No description available.

No description available.

The purpose of the present research was threefold: 1) to investigate whether natural aging affects the movements to remembered targets when participants make reaching movements under closed-loop feedback conditions, 2) to determine if experience with visually-guided movements facilitates memory-guided reaching and 3) to determine if age affects this facilitation. Two groups of 10 participants (healthy older and healthy younger) performed a manual aiming task with a mouse on a graphics tablet. A target appeared in one of 6 possible locations on a computer screen and participants had to make aiming movements with a visible cursor in 3 different visual conditions: full vision, immediate recall and delayed recall. In the full vision condition vision of the target was available throughout the movement. In the delay conditions the target disappeared either at the initiation of the aiming movement (immediate recall) or 2 seconds before movement onset (delayed recall). Vision of the hand (cursor) was available in all conditions. Each memory condition was divided into 2 blocks; block 1 was presented before the full vision condition and block 2 was presented after. Endpoint accuracy and variability were measured along with movement kinematics. Results showed no age differences in the kinematics in the full vision condition. For memory dependent pointing age also did not affect the movement kinematics or endpoint accuracy. Movements to remembered targets were significantly more variable in the delay recall compared to the immediate recall condition. A Block by Condition effect showed that the delay effect was present in the first block, but not in the second block, suggesting that variability did not increase with memory delay once participants had experience from full vision reaching. A Group by Condition effect showed the older adults were more variable than younger, although this difference was smaller in the delay condition due to the increase in variability as a function of delay seen in younger but not older adults. These findings suggest that aging does not affect how movements are controlled whether pointing to visible or remembered targets. They also suggest that aging does not affect the accuracy in pointing to remembered targets. Aging does affect the variability of these pointing movements. Finally, experience in pointing at targets with full vision modulates the increase in variability of pointing as a function of delay.

The purpose of the present research was threefold: 1) to investigate whether natural aging affects the movements to remembered targets when participants make reaching movements under closed-loop feedback conditions, 2) to determine if experience with visually-guided movements facilitates memory-guided reaching and 3) to determine if age affects this facilitation. Two groups of 10 participants (healthy older and healthy younger) performed a manual aiming task with a mouse on a graphics tablet. A target appeared in one of 6 possible locations on a computer screen and participants had to make aiming movements with a visible cursor in 3 different visual conditions: full vision, immediate recall and delayed recall. In the full vision condition vision of the target was available throughout the movement. In the delay conditions the target disappeared either at the initiation of the aiming movement (immediate recall) or 2 seconds before movement onset (delayed recall). Vision of the hand (cursor) was available in all conditions. Each memory condition was divided into 2 blocks; block 1 was presented before the full vision condition and block 2 was presented after. Endpoint accuracy and variability were measured along with movement kinematics. Results showed no age differences in the kinematics in the full vision condition. For memory dependent pointing age also did not affect the movement kinematics or endpoint accuracy. Movements to remembered targets were significantly more variable in the delay recall compared to the immediate recall condition. A Block by Condition effect showed that the delay effect was present in the first block, but not in the second block, suggesting that variability did not increase with memory delay once participants had experience from full vision reaching. A Group by Condition effect showed the older adults were more variable than younger, although this difference was smaller in the delay condition due to the increase in variability as a function of delay seen in younger but not older adults. These findings suggest that aging does not affect how movements are controlled whether pointing to visible or remembered targets. They also suggest that aging does not affect the accuracy in pointing to remembered targets. Aging does affect the variability of these pointing movements. Finally, experience in pointing at targets with full vision modulates the increase in variability of pointing as a function of delay.

It has been consistently shown in the literature that when gaze is directed to a specific location, for a long enough duration, at the correct time relative to the motor execution, high-levels of performance are possible. In recent years, a particular gaze called quiet eye (QE) has gained growing attention among researchers investigating aiming tasks and has become accepted within the literature as a measure of optimal attentional control. Previous studies consistently displayed that longer QE is associated with superior performance however there is lack of understanding how QE exerts its positive effect on performance. Therefore the overriding aim of the current program of research was to explore the mechanisms behind the QE by experimentally manipulating the separate aspects of the QE definition in ways that have not been explored by previous researchers. In study 1 (Chapter 2), two experiments were conducted to examine the key characteristics of the QE in golf putting; duration and location. Novice participants were randomly allocated to training groups of experimentally longer or shorter QE durations (experiment 1) and training groups of different QE locations (experiment 2). A retention-pressure-retention design was adopted, and measures of performance and QE were recorded. All groups improved performance after training and the levels of performance achieved were robust in a pressure test. However there were no significant group effects. Study 1 provided partial support for the efficacy of QE training, but did not clarify how the QE itself underpins the performance advantage revealed in earlier studies and suggested that the QE should perhaps not be reported simply as a function of its duration or its location. In study 2 (Chapter 3) an examination of the timing of the QE was performed, using an occlusion paradigm. This provided an experimental manipulation of the availability of visual information during the putting action. Expert participants performed a putting task under three different conditions, namely full, early, and late vision conditions. The results showed that putting accuracy was the poorest when late visual information was occluded (early vision condition). Therefore study 2 suggested that the correct temporal placement of gaze might be more crucial to successful performance, and that putting accuracy was poorer when the latter component of QE which ensures precise control of movement was occluded. Previous research has revealed that anxiety can attenuate the QE duration, shortening the latter component which was shown to be important in study 2. Therefore the final study in the thesis examined the influence of anxiety on attentional control (QE). Expert golfers participated in a putting shootout competition designed to increase levels of anxiety and continued putting until a missed putt occurred. The results revealed that duration of QE was shorter on the missed putt, while there was no difference in QE duration for successful putts (first and penultimate putts). The results are therefore supportive of the predictions of attentional control theory. Furthermore this reduction in QE duration was result of latter component of QE being attenuated, supporting models of motor control that point to the importance of online visual information for regulating control of movements. The results of this series of studies conclude that the timing of the QE – maintaining a steady fixation through the unfolding movement to ensure precise online control - seems to be the strong candidate for how QE exerts a positive effect on performance.

No description available.

Football penalty kicks are having increasing influence in today’s professional game. Despite this, little scientific evidence currently exists to ascertain the mechanisms behind performance failure in this task and/or the efficacy of training designed to improve penalty shooting. In a football penalty kick it has been reported that the majority of kickers do not look to the area they wish to place the ball; preferring to focus on the ‘keeper and predict anticipatory movements before shooting. Such a strategy seems counterproductive and contradictory to current research findings regarding visually guided aiming. Coordination of eye and limb movements has been shown to be essential for the production of accurate motor responses. A disruption to this coordination not only seems to negatively affect performance, but subsequent motor responses seem to follow direction of gaze. Thus, where the eyes lead actions tend to follow. In study 1, ten participants were asked to kick a standard sized football to alternate corners of a goal, whilst looking centrally and whilst looking where they intended to hit. This disruption of eye-limb coordination brought about a 15% reduction in kicking accuracy. When participants were asked to fixate centrally, their shots hit more centrally (17cm) than when they were allowed to look where they intended to hit. These results were in spite of no significant differences between the number of missed shots, preparation time and ball speed data across conditions. We concluded that centrally focused fixations dragged resultant motor actions inwards towards more central target locations. Put simply, where the eyes looked shots tended to follow. The second study sought to test the predictions of attentional control theory (ACT) in a sporting environment in order to establish how anxiety affects performance in penalty kicks. Fourteen experienced footballers took penalty kicks under low- and high-threat counterbalanced conditions while wearing a gaze registration system. Fixations to target locations (goalkeeper and goal area) were determined using frame-by-frame analysis. When anxious, footballers made faster first fixations and fixated for significantly longer toward the goalkeeper. This disruption in gaze behaviour brought about significant reductions in shooting accuracy, with shots becoming significantly centralized and within the goalkeeper’s reach. These findings support the predictions of ACT, as anxious participants were more likely to focus on the “threatening” goalkeeper, owing to an increased influence of the stimulus-driven attentional control system. A further prediction of ACT is that when anxious, performers are more likely to be distracted, particularly if the distracter is threat related. When facing penalty kicks in football (soccer), goalkeepers frequently incorporate strategies that are designed to distract the kicker. However, no direct empirical evidence exists to ascertain what effect such visual distractions have on the attentional control, and performance, of footballers. In the third study, eighteen experienced footballers took five penalty kicks under counterbalanced conditions of threat (low vs. high) and goalkeeper movement (stationary vs. waving arms) while wearing eye-tracking equipment. Results suggested that participants were more distracted by a moving goalkeeper than a stationary one and struggled to disengage from a moving goalkeeper under situations of high threat. Significantly more penalties were saved on trials when the goalkeeper was moving and shots were also generally hit closer to the goalkeeper (centrally) on these trials. The results provide partial support for the predictions of attentional control theory and implications for kickers and goalkeepers are discussed. The previous studies showed that anxiety can disrupt visual attention, visuomotor control and subsequent shot location in penalty kicks. However, optimal visual attention has been trained in other far aiming skills, improving performance and resistance to pressure. In study 4, we therefore asked a team of ten university soccer players to follow a quiet eye (QE; Vickers, 1996) training program, designed to align gaze with aiming intention to optimal scoring zones, over a seven week period. Performance and gaze parameters were compared to a placebo group (ten players) who received no instruction, but practiced the same number of penalty kicks over the same time frame. Results from a retention test indicated that the QE trained group had more effective visual attentional control; were significantly more accurate; and had 50% fewer shots saved by the goalkeeper than the placebo group. Both groups then competed in a penalty shootout to explore the influence of anxiety on attentional control and shooting accuracy. Under the pressure of the shootout the QE trained group failed to maintain their accuracy advantage, despite maintaining more distal aiming fixations of longer duration. The results therefore provide only partial support for the effectiveness of brief QE training interventions for experienced performers. This series of studies are the first to explore the gaze behaviour of football penalty takers in a quest to uncover and understand anxiety’s negative influence on attentional control and performance. They are also the first to explore the efficacy of goalkeeper distractions and training in improving performance from both the goalkeeper’s and kicker’s perspective. The results of these studies conclude that when anxious, penalty takers show an attentional bias toward the ‘threatening’ goalkeeper that can be increased and utilised by a goalkeeper employing distraction techniques and that penalty takers do benefit, to some extent, from a gaze-based pre-shot routine

La locomotion est une composante fondamentale de la vie animale : elle permet l’accès continu aux ressources nécessaires à la survie ainsi que l’évitement de périls variés. Les milieux naturels comme anthropiques regorgent toutefois d’obstacles s’élevant contre notre progression. Pour l’humain et les autres mammifères terrestres naviguant principalement par la vision, le franchissement efficace de ces obstacles repose critiquement sur la capacité de modifier proactivement le positionnement et la trajectoire des pas en fonction des informations visuelles extraites durant leur approche. Au niveau du système nerveux, cette capacité implique un processus complexe où le traitement des signaux visuels reflétant les paramètres de l’obstacle spécifie un cours d’action sécurisant son franchissement, lequel est ultimement exécuté par des altérations précises à l’activité musculaire. Des études approfondies chez le chat, l’un des modèles animaux les plus développés et investigués vis-à-vis du contrôle locomoteur, ont présentement impliqué deux structures corticales dans ce processus. Le cortex pariétal postérieur contribuerait ainsi à déterminer la position relative de l’obstacle et le cortex moteur primaire serait central à l’exécution des modifications de la démarche. Cependant, notre compréhension du substrat neural impliqué dans la transformation sensorimotrice joignant ces deux étapes est extrêmement limitée. Plusieurs lignes d’évidences, particulièrement dérivées de travaux chez le primate investiguant le contrôle des mouvements volontaires du bras, pointent cependant vers une contribution potentiellement majeure du cortex prémoteur à cette fonction. Cette thèse entreprend de déterminer directement la contribution prémotrice aux modifications de la démarche. Deux études rapportent ainsi l’activité de neurones individuels enregistrés dans deux larges subdivisions du cortex prémoteur, les aires 6iffu et 4delta, chez le chat éveillé accomplissant librement une tâche de négociation d’obstacles sur tapis roulant. Ces études font état de changements d’activité distincts d’une subdivision à l’autre et corrélés à des aspects spécifiques de la tâche, incluant des changements préparatoires liés à l’approche finale de l’obstacle et d’autres liés à une ou plusieurs étapes des ajustements locomoteurs séquentiels entourant sa négociation. Une troisième étude investigue par microstimulation intracorticale la capacité des différentes subdivisions prémotrices du chat à modifier la démarche. Cette étude expose une variété de réponses électromyographiques complexes s’intégrant en phase avec la marche, où plusieurs subdivisions présentent des signatures distinctes d’effets multi-membres contrastant avec l’influence focale du cortex moteur primaire. Chacune de ces trois études est finalement complémentée d’investigations par traçage rétrograde de connexions anatomiques décisives à l’interprétation fonctionnelle des subdivisions investiguées. Ensemble, ces travaux soutiennent et précisent une contribution centrale du cortex prémoteur aux modifications de la démarche sous guidage visuel. D’une part, ils rapportent pour la première fois que l’activité neuronale de multiples subdivisions du cortex prémoteur reflète différentes étapes de la planification locomotrice stipulant les altérations à entreprendre à l’approche d’un obstacle et durant son franchissement. D’autre part, ils révèlent complémentairement que l’activation de ces subdivisions a le pouvoir d’influencer profondément la marche. Les données collectées soulignent finalement plusieurs points de comparaison entre les aires prémotrices du chat et du primate, suggérant un degré d’analogie fonctionnelle extensible à la locomotion humaine. / Locomotion is a fundamental component of animal life: it provides continuous access to the resources necessary for survival as well as the means to elude potential perils. However, both natural and built environments teem with obstacles impeding one’s progress. For humans and other terrestrial mammals navigating primarily through vision, efficiently negotiating these obstacles critically requires the capacity to proactively adapt the positioning and trajectory of each step on the basis of visual information extracted during their approach. In the nervous system, this capacity involves a complex process through which the integration of visual signals reflecting the parameters and location of an obstacle specifies a course of action to ensure its negotiation, Extensive studies in the cat, one of the most common models used to study the neural mechanisms involved in the control of locomotion, have currently implicated two cortical structures to this process. The posterior parietal cortex is suggested to contribute to the determination of the obstacle’s relative position (with respect to the body) while the primary motor cortex is central to the execution of the gait modifications. However, our comprehension of the neural substrate implicated in the sensorimotor transformation linking these defined stages is extremely limited. Several lines of evidence, predominantly derived from work in the primate investigating the voluntary control of arm movements, nonetheless point towards a potentially major contribution of the premotor cortex to this function. This thesis sets out to directly determine the premotor contribution to the control of gait modifications. Two studies report the activity of individual neurons recorded in two large subdivisions of premotor cortex, areas 6iffu and 4delta, in awake cats freely performing an obstacle negotiation task on treadmill. These studies describe distinct changes in activity across subdivisions that correlate with specific aspects of the task, including preparatory changes related to the final approach of the obstacle and others related to one or more stages of the sequential locomotor adjustments surrounding its negotiation. A third study used intracortical microstimulation to investigate the capacity of different premotor subdivisions of the cat to modify gait. This study reveals a variety of complex electromyographic responses that are integrated into the gait cycle. Moreover, several subdivisions show distinct signatures of multi-limb effects that contrast with the focal influence of the primary motor cortex. Each of these three studies is finally complemented by retrograde tracing investigations of anatomical connections critical to the functional interpretation of the subdivisions examined. Together, these studies support and clarify a central contribution of the premotor cortex to the modification of gait under visual guidance. We report for the first time that the neural activity of multiple subdivisions of the premotor cortex reflects different stages of the locomotor plan specifying the gait alterations to perform during the approach and crossing of an obstacle. In addition, we reveal that activation of these subdivisions has the power to profoundly influence walking. The data collected finally highlight several points of comparison between the premotor areas of the cat and the primate, suggesting a degree of functional analogy extensible to human locomotion.

L'intelligence artificielle (IA) moderne a ouvert de nouvelles perspectives prometteuses pour la création de robots intelligents. En particulier, les architectures d'apprentissage basées sur le gradient (réseaux neuronaux profonds) ont considérablement amélioré la compréhension des scènes 3D en termes de perception, de raisonnement et d'action. Cependant, ces progrès ont affaibli l'attrait de nombreuses techniques ``classiques'' développées au cours des dernières décennies. Nous postulons qu'un mélange de méthodes ``classiques'' et ``apprises'' est la voie la plus prometteuse pour développer des modèles du monde flexibles, interprétables et exploitables : une nécessité pour les agents intelligents incorporés. La question centrale de cette thèse est : ``Quelle est la manière idéale de combiner les techniques classiques avec des architectures d'apprentissage basées sur le gradient pour une compréhension riche du monde 3D ?''. Cette vision ouvre la voie à une multitude d'applications qui ont un impact fondamental sur la façon dont les agents physiques perçoivent et interagissent avec leur environnement. Cette thèse, appelée ``programmes différentiables pour modèler l'environnement'', unifie les efforts de plusieurs domaines étroitement liés mais actuellement disjoints, notamment la robotique, la vision par ordinateur, l'infographie et l'IA. Ma première contribution---gradSLAM--- est un système de localisation et de cartographie simultanées (SLAM) dense et entièrement différentiable. En permettant le calcul du gradient à travers des composants autrement non différentiables tels que l'optimisation non linéaire par moindres carrés, le raycasting, l'odométrie visuelle et la cartographie dense, gradSLAM ouvre de nouvelles voies pour intégrer la reconstruction 3D classique et l'apprentissage profond. Ma deuxième contribution - taskography - propose une sparsification conditionnée par la tâche de grandes scènes 3D encodées sous forme de graphes de scènes 3D. Cela permet aux planificateurs classiques d'égaler (et de surpasser) les planificateurs de pointe basés sur l'apprentissage en concentrant le calcul sur les attributs de la scène pertinents pour la tâche. Ma troisième et dernière contribution---gradSim--- est un simulateur entièrement différentiable qui combine des moteurs physiques et graphiques différentiables pour permettre l'estimation des paramètres physiques et le contrôle visuomoteur, uniquement à partir de vidéos ou d'une image fixe. / Modern artificial intelligence (AI) has created exciting new opportunities for building intelligent robots. In particular, gradient-based learning architectures (deep neural networks) have tremendously improved 3D scene understanding in terms of perception, reasoning, and action. However, these advancements have undermined many ``classical'' techniques developed over the last few decades. We postulate that a blend of ``classical'' and ``learned'' methods is the most promising path to developing flexible, interpretable, and actionable models of the world: a necessity for intelligent embodied agents. ``What is the ideal way to combine classical techniques with gradient-based learning architectures for a rich understanding of the 3D world?'' is the central question in this dissertation. This understanding enables a multitude of applications that fundamentally impact how embodied agents perceive and interact with their environment. This dissertation, dubbed ``differentiable world programs'', unifies efforts from multiple closely-related but currently-disjoint fields including robotics, computer vision, computer graphics, and AI. Our first contribution---gradSLAM---is a fully differentiable dense simultaneous localization and mapping (SLAM) system. By enabling gradient computation through otherwise non-differentiable components such as nonlinear least squares optimization, ray casting, visual odometry, and dense mapping, gradSLAM opens up new avenues for integrating classical 3D reconstruction and deep learning. Our second contribution---taskography---proposes a task-conditioned sparsification of large 3D scenes encoded as 3D scene graphs. This enables classical planners to match (and surpass) state-of-the-art learning-based planners by focusing computation on task-relevant scene attributes. Our third and final contribution---gradSim---is a fully differentiable simulator that composes differentiable physics and graphics engines to enable physical parameter estimation and visuomotor control, solely from videos or a still image.