Classification of motor commands using a modified self-organising feature map

In this paper, a control system for an advanced prosthesis is proposed and has been investigated in two different biological systems: (1) the spinal withdrawal reflex system of a rat and (2) voluntary movements in two human males: one normal subject and one subject with a traumatic hand amputation. The small-animal system was used as a model system to test different processing methods for the prosthetic control system. The best methods were then validated in the human set-up. The recorded EMGs were classified using different ANN algorithms, and it was found that a modified self-organising feature map (SOFM) composed of a combination of a Kohonen network and the conscience mechanism algorithm (KNC) was superior in performance to the reference networks (e.g. multi-layer perceptrons) as regards training time, low memory consumption, and simplicity in finding optimal training parameters and architecture. The KNC network classified both experimental set-ups with high accuracy, including five movements for the animal set-up and seven for the human set-up.     

Hyperstriatal lesions and classical conditioning in the pigeon

In two experiments, pigeons with bilateral lesions of the hyperstriatum were compared with unoperated control birds and operated control subjects having bilateral lesions of the neostriatum on tasks designed to test two hypotheses of hyperstriatal function. In Experiment 1, hyperstriatal lesions impaired both the acquisition and maintenance of autoshaped responding as well as maintenance responding in response-omission training. In Experiment 2, hyperstriatal birds displayed depressed levels of responding relative to control birds on a classical go-no-go alternation schedule. These results support the view that hyperstriatal selectively disrupt classical conditioning and go against the view that hyperstriatal lesions exaggerate perseverative responding. A surprising aspect of the results was the performance of the operated control subjects that showed better performance in the acquisition of autoshaping and superior differentiation on the go-no-go schedule compared with the unoperated control birds.     

Functional imaging of changes in cerebellar activity related to learning during a novel eye–hand tracking task

Coordination between the eyes and the hand is likely to be based on a process of motor learning, so that the interactions between the two systems can be accurately controlled. By using an unusual tracking task we measured the change in brain activation levels, as recorded with 3T functional magnetic resonance imaging (fMRI), between naïve human subjects and the same subjects after a period of extended training. Initially the performance of the two groups was similar. One subject group was then trained in a synchronous, coordinated, eye–hand task; the other group trained with a 304 ms temporal offset between hand and eye tracking movements. After training, different patterns of performance were observed for the groups, and different functional activation profiles. Significant change in the relationship between functional activation levels and eye–hand task conditions was predominantly restricted to visuo-motor areas of the lateral and vermal cerebellum. In an additional test with one of the subject groups, we show that there was increased cerebellar activation after learning, irrespective of change in performance error. These results suggest that two factors contribute to the measured blood oxygen level-dependent (BOLD) signal. One declined with training and may be directly related to performance error. The other increased after training, in the test conditions nearest to the training condition, and may therefore be related to acquisition of experience in the task. The loci of activity changes suggest that improved performance is because of selective modified processing of ocular and manual control signals within the cerebellum. These results support the suggestion that coordination between eye and hand movement is based on an internal model acquired by the cerebellum that provides predictive signals linking the control of the two effectors.     

Sustained effects for training of smooth pursuit plasticity

Maintaining orientation in space is a multisensory process, with the vestibular, visual, auditory and somatosensory systems as inputs. Since the input from each individual system changes, for example due to aging, the central nervous system must continuously adapt to these changes to maintain proper system performance. Changes can also be elicited by targeted modifications of the inputs, or by controlled training of sensory systems. While the effects of adaptation on eye movements elicited by the vestibulo-ocular reflex are well established, modifications of the efficacy of smooth pursuit eye movements are less well understood. We have investigated whether two 6-min training sessions on three subsequent days can induce lasting changes in the open- and closed-loop smooth pursuit performance of healthy, adult subjects. Ten subjects practiced making pursuit eye movements by tracking a target cross which moved quasi-randomly on a computer screen. Smooth pursuit performance was tested with a step-ramp paradigm immediately before and after the training, as well as 5 days after the last training session. Our results show that even such short training sessions can induce significant, lasting improvements in closed-loop smooth pursuit performance if the pursuit system of the subjects is challenged sufficiently during training. Control experiments on ten additional adult subjects who had their pursuit performance tested before and after a 20 min break without visual training confirmed that the pursuit enhancement is due to the visual training and not due to perceptual learning.     

Improvement and generalization of arm motor performance through motor imagery practice

This study compares the improvement and generalization of arm motor performance after physical or mental training in a motor task requiring a speed-accuracy tradeoff. During the pre- and post-training sessions, 40 subjects pointed with their right arm as accurately and as fast as possible toward targets placed in the frontal plane. Arm movements were performed in two different workspaces called right and left paths. During the training sessions, which included only the right path, subjects were divided into four training groups (n = 10): (i) the physical group, subjects overtly performed the task; (ii) the mental group, subjects imagined themselves performing the task; (iii) the active control group, subjects performed eye movements through the targets, (iv) the passive control group, subjects did not receive any specific training. We recorded movement duration, peak acceleration and electromyographic signals from arm muscles. Our findings showed that after both physical and mental training on the right path (training path), hand movement duration and peak acceleration respectively decreased and increased for this path. However, motor performance improvement was greater after physical compared with mental practice. Interestingly, we also observed a partial learning generalization, namely an enhancement of motor performance for the left path (non-training path). The amount of this generalization was roughly similar for the physical and mental groups. Furthermore, while arm muscle activity progressively increased during the training period for the physical group, the activity of the same muscles for the mental group was unchanged and comparable with that of the rest condition. Control groups did not exhibit any improvement. These findings put forward the idea that mental training facilitates motor learning and allows its partial transfer to nearby workspaces. They further suggest that motor prediction, a common process during both actual and imagined movements, is a fundamental operation for both sensorimotor control and learning.     

Control of multifunctional prosthetic hands by processing the electromyographic signal

The human hand is a complex system, with a large number of degrees of freedom (DoFs), sensors embedded in its structure, actuators and tendons, and a complex hierarchical control. Despite this complexity, the efforts required to the user to carry out the different movements is quite small (albeit after an appropriate and lengthy training). On the contray, prosthetic hands are just a pale replication of the natural hand, with significantly reduced grasping capabilities and no sensory information delivered back to the user. Several attempts have been carried out to develop multifunctional prosthetic devices controlled by electromyographic (EMG) signals (myoelectric hands), harness (kinematic hands), dimensional changes in residual muscles, and so forth, but none ofthese methods permits the "natural" control of more than two DoFs. This article presents a review of the traditional methods used to control artificial hands by means of EMG signal, in both the clinical and research contexts, and introduces what could be the future developments in the control strategy of these devices.     

Application of robotics to prosthetic control

Multiaxis robotic and prosthetic systems have a number of common characteristics such that control techniques for robots may sometimes be adapted for prosthetic devices. One characteristic that is particularly important for these systems is that good performance over a wide range of motion and loads is difficult since it is dependent upon complex time-varying, nonlinear dynamics. This paper gives an example of the complexity of the dynamics for a two degree-of-freedom, double-inverted pendulum which has been used as a model for study of biped locomotion. Two different methods called the “inverse plant” and State Space Memory, now being considered for robotic mechanisms, are then presented which base the control on the dynamics of the system and may have application for prosthetic devices.     

The specificity of strength training: the effect of posture

There is a paucity of research into the importance of performing strength training exercises in postures specific to the movements they are attempting to facilitate. In addressing this question, 27 previously trained subjects were randomly allocated into heavy weight training and control groups. The weight training group performed 4–6 sets of 6–10 repetitions of the squat and bench press lifts twice a week for 8 weeks. Prior to and after the training period the following tests were conducted: bench press throw at 30% of maximal load, vertical jump, maximal squat and bench press lifts, push-up test performed over a force platform, 40-m sprint, 6-s cycle, and isokinetic tests assessing upper and lower body musculature in varying actions. The results supported the concept that posture is important in training as those exercises conducted in similar postures to the training recorded the greatest improvement in performance. For example, after completion of the training the weight training subjects significantly increased by approximately 12% the maximal load lifted in the bench press exercise and the peak torque in the isokinetic bench press test. However, performance in the isokinetic horizontal arm adduction test was not significantly changed. We speculate that the phenomenon of posture specificity may, at least in part, be caused by the differing postures altering the neural input to the musculature. The results stress the importance of selecting exercises in which the posture closely resembles that of the movements they are attempting to facilitate.     

Learned dynamics of reaching movements generalize from dominant to nondominant arm

Accurate performance of reaching movements depends on adaptable neural circuitry that learns to predict forces and compensate for limb dynamics. In earlier experiments, we quantified generalization from training at one arm position to another position. The generalization patterns suggested that neural elements learning to predict forces coded a limb's state in an intrinsic, muscle-like coordinate system. Here, we test the sensitivity of these elements to the other arm by quantifying inter-arm generalization. We considered two possible coordinate systems: an intrinsic (joint) representation should generalize with mirror symmetry reflecting the joint's symmetry and an extrinsic representation should preserve the task's structure in extrinsic coordinates. Both coordinate systems of generalization were compared with a na?ve control group. We tested transfer in right-handed subjects both from dominant to nondominant arm (D-->ND) and vice versa (ND-->D). This led to a 2 x 3 experimental design matrix: transfer direction (D-->ND/ND-->D) by coordinate system (extrinsic, intrinsic, control). Generalization occurred only from dominant to nondominant arm and only in extrinsic coordinates. To assess the dependence of generalization on callosal inter-hemispheric communication, we tested commissurotomy patient JW. JW showed generalization from dominant to nondominant arm in extrinsic coordinates. The results suggest that when the dominant right arm is used in learning dynamics, the information could be represented in the left hemisphere with neural elements tuned to both the right arm and the left arm. In contrast, learning with the nondominant arm seems to rely on the elements in the nondominant hemisphere tuned only to movements of that arm.     

Remapping hand movements in a novel geometrical environment

The issue of how the Euclidean properties of space are represented in the nervous system is a main focus in the study of visual perception, but is equally relevant to motor learning. The goal of our experiments was to investigate how the properties of space guide the remapping of motor coordination. Subjects wore an instrumented data glove that recorded the finger motions. Signals generated by the glove operated a remotely controlled endpoint: a cursor on a computer monitor. The subjects were instructed to execute movements of this endpoint with controlled motions of the fingers. This required inverting a highly redundant map from fingers to cursor motions. We found that 1) after training with visual feedback of the final error (but not of the ongoing cursor motion), subjects learned to map cursor locations into configurations of the fingers; 2) extended practice of movement led to more rectilinear cursor movement, a trend facilitated by training under continuous visual feedback of cursor motions; 3) with practice, subjects reduced motion in the degrees of freedom that did not contribute to the movements of the cursor; 4) with practice, subjects reduced variability of both cursor and hand movements; and 5) the reduction of errors and the increase in linearity generalized beyond the set of movements used for training. These findings suggest that subjects not only learned to produce novel coordinated movement to control the placement of the cursor, but they also developed a representation of the Euclidean space on which hand movements were remapped.     

Problems in Design of Neuro-controlled Prosthetic Devices

One of the most promising ways of constructing modern prosthetic control systems would be to use electrical control signals recorded from peripheral neural tissue. In this work the construction of such systems is analyzed in order to upgrade modern technical devices and methods that are used to design these systems. The requirements for technical devices and methods used to record and process neural signals and rules are also considered. The efficiency of the systems and methods is estimated and described. A generalized structure of neural prosthetic control systems and a way of preprocessing control signals to provide more efficient information transfer are suggested.     

Creativity and cortical activation during creative, intellectual and EEG feedback tasks.

Thirty-two male subjects were divided into four groups based on their performance on the remote associates test and alternate uses test, two measures of creativity. Right EEG alpha presence was monitored under basal conditions, while subjects took tests of creativity and intelligence, and while they attempted to enhance and suppress the amount of alpha in a feedback situation. High scorers on the alternate uses test operated at a high percentage of basal alpha during all tests while high scorers on the remote associates test showed differential amounts of alpha presence across tests, with the highest percentage of basal alpha during tests of creativity and the lowest percentage during an intellectual test. Both high creative groups tended to show increases in amount of alpha across trials when trying to suppress alpha as well as when trying to enhance it, but did not differ in overall control from the low creative groups.     

Oculo-manual tracking of visual targets: control learning,coordination control and coordination model

Summary The processes which develop to coordinate eye and hand movements in response to motion of a visual target were studied in young children and adults. We have shown that functional maturation of the coordination control between eye and hand takes place as a result of training. We observed, in the trained child and in the adult, that when the hand is used either as a target or to track a visual target, the dynamic characteristics of the smooth pursuit system are markedly improved: the eye to target delay is decreased from 150 ms in eye alone tracking to 30 ms, and smooth pursuit maximum velocity is increased by 100%. Coordination signals between arm and eye motor systems may be responsible for smooth pursuit eye movements which occur during self-tracking of hand or finger in darkness. These signals may also account for the higher velocity smooth pursuit eye movements and the shortened tracking delay when the hand is used as a target, as well as for the synkinetic eye-arm motions observed at the early stage of oculo-manual tracking training in children. We propose a model to describe the interaction which develops between two systems involved in the execution of a common sensorimotor task. The model applies to the visuo-oculo-manual tracking system, but it may be generalized to other coordinated systems. According to our definition, coordination control results from the reciprocal transfer of sensory and motor information between two or more systems involved in the execution of single, goal-directed or conjugate actions. This control, originating in one or more highly specialized structures of the central nervous system, combines with the control processes normally operating in each system. Our model relies on two essential notions which describe the dynamic and static aspects of coordination control: timing and mutual coupling.     

Low-level Taekwondo practitioners have better somatosensory organisation in standing balance than sedentary people

Sports training, especially for those requiring fast and skilled movements have been reported to improve one’s postural control, but the underlying sensory integration mechanism is unknown. The purpose is to explore the sensory organisation strategies for maintaining standing balance in Taekwondo practitioners, and to examine the quasi-static and dynamic balance performance in subjects with and without TKD training. Case–control study was used as a study design. Eleven subjects with low level of Taekwondo training for 1–3 years, and eleven sedentary healthy subjects were assessed with the sensory organisation tests (SOT) under six visual and somatosensory input conditions and their balance upon landing from self- or operator-triggered drop test with the eyes closed condition. The SOT measured the equilibrium scores, whereas the drop test assessed the time to stabilisation (TTS), normalised peak force and distance of antero-posterior and medial–lateral centre of pressure on landing. Results for the SOT test revealed that Taekwondo subjects performed better during stance with eyes closed on a fixed support than the untrained group (p = 0.011). For the drop tests, the untrained group was slower in postural correction as revealed by the longer TTS than the Taekwondo group after the operator-triggered drops (p = 0.018). All subjects had a larger normalised peak force in operator-triggered than self-triggered drops. In conclusion, we observed that people with low-level Taekwondo training have better balance performance than untrained subjects as shown in the SOT results and shorter TTS with the drop test. They may rely more on the somatosensory and vestibular inputs for maintaining balance. People with balance problems may benefit from Taekwondo training.     

Adaptive changes in vergence eye movements induced by vergence-vestibular interaction training in monkeys

Clear vision of objects moving in three-dimensional space near an observer is attained by a combination of smooth-pursuit and vergence eye movements. The two systems must interact with the vestibular system to maintain the image of the object on the fovea. Previous studies showed that training with smooth-pursuit vestibular interactions resulted in adaptive changes in the smooth-pursuit response. Although vergence and smooth-pursuit systems are thought to have separate neural substrates, recent studies indicate that the caudal parts of the frontal eye fields that receive vestibular inputs contain neurons that discharge in response to combinations of smooth-pursuit and vergence. This combination of discharge sensitivities suggests the possibility that adaptive changes may be induced in the vergence system by vestibular inputs during vergence-pursuit training. To explore this possibility, we examined the effects of training with conflicting vestibular and vergence tracking in four head-stabilized monkeys. Animals were rewarded for tracking a laser spot that moved towards or away from them at 1 Hz in phase with sinusoidal whole-body rotation (±5°) in the pitch plane; the spot moved closer when the monkeys nose moved downward. From the monkeys point of view, the spot moved sinusoidally 10–66 cm in front of them along the mid-sagittal plane, requiring symmetrical vergence eye movements of 4.8° for each eye. Eye movements induced by equivalent spot motion at 0.3–1.0 Hz with or without chair rotation were examined before and after training for each session (0.5–1.0 h). Before training, pitch rotation alone in complete darkness did not induce vergence eye movements in any of the monkeys tested. Vergence tracking without chair rotation showed decreased gain and increased phase lag (re vergence target velocity) at frequencies above 0.5 Hz. After training, the vergence response during chair rotation with the spot showed significantly higher gains and smaller phase lags at 0.3–1.0 Hz in all monkeys. Pitch rotation alone in complete darkness induced vergence eye movements with gains (eye vergence/chair) of 0.15–0.35 after training in two monkeys. These results suggest that vestibular information can be used effectively to modify vergence tracking.     

Stroke Survivors Control the Temporal Structure of Variability During Reaching in Dynamic Environments

Learning to control forces is known to reduce the amount of movement variability (e.g., standard deviation; SD) while also altering the temporal structure of movement variability (e.g., approximate entropy; ApEn). Such variability control has not been explored in stroke survivors during reaching movements in dynamic environments. Whether augmented feedback affects such variability control, is also unknown. Chronic stroke survivors, assigned randomly to a control/experimental group, learned reaching movements in a dynamically changing environment while receiving either true feedback of their movement (control) or augmented visual feedback (experimental). Hand movement variability was analyzed using SD and ApEn. A significant change in variability was determined for both SD and ApEn. Post hoc tests revealed that the significant decrease in SD was not retained after a week. However, the significant increase in ApEn, determined on both days of training, showed significant retention effects. In dynamically changing environments, chronic stroke survivors reduced the amount of movement variability and made their movement patterns less repeatable and possibly more flexible. These changes were not affected by augmented visual feedback. Moreover, the learning patterns characteristically involved the control of the nonlinear dynamics rather than the amount of hand movement variability. The absence of transfer effects demonstrated that variability control of hand movement after a stroke is specific to the task and the environment.     

A closed-loop human simulator for investigating the role of feedback control in brain-machine interfaces

Neural prosthetic systems seek to improve the lives of severely disabled people by decoding neural activity into useful behavioral commands. These systems and their decoding algorithms are typically developed "offline," using neural activity previously gathered from a healthy animal, and the decoded movement is then compared with the true movement that accompanied the recorded neural activity. However, this offline design and testing may neglect important features of a real prosthesis, most notably the critical role of feedback control, which enables the user to adjust neural activity while using the prosthesis. We hypothesize that understanding and optimally designing high-performance decoders require an experimental platform where humans are in closed-loop with the various candidate decode systems and algorithms. It remains unexplored the extent to which the subject can, for a particular decode system, algorithm, or parameter, engage feedback and other strategies to improve decode performance. Closed-loop testing may suggest different choices than offline analyses. Here we ask if a healthy human subject, using a closed-loop neural prosthesis driven by synthetic neural activity, can inform system design. We use this online prosthesis simulator (OPS) to optimize "online" decode performance based on a key parameter of a current state-of-the-art decode algorithm, the bin width of a Kalman filter. First, we show that offline and online analyses indeed suggest different parameter choices. Previous literature and our offline analyses agree that neural activity should be analyzed in bins of 100- to 300-ms width. OPS analysis, which incorporates feedback control, suggests that much shorter bin widths (25-50 ms) yield higher decode performance. Second, we confirm this surprising finding using a closed-loop rhesus monkey prosthetic system. These findings illustrate the type of discovery made possible by the OPS, and so we hypothesize that this novel testing approach will help in the design of prosthetic systems that will translate well to human patients.     

Goal directed motor behavior and its adaptation following reversed tactile perception in man

Summary When two adjacent fingers are crossed over each other and two tactile stimuli are touched to the two crossed fingertips, the two stimuli are perceived to be inverted in space. This phenomenon of tactile reversal was used in the present work to study the sensorimotor transformation occurring in goal directed motor behavior. When the subjects had to perform a movement toward a tactile stimulus (target) in tactile reversal conditions, the stimulus directed movements were performed wrongly, that is, away from the target. However, not all the subjects perceived a complete inversion of the stimuli; in this case, the target stimulus was perceived to be on the same side as its actual position, although with an error. In these conditions, the stimulus directed movements were performed correctly, that is, toward the target. These results show that the illusory spatial perception of the stimuli controls motor behavior on the basis of the amount of the perceptual error. Within the first hour of training, several compensatory responses occurred so as to produce correct motor performance. Despite this motor learning, reversed tactile perception remained reversed. Therefore, what subjects learned was the execution of the movements opposite to those necessary for reaching the target stimulus, without any change in perception. In the context of theories concerning the relationship between motor learning and perceptual adaptation, the present study shows that, with this experimental paradigm, motor learning during the first hour was mainly cognitive and did not have short-term effects upon perceptual processes.     

Spike shape analysis of surface electromyographic activity in wrist flexor and extensor muscles of the world's fastest drummer

Spike shape analysis (SSA) is a method to infer motor unit (MU) activity by examining interference pattern of surface electromyography (sEMG). SSA has succeeded to assess neuromuscular adaptations after dynamic training; however, it has not been used to assess muscle activities during the dynamic movements as seen in music performance. The present study used SSA to investigate sEMG activities of wrist flexor and extensor muscles in the winner of a contest to find the world's fastest drummer (WFD) during performing rhythmic wrist flexion/extension movements with one hand using a handheld drumstick. SSA measures of the WFD were compared with those in the two control groups: non-drummers (NDs) and ordinary drummers (ODs). We found that the WFD showed significantly high mean spike frequency (MSF), short mean spike duration (MSD), and small mean number of peaks per spike (MNPPS) compared with the control groups. These results suggest that the WFD had exceptional MU activity such as higher MU discharge rate, more MU recruitment, and/or higher MU synchronization to achieve extraordinary fast 10-Hz drumming performance. SSA will be useful to investigate the muscle activity seen in music performance.