Danielle S. Bassett - US grants

Brain stimulation is currently being used to treat a variety of cognitive, neurological, and psychiatric disorders, including depression, Parkinson's disease, and schizophrenia. Despite this broad use, how brain stimulation works remains largely a mystery. This project aims to address this gap in understanding by studying and modeling how neural processes change with brain stimulation. The investigators aim to develop criteria by which to evaluate and optimize stimulation-based treatments of neurological and psychiatric disorders. To accompany the scientific advances, the investigators will engage in educational efforts to bring the research to the classroom and to enhance cross-institutional opportunities for students. The investigators will place special emphasis on mentoring and encouraging women and minorities on the academic path in science and engineering. In addition, the investigators are combining their efforts in the Skirkanich Internship in Network Visualization, which hosts undergraduate art students each summer.

The investigators propose a new set of mathematical techniques to describe and predict neural processes and how they change with brain stimulation. Investigators at the University of California at Riverside (one of the most ethnically diverse research-intensive institutions in the U.S.) and the University of Pennsylvania will collaborate on theoretical, computational, and experimental research with three main research initiatives: (1) static and dynamic modeling of complex neural dynamics, identification of multi-resolution regions of interest in the human brain, the nature of their interconnections, and their function in observed cognitive dynamics; (2) analysis of network-wide dynamic properties of neural systems, characterization of brain regions based on controllability metrics, and design of non-disruptive control algorithms for the modulation of complex neural processes; (3) validation of models and control strategies in a wide variety of empirical settings to engineer and predict the outcomes of clinical interventions in neurological and psychiatric disorders. The success of this project will enable a deepened understanding of complex neurobiological systems, construct novel maps of the human brain and its dynamic processes, and develop non-disruptive control techniques for therapeutic brain stimulation protocols.

The development of quantitative theories of learning, memory, and prediction is fundamental to understanding human cognitive processing. This workshop, to take place in Arlington VA, May 8-9, 2014, tackles a key scientific need: to integrate modern complex systems and network approaches with understanding cognitive function. Predictive models of higher order cognitive processes could inform the development of neuroprosthetics, facilitate advances in brain-computer interfaces, and assist in the construction of intervention protocols for cognitive deficits that accompany neurological disorders and psychiatric disease.


Understanding how the human brain works has emerged as a major international focus of research in the coming decade, identified as such in President Obama's State of the Union Address in February 2013 and further developed in President Obama's BRAIN initiative announced on April 2, 2013. This workshop will bring together systems neuroscientists, cognitive scientists, applied mathematicians, and theoretical physicists. The aim is to identify a set of achievable goals that integrate dynamic, quantitative theories of cognition with neuroscientific and theoretical avenues of research.

? DESCRIPTION (provided by applicant): This project will bring together expertise in computational and experimental neuroscience, signal processing and network science, statistics, modeling and simulation, to establish innovative methods to model and analyze temporally dynamic brain networks, and to apply these tools to develop predictive models of brain-computer interface (BCI) skill acquisition that can be used to improve performance. Leveraging experimental data and interdisciplinary theoretical techniques, this project will characterize brain networks at multiple temporal and spatial scales, and will develop models to predict the ability to control the BCI as well as methods to engineer BCI frameworks for adapting to neural plasticity. This project will enable a comprehensive understanding of the neural mechanisms of BCI learning, and will foster the design of viable BCI frameworks that improve usability and performance. Intellectual Merit: As a critical innovation, this project proposes to develop a systematic and rigorous approach based on neuroimaging techniques, signal processing, and network science for the modeling and analysis of temporally dynamic neural processes that characterize BCI skill learning. To achieve these goals, we will organize our research around the following objectives: (i) characterizing multiple spatio-temporal scales of dynamic functional brain networks, (ii) modeling BCI skill acquisition and predicting performance from brain network properties, (iii) simulating coadaptive BCI frameworks using dynamic network-based neural features. Results will first be characterized from pure graph-theoretic and neuroscience perspectives, so as to highlight fundamental research challenges, and then validated to clarify the importance and the applicability of our findings to translational efforts in practical BCI scenarios. Our results wil (i) unveil multi-resolution properties of dynamic brain networks, (ii) identify predictive neuromarkers for BCI learning, and ultimately (iii) inform the development of coadaptive BCI frameworks sensitive to subject-specific neural plasticity. The two young PIs - one from the Department of Bioengineering at the University of Pennsylvania and one from the ARAMIS team of the Institut National de Recherche en Informatique et en Automatique (INRIA) located at the Institut du Cerveau et de la Moelle epiniere (ICM) in Paris - bring complementary and interdisciplinary backgrounds to this research project, with a strong track record in network analysis, network neuroscience, multimodal neuroimaging and BCI applications. Their experience and resources will enable the success of this new approach to analyze dynamic networks in BCI learning, design co-adaptive BCI frameworks, and facilitate the use of non-invasive BCI technology for both control of external devices (e.g. neuroprosthetics) as well as neurofeedback applications (e.g. MI-based neurorehabilitation after stroke). Broader Impacts: This interdisciplinary project proposes a transformative approach to analyze large-scale neural systems, and to model and predict BCI skill acquisition. This research provides novel insights into the temporal interconnection structure of the human brain, and proposes entirely new methods to construct dynamic network-based models of neural plasticity from multimodal neuroimaging data. Results will foster the development of innovative predictive neuromarkers for the diagnosis and treatment of neurological disorders and psychiatric disease. The PIs will bring their findings and innovative techniques to the undergrad and graduate programs at their institutions, disseminate findings via dedicated courses, workshops, and publications, and to the community and local middle/highschools via lectures and STEAM outreach events.

The ability to learn relational data is critical to human life as we know it. By learning the relationships between syllables and words, or scientific and mathematical concepts, we produce language, form lexical knowledge, develop physical intuition, exercise logical deduction, and attain expertise in our line of work. Collectively, these relational data can be described as a graph in which nodes might represent syllables or concepts, and edges might represent shared content or conditional probabilities. Yet, how the organization of such a graph impacts our ability to learn the data or the neural processes that affect learning is far from understood. In this project the PI will use network science as a mathematical framework within which she will study the human learning of relational patterns, and answer the question of whether graphs that are complex in the mathematical or naturalistic senses are more or less difficult to learn, or require different neural processes. To facilitate broader impacts, these efforts incorporate art to transform STEM to STEAM, a recent international innovation that improves long-term retention of content and scientific reasoning. The goals of this program are (i) to create a local community - from preschoolers to adults - who are generally inspired by cutting edge science, and who more specifically appreciate the concepts of network architectures in natural information and in their brain?s ability to learn that information, (ii) to produce undergraduate and graduate students trained at the interdisciplinary boundary between network science and neuroscience to address critical and timely scientific questions that transcend national boundaries, (iii) to develop course material that incorporates these timely research questions, and (iv) to polish and release teaching materials developed in these aims to international and global collaborators, and to the public. The PI complements these efforts with extensive mentorship for women and underrepresented minorities in STEM fields, and with educational outreach efforts in under-served inner-city Philadelphia schools.

In particular the PI will use a 3-pronged approach that employs (i) engineering-based tools from network science to systematically define graph ensembles of relational information with dissociable topologies, (ii) behavioral studies to determine which graph topologies are easier or harder to learn, and (iii) functional neuroimaging to identify predictors of individual differences in learning. Exploratory work seeks to translate the knowledge gained in these areas to instructed learning of scientific concepts. In this proposal, the PI brings together her background in theoretical physics and network science, her expertise in multimodal human neuroimaging, her current research program intersecting engineering and cognitive neuroscience, and her recently developed methods to predict individual differences in learning from the dynamics of human brain functional connectivity to determine how the graph topology of relational information maps to individual differences in human learning behavior as produced by dissociable neurophysiological processes.

? DESCRIPTION (provided by applicant): The rapid development of executive function is a hallmark of adolescence, and requires the integrated recruitment of large-scale neural circuitry spanning multiple brain regions. Here we propose to develop and apply methods from network control theory to brain imaging data in order to allow us to understand how executive function develops in youth. Recent advances in control and dynamical systems theory have provided quantitative diagnostics of network controllability, which collectively define how external input t network nodes (in this case brain areas) can move the entire system (in this case cognitive function). Further, these methods allow quantitative estimation of the costs of disparate control structures. While these techniques have not previously been applied to brain imaging data, they provide an intuitive mechanism for executive function, and therefore have the potential to be putative biomarkers of executive capability. In this proposal, we capitalize upon existing diffusion imaging and working-memory task fMRI data acquired in a large sample of youth ages 8-22 imaged as part of the Philadelphia Neurodevelopmental Cohort (PNC). In Aim 1, we will describe the network control structure of the brain's structural connectome by (i) identifying driver nodes of network control and mapping their relationship to known executive networks and (ii) quantifying test-retest reliability of network control diagnostics. In Aim 2, we will chart th evolution of network control in adolescence by characterizing the development of network control diagnostics using cross-sectional (n=968) and longitudinal data (n=350) PNC data. In Aim 3, we will determine how network control is associated with executive function by examining whether individuals with higher levels of network control demonstrate better executive functioning, and investigating if baseline controllability predicts subsequent longitudinal improvement of executive function. In Aim 4, we will provide a publically available toolbox for measurement of network controllability as a resource to the neuroimaging community. This multi-disciplinary research has the potential to yield high-impact discoveries, and therefore represents a good fit for the R21 mechanism. Risks associated with the innovation in this project are tempered by the strength of our team (expertise in network science, developmental neuroimaging, and neuropsychiatry), the multiple levels of hypotheses to be tested, convergent preliminary data, and our intimate familiarity with the PNC dataset.

This project, developing a new instrument to enable an accurate quantitative analysis of the movement of animals and vocal expressions in real world scenes, aims to facilitate innovative research in the study of animal behavior and neuroscience in complex realistic environments. While much progress has been made investigating brain mechanisms of behavior, these have been limited primarily to studying individual subjects in relatively simple settings. For many social species, including humans, understanding neurobiological processes within the confines of these more complex environments is critical because their brains have evolved to perceive and evaluate signals within a social context. Indeed, today's advances in video capture hardware and storage and in algorithms in computer vision and network science make this facilitation with animals possible. Past work has relied on subjective and time-consuming observations from video streams, which suffer from imprecision, low dimensionality, and the limitations of the expert analyst's sensory discriminability. This instrument will not only automate the process of detecting behaviors but also provide an exact numeric characterization in time and space for each individual in the social group. While not explicitly part of the instrument, the quantitative description provided by our system will allow the ability to correlate social context with neural measurements, a task that may only be accomplished when sufficient spatiotemporal precision has been achieved.

The instrument enables research in the behavioral and neural sciences and development of novel algorithms in computer vision and network theory. In the behavioral sciences, the instrumentation allows the generation of network models of social behavior in small groups of animals or humans that can be used to ask questions that can range from how the dynamics of the networks influence sexual selection, reproductive success, and even health messaging to how vocal decision making in individuals gives rise to social dominance hierarchies. In the neural sciences, the precise spatio-temporal information the system would provide can be used to evaluate the neural bases of sensory processing and behavioral decision under precisely defined social contexts. Sensory responses to a given vocal stimulus, for example, can be evaluated by the context in which the animal heard the stimulus and both his and the sender's prior behavioral history in the group. In computer vision, we propose novel approaches for the calibration of multiple cameras "in the wild", the combination of appearance and geometry for the extraction of exact 3D pose and body parts from video, the learning of attentional focus among animals in a group, and the estimation of sound source and the classification of vocalizations. New approaches will be used on hierarchical discovery of behaviors in graphs, the incorporation of interactions beyond the pairwise level with simplicial complices, and a novel theory of graph dynamics for the temporal evolution of social behavior. The instrumentation benefits behavioral and neural scientists. Therefore, the code and algorithms developed will be open-source so that the scientific community can extend them based on the application. The proposed work also impacts computer vision and network science because the fundamental algorithms designed should advance the state of the art. For performance evaluation of other computer vision algorithms, established datasets will be employed.

Cognitive control is the ability to guide our thoughts and actions in accord with our internal intentions. It enables us to make good decisions, balance options, choose appropriate behaviors and inhibit inappropriate behaviors. Yet our understanding of how cognitive control works in the brain is critically lacking. The research outlined in this proposal will address this outstanding problem by developing and validating a mechanistic model to explain the fundamental principles enabling cognitive control. This problem is of urgent national interest and clinical relevance: greater understanding of how brain structure gives rise to cognitive control may be critical for the development of earlier and more effective treatments of the many neuropsychiatric disorders where cognitive control deficits are present. In addition, this project will create new research opportunities for undergraduate and graduate students in neuroscience, network theory, data sciences, and mathematics. The investigators will integrate the research into undergraduate and graduate teaching activities, providing a powerful bridge between theoretical and experimental applications for students at the University of Pennsylvania and the University of California at Riverside, one of America's most ethnically diverse research-intensive institutions. The investigators will also incorporate this material in extensive community and educational outreach efforts, in addition to translating this knowledge to mental health clinics.

In this research project, the investigators seek to develop, validate, and test a mechanistic theory of cognitive control. They postulate that the regulation of cognitive function is driven by a network-level control process akin to those utilized in technological, cyberphysical, and social systems. Their approach is grounded in network control theory, a relatively new subdiscipline of control and dynamical systems. In contrast to the descriptive statistics of graph theory, network control theory offers a principled mathematical modeling framework to inject energy into a networked system leading to a predictable alteration in the system's dynamics. Traditionally applied to mechanical and technological systems, this field builds on notions of structural controllability to ask specific questions about the difficulty of the control task and how to design realistic control strategies in finite time, with limited energy resources. The work will (i) develop a network-based theory of cognitive control informed by neuroimaging data, (ii) validate a network-based theory of cognitive control using data-informed computational models, (iii) define how network structure impacts individual differences in cognitive control performance in adults undergoing cognitive training, and (iv) release a publicly available toolbox for network controllability analysis. These theories and tools are the result of a truly integrated and cross-disciplinary approach to cognitive control, which blends the engineering and data sciences with empirical methodologies in neuroscience.

Epilepsy affects 65 million people worldwide. While medications control many, over 20 million patients continue to have seizures despite maximal medical therapy. New surgical techniques, laser thermal ablation and responsive devices are exciting options for these patients, but their effectiveness is limited by our inability to accurately map which brain regions should be removed or treated with electrical stimulation. Currently, this mapping is done manually, but seizure onset patterns on intracranial EEG (IEEG) are frequently not well localized, and clinicians often disagree on seizure onset time, location, and what regions should be targeted. Finally, most patient evaluations present a number of viable options for surgery and device placement. There is currently no way to test the effects of a specific therapeutic approach- an operation or device placement- on outcome other than actually doing the procedure. A technique that could simulate these interventions and pick the best approach for individual patients would be a tremendous step forward in clinical care. In this proposal we develop and validate exciting new methods to localize epileptic networks from intracranial EEG that: (1) replace manual marking by clinicians with automated, objective tools, (2) remove the need for precipitating acute seizures during evaluation to localize them and (3) allow clinicians to simulate the effects of different brain surgeries or device placements for individual patients to select the treatment that will work best for them. This work marries new graph theoretical computational methods to model brain networks from IEEG with state of the art neuroimaging techniques to precisely localize implanted electrodes, devices and brain structure. Adult and pediatric patients undergoing brain implants during evaluation for epilepsy surgery or NeuroPace Responsive Neurostimulator (RNS) device placement will be enrolled at the Hospital of the University of Pennsylvania and Children's Hospital of Philadelphia. We will obtain high-resolution brain imaging before and after electrode implant and after surgery or device placement. Our models, recently published, will be applied to each patient's data and brain regions that drive seizures will be quantitatively identified and mapped to their brain images. Patients will undergo standard invasive therapy, either resection or device implant, and outcome- reduction in seizure frequency- will be compared to the amount of the epileptic network that is removed or stimulated by an implanted device. Finally, we will test our ?virtual resection? technique against each patient's data to predict which therapeutic intervention will be most effective, and compare this prediction to the performed procedure and patient outcome. This work differs from many computational studies in that its focus is on developing practical tools to guide invasive treatment for medication resistant epilepsy. It leverages an established collaboration between experienced clinicians in adult and pediatric epilepsy with experts in neuroimaging, bioengineering, functional neurosurgery and a MacArthur-award-winning computational neuroscientist at the University of Pennsylvania.

ABSTRACT Executive function (EF) undergoes dramatic development during adolescence, and is impaired across multiple psychiatric disorders such as ADHD and psychosis. Despite this fact, the neural substrates of EF development remain incompletely understood. Here, we propose to study the development of EF using cutting- edge techniques from network science. In this proposal, we will recruit 140 participants ages 10-16. This sample will include 50 with ADHD, 50 with psychosis-spectrum diagnoses, and 40 typically developing comparators. Using an accelerated longitudinal design, all participants will be followed and undergo cognitive testing, clinical assessment, and advanced multi-modal neuroimaging at 18 month intervals, yielding an average of 2.5 sessions per participant. This design will allow us to chart the development of structural and functional brain networks during adolescence, and delineate how abnormalities of brain network development are associated with deficits in EF performance, activation, and dynamics. Our overarching hypothesis is that the development of modular yet integrated brain networks during adolescence allows for specific patterns of EF activation and dynamics, and represents a fundamental mechanism for EF development. We posit that abnormalities of network development will be associated with executive dysfunction that is dimensionally present across psychiatric disorders such as ADHD and psychosis. Accordingly, in Aim 1 of this proposal, we will chart the longitudinal development of both structural and functional brain networks, and define how abnormalities of network development are associated with dimensional EF deficits in youth with ADHD and psychosis. In Aim 2, we will define how abnormal development of brain network topology is associated with alterations of executive activation and dynamics. In Aim 3, we will integrate high-dimensional imaging data using advanced multivariate analytic techniques to create a dimensional predictor of executive dysfunction. Finally, in Aim 4, we will share both raw and processed data, creating a valuable new resource for the neuroscience community. This proposal capitalizes on complementary skills of the PIs and the research team, including expertise in brain development, network science, psychopathology, cognitive science, and high dimensional imaging statistics. Through the proposed multi-level analysis, this innovative research will provide a substantial advance in our understanding of the neurodevelopmental substrates of executive dysfunction across psychiatric disorders in adolescence.

ABSTRACT Despite the increasing use of transcranial magnetic stimulation (TMS) in both research and clinical practice, the field nonetheless lacks a theoretical framework to predict the impact of TMS on circuits. In this application, we propose to test the over-arching hypothesis that brain responses to TMS are governed by both the network control properties of the stimulation site and the functional context of the network during stimulation. Recent advances in network control theory have provided quantitative diagnostics of network controllability, which collectively define how external input (e.g., TMS) to network nodes (e.g., brain regions) can move the entire system. In Aim 1, we will test the hypothesis that TMS targeted to regions of high network control will produce greater brain responses than TMS targeted to regions of low network control. Specifically, we will recruit healthy young adults (n=40) and use ultra-high resolution diffusion imaging to identify control points that are topologically situated to drive network reconfiguration. We will use cutting-edge interleaved TMS/fMRI to test the hypothesis that individually-targeted TMS at control points will produce greater network segregation, consisting of fronto-parietal network activation, DMN de-activation, and reduced connectivity between the two. In Aim 2, we will examine the impact of the functional context of TMS. We predict that TMS simulation during conditions of high working memory (WM) load will result in greater network segregation responses than in conditions of lower load. In Aims 3 & 4, we will examine the degree to which individually targeted stimulation during WM task performance augments behavioral WM performance following repetitive TMS. We predict that the behavioral impact of neuromodulation on WM performance will scale with observed increases in network segregation versus baseline in both our sample of healthy young adults as well as age-matched patients with ADHD who have documented executive deficits (n=35). This proposal leverages our group's unique expertise in advanced TMS/fMRI, network science, and multi-modal imaging. Together, this research will elucidate basic mechanisms of neuromodulation that will accelerate translation of these therapies to clinical practice and more definitive links between brain functional modules and brain functioning.

The field of network neuroscience has developed powerful analysis tools for studying brain networks and holds promise for deepening our understanding of the role played by brain networks in health, disease, development, and cognition. Despite widespread interest, barriers exist that prevent these tools from having broader impact. These include (1) unstandardized practices for sharing and documenting software, (2) long delays from when a method is first introduced to when it becomes publicly available, and (3) gaps in theoretic knowledge and understanding leading to incorrect, delays due to mistakes, and errors in reported results. These barriers ultimately slow the rate of neuroscientific discovery and stall progress in applied domains. To overcome these challenges, we will use open science methods and cloud-computing, to increase the availability of network neuroscience tools. We will use the platform brainlife.io for sharing these tools, which will be packaged into self-contained, standardized, reproducible Apps, shared with and modified by a community of users, and integrated into existing brainlife.io analysis pipelines. Apps will also be accompanied by links to primary sources, in-depth tutorials, and documentation, and worked-through examples, highlighting their correct usage and offering solutions for mitigating possible pitfalls. In standardizing and packaging network neuroscience tools as Apps, this proposed research will engage a new generation of neuroscientists, providing them powerful new and leading to new discoveries. Second, the proposed research will contribute growing suite of modeling analysis that can be modified to suit specialized purposes. Finally, the Brainlife.io platform will serve as part of the infrastructure supporting neuroscience research. Altogether, these advances will lead to new opportunities in network neuroscience research and further stimulate its growth while increasing synergies with other domains in neuroscience.

Understanding the relations between the anatomical structure of the human brain and its functions in healthy and diseased states can not only lead to the design of novel, targeted, non-invasive, and highly-effective treatments for neurological disorders, but also inform the application of innovative stimulation schemes to enhance cognitive performance and executive capabilities. Leveraging data obtained with state-of-the-art sensing and imaging technologies, this project pursues these objectives by innovatively studying the human brain as a dynamic network system comprising neuronal ensembles and white-matter fibers, and as governed by principles similar to social and technological cyber-physical networks. This project develops and validates new rigorous theories and tools to address an outstanding problem in network neuroscience. Namely, to leverage the brain anatomical structure to characterize, predict, and control patterns of synchronized neural activity, and to validate the methods with realistic brain data. This project will not only contribute to the theories of networks, controls, and neuroscience, but also to their integration, by leveraging different levels of abstraction (brain representations from diffusion imaging data, electrocorticography time series, mathematical models) and distinct disciplinary approaches. In addition to new methods to study synchronized activity in the brain and inform the next generation of diagnostics, this project pursues far-reaching teaching and outreach activities, including (i) a number of university-level initiatives at the graduate and undergraduate levels, (ii) outreach activities that will engage young people from the local communities in Philadelphia and Riverside, and (iii) dissemination activities that will bring together traditionally separated communities and promote multi-disciplinary initiatives to tackle some of the most pressing problems in neuroscience.

The central hypothesis of this project is that the interconnected structure of the brain determines its performance and controls its transitions between healthy and diseased states. Building on this hypothesis, this project addresses the unsolved problems of characterizing, predicting, and controlling patterns of synchronized neural activity in the human brain from sparse and coarse temporal measurements and interventions. Additionally, to support the hypothesis and validate the theories of neural synchronization, the project leverages three unique and extensive multimodal neuroimaging datasets combining high-resolution electrocorticography and diffusion imaging that will allow to assess the relations between synchronization patterns and underlying structural network architecture. Specifically, this project is organized around two main tasks. Task 1, abstracts the problem of controlling patterns of neural activity as the problem of controlling the degree of synchronization among interconnected nonlinear oscillators, where oscillators represent brain regions and their interconnections reflect the anatomy of the human brain as reconstructed by diffusion magnetic resonance imaging. The idea is put forth that altered synchronization patterns are the results of, possibly small, modifications to the oscillators' interconnection structure and weights, and that desirable patterns can be restored by minimal and localized structural interventions. Task 2 uses empirical data to obtain inferences complementing those acquired in the formal theoretical and modeling work in Task 1. Because the focus here is the analysis, prediction, and control of cluster synchronization, the empirical efforts remain constrained to the study of functional neuroimaging data with clear electrographic signatures of synchronization. Specifically, the project uses electrocorticography data, which boasts markedly greater temporal resolution than functional magnetic resonance imaging and does not suffer from the issues of volume conduction that are more common in electroencephalography and magnetoencephalography. The project blends and extends tools from control and network theories, dynamical systems, data analysis, and network neuroscience. While this project focuses on synchronization problems in neural activity, the methods have broad applicability in engineering, for instance to design optimized networks and sparse controllers, network neuroscience, and network science.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

As humans navigate their environment, anticipation, planning, and perception all require an accurate map of the statistical regularities governing their visual, linguistic, auditory, and social experiences. In each context, hu- man experience consists of a sequence of events. Each event succeeds another according to a set of underlying rules codifying possible event-to-event transitions, and the likelihood of each. To make predictions about the fu- ture and respond to the environment with flexible behavior, humans must infer this network of transitions, forming a cognitive map of causes and effects. Such maps and inferences are made possible by statistical learning. The study of statistical learning represents a major opportunity for computational psychiatry for three reasons. First, statistical learning shows differential accuracy across psychiatric conditions, task domains, and temporal scales of experience. Second, statistical learning has marked potential for back-translation; multiple features of statistical learning behavior and its neural underpinnings are conserved in non-human primates, and simpler forms of sequence learning exist in other mammals (rats and mice) as well as birds. Third, ? as we describe in depth in our proposal ? statistical learning can be formally modeled mathematically. It is now timely to develop a flexible computational model of statistical learning. To serve the goals of com- putational psychiatry, the functional form of such a model should reflect general principles of statistical learning and the parameters should be sensitive to variability in behavior across the many specific disorders where deficits appear. In preliminary experimental, computational, and theoretical work, we have uncovered a novel behavioral signature of statistical learning; we have also translated that behavior into a formal model ? inspired by principles of statistical physics ? with mathematically well-defined parameters, thereby deriving a theory that is grounded in our previous experimental findings. Finally, we have experimentally validated the model by making accurate predictions of behavior in a novel experiment. Here we assemble a complementary set of co-investigators who have co-authored 31 papers in pairs or triplets, with expertise in mathematical modeling and statistical physics (Bassett), statistical models of behavior (Moore), intensive longitudinal experiments (Lydon-Staley), statistical learning (Thompson-Schill), and sensory process- ing in psychiatry (Wolf). Together, we offer a well-integrated theoretical and experimental plan to hone our math- ematical model of an aspect of human behavior that has not been extensively analyzed computationally, and in which the underlying dimensional process is affected in psychiatric disorders. We distill our aims into reliability, relevance, and generalizability of our model. Our approach is three-pronged, with innovations in experiment, computation, and theory building on our team?s diverse expertise. Each prong will address all three aims, thereby integrating our efforts to build a computational model of statistical learning behavior supporting future advances in computational psychiatry. Our proposed efforts provide the foundation for an R01 extending to patients.