Testing hypotheses of context-dependent, distributed, and non-hierarchical neural representation using hypermodal MRI

该项目旨在使用新的“超模态”MRI数据集评估上下文依赖的听觉和视觉表示、注意力和集成的特定计算模型。MOVIE II项目使用高时间和空间分辨率的功能性MRI来全面绘制参与者对全长电影的血氧水平依赖性反应，以及更受控制的视觉、听觉和体感映射刺激，大脑区域特异性血液动力学反应，大分子、水和铁含量的高分辨率地图，以及参与者凝视和生理的时间锁定记录。博士生将评估模型的解释能力，这些模型考虑了模态内部和跨模态的感官信息的时间积累，以及更抽象的特征，如事件流和音乐主题或短语张力的建立。在博士二年级，在Skipper和Dick的指导下，学生将建立fMRI活动的解释性计算模型，该模型建立在声学和视觉特征的上下文调制程度不断增加的基础上，并测试这些功能和骨髓结构定义的大脑区域，这些区域位于不同的输入皮层层次。在3-4年级，在Tierney、Skipper和Dick的指导下，该学生还将进行一个“衍生”脑电图研究，参与者将再次观看MOVIE II数据库中使用的同一部电影。这个新的数据集将用于生成用于fMRI分析的特征子集的时间响应函数；这些将产生更详细的神经反应的时间估计，这些神经反应将用于进一步调整MOVIE II数据的初始fMRI模型。学生还将设计和运行小n功能磁共振成像研究，改变所呈现的电影的语言或声学内容，以测试关于通过更抽象的认知结构调节感知加工的假设。

The project aims to evaluate specific computational models of context-dependent auditoryand visual representation, attention, and integration using a new 'hypermodal' MRI dataset.The MOVIE II project uses high temporal and spatial resolution functional MRI tocomprehensively map out participants' blood-oxygen-level-dependent responses to a fulllength movie, as well as more controlled visual, auditory, and somatosensory mappingstimuli, brain-region-specific hemodynamic responses, high-resolution maps ofmacromolecular, water, and iron content, and time-locked recordings of participants'eyegaze, and physiology.The PhD student will evaluate the explanatory power of models that take into accounttemporal buildup of sensory information within and across modalities, as well as moreabstract characterizations, such as event flow and the establishment of musical motifs orphrasal tension. In Year 2 of the PhD, under the supervision of Skipper and Dick, the studentwill build explanatory computational models for fMRI activity that build in an increasingdegree of contextual modulation on acoustic and visual features, and test these infunctionally and myeloarchitectonically defined brain regions that are at different tiers inputative cortical hierarchies. In Year 3-4, supervised by Tierney, Skipper, and Dick, thestudent will also conduct a 'spin-off' EEG study, where participants will again view the samemovie used in the MOVIE II database.This new dataset will be used to generate temporalresponse functions to a subset of the features used for fMRI analyses; these will in turngenerate more temporally detailed estimates of neural responses that will be used to furthercondition the initial fMRI models of the MOVIE II data. The student will also design and runsmall-N fMRI studies that vary the linguistic or acoustic content of the movie presented totest hypotheses about the modulation of perceptual processing by more abstract cognitiveconstructs.