Sensorimotor learning in complex and uncertain environments

Humans have the remarkable capacity to generate accurate motor behaviors under many different complex and often uncertain environmental conditions. There is a rich history of studies investigating how the various components of the motor system deal with these sources of uncertainty leading to optimal feedback control models of a wide range of motor behaviors such as bimanual movements, object manipulation, adaptation to perturbations, and the generalization of learned skill to novel contexts. In this project we seek to shed further light on the nature of sensorimotor learning in the context of complex and uncertain environments, and to deepen our understanding of modelling approaches in sensorimotor neuroscience. We employ visuomotor adaptation paradigms that track the rate of adaptation to a visual perturbation, the degree to which noisy visual feedback is integrated during a goal directed movement, and the degree to which adaptation generalizes to novel contexts. To comprehensively interpret our adaptation and generalization results, we fit our data to a variety of state-space models that predict the rate of adaptation, integration, and generalization within complex and uncertain sensorimotor learning contexts. In the first half of the project, we investigate the process of integrating uncertain sensory information in the context of visuomotor adaptation. Both the degree to which sensory feedback is integrated into an ongoing movement and the degree to which movement errors drive adaptive changes in feedforward motor plans have been shown to scale with the uncertainty of the sensory signal. However, little is known about whether feedback and feedforward processes interact when dealing with sensory uncertainty. We report the results from several experiments which indicate that sensory uncertainty can modulate one process without having any effect on the other. Computational modelling reveals that sensory uncertainty may influence feedback integration and feedforward adaptation differently and independently, in a manner that depends on the magnitude of movement errors. They also reveal a fundamental shift in how sensory uncertainty scales feedforward adaptation in contexts where feedback and feedforward processes co-occur. In the second half of the project we shift focus toward the capacity to generate highly skilled motor behaviors within complex environmental conditions. It remains unknown whether group differences in visuomotor adaption and generalization are present in expert cohorts that must routinely cope with and learn new visuomotor mappings. Here we prioritize an applied approach in order to gain a perspective on how key motor learning processes differ between populations of skilled motor experts and novices. Our experimental cohort consists of expert minimally invasive surgeons. If MIS introduces challenges for learning appropriate visuomotor transformations, this suggests that expert surgeons, who have successfully overcome these challenges, might perform better than naive controls in a standard visuomotor adaptation. This is either because (1) they have spent much more time practicing to compensate for visuomotor perturbations, (2) they are inherently more adept at compensating for visuomotor perturbations and this is part of what makes them good surgeons in the first place, or (3) some combination of these factors. The current literature does not address the plausibility of either of these possibilities. For instance, it is unknown to what degree the basic neural and cognitive processes that drive visuomotor adaptation are capable of enhancements in learning and performance in the first place. By employing a visuomotor rotation and generalization task, our first set of experiments sought to categorise the differential ability of expert MIS surgeons from naive controls. We found that expert surgeons compensate for a visuomotor perturbation more rapidly and to a greater extent than naive controls. Modelling indicates that these differences in expert behavioural performance reflects greater trial-to-trial retention, as opposed to greater trial-to-trial learning rate. We also found that surgeons generalize to novel reach directions more broadly than controls, a result confirmed by our modelling. Our findings show that minimally invasive surgeons exhibit enhanced visuomotor learning and spatial generalization. In a follow up study, we investigated the effects of realigned visual feedback on performance in a laparoscopic task for both naive controls and experienced surgeons. We observed that realignment imposed a reliable cost on task performance across both groups. The wide and varied terrain traversed in this project demonstrate how crucial, carefully designed behavioural experiments are in our efforts to understand how the brain supports complex, adaptive behaviour. We describe several behavioural experiments that help to provide a much more finegrained delineation of how sensory uncertainty influences feedback integration and feedforward adaptation in the hope that these efforts will lay the foundation for future scientific work to reveal how these processes are implemented in the brain.