture anatomique en double contre-courant / This thesis describes face-voice multimodal gender integration using complementary behavioral and brain imaging techniques. In a first psychophysical study, observers judged pairs of face-voice stimuli according to face, voice or stimulus (no specific instruction given) gender. A second study tested the bottom-up effect of adding visual and/or auditory noise in the same paradigm. Top-down task and bottom-up noise could both be modeled as weighting effects, as predicted by our theoretical results. Both effects might reflect similar shifts in functional hierarchy. Communication through coherence offers a potential explanation for the neural basis of such a mechanism in terms of selective modulation of segregated cortical streams by oscillatory rhythm synchronization. An asymmetry in favor of the auditory modality was found in both behavioral experiments as well as two interaction effects, first a multiplicative gender effect in the face and stimulus tasks, second an effect of gender coherence in the face and voice tasks. In a third experiment we used fMRI to investigate effective connectivity modulations between the Fusiform Face Area and Temporal Voice Area during the presentation of face-voice stimuli while attending to either face, voice or any gender information. We found a change in effective connectivity for stimulus and face tasks in response to gender information, and for face and voice tasks in response to gender incoherence. These two independent modulations could be supported by the anatomical dual counterstream architecture

As nomads, humanity constantly moved and relocated for hundred thousands of years. Thereby, individuals or small groups of people had to navigate over very long distances in order to survive. As a result, successful spatial navigation was one of the key cognitive abilities, which ensured our survival. Although navigation has nowadays become less life-threatening, exploring our environment and efficiently navigating between places are still very important aspects in our everyday life. However, in order to be able to navigate efficiently, our brain has to perform a series of spatial cognitive operations. This dissertation is structured into three sections, which explore these cognitive operations from three different perspectives. In the first section I will elaborate about the role of reference frames in human spatial navigation. Specifically, in an online navigation study (study one) I will show that humans have distinct but stable reference frame proclivities. Furthermore, this study demonstrates the existence of a spatial strategy, in which the preference to use a particular reference frame is dependent on the axis of rotation (horizontal vs. vertical). In a follow-up study (study two) I will then analyze the factors underlying performance differences in navigation, as well as individual preferences using one or another spatial strategy. Interestingly, the results suggest that performance measures (reaction time and error rate) are influenced mostly by the factors gender and age. However, even more importantly, I will show that the prevalent factor, which influences the choice for an individual navigation strategy, is the cultural background of the participant. This underlines the importance of socio-economic aspects in human spatial navigation. In the second part of this thesis I will then discuss aspects of learning and memorizing spatial information. In this respect, the alignment study (study three) will show that humans are able to recall object-to-object relations (e.g. how to get from A to B) in a very brief time, indicating that such information is directly stored in memory. This supports an embodied (action-oriented) perspective of human spatial cognition. Following this approach, in the feelSpace study (study four) I will then investigate the long-term training effects with a sensory augmentation device. Most importantly, the respective results will demonstrate substantial changes in the subjective perception of space, in sleep stage architecture, and in neural oscillations during sleep. In the third and last section I will describe the importance of multimodal processes in spatial cognitive operations. Most importantly, in the platform study (study five) I will combine the topics of sensory augmentation and Bayesian cue combination. The results of this study show that untrained adult participants alternate rather than integrate between augmented and native sensory information. Interestingly, this alternation is based on a subjective evaluation of cue reliability. In summary, this thesis will present relevant and new findings for better understanding spatial strategy formation, learning and representing spatial relations in memory, and multimodal cue combination. An important and overarching aspect of this thesis is the characterization of individual differences in the context of human spatial navigation. Specifically, my research revealed individual differences in three areas: First, in utilizing egocentric or allocentric reference frames for spatial updating, second in individualized qualitative changes of space perception during long-term sensory augmentation, and third, in preferences to use native or augmented information in a cue combination task. Most importantly, I will provide a better definition and understanding of these individual differences, by combining qualitative and quantitative measures and using latest technologies such as online data recordings and interactive experimental setups. In fact, in the real world, humans are very active beings who follow individualized spatial cognitive strategies. Studying such interactive and individualized behavior will ultimately lead to more coherent and meaningful insights within the human sciences.