How sleep deprivation degrades task performance: combining experimental analysis with simulations of adenosinergic effects of basal ganglia and cortical circuits
St. Hilaire, Melissa April
MetadataShow full item record
Humans configure themselves into "neural machines" to perform optimally on distinct tasks, and they excel at maintaining such configurations for brief episodes. The neural configuration needed for peak performance, however, is subject to perturbations on multiple time scales. This thesis reports new empirical analyses and computational modeling to advance understanding of the variations in reaction time (RT) on simple RT tasks that are associated with the duration of the preceding inter-stimulus interval (order of seconds); the time-on-task duration (order of minutes); and sleep deprivation duration (order of hours to days). Responses from the psychomotor vigilance task (PVT), including anticipations (false alarms), normal RTs, and very long RTs (lapses in attention), were analyzed to discover the effects of: the 1 - 9 second inter-stimulus interval (ISI); the 10-minute task session; up to 50 hours of sleep deprivation (SD); and wake-promoting agents, caffeine and modafinil. Normal RTs and lapses in attention were negatively correlated with ISI length, whereas anticipations were positively correlated. Anticipations, normal RTs, and lapses increased as time-on-task increased, and during SD. Both caffeine and modafinil reduced lapses and anticipations during SD and decreased RT variability. A simple neural network model incorporating both a time-dependent inhibitory process and a time-dependent excitatory process was developed. The model robustly simulated the ISI effect on behavior. The SD effects were reproducible with two parameter adjustments. Informed modeling of drug effects required greater neurobiological detail. In the basal ganglia (BG), adenosine accumulation during SD has two notable effects: it antagonizes dopamine to reduce BG responsiveness to incoming cortical signals, and it reduces cholinergic transmission to parietal and prefrontal cortices, thus reducing attention to visual signals. A detailed computational model of interactions between BG and cortex during PVT was developed to simulate effects of adenosine and their amelioration by caffeine. The model simulates drug, ISI and SD effects on anticipations, RTs, and lapses. This model can be used to describe the effects of SD over a wide range of tasks requiring planned and reactive movements, and can predict and model effects of pharmacological agents acting on the adenosinergic, cholinergic and dopaminergic systems.
Thesis (Ph.D.)--Boston University