Richard A. Andersen - US grants

Two of the most interesting functions of the visual motion system are the analysis of optic flow for heading perception and the determination of 3D structure-from-motion (SFM). This proposal will examine how these functions are accomplished by visual cortex using single cell recording techniques in behaving monkeys. The first specific aim is to examine the unsolved problem of how flow generated by eye and head movements is subtracted from flow generated by observer translation in order to recover the direction of heading. Pilot experiments for the proposal indicate that eye pursuit signals shift the spatial tuning curves of flow sensitive neurons in the dorsal division of the medial superior temporal area (MSTd). This shift enables MSTd neurons to code heading direction irrespective of whether the eyes are still or moving. Experiments are also proposed to determine whether head movements also lead to spatial tuning shifts, and to examine whether the head movement signals are derived from efference copy, vestibular, and/or proprioceptive sources. It will also be determined whether MSTd neurons code heading in eye-, head-, body-, or world-centered coordinates. The second specific aim is to examine the neural mechanisms for 3D SFM perception. Monkeys will be trained to report the perceived direction of rotation of cylinders defined by disparity and motion cues and cylinders defined only by motion cues. These latter cylinders are perceived as 3D, but the perceived direction of rotation is bistable. Recordings will be made from the middle temporal area (MT) while animals perform this task. Pilot studies indicate that MT activity is correlated with the perceived rotation direction of the bistable stimulus, with cells giving different responses for the same physical stimulus depending on the percept. Experiments are also planned to examine a 3 stage model for SFM processing: the first stage measures motions and likely occurs in V1; the second stage segregates and depth orders surfaces and likely occurs in MT; the third stage uses speed gradients to calculate 3D shape and may occur in MT or MST. Recordings will be made in V1, MT and MST using variations of the bistable cylinder paradigm to determine where these three stages are located in the motion pathway. The experiments in this proposal will determine the neural mechanism for heading computation during eye and head movements and whether MSTd is directly involved. They will also determine if monkeys perceive SFM and the role of V1, MT, and MST in this important perceptual process.

DESCRIPTION: Lesions of the posterior parietal cortex (PPC) in humans and monkeys produce deficits in visual attention, spatial perception, and the ability to make accurate movements. These deficits are the result of losing the cortical pathway which enables visual information to be transformed into action. This proposal will examine how this transformation is accomplished by recording the activity of neurons in the PPC of monkeys while they perform various visual-motor tasks. The proposal will focus on whether a component in PPC is already coding movement intentions, whether these intentions are coded in the motor coordinates of the movement being planned, and how the different sensory signals that converge on PPC are transformed into various spatial coordinate frames. The role of gain fields (the modulations of sensory signals by eye, head and limb position) in transforming between coordinate frames will be examined. The first specific aim will examine movement intention and will determine if cells in area LIP are specifically engaged in saccade tasks, and if cortical areas around LIP are specifically engaged in reach tasks. The second specific aim will determine if reach-related areas are coding visually derived signals in limb coordinates. Such a finding would indicate that visual signals have been transformed into motor coordinates for the purpose of moving the limbs. The third specific aim will examine how head position signals are combined with eye position and visual (retinal position) signals to code the spatial locations of objects. It will be determined if vestibular signals, which convey information about the location of the head in the world, affect the visual response of PPC neurons; these vestibular gain fields could potentially encode locations in world-centered coordinates. Likewise it will be determined if neck proprioceptive or efference copy signals, conveying the orientation of the head on the body, modulate visual responsiveness and potentially encode locations in body-centered coordinates. Finally the role of optical flow and visual landmark cues in coding the locations of visual stimuli in world-centered coordinates will be examined. The fourth specific aim will be to determine how visual and auditory signals are combined in PPC to code spatial locations independent of the modality of the stimulus. These experiments, particularly the third and fourth aims, are designed to answer the long standing question of how different modalities are combined into a common framework in parietal association cortex. Overall these experiments will significantly further our understanding of how perceptions are transformed into actions.

DESCRIPTION (provided by applicant): Advances are being made in the field of neural prosthetics for applications directed toward restoring function to those suffering from paralysis. A central question in this research is which areas to target for control signals. Initial efforts hve naturally focused on the primary motor cortex (M1) given its strong linkage to motor execution. These studies have extracted motor command signals for reconstructing the moment-by-moment kinematics of limb movement. Sensorimotor cortical areas one or two steps removed from motor cortex, including the posterior parietal cortex (PPC), have been examined in recent years for prosthetic control signals of a more cognitive nature related to the goals and context of movement. Despite the relatively high correlation between the M1 neural activity and kinematics of actual movements, controlling a virtual prosthetic device using this activity in brain-control experiments has produced much less accurate and slower movements than natural limb movements. Increasingly more studies report that accuracy and speed of the prosthetic movement can be significantly improved by incorporating more cognitive signals such as the intended goal when decoding the moment-by-moment kinematic information. We have found two regions of PPC, the parietal reach region (PRR) and the dorsal aspect of area 5 (area 5d) that, besides providing goal signals, also provide trajectory signals. The dynamics of the trajectory signal in PPC suggest that, rather than being a movement command signal similar to M1, it represents a state estimate of the limb movement. Although the decode performance for trajectories appears good in PPC, it is difficult to compare it to trajectory decodes from M1 from previous studies due to differences in the tasks, experimental conditions, and data analysis methods. Aim 1 will directly compare the representation of trajectories in PPC (PRR and area 5d) with M1as a benchmark. These studies will be performed in the same animals performing the same tasks under the same experimental conditions. If PPC can provide trajectory signals of similar fidelity to M1 then it would be an ideal location for obtaining both trajectory and goal signals. If the number of implant sites in patients is limited, these experiments would provide insight into the best sites to extract a variety of control signals. Aim 2 will compare neural adaptations in M1 and PPC to novel motor effector dynamics to infer the functional role of the trajectory signal of each area in motor skill learning. Understanding how the different brain areas adapt will be important for choosing sites for neural prosthetics that require learning the dynamics of mechanical devices.

DESCRIPTION (From the Applicant's Abstract): Although the posterior parietal cortex (PPC) has long been appreciated to be an area for association of different sensory modalities for spatial awareness and for spatial attention, recent studies have pointed to an additional, and very major, role of this area in movement planning. One important advance has been the fmding that there is an anatomical map of intentions within the PPC, with different areas specialized for different behaviors. In Aim 1 we will determine the inter-relationship between some of these areas in movement planning. In particular, we will examine whether area LIP has an executive role in decision making for both eye and limb movements, or if LIP is primarily involved in the former, and areas MIP and 5 in the latter. Another important advance has been the elucidation of the spatial representations in some PPC cortical areas. These results have led to a general proposal that the early stages of movement planning are performed in eye-centered coordinates in primates, regardless of the sensory input or eventual motor output. However, the activity of many PPC neurons is gain modulated by eye, head and limb position. Aim 2 will examine whether these PPC "gain fields" may effect the decisions to make eye and hand movements. Classically it has been believed that the coordinate transformation for reaching results from the transformation of the traget object from retinal to body coordinates, and then the subtraction of the object location in body-coordinates from the location of the hand in body coordinates to generate the motor error in limb coordinates. Our finding that early movement planning occurs in eye-coordinates in PPC suggests an alternative scheme, in which the hand is represented in eye coordinates and is subtracted from the location of the target in eye coordinates to produce the motor error signal. We will test this new idea in aim 3. Lastly, in aim 4 we will examine the spatial representation of sound location in an auditory area, Tpt, that provides input to the PPC. These experiments are designed to determine the stage in auditory processing pathway where head-centered auditory fields are converted into eye-centered auditory fields similar to those encountered in the PPC.

DESCRIPTION (PROVIDED BY APPLICANT): The purpose of this proposal is to design and build a smart MEMs device for recording multicellular activity in the visual nervous system. This miniaturized and implantable microelectrode array system will automatically adjust the depths of the electrodes to optimize recording performance. The rationale for this system is to improve long term, chronic recording from populations of neurons in the visual system for scientific and neuroprosthetic applications. Current systems with fixed electrode geometries cannot be adjusted to optimize recordings in terms of yield or cell type. Moreover, these systems cannot "follow" cells over time to overcome loss of signal due to movement of the tissue with respect to the electrodes. Manual systems have the drawback of becoming unmanageable for large arrays in scientific studies, and unacceptable for permanent implants for prosthetics applications. The proposed system overcomes all of these drawbacks by automating the position of each electrode based on recorded signal quality. The first aim of this study is to develop algorithms to automatically search for, and hold, recorded signals from neurons. Preliminary data show that this is possible for cortical neurons recorded from awake, behaving monkeys and anesthetized rats. The second aim is to develop MEMS actuators that have low heat dissipation, are lockable with minimal energy application, and can provide large displacements. Preliminary studies indicate that electrolysis actuators developed by one of the co-investigators are ideal for this application. Aim three will integrate all of the components developed in the earlier aims into a single system. This device will consist of a MEMS-based multielectrode array with independently moveable electrodes, and on-board control, monitoring and communications circuits. Although this is a proposal to develop technology, an underlying hypothesis is that this system will improve multiunit, chronic recordings substantially over what can be achieved with fixed geometry systems. This hypothesis will be tested initially with cortical recordings in anesthetized rats followed by recordings from visual cortical areas in behaving monkeys.

This three-year grant will purchase two pieces of equipment for magnetic resonance imaging of the brain at the California Institute of Technology. One equipment piece provides the best resolution in a horizontal orientation; the second provides imaging of behaving monkeys in vertical position. This will provide state-of-the-art tools for investigating brain structure and function in monkeys with noninvasive methods, and also provide opportunities for imaging post-mortem human brains. The technology will make possible a set of research studies on how the brain processes information, including how it sees faces, how it weighs different choices, and how it makes decisions and guides action. These are important questions in neuroscience, and the new equipment will greatly enhance science at the Caltech Brain Imaging Center. The grant will also provide opportunities for training of students and post-docs on the new equipment. This will include classes taught at Caltech as well as participation in individual research projects. The development of these new scientific tools will lead to a better understanding of how the brain works, and how it is "wired up." That knowledge, in turn, will contribute to efforts to build artificially intelligent systems. Taken together, the cutting-edge science enabled by the new equipment, and the training of the next generation of young scientists on it, will contribute substantially to cognitive neuroscience in America and worldwide.

DESCRIPTION (provided by applicant): This application is to study effector choice in parietal-frontal circuits. It focuses on movement plans and decisions regarding which part of the body to use to obtain a goal. Although this is a very common behavior and important for daily activities such as typing, playing musical instruments, sports, and operating an automobile, to our knowledge effector choice has generally not been studied at the neuronal level. Aim 1 will examine 2 models of cortical processing for effector selectivity. The parallel model hypothesizes that all cortical areas that represent potential movement plans before selection are the same areas that represent the outcomes of those decisions. The serial model hypothesizes that some cortical areas represent potential plans and outcomes and later cortical areas only represent the outcomes. Aim 2 will examine the dorsal premotor cortex (PMd) to see if it is similar to the parietal reach region (PRR) in representing both potential plans and outcomes. It will also determine whether PMd cells are mostly selective for reach outcomes (like PRR) or if it contains a mixture of reach and saccade selective neurons. The latter finding would suggest that PMd may be further along in the pathway for hand-eye coordination than PRR, which is suggested by other recent studies examining coordinate frames. Although PRR and the lateral intraparietal area (LIP) have recently been found to have activity consistent with their representing potential plans and outcomes of effector decisions, there is no evidence that they are actually necessary for effector decision making. Aim 3 will test whether LIP and PRR are necessary by inactivating them during effector decision making. If inactivation of these two areas produce a bias in the decision task (LIP for saccades and PRR for reaches), then these areas would be within the network for effector decision making. The three aims will provide important new information regarding how effector planning and decisions are processed including the functional hierarchy of the areas involved (Aim 1), the role of dorsal premotor cortex (Aim 2), and whether areas in the parietal cortex are involved in the decision process or only reflect the potential plans and the outcome of the decision (Aim 3). PUBLIC HEALTH RELEVANCE: Results from this study can be used to help design therapies for patients suffering from damage to frontal and parietal cortex from strokes and traumatic brain injuries. They will help in understanding deficits that result from neurological diseases that effect cortical functioning, and in guiding diagnoses and treatments for these diseases. They will also be useful for guiding the design of neural prosthetics. By determining the types of signals that can be obtained from different brain regions, neural implants can be made to read out these signals in order to control assistive external devices for paralyzed patients.

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).

This NSF Major Research Instrumentation (MRI-R2) Award will enable a three-year grant to purchase an upgrade for a single piece of equipment for imaging the human brain at the California Institute of Technology. The upgrade, a 32-channel Total Imaging Matrix upgrade of a Siemens 3.0 Tesla MRI scanner, will substantially improve the resolution, the speed with which experiments can be done, and the kinds of imaging sequences that can be programmed. Taken together, these major enhancements will enable a range of questions about the structure, connectivity, and functioning of the human brain. Researchers at Caltech, in collaboration with a national and international consortium of scientists, will use the equipment to investigate how the brain makes financial decisions, how social information such as faces are processed, and how brain-machine interfaces can be built to decipher information from the brain to guide robotic prostheses. These are important, big open questions in neuroscience, and the new equipment will greatly enhance science at the Caltech Brain Imaging Center.

The grant will also provide opportunities for training of students and post-docs on the new equipment. This will include classes taught at Caltech as well as participation in individual research projects. The development of these new scientific tools will lead to a better understanding of how the brain works, how it is wired up, and how it may dysfunction in disease. That knowledge, in turn, will contribute to efforts to build artificially intelligent systems. Taken together, the cutting-edge science enabled by the new equipment, and the training of the next generation of young scientists on it, will contribute substantially to cognitive neuroscience in America and worldwide.

DESCRIPTION (provided by applicant): The objective of this application is to assess, in human posterior parietal cortex (PPC), the efficacy of both microwire-based array technology and decoding algorithms for neural prosthetic applications. An outcome of this work is a human-approved microwire array technology capable of reaching deep sulcal areas of the cortex. In animals, the posterior parietal cortex is an area that we have shown to encode both the reach target (goal) location and real time dynamics in point-to-point arm reaching tasks. It is our intention to show that, in the human, these features are also encoded and that goal and dynamic information can be combined for more accurate decoding of movement intentions. In addition, our research with animals has demonstrated that the PPC encodes a number of cognitive variables that could be potentially useful for neural prosthetic applications. Tasks will be designed to see if these cognitive signals can also be recorded and decoded from human PPC. These tasks will examine 1) rapid decoding of movement sequences; 2) decoding higher level aspects of goal information that are symbolic and non-spatial; 3) local field potentials to increase decode accuracy and provide a foundation for cognitive state decoding; 4) context decoding; 5) learning as a function of practice for goal and trajectory decoding and 6) learning novel effecter dynamics. We hypothesize that these cognitive aspects of animals' PPC are also available in the corresponding human PPC. Our five year goal for this grant is to complete the preclinical testing for an investigational device exemption (IDE) to the Food and Drug Administration, submit and gain approval for an IDE, obtain IRB approvals and design the behavioral tasks and data analysis that will be used in subsequent human clinical studies, and perform a clinical trial with two subjects. To this end, we have put forth five specific aims: (1) to perform a biocompatibility assessment of the technology per recommended standards, (2) to perform a histological assessment of the technology following chronic implantation, (3) to perform a safety and efficacy assessment in a non-human primate model, (4) to test the performance of decoding algorithms that will be used in humans, and (5) to assess the performance of our technology and cognitive decoding algorithms in paralyzed individuals under an FDA approved Investigational Device Exemption Clinical Trial. PUBLIC HEALTH RELEVANCE: This application has direct relevance to public health since its goal is to perform clinical trials to test a neuroprosthetic medical device for implantation in posterior parietal cortex. The clinical trials are designed to help patients with severe paralysis, which can result from spinal cord lesion and other traumatic accidents, peripheral neuropathies, amyotrophic lateral sclerosis, multiple sclerosis, and stroke.

This Project investigates the basic computations at work in making simple social decisions, and contrasts them with simple non-social decisions (e.g., ones based on the value of juice or money, rather than the value of other people). It sets the stage for all the others in investigating how social reward is represented and compares to nonsocial reward. An example of a non-social decision is choosing what to drink by pushing one of several buttons on a soda dispensing machine, an example of a social decision is choosing what person to call to go on a date. Here we address these questions: Are there regions in the amygdala and prefrontal cortex that encode stimulus values at the time of choice, and experienced (hedonic) values at the time of outcome, in the social domain (seeing smiling or beautiful faces), as they do in the nonsocial case (getting juice when thirsty)? Are there neurons specialized for valuation of social stimuli, or do the same cells encode value in social and non-social decisions? Does the valuation of different types of social stimuli require specific sub-circuits? And how are individual differences between people reflected in these processes? We will address these questions by carrying out parallel experiments in humans and rhesus monkeys, using the complementary techniques of fMRI and electrophysiological recording in both species, and using a variety of basic social and non-social stimuli. Comparisons will be made across species and across single-unit, local field potential, and BOLD-fMRI data, as well as with data from the other Projects and across individual differences.

Brain-machine interfaces (BMI) read signals directly from the brain to control external devices such as robotic limbs. While this technology has great potential to benefit people who are paralyzed, BMIs often have poor performance because they use noisy, low-level signals to simultaneously control many aspects of the robotic limb's movements. In contrast, this project will address this shortcoming by reading high-level intents from the brain in order to control an intelligent robotic system. These changes reflect cutting-edge advances in neuroscience and machine intelligence and will require close cooperation between scientists, engineers, and physicians. The project aims to leverage expertise across these diverse fields in order to generate significant improvements in BMI technology to advance the national health, increase scientific understanding of the brain, and lead to dramatic improvements in the quality of life for these severely disabled persons.

This collaborative project will decode high-level cognitive actions from neural signals recorded in the parietal cortex of a tetraplegic human, then carry out those intents using a smart robotic prosthesis. Persons with tetraplegia who have multielectrode arrays (MEA) implanted in reach and grasp areas of the posterior parietal cortex (PPC), will participate in experiments to explore the neural representation of cognitive intentions in human PPC including object selection, action intention, and neural control of robotic limbs. Experimental results will be used to construct BMI control algorithms optimized to decode these cognitive signals. In parallel, a modular, semi-autonomous robotic prosthesis will be developed that can identify household objects and plan reach-and-grasp movements to manipulate or transport the objects. These scientific and technological efforts will be supported by continued clinical care of the tetraplegic participants. The study will explore increasingly capable iterations of the BMI system, culminating in testing of the fully developed BMI system in the participants' own home environment where they will practice activities of daily living. The resulting system will leverage deep insights in cognitive neuroscience and advanced capabilities in machine sensing and robotic control systems to substantially improve the ease of use and capability of brain-machine interfaces.

Project summary/abstract Reach-to-grasp and hand manipulation will be studied in tetraplegic humans with neural recordings from multielectrode arrays (MEAs) and intracortical microstimulation (ICMS) of somatosensory cortex. Recordings will be performed within the cortical grasp circuit with MEAs implanted in two grasp-related areas, the ventral premotor cortex (PMv) and the anterior intraparietal area (AIP) of the posterior parietal cortex (PPC). ICMS will be delivered to Brodmann's area 1 (BA1) of somatosensory cortex. Aim 1 will compare recordings from PMv and AIP to determine how grasping is processed by these two areas. We will test the hypothesis that AIP is more concerned with high-level goals and PMv more with motor trajectories. These results will be important for selecting areas for implantation for future neurosprosthetics development. Aim 2 will examine the tactile cues that can be provided by ICMS. We hypothesize that detection, discrimination, and location of stimuli will be perceived by varying parameters of frequency, amplitude and location of the ICMS. These findings will be used for providing stimulation induced tactile feedback for brain-machine interface (BMI) applications. In aim 3 we will use three tasks that combine decoding neural signals from AIP and PMv with ICMS for somesthetic feedback for a bidirectional BMI. The tasks will be used to test BMI performance (1) for frequency and amplitude discrimination of ICMS without visual feedback (?handbag task?), (2) for learning-based proprioception, and (3) a manipulandum task that implements force feedback for adjusting grasp. We hypothesize that the patients will successfully perform these tasks, indicating that bidirectional BMIs are feasible and can deliver tactile and proprioceptive information for grasp and hand manipulation. We have already developed artifact rejection techniques and demonstrated their application in a non-human primate (NHP) bidirectional BMI task. We also have all of the regulatory approvals for these studies and already have had success with BMI applications in tetraplegic patients using PPC recordings. Thus the outlook is good for successful completion of these aims within 3 years. The results of these aims will advance both the scientific understanding of grasp processing in human cortex and the development and implementation of bidirectional BMIs to assist patients with movement disorders.

Project 4. Project Summary. Project 4 is a continuation of our basic-research intracranial recordings in humans, which were scattered across Aims in multiple Projects in the current funding period and have now been collected in one focused Project. It is co-directed by Richard Andersen and Ueli Rutishauser, both of whom are also on our current Conte Center. Its overarching Aim is to use single-unit recordings in humans to understand the representations and circuits for social inference and its contextual modulation, with an emphasis on comparing this to representations of one's own states. It links to other Projects: Aim 1 will investigate the responses of amygdala neurons to social threat, a question linked to Project 3. Aim 2 will investigate the responses of posterior parietal neurons in representing another person's actions from which we could learn, a question linked to Project 1. These links are reflected in the personnel of Project 4, which includes PIs from other Projects (Adolphs, O'Doherty, Mobbs) as well as shared post-docs and students. The Aims of Project 4 map onto recordings in two distinct patient populations. Aims 1 and 3 will be led by Rutishauser, and tested with single-unit recordings from the amygdala and prefrontal cortex in epilepsy patients (10 per year), through a subcontract to Cedars-Sinai Medical Center. Aims 2 and 4 will be led primarily by Richard Andersen, and tested with single-unit recordings from the posterior parietal cortex in rare patients who have brain-machine interfaces implanted chronically for control of neural prosthetics (tested at a rehabilitation center close to Caltech). Aim 1 will record the responses of neurons in the prefrontal cortex and amygdala to social stimuli such as faces and a range of threats. It will investigate the category-selectivity of neurons, and ask how this is modulated by attention, using concurrent eyetracking. Aim 2 will record primarily in the posterior parietal cortex, but also include some experiments with recordings in amygdala and prefrontal cortex, and examine how the observation and execution of actions are represented by these neurons (a question related to so-called ?mirror neurons?). It also includes a close link to Project 1 in testing whether single neurons encode signals for observational learning. Aim 3 will investigate error signals in a Stroop task by recording from neurons in the anterior cingulate cortex and supplementary motor cortex in epilepsy patients; it will also feature an observational Stroop task, where the subject sees the errors made by another person. This Aim will also be conducted, in the same patients, using fMRI (prior to their electrode implantations). Aim 4 will capitalize on the ability record in posterior parietal cortex in the brain-machine interface patients from over a year, and will compare recordings done over this long period as a function of state variables such as mood.

Loss of walking function and leg sensation are devastating consequences of spinal cord injury (SCI). These deficits have a profoundly negative impact on independence and quality of life of those affected. Moreover, wheelchair reliance after SCI increases the risk of medical complications. The healthcare costs associated with SCI are ~$50 billion/year, presenting a significant public health concern. Currently, there are no biomedical solutions capable of restoring walking and leg sensation after SCI. Clinically practical and socially acceptable solutions to these important problems are desperately needed. Employing a cyber-physical system (CPS) to bypass the damaged spinal cord may be a novel way to restore walking and leg sensation to those with leg paralysis due to SCI. The proposed multi-disciplinary effort will inspire students from traditionally underprivileged and underrepresented groups to pursue college education in STEM fields by demonstrating how engineering and science can make a difference in the well-being of those with disabilities. In addition, it will engage individuals with disabilities, their family members, friends, and caregivers, in educational opportunities in order to increase their scientific and technical literacy. The outreach to these communities will be accomplished by leveraging diverse ethnic makeup of Orange and Los Angeles Counties, geographic proximity of the three study sites, which makes outreach activities amenable to integration, and the high societal significance and visibility of the project.


Impairment or complete loss of gait function and lower extremity sensation are common after spinal cord injury (SCI). A new cyberphysical system, CPS, can be realized as a permanently implantable bi-directional (BD) brain-computer interface (BCI), which translates walking intentions from brain signals into commands for a leg prosthesis, and converts prosthesis sensor signals into electrical stimulation of the brain for artificial leg sensation. This closed-loop operation would come close to restoring able-body-like walking and leg sensation after SCI. Before such an implantable CPS is deployed in humans, its feasibility and safety must be established. The main objective of this Frontier project is to design, develop, and test a wearable analogue of a fully implantable electrocorticogram (ECoG)-based BD-BCI for walking and leg sensation. The BD-BCI CPS will be designed as an ultra-low power modular system with revolutionary techniques for interference mitigation to enable simultaneous electrical stimulation and recording. The first module will consist of a custom brain signal acquisition system that exploits ECoG signal attributes to significantly reduce power consumption. The second module will consist of a low-power processing unit, brain stimulator, and wireless communication transceiver. This module will internally execute optimized BCI algorithms and wirelessly transmit commands to a robotic gait exoskeleton for walking. Comprehensive benchtop and bedside tests will be conducted to assess proper system function. Finally, subjects with paraplegia due to SCI will be recruited to undergo a 30-day ECoG implantation to test the BD-BCI's ability to restore brain-controlled walking and leg sensation. The goals of transition to practice (TTP) are to: (1) develop a fully implantable version of the BD-BCI, (2) perform a series of industrial-standard medical device benchtop tests, and (3) test the implants safety.

Abstract: The long-term objective of this application is to understand cortical processing of sensory to motor transformations within the human cerebral cortex. A vast number of computations must be performed to achieve sensory-guided motor control. Standing out among these computations, visual information of the goals of action must be transformed from the coordinates of the retina to the coordinates of effectors used for movement, for instance limb coordinates for reaching under visual guidance and to world coordinates for interactions in the environment. Once an object is grasped, somatosensory signals from the hand are required for dexterous manipulation of grasped objects. Internal models within the sensory motor pathway are essential for estimating the current state of the body and the external environment, accounting for lags in sensory feedback, and calibrating the body to the environment. We will use the rare opportunity of being able to record from populations of single neurons in a clinical study designed to develop neural prosthetics for tetraplegic participants paralyzed by spinal cord injuries. Cortical implants of microelectrode arrays will be made within three key locations in the sensorimotor system: primary motor cortex, primary somatosensory cortex, and posterior parietal cortex. These microelectrode arrays enable both recording and intracortical microstimulation. We will test the hypothesis that somatosensory and motor cortex represent imagined reaches in hand coordinates, but posterior parietal cortex is task dependent, and its population neural activity can flexibly change coordinate frames to enable encoding of the spatial relations within the body (arm and eyes), between the body and world (arm and reach targets; objects relative to self), and within the world (relative position of objects in the world) as required by task demands. Percepts evoked by intracortical microstimulation and imagined sensations will be used to understand the representation of cutaneous and proprioceptive information within primary somatosensory cortex and posterior parietal cortex. The hypothesis to be tested is that imagined sensation and electrically evoked sensations are highly overlapping?not just in primary somatosensory cortex but also in posterior parietal cortex. Lastly, we hypothesize that the posterior parietal cortex contains in humans an internal model of state estimation that shows plasticity for both natural and brain-control behaviors and transfers this learning to motor cortex. These studies will not only greatly advance our understanding of the human sensorimotor cortical circuit, but also will provide basic knowledge for the design of future neural prosthetics.

SUMMARY Controlling specific neural circuits across large areas of the brain is a major technology goal of the BRAIN Initiative. To achieve this goal, technologies should ideally provide a combination of spatial, temporal and cell- type specificity and be noninvasive to facilitate their translation across animal models and, ultimately, human patients. Here, we propose an approach to modulating neural circuits noninvasively with spatial, cell-type and temporal specificity. This approach, which we have named Acoustically Targeted Chemogenetics, or ATAC, uses transient focused ultrasound (FUS) blood brain barrier opening (BBBO) to transduce neurons at specific locations in the brain with virally-encoded engineered receptors, which subsequently respond to systemically administered bio-inert compounds to activate or inhibit the activity of these neurons. This technology allows a brief, noninvasive procedure to make one or more specific brain regions capable of being selectively modulated using orally bioavailable compounds. In preliminary experiments, we have implemented this concept in mice by using ATAC to noninvasively target AAV9 viral vectors encoding chemogenetic DREADD receptors to excitatory neurons in the hippocampus, and showing that this enables pharmacological inhibition of memory formation. Building on this proof of concept, we will now scale ATAC to work in non-human primates. This goal is particularly important given the relatively limited success of existing technologies, including optogenetics and conventional chemogenetics, in robust behavioral neuromodulation in larger animals. Scaling ATAC to larger animals requires several innovations beyond the core concept, including evolving viral vectors for more efficient and intersectional transfection of neurons with FUS-BBBO, developing ultrasound methods to overcome skull aberrations and enable precise targeting in large animals, establishing ways of confirming the functionality of ATAC non- invasively with functional imaging, and optimizing the selection and pharmacological administration of chemogenetic ligands for large-animal behavioral studies. In this project, we will first establish the basic capabilities of ATAC in NHPs and integrate them with non-invasive functional imaging, setting a baseline for ATAC performance. Then, we will use a pioneering technology for in vivo evolution of viral vectors to develop AAV viruses specifically optimized to efficiently deliver chemogenetic receptors to brain regions targeted with FUS-BBBO. In parallel, we will develop non-clinical image guidance and aberration correction methods to enable precise targeting and verification of FUS-BBBO in NHPs. This will make it possible for academic groups without access to expensive clinical FUS systems to perform ATAC in larger organisms. Finally, as motivating example applications, we will demonstrate that the optimized ATAC paradigm can be used to inhibit multiple distinct brain regions in macaques, reversibly and repeatably modulating their ability to recognize faces and also apply it in a sensorimotor circuit to alter functional connectivity. We will also show its stability, reliability and non-toxicity.

The objective of the proposed research is to obtain scientific knowledge of visuomotor transformations in posterior parietal cortex (PPC) and primary motor cortex (M1) from tetraplegic subjects in a clinical trial to advance the development of neural prosthetics. We have shown in clinical trials conducted over the past 6 years that PPC can control neural prosthetics for assisting tetraplegic subjects. Other groups have concentrated on M1 and likewise find control for neural prosthetics. In our studies of PPC we have found that besides trajectory signals to move robotic limbs or control computer cursors, there are a plethora of visuomotor signals that represent intended movements of most of the body, movement goals, cognitive strategies, and even memory signals. Our central hypothesis is that PPC and M1 will encode visuomotor parameters in both similar and different ways, and that algorithms can be developed to leverage those signals from the two areas that are complimentary to improve prosthetic range and performance. Implants will be made in both M1 and PPC, enabling simultaneous recording in the same subjects, elevating concerns of comparing data from different labs collected in different individuals with different implants and different tasks. This central hypothesis will be tested in two broad aims, for which we have substantial preliminary data. Aim 1 will examine the control of the body by the two areas. It is hypothesized that M1 will demonstrate strong specificity for the contralateral limb (implants will be made in the hand knob) whereas PPC will code movements for most of the body and on both contra and ipsilateral sides by leveraging its partially mixed encoding of parameters (subaim 1a). Whereas M1 is hypothesized to code spatial variables exclusively during attempted or imagined actions, it is hypothesized that PPC also encodes cognitive spatial variables in task appropriate reference frames (subaim 1b). In subaim 1c we will examine how multiple body parts are combined in movement representations, hypothesizing that M1 and PPC will employ a diverse set of mechanisms including linear summation, non-linear combinations, and movement suppression expressed in different ways as a function of brain area and the specific movement set. Aim 2 will examine the temporal aspects of encoding in the two areas. In subaim 2a we will test the hypothesis that the neural dynamics during sustained periods of movement are largely unchanging in both areas. In subaim 2b we hypothesize that, during sequential movements, M1 codes only the ongoing movement whereas PPC codes both the current and subsequent movements. Finally, in subaim 2c we will examine the coding of movement speed, with the hypothesis that there are separate subspaces in both M1 and PPC for direction and speed of movement.