The human sense of touch integrates feedback from a multitude of touch receptors, but how this information is represented in the neural responses such that it can be extracted quickly and reliably is still largely an open question. At the same time, dexterous robots equipped with touch sensors are becoming more common, necessitating better methods for representing sequentially updated information and new control strategies that aid in extracting relevant features for object manipulation from the data. This thesis uses information theoretic methods for two main aims: First, the neural code for tactile processing in humans is analyzed with respect to how much information is transmitted about tactile features. Second, machine learning approaches are used in order to influence both what data is gathered by a robot and how it is represented by maximizing information theoretic quantities. The first part of this thesis contains an information theoretic analysis of data recorded from primary tactile neurons in the human peripheral somatosensory system. We examine the differences in information content of two coding schemes, namely spike timing and spike counts, along with their spatial and temporal characteristics. It is found that estimates of the neurons’ information content based on the precise timing of spikes are considerably larger than for spikes counts. Moreover, the information estimated based on the timing of the very first elicited spike is at least as high as that provided by spike counts, but in many cases considerably higher. This suggests that first spike latencies can serve as a powerful mechanism to transmit information quickly. However, in natural object manipulation tasks, different tactile impressions follow each other quickly, so we asked whether the hysteretic properties of the human fingertip affect neural responses and information transmission. We find that past stimuli affect both the precise timing of spikes and spike counts of peripheral tactile neurons, resulting in increased neural noise and decreased information about ongoing stimuli. Interestingly, the first spike latencies of a subset of afferents convey information primarily about past stimulation, hinting at a mechanism to resolve ambiguity resulting from mechanical skin properties. The second part of this thesis focuses on using machine learning approaches in a robotics context in order to influence both what data is gathered and how it is represented by maximizing information theoretic quantities. During robotic object manipulation, often not all relevant object features are known, but have to be acquired from sensor data. Touch is an inherently active process and the question arises of how to best control the robot’s movements so as to maximize incoming information about the features of interest. To this end, we develop a framework that uses active learning to help with the sequential gathering of data samples by finding highly informative actions. The viability of this approach is demonstrated on a robotic hand-arm setup, where the task involves shaking bottles of different liquids in order to determine the liquid’s viscosity from tactile feedback only. The shaking frequency and the rotation angle of shaking are optimized online. Additionally, we consider the problem of how to better represent complex probability distributions that are sequentially updated, as approaches for minimizing uncertainty depend on an accurate representation of that uncertainty. A mixture of Gaussians representation is proposed and optimized using a deterministic sampling approach. We show how our method improves on similar approaches and demonstrate its usefulness in active learning scenarios. The results presented in this thesis highlight how information theory can provide a principled approach for both investigating how much information is contained in sensory data and suggesting ways for optimization, either by using better representations or actively influencing the environment.

<p>Touch plays a key role in facilitating social communication and is often presented in conjunction with auditory stimuli such as speech. Individuals with autism spectrum disorder (ASD) frequently show atypical behavioral responsivity to both tactile and auditory stimuli, which is associated with increased ASD symptomatology. However, as discussed throughout Chapter 1, the neural mechanisms associated with responsivity to tactile and auditory stimuli in ASD are not fully understood. For example, some have argued that differences in responding to tactile and auditory stimuli may be attributed to sensory and perceptual factors, whereas others suggest that these differences could be related to atypicalities in allocation of attention to incoming stimuli. In Chapter 2, I address these competing hypotheses by examining early and late ERP components (indicative of perceptual and attentional processing respectively) in response to tactile and auditory stimuli. Next, despite the evidence suggesting that touch plays a role in modulating attention in typical development (TD), it is unclear whether touch cues affect the response of the phasic alerting network – a subcomponent of attention – in ASD and TD, and whether the alerting response may be atypical in children with ASD. In Chapter 3, I address this gap in the literature by examining whether tactile cues presented at different intervals before auditory targets facilitate reaction times differently in children with ASD and TD. Lastly, because prior research has shown associations between sensory and attentional processes and ASD symptomatology, in Chapters 2 and 3, I examine the associations of neural and behavioral indices of tactile and auditory processing with ASD symptomatology and language skills in children with ASD and TD. </p><p>In Chapter 2, I show that children in both the ASD and TD groups do not exhibit differences in both early and later neurological responses to tactile and auditory stimuli, suggesting that under certain experimentally-controlled conditions, behavioral differences to tactile and auditory stimuli may not be attributable to atypicalities in perceiving or attending to the incoming sensory input. However, neural responsivity to tactile and auditory stimuli is linked with sensory responsivity and social skills in all children. Specifically, reduced early contralateral activation to tactile stimuli is related to increased tactile symptoms, and reduced early amplitudes to auditory oddball stimuli are associated with impairments in reciprocal social communication in children with ASD as well as when examined across all children, and greater tendency of overall sensory hyper-reactivity. Additionally, in the TD group, greater later amplitudes to touch and auditory oddball stimuli are related to differences in reciprocal social communication and sensory reactivity respectively, indicating that patterns of allocation of attention may be related to ASD-like traits in typical development. Lastly, there is an association between greater sensitivity to changes to a stream of auditory stimuli and expressive language skills in all children. These results suggest that, although there are no group differences between neurological responses to tactile and auditory stimuli in ASD and TD, individual neural differences may be related to sensory and socio-communicative skills in all children. </p><p>In Chapter 3, I show that although children with ASD responded more slowly than children with TD, both groups displayed faster reaction times as a result of tactile cues before auditory targets, suggesting equivalent phasic alerting in response to tactile stimuli. Longer intervals between cues and targets benefitted children in both groups resulting in faster reaction times. Contrary to my hypotheses, touch-related behavioral facilitation was not associated with ASD symptomatology and language skills. </p>Taken together, the results of these studies suggest that, at least in certain contexts and with certain cues, children with ASD may show typical neurological processing in response to tactile and auditory stimuli, and that touch may facilitate the response of the alerting network similarly in ASD and TD. Therefore, everyday behavioral differences in response to tactile and auditory stimuli may be related to the specific nature of the stimuli as well as social contexts in which such stimuli are more likely to be encountered. Differences between processing rich and dynamic sensory stimuli experienced in the outside world vs experimentally-controlled sensory stimuli presented in the laboratory settings are discussed in Chapter 4. Additionally, I argue that individual responses expected in social vs non-social experimental settings may affect neural and behavioral responses in individuals with ASD. Finally, future research directions are discussed.

Cette étude explore les réponses évoquées, l'activité intrinsèque et spontanée de deux populations neuronales différentes dans la région du cerveau correspondant à la patte arrière des souris. Dans cet article, nous nous sommes concentrés sur un modèle murin du syndrome de l'X fragile (SXF), qui est la forme la plus commune de syndrome de retard mental héréditaire et une cause fréquente de troubles du spectre autistique (TSA). SXF est un trouble à gène unique (Fmr1), qui peut être modélisé de manière fiable par un modèle murin transgénique : la souris Fmr1-/y déficiente pour le gène codant Fmr1. L'hyperexcitabilité des réseaux néocorticaux et l'hypersensibilité aux stimuli sensoriels sont des caractéristiques importantes du SXF et des TSA.Ceci est directement lié à un changement du nombre de synapses locales, de canaux ioniques, de l'excitabilité membranaire et de la connectivité des circuits de cellules individuelles. Précédemment, nous avons identifié un défaut dans les canaux ioniques, comme pouvant contribuer à ces phénotypes. Nous avons testé cette hypothèse comme un mécanisme contribuant aux défauts de traitement sensoriel chez les souris Fmr1-/y. Le cortex somatosensoriel primaire de la souris (S1) traite différentes informations sensorielles et constitue la plus grande zone du néocortex, soulignant l'importance de la modalité sensorielle pour le comportement des rongeurs. Nos connaissances concernant le traitement de l'information dans S1 proviennent d'études du cortex en tonneaux lié aux moustaches, mais le traitement des entrées sensorielles des pattes postérieures est mal compris. Par l’utilisation de la technique d’enregistrement de cellule entière par patch clamp in vivo, nous avons classes les cellules en répondeurs supraliminaires (cellules qui répondaient aux stimulations de la patte arrière avec un potentiel d'action), les répondeurs subliminaires (les cellules qui répondaient sans déclencher un potentiel d'action) et les cellules non répondeuses qui ne présentaient aucune réponse. Puis, nous avons comparé les réponses évoquées sub et supraliminaires, les propriétés intrinsèques et l’activité spontanée des neurones pyramidaux de la couche 2/3 (L2/3) de la region S1 de la patte arrière (S1-HP) d’animaux anesthésiés sauvage (WT) et Fmr1-/y. Nous avons identifié des altérations de réponse spontanée, intrinsèque et évoquée chez les souris Fmr1-/y. L’application d’un ouvreur de canaux ioniques BKCa a restauré certaines de ces propriétés altérées chez les souris Fmr1-/y / This study explores the evoked responses, intrinsic and spontaneous activity of two different neuronal populations in the hind paw region of the primary somatosensory cortex (S1) of mice. Initially, we explored information processing in these neurons under normal physiological conditions, and subsequently in a mouse model of Fragile X Syndrome (FXS). FXS is the most common form of inherited mental retardation syndrome and a frequent cause of autism spectrum disorders (ASD). FXS is a single gene (Fmr1) disorder, which can be reliably modeled by a mutant mouse model, the Fmr1 knockout (Fmr1-/y) mouse. Hyperexcitability of neocortical networks and hypersensibility to sensory stimuli are prominent features of FXS and ASD. We previously established a strong causal link between a channelopathy, hyperexcitability of neurons in the primary sensory region of the neocortex and sensory hypersensitivity in this mouse model. In the current study, we extended these findings, by conducting a detailed exploration of the processing of tactile sensory information (evoked by hind paw stimulation) in the neocortex of these mice.Most of our knowledge regarding information processing in S1 comes from studies of the whisker-related barrel cortex (which processes tactile-related sensory information derived from the whiskers), yet the processing of sensory inputs from the hind-paws is poorly understood. Using in vivo whole-cell patch-clamp recordings, we classified the cells into suprathreshold responders (the cells which responded to the hind-paw stimulations with an action potential), subthreshold responders (the cells responded without eliciting an action potential) and non-responder cells (neurons which did not show any response). We then compared the evoked sub- and supra-threshold responses, intrinsic properties, and spontaneous activity of layer (L) 2/3 pyramidal neurons of the S1 hind-paw (S1-HP) region of anaesthetized wild type (WT) and Fmr1-/y mice. We identified spontaneous, intrinsic and evoked response alterations in Fmr1-/y mice. We probed possible mechanisms contributing to this sensory impairment in Fmr1-/y mice. Finally, we tested the possibility of correcting pathophysiological alterations in these neurons using specific pharmacological agents targeting the ion channel defects described previously by our team.