Marianna Semprini

New perspectives on the dialogue between brains and machines

Brain-machine interfaces (BMIs) are mostly investigated as a means to provide paralyzed people with new communication channels with the external world. However, the communication between brain and artificial devices also offers a unique opportunity to study the dynamical properties of neural systems. This review focuses on bidirectional interfaces, which operate in two ways by translating neural signals into input commands for the device and the output of the device into neural stimuli. We discuss how bidirectional BMIs help investigating neural information processing and how neural dynamics may participate in the control of external devices. In this respect, a bidirectional BMI can be regarded as a fancy combination of neural recording and stimulation apparatus, connected via an artificial body. The artificial body can be designed in virtually infinite ways in order to observe different aspects of neural dynamics and to approximate desired control policies.

Shaping the Dynamics of a Bidirectional Neural Interface

Progress in decoding neural signals has enabled the development of interfaces that translate cortical brain activities into commands for operating robotic arms and other devices. The electrical stimulation of sensory areas provides a means to create artificial sensory information about the state of a device. Taken together, neural activity recording and microstimulation techniques allow us to embed a portion of the central nervous system within a closed-loop system, whose behavior emerges from the combined dynamical properties of its neural and artificial components. In this study we asked if it is possible to concurrently regulate this bidirectional brain-machine interaction so as to shape a desired dynamical behavior of the combined system. To this end, we followed a well-known biological pathway. In vertebrates, the communications between brain and limb mechanics are mediated by the spinal cord, which combines brain instructions with sensory information and organizes coordinated patterns of muscle forces driving the limbs along dynamically stable trajectories. We report the creation and testing of the first neural interface that emulates this sensory-motor interaction. The interface organizes a bidirectional communication between sensory and motor areas of the brain of anaesthetized rats and an external dynamical object with programmable properties. The system includes (a) a motor interface decoding signals from a motor cortical area, and (b) a sensory interface encoding the state of the external object into electrical stimuli to a somatosensory area. The interactions between brain activities and the state of the external object generate a family of trajectories converging upon a selected equilibrium point from arbitrary starting locations. Thus, the bidirectional interface establishes the possibility to specify not only a particular movement trajectory but an entire family of motions, which includes the prescribed reactions to unexpected perturbations.

Proprioceptive assessment of the wrist joint across both joint degrees of freedom

Neurological movement disorders such as stroke or sensory neuropathy are associated with somatosensory deficits. From a neurorehabilitation perspective, the assessment of proprioceptive function is important for planning and applying adequate rehabilitation treatments. Numerous behavioral and psychophysical methods are available to measure proprioceptive acuity. However, no universally accepted and adopted protocol exists. In recent years robotic devices have increasingly been used to investigate and assess proprioceptive function, but few studies have focused on the wrist joint. To fill this knowledge gap, this study aimed to systematically map the proprioceptive acuity of the wrist for its two joint degrees of freedom (DoF) - flexion/extension (FE) and abduction/adduction (AA). Twenty eight healthy young adults performed an ipsilateral, active joint position matching task using a robotic device. As a measure of proprioceptive acuity we determined the error between target position and the matched joint position. Results showed that: first, proprioceptive acuity varied between the two joint DoF with the matching error for AA being lower than the FE. Second, within each DoF, the motion direction did not affect the accuracy. Third, the radial component of the matching error showed DoF dependence: the FE movements tended to undershoot, while the AA movement overshot the target position. Results are indicative of a joint DoF dependent anisotropy of wrist proprioceptive acuity.

A parametric study of intracortical microstimulation in behaving rats for the development of artificial sensory channels

In the framework of developing new brain-machine interfaces, many valuable results have been obtained in understanding which features of neural activity can be used in controlling an external device. Somatosensory real-time feedback is crucial for motor planning and for executing “online” errors correction during the movement. In people with sensory motor disabilities cortical microstimulation can be used as sensory feedback to elicit an artificial sensation providing the brain with information about the external environment. Even if intracortical microstimulation (ICMS) is broadly used in several experiments, understanding the psychophysics of such artificial sensory channel is still an open issue. Here we present the results of a parametric study that aims to define which stimulation parameters are needed to create an artificial sensation. Behaving rats were trained to report by pressing a lever the presence of ICMS delivered through microwire electrodes chronically implanted in the barrel cortex. Psychometric curves obtained by varying pulse amplitude, pulse frequency and train duration, demonstrate that in freely moving animals the perception threshold of microstimulation increased with respect to previous studies with head-restrained rats.

A Bidirectional Brain-Machine Interface Algorithm That Approximates Arbitrary Force-Fields

We examine bidirectional brain-machine interfaces that control external devices in a closed loop by decoding motor cortical activity to command the device and by encoding the state of the device by delivering electrical stimuli to sensory areas. Although it is possible to design this artificial sensory-motor interaction while maintaining two independent channels of communication, here we propose a rule that closes the loop between flows of sensory and motor information in a way that approximates a desired dynamical policy expressed as a field of forces acting upon the controlled external device. We previously developed a first implementation of this approach based on linear decoding of neural activity recorded from the motor cortex into a set of forces (a force field) applied to a point mass, and on encoding of position of the point mass into patterns of electrical stimuli delivered to somatosensory areas. However, this previous algorithm had the limitation that it only worked in situations when the position-to-force map to be implemented is invertible. Here we overcome this limitation by developing a new non-linear form of the bidirectional interface that can approximate a virtually unlimited family of continuous fields. The new algorithm bases both the encoding of position information and the decoding of motor cortical activity on an explicit map between spike trains and the state space of the device computed with Multi-Dimensional-Scaling. We present a detailed computational analysis of the performance of the interface and a validation of its robustness by using synthetic neural responses in a simulated sensory-motor loop.

Biofeedback Signals for Robotic Rehabilitation: Assessment of Wrist Muscle Activation Patterns in Healthy Humans

Electrophysiological recordings from human muscles can serve as control signals for robotic rehabilitation devices. Given that many diseases affecting the human sensorimotor system are associated with abnormal patterns of muscle activation, such biofeedback can optimize human–robot interaction and ultimately enhance motor recovery. To understand how mechanical constraints and forces imposed by a robot affect muscle synergies, we mapped the muscle activity of seven major arm muscles in healthy individuals performing goal-directed discrete wrist movements constrained by a wrist robot. We tested six movement directions and four force conditions typically experienced during robotic rehabilitation. We analyzed electromyographic (EMG) signals using a space-by-time decomposition and we identified a set of spatial and temporal modules that compactly described the EMG activity and were robust across subjects. For each trial, coefficients expressing the strength of each combination of modules and representing the underlying muscle recruitment, allowed for a highly reliable decoding of all experimental conditions. The decomposition provides compact representations of the observable muscle activation constrained by a robotic device. Results indicate that a low-dimensional control scheme incorporating EMG biofeedback could be an effective add-on for robotic rehabilitative protocols seeking to improve impaired motor function in humans.

A bidirectional brain-machine interface connecting alert rodents to a dynamical system

We present a novel experimental framework that implements a bidirectional brain-machine interface inspired by the operation of the spinal cord in vertebrates that generates a control policy in the form of a force field. The proposed experimental set-up allows connecting the brain of freely moving rats to an external device. We tested this apparatus in a preliminary experiment with an alert rat that used the interface for acquiring a food reward. The goal of this approach to bidirectional interfaces is to explore the role of voluntary neural commands in controlling a dynamical system represented by a small cart moving on vertical plane and connected to a water/pellet dispenser.

Dynamic brain-machine interface: A novel paradigm for bidirectional interaction between brains and dynamical systems

Brain-Machine Interfaces (BMIs) are systems which mediate communication between brains and artificial devices. Their long term goal is to restore motor functions, and this ultimately demands the development of a new generation of bidirectional brain-machine interfaces establishing a two-way brain-world communication channel, by both decoding motor commands from neural activity and providing feedback to the brain by electrical stimulation. Taking inspiration from how the spinal cord of vertebrates mediates communication between the brain and the limbs, here we present a model of a bidirectional brain-machine interface that interacts with a dynamical system by generating a control policy in the form of a force field. In our model, bidirectional communication takes place via two elements: (a) a motor interface decoding activities recorded from a motor cortical area, and (b) a sensory interface encoding the state of the controlled device into electrical stimuli delivered to a somatosensory area. We propose a specific mathematical model of the sensory and motor interfaces guiding a point mass moving in a viscous medium, and we demonstrate its performance by testing it on realistically simulated neural responses.

Robot-Assisted Proprioceptive Training with Added Vibro-Tactile Feedback Enhances Somatosensory and Motor Performance

This study examined the trainability of the proprioceptive sense and explored the relationship between proprioception and motor learning. With vision blocked, human learners had to perform goal-directed wrist movements relying solely on proprioceptive/haptic cues to reach several haptically specified targets. One group received additional somatosensory movement error feedback in form of vibro-tactile cues applied to the skin of the forearm. We used a haptic robotic device for the wrist and implemented a 3-day training regimen that required learners to make spatially precise goal-directed wrist reaching movements without vision. We assessed whether training improved the acuity of the wrist joint position sense. In addition, we checked if sensory learning generalized to the motor domain and improved spatial precision of wrist tracking movements that were not trained. The main findings of the study are: First, proprioceptive acuity of the wrist joint position sense improved after training for the group that received the combined proprioceptive/haptic and vibro-tactile feedback (VTF). Second, training had no impact on the spatial accuracy of the untrained tracking task. However, learners who had received VTF significantly reduced their reliance on haptic guidance feedback when performing the untrained motor task. That is, concurrent VTF was highly salient movement feedback and obviated the need for haptic feedback. Third, VTF can be also provided by the limb not involved in the task. Learners who received VTF to the contralateral limb equally benefitted. In conclusion, somatosensory training can significantly enhance proprioceptive acuity within days when learning is coupled with vibro-tactile sensory cues that provide feedback about movement errors. The observable sensory improvements in proprioception facilitates motor learning and such learning may generalize to the sensorimotor control of the untrained motor tasks. The implications of these findings for neurorehabilitation are discussed.

Robot-assisted training to improve proprioception does benefit from added vibro-tactile feedback.

Proprioception is central for motor control and its role must also be taken into account when designing motor rehabilitation training protocols. This is particularly important when dealing with motor deficits due to proprioceptive impairment such as peripheral sensory neuropathy. In these cases substituting or augmenting diminished proprioceptive sensory information might be beneficial for improving motor function. However it still remains to be understood how proprioceptive senses can be improved by training, how this would translate into motor improvement and whether additional sensory modalities during motor training contribute to the sensorimotor training process. This preliminary study investigated how proprioceptive/haptic training can be augmented by providing additional sensory information in the form of vibro-tactile feedback. We tested the acuity of the wrist proprioceptive position sense before and after robotic training in two groups of healthy subjects, one trained only with haptic feedback and one with haptic and vibro-tactile feedback. We found that only the group receiving the multimodal feedback significantly improved proprioceptive acuity. This study demonstrates that non-proprioceptive position feedback derived from another somatosensory modality is easily interpretable for humans and can contribute to an increased precision of joint position. The clinical implications of this finding will be outlined.