Inferring visuomotor priors for sensorimotor learning

Sensorimotor learning has been shown to depend on both prior expectations and sensory evidence in a way that is consistent with Bayesian integration. Thus, prior beliefs play a key role during the learning process, especially when only ambiguous sensory information is available. Here we develop a novel technique to estimate the covariance structure of the prior over visuomotor transformations--the mapping between actual and visual location of the hand--during a learning task. Subjects performed reaching movements under multiple visuomotor transformations in which they received visual feedback of their hand position only at the end of the movement. After experiencing a particular transformation for one reach, subjects have insufficient information to determine the exact transformation, and so their second reach reflects a combination of their prior over visuomotor transformations and the sensory evidence from the first reach. We developed a Bayesian observer model in order to infer the covariance structure of the subjects' prior, which was found to give high probability to parameter settings consistent with visuomotor rotations. Therefore, although the set of visuomotor transformations experienced had little structure, the subjects had a strong tendency to interpret ambiguous sensory evidence as arising from rotation-like transformations. We then exposed the same subjects to a highly-structured set of visuomotor transformations, designed to be very different from the set of visuomotor rotations. During this exposure the prior was found to have changed significantly to have a covariance structure that no longer favored rotation-like transformations. In summary, we have developed a technique which can estimate the full covariance structure of a prior in a sensorimotor task and have shown that the prior over visuomotor transformations favor a rotation-like structure. Moreover, through experience of a novel task structure, participants can appropriately alter the covariance structure of their prior.     

Acute Alcohol Consumption Impairs Controlled but Not Automatic Processes in a Psychophysical Pointing Paradigm

Numerous studies have investigated the effects of alcohol consumption on controlled and automatic cognitive processes. Such studies have shown that alcohol impairs performance on tasks requiring conscious, intentional control, while leaving automatic performance relatively intact. Here, we sought to extend these findings to aspects of visuomotor control by investigating the effects of alcohol in a visuomotor pointing paradigm that allowed us to separate the influence of controlled and automatic processes. Six male participants were assigned to an experimental “correction” condition in which they were instructed to point at a visual target as quickly and accurately as possible. On a small percentage of trials, the target “jumped” to a new location. On these trials, the participants’ task was to amend their movement such that they pointed to the new target location. A second group of 6 participants were assigned to a “countermanding” condition, in which they were instructed to terminate their movements upon detection of target “jumps”. In both the correction and countermanding conditions, participants served as their own controls, taking part in alcohol and no-alcohol conditions on separate days. Alcohol had no effect on participants’ ability to correct movements “in flight”, but impaired the ability to withhold such automatic corrections. Our data support the notion that alcohol selectively impairs controlled processes in the visuomotor domain.     

The 50s Cliff: A Decline in Perceptuo-Motor Learning,Not a Deficit in Visual Motion Perception

Previously, we measured perceptuo-motor learning rates across the lifespan and found a sudden drop in learning rates between ages 50 and 60, called the “50s cliff.” The task was a unimanual visual rhythmic coordination task in which participants used a joystick to oscillate one dot in a display in coordination with another dot oscillated by a computer. Participants learned to produce a coordination with a 90° relative phase relation between the dots. Learning rates for participants over 60 were half those of younger participants. Given existing evidence for visual motion perception deficits in people over 60 and the role of visual motion perception in the coordination task, it remained unclear whether the 50s cliff reflected onset of this deficit or a genuine decline in perceptuo-motor learning. The current work addressed this question. Two groups of 12 participants in each of four age ranges (20s, 50s, 60s, 70s) learned to perform a bimanual coordination of 90° relative phase. One group trained with only haptic information and the other group with both haptic and visual information about relative phase. Both groups were tested in both information conditions at baseline and post-test. If the 50s cliff was caused by an age dependent deficit in visual motion perception, then older participants in the visual group should have exhibited less learning than those in the haptic group, which should not exhibit the 50s cliff, and older participants in both groups should have performed less well when tested with visual information. Neither of these expectations was confirmed by the results, so we concluded that the 50s cliff reflects a genuine decline in perceptuo-motor learning with aging, not the onset of a deficit in visual motion perception.     

Credit Assignment during Movement Reinforcement Learning

We often need to learn how to move based on a single performance measure that reflects the overall success of our movements. However, movements have many properties, such as their trajectories, speeds and timing of end-points, thus the brain needs to decide which properties of movements should be improved; it needs to solve the credit assignment problem. Currently, little is known about how humans solve credit assignment problems in the context of reinforcement learning. Here we tested how human participants solve such problems during a trajectory-learning task. Without an explicitly-defined target movement, participants made hand reaches and received monetary rewards as feedback on a trial-by-trial basis. The curvature and direction of the attempted reach trajectories determined the monetary rewards received in a manner that can be manipulated experimentally. Based on the history of action-reward pairs, participants quickly solved the credit assignment problem and learned the implicit payoff function. A Bayesian credit-assignment model with built-in forgetting accurately predicts their trial-by-trial learning.     

Bimanual Coordination Learning with Different Augmented Feedback Modalities and Information Types

Previous studies have shown that bimanual coordination learning is more resistant to the removal of augmented feedback when acquired with auditory than with visual channel. However, it is unclear whether this differential “guidance effect” between feedback modalities is due to enhanced sensorimotor integration via the non-dominant auditory channel or strengthened linkage to kinesthetic information under rhythmic input. The current study aimed to examine how modalities (visual vs. auditory) and information types (continuous visuospatial vs. discrete rhythmic) of concurrent augmented feedback influence bimanual coordination learning. Participants either learned a 90°-out-of-phase pattern for three consecutive days with Lissajous feedback indicating the integrated position of both arms, or with visual or auditory rhythmic feedback reflecting the relative timing of the movement. The results showed diverse performance change after practice when the feedback was removed between Lissajous and the other two rhythmic groups, indicating that the guidance effect may be modulated by the type of information provided during practice. Moreover, significant performance improvement in the dual-task condition where the irregular rhythm counting task was applied as a secondary task also suggested that lower involvement of conscious control may result in better performance in bimanual coordination.     

Differences in Context and Feedback Result in Different Trajectories and Adaptation Strategies in Reaching

Computational models of motor control have often explained the straightness of horizontal planar reaching movements as a consequence of optimal control. Departure from rectilinearity is thus regarded as sub-optimal. Here we examine if subjects may instead select to make curved trajectories following adaptation to force fields and visuomotor rotations. Separate subjects adapted to force fields with or without visual feedback of their hand trajectory and were retested after 24 hours. Following adaptation, comparable accuracies were achieved in two ways: with visual feedback, adapted trajectories in force fields were straight whereas without it, they remained curved. The results suggest that trajectory shape is not always straight, but is also influenced by the calibration of available feedback signals for the state estimation required by the task. In a follow-up experiment, where additional subjects learned a visuomotor rotation immediately after force field, the trajectories learned in force fields (straight or curved) were transferred when directions of the perturbations were similar but not when directions were opposing. This demonstrates a strong bias by prior experience to keep using a recently acquired control policy that continues to produce successful performance inspite of differences in tasks and feedback conditions. On relearning of force fields on the second day, facilitation by intervening visuomotor rotations occurred only when required motor adjustments and calibration of feedback signals were similar in both tasks. These results suggest that both the available feedback signals and prior history of learning influence the choice and maintenance of control policy during adaptations.     

Distributed neural plasticity for shape learning in the human visual cortex

Expertise in recognizing objects in cluttered scenes is a critical skill for our interactions in complex environments and is thought to develop with learning. However, the neural implementation of object learning across stages of visual analysis in the human brain remains largely unknown. Using combined psychophysics and functional magnetic resonance imaging (fMRI), we show a link between shape-specific learning in cluttered scenes and distributed neuronal plasticity in the human visual cortex. We report stronger fMRI responses for trained than untrained shapes across early and higher visual areas when observers learned to detect low-salience shapes in noisy backgrounds. However, training with high-salience pop-out targets resulted in lower fMRI responses for trained than untrained shapes in higher occipitotemporal areas. These findings suggest that learning of camouflaged shapes is mediated by increasing neural sensitivity across visual areas to bolster target segmentation and feature integration. In contrast, learning of prominent pop-out shapes is mediated by associations at higher occipitotemporal areas that support sparser coding of the critical features for target recognition. We propose that the human brain learns novel objects in complex scenes by reorganizing shape processing across visual areas, while taking advantage of natural image correlations that determine the distinctiveness of target shapes.     

Multiple Motor Learning Strategies in Visuomotor Rotation

Background

When exposed to a continuous directional discrepancy between movements of a visible hand cursor and the actual hand (visuomotor rotation), subjects adapt their reaching movements so that the cursor is brought to the target. Abrupt removal of the discrepancy after training induces reaching error in the direction opposite to the original discrepancy, which is called an aftereffect. Previous studies have shown that training with gradually increasing visuomotor rotation results in a larger aftereffect than with a suddenly increasing one. Although the aftereffect difference implies a difference in the learning process, it is still unclear whether the learned visuomotor transformations are qualitatively different between the training conditions.

Methodology/Principal Findings

We examined the qualitative changes in the visuomotor transformation after the learning of the sudden and gradual visuomotor rotations. The learning of the sudden rotation led to a significant increase of the reaction time for arm movement initiation and then the reaching error decreased, indicating that the learning is associated with an increase of computational load in motor preparation (planning). In contrast, the learning of the gradual rotation did not change the reaction time but resulted in an increase of the gain of feedback control, suggesting that the online adjustment of the reaching contributes to the learning of the gradual rotation. When the online cursor feedback was eliminated during the learning of the gradual rotation, the reaction time increased, indicating that additional computations are involved in the learning of the gradual rotation.

Conclusions/Significance

The results suggest that the change in the motor planning and online feedback adjustment of the movement are involved in the learning of the visuomotor rotation. The contributions of those computations to the learning are flexibly modulated according to the visual environment. Such multiple learning strategies would be required for reaching adaptation within a short training period.     

Adaptation to visual feedback delay influences visuomotor learning

Computational theory of motor control suggests that the brain continuously monitors motor commands, to predict their sensory consequences before actual sensory feedback becomes available. Such prediction error is a driving force of motor learning, and therefore appropriate associations between motor commands and delayed sensory feedback signals are crucial. Indeed, artificially introduced delays in visual feedback have been reported to degrade motor learning. However, considering our perceptual ability to causally bind our own actions with sensory feedback, demonstrated by the decrease in the perceived time delay following repeated exposure to an artificial delay, we hypothesized that such perceptual binding might alleviate deficits of motor learning associated with delayed visual feedback. Here, we evaluated this hypothesis by investigating the ability of human participants to adapt their reaching movements in response to a novel visuomotor environment with 3 visual feedback conditions--no-delay, sudden-delay, and adapted-delay. To introduce novelty into the trials, the cursor position, which originally indicated the hand position in baseline trials, was rotated around the starting position. In contrast to the no-delay condition, a 200-ms delay was artificially introduced between the cursor and hand positions during the presence of visual rotation (sudden-delay condition), or before the application of visual rotation (adapted-delay condition). We compared the learning rate (representing how the movement error modifies the movement direction in the subsequent trial) between the 3 conditions. In comparison with the no-delay condition, the learning rate was significantly degraded for the sudden-delay condition. However, this degradation was significantly alleviated by prior exposure to the delay (adapted-delay condition). Our data indicate the importance of appropriate temporal associations between motor commands and sensory feedback in visuomotor learning. Moreover, they suggest that the brain is able to account for such temporal associations in a flexible manner.     

Eye-Hand Coordination during Visuomotor Adaptation with Different Rotation Angles: Effects of Terminal Visual Feedback

This study examined adaptive changes of eye-hand coordination during a visuomotor rotation task under the use of terminal visual feedback. Young adults made reaching movements to targets on a digitizer while looking at targets on a monitor where the rotated feedback (a cursor) of hand movements appeared after each movement. Three rotation angles (30°, 75° and 150°) were examined in three groups in order to vary the task difficulty. The results showed that the 30° group gradually reduced direction errors of reaching with practice and adapted well to the visuomotor rotation. The 75° group made large direction errors of reaching, and the 150° group applied a 180° reversal shift from early practice. The 75°and 150° groups, however, overcompensated the respective rotations at the end of practice. Despite these group differences in adaptive changes of reaching, all groups gradually adapted gaze directions prior to reaching from the target area to the areas related to the final positions of reaching during the course of practice. The adaptive changes of both hand and eye movements in all groups mainly reflected adjustments of movement directions based on explicit knowledge of the applied rotation acquired through practice. Only the 30° group showed small implicit adaptation in both effectors. The results suggest that by adapting gaze directions from the target to the final position of reaching based on explicit knowledge of the visuomotor rotation, the oculomotor system supports the limb-motor system to make precise preplanned adjustments of reaching directions during learning of visuomotor rotation under terminal visual feedback.     

The Effect of Gaze Position on Reaching Movements in an Obstacle Avoidance Task

Numerous studies have addressed the issue of where people look when they perform hand movements. Yet, very little is known about how visuomotor performance is affected by fixation location. Previous studies investigating the accuracy of actions performed in visual periphery have revealed inconsistent results. While movements performed under full visual-feedback (closed-loop) seem to remain surprisingly accurate, open-loop as well as memory-guided movements usually show a distinct bias (i.e. overestimation of target eccentricity) when executed in periphery. In this study, we aimed to investigate whether gaze position affects movements that are performed under full-vision but cannot be corrected based on a direct comparison between the hand and target position. To do so, we employed a classical visuomotor reaching task in which participants were required to move their hand through a gap between two obstacles into a target area. Participants performed the task in four gaze conditions: free-viewing (no restrictions on gaze), central fixation, or fixation on one of the two obstacles. Our findings show that obstacle avoidance behaviour is moderated by fixation position. Specifically, participants tended to select movement paths that veered away from the obstacle fixated indicating that perceptual errors persist in closed-loop vision conditions if they cannot be corrected effectively based on visual feedback. Moreover, measuring the eye-movement in a free-viewing task (Experiment 2), we confirmed that naturally participants’ prefer to move their eyes and hand to the same spatial location.     

Dynamics of Dual Prism Adaptation: Relating Novel Experimental Results to a Minimalistic Neural Model

In everyday life, humans interact with a dynamic environment often requiring rapid adaptation of visual perception and motor control. In particular, new visuo–motor mappings must be learned while old skills have to be kept, such that after adaptation, subjects may be able to quickly change between two different modes of generating movements (‘dual–adaptation’). A fundamental question is how the adaptation schedule determines the acquisition speed of new skills. Given a fixed number of movements in two different environments, will dual–adaptation be faster if switches (‘phase changes’) between the environments occur more frequently? We investigated the dynamics of dual–adaptation under different training schedules in a virtual pointing experiment. Surprisingly, we found that acquisition speed of dual visuo–motor mappings in a pointing task is largely independent of the number of phase changes. Next, we studied the neuronal mechanisms underlying this result and other key phenomena of dual–adaptation by relating model simulations to experimental data. We propose a simple and yet biologically plausible neural model consisting of a spatial mapping from an input layer to a pointing angle which is subjected to a global gain modulation. Adaptation is performed by reinforcement learning on the model parameters. Despite its simplicity, the model provides a unifying account for a broad range of experimental data: It quantitatively reproduced the learning rates in dual–adaptation experiments for both direct effect, i.e. adaptation to prisms, and aftereffect, i.e. behavior after removal of prisms, and their independence on the number of phase changes. Several other phenomena, e.g. initial pointing errors that are far smaller than the induced optical shift, were also captured. Moreover, the underlying mechanisms, a local adaptation of a spatial mapping and a global adaptation of a gain factor, explained asymmetric spatial transfer and generalization of prism adaptation, as observed in other experiments.     

People Efficiently Explore the Solution Space of the Computationally Intractable Traveling Salesman Problem to Find Near-Optimal Tours

Humans need to solve computationally intractable problems such as visual search, categorization, and simultaneous learning and acting, yet an increasing body of evidence suggests that their solutions to instantiations of these problems are near optimal. Computational complexity advances an explanation to this apparent paradox: (1) only a small portion of instances of such problems are actually hard, and (2) successful heuristics exploit structural properties of the typical instance to selectively improve parts that are likely to be sub-optimal. We hypothesize that these two ideas largely account for the good performance of humans on computationally hard problems. We tested part of this hypothesis by studying the solutions of 28 participants to 28 instances of the Euclidean Traveling Salesman Problem (TSP). Participants were provided feedback on the cost of their solutions and were allowed unlimited solution attempts (trials). We found a significant improvement between the first and last trials and that solutions are significantly different from random tours that follow the convex hull and do not have self-crossings. More importantly, we found that participants modified their current better solutions in such a way that edges belonging to the optimal solution (“good” edges) were significantly more likely to stay than other edges (“bad” edges), a hallmark of structural exploitation. We found, however, that more trials harmed the participants'' ability to tell good from bad edges, suggesting that after too many trials the participants “ran out of ideas.” In sum, we provide the first demonstration of significant performance improvement on the TSP under repetition and feedback and evidence that human problem-solving may exploit the structure of hard problems paralleling behavior of state-of-the-art heuristics.     

Egocentric Representation Acquired from Offline Map Learning

It is widely accepted that people establish allocentric spatial representation after learning a map. However, it is unknown whether people can directly acquire egocentric representation after map learning. In two experiments, the participants learned a distal environment through a map and then performed the egocentric pointing tasks in that environment under three conditions: with the heading aligned with the learning perspective (baseline), after 240° rotation from the baseline (updating), and after disorientation (disorientation). Disorientation disrupted the internal consistency of pointing among objects when the participants learned the sequentially displayed map, on which only one object name was displayed at a time while the location of “self” remained on the screen all the time. However, disorientation did not affect the internal consistency of pointing among objects when the participants learned the simultaneously displayed map. These results suggest that the egocentric representation can be acquired from a sequentially presented map.     

BAC-HAPPY Mapping (BAP Mapping): A New and Efficient Protocol for Physical Mapping

Physical and linkage mapping underpin efforts to sequence and characterize the genomes of eukaryotic organisms by providing a skeleton framework for whole genome assembly. Hitherto, linkage and physical “contig” maps were generated independently prior to merging. Here, we develop a new and easy method, BAC HAPPY MAPPING (BAP mapping), that utilizes BAC library pools as a HAPPY mapping panel together with an Mbp-sized DNA panel to integrate the linkage and physical mapping efforts into one pipeline. Using Arabidopsis thaliana as an exemplar, a set of 40 Sequence Tagged Site (STS) markers spanning ∼10% of chromosome 4 were simultaneously assembled onto a BAP map compiled using both a series of BAC pools each comprising 0.7x genome coverage and dilute (0.7x genome) samples of sheared genomic DNA. The resultant BAP map overcomes the need for polymorphic loci to separate genetic loci by recombination and allows physical mapping in segments of suppressed recombination that are difficult to analyze using traditional mapping techniques. Even virtual “BAC-HAPPY-mapping” to convert BAC landing data into BAC linkage contigs is possible.     

Visual Acuity Testing: Feedback Affects Neither Outcome nor Reproducibility,but Leaves Participants Happier

Assessment of visual acuity is a well standardized procedure at least for expert opinions and clinical trials. It is often recommended not giving patients feedback on the correctness of their responses. As this viewpoint has not been quantitatively examined so far, we quantitatively assessed possible effects of feedback on visual acuity testing. In 40 normal participants we presented Landolt Cs in 8 orientations using the automated Freiburg Acuity Test (FrACT, <michaelbach.de/fract. Over a run comprising 24 trials, the acuity threshold was measured with an adaptive staircase procedure. In an ABCDDCBA scheme, trial-by-trial feedback was provided in 2 x 4 conditions: (A) no feedback, (B) acoustic signals indicating correctness, (C)visual indication of correct orientation, and (D) a combination of (B) and (C). After each run the participants judged comfort. Main outcome measures were absolute visual acuity (logMAR), its test-retest agreement (limits of agreement) and participants’ comfort estimates on a 5-step symmetric Likert scale. Feedback influenced acuity outcome significantly (p = 0.02), but with a tiny effect size: 0.02 logMAR poorer acuity for (D) compared to (A), even weaker effects for (B) and (C). Test-retest agreement was high (limits of agreement: ± 1.0 lines) and did not depend on feedback (p>0.5). The comfort ranking clearly differed, by 2 steps on the Likert scale: the condition (A)–no feedback–was on average “slightly uncomfortable”, the other three conditions were “slightly comfortable” (p<0.0001). Feedback affected neither reproducibility nor the acuity outcome to any relevant extent. The participants, however, reported markedly greater comfort with any kind of feedback. We conclude that systematic feedback (as implemented in FrACT) offers nothing but advantages for routine use.     

Slowness: An Objective for Spike-Timing–Dependent Plasticity?

Our nervous system can efficiently recognize objects in spite of changes in contextual variables such as perspective or lighting conditions. Several lines of research have proposed that this ability for invariant recognition is learned by exploiting the fact that object identities typically vary more slowly in time than contextual variables or noise. Here, we study the question of how this “temporal stability” or “slowness” approach can be implemented within the limits of biologically realistic spike-based learning rules. We first show that slow feature analysis, an algorithm that is based on slowness, can be implemented in linear continuous model neurons by means of a modified Hebbian learning rule. This approach provides a link to the trace rule, which is another implementation of slowness learning. Then, we show analytically that for linear Poisson neurons, slowness learning can be implemented by spike-timing–dependent plasticity (STDP) with a specific learning window. By studying the learning dynamics of STDP, we show that for functional interpretations of STDP, it is not the learning window alone that is relevant but rather the convolution of the learning window with the postsynaptic potential. We then derive STDP learning windows that implement slow feature analysis and the “trace rule.” The resulting learning windows are compatible with physiological data both in shape and timescale. Moreover, our analysis shows that the learning window can be split into two functionally different components that are sensitive to reversible and irreversible aspects of the input statistics, respectively. The theory indicates that irreversible input statistics are not in favor of stable weight distributions but may generate oscillatory weight dynamics. Our analysis offers a novel interpretation for the functional role of STDP in physiological neurons.     

Dual-Force ISOMAP: A New Relevance Feedback Method for Medical Image Retrieval

With great potential for assisting radiological image interpretation and decision making, content-based image retrieval in the medical domain has become a hot topic in recent years. Many methods to enhance the performance of content-based medical image retrieval have been proposed, among which the relevance feedback (RF) scheme is one of the most promising. Given user feedback information, RF algorithms interactively learn a user’s preferences to bridge the “semantic gap” between low-level computerized visual features and high-level human semantic perception and thus improve retrieval performance. However, most existing RF algorithms perform in the original high-dimensional feature space and ignore the manifold structure of the low-level visual features of images. In this paper, we propose a new method, termed dual-force ISOMAP (DFISOMAP), for content-based medical image retrieval. Under the assumption that medical images lie on a low-dimensional manifold embedded in a high-dimensional ambient space, DFISOMAP operates in the following three stages. First, the geometric structure of positive examples in the learned low-dimensional embedding is preserved according to the isometric feature mapping (ISOMAP) criterion. To precisely model the geometric structure, a reconstruction error constraint is also added. Second, the average distance between positive and negative examples is maximized to separate them; this margin maximization acts as a force that pushes negative examples far away from positive examples. Finally, the similarity propagation technique is utilized to provide negative examples with another force that will pull them back into the negative sample set. We evaluate the proposed method on a subset of the IRMA medical image dataset with a RF-based medical image retrieval framework. Experimental results show that DFISOMAP outperforms popular approaches for content-based medical image retrieval in terms of accuracy and stability.     

Reinforcement Learning of Targeted Movement in a Spiking Neuronal Model of Motor Cortex

Sensorimotor control has traditionally been considered from a control theory perspective, without relation to neurobiology. In contrast, here we utilized a spiking-neuron model of motor cortex and trained it to perform a simple movement task, which consisted of rotating a single-joint “forearm” to a target. Learning was based on a reinforcement mechanism analogous to that of the dopamine system. This provided a global reward or punishment signal in response to decreasing or increasing distance from hand to target, respectively. Output was partially driven by Poisson motor babbling, creating stochastic movements that could then be shaped by learning. The virtual forearm consisted of a single segment rotated around an elbow joint, controlled by flexor and extensor muscles. The model consisted of 144 excitatory and 64 inhibitory event-based neurons, each with AMPA, NMDA, and GABA synapses. Proprioceptive cell input to this model encoded the 2 muscle lengths. Plasticity was only enabled in feedforward connections between input and output excitatory units, using spike-timing-dependent eligibility traces for synaptic credit or blame assignment. Learning resulted from a global 3-valued signal: reward (+1), no learning (0), or punishment (−1), corresponding to phasic increases, lack of change, or phasic decreases of dopaminergic cell firing, respectively. Successful learning only occurred when both reward and punishment were enabled. In this case, 5 target angles were learned successfully within 180 s of simulation time, with a median error of 8 degrees. Motor babbling allowed exploratory learning, but decreased the stability of the learned behavior, since the hand continued moving after reaching the target. Our model demonstrated that a global reinforcement signal, coupled with eligibility traces for synaptic plasticity, can train a spiking sensorimotor network to perform goal-directed motor behavior.