The brain represents sensory variables in the coordinated activity of neural populations, in which tuning curves of single neurons define the geometry of the population code. Whether the same coding principle holds for dynamic cognitive variables remains unknown because internal cognitive processes unfold with a unique time course on single trials observed only in the irregular spiking of heterogeneous neural populations. Here we show the existence of such a population code for the dynamics of choice formation in the primate premotor cortex. We developed an approach to simultaneously infer population dynamics and tuning functions of single neurons to the population state. Applied to spike data recorded during decision-making, our model revealed that populations of neurons encoded the same dynamic variable predicting choices, and heterogeneous firing rates resulted from the diverse tuning of single neurons to this decision variable. The inferred dynamics indicated an attractor mechanism for decision computation. Our results reveal a common geometric principle for neural encoding of sensory and dynamic cognitive variables.

High-density, integrated silicon electrodes have begun to transform systems neuroscience, by enabling large-scale neural population recordings with single cell resolution. Existing technologies, however, have provided limited functionality in nonhuman primate species such as macaques, which offer close models of human cognition and behavior. Here, we report the design, fabrication, and performance of Neuropixels 1.0-NHP, a high channel count linear electrode array designed to enable large-scale simultaneous recording in superficial and deep structures within the macaque or other large animal brain. These devices were fabricated in two versions: 4416 electrodes along a 45 mm shank, and 2496 along a 25 mm shank. For both versions, users can programmatically select 384 channels, enabling simultaneous multi-area recording with a single probe. We demonstrate recording from over 3000 single neurons within a session, and simultaneous recordings from over 1000 neurons using multiple probes. This technology represents a significant increase in recording access and scalability relative to existing technologies, and enables new classes of experiments involving fine-grained electrophysiological characterization of brain areas, functional connectivity between cells, and simultaneous brain-wide recording at scale.

How does the motor cortex combine simple movements (such as single finger flexion/extension) into complex movements (such hand gestures or playing piano)? Motor cortical activity was recorded using intracortical multi-electrode arrays in two people with tetraplegia as they attempted single, pairwise and higher order finger movements. Neural activity for simultaneous movements was largely aligned with linear summation of corresponding single finger movement activities, with two violations. First, the neural activity was normalized, preventing a large magnitude with an increasing number of moving fingers. Second, the neural tuning direction of weakly represented fingers (e.g. middle) changed significantly as a result of the movement of other fingers. These deviations from linearity resulted in non-linear methods outperforming linear methods for neural decoding. Overall, simultaneous finger movements are thus represented by the combination of individual finger movements by pseudo-linear summation.

The study of the cortical control of movement experienced a conceptual shift over recent decades, as the basic currency of understanding shifted from single-neuron tuning towards population-level factors and their dynamics. This transition was informed by a maturing understanding of recurrent networks, where mechanism is often characterized in terms of population-level factors. By estimating factors from data, experimenters could test network-inspired hypotheses. Central to such hypotheses are 'output-null' factors that do not directly drive motor outputs yet are essential to the overall computation. In this Review, we highlight how the hypothesis of output-null factors was motivated by the venerable observation that motor-cortex neurons are active during movement preparation, well before movement begins. We discuss how output-null factors then became similarly central to understanding neural activity during movement. We discuss how this conceptual framework provided key analysis tools, making it possible for experimenters to address long-standing questions regarding motor control. We highlight an intriguing trend: as experimental and theoretical discoveries accumulate, the range of computational roles hypothesized to be subserved by output-null factors continues to expand.

Flexible computation is a hallmark of intelligent behavior. Yet, little is known about how neural networks contextually reconfigure for different computations. Humans are able to perform a new task without extensive training, presumably through the composition of elementary processes that were previously learned. Cognitive scientists have long hypothesized the possibility of a compositional neural code, where complex neural computations are made up of constituent components; however, the neural substrate underlying this structure remains elusive in biological and artificial neural networks. Here we identified an algorithmic neural substrate for compositional computation through the study of multitasking artificial recurrent neural networks. Dynamical systems analyses of networks revealed learned computational strategies that mirrored the modular subtask structure of the task-set used for training. Dynamical motifs such as attractors, decision boundaries and rotations were reused across different task computations. For example, tasks that required memory of a continuous circular variable repurposed the same ring attractor. We show that dynamical motifs are implemented by clusters of units and are reused across different contexts, allowing for flexibility and generalization of previously learned computation. Lesioning these clusters resulted in modular effects on network performance: a lesion that destroyed one dynamical motif only minimally perturbed the structure of other dynamical motifs. Finally, modular dynamical motifs could be reconfigured for fast transfer learning. After slow initial learning of dynamical motifs, a subsequent faster stage of learning reconfigured motifs to perform novel tasks. This work contributes to a more fundamental understanding of compositional computation underlying flexible general intelligence in neural systems. We present a conceptual framework that establishes dynamical motifs as a fundamental unit of computation, intermediate between the neuron and the network. As more whole brain imaging studies record neural activity from multiple specialized systems simultaneously, the framework of dynamical motifs will guide questions about specialization and generalization across brain regions.

Brain-computer interfaces have so far focused largely on enabling the control of a single effector, for example a single computer cursor or robotic arm. Restoring multi-effector motion could unlock greater functionality for people with paralysis (e.g., bimanual movement). However, it may prove challenging to decode the simultaneous motion of multiple effectors, as we recently found that a compositional neural code links movements across all limbs and that neural tuning changes nonlinearly during dual-effector motion. Here, we demonstrate the feasibility of high-quality bimanual control of two cursors via neural network (NN) decoders. Through simulations, we show that NNs leverage a neural ‘laterality’ dimension to distinguish between left and right-hand movements as neural tuning to both hands become increasingly correlated. In training recurrent neural networks (RNNs) for two-cursor control, we developed a method that alters the temporal structure of the training data by dilating/compressing it in time and re-ordering it, which we show helps RNNs successfully generalize to the online setting. With this method, we demonstrate that a person with paralysis can control two computer cursors simultaneously. Our results suggest that neural network decoders may be advantageous for multi-effector decoding, provided they are designed to transfer to the online setting.

BACKGROUND AND OBJECTIVES:Brain-computer interfaces (BCIs) are being developed to restore mobility, communication, and functional independence to people with paralysis. Though supported by decades of preclinical data, the safety of chronically implanted microelectrode array BCIs in humans is unknown. We report safety results from the prospective, open-label, nonrandomized BrainGate feasibility study (NCT00912041), the largest and longest-running clinical trial of an implanted BCI. METHODS:Adults aged 18-75 years with quadriparesis from spinal cord injury, brainstem stroke, or motor neuron disease were enrolled through 7 clinical sites in the United States. Participants underwent surgical implantation of 1 or 2 microelectrode arrays in the motor cortex of the dominant cerebral hemisphere. The primary safety outcome was device-related serious adverse events (SAEs) requiring device explantation or resulting in death or permanently increased disability during the 1-year postimplant evaluation period. The secondary outcomes included the type and frequency of other adverse events and the feasibility of the BrainGate system for controlling a computer or other assistive technologies. RESULTS:From 2004 to 2021, 14 adults enrolled in the BrainGate trial had devices surgically implanted. The average duration of device implantation was 872 days, yielding 12,203 days of safety experience. There were 68 device-related adverse events, including 6 device-related SAEs. The most common device-related adverse event was skin irritation around the percutaneous pedestal. There were no safety events that required device explantation, no unanticipated adverse device events, no intracranial infections, and no participant deaths or adverse events resulting in permanently increased disability related to the investigational device. DISCUSSION:The BrainGate Neural Interface system has a safety record comparable with other chronically implanted medical devices. Given rapid recent advances in this technology and continued performance gains, these data suggest a favorable risk/benefit ratio in appropriately selected individuals to support ongoing research and development. TRIAL REGISTRATION INFORMATION:ClinicalTrials.gov Identifier: NCT00912041. CLASSIFICATION OF EVIDENCE:This study provides Class IV evidence that the neurosurgically placed BrainGate Neural Interface system is associated with a low rate of SAEs defined as those requiring device explantation, resulting in death, or resulting in permanently increased disability during the 1-year postimplant period.

Speech brain-computer interfaces (BCIs) have the potential to restore rapid communication to people with paralysis by decoding neural activity evoked by attempted speech into text(1,2) or sound(3,4). Early demonstrations, although promising, have not yet achieved accuracies sufficiently high for communication of unconstrained sentences from a large vocabulary(1-7). Here we demonstrate a speech-to-text BCI that records spiking activity from intracortical microelectrode arrays. Enabled by these high-resolution recordings, our study participant-who can no longer speak intelligibly owing to amyotrophic lateral sclerosis-achieved a 9.1% word error rate on a 50-word vocabulary (2.7times fewer errors than the previous state-of-the-art speech BCI2) and a 23.8% word error rate on a 125,000-word vocabulary (the first successful demonstration, to our knowledge, of large-vocabulary decoding). Our participant's attempted speech was decoded at 62words per minute, which is 3.4times as fast as the previous record(8) and begins to approach the speed of natural conversation (160words per minute(9)). Finally, we highlight two aspects of the neural code for speech that are encouraging for speech BCIs: spatially intermixed tuning to speech articulators that makes accurate decoding possible from only a small region of cortex, and a detailed articulatory representation of phonemes that persists years after paralysis. These results show a feasible path forward for restoring rapid communication to people with paralysis who can no longer speak.

We used a dynamical systems perspective to understand decision-related neural activity, a fundamentally unresolved problem. This perspective posits that time-varying neural activity is described by a state equation with an initial condition and evolves in time by combining at each time step, recurrent activity and inputs. We hypothesized various dynamical mechanisms of decisions, simulated them in models to derive predictions, and evaluated these predictions by examining firing rates of neurons in the dorsal premotor cortex (PMd) of monkeys performing a perceptual decision-making task. Prestimulus neural activity (i.e., the initial condition) predicted poststimulus neural trajectories, covaried with RT and the outcome of the previous trial, but not with choice. Poststimulus dynamics depended on both the sensory evidence and initial condition, with easier stimuli and fast initial conditions leading to the fastest choice-related dynamics. Together, these results suggest that initial conditions combine with sensory evidence to induce decision-related dynamics in PMd.

Intracortical brain-computer interfaces (iBCIs) require frequent recalibration to maintain robust performance due to changes in neural activity that accumulate over time. Compensating for this nonstationarity would enable consistently high performance without the need for supervised recalibration periods, where users cannot engage in free use of their device. Here we introduce a hidden Markov model (HMM) to infer what targets users are moving toward during iBCI use. We then retrain the system using these inferred targets, enabling unsupervised adaptation to changing neural activity. Our approach outperforms the state of the art in large-scale, closed-loop simulations over two months and in closed-loop with a human iBCI user over one month. Leveraging an offline dataset spanning five years of iBCI recordings, we further show how recently proposed data distribution-matching approaches to recalibration fail over long time scales; only target-inference methods appear capable of enabling long-term unsupervised recalibration. Our results demonstrate how task structure can be used to bootstrap a noisy decoder into a highly-performant one, thereby overcoming one of the major barriers to clinically translating BCIs.

Methylphenidate (MPH, brand: Ritalin) is a common stimulant used both medically and non-medically. Though typically prescribed for its cognitive effects, MPH also affects movement. While it is known that MPH noncompetitively blocks the reuptake of catecholamines through inhibition of dopamine and norepinephrine transporters, a critical step in exploring how it affects behavior is to understand how MPH directly affects neural activity. This would establish an electrophysiological mechanism of action for MPH. Since we now have biologically-grounded network-level hypotheses regarding how populations of motor cortical neurons plan and execute movements, there is a unique opportunity to make testable predictions regarding how systemic MPH administration – a pharmacological perturbation – might affect neural activity in motor cortex. To that end, we administered clinically-relevant doses of MPH to Rhesus monkeys as they performed an instructed-delay reaching task. Concomitantly, we measured neural activity from dorsal premotor and primary motor cortex. Consistent with our predictions, we found dose-dependent and significant effects on reaction time, trial-by-trial variability, and movement speed. We confirmed our hypotheses that changes in reaction time and variability were accompanied by previously established population-level changes in motor cortical preparatory activity and the condition-independent signal that precedes movements. We expected changes in speed to be a result of changes in the amplitude of motor cortical dynamics and/or a translation of those dynamics in activity space. Instead, our data are consistent with a mechanism whereby the neuromodulatory effect of MPH is to increase the gain and/or the signal-to-noise of motor cortical dynamics during reaching. Continued work in this domain to better understand the brain-wide electrophysiological mechanism of action of MPH and other psychoactive drugs could facilitate more targeted treatments for a host of cognitive-motor disorders.

Intracortical brain computer interfaces (iBCIs) decode neural activity from the cortex and enable motor and communication prostheses, such as cursor control, handwriting and speech, for people with paralysis. This paper introduces a new iBCI communication prosthesis using a 3D keyboard interface for typing using continuous, closed loop movement of multiple fingers. A participant-specific BCI keyboard prototype was developed for a BrainGate2 clinical trial participant (T5) using neural recordings from the hand-knob area of the left premotor cortex. We assessed the relative decoding accuracy of flexion/extension movements of individual single fingers (5 degrees of freedom (DOF)) vs. three groups of fingers (thumb, index-middle, and ring-small fingers, 3 DOF). Neural decoding using 3 independent DOF was more accurate (95%) than that using 5 DOF (76%). A virtual keyboard was then developed where each finger group moved along a flexion-extension arc to acquire targets that corresponded to English letters and symbols. The locations of these letter/symbols were optimized using natural language statistics, resulting in an approximately a $2\times$ reduction in distance traveled by fingers on average compared to a random keyboard layout. This keyboard was tested using a simple real-time closed loop decoder enabling T5 to type with 31 symbols at 90% accuracy and approximately 2.3 sec/symbol (excluding a 2 second hold time) on average.

Advances in deep learning have given rise to neural network models of the relationship between movement and brain activity that appear to far outperform prior approaches. Brain-computer interfaces (BCIs) that enable people with paralysis to control external devices, such as robotic arms or computer cursors, might stand to benefit greatly from these advances. We tested recurrent neural networks (RNNs) on a challenging nonlinear BCI problem: decoding continuous bimanual movement of two computer cursors. Surprisingly, we found that although RNNs appeared to perform well in offline settings, they did so by overfitting to the temporal structure of the training data and failed to generalize to real-time neuroprosthetic control. In response, we developed a method that alters the temporal structure of the training data by dilating/compressing it in time and re-ordering it, which we show helps RNNs successfully generalize to the online setting. With this method, we demonstrate that a person with paralysis can control two computer cursors simultaneously, far outperforming standard linear methods. Our results provide evidence that preventing models from overfitting to temporal structure in training data may, in principle, aid in translating deep learning advances to the BCI setting, unlocking improved performance for challenging applications.

High-density, integrated silicon electrodes have begun to transform systems neuroscience, by enabling large-scale neural population recordings with single cell resolution. Existing technologies, however, have provided limited functionality in nonhuman primate species such as macaques, which offer close models of human cognition and behavior. Here, we report the design, fabrication, and performance of Neuropixels 1.0-NHP, a high channel count linear electrode array designed to enable large-scale simultaneous recording in superficial and deep structures within the macaque or other large animal brain. These devices were fabricated in two versions: 4416 electrodes along a 45 mm shank, and 2496 along a 25 mm shank. For both versions, users can programmably select 384 channels, enabling simultaneous multi-area recording with a single probe. We demonstrate recording from over 3000 single neurons within a session, and simultaneous recordings from over 1000 neurons using multiple probes. This technology represents a significant increase in recording access and scalability relative to existing technologies, and enables new classes of experiments involving fine-grained electrophysiological characterization of brain areas, functional connectivity between cells, and simultaneous brain-wide recording at scale.

Intracortical brain-computer interfaces (iBCIs) have shown promise for restoring rapid communication to people with neurological disorders such as amyotrophic lateral sclerosis (ALS). However, to maintain high performance over time, iBCIs typically need frequent recalibration to combat changes in the neural recordings that accrue over days. This requires iBCI users to stop using the iBCI and engage in supervised data collection, making the iBCI system hard to use. In this paper, we propose a method that enables self-recalibration of communication iBCIs without interrupting the user. Our method leverages large language models (LMs) to automatically correct errors in iBCI outputs. The self-recalibration process uses these corrected outputs ("pseudo-labels") to continually update the iBCI decoder online. Over a period of more than one year (403 days), we evaluated our Continual Online Recalibration with Pseudo-labels (CORP) framework with one clinical trial participant. CORP achieved a stable decoding accuracy of 93.84% in an online handwriting iBCI task, significantly outperforming other baseline methods. Notably, this is the longest-running iBCI stability demonstration involving a human participant. Our results provide the first evidence for long-term stabilization of a plug-and-play, high-performance communication iBCI, addressing a major barrier for the clinical translation of iBCIs.

Neuropixels are silicon-based electrophysiology-recording probes with high channel count and recording-site density. These probes offer a turnkey platform for measuring neural activity with single-cell resolution and at a scale that is beyond the capabilities of current clinically approved devices. Our team demonstrated the first-in-human use of these probes during resection surgery for epilepsy or tumors and deep brain stimulation electrode placement in patients with Parkinson's disease. Here, we provide a better understanding of the capabilities and challenges of using Neuropixels as a research tool to study human neurophysiology, with the hope that this information may inform future efforts toward regulatory approval of Neuropixels probes as research devices. In perioperative procedures, the major concerns are the initial sterility of the device, maintaining a sterile field during surgery, having multiple referencing and grounding schemes available to de-noise recordings (if necessary), protecting the silicon probe from accidental contact before insertion and obtaining high-quality action potential and local field potential recordings. The research team ensures that the device is fully operational while coordinating with the surgical team to remove sources of electrical noise that could otherwise substantially affect the signals recorded by the sensitive hardware. Prior preparation using the equipment and training in human clinical research and working in operating rooms maximize effective communication within and between the teams, ensuring high recording quality and minimizing the time added to the surgery. The perioperative procedure requires ~4 h, and the entire protocol requires multiple weeks.

The rich repertoire of skilled mammalian behavior is the product of neural circuits that generate robust and flexible patterns of activity distributed across populations of neurons. Decades of associative studies have linked many behaviors to specific patterns of population activity, but association alone cannot reveal the dynamical mechanisms that shape those patterns. Are local neural circuits high-dimensional dynamical reservoirs able to generate arbitrary superpositions of patterns with appropriate excitation? Or might circuit dynamics be shaped in response to behavioral context so as to generate only the low-dimensional patterns needed for the task at hand? Here, we address these questions within primate motor cortex by delivering optogenetic and electrical microstimulation perturbations during reaching behavior. We develop a novel analytic approach that relates measured activity to theoretically tractable, dynamical models of excitatory and inhibitory neurons. This computational model captures the dynamical effects of these perturbations and demonstrates that motor cortical activity during reaching is shaped by a self-contained, low-dimensional dynamical system. The subspace containing task-relevant dynamics proves to be oriented so as to be robust to strong non-normal amplification within cortical circuits. This task dynamics space exhibits a privileged causal relationship with behavior, in that stimulation in motor cortex perturb reach kinematics only to the extent that it alters neural states within this subspace. Our results resolve long-standing questions about the dynamical structure of cortical activity associated with movement, and illuminate the dynamical perturbation experiments needed to understand how neural circuits throughout the brain generate complex behavior.

Brain–computer interfaces (BCIs) can restore com-munication to people who have lost the ability to move or speak. So far, a major focus of BCI research has been on restoring gross motor skills, such as reaching and grasping1–5 or point-and-click typing with a computer cursor. However, rapid sequences of highly dexterous behaviours, such as handwriting or touch typing, might enable faster rates of communication. Here we developed an intracortical BCI that decodes attempted handwriting movements from neural activity in the motor cortex and translates it to text in real time, using a recurrent neural net-work decoding approach. With this BCI, our study participant, whose hand was paralysed from spinal cord injury, achieved typing speeds of 90 characters per minute with 94.1 percent raw accuracy online, and greater than 99 percent accuracy offline with a general-purpose autocorrect. To our knowledge, these typing speeds exceed those reported for any other BCI, and are comparable to typical smartphone typing speeds of individuals in the age group of our participant (115 characters per minute). Fi-nally, theoretical considerations explain why temporally complex movements, such as handwriting, may be fundamentally easier to decode than point-to-point movements. Our results open a new approach for BCIs and demonstrate the feasibility of accurately decoding rapid, dexterous movements years after paralysis. We recently reported this research in Willett et al. Nature 2021 [1].

Computational analysis of cellular activity has developed largely independently of modern transcriptomic cell typology, but integrating these approaches may be essential for full insight into cellular-level mechanisms underlying brain function and dysfunction. Applying this approach to the habenula (a structure with diverse, intermingled molecular, anatomical, and computational features), we identified encoding of reward-predictive cues and reward outcomes in distinct genetically defined neural populations, including TH+ cells and Tac1+ cells. Data from genetically targeted recordings were used to train an optimized nonlinear dynamical systems model and revealed activity dynamics consistent with a line attractor. High-density, cell-type-specific electrophysiological recordings and optogenetic perturbation provided supporting evidence for this model. Reverse-engineering predicted how Tac1+ cells might integrate reward history, which was complemented by in vivo experimentation. This integrated approach describes a process by which data-driven computational models of population activity can generate and frame actionable hypotheses for cell-type-specific investigation in biological systems.

The brain's remarkable ability to learn and execute various motor behaviours harnesses the capacity of neural populations to generate a variety of activity patterns. Here we explore systematic changes in preparatory activity in motor cortex that accompany motor learning. We trained rhesus monkeys to learn an arm-reaching task1 in a curl force field that elicited new muscle forces for some, but not all, movement directions2,3. We found that in a neural subspace predictive of hand forces, changes in preparatory activity tracked the learned behavioural modifications and reassociated4 existing activity patterns with updated movements. Along a neural population dimension orthogonal to the force-predictive subspace, we discovered that preparatory activity shifted uniformly for all movement directions, including those unaltered by learning. During a washout period when the curl field was removed, preparatory activity gradually reverted in the force-predictive subspace, but the uniform shift persisted. These persistent preparatory activity patterns may retain a motor memory of the learned field5,6 and support accelerated relearning of the same curl field. When a set of distinct curl fields was learned in sequence, we observed a corresponding set of field-specific uniform shifts which separated the associated motor memories in the neural state space7-9. The precise geometry of these uniform shifts in preparatory activity could serve to index motor memories, facilitating the acquisition, retention and retrieval of a broad motor repertoire.