Search

Across many brain areas, neural population activity appears to be constrained to a low-dimensional manifold within a neural state space of considerably higher dimension. Recent studies of the primary motor cortex (M1) suggest that the activity within the low-dimensional manifold, rather than the activity of individual neurons, underlies the computations required for planning and executing movements. To date, these studies have been limited to data obtained in constrained laboratory settings where monkeys executed repeated, stereotyped tasks. An open question is whether the observed low dimensionality of the neural manifolds is due to these constraints; the dimensionality of M1 activity during the execution of more natural and unconstrained movements, like walking and picking food, remains unknown. We have now found similarly low-dimensional manifolds associated with various unconstrained natural behaviors, with dimensionality only slightly higher than those associated with constrained laboratory behaviors. To quantify the extent to which these low-dimensional manifolds carry task-relevant information, we built task-specific linear decoders that predicted EMG activity from M1 manifold activity. In both settings, decoding performance based on activity within the estimated low-dimensional manifold was the same as decoding performance based on the activity of all recorded neurons. These results establish functional links between task-specific manifolds and motor behaviors, and highlight that both constrained and unconstrained behaviors are associated with low-dimensional M1 manifolds.

Existing intracortical brain computer interfaces (iBCIs) transform neural activity into control signals capable of restoring movement to persons with paralysis. However, the accuracy of the “decoder” at the heart of the iBCI typically degrades over time due to turnover of recorded neurons. To compensate, decoders can be recalibrated, but this requires the user to spend extra time and effort to provide the necessary data, then learn the new dynamics. As the recorded neurons change, one can think of the underlying movement intent signal being expressed in changing coordinates. If a mapping can be computed between the different coordinate systems, it may be possible to stabilize the original decoder’s mapping from brain to behavior without recalibration. We previously proposed a method based on Generalized Adversarial Networks (GANs), called “Adversarial Domain Adaptation Network” (ADAN), which aligns the distributions of latent signals within underlying low-dimensional neural manifolds. However, ADAN was tested on only a very limited dataset. Here we propose a method based on Cycle-Consistent Adversarial Networks (Cycle-GAN), which aligns the distributions of the full-dimensional neural recordings. We tested both Cycle-GAN and ADAN on data from multiple monkeys and behaviors and compared them to a linear method based on Procrustes Alignment of axes provided by Factor Analysis (PAF). Both GAN-based methods outperformed PAF. Cycle-GAN and ADAN (like PAF) are unsupervised and require little data, making them practical in real life. Overall, Cycle-GAN had the best performance and was easier to train and more robust than ADAN, making it ideal for stabilizing iBCI systems over time.Significance StatementThe inherent instabilities in the neural signals acquired by intracortical microelectrode arrays cause the performance of an intracortical brain computer interface (iBCI) decoder to drop over time, as the movement intent signal must essentially be recorded from neurons representing an ever-changing coordinate system. Here, we address this problem using Generative Adversarial Networks (GANs) to align these coordinates and compare their success to another, recently proposed linear method that uses Factor Analysis and Procrustes alignment. Our proposed methods are fully unsupervised, can be trained quickly, and require remarkably little new data. These methods should give iBCI users access to decoders with unchanging dynamics, and without the need for periodic supervised recalibration.

Ankle sprains are the most common musculoskeletal injury, typically resulting from excessive inversion of the ankle. One way to prevent excessive inversion and maintain ankle stability is to generate a stiffness that is sufficient to resist externally imposed rotations. Frontal-plane ankle stiffness increases as participants place more weight on their ankle, but whether this effect is due to muscle activation or axial loading of the ankle is unknown. Identifying whether and to what extent axial loading affects ankle stiffness is important in understanding what role the passive mechanics of the ankle joint play in maintaining its stability. The objective of this study was to determine the effect of passive axial load on frontal-plane ankle stiffness. We had subjects seated in a chair as an axial load was applied to the ankle ranging from 10% to 50% body weight. Small rotational perturbations were applied to the ankle in the frontal plane to estimate stiffness. We found a significant, linear, 3-fold increase in ankle stiffness with axial load from the range of 0% bodyweight to 50% bodyweight. This increase could not be due to muscle activity as we observed no significant axial-load-dependent change in any of the recorded muscle activations. These results demonstrate that axial loading is a significant contributor to maintaining frontal-plane ankle stability, and that disruptions to the mechanism mediating this sensitivity of stiffness to axial loading may result in pathological cases of ankle instability.

The spinal stretch reflex is a fundamental building block of motor function, with a sensitivity that varies continuously during movement and when changing between movement and posture. Many have investigated task-dependent reflex sensitivity, but few have provided simple, quantitative analyses of the relationship between the volitional control and stretch reflex sensitivity throughout tasks that require coordinated activity of several muscles. Here we develop such an analysis and use it to test the hypothesis that modulation of reflex sensitivity during movement can be explained by the balance of activity within agonist and antagonist muscles better than by activity only in the muscle homonymous with the reflex. Subjects completed hundreds of flexion and extension movements as small, pseudo-random perturbations of elbow angle were applied to obtain estimates of stretch reflex amplitude throughout the movement. A subset of subjects performed a postural control task at muscle activities matched to those during movement. We found that reflex modulation during movement can be described by background activity in antagonist muscles about the elbow much better than by activity only in the muscle homonymous to the reflex (pNew and NoteworthyThis study examined the sensitivity of the stretch reflexes elicited in elbow muscles to the background activity in these same muscles during movement and postural tasks. We found a heightened reciprocal control of reflex sensitivity during movement that was not present during maintenance of posture. These results help explain previous discrepancies in reflex sensitivity measured during movement and posture and provide a simple model for assessing their contributions to muscle activity in both tasks.

Ultrasound shear wave elastography can be used to characterize mechanical properties of unstressed tissue by measuring shear wave velocity (SWV), which increases with increasing tissue stiffness. Measurements of SWV have often been assumed to be directly related to the stiffness of muscle. Some have also used measures of SWV to estimate stress, since muscle stiffness and stress covary during active contractions. However, few have considered the direct influence of muscle stress on SWV, independent of the stress-dependent changes in muscle stiffness, even though it is well known that stress alters shear wave propagation. The objective of this study was to determine how well the theoretical dependency of SWV on stress can account for measured changes of SWV in passive and active muscle. Data were collected from six isoflurane-anesthetized cats; three soleus muscles and three medial gastrocnemius muscles. Muscle stress and stiffness were measured directly along with SWV. Measurements were made across a range of passively and actively generated stresses, obtained by varying muscle length and activation, which was controlled by stimulating the sciatic nerve. Our results show that SWV depends primarily on the stress in a passively stretched muscle. In contrast, the SWV in active muscle is higher than would be predicted by considering only stress, presumably due to activation-dependent changes in muscle stiffness. Our results demonstrate that while SWV is sensitive to changes in muscle stress and activation, there is not a unique relationship between SWV and either of these quantities when considered in isolation.

The material properties of muscle play a central role in how muscle resists joint motion, transmits forces internally, and repairs itself. While many studies have evaluated muscle’s tensile material properties, few have investigated muscle’s shear properties. None of which have taken into account muscle’s anisotropic structure or investigated how different muscle architecture affect muscle’s shear properties. The objective of this study was to quantify the shear moduli of skeletal muscle in three orientations relevant to the function of whole muscle. We collected data from the extensor digitorum longus, tibialis anterior, and soleus harvested from both hindlimbs of 12 rats. These muscles were chosen to further evaluate the consistency of shear moduli across muscles with different architectures. We calculated the shear modulus in three orientations: parallel, perpendicular, and across with respect to muscle fiber alignment; while the muscle was subjected to increasing shear strain. For all muscles and orientations, the shear modulus increased with increasing strain. The shear modulus measured perpendicular to fibers was greater than in any other orientation. Despite architectural differences between muscles, we did not find a significant effect of muscle type on shear modulus. Our results show that in rat, muscles’ shear moduli vary with respect to fiber orientation and are not influenced by architectural differences in muscles.

There is an increasing use of shear wave ultrasound elastography to quantify mechanical properties of muscles under various conditions such as changes muscle length and levels of activation in healthy and pathological muscle. However, little is known about the variability in shear wave velocity among muscles as most studies investigate one specific muscle. The purpose of this study was to determine if the relationship between SWV and muscle activation is consistent across muscles with different architectures: biceps brachii, tibialis anterior, and medial gastrocnemius, All measures were made at matching levels of activation and approximately at the optimal length for each muscle to control for length-dependent changes in the relationship between activation and force or stiffness. We also conducted a control experiment to determine how the passive force within a muscle alters the relationship between muscle activation and shear wave velocity. The relationship between shear wave velocity-squared and activation above 10% MVC differed across muscles, with biceps brachii and medial gastrocnemius showing a lower slope than tibialis anterior. Shear wave velocity-squared also differed between muscles at the shortest length (p

Behaviors we perform every day, such as manipulating an object or walking, require precise control of interaction forces between our bodies and the environment. These forces are generated by active muscle contractions, specified by the nervous system, and by joint mechanics, determined by the intrinsic properties of the musculoskeletal system. Depending on behavioral goals, joint mechanics might simplify or complicate control of movement by the nervous system. However, whether humans can exploit joint mechanics to simplify neural control remains unclear. Here we evaluated if leveraging joint mechanics can simplify neural control by comparing performance in three tasks that required subjects to generate specified torques about the ankle during imposed sinusoidal movements; only one task required torques that could be generated by leveraging the intrinsic mechanics of the joint. We developed a novel approach that used continuous estimates of impedance, a quantitative description of joint mechanics, and measures of muscle activity to determine the mechanical and neural contributions to the net ankle torque generated in each motor task. We found that the torque resulting from changes in neural control was reduced when ankle impedance was consistent with the task being performed, resulting in a task that required less muscular effort. Subjects perceived this task to be easier than those that were not consistent with the impedance of the ankle and were able to perform it with the highest level of consistency. These results demonstrate that leveraging the mechanical properties of a joint can simplify task completion and improve performance.KEY POINTSInteracting with our environment requires production of interaction forces, which are generated by muscle contractions, specified by the nervous system, and by joint mechanics, determined by the intrinsic properties of the musculoskeletal system.We assessed whether leveraging joint mechanics can simplify neural control by having subjects complete 3 tasks, only one of which could be accomplished by leveraging joint mechanics.We found that subjects reduced their muscular effort, perceived the task to be easier and completed the task more consistently when the mechanics of the ankle were consistent with the task.These results highlight the importance of considering limb mechanics when interpreting measures of neural control related to movement and may benefit the design of mechanical interfaces that optimize human performance.

Many motor tasks involve transitions from posture to movement such as shifting from holding an object to setting it on a surface. Although many movements are voluntary processes, they are facilitated through motor pathways that span the continuum between reflexive and voluntary control. The mechanisms driving the long-latency stretch reflex (LLSR) have remained hotly contested. Recently, startReact has been shown to influence the fast muscular response to perturbations (Ravichandran et. al. 2013). The objective of this study was to evaluate how the LLSR and startReact impact the muscular response during the transition from posture to movement. We hypothesized that both the LLSR and startReact would be involved throughout the transition from posture to movement; however there would be a clear transition from LLSR dominance during posture to startReact dominance near/during movement. We tested this hypothesis using perturbations of elbow posture at various times before a fast ballistic extension movement. We found clear evidence that both the LLSR and startReact components were influenced changes in the late long-latency time window. The results provide insights on how the nervous system regulates involuntary responses to perturbations during the transition from maintenance of arm posture to movement.New and NoteworthyWe recently demonstrated that the startReact effect, the startled release of a planned movement, can influence the muscular response during the LLSR. Here we observe how the influence of startReact can change from the transition from posture (no/little influence) to movement (strong influence). While not all paradigms trigger a startReact effect, this work demonstrates that when present startRect can have a profound effect on the overall muscle activity - even obscuring the traditional LLSR response.