Rotational dynamics in motor cortex are consistent with a feedback controller.

Recent studies have identified rotational dynamics in motor cortex (MC), which many assume arise from intrinsic connections in MC. However, behavioral and neurophysiological studies suggest that MC behaves like a feedback controller where continuous sensory feedback and interactions with other brain areas contribute substantially to MC processing. We investigated these apparently conflicting theories by building recurrent neural networks that controlled a model arm and received sensory feedback from the limb. Networks were trained to counteract perturbations to the limb and to reach toward spatial targets. Network activities and sensory feedback signals to the network exhibited rotational structure even when the recurrent connections were removed. Furthermore, neural recordings in monkeys performing similar tasks also exhibited rotational structure not only in MC but also in somatosensory cortex. Our results argue that rotational structure may also reflect dynamics throughout the voluntary motor system involved in online control of motor actions.

Convection Enhanced Delivery of Optogenetic Adeno-associated Viral Vector to the Cortex of Rhesus Macaque Under Guidance of Online MRI Images.

In non-human primate (NHP) optogenetics, infecting large cortical areas with viral vectors is often a difficult and time-consuming task. Here, we demonstrate the use of magnetic resonance (MR)-guided convection enhanced delivery (CED) of optogenetic viral vectors into primary somatosensory (S1) and motor (M1) cortices of macaques to obtain efficient, widespread cortical expression of light-sensitive ion channels. Adeno-associated viral (AAV) vectors encoding the red-shifted opsin C1V1 fused to yellow fluorescent protein (EYFP) were injected into the cortex of rhesus macaques under MR-guided CED. Three months post-infusion, epifluorescent imaging confirmed large regions of optogenetic expression (>130 mm2) in M1 and S1 in two macaques. Furthermore, we were able to record reliable light-evoked electrophysiology responses from the expressing areas using micro-electrocorticographic arrays. Later histological analysis and immunostaining against the reporter revealed widespread and dense optogenetic expression in M1 and S1 corresponding to the distribution indicated by epifluorescent imaging. This technique enables us to obtain expression across large areas of the cortex within a shorter period of time with minimal damage compared to the traditional techniques and can be an optimal approach for optogenetic viral delivery in large animals such as NHPs. This approach demonstrates great potential for network-level manipulation of neural circuits with cell-type specificity in animal models evolutionarily close to humans.

Cortical Stimulation Induces Network-Wide Coherence Change In Non-Human Primate Somatosensory Cortex.

Stimulation of the cortex can modulate the connectivity between brain regions. Although targeted neuroplasticity has been demonstrated in-vitro, in-vivo models have been inconsistent in their response to stimulation. In this paper, we tested various stimulation protocols to characterize the effect of stimulation on coherence in the non-human primate cortex in-vivo. We found that the stimulation latency, the state of the cortex during stimulation, and the stimulation site all affected the modulation of cortical coherence. We further investigated features of a resting-state network that could predict how a connection is likely to change with stimulation.

Novel techniques for large-scale manipulations of cortical networks in non-human primates.

Optogenetics is a powerful tool that enables millisecond-level control of the activity of specific groups of neurons. Furthermore, it has the great advantage of artifact free recordings. These characteristics make this technique ideal for relating brain function to behavior in animals with great behavioral capabilities such as non-human primates (NHPs). We recently introduced a practical, stable interface for optogenetic stimulation and recording of large-scale cortical circuits in NHPs. Here we present the various potentials of this interface for studying circuits and connectivity at a large-scale and for relating it to behavior.

Targeted cortical reorganization using optogenetics in non-human primates.

Brain stimulation modulates the excitability of neural circuits and drives neuroplasticity. While the local effects of stimulation have been an active area of investigation, the effects on large-scale networks remain largely unexplored. We studied stimulation-induced changes in network dynamics in two macaques. A large-scale optogenetic interface enabled simultaneous stimulation of excitatory neurons and electrocorticographic recording across primary somatosensory (S1) and motor (M1) cortex (Yazdan-Shahmorad et al., 2016). We tracked two measures of network connectivity, the network response to focal stimulation and the baseline coherence between pairs of electrodes; these were strongly correlated before stimulation. Within minutes, stimulation in S1 or M1 significantly strengthened the gross functional connectivity between these areas. At a finer scale, stimulation led to heterogeneous connectivity changes across the network. These changes reflected the correlations introduced by stimulation-evoked activity, consistent with Hebbian plasticity models. This work extends Hebbian plasticity models to large-scale circuits, with significant implications for stimulation-based neurorehabilitation.

Widespread optogenetic expression in macaque cortex obtained with MR-guided, convection enhanced delivery (CED) of AAV vector to the thalamus.

BACKGROUND: In non-human primate (NHP) optogenetics, infecting large cortical areas with viral vectors is often a difficult and time-consuming task. Previous work has shown that parenchymal delivery of adeno-associated virus (AAV) in the thalamus by convection-enhanced delivery (CED) can lead to large-scale transduction via axonal transport in distal areas including cortex. We used this approach to obtain widespread cortical expression of light-sensitive ion channels. NEW METHOD: AAV vectors co-expressing channelrhodopsin-2 (ChR2) and yellow fluorescent protein (YFP) genes were infused into thalamus of three rhesus macaques under MR-guided CED. After six to twelve weeks recovery, in vivo optical stimulation and single cell recording in the cortex was carried out using an optrode in anesthetized animals. Post-mortem immunostaining against YFP was used to estimate the distribution and level of expression of ChR2 in thalamus and cortex. RESULTS: Histological analysis revealed high levels of transduction in cortical layers. The patterns of expression were consistent with known thalamo-cortico-thalamic circuits. Dense expression was seen in thalamocortiocal axonal fibers in layers III, IV and VI and in pyramidal neurons in layers V and VI, presumably corticothalamic neurons. In addition we obtained reliable in vivo light-evoked responses in cortical areas with high levels of expression. COMPARISON WITH EXISTING METHODS: Thalamic CED is very efficient in achieving large expressing areas in comparison to convectional techniques both in minimizing infusion time and in minimizing damage to the brain. CONCLUSION: MR-guided CED infusion into thalamus provides a simplified approach to transduce large cortical areas by thalamo-cortico-thalamic projections in primate brain.

Superior arm-movement decoding from cortex with a new, unsupervised-learning algorithm.

OBJECTIVE: The aim of this work is to improve the state of the art for motor-control with a brain-machine interface (BMI). BMIs use neurological recording devices and decoding algorithms to transform brain activity directly into real-time control of a machine, archetypically a robotic arm or a cursor. The standard procedure treats neural activity-vectors of spike counts in small temporal windows-as noisy observations of the kinematic state (position, velocity, acceleration) of the fingertip. Inferring the state from the observations then takes the form of a dynamical filter, typically some variant on Kalman's (KF). The KF, however, although fairly robust in practice, is optimal only when the relationships between variables are linear and the noise is Gaussian, conditions usually violated in practice. APPROACH: To overcome these limitations we introduce a new filter, the 'recurrent exponential-family harmonium' (rEFH), that models the spike counts explicitly as Poisson-distributed, and allows for arbitrary nonlinear dynamics and observation models. Furthermore, the model underlying the filter is acquired through unsupervised learning, which allows temporal correlations in spike counts to be explained by latent dynamics that do not necessarily correspond to the kinematic state of the fingertip. MAIN RESULTS: We test the rEFH on offline reconstruction of the kinematics of reaches in the plane. The rEFH outperforms the standard, as well as three other state-of-the-art, decoders, across three monkeys, two different tasks, most kinematic variables, and a range of bin widths, amounts of training data, and numbers of neurons. SIGNIFICANCE: Our algorithm establishes a new state of the art for offline decoding of reaches-in particular, for fingertip velocities, the variable used for control in most online decoders.

Neural Representation and Causal Models in Motor Cortex.

Dorsal premotor (PMd) and primary motor (M1) cortices play a central role in mapping sensation to movement. Many studies of these areas have focused on correlation-based tuning curves relating neural activity to task or movement parameters, but the link between tuning and movement generation is unclear. We recorded motor preparatory activity from populations of neurons in PMd/M1 as macaque monkeys performed a visually guided reaching task and show that tuning curves for sensory inputs (reach target direction) and motor outputs (initial movement direction) are not typically aligned. We then used a simple, causal model to determine the expected relationship between sensory and motor tuning. The model shows that movement variability is minimized when output neurons (those that directly drive movement) have target and movement tuning that are linearly related across targets and cells. In contrast, for neurons that only affect movement via projections to output neurons, the relationship between target and movement tuning is determined by the pattern of projections to output neurons and may even be uncorrelated, as was observed for the PMd/M1 population as a whole. We therefore determined the relationship between target and movement tuning for subpopulations of cells defined by the temporal duration of their spike waveforms, which may distinguish cell types. We found a strong correlation between target and movement tuning for only a subpopulation of neurons with intermediate spike durations (trough-to-peak ∼350 µs after high-pass filtering), suggesting that these cells have the most direct role in driving motor output.SIGNIFICANCE STATEMENT This study focuses on how macaque premotor and primary motor cortices transform sensory inputs into motor outputs. We develop empirical and theoretical links between causal models of this transformation and more traditional, correlation-based "tuning curve" analyses. Contrary to common assumptions, we show that sensory and motor tuning curves for premovement preparatory activity do not generally align. Using a simple causal model, we show that tuning-curve alignment is only expected for output neurons that drive movement. Finally, we identify a physiologically defined subpopulation of neurons with strong tuning-curve alignment, suggesting that it contains a high concentration of output cells. This study demonstrates how analysis of movement variability, combined with simple causal models, can uncover the circuit structure of sensorimotor transformations.

Encoding and Decoding of Multi-Channel ICMS in Macaque Somatosensory Cortex.

Naturalistic control of brain-machine interfaces will require artificial proprioception, potentially delivered via intracortical microstimulation (ICMS). We have previously shown that multi-channel ICMS can guide a monkey reaching to unseen targets in a planar workspace. Here, we expand on that work, asking how ICMS is decoded into target angle and distance by analyzing the performance of a monkey when ICMS feedback was degraded. From the resulting pattern of errors, we found that the animal's estimate of target direction was consistent with a weighted circular-mean strategy-close to the optimal decoding strategy given the ICMS encoding. These results support our previous finding that animals can learn to use this artificial sensory feedback in an efficient and naturalistic manner.

A Large-Scale Interface for Optogenetic Stimulation and Recording in Nonhuman Primates.

While optogenetics offers great potential for linking brain function and behavior in nonhuman primates, taking full advantage of that potential will require stable access for optical stimulation and concurrent monitoring of neural activity. Here we present a practical, stable interface for stimulation and recording of large-scale cortical circuits. To obtain optogenetic expression across a broad region, here spanning primary somatosensory (S1) and motor (M1) cortices, we used convection-enhanced delivery of the viral vector, with online guidance from MRI. To record neural activity across this region, we used a custom micro-electrocorticographic (µECoG) array designed to minimally attenuate optical stimuli. Lastly, we demonstrated the use of this interface to measure spatiotemporal responses to optical stimulation across M1 and S1. This interface offers a powerful tool for studying circuit dynamics and connectivity across cortical areas, for long-term studies of neuromodulation and targeted cortical plasticity, and for linking these to behavior.

Learning to Estimate Dynamical State with Probabilistic Population Codes.

Tracking moving objects, including one's own body, is a fundamental ability of higher organisms, playing a central role in many perceptual and motor tasks. While it is unknown how the brain learns to follow and predict the dynamics of objects, it is known that this process of state estimation can be learned purely from the statistics of noisy observations. When the dynamics are simply linear with additive Gaussian noise, the optimal solution is the well known Kalman filter (KF), the parameters of which can be learned via latent-variable density estimation (the EM algorithm). The brain does not, however, directly manipulate matrices and vectors, but instead appears to represent probability distributions with the firing rates of population of neurons, "probabilistic population codes." We show that a recurrent neural network-a modified form of an exponential family harmonium (EFH)-that takes a linear probabilistic population code as input can learn, without supervision, to estimate the state of a linear dynamical system. After observing a series of population responses (spike counts) to the position of a moving object, the network learns to represent the velocity of the object and forms nearly optimal predictions about the position at the next time-step. This result builds on our previous work showing that a similar network can learn to perform multisensory integration and coordinate transformations for static stimuli. The receptive fields of the trained network also make qualitative predictions about the developing and learning brain: tuning gradually emerges for higher-order dynamical states not explicitly present in the inputs, appearing as delayed tuning for the lower-order states.

A learning-based approach to artificial sensory feedback leads to optimal integration.

Proprioception-the sense of the body's position in space-is important to natural movement planning and execution and will likewise be necessary for successful motor prostheses and brain-machine interfaces (BMIs). Here we demonstrate that monkeys were able to learn to use an initially unfamiliar multichannel intracortical microstimulation signal, which provided continuous information about hand position relative to an unseen target, to complete accurate reaches. Furthermore, monkeys combined this artificial signal with vision to form an optimal, minimum-variance estimate of relative hand position. These results demonstrate that a learning-based approach can be used to provide a rich artificial sensory feedback signal, suggesting a new strategy for restoring proprioception to patients using BMIs, as well as a powerful new tool for studying the adaptive mechanisms of sensory integration.

Motor variability arises from a slow random walk in neural state.

Even well practiced movements cannot be repeated without variability. This variability is thought to reflect "noise" in movement preparation or execution. However, we show that, for both professional baseball pitchers and macaque monkeys making reaching movements, motor variability can be decomposed into two statistical components, a slowly drifting mean and fast trial-by-trial fluctuations about the mean. The preparatory activity of dorsal premotor cortex/primary motor cortex neurons in monkey exhibits similar statistics. Although the neural and behavioral drifts appear to be correlated, neural activity does not account for trial-by-trial fluctuations in movement, which must arise elsewhere, likely downstream. The statistics of this drift are well modeled by a double-exponential autocorrelation function, with time constants similar across the neural and behavioral drifts in two monkeys, as well as the drifts observed in baseball pitching. These time constants can be explained by an error-corrective learning processes and agree with learning rates measured directly in previous experiments. Together, these results suggest that the central contributions to movement variability are not simply trial-by-trial fluctuations but are rather the result of longer-timescale processes that may arise from motor learning.

Learning multisensory integration and coordinate transformation via density estimation.

Sensory processing in the brain includes three key operations: multisensory integration-the task of combining cues into a single estimate of a common underlying stimulus; coordinate transformations-the change of reference frame for a stimulus (e.g., retinotopic to body-centered) effected through knowledge about an intervening variable (e.g., gaze position); and the incorporation of prior information. Statistically optimal sensory processing requires that each of these operations maintains the correct posterior distribution over the stimulus. Elements of this optimality have been demonstrated in many behavioral contexts in humans and other animals, suggesting that the neural computations are indeed optimal. That the relationships between sensory modalities are complex and plastic further suggests that these computations are learned-but how? We provide a principled answer, by treating the acquisition of these mappings as a case of density estimation, a well-studied problem in machine learning and statistics, in which the distribution of observed data is modeled in terms of a set of fixed parameters and a set of latent variables. In our case, the observed data are unisensory-population activities, the fixed parameters are synaptic connections, and the latent variables are multisensory-population activities. In particular, we train a restricted Boltzmann machine with the biologically plausible contrastive-divergence rule to learn a range of neural computations not previously demonstrated under a single approach: optimal integration; encoding of priors; hierarchical integration of cues; learning when not to integrate; and coordinate transformation. The model makes testable predictions about the nature of multisensory representations.

How each movement changes the next: an experimental and theoretical study of fast adaptive priors in reaching.

Most voluntary actions rely on neural circuits that map sensory cues onto appropriate motor responses. One might expect that for everyday movements, like reaching, this mapping would remain stable over time, at least in the absence of error feedback. Here we describe a simple and novel psychophysical phenomenon in which recent experience shapes the statistical properties of reaching, independent of any movement errors. Specifically, when recent movements are made to targets near a particular location subsequent movements to that location become less variable, but at the cost of increased bias for reaches to other targets. This process exhibits the variance-bias tradeoff that is a hallmark of Bayesian estimation. We provide evidence that this process reflects a fast, trial-by-trial learning of the prior distribution of targets. We also show that these results may reflect an emergent property of associative learning in neural circuits. We demonstrate that adding Hebbian (associative) learning to a model network for reach planning leads to a continuous modification of network connections that biases network dynamics toward activity patterns associated with recent inputs. This learning process quantitatively captures the key results of our experimental data in human subjects, including the effect that recent experience has on the variance-bias tradeoff. This network also provides a good approximation of a normative Bayesian estimator. These observations illustrate how associative learning can incorporate recent experience into ongoing computations in a statistically principled way.

Sensory integration for reaching: models of optimality in the context of behavior and the underlying neural circuits.

Although multisensory integration has been well modeled at the behavioral level, the link between these behavioral models and the underlying neural circuits is still not clear. This gap is even greater for the problem of sensory integration during movement planning and execution. The difficulty lies in applying simple models of sensory integration to the complex computations that are required for movement control and to the large networks of brain areas that perform these computations. Here I review psychophysical, computational, and physiological work on multisensory integration during movement planning, with an emphasis on goal-directed reaching. I argue that sensory transformations must play a central role in any modeling effort. In particular, the statistical properties of these transformations factor heavily into the way in which downstream signals are combined. As a result, our models of optimal integration are only expected to apply "locally," that is, independently for each brain area. I suggest that local optimality can be reconciled with globally optimal behavior if one views the collection of parietal sensorimotor areas not as a set of task-specific domains, but rather as a palette of complex, sensorimotor representations that are flexibly combined to drive downstream activity and behavior.

Heterogeneous representations in the superior parietal lobule are common across reaches to visual and proprioceptive targets.

The planning and control of sensory-guided movements requires the integration of multiple sensory streams. Although the information conveyed by different sensory modalities is often overlapping, the shared information is represented differently across modalities during the early stages of cortical processing. We ask how these diverse sensory signals are represented in multimodal sensorimotor areas of cortex in macaque monkeys. Although a common modality-independent representation might facilitate downstream readout, previous studies have found that modality-specific representations in multimodal cortex reflect upstream spatial representations. For example, visual signals have a more eye-centered representation. We recorded neural activity from two parietal areas involved in reach planning, area 5 and the medial intraparietal area (MIP), as animals reached to visual, combined visual and proprioceptive, and proprioceptive targets while fixing their gaze on another location. In contrast to other multimodal cortical areas, the same spatial representations are used to represent visual and proprioceptive signals in both area 5 and MIP. However, these representations are heterogeneous. Although we observed a posterior-to-anterior gradient in population responses in parietal cortex, from more eye-centered to more hand- or body-centered representations, we do not observe the simple and discrete reference frame representations suggested by studies that focused on identifying the "best-match" reference frame for a given cortical area. In summary, we find modality-independent representations of spatial information in parietal cortex, although these representations are complex and heterogeneous.

Sensory transformations and the use of multiple reference frames for reach planning.

The sensory signals that drive movement planning arrive in a variety of 'reference frames', and integrating or comparing them requires sensory transformations. We propose a model in which the statistical properties of sensory signals and their transformations determine how these signals are used. This model incorporates the patterns of gaze-dependent errors that we found in our human psychophysics experiment when the sensory signals available for reach planning were varied. These results challenge the widely held ideas that error patterns directly reflect the reference frame of the underlying neural representation and that it is preferable to use a single common reference frame for movement planning. We found that gaze-dependent error patterns, often cited as evidence for retinotopic reach planning, can be explained by a transformation bias and are not exclusively linked to retinotopic representations. Furthermore, the presence of multiple reference frames allows for optimal use of available sensory information and explains task-dependent reweighting of sensory signals.

Calibration of visually guided reaching is driven by error-corrective learning and internal dynamics.

The sensorimotor calibration of visually guided reaching changes on a trial-to-trial basis in response to random shifts in the visual feedback of the hand. We show that a simple linear dynamical system is sufficient to model the dynamics of this adaptive process. In this model, an internal variable represents the current state of sensorimotor calibration. Changes in this state are driven by error feedback signals, which consist of the visually perceived reach error, the artificial shift in visual feedback, or both. Subjects correct for > or =20% of the error observed on each movement, despite being unaware of the visual shift. The state of adaptation is also driven by internal dynamics, consisting of a decay back to a baseline state and a "state noise" process. State noise includes any source of variability that directly affects the state of adaptation, such as variability in sensory feedback processing, the computations that drive learning, or the maintenance of the state. This noise is accumulated in the state across trials, creating temporal correlations in the sequence of reach errors. These correlations allow us to distinguish state noise from sensorimotor performance noise, which arises independently on each trial from random fluctuations in the sensorimotor pathway. We show that these two noise sources contribute comparably to the overall magnitude of movement variability. Finally, the dynamics of adaptation measured with random feedback shifts generalizes to the case of constant feedback shifts, allowing for a direct comparison of our results with more traditional blocked-exposure experiments.

Modeling sensorimotor learning with linear dynamical systems.

Recent studies have employed simple linear dynamical systems to model trial-by-trial dynamics in various sensorimotor learning tasks. Here we explore the theoretical and practical considerations that arise when employing the general class of linear dynamical systems (LDS) as a model for sensorimotor learning. In this framework, the state of the system is a set of parameters that define the current sensorimotor transformation-the function that maps sensory inputs to motor outputs. The class of LDS models provides a first-order approximation for any Markovian (state-dependent) learning rule that specifies the changes in the sensorimotor transformation that result from sensory feedback on each movement. We show that modeling the trial-by-trial dynamics of learning provides a substantially enhanced picture of the process of adaptation compared to measurements of the steady state of adaptation derived from more traditional blocked-exposure experiments. Specifically, these models can be used to quantify sensory and performance biases, the extent to which learned changes in the sensorimotor transformation decay over time, and the portion of motor variability due to either learning or performance variability. We show that previous attempts to fit such models with linear regression have not generally yielded consistent parameter estimates. Instead, we present an expectation-maximization algorithm for fitting LDS models to experimental data and describe the difficulties inherent in estimating the parameters associated with feedback-driven learning. Finally, we demonstrate the application of these methods in a simple sensorimotor learning experiment: adaptation to shifted visual feedback during reaching.