Frontal/Prefrontal control of cortical rhythms during auditory active sensi

我们如何从不断变化和嘈杂的环境中提取重要信息？项目三地址

How do we extract salient information from an ever-changing and noisy environment? Project 3 addresses this fundamental question in perception using direct brain recordings in humans (electrocorticography; ECoG) to assess two models of sensory acquisition. The Active Sensing model posits that high-level inputs act to rhythmically sample the sensory word and filter out noise. The related predictive coding model theory posits that prior knowledge enhances perception with the brain making predictions about upcoming stimuli to sharpen low-level sensory processing. We propose that both processes share similar neural substrates - neuronal rhythm-based engagement of frontal, premotor, motor and sensory cortical networks to enable active and predictive sampling of the world to enhance perception. We employ ECoG to measure neural oscillations and high frequency activity (HG; 70-200 Hz; surrogate for intracortical SUA activity) and employ network analysis approaches to define the role of top-down control of active sensing and predictive coding in the human brain. Two or our proposed human ECoG studies are performed in monkeys in Project 4 permitting a rich inter-species comparison of the neural substrates of sensory acquisition. AIM 1 tests the hypotheses that motor/premotor systems control auditory sampling rhythms and actively suppress distracting information. This aim also explores whether lateral prefrontal regions provide additional control to the motor/premotor-auditory active-sensing network. AIM 2 addresses how prior knowledge enhances speech perception and `fills-in' degraded speech representations in auditory cortices. Given the use of speech stimuli this study will only be performed in humans. This Aim directly tests the predictive coding model and examines if similar neural substrates support both predictive coding and active sensing. AIM 3 compares our ECoG data to the laminar LFP/CSD and MUA profiles and network parameters obtained in parallel monkey auditory Project 4. These unique cross-species data will be used to identify the cell populations and physiological processes that generate ECoG components in monkeys and humans providing unprecedented insights into cortical physiology in humans. We predict that active sensing mechanisms are modality independent and will also compare our finding from the auditory monkey-man to the visual human and monkey active sensing studies in Projects 1 and 2. Core C provides critical DTI and resting state fMRI to correlate with our ECoG network and HG data and Core B provides for data standardization and sharing. Finally, Project 5 provides the computational and modeling infrastructure necessary to build and refine cell and systems level models of the world is sampled. Active sensing and predictive coding are likely impaired in a host of disabling psychiatric, neurological and developmental disorders making the understanding of these processes central to the mission of the NIMH.