Brain-machine interface learning is facilitated by specific patterning of distributed cortical feedback.

Neuroprosthetics offer great hope for motor-impaired patients. One obstacle is that fine motor control requires near-instantaneous, rich somatosensory feedback. Such distributed feedback may be recreated in a brain-machine interface using distributed artificial stimulation across the cortical surface. Here, we hypothesized that neuronal stimulation must be contiguous in its spatiotemporal dynamics to be efficiently integrated by sensorimotor circuits. Using a closed-loop brain-machine interface, we trained head-fixed mice to control a virtual cursor by modulating the activity of motor cortex neurons. We provided artificial feedback in real time with distributed optogenetic stimulation patterns in the primary somatosensory cortex. Mice developed a specific motor strategy and succeeded to learn the task only when the optogenetic feedback pattern was spatially and temporally contiguous while it moved across the topography of the somatosensory cortex. These results reveal spatiotemporal properties of the sensorimotor cortical integration that set constraints on the design of neuroprosthetics.

Imaging the brain in action: a motorized optical rotary joint for wide field fibroscopy in freely moving animals.

Significance: The study of neuronal processes governing behavior in awake behaving mice is constantly boosted by the development of technological strategies, such as miniaturized microscopes and closed-loop virtual reality systems. However, the former limits the quality of recorded signals due to constrains in size and weight and the latter suffers from the restriction of the movement repertoire of the animal, therefore, hardly reproducing the complexity of natural multisensory scenes. Aim: Another strategy that takes advantage of both approaches consists of the use of a fiber-bundle interface to carry optical signals from a moving animal to a conventional imaging system. However, as the bundle is usually fixed below the optics, its torsion resulting from rotations of the animal inevitably constrains the behavior over long recordings. Our aim was to overcome this major limitation of fibroscopic imaging. Approach: We developed a motorized optical rotary joint controlled by an inertial measurement unit at the animal's head. Results: We show its principle of operation, demonstrate its efficacy in a locomotion task, and propose several modes of operation for a wide range of experimental designs. Conclusions: Combined with an optical rotary joint, fibroscopic approaches represent an outstanding tool to link neuronal activity with behavior in mice at the millisecond timescale.

Learning in a closed-loop brain-machine interface with distributed optogenetic cortical feedback.

Objective.Distributed microstimulations at the cortical surface can efficiently deliver feedback to a subject during the manipulation of a prosthesis through a brain-machine interface (BMI). Such feedback can convey vast amounts of information to the prosthesis user and may be key to obtain an accurate control and embodiment of the prosthesis. However, so far little is known of the physiological constraints on the decoding of such patterns. Here, we aimed to test a rotary optogenetic feedback that was designed to encode efficiently the 360° movements of the robotic actuators used in prosthetics. We sought to assess its use by mice that controlled a prosthesis joint through a closed-loop BMI.Approach.We tested the ability of mice to optimize the trajectory of a virtual prosthesis joint in order to solve a rewarded reaching task. They could control the speed of the joint by modulating the activity of individual neurons in the primary motor cortex. During the task, the patterned optogenetic stimulation projected on the primary somatosensory cortex continuously delivered information to the mouse about the position of the joint.Main results.We showed that mice are able to exploit the continuous, rotating cortical feedback in the active behaving context of the task. Mice achieved better control than in the absence of feedback by detecting reward opportunities more often, and also by moving the joint faster towards the reward angular zone, and by maintaining it longer in the reward zone. Mice controlling acceleration rather than speed of the joint failed to improve motor control.Significance.These findings suggest that in the context of a closed-loop BMI, distributed cortical feedback with optimized shapes and topology can be exploited to control movement. Our study has direct applications on the closed-loop control of rotary joints that are frequently encountered in robotic prostheses.

Continuity within the somatosensory cortical map facilitates learning.

The topographic organization is a prominent feature of sensory cortices, but its functional role remains controversial. Particularly, it is not well determined how integration of activity within a cortical area depends on its topography during sensory-guided behavior. Here, we train mice expressing channelrhodopsin in excitatory neurons to track a photostimulation bar that rotated smoothly over the topographic whisker representation of the primary somatosensory cortex. Mice learn to discriminate angular positions of the light bar to obtain a reward. They fail not only when the spatiotemporal continuity of the photostimulation is disrupted in this area but also when cortical areas displaying map discontinuities, such as the trunk and legs, or areas without topographic map, such as the posterior parietal cortex, are photostimulated. In contrast, when cortical topographic continuity enables to predict future sensory activation, mice demonstrate anticipation of reward availability. These findings could be helpful for optimizing feedback while designing cortical neuroprostheses.

An elaborate sweep-stick code in rat barrel cortex.

In rat barrel cortex, feature encoding schemes uncovered during broadband whisker stimulation are hard to reconcile with the simple stick-slip code observed during natural tactile behaviors, and this has hindered the development of a generalized computational framework. By designing broadband artificial stimuli to sample the inputs encoded under natural conditions, we resolve this disparity while markedly increasing the percentage of deep layer neurons found to encode whisker movements, as well as the diversity of these encoded features. Deep layer neurons encode two main types of events, sticks and sweeps, corresponding to high angular velocity bumps and large angular displacements with high velocity, respectively. Neurons can exclusively encode sticks or sweeps, or they can encode both, with or without direction selectivity. Beyond unifying coding theories from naturalistic and artificial stimulation studies, these findings delineate a simple and generalizable set of whisker movement features that can support a range of perceptual processes.

Spatiotemporal properties of whisker-evoked tactile responses in the mouse secondary somatosensory cortex.

The representation of rodents' mystacial vibrissae within the primary somatosensory (S1) cortex has become a major model for studying the cortical processing of tactile sensory information. However, upon vibrissal stimulation, tactile information first reaches S1 but also, almost simultaneously, the secondary somatosensory cortex (S2). To further understand the role of S2 in the processing of whisker inputs, it is essential to characterize the spatio-temporal properties of whisker-evoked response dynamics in this area. Here we describe the topography of the whiskers representation in the mouse S2 with voltage sensitive dye imaging. Analysis of the spatial properties of the early S2 responses induced by stimulating individually 22 to 24 whiskers revealed that they are spatially ordered in a mirror symmetric map with respect to S1 responses. Evoked signals in S2 and S1 are of similar amplitude and closely correlated at the single trial level. They confirm a short delay (~3 ms) between S1 and S2 early activation. In both S1 and S2 caudo-dorsal whiskers induce stronger responses than rostro-ventral ones. Finally, analysis of early C2-evoked responses indicates a faster activation of neighboring whisker representations in S2 relative to S1, probably due to the reduced size of the whisker map in S2.

Mechanical coupling through the skin affects whisker movements and tactile information encoding.

Rats use their whiskers to extract sensory information from their environment. While exploring, they analyze peripheral stimuli distributed over several whiskers. Previous studies have reported cross-whisker integration of information at several levels of the neuronal pathways from whisker follicles to the somatosensory cortex. In the present study, we investigated the possible coupling between whiskers at a preneuronal level, transmitted by the skin and muscles between follicles. First, we quantified the movement induced on one whisker by deflecting another whisker. Our results show significant mechanical coupling, predominantly when a given whisker's caudal neighbor in the same row is deflected. The magnitude of the effect was correlated with the diameter of the deflected whisker. In addition to changes in whisker angle, we observed curvature changes when the whisker shaft was constrained distally from the base. Second, we found that trigeminal ganglion neurons innervating a given whisker follicle fire action potentials in response to high-magnitude deflections of an adjacent whisker. This functional coupling also shows a bias toward the caudal neighbor located in the same row. Finally, we designed a two-whisker biomechanical model to investigate transmission of forces across follicles. Analysis of the whisker-follicle contact forces suggests that activation of mechanoreceptors in the ring sinus region could account for our electrophysiological results. The model can fully explain the observed caudal bias by the gradient in whisker diameter, with possible contribution of the intrinsic muscles connecting follicles. Overall, our study demonstrates the functional relevance of mechanical coupling on early information processing in the whisker system.NEW & NOTEWORTHY Rodents explore their environment actively by touching objects with their whiskers. A major challenge is to understand how sensory inputs from different whiskers are merged together to form a coherent tactile percept. We demonstrate that external sensory events on one whisker can influence the position of another whisker and, importantly, that they can trigger the activity of mechanoreceptors at its base. This cross-whisker interaction occurs pre-neuronally, through mechanical transmission of forces in the skin.

Rich spatio-temporal stimulus dynamics unveil sensory specialization in cortical area S2.

Tactile perception in rodents depends on simultaneous, multi-whisker contacts with objects. Although it is known that neurons in secondary somatosensory cortex (wS2) respond to individual deflections of many whiskers, wS2's precise function remains unknown. The convergence of information from multiple whiskers into wS2 neurons suggests that they are good candidates for integrating multi-whisker information. Here, we apply stimulation patterns with rich dynamics simultaneously to 24 macro-vibrissae of rats while recording large populations of single neurons. Varying inter-whisker correlations without changing single whisker statistics, we observe pronounced supra-linear multi-whisker integration. Using novel analysis methods, we show that continuous multi-whisker movements contribute to the firing of wS2 neurons over long temporal windows, facilitating spatio-temporal integration. In contrast, primary cortex (wS1) neurons encode fine features of whisker movements on precise temporal scales. These results provide the first description of wS2's representation during multi-whisker stimulation and outline its specialized role in parallel to wS1 tactile processing.

A fast intracortical brain-machine interface with patterned optogenetic feedback.

OBJECTIVE: The development of brain-machine interfaces (BMIs) brings new prospects to patients with a loss of autonomy. By combining online recordings of brain activity with a decoding algorithm, patients can learn to control a robotic arm in order to perform simple actions. However, in contrast to the vast amounts of somatosensory information channeled by limbs to the brain, current BMIs are devoid of touch and force sensors. Patients must therefore rely solely on vision and audition, which are maladapted to the control of a prosthesis. In contrast, in a healthy limb, somatosensory inputs alone can efficiently guide the handling of a fragile object, or ensure a smooth trajectory. We have developed a BMI in the mouse that includes a rich artificial somatosensory-like cortical feedback. APPROACH: Our setup includes online recordings of the activity of multiple neurons in the whisker primary motor cortex (vM1) and delivers feedback simultaneously via a low-latency, high-refresh-rate, spatially structured photo-stimulation of the whisker primary somatosensory cortex (vS1), based on a mapping obtained by intrinsic imaging. MAIN RESULTS: We demonstrate the operation of the loop and show that mice can detect the neuronal spiking in vS1 triggered by the photo-stimulations. Finally, we show that the mice can learn a behavioral task relying solely on the artificial inputs and outputs of the closed-loop BMI. SIGNIFICANCE: This is the first motor BMI that includes a short-latency, intracortical, somatosensory-like feedback. It will be a useful platform to discover efficient cortical feedback schemes towards future human BMI applications.

Supra-barrel Distribution of Directional Tuning for Global Motion in the Mouse Somatosensory Cortex.

Rodents explore their environment with an array of whiskers, inducing complex patterns of whisker deflections. Cortical neuronal networks can extract global properties of tactile scenes. In the primary somatosensory cortex, the information relative to the global direction of a spatiotemporal sequence of whisker deflections can be extracted at the single neuron level. To further understand how the cortical network integrates multi-whisker inputs, we imaged and recorded the mouse barrel cortex activity evoked by sequences of multi-whisker deflections generating global motions in different directions. A majority of barrel-related cortical columns show a direction preference for global motions with an overall preference for caudo-ventral directions. Responses to global motions being highly sublinear, the identity of the first deflected whiskers is highly salient but does not seem to determine the global direction preference. Our results further demonstrate that the global direction preference is spatially organized throughout the barrel cortex at a supra-columnar scale.

Representation of tactile scenes in the rodent barrel cortex.

After half a century of research, the sensory features coded by neurons of the rodent barrel cortex remain poorly understood. Still, views of the sensory representation of whisker information are increasingly shifting from a labeled line representation of single-whisker deflections to a selectivity for specific elements of the complex statistics of the multi-whisker deflection patterns that take place during spontaneous rodent behavior - so called natural tactile scenes. Here we review the current knowledge regarding the coding of patterns of whisker stimuli by barrel cortex neurons, from responses to single-whisker deflections to the representation of complex tactile scenes. A number of multi-whisker tunings have already been identified, including center-surround feature extraction, angular tuning during edge-like multi-whisker deflections, and even tuning to specific statistical properties of the tactile scene such as the level of correlation across whiskers. However, a more general model of the representation of multi-whisker information in the barrel cortex is still missing. This is in part because of the lack of a human intuition regarding the perception emerging from a whisker system, but also because in contrast to other primary sensory cortices such as the visual cortex, the spatial feature selectivity of barrel cortex neurons rests on highly nonlinear interactions that remained hidden to classical receptive field approaches.

Bilateral Discrimination of Tactile Patterns without Whisking in Freely Running Rats.

A majority of whisker discrimination tasks in rodents are performed on head-fixed animals to facilitate tracking or control of the sensory inputs. However, head fixation critically restrains the behavior and thus the incoming stimuli compared with those occurring in natural conditions. In this study, we investigated whether freely behaving rats can discriminate fine tactile patterns while running, in particular when stimuli are presented simultaneously on both sides of the snout. We developed a two-alternative forced-choice task in an automated modified T-maze. Stimuli were either a surface with no bars (smooth) or with vertical bars spaced irregularly or regularly. While running at full speed, rats encountered simultaneously the two discriminanda placed on the two sides of the central aisle. Rats learned to recognize regular bars versus a smooth surface in 8 weeks. They solved the task while running at an average speed of 1 m/s, so that the contact with the stimulus lasted <1 typical whisking cycle, precluding the use of active whisking. Whisker-tracking analysis revealed an asymmetry in the position of the whiskers: they oriented toward the rewarded stimulus during successful trials as early as 60 ms after the first possible contact. We showed that the whiskers and activity in the primary somatosensory cortex are involved during the discrimination process. Finally, we identified irregular patterns of bars that the rats can discriminate from the regular one. This novel task shows that freely moving rodents can make simultaneous bilateral tactile discrimination without whisking.SIGNIFICANCE STATEMENT The whisker system of rodents is a widely used model to study tactile processing. Rats show remarkable abilities in discriminating surfaces by actively moving their whiskers (whisking) against stimuli, typically sampling them several times. This motor strategy affects considerably the way that tactile information is acquired and thus the way that neuronal networks process the information. However, when rats run at high speed, they protract their whiskers in front of the snout without large movements. Here, we investigated whether rats are able to discriminate regular and irregular patterns of vertical bars while running without whisking. We found that the animals can perform a bilateral simultaneous discrimination without whisking and that this involves both whiskers and barrel cortex activity.

A radial map of multi-whisker correlation selectivity in the rat barrel cortex.

In the barrel cortex, several features of single-whisker stimuli are organized in functional maps. The barrel cortex also encodes spatio-temporal correlation patterns of multi-whisker inputs, but so far the cortical mapping of neurons tuned to such input statistics is unknown. Here we report that layer 2/3 of the rat barrel cortex contains an additional functional map based on neuronal tuning to correlated versus uncorrelated multi-whisker stimuli: neuron responses to uncorrelated multi-whisker stimulation are strongest above barrel centres, whereas neuron responses to correlated and anti-correlated multi-whisker stimulation peak above the barrel-septal borders, forming rings of multi-whisker synchrony-preferring cells.

An automated workflow for the anatomo-functional mapping of the barrel cortex.

BACKGROUND: The rodent barrel cortex is a widely used model to study the cortical processing of tactile sensory information. It is notable by the cytoarchitecture of its layer IV, which contains distinguishable structural units called barrels that can be considered as anatomical landmarks of the functional columnar organization of the cerebral cortex. To study sensory integration in the barrel cortex it is therefore essential to map recorded functional data onto the underlying barrel topography, which can be reconstructed from the post hoc alignment of tangential brain slices stained for cytochrome oxidase. NEW METHOD: This article presents an automated workflow to perform the registration of histological slices of the barrel cortex followed by the 2-D reconstruction of the barrel map from the registered slices. The registration of two successive slices is obtained by computing a rigid transformation to align sets of detected blood vessel cross-sections. This is achieved by using a robust variant of the classical iterative closest point method. A single fused image of the barrel field is then generated by computing a nonlinear merging of the gradients from the registered images. COMPARISON WITH EXISTING METHODS: This novel anatomo-functional mapping tool leads to a substantial gain in time and precision compared to conventional manual methods. It provides a flexible interface for the user with only a few parameters to tune. CONCLUSIONS: We demonstrate here the usefulness of the method for voltage sensitive dye imaging of the mouse barrel cortex. The method could also benefit other experimental approaches and model species.

An amplitude modulation/demodulation scheme for whisker-based texture perception.

Whisking rodents can discriminate finely textured objects using their vibrissae. The biomechanical and neural processes underlying such sensory tasks remain elusive. Here we combine the use of model micropatterned substrates and high-resolution videography of rats' whiskers during tactile exploration to study how texture information is mechanically encoded in the whisker motion. A biomechanical modeling of the whisker is developed, which yields quantitative predictions of the spectral and temporal characteristics of the observed whisker kinetics, for any given topography. These texture-induced whisker vibrations are then replayed via a multiwhisker stimulator while recording neuronal responses in the barrel field of the primary somatosensory cortex (S1bf). These results provide a comprehensive description of the transduction process at play during fine texture sensing in rats. They suggest that the sensory system operates through a vibratory amplitude modulation/demodulation scheme. Fine textural properties are encoded in the time-varying envelope of the whisker-resonant vibrations. This quantity is then recovered by neural demodulation, as it effectively drives the spiking-rate signal of a large fraction of S1 cortical neurons. This encoding/decoding scheme is shown to be robust against variations in exploratory conditions, such as the scanning speed or pad-to-substrate distance, thus allowing for reliable tactile discrimination in realistic conditions.

Bidirectional control of a one-dimensional robotic actuator by operant conditioning of a single unit in rat motor cortex.

The design of efficient neuroprosthetic devices has become a major challenge for the long-term goal of restoring autonomy to motor-impaired patients. One approach for brain control of actuators consists in decoding the activity pattern obtained by simultaneously recording large neuronal ensembles in order to predict in real-time the subject's intention, and move the prosthesis accordingly. An alternative way is to assign the output of one or a few neurons by operant conditioning to control the prosthesis with rules defined by the experimenter, and rely on the functional adaptation of these neurons during learning to reach the desired behavioral outcome. Here, several motor cortex neurons were recorded simultaneously in head-fixed awake rats and were conditioned, one at a time, to modulate their firing rate up and down in order to control the speed and direction of a one-dimensional actuator carrying a water bottle. The goal was to maintain the bottle in front of the rat's mouth, allowing it to drink. After learning, all conditioned neurons modulated their firing rate, effectively controlling the bottle position so that the drinking time was increased relative to chance. The mean firing rate averaged over all bottle trajectories depended non-linearly on position, so that the mouth position operated as an attractor. Some modifications of mean firing rate were observed in the surrounding neurons, but to a lesser extent. Notably, the conditioned neuron reacted faster and led to a better control than surrounding neurons, as calculated by using the activity of those neurons to generate simulated bottle trajectories. Our study demonstrates the feasibility, even in the rodent, of using a motor cortex neuron to control a prosthesis in real-time bidirectionally. The learning process includes modifications of the activity of neighboring cortical neurons, while the conditioned neuron selectively leads the activity patterns associated with the prosthesis control.

[What the whiskers tell the tactile brain]. / Ce que les vibrisses disent au cerveau tactile.

The rodent whisker system became one of the main system models for the study of the functional properties of sensory neurons. This is due on one hand to the detailed knowledge that we have on the afferent pathways linking the mechanoreceptors in the follicles to the primary somatosensory cortex and on the other hand to the possibility of controlling the sensory input at a micrometer and millisecond scale. The observation of the natural use of the whiskers by rodents indicates that exploration of objects and textures imply multiple contacts with tens of whiskers simultaneously. We have studied the neural code in the barrel cortex, which receives tactile information from the whiskers. By combining multi-electrode recordings and controlled multiwhisker tactile stimulation with theoretical analysis, we have observed a dependence of neural responses on the statistics of the sensory input. Several classes of neuronal responses, similar to those described in a number of cortical visual areas, were observed in the same cortical volume, indicating that various coding schemes are implemented in the same cortical network and can be put into play differentially to cope with the changing statistics of the peripheral stimuli.

"Master" neurons induced by operant conditioning in rat motor cortex during a brain-machine interface task.

Operant control of a prosthesis by neuronal cortical activity is one of the successful strategies for implementing brain-machine interfaces (BMI), by which the subject learns to exert a volitional control of goal-directed movements. However, it remains unknown if the induced brain circuit reorganization affects preferentially the conditioned neurons whose activity controlled the BMI actuator during training. Here, multiple extracellular single-units were recorded simultaneously in the motor cortex of head-fixed behaving rats. The firing rate of a single neuron was used to control the position of a one-dimensional actuator. Each time the firing rate crossed a predefined threshold, a water bottle moved toward the rat, until the cumulative displacement of the bottle allowed the animal to drink. After a learning period, most (88%) conditioned neurons raised their activity during the trials, such that the time to reward decreased across sessions: the conditioned neuron fired strongly, reliably and swiftly after trial onset, although no explicit instruction in the learning rule imposed a fast neuronal response. Moreover, the conditioned neuron fired significantly earlier and more strongly than nonconditioned neighboring neurons. During the first training sessions, an increase in firing rate variability was seen only for the highly conditionable neurons. This variability then decreased while the conditioning effect increased. These findings suggest that modifications during training target preferentially the neuron chosen to control the BMI, which acts then as a "master" neuron, leading in time the reconfiguration of activity in the local cortical network.

Correlated input reveals coexisting coding schemes in a sensory cortex.

As in other sensory modalities, one function of the somatosensory system is to detect coherence and contrast in the environment. To investigate the neural bases of these computations, we applied different spatiotemporal patterns of stimuli to rat whiskers while recording multiple neurons in the barrel cortex. Model-based analysis of the responses revealed different coding schemes according to the level of input correlation. With uncorrelated stimuli on 24 whiskers, we identified two distinct functional categories of neurons, analogous in the temporal domain to simple and complex cells of the primary visual cortex. With correlated stimuli, however, a complementary coding scheme emerged: two distinct cell populations, similar to reinforcing and antagonist neurons described in the higher visual area MT, responded specifically to correlations. We suggest that similar context-dependent coexisting coding strategies may be present in other sensory systems to adapt sensory integration to specific stimulus statistics.

Whisker encoding of mechanical events during active tactile exploration.

Rats use their whiskers to extract a wealth of information about their immediate environment, such as the shape, position or texture of an object. The information is conveyed to mechanoreceptors located within the whisker follicle in the form of a sequence of whisker deflections induced by the whisker/object contact interaction. How the whiskers filter and shape the mechanical information and effectively participate in the coding of tactile features remains an open question to date. In the present article, a biomechanical model was developed that provides predictions of the whisker dynamics during active tactile exploration, amenable to quantitative experimental comparison. This model is based on a decomposition of the whisker profile into a slow, quasi-static sequence and rapid resonant small-scale vibrations. It was applied to the typical situation of a rat actively whisking across a solid object. Having derived the quasi-static sequence of whisker deformation, the resonant properties of the whisker were analyzed, taking into account the boundary conditions imposed by the whisker/surface contact. We then focused on two elementary mechanical events that are expected to trigger significant neural responses, namely (1) the whisker/object first contact and (2) the whisker detachment from the object. Both events were found to trigger a deflection wave propagating upward to the mystacial pad at constant velocity of ≈3-5 m/s. This yielded a characteristic mechanical signature at the whisker base, in the form of a large peak of negative curvature occurring ≈4 ms after the event has been triggered. The dependence in amplitude and lag of this mechanical signal with the main contextual parameters (such as radial or angular distance) was investigated. The model was validated experimentally by comparing its predictions to high-speed video recordings of shock-induced whisker deflections performed on anesthetized rats. The consequences of these results on possible tactile encoding schemes are briefly discussed.