The main focus of the lab is on network analysis and modeling of connectome data sets. These data mainly result from noninvasive neuroimaging studies of the human brain, using diffusion as well as functional MRI. In addition, we work with data from tract tracing studies of the brains of other mammalian species, as well as microscopic reconstructions of the nervous systems of invertebrates (Drosophila, C. elegans). We develop and apply a broad range of analysis methods to understand the architecture and function of these networks, ranging from the fractal dimension of gray matter volumes to measures of dynamic information and complexity. In collaboration with a number of labs we investigate individual differences in network structure across development, genetic variation, personality traits, brain injury and disease states. A main focus is to relate structural connectivity patterns to dynamic interactions recorded in functional and effective connectivity.

Many brain networks studied so far exhibit community structure, i.e. they can be decomposed into structural and functional modules, at (meso)scales that are intermediate between single elements and the whole network. Charting the brain's community structure discloses structurally and functionally distinct sets of neural elements and connections, corresponding to specialized anatomical systems and neurocognitive networks, as well as highly central hub regions. We study network communities in a wide range of brain networks using approaches such as modularity maximization, consensus clustering, and stochastic block modeling. One research focus is on how community structure varies across resting or cognitive state, over development and life span, as well as in disorders.

Brain connectivity supports neural computation and information flow. The topology of the brain's structural connections plays a major role in shaping how information is exchanged between neurons and brain regions, and how it is transformed and integrated. How does network topology promote efficient communication? Using models of communication processes that incorporate aspects of information routing and diffusion, we explore how different network topologies can facilitate global information flow. We are also interested in how the combination of network topology and communication dynamics can predict empirically measured patterns of functional connectivity and behavior.

Various disease pathologies that manifest in the brain can be probed with noninvasive neuroimaging modalities such as MRI and PET. Network and graph theory methodologies can offer unique insight into disease processes in the human brain. Our interests are in development, application, and validation of novel and established network methodologies to clinical datasets, with the aim of identifying brain network properties that can improve our understanding of and ultimately drive intervention development in disease. Our primary focus in this area is on neurodegenerative diseases such as Alzheimer's Disease and related dementias.

How does activity in the brain relate to complex mental phenomena such as awareness or perception? Can we understand why consciousness fades during anesthesia, or persists during REM sleep? A current focus is using networks to analyze complex physiological time-series data, such as EEG, MEG, and fMRI signals. By combining techniques from nonlinear dynamical systems (embeddings, attractor reconstruction) with analyses from network science, information theory, and algebraic topology, we can understand the connections between the quantity/quality of consciousness and the dynamics driving brain activity. Specific projects have looked at disorders of consciousness, anesthesia, and psychedelics.

The human brain can be seen as large-scale, nonlinear dynamical system. Studying brain dynamics can shade new lights on characterizing the different states of consciousness, cognition, and perception that arise during the recording sessions. Algebraic topology has been used to study time series in signal processing and chaos theory. Using newly developed topological data analysis (TDA) methods we explore the topological features of brain dynamics at different time scales. Our aim is to develop interpretable tools for characterizing the different aspects of brain dynamics complementary to the current methodology.

Brain networks operate efficiently, and they are built economically, maximizing performance while minimizing cost. Given an economic trade-off between efficiency and economy, how optimal is the network organization encountered in biological nervous systems? Can we gain insights into major trends in network evolution by examining aspects of network topology and geometry? We are investigating these questions through comparative analysis of anatomical networks obtained from a broad range of mammalian species, as well as in neurophysiological recordings.

Working closely with the Brain Networkk and Behavior Lab we are pursuing a method to characterize the co-fluctuation patterns of functional connections across time. Using this approach, we can construct an edge-by-edge co-fluctuation similarity matrix, which we call "edge functional connectivity" (eFC). Whereas traditional (node) functional connectivity indexes the similarity among node activations, the eFC matrix provides a network account of edge relationships based on time-varying patterns of between-node communication. We are actively exploring the utility of this 'edge-centric' framework for understanding time-varying functional connectivity, network communities, individual differences, and clinical disorders.