Inference from populations: going beyond models

Cortical Control of Virtual Self-Motion Using Task-Specific Subspaces

Brain-machine interfaces (BMIs) for reaching have enjoyed continued performance improvements, yet there remains significant need for BMIs that control other movement classes. Recent scientific findings suggest that the intrinsic covariance structure of neural activity depends strongly on movement class, potentially necessitating different decode algorithms across classes. To address this possibility, we developed a self-motion BMI based on cortical activity as monkeys cycled a hand-held pedal to progress along a virtual track. Unlike during reaching, we found no high-variance dimensions that directly correlated with to-be-decoded variables. This was due to no neurons having consistent correlations between their responses and kinematic variables. Yet we could decode a single variable-self-motion-by nonlinearly leveraging structure that spanned multiple high-variance neural dimensions. Resulting online BMI-control success rates approached those during manual control. These findings make two broad points regarding how to build decode algorithms that harmonize with the empirical structure of neural activity in motor cortex. First, even when decoding from the same cortical region (e.g., arm-related motor cortex), different movement classes may need to employ very different strategies. Although correlations between neural activity and hand velocity are prominent during reaching tasks, they are not a fundamental property of motor cortex and cannot be counted on to be present in general. Second, although one generally desires a low-dimensional readout, it can be beneficial to leverage a multidimensional high-variance subspace. Fully embracing this approach requires highly nonlinear approaches tailored to the task at hand, but can produce near-native levels of performance.SIGNIFICANCE STATEMENT Many brain-machine interface decoders have been constructed for controlling movements normally performed with the arm. Yet it is unclear how these will function beyond the reach-like scenarios where they were developed. Existing decoders implicitly assume that neural covariance structure, and correlations with to-be-decoded kinematic variables, will be largely preserved across tasks. We find that the correlation between neural activity and hand kinematics, a feature typically exploited when decoding reach-like movements, is essentially absent during another task performed with the arm: cycling through a virtual environment. Nevertheless, the use of a different strategy, one focused on leveraging the highest-variance neural signals, supported high performance real-time brain-machine interface control.

Optimizing the Usability of Brain-Computer Interfaces

Brain-computer interfaces are in the process of moving from the laboratory to the clinic. These devices act by reading neural activity and using it to directly control a device, such as a cursor on a computer screen. An open question in the field is how to map neural activity to device movement in order to achieve the most proficient control. This question is complicated by the fact that learning, especially the long-term skill learning that accompanies weeks of practice, can allow subjects to improve performance over time. Typical approaches to this problem attempt to maximize the biomimetic properties of the device in order to limit the need for extensive training. However, it is unclear if this approach would ultimately be superior to performance that might be achieved with a nonbiomimetic device once the subject has engaged in extended practice and learned how to use it. Here we approach this problem using ideas from optimal control theory. Under the assumption that the brain acts as an optimal controller, we present a formal definition of the usability of a device and show that the optimal postlearning mapping can be written as the solution of a constrained optimization problem. We then derive the optimal mappings for particular cases common to most brain-computer interfaces. Our results suggest that the common approach of creating biomimetic interfaces may not be optimal when learning is taken into account. More broadly, our method provides a blueprint for optimal device design in general control-theoretic contexts.

Tensor Analysis Reveals Distinct Population Structure that Parallels the Different Computational Roles of Areas M1 and V1

Cortical firing rates frequently display elaborate and heterogeneous temporal structure. One often wishes to compute quantitative summaries of such structure—a basic example is the frequency spectrum—and compare with model-based predictions. The advent of large-scale population recordings affords the opportunity to do so in new ways, with the hope of distinguishing between potential explanations for why responses vary with time. We introduce a method that assesses a basic but previously unexplored form of population-level structure: when data contain responses across multiple neurons, conditions, and times, they are naturally expressed as a third-order tensor. We examined tensor structure for multiple datasets from primary visual cortex (V1) and primary motor cortex (M1). All V1 datasets were ‘simplest’ (there were relatively few degrees of freedom) along the neuron mode, while all M1 datasets were simplest along the condition mode. These differences could not be inferred from surface-level response features. Formal considerations suggest why tensor structure might differ across modes. For idealized linear models, structure is simplest across the neuron mode when responses reflect external variables, and simplest across the condition mode when responses reflect population dynamics. This same pattern was present for existing models that seek to explain motor cortex responses. Critically, only dynamical models displayed tensor structure that agreed with the empirical M1 data. These results illustrate that tensor structure is a basic feature of the data. For M1 the tensor structure was compatible with only a subset of existing models.

Neuroscientists commonly measure the time-varying activity of neurons in the brain. Early studies explored how such activity directly encodes sensory stimuli. Since then neural responses have also been found to encode abstract parameters such as expected reward. Yet not all aspects of neural activity directly encode identifiable parameters: patterns of activity sometimes reflect the evolution of underlying internal computations, and may be only obliquely related to specific parameters. For example, it remains debated whether cortical activity during movement relates to parameters such as reach velocity, to parameters such as muscle activity, or to underlying computations that culminate in the production of muscle activity. To address this question we exploited an unexpected fact. When activity directly encodes a parameter it tends to be mathematically simple in a very particular way. When activity reflects the evolution of a computation being performed by the network, it tends to be mathematically simple in a different way. We found that responses in a visual area were simple in the first way, consistent with encoding of parameters. We found that responses in a motor area were simple in the second way, consistent with participation in the underlying computations that culminate in movement.

Reorganization between preparatory and movement population responses in motor cortex

Neural populations can change the computation they perform on very short timescales. Although such flexibility is common, the underlying computational strategies at the population level remain unknown. To address this gap, we examined population responses in motor cortex during reach preparation and movement. We found that there exist exclusive and orthogonal population-level subspaces dedicated to preparatory and movement computations. This orthogonality yielded a reorganization in response correlations: the set of neurons with shared response properties changed completely between preparation and movement. Thus, the same neural population acts, at different times, as two separate circuits with very different properties. This finding is not predicted by existing motor cortical models, which predict overlapping preparation-related and movement-related subspaces. Despite orthogonality, responses in the preparatory subspace were lawfully related to subsequent responses in the movement subspace. These results reveal a population-level strategy for performing separate but linked computations.

Single neuron responses are highly complex and dynamic yet they are able to flexibly represent behaviour through their collective activity. Here the authors demonstrate that population activity patterns of motor cortex neurons are orthogonal during successive task epochs that are linked through a simple linear function.

Clinical applications of penetrating neural interfaces and Utah Electrode Array technologies

This paper briefly describes some of the recent progress in the development of penetrating microelectrode arrays and highlights the use of two of these devices, Utah electrode arrays and Utah slanted electrode arrays, in two therapeutic interventions: recording volitional skeletal motor commands from the central nervous system, and recording motor commands and evoking somatosensory percepts in the peripheral nervous system (PNS). The paper also briefly explores other potential sites for microelectrode array interventions that could be profitably pursued and that could have important consequences in enhancing the quality of life of patients that has been compromised by disorders of the central and PNSs.

The Largest Response Component in the Motor Cortex Reflects Movement Timing but Not Movement Type

Neural activity in monkey motor cortex (M1) and dorsal premotor cortex (PMd) can reflect a chosen movement well before that movement begins. The pattern of neural activity then changes profoundly just before movement onset. We considered the prediction, derived from formal considerations, that the transition from preparation to movement might be accompanied by a large overall change in the neural state that reflects when movement is made rather than which movement is made. Specifically, we examined “components” of the population response: time-varying patterns of activity from which each neuron’s response is approximately composed. Amid the response complexity of individual M1 and PMd neurons, we identified robust response components that were “condition-invariant”: their magnitude and time course were nearly identical regardless of reach direction or path. These condition-invariant response components occupied dimensions orthogonal to those occupied by the “tuned” response components. The largest condition-invariant component was much larger than any of the tuned components; i.e., it explained more of the structure in individual-neuron responses. This condition-invariant response component underwent a rapid change before movement onset. The timing of that change predicted most of the trial-by-trial variance in reaction time. Thus, although individual M1 and PMd neurons essentially always reflected which movement was made, the largest component of the population response reflected movement timing rather than movement type.

Movement: How the Brain Communicates with the World

Voluntary movement is a result of signals transmitted through a communication channel that links the internal world in our minds to the physical world around us. Intention can be considered the desire to effect change on our environment, and this is contained in the signals from the brain, passed through the nervous system to converge on muscles that generate displacements and forces on our surroundings. The resulting changes in the world act to generate sensations that feed back to the nervous system, closing the control loop. This Perspective discusses the experimental and theoretical underpinnings of current models of movement generation and the way they are modulated by external information. Movement systems embody intentionality and prediction, two factors that are propelling a revolution in engineering. Development of movement models that include the complexities of the external world may allow a better understanding of the neuronal populations regulating these processes, as well as the development of solutions for autonomous vehicles and robots, and neural prostheses for those who are motor impaired.

What single-cell stimulation has told us about neural coding

In recent years, single-cell stimulation experiments have resulted in substantial progress towards directly linking single-cell activity to movement and sensation. Recent advances in electrical recording and stimulation techniques have enabled control of single neuron spiking in vivo and have contributed to our understanding of neuronal coding schemes in the brain. Here, we review single neuron stimulation effects in different brain structures and how they vary with artificially inserted spike patterns. We briefly compare single neuron stimulation with other brain stimulation techniques. A key advantage of single neuron stimulation is the precise control of the evoked spiking patterns. Systematically varying spike patterns and measuring evoked movements and sensations enables 'decoding' of the single-cell spike patterns and provides insights into the readout mechanisms of sensory and motor cortical spikes.

Advancing brain-machine interfaces: moving beyond linear state space models

Advances in recent years have dramatically improved output control by Brain-Machine Interfaces (BMIs). Such devices nevertheless remain robotic and limited in their movements compared to normal human motor performance. Most current BMIs rely on transforming recorded neural activity to a linear state space composed of a set number of fixed degrees of freedom. Here we consider a variety of ways in which BMI design might be advanced further by applying non-linear dynamics observed in normal motor behavior. We consider (i) the dynamic range and precision of natural movements, (ii) differences between cortical activity and actual body movement, (iii) kinematic and muscular synergies, and (iv) the implications of large neuronal populations. We advance the hypothesis that a given population of recorded neurons may transmit more useful information than can be captured by a single, linear model across all movement phases and contexts. We argue that incorporating these various non-linear characteristics will be an important next step in advancing BMIs to more closely match natural motor performance.

Recasting brain-machine interface design from a physical control system perspective

With the goal of improving the quality of life for people suffering from various motor control disorders, brain-machine interfaces provide direct neural control of prosthetic devices by translating neural signals into control signals. These systems act by reading motor intent signals directly from the brain and using them to control, for example, the movement of a cursor on a computer screen. Over the past two decades, much attention has been devoted to the decoding problem: how should recorded neural activity be translated into the movement of the cursor? Most approaches have focused on this problem from an estimation standpoint, i.e., decoders are designed to return the best estimate of motor intent possible, under various sets of assumptions about how the recorded neural signals represent motor intent. Here we recast the decoder design problem from a physical control system perspective, and investigate how various classes of decoders lead to different types of physical systems for the subject to control. This framework leads to new interpretations of why certain types of decoders have been shown to perform better than others. These results have implications for understanding how motor neurons are recruited to perform various tasks, and may lend insight into the brain's ability to conceptualize artificial systems.

Brain computer interface learning for systems based on electrocorticography and intracortical microelectrode arrays

A brain-computer interface (BCI) system transforms neural activity into control signals for external devices in real time. A BCI user needs to learn to generate specific cortical activity patterns to control external devices effectively. We call this process BCI learning, and it often requires significant effort and time. Therefore, it is important to study this process and develop novel and efficient approaches to accelerate BCI learning. This article reviews major approaches that have been used for BCI learning, including computer-assisted learning, co-adaptive learning, operant conditioning, and sensory feedback. We focus on BCIs based on electrocorticography and intracortical microelectrode arrays for restoring motor function. This article also explores the possibility of brain modulation techniques in promoting BCI learning, such as electrical cortical stimulation, transcranial magnetic stimulation, and optogenetics. Furthermore, as proposed by recent BCI studies, we suggest that BCI learning is in many ways analogous to motor and cognitive skill learning, and therefore skill learning should be a useful metaphor to model BCI learning.

Reprogramming movements: extraction of motor intentions from cortical ensemble activity when movement goals change

The ability to inhibit unwanted movements and change motor plans is essential for behaviors of advanced organisms. The neural mechanisms by which the primate motor system rejects undesired actions have received much attention during the last decade, but it is not well understood how this neural function could be utilized to improve the efficiency of brain-machine interfaces (BMIs). Here we employed linear discriminant analysis (LDA) and a Wiener filter to extract motor plan transitions from the activity of ensembles of sensorimotor cortex neurons. Two rhesus monkeys, chronically implanted with multielectrode arrays in primary motor (M1) and primary sensory (S1) cortices, were overtrained to produce reaching movements with a joystick toward visual targets upon their presentation. Then, the behavioral task was modified to include a distracting target that flashed for 50, 150, or 250 ms (25% of trials each) followed by the true target that appeared at a different screen location. In the remaining 25% of trials, the initial target stayed on the screen and was the target to be approached. M1 and S1 neuronal activity represented both the true and distracting targets, even for the shortest duration of the distracting event. This dual representation persisted both when the monkey initiated movements toward the distracting target and then made corrections and when they moved directly toward the second, true target. The Wiener filter effectively decoded the location of the true target, whereas the LDA classifier extracted the location of both targets from ensembles of 50–250 neurons. Based on these results, we suggest developing real-time BMIs that inhibit unwanted movements represented by brain activity while enacting the desired motor outcome concomitantly.