Garrett B. Stanley - US grants

computational neuroscience; visual stimulus; brain electrical activity; lateral geniculate body; visual pathways; electrophysiology; cats;

DESCRIPTION (provided by applicant): The role of thalamic synchrony in gating information flow to cortex A ubiquitous property of sensory pathways is that they continuously adapt to changes in the properties of the sensory input. Adaptation is not simply fatigue or attenuation of neural activity, but instead can fundamentally change the features to which the sensory pathway are sensitive and thus changes what the pathway encodes. We have recently shown that adaptation through ongoing tactile stimulation in the rat vibrissa system induces a fundamental change in what behaviorally relevant features the cortex encodes: from detection of a tactile contact to discrimination between different speeds of the tactile/whisker input. To what extent this observation is general and whether it manifests perceptually has not been well studied. In this project, using a combination of electrophysiology and related functional measures and behavior, we will determine whether adaptation enhances discrimination at the expense of detection through modulation of the timing synchrony of thalamic input to cortex. We do so through an integrated, parallel study using multiple recording techniques in the vibrissa pathway of the anesthetized rat, and using behavioral tasks focused on detection of and discrimination between tactile stimuli. Specifically, we will test whether adaptation switches the fundamental cortical processing mode from detection to discrimination through modulation of synchrony of projecting thalamic input (AIM 1). We will directly test the generality of this central hypothesis across whisker deflection speed and direction, and in the context of spatial acuity across whiskers, and determine how this phenomenon is shaped by the statistics of the adapting tactile input. Second, we will directly test whether adaptation switches perceptual performance from detector to discriminator during behavior and whether modulation of thalamic synchrony mediates the switch (AIM 2). Rats will be trained to either detect whisker contact or discriminate between contacts of nearby whiskers, while we monitor population activity in VPm thalamus through chronically implanted multi-electrodes within and across vibrissa-specific regions. Significance: It has long been posited that adaptation acts to enhance information flow in sensory pathways and human psychophysical studies have indeed shown that in certain circumstances adaptation acts to enhance discriminability of tactile inputs. However, the precise link between the underlying neural representations in the thalamocortical circuit and the resultant percept remains a major open question in neuroscience. Success of our aims will directly determine how the cortical representation changes through modulation in thalamic input and the consequences of this on perception. Broad Impacts: Various brain pathologies such as autism, concussive injuries, and alcoholism result in an impaired effect of adaptation on spatial acuity. Success of our aims will provide a more solid foundation for sensory-based diagnostics, and may help to shed light on the pathophysiology associated with these life-altering disorders. Further, one mechanism that we explore is that of thalamic synchrony and its effect on cortical activation, which has been implicated in seizure generation and thalamic pain. PUBLIC HEALTH RELEVANCE: Sensory input is critical in our daily lives, for both the perception of the world around us, and in providing feedback for our muscle systems that help us interact with the external world. How the nervous system performs this feat, and precisely how this helps us in our environment, is unknown. Discovery in this area can potentially help us understand a number of disorders/diseases of the nervous system for which individuals exhibit loss of acuity/sensitivity, and do not have the ability to adapt to changes in the sensory environment.

DESCRIPTION (provided by applicant): Emory and Georgia Tech have steadily grown the number of faculty involved in computational neuroscience over the past 15 years. The research of these faculty stretch from cellular to systems and theoretical approaches. In 1997 the two Institutions formed a joint Department of Biomedical Engineering, further strengthening the highly collaborative atmosphere between researchers on both campuses. In addition both campuses have a strong track record both in undergraduate and graduate teaching. The proposed training program in computational neuroscience aims to capitalize on these strengths by formalizing an integrated approach to class work and research on both undergraduate and graduate levels. The strong NIH and NSF funded research programs of more than 15 principal investigators identified as computational neuroscientists range from detailed cellular computer simulations of neural dynamics to engineering approaches and the quantitative study of disease mechanisms underlying important disorders such as epilepsy and Parkinson's disease using computational methods. Therefore students will be exposed to multiple levels of approaches aimed ultimately at addressing medical questions. A highly qualified and diverse applicant pool for student fellowships under this program exists on both undergraduate and graduate levels, and will bring some applicants with a primarily background in the biological sciences to integrate computational approaches into their research, and vice versa brings more computational or theoretically oriented applicants in touch with biological experimental research. The program encompasses a cohort of 6 undergraduate and 6 graduate student fellows, who will absolve a rigorous curriculum in neurobiology and mathematical and computational methods through a core sequence of required classes as well as individually chosen electives. Undergraduate fellows will be funded for a period of two years in their junior and senior years, during which they will undertake specialized class work and research in a computational neuroscience lab. Undergraduate trainees will be primarily recruited from the Emory Neuroscience and Behavioral Biology and the Georgia Tech Biomedical Engineering majors, who bring a biological and quantitative strength to the program, respectively. Over 200 students join these majors annually, and we will only take applicants with a GPA of 3.5 or better and expressing an interest in future research graduate training. The graduate students in this program will be recruited from the applicant pools for the Neuroscience and Biomedical Engineering programs, which together receive more than 120 highly qualified applications each year. A special track for fellows in computational neuroscience will be announced on the program websites that will also link to an extensive independent website describing this program. Graduate students will be funded for the first two years of their education, and then obtain individual training grants or be funded by research grants. PHS 398/2590 (Rev. 06/09) Page Continuation Format Page

Emory and Georgia Tech have steadily grown the number of faculty involved in computational neuroscience over the past 15 years. The research of these faculty stretch from cellular to systems and theoretical approaches. In 1997 the two Institutions formed a joint Department of Biomedical Engineering, further strengthening the highly collaborative atmosphere between researchers on both campuses. In addition both campuses have a strong track record both in undergraduate and graduate teaching. The proposed training program in computational neuroscience aims to capitalize on these strengths by formalizing an integrated approach to class work and research on both undergraduate and graduate levels. The strong NIH and NSF funded research programs of more than 15 principal investigators identified as computational neuroscientists range from detailed cellular computer simulations of neural dynamics to engineering approaches and the quantitative study of disease mechanisms underlying important disorders such as epilepsy and Parkinson's disease using computational methods. Therefore students will be exposed to multiple levels of approaches aimed ultimately at addressing medical questions. A highly qualified and diverse applicant pool for student fellowships under this program exists on both undergraduate and graduate levels, and will bring some applicants with a primarily background in the biological sciences to integrate computational approaches into their research, and vice versa brings more computational or theoretically oriented applicants in touch with biological experimental research. The program encompasses a cohort of 6 undergraduate and 6 graduate student fellows, who will absolve a rigorous curriculum in neurobiology and mathematical and computational methods through a core sequence of required classes as well as individually chosen electives. Undergraduate fellows will be funded for a period of two years in their junior and senior years, during which they will undertake specialized class work and research in a computational neuroscience lab. Undergraduate trainees will be primarily recruited from the Emory Neuroscience and Behavioral Biology and the Georgia Tech Biomedical Engineering majors, who bring a biological and quantitative strength to the program, respectively. Over 200 students join these majors annually, and we will only take applicants with a GPA of 3.5 or better and expressing an interest in future research graduate training. The graduate students in this program will be recruited from the applicant pools for the Neuroscience and Biomedical Engineering programs, which together receive more than 120 highly qualified applications each year. A special track for fellows in computational neuroscience will be announced on the program websites, that will also link to an extensive independent website describing this program. Graduate students will be funded for the first two years of their education, and then obtain individual training grants or be funded by research grants.

DESCRIPTION (provided by applicant): In-vivo control of information flow by artificial stimulation: ephys and behavior Sensory pathways in the brain exist to extract and process information about the outside world, so that we may perceive our environment and make decisions about our actions. In contrast to motor control, where a rich history of investigation has formed a foundation for linking neuronal population activity in motor regions to elements of motor output, we do not have an analogous general framework for providing surrogate signals to sensory pathways. Using a combination of acute and awake-behaving experiments in the rat vibrissa system, coupled with sub-cortical patterned electrical and optical (optogenetic) stimulation, voltage sensitive dye (VSD) imaging of cortical activation, ideal observer and information theoretic analyses, and precise sensory behavioral tasks, we are uniquely positioned to develop a framework for characterizing, optimizing, and controlling information flow in sensory pathways induced through artificial activation of neural structures. In this project, we will 1) utilize electrical and optical stimulation of the thalamus in conjunction with n-vivo VSD imaging in cortex of the anesthetized rat to assess the detectability and discriminability of a range of stimuli from the perspective of an ideal observer of cortical activiy, and formulate a principled basis for stimulation design for optimizing discriminability, 2) manipulate the level of thalamic depolarization/synchrony through background depolarization/hyperpolarization using optogenetic techniques to shape the nonlinear response properties of the circuit and develop control of the level of detectability and discriminability among sets of inputs, and 3) develop a behavioral framework for artificial manipulation of the trade-off between detectability and discriminability to optimize performance in different contexts. Significance: The development of artificial means by which to activate sensory circuits for impaired individuals is thus an extremely important public health issue, and the development of principles for controlling information flow in sensory pathways is of paramount importance, but currently does not exist. Furthermore, being able to activate downstream brain structures in a precise manner is also critical in the basic scientific investigation of how populations of neurons represent information and how these representations are propagated across brain regions - to understand the neural code. Innovation: The conceptual innovation here is the use of the link between performance of an ideal observer of cortical activity and behavior as a design and optimization tool for artificial stimulation. We use an innovative combination of in-vivo electrophysiology with multi-electrode, mult-site recording, voltage sensitive dye imaging, patterned microstimulation and optogenetics, highly specific behavioral tasks, and a control-theoretic framework, which together do not exist in the scientific community.

? DESCRIPTION (provided by applicant): Neurons communicate information through fluctuations in the electrical potentials across their cellular membranes. Whole-cell patch clamping, the gold standard technique for measuring these fluctuations, is something of an art form, requiring great skill to perform on only a few cells per day. Thus, it has been primarily limited to in vitro experiments, a few in vivo experiments, and very limited applications in the awake brain. Dr. Forest (and collaborator Dr. Boyden at MIT) developed a robot that automatically performs patch clamping in the living brains of mice by algorithmically detecting cells through analysis of a temporal sequence of electrode impedance changes. Using it, they have demonstrated good yield, throughput, and quality of recording in mouse cortex and hippocampus. With this 'autopatching' robot enabling routine access to electrical and molecular properties of neurons, systematic and scalable in vivo experiments as well as fundamentally new kinds of single-cell analyses have become possible. In the past 12 months, the team has installed 15 autopatchers in academic research laboratories, garnered worldwide media coverage, and led to Dr. Forest's and Dr. Boyden's invitations to President Barack Obama's announcement of the BRAIN Initiative. There are currently no published experiments demonstrating in vivo intracellular recordings of two or more neurons that are synaptically connected. We propose to utilize the autopatcher to target anatomically well-studied sub-circuits to significantly increase the odds of identifying synaptically connected pairs. Specifically, we wil utilize the thalamocortical circuit in the mouse vibrissa/whisker pathway as a model experimental system, where there is a substantial convergence of projections from the thalamus to the input layer in the somatosensory (tactile) cortex. The Stanley Laboratory has extensive experience with stimulation and electrophysiological recordings in this circuit, and is one of only a few laboratories that has successfully recorded from synaptically connected pairs of neurons using extracellular techniques. Thus we aim to demonstrate and characterize the first simultaneous intracellular recording of a functional circuit in the anesthetized and awake living mouse brain to reveal its neural network dynamics. In this 36 month program, the labs of Prof. Stanley and Forest, supported by two postdoctoral researchers, two graduate research assistants, a research engineer and five undergraduates, with assistance from ten graduate students working on related projects, will develop single (Aim 1) and dual (Aim 2,3) autopatching robots for the anesthetized and awake brain. Success will allow, for the first time, quantification of synaptic efficacy in the living brain, crucial for understanding normal and pathological function. Just as molecular biology has greatly benefited from the revolution in in vitro automation, we believe that neuroscience will greatly benefit from the revolution in in vivo automation that we have launched, and here propose to extend.

? DESCRIPTION (provided by applicant): To address the core question underlying the Obama Brain Initiative to better understand the function of complex brain circuits, we propose a multi-scale recording and data analysis project to study the dynamical interactions between sensory cortex, motor cortex, and the basal ganglia in the process of motor planning and execution. The multi-scale approach will involve simultaneous recordings at the cellular, network, and systems level in head-fixed behaving mice trained to perform a rewarded locomotor task. Sensory stimuli delivered to the whiskers will denote GO or STOP cues, and resulting brain processes initiating or suppressing movement will be analyzed. At the cellular level, in vivo whole cell recordings employing autopatcher technology will yield detailed information on the membrane potential trajectory of individual neurons in the sensory and motor cortex in this task. At the network level multiple single unit and local field potential (LFP) recordings will allow the assessment of local population dynamics across multiple layers of cortex and for thalamo-cortical interactions. At the systems level, voltage imaging of the cortical surface using novel transgenic voltage sensing proteins will allow the study of spatio-temporal dynamics of macroscopic activity patterns with a frequency resolution of up to 200 Hz. Recording data simultaneously will allow for a multi-scale analysis of the relations between cellular and network dynamics. For example, the relationship between fluctuations in the field potential and the membrane dynamics of single neurons will be analyzed and is expected to yield important insights into population coding. Similarly, the relation between activity maps obtained with imaging and oscillatory network activity revealed by LFP recordings of cortex is expected to result in important insights into the organization of motor planning. Our work will pay specific attention to the role of beta band (12-35 Hz) oscillations in the control of the observed behavior, because beta oscillations have been implicated convincingly both in cortical sensory processes as well as motor control. Further, beta oscillations are pathologically overexpressed in the basal ganglia of Parkinson's patients and 6OHDA lessoned rodent animal models of Parkinsonism with a likely source in motor cortex. Thus, our guiding hypothesis is that beta oscillations provide an important scaffold to the communication between brain areas in the process of motor planning and execution. To test the causal relation between beta oscillations and motor processing we will artificially induce beta band activity with ontogenetic stimulation of basal ganglia efferent, sensory cortex, or motor cortex and analyze resulting changes in behavior and brain dynamics in stimulated and non-stimulated areas. Overall, these studies will raise the level of systems neurophysiology of motor processing in the behaving rodent to a new level, and are expected to provide fundamental insights into the organization of brain activity across multiple scales. These insights will be invaluable in studies of pathological brain dynamics in neurological disorders affecting the basal ganglia such as Parkinson's disease, Huntington's disease and OCD.

Emory and Georgia Tech have steadily grown the number of faculty involved in computational neuroscience over the past 15 years. The research of these faculty stretch from cellular to systems and theoretical approaches. In 1997 the two Institutions formed a joint Department of Biomedical Engineering, further strengthening the highly collaborative atmosphere between researchers on both campuses. In addition both campuses have a strong track record both in undergraduate and graduate teaching. The proposed training program in computational neuroscience aims to capitalize on these strengths by formalizing an integrated approach to class work and research on both undergraduate and graduate levels. The strong NIH and NSF funded research programs of more than 15 principal investigators identified as computational neuroscientists range from detailed cellular computer simulations of neural dynamics to engineering approaches and the quantitative study of disease mechanisms underlying important disorders such as epilepsy and Parkinson's disease using computational methods. Therefore students will be exposed to multiple levels of approaches aimed ultimately at addressing medical questions. A highly qualified and diverse applicant pool for student fellowships under this program exists on both undergraduate and graduate levels, and will bring some applicants with a primarily background in the biological sciences to integrate computational approaches into their research, and vice versa brings more computational or theoretically oriented applicants in touch with biological experimental research. The program encompasses a cohort of 6 undergraduate and 6 graduate student fellows, who will absolve a rigorous curriculum in neurobiology and mathematical and computational methods through a core sequence of required classes as well as individually chosen electives. Undergraduate fellows will be funded for a period of two years in their junior and senior years, during which they will undertake specialized class work and research in a computational neuroscience lab. Undergraduate trainees will be primarily recruited from the Emory Neuroscience and Behavioral Biology and the Georgia Tech Biomedical Engineering majors, who bring a biological and quantitative strength to the program, respectively. Over 200 students join these majors annually, and we will only take applicants with a GPA of 3.5 or better and expressing an interest in future research graduate training. The graduate students in this program will be recruited from the applicant pools for the Neuroscience and Biomedical Engineering programs, which together receive more than 120 highly qualified applications each year. A special track for fellows in computational neuroscience will be announced on the program websites, that will also link to an extensive independent website describing this program. Graduate students will be funded for the first two years of their education, and then obtain individual training grants or be funded by research grants.

Thalamocortical state control of tactile sensing: Mechanisms, Models, and Behavior Despite the fact that the sensory thalamus plays a major role in shaping sensory representations in cortex, and thus shaping our percepts, most of what we know has been determined through electrophysiological investigation of the thalamus in-vitro or in the anesthetized brain. Properties of thalamic activity such as mean firing rates, timing and synchrony, and tonic/burst firing directly determine how sensory inputs are represented in the spatiotemporal activation of cortex. Modulations in thalamic ?state? through changes in baseline levels of depolarization strongly influence the gating properties of the dynamic relay of sensory signals to cortex during normal behavior. Using the vibrissa pathway of the awake mouse, our team is uniquely positioned to precisely quantify and control thalamic state, and measure the downstream impact on spatiotemporal cortical representations, using a range of multi-scale electrophysiological and optical measurements, causal manipulations, modeling, and sensory behavioral tasks. We will first Determine Thalamic State Control of Sensory Information Processing in the awake mouse (Aim 1). A range of separate experiments will utilize single unit and LFP recording across thalamus and S1 and widefield genetically expressed voltage sensor imaging in S1 to fully capture and model the range of modulations in thalamic firing, synchronization, and tonic/burst firing in the awake brain. We will then conduct experiments that parallel Aim 1 in which we optogenetically manipulate thalamic state, to determine the causal role of thalamic firing modes on spatiotemporal representations of sensory inputs in cortex (Aim 2). Finally, we will Determine Thalamic State Control of Sensory Percepts in behavior, in a well-defined whisker detection and spatial (two-whisker) discrimination tasks (Aim 3), employing the same electrophysiology/imaging and optogenetic manipulation approaches across thalamus and cortex as in Aims 1 and 2. Significance: The thalamocortical circuit is continuously controlled by modulatory inputs that fundamentally shape information processing relevant for perception and behavior. However, the precise link between thalamic state and the resultant percept remains a major open question in neuroscience. We will determine how cortical representations changes through modulation in thalamic input and the consequences of this on perception. Broad Impacts: Dysfunction of brain state has been implicated in an incredibly wide range of neurological disorders ranging from dysfunction of arousal in narcolepsy to dysfunction of neuromodulators in mood disorders such as depression. Furthermore, recent genome wide association studies have implicated voltage-gated calcium channels found in brain structures including thalamus and cortex as risk loci for both schizophrenia and bipolar disorder. Finally, we also assert that understanding of the interaction between brain state and sensory representations is requisite for delivery of surrogate inputs in prostheses.

Despite substantial progress characterizing neural responses, it is particularly challenging to determine causal interactions within recurrently connected circuits due to the confounding influence of the interconnections. This proposed project pioneers a nascent field of closed-loop computational neuroscience that enables real-time feedback stimulation during experiments to decouple recurrently connected elements and make stronger causal inferences about their interactions. Specifically, the contributions of this project will include: Aim 1) Using modern unsupervised machine learning methods to fit latent state dynamical system models of population responses under closed-loop stimulation. The developed techniques will be used to clamp firing rate in genetically targeted inhibitory interneurons across S 1 cortical laminae in the mouse to map the causal effect of inhibitory cells on the sensory gain in excitatory cells. Aim 2) Merging and extending tools from network feedback control and causal inference to identify functional connections between network nodes using realistic experimental constraints. These techniques will be used to clamp firing rate in different S1 laminae of the mouse, using distributed perturbations to identify the functional connectivity between microcircuit layers during sensory stimulation. Aim 3) Developing a large-scale computational modeling environment to serve as an in si\ico testbed for the community. Significance: The proposed project changes the de facto standard use of stimulation in experiments to leverage the full power of new recording and s.timulation technology for decoupling recurrently connected variables and making stronger causal inferences. Broader impacts: While the project uses rodent somatosensation as a model system, the results of this project will provide new techniques to study neurologic disorders involving disfunction of recurrent circuits (e.g., epilepsy, Parkinson's disease and depression). The open-source implementations will constitute critical algorithmic infrastructure for closed-loop stimulation experiments. This project will also result in the production of new trainees in an emerging new interdisciplinary field of closed-loop computational neuroscience. RELEVANCE (See instructions): The neural circuits that fail in many neurologic disorders (e.g., epilepsy, Parkinson's disease and depression) are difficult to study because they involve complex feedback loops. This project will develop algorithms that combine measurements and stimulation in real-time to provide powerful new tools to uncover the operating principles of these circuits and change their operation. Discovery in this area can help improve understanding of neurologic disorders and development of new stimulation therapies.

The proposal initiates a new, innovative training program in Computational Neural Engineering at Georgia Tech and Emory University to train the next generation of researchers at the intersection of computational neuroscience, data science, and clinical neurophysiology. It addresses the opportunities provided by the explosion of tools for measurement and manipulation of nervous system function and the challenges posed by the growing threat of neurological diseases and disorders on an expanding senior population. The program leverages past successes in federally-funded training efforts that have helped to catalyze rapid and recent growth in research and education in Computational Neural Engineering across Emory and Georgia. Early exposure to the intersection of fields is critical to the program mission. Interdisciplinary training in the first two years of the PhD program will provide trainees with unique opportunities for training across axes that span basic to clinical neuroscience, and from neural engineering to computational neuroscience, data science, and machine learning. Two graduate students per will be recruited from the applicant pools for the Biomedical Engineering (GT and Emory), Bioengineering (GT), Electrical and Computer Engineering (GT), and Machine Learning PhD programs, which collectively enroll over 200 PhD students per year. None of the participating programs offer research rotations or funding in the first year of graduate school. We will support a total of four students per year over a five-year period, providing two years of support for two entering students per year. funding such support will help attract the highest quality students to the program, and offer trainees the unique opportunity to rotate through research labs and establish new collaborative research projects. In our prior training experience, such interactions have led to new collaborations funded through fellowships and new research grants. Didactic training will complement core training in each PhD program with existing and new courses in computational neuroscience, neuropathology and neuroengineering, and a new course providing students with an immersive clinical experience at the Emory Brain Health Center. Extracurricular training includes Innovation Forums for clinician/engineering interactions, and a wide variety of seminars, methods clinics, and journal clubs. Trainees will also be provided with professional development for this new generation of researchers, including training in leadership, mentorship, neuroethics, and public scholarship. Trainees will also learn the growing industry in neural engineering, and will have opportunities for internships. Importantly, with solid preliminary evidence for the success in all of these ventures, this program targets an imperative area for growth.

Project Summary/Abstract The brain comprises two distinct hemispheres mainly connected to each other by the corpus callosum. Despite the proven role of callosal connections in sensorimotor behaviors involving both sides of the body, the nature, area and cell-type specificity, and the function of interhemispheric interactions in sensory perception are far from understood. The proposed project will investigate those issues in the context of active bilateral vibrissa-mediated somatosensation in the mouse, focusing on primary (S1) and secondary (S2) somatosensory cortices and their associated callosal projections. The work will combine laminar extracellular recordings in S1 and S2, optogenetic identification of callosal neurons, and chemogenetic modulation of callosal inputs with decoding and signal detection analytic approaches to relate neuronal activity to stimulus encoding and perception while the mouse actively whisks to contact static poles located on each side of the face. The first aim of the work is to investigate the existence and nature of bilateral tactile signal interactions in S1 and S2 and their relation to sensory-evoked callosal activity, as well as to establish whether bilateral features of tactile stimuli can be decoded from S1 and S2 activity. The second aim of the work is to reveal the contribution of homologous callosal neurons of S1 and S2 to the discrimination of bilateral tactile stimuli, and to determine whether S1 and S2 activity encodes behaviorally relevant bilateral stimulus properties predictive of stimulus perception. Significance: The proposed work will provide unprecedented information about the functional role of specific populations of neurons constituting the corpus callosum. It will allow to reconcile investigations about the locus and the nature of interhemispheric interactions conducted in reduced preparations with studies indirectly probing their behavioral relevance. Furthermore, this work will provide a more complete understanding of somatosensation, identifying area and cell types critical for bilateral tactile perception. It will also establish a novel sensory coding framework, encompassing tactile signals arising from both sides of the body. Broad Impact: This project lays the groundwork for future investigations on the role of interhemispheric interactions in the coordination of motor behaviors and in cognitive processes implicating the left and right cerebral hemispheres. Additionally, it will enable targeted functional investigations at the cellular and network level of several mental disorders characterized by abnormal anatomical and functional interhemispheric connectivity, such as ADHD, autism and schizophrenia. Separately, knowledge about cerebral cross-areal communication and its detailed computations will also be beneficial to neuroengineering applications, particularly those regarding movement coordination and natural integration of sensory feedback for bilateral neuroprostheses.

Despite substantial progress characterizing neural responses, it is particularly challenging to determine causal interactions within recurrently connected circuits due to the confounding influence of the interconnections. This proposed project pioneers a nascent field of closed-loop computational neuroscience that enables real-time feedback stimulation during experiments to decouple recurrently connected elements and make stronger causal inferences about their interactions. Specifically, the contributions of this project will include: Aim 1) Using modern unsupervised machine learning methods to fit latent state dynamical system models of population responses under closed-loop stimulation. The developed techniques will be used to clamp firing rate in genetically targeted inhibitory interneurons across S 1 cortical laminae in the mouse to map the causal effect of inhibitory cells on the sensory gain in excitatory cells. Aim 2) Merging and extending tools from network feedback control and causal inference to identify functional connections between network nodes using realistic experimental constraints. These techniques will be used to clamp firing rate in different S1 laminae of the mouse, using distributed perturbations to identify the functional connectivity between microcircuit layers during sensory stimulation. Aim 3) Developing a large-scale computational modeling environment to serve as an in si\ico testbed for the community. Significance: The proposed project changes the de facto standard use of stimulation in experiments to leverage the full power of new recording and s.timulation technology for decoupling recurrently connected variables and making stronger causal inferences. Broader impacts: While the project uses rodent somatosensation as a model system, the results of this project will provide new techniques to study neurologic disorders involving disfunction of recurrent circuits (e.g., epilepsy, Parkinson's disease and depression). The open-source implementations will constitute critical algorithmic infrastructure for closed-loop stimulation experiments. This project will also result in the production of new trainees in an emerging new interdisciplinary field of closed-loop computational neuroscience. RELEVANCE (See instructions): The neural circuits that fail in many neurologic disorders (e.g., epilepsy, Parkinson's disease and depression) are difficult to study because they involve complex feedback loops. This project will develop algorithms that combine measurements and stimulation in real-time to provide powerful new tools to uncover the operating principles of these circuits and change their operation. Discovery in this area can help improve understanding of neurologic disorders and development of new stimulation therapies.

The brain is a complex dynamical system, with a hierarchy of spatial and temporal scales ranging from microns and milliseconds to centimeters and years. Activity at any given scale contributes to activity at the scales above it and can influence activity at smaller scales. Thus a true understanding of the brain requires the ability to understand how each level contributes to the system as a whole. Most brain research focuses on a single scale (single unit firing, activity in a circuit), which cannot account for the constraints imposed by activities at other scales. The goal of this proposal is to develop a framework for the integration of multiscalar, multimodal measurements of brain activity. One of the challenges in understanding how activity translates across scales is that features that are relevant at one scale (e.g., firing rate) do not have clear analogues at other scales. We address this issue by defining trajectories in ?state space? at each scale, where the state space is defined by parameters and time scales appropriate to each type of data. The trajectory of brain activity through state space can uncover features like attractor dynamics and limit cycles that characterize the evolution of activity. Using machine learning along with new and existing multimodal measurements of brain activity (MRI, optical, and electrophysiological), we propose to establish methods that relate trajectories across scales while handling the mismatch in temporal sampling rates inherent in multi-scale data. Specific aims are 1. Create and test a tool for learning how trajectories at fast scales influence activity at slower scales. Different modalities have different inherent temporal resolutions in addition to different types of contrast. Current methods generally downsample the faster modality in some way, losing much information in the process. We will leverage variants on long short-term memory (LSTM) network architectures to learn the relationship between state space trajectories acquired simultaneously with population recording and optical imaging, and with optical imaging and fMRI. 2. Create and test a tool for learning how trajectories at slow scales influence activity at faster scales. Leveraging the same LSTM-based approach, we will learn how slower, larger scale activity affects activity at smaller scales, using whisker stimulation as a test case. We anticipate inclusion of the large scale activity (measured with fMRI or optical imaging) will improve prediction of the response at smaller scales (measured with optical imaging or population recording). Our work will allow us to begin to answer a wide range of questions about how the brain functions (e.g., what type of localized stimulation that will drive the brain to a desired global state? How does modulation of the global brain state affect local information processing?) and provide guidance for future experiments by identifying key features that influence activity across scales. By approaching the whole brain as a complex dynamical system, we will break free from the limitations of previous studies that focus on individual cells or circuits. We also expect our work to stimulate new theories that incorporate multiple scales of activity.