This project focuses on developing new analysis tools to investigate the mechanisms of information routing in primate fronto-striatal circuits during goal-directed behavior. These tools are part of a seamless analysis pipeline that will enable neuroscience researchers to rigorously test hypotheses fast and with a minimum of translation between incompatible strategies and tools. The rationale for developing these tools is that evidence from recent studies suggests that information routing and changes in routing happen during transient bursts of activity. This project develops tools to test this hypothesis under three specific aims. First, we employ a new time-frequency-scale analysis algorithm to detect transient bursts. Transient bursts have a wide range of variable durations, suggesting that a time-frequency-scale dictionary composed of elementary signal atoms would prove useful to accurately match recorded bursts. Generating a sparse code in overcomplete dictionaries is a difficult computational problem, and existing adaptive algorithms are slow and computationally intensive. The algorithm we develop and deploy here -- the Multiscale Adaptive Gabor Expansion or MAGE -- accomplishes this data-driven time-frequency-scale decomposition of a raw signal into extracted bursts and inter-burst periods. The MAGE algorithm developed by our lab is a methodological advance over existing methods, estimating ongoing amplitude and phase of burst activity with precision and accuracy. Second, we characterize how burst events coordinate distinct regions to act as one functional network. Anatomically-connected brain areas form the scaffolding upon which multi-site transient bursts rapidly create and dissolve transient functional networks. Unfortunately, phase coupling between neurophysiological signals has often been investigated using suboptimal techniques. We will show that an extension of a well characterized multivariate probabilistic model (Cadieu-Koepsell PCE) can be used to investigate changes in network phase coupling during burst periods, and we detail the advantages this model has over the commonly-employed phase- locking value (PLV) ? including increased sensitivity, specificity, and sparsity. Third, we repurpose an existing dataset that employed electrical stimulation to address the impact network coordination has on the direction and strength of information flow through an active network. By measuring the trial-by-trial variation in cortico-cortico evoked potential (CCEPs) amplitude and waveform shape, the directionality and transmission gain of network links can be estimated. Electrical stimulation is an approach that will be used by more labs in the future in the development of novel clinical protocols and the investigation of causal mechanisms of network control. In sum, this project will develop original tools to study an important problem via a rigorous framework. The investigators have successfully completed similar projects with good results, and the strong empirical focus of our laboratory at Vanderbilt University makes it a perfect place to develop novel and useful analytic tools.
How can a distributed brain network with relatively fixed anatomy rapidly reconfigure itself to perform one task versus another? We hold that the rapid remapping of functional roles, inter-area communication, and local computation are accomplished through changes in patterns of dynamic activity that effectively control information flow and functional processes. Using simulation models as well as empirical data, our research will validate novel tools and analysis pipelines that can be used to investigate this hypothesis that the causal mechanisms guiding information processing in fronto-striatal circuits are brief, transient changes in distributed patterns of dynamic ongoing activity.
