Collaborative: Direct Impacts of Executive Functions on Language Comprehension: Evidence from Eye Movements and Electrophysiology

人类在听和读的过程中会遇到语言中普遍存在的歧义。例如，两个发音几乎相同的单词（例如，“beach”和“peach”）；句子可以用多种方式进行语法分析；而且说话的声音常常被周围环境的声音所掩盖，比如在拥挤的餐馆里。尽管歧义通常会增加获得错误解释的风险，甚至会增加无法解释信息的风险，但健康的年轻人在理解语言方面非常熟练，大大超过了自动系统的能力，尽管机器学习取得了重大进展。解释人类在面对歧义时是如何理解语言的，以及为什么有些人无法理解语言，是认知科学的一个基本挑战，对于帮助在歧义条件下加剧语言障碍的人（包括中风患者和听力损失的健康老年人）至关重要。本研究将使用先进的神经生理学和眼动追踪方法来阐明人们如何通过执行功能（EF）来调节他们对言语和语言输入的解释。EF是一系列认知机制，使大脑中的信息处理和目标导向行为的灵活指导成为可能。EF被认为对于帮助人们在可能有多种解释的情况下从不同的句子解释中进行选择，以及在环境“嘈杂”时引导大脑感知单词至关重要。然而，人们对EF促进语言理解的神经和认知机制知之甚少。本项目将阐明英语在语言理解中的作用。该项目将测试两种特定形式的执行功能——注意控制和认知控制——在面对歧义时支持语言理解方面发挥着基本而独特的作用的假设。注意控制被假设为通过将知觉处理导向预期信号来调节信息的收集，并在次优听力条件下将其与噪声隔离。假设认知控制通过指导内部表征在多个线索竞争时与相关证据来源保持一致来帮助纠正误解。研究人员将利用头皮脑电图记录的神经振荡活动来提供认知任务执行过程中认知控制和注意控制的新标记。通过记录听众在听句子和询问眼前场景时的毫秒级凝视动态，眼球运动模式将被用来实时跟踪参与者对语言输入的解释。脑电图和眼动追踪的结合将提供语言理解过程中神经过程和时间性质的多模态表征。为了阐明注意力和认知控制如何分别影响人类处理歧义的能力，实验将通过让参与者执行任务（如Flanker任务）来操纵这些EF亚型在处理语言之前的参与状态，这些任务已知会调动认知控制过程。这种实验方法将建立英语学习和语言加工之间的因果关系，并将确定认知控制和注意控制对理解的独特贡献。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Human beings confront pervasive ambiguity in language as they listen and read. For example, two spoken words can sound nearly identical (e.g., “beach” and “peach”); sentences can be grammatically analyzed in multiple ways; and speech is often obscured by environmental sounds, as in a crowded restaurant. Although ambiguity routinely increases the risk of arriving at the wrong interpretation, or even no interpretation of a message, healthy young adults are marvelously adept at understanding language, vastly surpassing the capabilities of automatic systems despite major advances in machine learning. Explaining how humans understand language in the face of ambiguity, and why some populations fail, is a fundamental challenge for cognitive science, and is crucial for assisting people with language impairments that intensify under conditions of ambiguity, including stroke patients and healthy older adults with hearing loss. This research will use advanced neurophysiological and eye-tracking methods to illuminate how people regulate their interpretations of speech and language input by using executive functions (EF): the family of cognitive mechanisms that enable information processing in the brain and the flexible guidance of goal-directed behavior. EF are thought to be critical to helping people select among different interpretations of a sentence when more than one is possible, and for guiding the mind toward perceiving a word when the environment is “noisy.” However, the neural and cognitive mechanisms that allow EF to contribute to language comprehension are poorly understood. This project will elucidate EF’s role in language comprehension. This project will test the hypothesis that two particular forms of executive function – attention control and cognitive control – play fundamental and distinct roles in supporting language comprehension in the face of ambiguity. Attention control is hypothesized to regulate the collection of information by directing perceptual processing toward the intended signal, isolating it from noise under suboptimal listening conditions. Cognitive control is hypothesized to assist the revision of misinterpretations by guiding internal representations to align with relevant sources of evidence when multiple cues compete. The investigators will use neural oscillatory activity in scalp-recorded EEG to provide novel markers of cognitive control and attention control during cognitive task performance. Eye-movement patterns will be used to track participants’ interpretations of language input in real time, by recording the millisecond-level dynamics of listeners’ gaze while they listen to sentences and interrogate a scene before them. The combination of EEG and eye tracking will provide a multi-modal characterization of the neural processes and temporal nature of interpretation procedures during language comprehension. In order to illuminate how attention and cognitive control separately affect humans’ ability to deal with ambiguity, the experiments will manipulate the engagement status of these EF subtypes just prior to processing language, by having participants perform tasks, such as the Flanker task, which are known to recruit cognitive-control processes. This experimental approach will establish causal relationships between EF and language processing and will identify the distinctive contributions of cognitive control and attention control to comprehension.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.