Journal of Neurophysiology

Patients with cerebellar damage experience various motor impairments, but the specific sequence of primary and compensatory processes that contribute to these deficits remains unclear. To clarify this, we reversibly blocked cerebellar outflow in monkeys engaged in planar reaching tasks. This intervention led to a spatially selective reduction in hand velocity, primarily due to decreased muscle torque, especially in movements requiring high inter-joint torque coupling. When examining repeated reaches to the same target, we found that the reduced velocity resulted from both an immediate deficit and a gradually developing compensatory slowing to reduce passive inter-joint interactions. However, the slowed hand velocity did not account for the fragmented and variable movement trajectories observed during the cerebellar block. Our findings indicate that cerebellar impairment results in motor deficits due to both inadequate muscle torque and an altered motor control strategy for managing impaired limb dynamics. Additionally, impaired motor control elevates noise, which cannot be entirely mitigated through compensatory strategies.

Motor adaptation is essential for maintaining coordination and precision in daily activities. Implicit motor adaptation--adaptation that occurs without conscious awareness--is thought to be primarily driven by sensory prediction errors. Here, we investigated how rapidly these unconscious changes in reaching behavior emerge as a function of error magnitude and the availability of task error signals. To this end, we employed a single-trial learning (STL) paradigm within a classical visuomotor rotation task. Participants made center-out reaching movements to either small (dot) or large (arc) targets while experiencing single perturbation trials with cursor rotations ranging from 1{degrees} to 90{degrees}, each followed by an aligned washout trial. By manipulating target size, we systematically modulated the presence of task error while holding sensory prediction error constant. We further compared these early implicit changes with those observed during standard prolonged adaptation to a fixed 20{degrees} rotation across >100 trials. Our results show that implicit adaptation emerges rapidly, even after a single exposure to small perturbations, and follows a saturating, fixed-rate response profile. Importantly, the magnitude of single-trial adaptation was greater when task error was present (small targets) compared with conditions in which only sensory prediction error was available (large targets). Moreover, STL-derived parameters moderately predicted the initial phase of adaptation during prolonged learning, suggesting that STL captures core dynamics of early implicit processes. These findings provide new insight into the mechanistic principles governing implicit motor adaptation. By identifying the parameters that drive early-stage error-based learning, this work refines current models of sensorimotor learning and highlights potential strategies for designing targeted training or rehabilitation protocols that leverage rapid adaptation processes to enhance motor performance and recovery.

Attention plays a crucial role in maintaining precision and effectiveness in goal-directed actions. Although there is evidence that dividing attention across tasks impairs performance in various domains, the impact of attention on sensorimotor adaptation remains inconclusive, with some studies reporting deficits and others showing no effects. Because sensorimotor adaptation arises from the interaction of explicit and implicit processes, this discrepancy may reflect differential effects of attention on each process. Here, we investigate how divided attention influences implicit sensorimotor adaptation using an error-clamp paradigm, coupled with a random dot kinematogram (RDK) motion coherence discrimination task. We also assessed whether the timing of the secondary task affects error processing during sensorimotor adaptation by presenting the RDK either during the outward movement (coinciding with error feedback), or the inward movement (following error feedback). We observed that attentional manipulation influenced implicit sensorimotor adaptation only when the RDK was presented on the outward movement, not the inward movement. Remarkably, implicit sensorimotor adaptation was enhanced when attention was divided, compared to when attention was focused entirely on the adaptation task. This suggests that implicit sensorimotor adaptation is sensitive to attentional demand, particularly during the time window where error feedback is received.

During deliberation, as the brain considers its options, the neural activity representing the goodness of each option rises toward a threshold, and the choice is often dictated by the option for which the rise is fastest. Here we report a surprising correlate of these activities: saccade vigor. We engaged human subjects in a decision-making task in which they considered effortful options, each requiring walking various durations and inclines. As they deliberated, they made saccades between the symbolic representations of those options. These saccades had no bearing on the effort that they would later expend, yet as they deliberated, saccade velocities increased. The rate of rise in vigor was faster for saccades toward the option that they later indicated as their choice, and encoded the difference in the subjective value of the two effortful options. Once deliberation ended, following a brief delay the subjects indicated their choice by making another saccade. Remarkably, vigor for this saccade dropped to baseline and no longer encoded subjective value. These results are consistent with an urgency model of decision-making in which a global signal in the brain drives both the neural circuits that make decisions, and the neural circuits that make movements. Paradoxically, this common drive is shared between the oculomotor circuits and the decision-making circuits, even when the decision involves effortful expenditure during a future event. SignificanceThere is a link between the decisions we make and the movements that follow. Not only do we prefer options of greater value, but we also move faster to acquire them. When deliberating between options, neural activity rises to a threshold and the option that wins this race is the one chosen. We report a potential correlate of this in the motor control circuits; during deliberation, saccade vigor to both options rise, but faster for the option ultimately chosen. Thus, our movements appear to mirror the neural activity conducting the decision-making process. Paradoxically, this is true even when the movements have no direct bearing on the decision at hand.

Sensorimotor adaptation is essential for keeping our movements well-calibrated in response to changes in the body and environment. For over a century, we have studied sensorimotor adaptation in highly controlled laboratory settings that typically involve small sample sizes. While this approach has proven useful to characterize different learning processes, laboratory studies are typically very underpowered to generate data suited for exploring the myriad of factors that may modulate motor performance. Here, using a citizen science website (testmybrain.org), we collected over 2000 sessions on a visuomotor rotation task. This unique dataset has allowed us to replicate classic motor findings, reconcile controversial findings in the learning and memory literature, and discover novel constraints underlying dissociable implicit and explicit learning processes supporting sensorimotor adaptation. Taken together, this study suggests that large-scale motor learning studies hold enormous potential to advance sensorimotor neuroscience.

Visuomotor adaptation typically involves an interplay between implicit and explicit processes. While explicit strategy development has often been characterized as an exponential function, recent work has shown that individual participants usually show different time courses. Here we identify 3 styles of explicit strategy development and test how these rely on rotations ranging from 20{degrees} to 60{degrees}. Participants self-reported their planned reach direction, allowing us to record a trial-by-trial strategy development time course. We used machine learning to determine the start and stabilization of strategy development. We then use descriptive statistics of this phase to cluster participants into stepwise, gradual, and exploratory strategy learning styles. First, larger rotations increased the proportion of participants who spontaneously developed a strategy. Crucially, the proportions of strategy learning styles also varied as a function of perturbation size; larger rotations (50{degrees}-60{degrees}) favored exploratory and stepwise strategies, whereas smaller rotations (20{degrees}-30{degrees}) predominantly yielded gradual learning, with no exploratory behaviour observed in the 20{degrees} group. These findings challenge the notion of explicit adaptation as a homogeneous process. They also suggest that rotation size may boost non-gradual strategy formation.

Human motor adaptation relies on both explicit conscious strategies and implicit unconscious updating of internal models to correct motor errors. Implicit adaptation is powerful, requiring less preparation time before executing adapted movements, but recent work suggests it is limited to some absolute magnitude regardless of the size of a visuomotor perturbation when the perturbation is introduced abruptly. It is commonly assumed that gradually introducing a perturbation should lead to improved implicit learning beyond this limit, but outcomes are conflicting. We tested whether introducing a perturbation in two distinct gradual methods can overcome the apparent limit and explain past conflicting findings. We found that gradually introducing a perturbation in a stepped manner, where participants were given time to adapt to each partial step before being introduced to a larger partial step, led to [~]80% higher implicit aftereffects of learning, but introducing it in a ramped manner, where participants adapted larger rotations on each subsequent reach, did not. Our results clearly show that gradual introduction of a perturbation can lead to substantially larger implicit adaptation, as well as identify the type of introduction that is necessary to do so.

The role of sensory feedback is well established in current models of motor control, evidenced by deficits in movement coordination resulting from impaired sensory function. When vision and touch are both available for object-oriented manual behaviors, these senses can be distinctly leveraged; vision guides movement planning while touch provides feedback on hand-object interactions. How eye-hand coordination changes with the loss of somatosensory feedback has not been well studied. Conceivably, vision is recruited to compensate for the feedback lost when touch is abolished. We tested healthy participants on a manual dexterity task, consisting of moving small metal pegs. The task was performed before and after administration of digital anesthesia that abolished cutaneous sensations while preserving motor function with the acting hand. During peg collection, transport, and placement epochs, we tracked gaze direction and hand positions while also recording forces applied to the pegboard. We hypothesized that the nervous system selectively adapts eye-hand coordination according to the dexterity demands of the task epochs. We found that participants maintained the ability to perform the pegboard task following the loss of cutaneous feedback, albeit with longer trial times and altered force profiles. Notably, somatosensory loss was accompanied by a shift in visual behavior marked by a closer alignment between gaze and hand positions during all task epochs, even those that did not involve object manipulation. Together, these data affirm the contributions of sensory feedback to force control in service of dexterous object manipulation and reveal the non-selective nature of compensatory eye-hand coordination processes. Significance StatementTouch and vision typically support distinct, but coordinated aspects of dexterous manual behaviors. Here, we evaluated how acute removal of tactile feedback using digital anesthesia affected performance and eye-hand coordination in a manual dexterity task. With insensate fingers and intact vision, participants continued to perform the task successfully, albeit with longer trial times and altered force profiles. We also observed closer alignment between gaze and hand positions during all task epochs, even those that did not involve object manipulation. Our results reveal the consequences of acute somatosensory loss and the general nature of compensatory eye-hand coordination processes.

The nervous system has a tremendous ability to modify motoneuron excitability according to task demands through neuromodulatory synaptic input to motoneurons. Neuromodulatory inputs adjust the response of the motoneuron to excitatory and inhibitory ionotropic input and can facilitate the induction of persistent inward currents (PICs). PICs amplify and prolong the motoneuron response to synaptic inputs, and PIC impairment may play a major role in motor deficits observed in pathological conditions. Noninvasive estimation of the magnitude of neuromodulatory input and persistent inward currents in human motoneurons is achieved through a paired motor unit analysis ({Delta}F) that quantifies hysteresis in the firing rates at motor unit recruitment and derecruitment. While the {Delta}F technique is commonly used for estimating motoneuron excitability, computational parameters used for the technique vary across studies. In the present study, we assessed the sensitivity of the {Delta}F technique to several criteria commonly used in selecting motor unit pairs for analysis, as well as to methods used for smoothing the instantaneous motor unit firing rates. Using HD-sEMG and motor unit decomposition we obtained 5,409 motor unit pairs from the triceps brachii of ten healthy individuals during submaximal triangle contractions. The mean (SD) {Delta}F was 4.9 (1.08) pps, consistent with previous work using intramuscular recordings. There was an exponential plateau relationship between {Delta}F and the recruitment time difference between the motor unit pairs, with the plateau occurring at approximately 1 s. There was an exponential decay relationship between {Delta}F and the derecruitment time difference between the motor unit pairs, with the decay stabilizing at approximately 1.5 s. We found that reducing or removing the minimum threshold for the correlation of the rate-rate slope for the two units did not affect {Delta}F values or variance. Additionally, we found that removing motor unit pairs in which the control unit was saturated had no significant effect on {Delta}F. Smoothing filter selection had no substantial effect on {Delta}F values and {Delta}F variance; however, the length and type of smoothing filter affected the minimum recruitment and derecruitment time differences. Our results facilitate interpretation of findings from studies that implement the {Delta}F approach but use different computational parameters.

Motor learning involves plasticity in a network of brain areas across the cortex and cerebellum. Such traces of learning have the potential to affect subsequent learning of other tasks. In some cases, prior learning can interfere with subsequent learning, but it may be possible to potentiate learning of one task with a prior task if they are sufficiently different. Because prism adaptation involves extensive neuroplasticity, we reasoned that the elevated excitability of neurons could increase their readiness to undergo structural changes, and in turn, create an optimal state for learning a subsequent task. We tested this idea, selecting two different forms of learning tasks, asking whether exposure to a sensorimotor adaptation task can improve subsequent de novo motor skill learning. Participants first learned a new visuomotor mapping induced by prism glasses in which prism strength varied trial-to-trial. Immediately after and the next day, we tested participants on a mirror tracing task, a form of de novo skill learning. Prism-trained and control participants both learned the mirror tracing task, with similar reductions in error and increases in distance traced. Both groups also showed evidence of offline performance gains between the end of day 1 and the start of day 2. However, we did not detect differences between groups. Overall, our results do not support the idea that prism adaptation learning can potentiate subsequent de novo learning. We discuss factors that may have contributed to this result.

During goal-directed movements, the magnitude of error correction by a person on a subsequent movement provides important insight into a persons motor learning dynamics. Observed differences in trial-by-trial adaptation rates might indicate different relative weighting placed on the various sources of information that inform a movement, e.g. sensory feedback, control predictions, or internal model expectations. Measuring this trial-by-trial adaptation rate is not straightforward, however, since externally observed data are masked by noise from several sources and influenced by inaccessible internal processes. Adaptation to perturbation has been used to measure error adaptation as the introduced external disturbance is sufficiently large to overshadow other noise sources. However, perturbation analysis is difficult to implement in real-world scenarios, requires a large number of movement trials to accommodate infrequent perturbations, and the paradigm itself might affect the movement dynamics being observed. Here we focus on error adaptation during unperturbed and naturalistic movements. With increasing motor noise, the conventional estimation of trial-by-trial adaptation increases, a counterintuitive finding that is the consequence of systematic bias in the estimate due to noise masking the learners intention. We present an analytic solution relying on stochastic signal processing to reduce this effect of noise, producing an estimate of motor adaptation with reduced bias. The result is an improved estimate of trial-by-trial adaptation in a human learner compared to conventional methods. We demonstrate the effectiveness of the new method in analyzing simulated and empirical movement data under different noise conditions. The analytic approach is applicable across different types of movements in varied contexts and should replace the regression analysis method in future motor analysis studies. Author SummaryWhen a person makes a movement, a motor error is typically observed that then drives motor planning corrections on subsequent movements. This error correction provides insight into how the nervous system is operating, particularly in regard to how much confidence a person places in different sources of information such as sensory feedback or motor command reproducibility. Traditional analysis of movement has required carefully controlled laboratory conditions, limiting the usefulness of motor analysis in clinical and everyday environments. Here we present a new computational method that can be accurately applied to typical movements. Counterintuitive findings of the established approach are corrected by the proposed method. This method will provide a common framework for researchers to analyze movements while extending dynamic motor adaptation analysis capabilities to clinical and non-laboratory settings.

Generating pure elbow rotation requires contracting muscles at both the shoulder and elbow joints to counter torques that arise at the shoulder when the forearm rotates (i.e., intersegmental dynamics). Previous work has shown that human participants learn to reduce their shoulder muscle activity if the same elbow movement is performed after the shoulder joint is mechanically locked, which is appropriate because locking the shoulder joint eliminates the torques that arise at the shoulder when the forearm rotates. However, this learning is slow (i.e., it unfolds over hundreds of trials) and incomplete (i.e., shoulder activity is not fully eliminated). Here we investigated whether and how the addition of explicit strategies and biofeedback modulate this type of learning. Three groups of human participants (N = 55) performed voluntary pure elbow rotations using a robotic exoskeleton that permits shoulder and elbow rotation in a horizontal plane. Participants did the task with the shoulder free to move (baseline), then with the shoulder joint locked by the robotic manipulandum (adaptation), and then with the shoulder free to move again (post-adaptation). The first group of participants performed this protocol and received no instructions about what to do after their shoulder was locked. The second group of participants received visual feedback about their shoulder muscle activity after each movement and was instructed to reduce their shoulder activity to zero. The third group of participants also received visual biofeedback, but it was removed part way through the experiment. We found that, although all groups learned, the rate and magnitude of learning was not reliably different across the groups. Taken together, our results suggest that learning new arm dynamics, unlike other motor learning paradigms, unfolds independent of explicit instructions, biofeedback and task instructions.

When a movement error occurs, the motor system updates its commands to improve performance on subsequent trials. A prominent feedback error learning hypothesis proposes that the feedback response that corrects movement within a trial serves as a teaching signal for the learning response, observed as changes in motor commands on the next trial. However, how the temporal pattern of the feedback response influences the learning response remains unclear. Here, we introduce an experimental paradigm that directly compares the temporal patterns of feedback and learning responses across different patterns of visual error. We show that although the feedback response closely tracks the temporal pattern of the visual error, this temporal pattern is not transferred to the learning response. Instead, the amplitude of the feedback response during the holding period, reflecting the temporal pattern of the visual error, primarily determines the magnitude of the learning response. These findings reveal how the motor learning system selectively extracts specific components of feedback responses to update future motor commands.

Accumulating evidence indicates that the humans proprioception map appears subject-specific. However, whether the idiosyncratic pattern persists across time with good within-subject consistency has not been quantitatively examined. Here we measured the proprioception by a hand visual-matching task in multiple sessions over two days. We found that people improved their proprioception when tested repetitively without performance feedback. Importantly, despite the reduction of average error, the spatial pattern of proprioception errors remained idiosyncratic. Based on individuals proprioceptive performance, a standard convolutional neural network classifier could identify people with good accuracy. We also found that subjects baseline proprioceptive performance could not predict their motor performance in a visual trajectory-matching task even though both tasks require accurate mapping of hand position to visual targets in the same workspace. Using a separate experiment, we not only replicated these findings but also ruled out the possibility that performance feedback during a few familiarization trials caused the observed improvement in proprioception. We conclude that the conventional proprioception test itself, even without feedback, can improve proprioception but leave the idiosyncrasy of proprioception unchanged.

Integrated multisensory feedback plays a crucial role in balance control. Minimal fingertip contact with a surface (light-touch), reduces center of pressure (CoP) by adding sensory information about postural orientation and balance state. Electrical vestibular stimulation (EVS) can increase sway by adding erroneous vestibular cues. This juxtaposition of conflicting sensory cues can be exploited to explore the dynamics of sensorimotor reweighting. We used continuous stochastic EVS (0-25Hz; {+/-}4mA; 200-300s) to evoke balance responses in CoP (Exp-1, Exp-2) and segment accelerations (Exp-2). Systems analyses (coherence, gain) quantified coupling and size of balance responses to EVS. We had participants either touch (TOUCH; <2N) or not touch (NO-TOUCH) a load cell during EVS (Exp-1, Exp-2), or we intermittently removed the touch surface (Exp-2) to measure the effects of light touch on vestibular-evoked balance responses. We hypothesized that coherence and gain between EVS and CoP would decrease, consistent with the CNS down-weighting vestibular cues that conflict with light touch. Light touch reduced CoP displacement, but increased variation in the CoP signal explained by EVS input. Significant coherence between EVS and CoP was observed up to [~]30Hz in both conditions but was significantly greater in the TOUCH condition from 12-28.5-Hz. Conversely, EVS-CoP gain was 63% lower in TOUCH, compared to NO-TOUCH. Our findings show that light touch can re-weight vestibular-evoked responses by reducing their size but also increasing high frequency vestibular contributions for sway. This suggests that the CNS can use novel sensory inputs to alter balance behavior but cannot completely ignore a salient balance cue. New and NoteworthyThis study reveals that minimal fingertip contact (light touch) during balance tasks not only diminishes the impact of electrical vestibular stimulation (EVS) on sway, but also enhances the central nervous systems ability to integrate high-frequency vestibular cues. Specifically, light touch decreases the magnitude of EVS-induced sway while increasing coherence with EVS at higher frequencies, illustrating the central nervous systems capacity to adaptively reweight sensory inputs for improved balance control without fully disregarding dominant cues.

Precision reaching tasks often require corrective submovements for successful completion. Most studies of reaching have focused on single initial movements, and the cortical encoding model was implied to be the same for all submovements. However, corrective submovements may show different encoding patterns from the initial submovement with distinct patterns of activation across the population. Two rhesus macaques performed a precision center-out-task with small targets. Neural activity from single units in primary motor cortex and associated behavioral data were recorded to evaluate movement characteristics. Neural population data and individual neuronal firing rates identified with a peak finding algorithm to identify peaks in hand speed were examined for encoding differences between initial and corrective submovements. Individual neurons were fitted with a regression model that included the reach vector, position, and speed to predict firing rate. For both initial and corrective submovements, the largest effect remained movement direction. We observed a large subset changed their preferred direction greater than 45{degrees} between initial and corrective submovements. Neuronal depth of modulation also showed considerable variation when adjusted for movement speed. By utilizing principal component analysis, neural trajectories of initial and corrective submovements progressed through different neural subspaces. These findings all suggest that different neural encoding patterns exist for initial and corrective submovements within the cortex. We hypothesize that this variation in how neurons change to encode small, corrective submovements might allow for a larger portion of the neural space being used to encode a greater range of movements with varying amplitudes and levels of precision. New and NoteworthyNeuronal recordings matched with kinematic behavior were collected in a precision center-out task that often required corrective movements. We reveal large differences in preferred direction and depth of modulation between initial and corrective submovements across the neural population. We then present a model of the neural population describing how these shifts in tuning create different subspaces for signaling initial and corrective movements likely to improve motor precision.

Responding to an external stimulus takes [~]200 ms, but this can be shortened to as little as [~]120 ms with the additional presentation of a startling acoustic stimulus. This phenomenon is hypothesized to arise from the involuntary release of a prepared movement (a StartReact effect). However, a startling acoustic stimulus also expedites rapid mid-flight, reactive adjustments to unpredictably displaced targets which could not have been prepared in advance. We surmise that for such rapid visuomotor transformations, intersensory facilitation may occur between auditory signals arising from the startling acoustic stimulus and visual signals relayed along a fast subcortical network. To explore this, we examined how a startling acoustic stimulus shortens reaction times in a task that produces express visuomotor responses, which are brief bursts of muscle activity that arise from a fast tectoreticulospinal network. We measured express visuomotor responses on upper limb muscles in humans as they reached either toward or away from a stimulus in blocks of trials where movements could either be fully prepared or not, occasionally pairing stimulus presentation with a startling acoustic stimulus. The startling acoustic stimulus reliably produced larger but fixed-latency express visuomotor responses in a target-selective manner, and also shortened reaction times, which were equally short for prepared and unprepared movements. Our results provide insights into how a startling acoustic stimulus shortens the latency of reactive movements without full motor preparation. We propose that the reticular formation is the likely node for intersensory convergence during the most rapid transformations of vision into targeted reaching actions. KEY POINTSO_LIA startling acoustic stimulus (SAS) shortens reaction times by releasing fully prepared motor programs (the StartReact effect), but can also hasten responses in reflexive tasks without any movement preparation C_LIO_LIHere we measure the effect of a SAS on reaction times and upper limb muscle recruitment in a reflexive reaching task, focusing on express visuomotor responses that are evoked by visual target presentation and demarcate activity along a subcortical tectoreticulospinal pathway C_LIO_LIA SAS robustly increased the magnitude of express visuomotor responses without changing their timing, and this increase was tightly related to the subsequent reaction time even in the absence of motor preparation C_LIO_LIOur results attest to intersensory facilitation within the tectoreticulospinal pathway, which provides the shortest pathway mediating visuomotor transformations for reaching C_LIO_LIThese results reconcile discrepant findings by emphasizing the importance of intersensory facilitation in SAS-induced hastening of reaction times in reflexive tasks C_LI

How do precision of movement and proprioception influence motor control and adaptation? Several theories--such as the exploration-exploitation hypothesis--propose that variability plays a key role in motor performance and learning. However, empirical measures of motor and proprioceptive precision are often limited by small sample sizes, and proprioceptive estimates, especially those relying on efferent signals, are difficult to isolate and quantify. In this study, we leveraged a large dataset of 270 participants--including a subsample of older adults (ages 54- 84)--to assess the precision of hand movements and proprioceptive estimates, and to examine whether these factors predict individual differences in motor learning and adaptation. We found that baseline reach variance did not predict learning or changes in hand localization. Although active hand localization (which includes efferent contributions) was slightly more precise--showing an 8.6% reduction in variance--this suggests that unseen hand estimates rely primarily on proprioception. Neither motor nor sensory precision varied with age. However, reach aftereffects were modestly associated with proprioceptive precision before training and proprioceptive recalibration after training. No other measure of learning or variance was reliably associated. These findings suggest that reach aftereffects may partly reflect changes in hand proprioception, but overall, we identified no predictors of adaptation to a rotated visual cursor.

The ability to coordinate movements between the hands is crucial for many daily tasks. However, the precise mechanisms governing the storage and utilization of bimanual movement and the distinct contributions of each limb in this process are currently not fully understood. Two key questions persist: 1) How is the neural representation of bimanual coordination stored in the brain, and 2) How is the information governing bimanual coordination shared between hemispheres? In this investigation, we used a virtual partner (VP) to systematically address these issues by allowing the same coordination pattern (CP) to be acquired with unimanual and bimanual movements. More specifically, we used four experimental groups: unimanual (left, right) VP, bimanual, and control conditions. For each condition, retention and transfer tests were administered immediately and 6 hours after the initial practice. The control condition employed the same protocol as unimanual conditions without practice. As anticipated, performance after practice and during retention sessions indicated that all groups learned to perform the target CP. Furthermore, generalization from unimanual to bimanual occurred when the same type of visual feedback (VF) was provided. Interestingly, the absence of VF impaired motor generalization from unimanual to bimanual condition unless the participants initially practiced the task bimanually. Taken together, our results demonstrated that both limbs could access the memory representation of the CP. However, this globally shared representation appeared to be encoded in the visual-spatial domain. The conditions without VF underscored the importance of proprioception in forming a motor representation in intrinsic coordinates. NEW & NOTEWORTHYConventional views on acquiring bimanual skills stress the need for simultaneous engagement of both hands. However, our study challenges this notion by demonstrating that the coordination pattern learned in unimanual conditions significantly boosts subsequent bimanual coordination--a novel approach to skill acquisition. Yet, this advantage diminishes without visual feedback, resulting in a breakdown of the intended bimanual coordination, highlighting the limitations of relying solely on unimanual practice.

The ventral lateral (VL) nucleus of the thalamus relays signals from the cerebellum (Cb) and basal ganglia (BG) to primary motor cortex (M1). In primates, glutamatergic Cb efferents from the deep cerebellar nuclei and GABAergic BG efferents from the internal segment of the globus pallidus (GPi) terminate in distinct subregions of VL: the posterior (VLp) and anterior (VLa) divisions, respectively. This anatomical segregation suggests that Cb- and BG-thalamocortical circuits may play distinct roles in motor control, which could be revealed by comparing movement-related activity in VLp and VLa. Here, we recorded single-unit activity from VLp and VLa, identified via electrical stimulation of superior cerebellar peduncle and GPi, during a choice reaction time reaching task. We also recorded from M1, which maintains bidirectional connections with both VLp and VLa. VLa neurons exhibited a significantly higher proportion of decrease-type responses compared with VLp and M1, consistent with inhibitory GPi input. Time-resolved general linear model analysis showed dynamic encoding of task parameters, particularly movement direction, in all three regions. Direction encoding was strongest in M1, moderate in VLp, and weakest in VLa. Direction encoding in VLa also lagged behind that in M1 and VLp. Clustering analysis of direction encoding strength and timing revealed a subpopulation of VLp neurons that encoded direction particularly strongly during the reaction time period. These results highlight a limitation of traditional assumptions that activity characteristics are distributed homogeneously across neural populations and point to a novel functional organization within VLp neurons.