Multivariate Representations of Emotion

描述（由申请人提供）：情感神经科学的一个中心目标是了解评估，体验和情感表达的大脑系统和机制。例如，该领域中一个被广泛研究和激烈争论的问题是，对情绪刺激的生物物理反应可以被表征的方式，无论是通过不同的类别还是沿着效价和唤醒的维度。心理学、神经科学和计算机科学领域的进步促进了在识别参与处理情绪的大脑区域方面的重大进展;然而，不同情感状态的一致和特定的神经标记还没有被发现。2这个提议使用了一个前沿的方法来研究这个核心，通过利用新兴的模式分类技术，能够检测来自一系列源的微妙但协调的信号，解决该领域尚未解决的问题。总体目标是确定多元模式的行为和生物响应特定的情感状态，并确定这些状态是否组织根据分类或维度架构。通过结合心理生理学（目标1）和功能性磁共振成像（fMRI）（目标2），这些研究将研究人类如何对持续时间，模态和分类性质不同的情绪刺激做出反应。研究1侧重于器乐和电影片段引起的不同情绪，而研究2侧重于面部和声音的影响引起的。总之，这些目标将提供一种综合的，计算严格的方法来识别通常被传统单变量统计方法忽略的特定情绪的生物标志物。以这种创新的方式应用机器学习算法，可以有效地识别情感障碍中的情感表征是如何改变的，并有可能开发出新的治疗靶点。此外，识别特定情绪类别之间的重叠模式可能会进一步帮助理解焦虑和情绪障碍中的共病问题。

DESCRIPTION (provided by applicant): A central goal of affective neuroscience is to understand the brain systems and mechanisms underlying the evaluation, experience, and expression of emotion. For example, a widely studied and hotly debated issue in the field is the manner in which biophysical responses to emotional stimuli can be characterized, whether by distinct categories or alternatively along dimensions of valence and arousal. Advances in the fields of psychology, neuroscience, and computer science have fostered significant progress in identifying brain regions involved in processing emotion generally; however, consistent and specific neural markers for distinct affective states have yet to be found. This proposal uses a cutting-edge approach to this core, unresolved question in the field by harnessing emerging pattern classification techniques that are capable of detecting subtle yet coordinated signals from an array of sources. The overarching goal is to identify multivariate patterns of behavioral and biological responding to specific affective states and determine whether these states are organized according to categorical or dimensional architectures. By combining psychophysiology (Aim 1) and functional magnetic resonance imaging (fMRI) (Aim 2), these studies will examine how humans respond to emotional stimuli that vary in duration, modality, and categorical nature. Study 1 focuses on distinct emotions elicited by instrumental music and movie clips whereas Study 2 focuses on those elicited by facial and vocal affect. Together, the aims will provide an integrative, computationally-rigorous method to identify biomarkers of specific emotions that are typically overlooked by conventional univariate statistical approaches. Applying machine learning algorithms in this innovative way could be fruitful for identifying how emotional representations are altered in affective disorders, with the potential for developing novel therapeutic targets. Moreover, identifying patterns of overlap between specific emotion categories may further aid efforts to understand comorbidity issues in anxiety and mood disorders.