bioRxiv Subject Collection: Neuroscience's Journal

Impaired Visuospatial Working Memory but Preserved Attentional Control in Bipolar Disorder
BACKGROUNDPersistent deficits in working memory (WM) and attention have considerable clinical and functional impact in people with bipolar disorder (PBD). Understanding the neurocognitive underpinnings of these interacting cognitive constructs might facilitate the discovery of more effective pro-cognitive interventions. Therefore, we employed a paradigm designed for jointly studying attentional control and WM encoding.

METHODSWe used a visuospatial change-detection task using four Gabor Patches with differing orientations in 63 euthymic PBD and 76 healthy controls (HCS), which investigated attentional competition during WM encoding. To manipulate bottom-up attention using stimulus salience, two Gabor patches flickered, which were designated as either targets or distractors. To manipulate top-down attention, the Gabor patches were preceded by either a predictive or a non-predictive cue for the target locations.

RESULTSAcross all task conditions, PBD stored significantly less information in visual WM than HCS (significant effect of group). However, we observed no significant group by salience or group by cue interactions. This indicates that impaired WM was not caused by deficits in attentional control.

CONCLUSIONSOur results imply that while WM is disturbed in PBD, attentional prioritization of salient targets and distractors as well as the utilization of external top-down cues were not compromised. Consequently, the control of attentional selection appears be intact. These findings provide important constraints for models of WM dysfunction in PBD by indicating that later stages of WM encoding are likely primarily impaired. We also demonstrate that selective attention is not among the main sources of cognitive impairment in PBD.

The neurodevelopmental trajectory of beta band oscillations: an OPM-MEG study
Neural oscillations mediate coordination of activity within and between brain networks, supporting cognition and behaviour. How these processes develop throughout childhood is not only a critical neuroscientific question but could also shed light on the mechanisms underlying neurological and psychiatric disorders. However, measuring the neurodevelopmental trajectory of oscillations has been hampered by confounds from instrumentation. In this paper, we investigate the suitability of a disruptive new imaging platform - Optically Pumped Magnetometer-based magnetoencephalography (OPM-MEG) - to study oscillations during brain development. We show how a unique 192-channel OPM-MEG device, which is adaptable to head size and robust to participant movement, can be used to collect high-fidelity electrophysiological data in individuals aged between 2 and 34 years. Data were collected during a somatosensory task, and we measured both stimulus-induced modulation of beta oscillations in sensory cortex, and whole-brain connectivity, showing that both modulate significantly with age. Moreover, we show that pan-spectral bursts of electrophysiological activity drive beta oscillations throughout neurodevelopment, and how their probability of occurrence and spectral content changes with age. Our results offer new insights into the developmental trajectory of oscillations and provide the first clear evidence that OPM-MEG is an ideal platform for studying electrophysiology in children.

Age differences in functional connectivity track dedifferentiation of category representations
With advancing age, the distinctiveness of neural representations of information declines. While the finding of this so-called age-related neural dedifferentiation in category-selective neural regions is well-described, the contribution of age-related changes in network organization to dedifferentiation is unknown. Here, we asked whether age differences in a) functional connectivity to category-selective neural regions and b) segregation of the visual network (i.e., network dedifferentiation) contribute to regional dedifferentiation of categorical representations. Younger and older adults viewed blocks of face and house stimuli in the fMRI scanner. We found an age-related decline in neural distinctiveness for faces in the fusiform gyrus (FG) and for houses in the parahippocampal gyrus (PHG). Functional connectivity analyses revealed age differences in connectivity between the FG and visual network as well as age-related dedifferentiation of the visual network. Interindividual correlations demonstrated that regional distinctiveness was related to connectivity of category-selective regions to the visual network as well as network segregation. Thus, dedifferentiation of categorical representations may be linked to age-related reorganization of functional networks.

Diabetes alters neuroeconomically dissociable forms of mental accounting
Those with diabetes mellitus are at high-risk of developing psychiatric disorders, yet the link between hyperglycemia and alterations in motivated behavior has not been explored in detail. We characterized value-based decision-making behavior of a streptozocin-induced diabetic mouse model on a naturalistic neuroeconomic foraging paradigm called Restaurant Row. Mice made self-paced choices while on a limited time-budget accepting or rejecting reward offers as a function of cost (delays cued by tone-pitch) and subjective value (flavors), tested daily in a closed-economy system across months. We found streptozocin-treated mice disproportionately undervalued less-preferred flavors and inverted their meal-consumption patterns shifted toward a more costly strategy that overprioritized high-value rewards. We discovered these foraging behaviors were driven by impairments in multiple decision-making systems, including the ability to deliberate when engaged in conflict and cache the value of the passage of time in the form of sunk costs. Surprisingly, diabetes-induced changes in behavior depended not only on the type of choice being made but also the salience of reward-scarcity in the environment. These findings suggest complex relationships between glycemic regulation and dissociable valuation algorithms underlying unique cognitive heuristics and sensitivity to opportunity costs can disrupt fundamentally distinct computational processes and could give rise to psychiatric vulnerabilities.

Genome binding properties of Zic transcription factors underlie their changing functions during neuronal maturation
BackgroundThe Zic family of transcription factors (TFs) promote both proliferation and maturation of cerebellar granule neurons (CGNs), raising the question of how a single, constitutively expressed TF family can support distinct developmental processes. Here we use an integrative experimental and bioinformatic approach to discover the regulatory relationship between Zic TF binding and changing programs of gene transcription during CGN differentiation.

ResultsWe first established a bioinformatic pipeline to integrate Zic ChIP-seq data from the developing mouse cerebellum with other genomic datasets from the same tissue. In newborn CGNs, Zic TF binding predominates at active enhancers that are co-bound by developmentally-regulated TFs including Atoh1, whereas in mature CGNs, Zic TF binding consolidates toward promoters where it co-localizes with activity-regulated TFs. We then performed CUT&RUN-seq in differentiating CGNs to define both the time course of developmental shifts in Zic TF binding and their relationship to gene expression. Mapping Zic TF binding sites to genes using chromatin looping, we identified the set of Zic target genes that have altered expression in RNA-seq from Zic1 or Zic2 knockdown CGNs.

ConclusionOur data show that Zic TFs are required for both induction and repression of distinct, developmentally regulated target genes through a mechanism that is largely independent of changes in Zic TF binding. We suggest that the differential collaboration of Zic TFs with other TF families underlies the shift in their biological functions across CGN development.

Altered Socio-Affective Communication and Amygdala Development in mice with Protocadherin10-deficient Interneurons
Autism Spectrum Disorder (ASD) is a group of neurodevelopmental conditions associated with deficits in social interaction and communication, together with repetitive behaviors. The cell adhesion molecule Protocadherin10 (Pcdh10) has been implicated in the etiology of ASD. Pcdh10 is expressed in the nervous system during embryonic and early postnatal development and has been linked to neural circuit formation. Here, we show strong expression of Pcdh10 in the ganglionic eminences and in the basolateral complex of the amygdala at mid and late embryonic stages, respectively. Both inhibitory and excitatory neurons expressed Pcdh10 in the basolateral complex at perinatal stages and genes linked to vocalization behavior were enriched in Pcdh10- expressing neurons in adult mice. To further investigate the involvement of Pcdh10 in neurodevelopment with relevance to ASD, and to assess the functional and behavioral consequences of loss of Pcdh10 in basolateral amygdala interneurons, we combined a ubiquitous and a conditional Pcdh10 knockout mouse model. Conditional knockout of Pcdh10 reduced the number of interneurons in the basolateral complex. Both models exhibited altered developmental trajectories of socio-affective communication through isolation-induced ultrasonic vocalizations in neonatal pups, characterized by increased emission rates in heterozygous pups. Furthermore, acoustic call features were affected and heterozygous conditional knockout pups emitted calls characterized by reduced peak frequencies but increased frequency modulation. Additionally, we identified distinct clusters of call subtypes with specific developmental trajectories, suggesting the vocalization repertoire is extensive and dynamic during early life. The nuanced alterations in socio-affective communication at the level of call emission rates, acoustic call features, and clustering of call subtypes were primarily seen in heterozygous pups of the conditional knockout and less prominent in the ubiquitous Pcdh10 knockout, suggesting that changes in anxiety levels associated with Gsh2-lineage interneurons might drive the observed behavioral effects. Together, this demonstrates that loss of Pcdh10 specifically in interneurons contributes to behavioral alterations in socio-affective communication with relevance to ASD.

Shaping the physical world to our ends: The left PF technical-cognition area
Our propensity to materiality, which consists in using, making, creating, and passing on technologies, has enabled us to shape the physical world according to our ends. To explain this proclivity, scientists have calibrated their lens to either low-level skills such as motor cognition or high-level skills such as language or social cognition. Yet, little has been said about the intermediate-level cognitive processes that are directly involved in mastering this materiality. We aim to focus on this intermediate level for contributing to building a cognitive framework of human technology. Here we show that a technical-reasoning process might be specifically at work in physical problem-solving situations. We found via two distinct neuroimaging studies that the area PF (parietal F) within the left parietal lobe is central for this reasoning process in both tool-use and non-tool-use physical problem-solving and can work along with social-cognitive skills to resolve day-to-day interactions that combine social and physical constraints. Our results demonstrate the existence of a specific cognitive module in the human brain dedicated to materiality, which might be the supporting pillar allowing the accumulation of technical knowledge over generations. Intensifying research on technical cognition could nurture a comprehensive framework that has been missing in fields interested in how early and modern humans have been interacting with the physical world through technology, and how this interaction has shaped our history and culture.

Plural molecular and cellular mechanisms of pore domain KCNQ2 encephalopathy
KCNQ2 variants in children with neurodevelopmental impairment are difficult to assess due their heterogeneity and unclear pathogenic mechanisms. We describe a child with neonatal-onset epilepsy, developmental impairment of intermediate severity, and KCNQ2 G256W heterozygosity. Analyzing prior KCNQ2 channel cryoelectron microscopy models revealed G256 as keystone of an arch-shaped non-covalent bond network linking S5, the pore turret, and the ion path. Co-expression with G256W dominantly suppressed conduction by wild-type subunits in heterologous cells. Ezogabine partly reversed this suppression. G256W/+ mice have epilepsy leading to premature deaths. Recorded in G256W/+ mouse brain slices, CA1 pyramidal cells show increased neuronal excitability. Hippocampal pyramidal cell KCNQ2 and KCNQ3 immunolabeling is significantly shifted from axon initial segments to neuronal somata. Despite normal mRNA levels, G256W/+ mouse neocortical KCNQ2 protein levels are reduced by about 50%. Our findings indicate that G256W pathogenicity results from multiplicative effects, including reductions in intrinsic conduction, subcellular targeting, and protein stability. These studies reveal pore "turret arch" bonding as a KCNQ structural novelty and introduce a valid animal model of KCNQ2 encephalopathy. Our results, spanning structure to behavior, may be broadly informative since the majority of KCNQ2 encephalopathy patients share variants near the selectivity filter.

Insights into Human Epileptogenesis with Proteomic Profiling
Epilepsy affects millions globally, and drug-resistant epilepsy remains a challenge. Molecular mechanisms underlying epilepsy remain elusive. Protein profiling through proteomics offers insight into biomarkers and therapeutic targets. Human brain tissue from epilepsy surgeries was analyzed using data-independent acquisition (DIA) proteomics. Samples were categorized into Core (epileptogenic focus), Border (marginal excision tissue), and Nonepileptic control groups. Differential expression proteins (DEPs) were identified and shared proteins were analyzed. 163 DEPs were identified which may has potential roles in the initiation of epileptic electrical firing 412 DEPs which indicating the difference between epilepsy and Nonepilepsy patients and 10 DEPs consistently altered in Core which indicating potential roles in epileptogenesis. Notably, P35754/GLRX, O75335/PPFIA4, and Q96KP4/CNDP2 were consistently expressed differently in all group pairs. From validation experiments, the expression of Kv3.2 significant reduced in the Core group compare to border group by immunohistochemistry and knockdown of Kv3.2 increased seizure susceptibility and altered neuronal excitability through our cellular and animal experimentation.

Microglia promote extracellular matrix deposition and restrict excitatory synapse numbers in the mesolimbic dopamine system during healthy aging
Synapse dysfunction has been definitively linked to cognitive impairments in the aging brain, and microglial physiology has emerged as a robust regulator of synapse status and cognitive aging outcomes. Hippocampal microglia have recently been shown to regulate synapse function via targeted remodeling of the extracellular matrix (ECM), yet the degree to which microglia-ECM interactions impact synapse function in the healthy aged brain remains virtually unexplored. This study combines high-resolution imaging and ECM-optimized tissue proteomics to examine the impact that microglial physiology has on ECM and synapse status in the basal ganglia of healthy aging mice. Our results demonstrate that deposition of the ubiquitous ECM scaffold hyaluronan increases during aging in the ventral tegmental area (VTA), but not its downstream target, the nucleus accumbens, and that VTA microglial tissue coverage correlates with local hyaluronan deposition. Proteomic mapping of core matrisome proteins showed prominent regional differences in ECM composition across basal ganglia nuclei that were significantly associated with abundance of chemokine receptors and synapse proteins. Finally, manipulation of microglial fractalkine signaling through Cx3Cr1 receptor deletion reversed age-associated ECM accumulation within the VTA and resulted in abnormally elevated synapse numbers in this brain region by middle age. These findings indicate that microglia promote age-related increases in ECM deposition in some, but not all, brain regions that may restrict local excitatory synapse numbers. This microglial function could represent an adaptive response to brain aging that helps to maintain appropriate activity patterns within basal ganglia circuits.

Assessing White Matter Engagement in Brain Networks through Functional and Structural Connectivity Mapping
Understanding the intricate interplay between gray matter (GM) and white matter (WM) is crucial for deciphering the complex activities of the brain. While diffusion tensor imaging (DTI) has advanced the mapping of these structural pathways, the relationship between structural connectivity (SC) and functional connectivity (FC) remains inadequately understood. This study addresses the need for a more integrative approach by mapping the importance of the inter-GM functional link to its structural counterparts in WM. This mapping yields a spatial distribution of engagement that is not only highly reproducible but also aligns with direct structural, functional, and bioenergetic measures within WM, illustrating a notable interdependence between the function of GM and the characteristics of WM. Additionally, our research has uncovered a set of unique engagement modes through a clustering analysis of window-wise engagement maps, highlighting the dyanmic nature of the engagement. The engagement along with their temporal variations revealed significant differences across genders and age groups. These findings suggest the potential of WM engagement as a biomarker for neurological and cognitive conditions, offering a more nuanced understanding of individualized brain activity and connectivity patterns.

Multi-parametric assays capture sex- and environment-dependent modifiers of behavioral phenotypes in autism mouse models
Current phenotyping approaches for murine autism models often focus on one selected behavioral feature, making the translation onto a spectrum of autistic characteristics in humans challenging. Furthermore, sex and environmental factors are rarely considered. Here, we aimed to capture the full spectrum of behavioral manifestations in three autism mouse models to develop a "behavioral fingerprint" that takes environmental and sex influences under consideration. To this end, we employed a wide range of classical standardized behavioral tests; and two multi-parametric behavioral assays: the Live Mouse Tracker and Motion Sequencing (MoSeq), on male and female Shank2, Tsc1 and Purkinje cell specific-Tsc1 mutant mice raised in standard or enriched environments. Our aim was to integrate our high dimensional data into one single platform to classify differences in all experimental groups along dimensions with maximum discriminative power. Multi-parametric behavioral assays enabled far more accurate classification of experimental groups compared to classical tests, and dimensionality reduction analysis demonstrated significant additional gains in classification accuracy, highlighting the presence of sex, environmental and genotype differences in our experimental groups. Together, our results provide a complete phenotypic description of all tested groups, suggesting multi-parametric assays can capture the entire spectrum of the heterogenous phenotype in autism mouse models.

An ecological investigation of the effect of background noise on speech processing in a Virtual Classroom
Many real-life situations can be extremely noisy. Psychoacoustic studies have shown that background noise can have a detrimental effect on the ability to process and understand speech. However, most studies use stimuli and task designs that are highly artificial, limiting their generalization to more realistic contexts. Moreover, to date, we do not fully understand the neurophysiological consequences of trying to pay attention to speech in a noisy place. To address this lab to real-life gap and increase the ecological validity of speech in noise research, here we introduce a novel audiovisual Virtual Reality (VR) experimental platform. Combined with neurophysiological measurements of neural activity (EEG), eye-gaze and skin conductance (GSR) we studied the effects of background noise in a realistic context where the ability to process and understand continuous speech is especially important: A VR Classroom. Participants (n=32) sat in a VR Classroom and were told to pay attention to mini-lecture segments by a virtual teacher. Trials were either Quiet or contained background construction noise, emitted from outside the classroom window, which was either Continuous (drilling) or Intermittent (air hammers). Result show that background noise had a detrimental effect on learning outcomes, which was also accompanied by reduced neural tracking of the teacher's speech. Comparison of the two noise types showed that the intermittent construction noise was more disruptive than continuous noise, as index by both behavioral and neural measures, and it also elicited higher skin-conductance levels, reflecting heightened arousal. Interesting, eye-gaze dynamics were not affected by the presence of noise. This study advances our understanding of the neurophysiological effects of background noise and extends it to more ecologically relevant contexts. It also emphasizes the role that temporal dynamics play for processing speech in noise, highlighting the need to consider the features of realistic noises, as we expand speech in noise research to increasingly realistic circumstances.

Efficient value encoding through convergence of tactile and visual value information in the primate putamen
The processing of diverse sensory values by a limited number of basal ganglia neurons raises the question of whether each value is processed independently or combined as the number of neurons decreases from the cortex to downstream structures. Here, we discovered that tactile and visual values were partially converged in the primate putamen, enhancing its efficiency in encoding values while preserving modality information. Humans and monkeys performed tactile and visual value discrimination tasks. Notably, the human putamen selectively represented both tactile and visual values in fMRI scans. Single-unit electrophysiology further revealed that half of the individual neurons in the macaque putamen encoded both tactile and visual values, and the other half encoded each value separately. The bimodal value neurons enable more efficient value encoding using fewer neurons than the modality-selective value neurons. Our data suggest that the basal ganglia system uses modality convergence to efficiently encode values with limited resources.

Small non-coding RNA content in plasma-derived extracellular vesicles distinguish ataxic SCA3 mutation carriers from pre-ataxic and control subjects
Spinocerebellar ataxia type 3 (SCA3), a neurodegenerative disorder caused by a CAG expansion in the ATXN3 gene, is the most common spinocerebellar ataxia subtype worldwide. Currently, there is no therapy to stop or prevent disease progression. Promising therapeutic strategies are emerging, but their translation into clinical practice requires sensitive and reliable biomarkers. Blood circulating extracellular vesicles constitute a promising source of biomarkers with potential to track alterations of the central nervous system due to their ability to cross the blood brain barrier. Here, we perform sequencing analysis of small RNAs from plasma-derived extracellular vesicles from SCA3 mutation carriers (10 pre-ataxic and 10 ataxic) and 12 control subjects to identify potential RNA biomarker candidates for this disease. Data showed that plasma-derived extracellular vesicles from ataxic SCA3 mutation carriers are enriched in mitochondrial, nuclear, and nucleolar RNA biotypes compared to pre-ataxic and control subjects. Moreover, ataxic mutation carriers could be discriminated from control and pre-ataxic subjects based on the miRNAs or piRNAs content, but not tRNA. Furthermore, we identified a subset of differentially expressed miRNAs and piRNAs that clearly differentiate ataxic mutation carriers from pre-ataxic and control subjects. These findings open new avenues for further investigation on the role of these RNAs in the pathogenesis of SCA3 and their potential as biomarkers for this disease.

Deep Learning Analysis on Images of iPSC-derived Motor Neurons Carrying fALS-genetics Reveals Disease-Relevant Phenotypes
Amyotrophic lateral sclerosis (ALS) is a devastating condition with very limited treatment options. It is a heterogeneous disease with complex genetics and unclear etiology, making the discovery of disease-modifying interventions very challenging. To discover novel mechanisms underlying ALS, we leverage a unique platform that combines isogenic, induced pluripotent stem cell (iPSC)-derived models of disease-causing mutations with rich phenotyping via high-content imaging and deep learning models. We introduced eight mutations that cause familial ALS (fALS) into multiple donor iPSC lines, and differentiated them into motor neurons to create multiple isogenic pairs of healthy (wild-type) and sick (mutant) motor neurons. We collected extensive high-content imaging data and used machine learning (ML) to process the images, segment the cells, and learn phenotypes. Self-supervised ML was used to create a concise embedding that captured significant, ALS-relevant biological information in these images. We demonstrate that ML models trained on core cell morphology alone can accurately predict TDP-43 mislocalization, a known phenotypic feature related to ALS. In addition, we were able to impute RNA expression from these image embeddings, in a way that elucidates molecular differences between mutants and wild-type cells. Finally, predictors leveraging these embeddings are able to distinguish between mutant and wild-type both within and across donors, defining cellular, ML-derived disease models for diverse fALS mutations. These disease models are the foundation for a novel screening approach to discover disease-modifying targets for familial ALS.

Pathogen infection induces sickness behaviors by recruiting neuromodulatory systems linked to stress and satiety in C. elegans
When animals are infected by a pathogen, peripheral sensors of infection signal to the brain to coordinate a set of adaptive behavioral changes known as sickness behaviors. While the pathways that signal from the periphery to the brain have been intensively studied in recent years, how central circuits are reconfigured to elicit sickness behaviors is not well understood. Here we find that neuromodulatory systems linked to stress and satiety are recruited upon infection to drive sickness behaviors in C. elegans. Upon chronic infection by the bacterium Pseudomonas aeruginosa PA14, C. elegans decrease their feeding behavior, then display reversible bouts of quiescence, and eventually die. The ALA neuron and its neuropeptides FLP-7, FLP-24, and NLP-8, which control stress-induced sleep in uninfected animals, promote the PA14-induced feeding reduction. However, the ALA neuropeptide FLP-13 instead acts to delay quiescence and death in infected animals. Cell-specific genetic perturbations show that the neurons that release FLP-13 to delay quiescence in infected animals are distinct from ALA. A brain-wide imaging screen reveals that infection-induced quiescence involves ASI and DAF-7/TGF-beta, which control satiety-induced quiescence in uninfected animals. Our results suggest that a common set of neuromodulators are recruited across different physiological states, acting from distinct neural sources and in distinct combinations to drive state-dependent behaviors.

Distribution of calbindin-positive neurons across areas and layers of the marmoset cerebral cortex
The calcium-binding protein calbindin is selectively expressed in specific neuronal populations of the cerebral cortex, including major classes of inhibitory interneurons. We have charted the distribution of calbindin-positive (CB+) neurons across areas and layers of the entire marmoset cortex using a combination of immunohistochemistry, AI-based image segmentation, 3-dimensional reconstruction, and cytoarchitecture-aware registration. CB+ neurons formed 10-20% of the cortical neuronal population, occurring in higher proportions in areas corresponding to low hierarchical levels of processing, such as sensory cortices. Although CB+ neurons concentrated in the supragranular layers, there were clear trends in laminar distribution: the relative density in infragranular layers increased with hierarchical level, and the density in layer 4 was lowest in areas involved in sensorimotor integration and action planning. These results reveal new aspects of the cytoarchitectural organization of the primate cortex and demonstrate an efficient approach to mapping the full distribution of neurochemically distinct cell types throughout the brain, readily applicable to most mammalian species and parts of the nervous system.

Neural representations of predicted events: Evidence from time-resolved EEG decoding
Through statistical learning, humans are able to extract temporal regularities, using the past to predict the future. Evidence suggests that learning relational structures makes it possible to anticipate the imminent future; yet, the neural dynamics of predicting the future and its time-course remain elusive. To examine whether future representations are denoted in a temporally discounted fashion, we used the high-temporal-resolution of electroencephalography (EEG). Observers were exposed to a fixed sequence of events at four unique spatial positions within the display. Using multivariate pattern analyses trained on independent pattern estimators, we were able to decode the spatial position of dots within full sequences, and within randomly intermixed partial sequences wherein only a single dot was presented. Crucially, within these partial sequences, subsequent spatial positions could be reliably decoded at their expected moment in time. These findings highlight the dynamic weight changes within the assumed spatial priority map and mark the first implementation of EEG to decode predicted, yet critically omitted events.

Impact statementUtilizing high-temporal-resolution EEG, the dynamic weight changes of assumed spatial priority map were visualized by decoding the spatial position of expected, yet omitted, events at their expected moment in time.

The volitional control of individual motor units is constrained within low-dimensional manifolds by common inputs
The implementation of low-dimensional movement control by the central nervous system has been debated for decades. In this study, we investigated the dimensionality of control signals received by spinal motor neurons in controlling one degree of freedom of either the ankle or knee joint. We hypothesized that the central nervous system would mainly adopt a rigid control of motor units; specifically, that the motor units active during these tasks would belong to a small number of synergies, each receiving common descending inputs. This hypothesis was tested using torque-matched isometric contractions, as well as with an operant-conditioning paradigm, where the firing activities of pairs of motor units were provided as visual feedback to the participants. The motor units of the gastrocnemius lateralis could be controlled largely independently from those of the gastrocnemius medialis during ankle plantarflexion. This dissociation of motor unit activity imposed similar behavior to the motor units that were not displayed in the feedback, leading to a two-dimensional control manifold, where each dimension represented a "synergistic" muscle. It was not possible to independently control the motor units within the gastrocnemius medialis muscle. During knee extension tasks, it was not possible to dissociate the activity of the motor units between the vastus lateralis and medialis muscles, which thus belonged to a one-dimensional manifold. Overall, individual motor units were never controlled independently of all others but rather belonged to synergistic groups. These results provide evidence for a synergistic low-dimensional control of motor units constrained by common inputs spanning one or more muscles.