Mechanisms of Attentional Control: Structure and Dynamics from Simultaneous EEG-fMRI and Machine Learning

Principal Investigator: Mingzhou Ding & George R. Mangun

Co-PI: Mingzhou Ding

Sponsor: National Institutes of Health NIMH

Start Date: June 8, 2018

End Date: February 28, 2023

Amount: $2,688,000


Selective attention is an essential cognitive ability that permits us to effectively process and act upon relevant information while ignoring distracting events. A network involving frontal and parietal cortex for top-down attentional control, referred to as the Dorsal Attention Network (DAN), is active during both spatial and non- spatial (feature-based) attention. However, we know very little about the fine structure of attentional control activity in the DAN, how this structure changes to represent different to-be-attended stimulus features, how the connectivity within the DAN, and between the DAN and sensory cortex shifts when attending different features, or how these top-down processes and their influence in sensory cortex unfold over time. This gap in our knowledge is a critical problem for our models and theories of attention, and because attentional deficits are involved in a wide variety of neuropsychiatric disorders including autism, attention deficit disorder, dementia, and schizophrenia. The working model guiding this research is that top-down attentional control, based on different to-be-attended stimulus attributes, is guided by a smaller-scale neural fine structure within the DAN and prefrontal cortex that makes specific connections with specialized areas of visual cortex coding the attended attributes. Moreover, the time course of activity within the DAN in relation to that in sensory cortex follows a top-down cascading model, being earliest in frontal, then parietal cortex, and finally sensory cortex for preparatory, voluntary, attentional control. To identify the functional networks for attentional control for different forms of attention, and to define their time courses, this project uses innovative simultaneous recording of electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) data. Advanced signal processing and modeling, including multivariate pattern analysis (MVPA), graph theoretic connectivity analysis, and Granger causality analysis will be used to reveal the fine functional anatomy and time course of attentional control and selection. The project includes three experiments that vary the to-be-attended stimulus attributes from spatial location to stimulus features (color and motion), and pursues three aims. Aim 1 is to reveal the fine structure of top-down preparatory attentional control for different to-be-attended stimulus features. Aim 2 is to elucidate the specific connectivity between fine structures for preparatory attentional control in the DAN and their target sensory structures in sensory cortex. Aim 3 is to reveal the time course of top-down attentional control for different to-be-attended stimulus attributes.