The manifold structure of limb coordination in walking Drosophila

Inhibitory control of locomotor statistics in walking Drosophila

In order to forage for food, many animals regulate not only specific limb movements but the statistics of locomotor behavior over time, switching between long-range dispersal and localized search depending on resource availability. How pre-motor circuits regulate such locomotor statistics is not clear. Here we analyze and model locomotor statistics in walking Drosophila , and their modulation by attractive food odor. Odor evokes three motor regimes in flies: baseline walking, upwind running during odor, and search behavior following odor loss. During search behavior, we find that flies adopt higher angular velocities and slower ground speeds, and tend to turn for longer periods of time in one direction. We further find that flies spontaneously adopt periods of different mean ground speed, and that these changes in state influence the length of odor-evoked runs. We next developed a simple model of neural locomotor control that suggests that contralateral inhibition plays a key role in regulating the statistical features of locomotion. As the fly connectome predicts decussating inhibitory neurons in the lateral accessory lobe (LAL), a pre-motor structure, we gained genetic access to a subset of these neurons and tested their effects on behavior. We identified one population of neurons whose activation induces all three signature of search and that bi-directionally regulates angular velocity at odor offset. We identified a second group of neurons, including a single LAL neuron pair, that bi-directionally regulate ground speed. Together, our work develops a biologically plausible computational architecture that captures the statistical features of fly locomotion across behavioral states and identifies potential neural substrates of these computations.

Natural variability and individuality of walking behavior in Drosophila

Insects use walking behavior in a large number of contexts, such as exploration, foraging, escape and pursuit, or migration. A lot is known about how nervous systems produce this behavior in general and also how certain parameters vary with regard to walking direction or speed, for instance. An aspect that has not received much attention is whether and how walking behavior varies across individuals of a particular species. To address this, we created a large corpus of kinematic walking data of many individuals of the fruit fly Drosophila. We only selected instances of straight walking in a narrow range of walking speeds to minimize the influence of high-level parameters, such as turning and walking speed, aiming to uncover more subtle aspects of variability. Using high-speed videography and automated annotation, we captured the positions of the six leg tips for thousands of steps and used principal components analysis to characterize the postural space individuals used during walking. Our analysis shows that the largest part of walking kinematics can be described by five principal components (PCs). Separation of these five PCs into a 2D and a 3D subspace divided the description of walking behavior into invariant features shared across individuals and features that relate to the specifics of individuals; the latter features can be regarded as idiosyncrasies. We also demonstrate that this approach can detect the effects of experimental interventions in an unbiased manner and that general aspects of individuality, such as the individual walking posture, can be described.

Fine-grained descending control of steering in walking Drosophila

Locomotion involves rhythmic limb movement patterns that originate in circuits outside the brain. Purposeful locomotion requires descending commands from the brain, but we do not understand how these commands are structured. Here, we investigate this issue, focusing on the control of steering in walking Drosophila. First, we describe different limb "gestures" associated with different steering maneuvers. Next, we identify a set of descending neurons whose activity predicts steering. Focusing on two descending cell types downstream of distinct brain networks, we show that they evoke specific limb gestures: one lengthens strides on the outside of a turn, while the other attenuates strides on the inside of a turn. Our results suggest that a single descending neuron can have opposite effects during different locomotor rhythm phases, and we identify networks positioned to implement this phase-specific gating. Together, our results show how purposeful locomotion emerges from specific, coordinated modulations of low-level patterns.

Miniature linear and split-belt treadmills reveal mechanisms of adaptive motor control in walking Drosophila

To navigate complex environments, walking animals must detect and overcome unexpected perturbations. One technical challenge when investigating adaptive locomotion is measuring behavioral responses to precise perturbations during naturalistic walking; another is that manipulating neural activity in sensorimotor circuits often reduces spontaneous locomotion. To overcome these obstacles, we introduce miniature treadmill systems for coercing locomotion and tracking 3D kinematics of walking Drosophila. By systematically comparing walking in three experimental setups, we show that flies compelled to walk on the linear treadmill have similar stepping kinematics to freely walking flies, while kinematics of tethered walking flies are subtly different. Genetically silencing mechanosensory neurons altered step kinematics of flies walking on the linear treadmill across all speeds. We also discovered that flies can maintain a forward heading on a split-belt treadmill by specifically adapting the step distance of their middle legs. These findings suggest that proprioceptive feedback contributes to leg motor control irrespective of walking speed and that the fly's middle legs play a specialized role in stabilizing locomotion.

Kinematics and coordination of moth flies walking on smooth and rough surfaces

The moth fly, Clogmia albipunctata, is a common synanthropic insect with a worldwide range that lives in nearly any area with moist, decaying organic matter. These habitats comprise both smooth, slippery substrates (e.g., bathroom drains) and heterogeneous, bumpy ground (e.g., soil in plant pots). By using terrain of varying levels of roughness, we focus specifically on how substrate roughness at the approximate size scale of the organism affects kinematics and coordination in adult moth flies. Finally, we compare and contrast our characterizations of locomotion in C. albipunctata with previous work of insect walking in naturalistic environments.

Keypoint-MoSeq: parsing behavior by linking point tracking to pose dynamics

Keypoint tracking algorithms can flexibly quantify animal movement from videos obtained in a wide variety of settings. However, it remains unclear how to parse continuous keypoint data into discrete actions. This challenge is particularly acute because keypoint data are susceptible to high-frequency jitter that clustering algorithms can mistake for transitions between actions. Here we present keypoint-MoSeq, a machine learning-based platform for identifying behavioral modules ('syllables') from keypoint data without human supervision. Keypoint-MoSeq uses a generative model to distinguish keypoint noise from behavior, enabling it to identify syllables whose boundaries correspond to natural sub-second discontinuities in pose dynamics. Keypoint-MoSeq outperforms commonly used alternative clustering methods at identifying these transitions, at capturing correlations between neural activity and behavior and at classifying either solitary or social behaviors in accordance with human annotations. Keypoint-MoSeq also works in multiple species and generalizes beyond the syllable timescale, identifying fast sniff-aligned movements in mice and a spectrum of oscillatory behaviors in fruit flies. Keypoint-MoSeq, therefore, renders accessible the modular structure of behavior through standard video recordings.

Sensorimotor delays constrain robust locomotion in a 3D kinematic model of fly walking

Walking animals must maintain stability in the presence of external perturbations, despite significant temporal delays in neural signaling and muscle actuation. Here, we develop a 3D kinematic model with a layered control architecture to investigate how sensorimotor delays constrain robustness of walking behavior in the fruit fly, Drosophila. Motivated by the anatomical architecture of insect locomotor control circuits, our model consists of three component layers: a neural network that generates realistic 3D joint kinematics for each leg, an optimal controller that executes the joint kinematics while accounting for delays, and an inter-leg coordinator. The model generates realistic simulated walking that matches real fly walking kinematics and sustains walking even when subjected to unexpected perturbations, generalizing beyond its training data. However, we found that the model's robustness to perturbations deteriorates when sensorimotor delay parameters exceed the physiological range. These results suggest that fly sensorimotor control circuits operate close to the temporal limit at which they can detect and respond to external perturbations. More broadly, we show how a modular, layered model architecture can be used to investigate physiological constraints on animal behavior.

Learning agile soccer skills for a bipedal robot with deep reinforcement learning

We investigated whether deep reinforcement learning (deep RL) is able to synthesize sophisticated and safe movement skills for a low-cost, miniature humanoid robot that can be composed into complex behavioral strategies. We used deep RL to train a humanoid robot to play a simplified one-versus-one soccer game. The resulting agent exhibits robust and dynamic movement skills, such as rapid fall recovery, walking, turning, and kicking, and it transitions between them in a smooth and efficient manner. It also learned to anticipate ball movements and block opponent shots. The agent's tactical behavior adapts to specific game contexts in a way that would be impractical to manually design. Our agent was trained in simulation and transferred to real robots zero-shot. A combination of sufficiently high-frequency control, targeted dynamics randomization, and perturbations during training enabled good-quality transfer. In experiments, the agent walked 181% faster, turned 302% faster, took 63% less time to get up, and kicked a ball 34% faster than a scripted baseline.

Mapping the neural dynamics of locomotion across the Drosophila brain

Locomotion engages widely distributed networks of neurons. However, our understanding of the spatial architecture and temporal dynamics of the networks that underpin walking remains incomplete. We use volumetric two-photon imaging to map neural activity associated with walking across the entire brain of Drosophila. We define spatially clustered neural signals selectively associated with changes in either forward or angular velocity, demonstrating that neurons with similar behavioral selectivity are clustered. These signals reveal distinct topographic maps in diverse brain regions involved in navigation, memory, sensory processing, and motor control, as well as regions not previously linked to locomotion. We identify temporal trajectories of neural activity that sweep across these maps, including signals that anticipate future movement, representing the sequential engagement of clusters with different behavioral specificities. Finally, we register these maps to a connectome and identify neural networks that we propose underlie the observed signals, setting a foundation for subsequent circuit dissection. Overall, our work suggests a spatiotemporal framework for the emergence and execution of complex walking maneuvers and links this brain-wide neural activity to single neurons and local circuits.

Neural mechanisms to incorporate visual counterevidence in self-movement estimation

In selecting appropriate behaviors, animals should weigh sensory evidence both for and against specific beliefs about the world. For instance, animals measure optic flow to estimate and control their own rotation. However, existing models of flow detection can be spuriously triggered by visual motion created by objects moving in the world. Here, we show that stationary patterns on the retina, which constitute evidence against observer rotation, suppress inappropriate stabilizing rotational behavior in the fruit fly Drosophila. In silico experiments show that artificial neural networks (ANNs) that are optimized to distinguish observer movement from external object motion similarly detect stationarity and incorporate negative evidence. Employing neural measurements and genetic manipulations, we identified components of the circuitry for stationary pattern detection, which runs parallel to the fly's local motion and optic-flow detectors. Our results show how the fly brain incorporates negative evidence to improve heading stability, exemplifying how a compact brain exploits geometrical constraints of the visual world.

Fine-grained descending control of steering in walking Drosophila

Locomotion involves rhythmic limb movement patterns that originate in circuits outside the brain. Purposeful locomotion requires descending commands from the brain, but we do not understand how these commands are structured. Here we investigate this issue, focusing on the control of steering in walking Drosophila. First, we describe different limb "gestures" associated with different steering maneuvers. Next, we identify a set of descending neurons whose activity predicts steering. Focusing on two descending cell types downstream from distinct brain networks, we show that they evoke specific limb gestures: one lengthens strides on the outside of a turn, while the other attenuates strides on the inside of a turn. Notably, a single descending neuron can have opposite effects during different locomotor rhythm phases, and we identify networks positioned to implement this phase-specific gating. Together, our results show how purposeful locomotion emerges from brain cells that drive specific, coordinated modulations of low-level patterns.

Direct comparison reveals algorithmic similarities in fly and mouse visual motion detection

Evolution has equipped vertebrates and invertebrates with neural circuits that selectively encode visual motion. While similarities in the computations performed by these circuits in mouse and fruit fly have been noted, direct experimental comparisons have been lacking. Because molecular mechanisms and neuronal morphology in the two species are distinct, we directly compared motion encoding in these two species at the algorithmic level, using matched stimuli and focusing on a pair of analogous neurons, the mouse ON starburst amacrine cell (ON SAC) and Drosophila T4 neurons. We find that the cells share similar spatiotemporal receptive field structures, sensitivity to spatiotemporal correlations, and tuning to sinusoidal drifting gratings, but differ in their responses to apparent motion stimuli. Both neuron types showed a response to summed sinusoids that deviates from models for motion processing in these cells, underscoring the similarities in their processing and identifying response features that remain to be explained.

Multilevel visuomotor control of locomotion in Drosophila

Vision is critical for the control of locomotion, but the underlying neural mechanisms by which visuomotor circuits contribute to the movement of the body through space are yet not well understood. Locomotion engages multiple control systems, forming distinct interacting "control levels" driven by the activity of distributed and overlapping circuits. Therefore, a comprehensive understanding of the mechanisms underlying locomotion control requires the consideration of all control levels and their necessary coordination. Due to their small size and the wide availability of experimental tools, Drosophila has become an important model system to study this coordination. Traditionally, insect locomotion has been divided into studying either the biomechanics and local control of limbs, or navigation and course control. However, recent developments in tracking techniques, and physiological and genetic tools in Drosophila have prompted researchers to examine multilevel control coordination in flight and walking.

Long-timescale anti-directional rotation in Drosophila optomotor behavior

Locomotor movements cause visual images to be displaced across the eye, a retinal slip that is counteracted by stabilizing reflexes in many animals. In insects, optomotor turning causes the animal to turn in the direction of rotating visual stimuli, thereby reducing retinal slip and stabilizing trajectories through the world. This behavior has formed the basis for extensive dissections of motion vision. Here, we report that under certain stimulus conditions, two Drosophila species, including the widely studied Drosophila melanogaster, can suppress and even reverse the optomotor turning response over several seconds. Such 'anti-directional turning' is most strongly evoked by long-lasting, high-contrast, slow-moving visual stimuli that are distinct from those that promote syn-directional optomotor turning. Anti-directional turning, like the syn-directional optomotor response, requires the local motion detecting neurons T4 and T5. A subset of lobula plate tangential cells, CH cells, show involvement in these responses. Imaging from a variety of direction-selective cells in the lobula plate shows no evidence of dynamics that match the behavior, suggesting that the observed inversion in turning direction emerges downstream of the lobula plate. Further, anti-directional turning declines with age and exposure to light. These results show that Drosophila optomotor turning behaviors contain rich, stimulus-dependent dynamics that are inconsistent with simple reflexive stabilization responses.

Synchronous multi-segmental activity between metachronal waves controls locomotion speed in Drosophila larvae

The ability to adjust the speed of locomotion is essential for survival. In limbed animals, the frequency of locomotion is modulated primarily by changing the duration of the stance phase. The underlying neural mechanisms of this selective modulation remain an open question. Here, we report a neural circuit controlling a similarly selective adjustment of locomotion frequency in Drosophila larvae. Drosophila larvae crawl using peristaltic waves of muscle contractions. We find that larvae adjust the frequency of locomotion mostly by varying the time between consecutive contraction waves, reminiscent of limbed locomotion. A specific set of muscles, the lateral transverse (LT) muscles, co-contract in all segments during this phase, the duration of which sets the duration of the interwave phase. We identify two types of GABAergic interneurons in the LT neural network, premotor neuron A26f and its presynaptic partner A31c, which exhibit segmentally synchronized activity and control locomotor frequency by setting the amplitude and duration of LT muscle contractions. Altogether, our results reveal an inhibitory central circuit that sets the frequency of locomotion by controlling the duration of the period in between peristaltic waves. Further analysis of the descending inputs onto this circuit will help understand the higher control of this selective modulation.

Flexible locomotion in complex environments: the influence of species, speed and sensory feedback on panarthropod inter-leg coordination

Panarthropods (a clade containing arthropods, tardigrades and onychophorans) can adeptly move across a wide range of challenging terrains and their ability to do so given their relatively simple nervous systems makes them compelling study organisms. Studies of forward walking on flat terrain excitingly point to key features in inter-leg coordination patterns that seem to be 'universally' shared across panarthropods. However, when movement through more complex, naturalistic terrain is considered, variability in coordination patterns - from the intra-individual to inter-species level - becomes more apparent. This variability is likely to be due to the interplay between sensory feedback and local pattern-generating activity, and depends crucially on species, walking speed and behavioral goal. Here, I gather data from the literature of panarthropod walking coordination on both flat ground and across more complex terrain. This Review aims to emphasize the value of: (1) designing experiments with an eye towards studying organisms in natural environments; (2) thoughtfully integrating results from various experimental techniques, such as neurophysiological and biomechanical studies; and (3) ensuring that data is collected and made available from a wider range of species for future comparative analyses.

Global change in brain state during spontaneous and forced walk in Drosophila is composed of combined activity patterns of different neuron classes

Movement-correlated brain activity has been found across species and brain regions. Here, we used fast whole brain lightfield imaging in adult Drosophila to investigate the relationship between walk and brain-wide neuronal activity. We observed a global change in activity that tightly correlated with spontaneous bouts of walk. While imaging specific sets of excitatory, inhibitory, and neuromodulatory neurons highlighted their joint contribution, spatial heterogeneity in walk- and turning-induced activity allowed parsing unique responses from subregions and sometimes individual candidate neurons. For example, previously uncharacterized serotonergic neurons were inhibited during walk. While activity onset in some areas preceded walk onset exclusively in spontaneously walking animals, spontaneous and forced walk elicited similar activity in most brain regions. These data suggest a major contribution of walk and walk-related sensory or proprioceptive information to global activity of all major neuronal classes.

Long timescale anti-directional rotation in Drosophila optomotor behavior

Locomotor movements cause visual images to be displaced across the eye, a retinal slip that is counteracted by stabilizing reflexes in many animals. In insects, optomotor turning causes the animal to turn in the direction of rotating visual stimuli, thereby reducing retinal slip and stabilizing trajectories through the world. This behavior has formed the basis for extensive dissections of motion vision. Here, we report that under certain stimulus conditions, two Drosophila species, including the widely studied D. melanogaster, can suppress and even reverse the optomotor turning response over several seconds. Such "anti-directional turning" is most strongly evoked by long-lasting, high-contrast, slow-moving visual stimuli that are distinct from those that promote syn-directional optomotor turning. Anti-directional turning, like the syn-directional optomotor response, requires the local motion detecting neurons T4 and T5. A subset of lobula plate tangential cells, CH cells, show involvement in these responses. Imaging from a variety of direction-selective cells in the lobula plate shows no evidence of dynamics that match the behavior, suggesting that the observed inversion in turning direction emerges downstream of the lobula plate. Further, anti-directional turning declines with age and exposure to light. These results show that Drosophila optomotor turning behaviors contain rich, stimulus-dependent dynamics that are inconsistent with simple reflexive stabilization responses.

Sphingolipids in neurodegenerative diseases

Neurodegenerative Diseases (NDDs) are a group of disorders that cause progressive deficits of neuronal function. Recent evidence argues that sphingolipid metabolism is affected in a surprisingly broad set of NDDs. These include some lysosomal storage diseases (LSDs), hereditary sensory and autonomous neuropathy (HSAN), hereditary spastic paraplegia (HSP), infantile neuroaxonal dystrophy (INAD), Friedreich's ataxia (FRDA), as well as some forms of amyotrophic lateral sclerosis (ALS) and Parkinson's disease (PD). Many of these diseases have been modeled in Drosophila melanogaster and are associated with elevated levels of ceramides. Similar changes have also been reported in vertebrate cells and mouse models. Here, we summarize studies using fly models and/or patient samples which demonstrate the nature of the defects in sphingolipid metabolism, the organelles that are implicated, the cell types that are initially affected, and potential therapeutics for these diseases.

Social behavioral profiling by unsupervised deep learning reveals a stimulative effect of dopamine D3 agonists on zebrafish sociality

It has been a major challenge to systematically evaluate and compare how pharmacological perturbations influence social behavioral outcomes. Although some pharmacological agents are known to alter social behavior, precise description and quantification of such effects have proven difficult. We developed a scalable social behavioral assay for zebrafish named ZeChat based on unsupervised deep learning to characterize sociality at high resolution. High-dimensional and dynamic social behavioral phenotypes are automatically classified using this method. By screening a neuroactive compound library, we found that different classes of chemicals evoke distinct patterns of social behavioral fingerprints. By examining these patterns, we discovered that dopamine D3 agonists possess a social stimulative effect on zebrafish. The D3 agonists pramipexole, piribedil, and 7-hydroxy-DPAT-HBr rescued social deficits in a valproic-acid-induced zebrafish autism model. The ZeChat platform provides a promising approach for dissecting the pharmacology of social behavior and discovering novel social-modulatory compounds.

Modularity in Nervous Systems—a Key to Efficient Adaptivity for Deep Reinforcement Learning

Modularity as observed in biological systems has proven valuable for guiding classical motor theories towards good answers about action selection and execution. New challenges arise when we turn to learning: Trying to scale current computational models, such as deep reinforcement learning (DRL), to action spaces, input dimensions, and time horizons seen in biological systems still faces severe obstacles unless vast amounts of training data are available. This leads to the question: does biological modularity also hold an important key for better answers to obtain efficient adaptivity for deep reinforcement learning? We review biological experimental work on modularity in biological motor control and link this with current examples of (deep) RL approaches. Analyzing outcomes of simulation studies, we show that these approaches benefit from forms of modularization as found in biological systems. We identify three different strands of modularity exhibited in biological control systems. Two of them—modularity in state (i) and in action (ii) spaces—appear as a consequence of local interconnectivity (as in reflexes) and are often modulated by higher levels in a control hierarchy. A third strand arises from chunking of action elements along a (iii) temporal dimension. Usually interacting in an overarching spatio-temporal hierarchy of the overall system, the three strands offer major “factors” decomposing the entire modularity structure. We conclude that modularity with its above strands can provide an effective prior for DRL approaches to speed up learning considerably and making learned controllers more robust and adaptive.

neuroWalknet, a controller for hexapod walking allowing for context dependent behavior

Decentralized control has been established as a key control principle in insect walking and has been successfully leveraged to account for a wide range of walking behaviors in the proposed neuroWalknet architecture. This controller allows for walking patterns at different velocities in both, forward and backward direction-quite similar to the behavior shown in stick insects-, for negotiation of curves, and for robustly dealing with various disturbances. While these simulations focus on the cooperation of different, decentrally controlled legs, here we consider a set of biological experiments not yet been tested by neuroWalknet, that focus on the function of the individual leg and are context dependent. These intraleg studies deal with four groups of interjoint reflexes. The reflexes are elicited by stimulation of the femoral chordotonal organ (fCO) or groups of campaniform sensilla (CS). Motor output signals are recorded from the alpha-joint, the beta-joint or the gamma-joint of the leg. Furthermore, the influence of these sensory inputs to artificially induced oscillations by application of pilocarpine has been studied. Although these biological data represent results obtained from different local reflexes in different contexts, they fit with and are embedded into the behavior shown by the global structure of neuroWalknet. In particular, a specific and intensively studied behavior, active reaction, has since long been assumed to represent a separate behavioral element, from which it is not clear why it occurs in some situations, but not in others. This question could now be explained as an emergent property of the holistic structure of neuroWalknet which has shown to be able to produce artificially elicited pilocarpine-driven oscillation that can be controlled by sensory input without the need of explicit innate CPG structures. As the simulation data result from a holistic system, further results were obtained that could be used as predictions to be tested in further biological experiments.

Identifying behavioral structure from deep variational embeddings of animal motion

Quantification and detection of the hierarchical organization of behavior is a major challenge in neuroscience. Recent advances in markerless pose estimation enable the visualization of high-dimensional spatiotemporal behavioral dynamics of animal motion. However, robust and reliable technical approaches are needed to uncover underlying structure in these data and to segment behavior into discrete hierarchically organized motifs. Here, we present an unsupervised probabilistic deep learning framework that identifies behavioral structure from deep variational embeddings of animal motion (VAME). By using a mouse model of beta amyloidosis as a use case, we show that VAME not only identifies discrete behavioral motifs, but also captures a hierarchical representation of the motif's usage. The approach allows for the grouping of motifs into communities and the detection of differences in community-specific motif usage of individual mouse cohorts that were undetectable by human visual observation. Thus, we present a robust approach for the segmentation of animal motion that is applicable to a wide range of experimental setups, models and conditions without requiring supervised or a-priori human interference.

Food supplementation with wheat gluten leads to climbing performance decline in Drosophila melanogaster

Gluten sensitivity is associated with digestive and neurological disorders, correlating with abnormal amino acid levels, innate immune responses, gut dysbiosis and movement incoordination. However, the molecular mechanisms linking dietary gluten and brain function remain incompletely understood. We used Drosophila melanogaster to test the effects of gluten ingestion in locomotion performance. Whereas flies on control food showed decreased climbing performance after five weeks, flies exposed to food supplemented with different gluten concentrations showed a significant locomotion decline after three weeks of treatment. Future studies will determine the mechanisms underlying the observed gluten-dependent phenotypes to establish Drosophila models for gluten sensitivity.

The evolutionary trajectory of drosophilid walking

Neural circuits must both execute the behavioral repertoire of individuals and account for behavioral variation across species. Understanding how this variation emerges over evolutionary time requires large-scale phylogenetic comparisons of behavioral repertoires. Here, we describe the evolution of walking in fruit flies by capturing high-resolution, unconstrained movement from 13 species and 15 strains of drosophilids. We find that walking can be captured in a universal behavior space, the structure of which is evolutionarily conserved. However, the occurrence of and transitions between specific movements have evolved rapidly, resulting in repeated convergent evolution in the temporal structure of locomotion. Moreover, a meta-analysis demonstrates that many behaviors evolve more rapidly than other traits. Thus, the architecture and physiology of locomotor circuits can execute precise individual movements in one species and simultaneously support rapid evolutionary changes in the temporal ordering of these modular elements across clades.

Walking strides direct rapid and flexible recruitment of visual circuits for course control in Drosophila

Flexible mapping between activity in sensory systems and movement parameters is a hallmark of motor control. This flexibility depends on the continuous comparison of short-term postural dynamics and the longer-term goals of an animal, thereby necessitating neural mechanisms that can operate across multiple timescales. To understand how such body-brain interactions emerge across timescales to control movement, we performed whole-cell patch recordings from visual neurons involved in course control in Drosophila. We show that the activity of leg mechanosensory cells, propagating via specific ascending neurons, is critical for stride-by-stride steering adjustments driven by the visual circuit, and, at longer timescales, it provides information about the moving body's state to flexibly recruit the visual circuit for course control. Thus, our findings demonstrate the presence of an elegant stride-based mechanism operating at multiple timescales for context-dependent course control. We propose that this mechanism functions as a general basis for the adaptive control of locomotion.

Neural mechanisms to exploit positional geometry for collision avoidance

Visual motion provides rich geometrical cues about the three-dimensional configuration of the world. However, how brains decode the spatial information carried by motion signals remains poorly understood. Here, we study a collision-avoidance behavior in Drosophila as a simple model of motion-based spatial vision. With simulations and psychophysics, we demonstrate that walking Drosophila exhibit a pattern of slowing to avoid collisions by exploiting the geometry of positional changes of objects on near-collision courses. This behavior requires the visual neuron LPLC1, whose tuning mirrors the behavior and whose activity drives slowing. LPLC1 pools inputs from object and motion detectors, and spatially biased inhibition tunes it to the geometry of collisions. Connectomic analyses identified circuitry downstream of LPLC1 that faithfully inherits its response properties. Overall, our results reveal how a small neural circuit solves a specific spatial vision task by combining distinct visual features to exploit universal geometrical constraints of the visual world.

NeuroMechFly, a neuromechanical model of adult Drosophila melanogaster

Animal behavior emerges from an interaction between neural network dynamics, musculoskeletal properties and the physical environment. Accessing and understanding the interplay between these elements requires the development of integrative and morphologically realistic neuromechanical simulations. Here we present NeuroMechFly, a data-driven model of the widely studied organism, Drosophila melanogaster. NeuroMechFly combines four independent computational modules: a physics-based simulation environment, a biomechanical exoskeleton, muscle models and neural network controllers. To enable use cases, we first define the minimum degrees of freedom of the leg from real three-dimensional kinematic measurements during walking and grooming. Then, we show how, by replaying these behaviors in the simulator, one can predict otherwise unmeasured torques and contact forces. Finally, we leverage NeuroMechFly's full neuromechanical capacity to discover neural networks and muscle parameters that drive locomotor gaits optimized for speed and stability. Thus, NeuroMechFly can increase our understanding of how behaviors emerge from interactions between complex neuromechanical systems and their physical surroundings.

Thorax-Segment- and Leg-Segment-Specific Motor Control for Adaptive Behavior

We have just started to understand the mechanisms underlying flexibility of motor programs among segmental neural networks that control each individual leg during walking in vertebrates and invertebrates. Here, we investigated the mechanisms underlying curve walking in the stick insect Carausius morosus during optomotor-induced turning. We wanted to know, whether the previously reported body-side specific changes in a two-front leg turning animal are also observed in the other thoracic leg segments. The motor activity of the three major leg joints showed three types of responses: 1) a context-dependent increase or decrease in motor neuron (MN) activity of the antagonistic MN pools of the thorax-coxa (ThC)-joint during inside and outside turns; 2) an activation of 1 MN pool with simultaneous cessation of the other, independent of the turning direction in the coxa-trochanteral (CTr)-joint; 3) a modification in the activity of both FTi-joint MN pools which depended on the turning direction in one, but not in the other thorax segment. By pharmacological activation of the meso- or metathoracic central pattern generating networks (CPG), we show that turning-related modifications in motor output involve changes to local CPG activity. The rhythmic activity in the MN pools of the ThC and CTr-joints was modified similarly to what was observed under control conditions in saline. Our results indicate that changes in meso- and metathoracic motor activity during curve walking are leg-joint- and thorax-segment-specific, can depend on the turning direction, and are mediated through changes in local CPG activity.

Using Artificial Intelligence for Assistance Systems to Bring Motor Learning Principles into Real World Motor Tasks

Humans learn movements naturally, but it takes a lot of time and training to achieve expert performance in motor skills. In this review, we show how modern technologies can support people in learning new motor skills. First, we introduce important concepts in motor control, motor learning and motor skill learning. We also give an overview about the rapid expansion of machine learning algorithms and sensor technologies for human motion analysis. The integration between motor learning principles, machine learning algorithms and recent sensor technologies has the potential to develop AI-guided assistance systems for motor skill training. We give our perspective on this integration of different fields to transition from motor learning research in laboratory settings to real world environments and real world motor tasks and propose a stepwise approach to facilitate this transition.

Gating attractor dynamics of frontal cortex under acupuncture via recurrent neural network

Acupuncture can regulate the functions of human body and improve the cognition of brain. However, the mechanism of acupuncture manipulations remains unclear. Here, we hypothesis that the frontal cortex plays a gating role in information routing of brain network under acupuncture. To that end, the gating effect of frontal cortex under acupuncture is analyzed in combination with EEG data of acupuncture at Zusanli acupoints. In addition, recurrent neural network (RNN) is used to reproduce the dynamics of frontal cortex under normal state and acupuncture state. From low-dimensional view, it is shown that the brain networks under acupuncture state can show stable attractor cycle dynamics, which may explain the regulation effect of acupuncture. Comparing with different manipulations, we find that the attractor of low-dimensional trajectory varies under different frequencies of acupuncture. Besides, a strip gated band of neural dynamics is found by changing the frequency of stimulation and excitatory-inhibitory balance of network. The attractor state is found to transport in the gating area under different stimulation frequencies, and the probability of attractor migration is different across acupuncture manipulations. This reverse engineering of brain network indicates that the differences among acupuncture manipulations are caused by interaction and separation in the neural activity space between attractors that encode acupuncture function. Consequently, our results may provide help for quantitative analysis of acupuncture, and benefit for the clinical guidance of acupuncture clinicians.

Descending neurons coordinate anterior grooming behavior in Drosophila

The brain coordinates the movements that constitute behavior, but how descending neurons convey the myriad of commands required to activate the motor neurons of the limbs in the right order and combinations to produce those movements is not well understood. For anterior grooming behavior in the fly, we show that its component head sweeps and leg rubs can be initiated separately, or as a set, by different descending neurons. Head sweeps and leg rubs are mutually exclusive movements of the front legs that normally alternate, and we show that circuits in the ventral nerve cord as well as in the brain can resolve competing commands. Finally, the left and right legs must work together to remove debris. The coordination for leg rubs can be achieved by unilateral activation of a single descending neuron, while a similar manipulation of a different descending neuron decouples the legs to produce single-sided head sweeps. Taken together, these results demonstrate that distinct descending neurons orchestrate the complex alternation between the movements that make up anterior grooming.

Shallow neural networks trained to detect collisions recover features of visual loom-selective neurons

Animals have evolved sophisticated visual circuits to solve a vital inference problem: detecting whether or not a visual signal corresponds to an object on a collision course. Such events are detected by specific circuits sensitive to visual looming, or objects increasing in size. Various computational models have been developed for these circuits, but how the collision-detection inference problem itself shapes the computational structures of these circuits remains unknown. Here, inspired by the distinctive structures of LPLC2 neurons in the visual system of Drosophila, we build anatomically-constrained shallow neural network models and train them to identify visual signals that correspond to impending collisions. Surprisingly, the optimization arrives at two distinct, opposing solutions, only one of which matches the actual dendritic weighting of LPLC2 neurons. Both solutions can solve the inference problem with high accuracy when the population size is large enough. The LPLC2-like solutions reproduces experimentally observed LPLC2 neuron responses for many stimuli, and reproduces canonical tuning of loom sensitive neurons, even though the models are never trained on neural data. Thus, LPLC2 neuron properties and tuning are predicted by optimizing an anatomically-constrained neural network to detect impending collisions. More generally, these results illustrate how optimizing inference tasks that are important for an animal's perceptual goals can reveal and explain computational properties of specific sensory neurons.

Transforming representations of movement from body- to world-centric space

When an animal moves through the world, its brain receives a stream of information about the body's translational velocity from motor commands and sensory feedback signals. These incoming signals are referenced to the body, but ultimately, they must be transformed into world-centric coordinates for navigation1,2. Here we show that this computation occurs in the fan-shaped body in the brain of Drosophila melanogaster. We identify two cell types, PFNd and PFNv3-5, that conjunctively encode translational velocity and heading as a fly walks. In these cells, velocity signals are acquired from locomotor brain regions6 and are multiplied with heading signals from the compass system. PFNd neurons prefer forward-ipsilateral movement, whereas PFNv neurons prefer backward-contralateral movement, and perturbing PFNd neurons disrupts idiothetic path integration in walking flies7. Downstream, PFNd and PFNv neurons converge onto hΔB neurons, with a connectivity pattern that pools together heading and translation direction combinations corresponding to the same movement in world-centric space. This network motif effectively performs a rotation of the brain's representation of body-centric translational velocity according to the current heading direction. Consistent with our predictions, we observe that hΔB neurons form a representation of translational velocity in world-centric coordinates. By integrating this representation over time, it should be possible for the brain to form a working memory of the path travelled through the environment8-10.

A connectome of the Drosophila central complex reveals network motifs suitable for flexible navigation and context-dependent action selection

Flexible behaviors over long timescales are thought to engage recurrent neural networks in deep brain regions, which are experimentally challenging to study. In insects, recurrent circuit dynamics in a brain region called the central complex (CX) enable directed locomotion, sleep, and context- and experience-dependent spatial navigation. We describe the first complete electron microscopy-based connectome of the Drosophila CX, including all its neurons and circuits at synaptic resolution. We identified new CX neuron types, novel sensory and motor pathways, and network motifs that likely enable the CX to extract the fly's head direction, maintain it with attractor dynamics, and combine it with other sensorimotor information to perform vector-based navigational computations. We also identified numerous pathways that may facilitate the selection of CX-driven behavioral patterns by context and internal state. The CX connectome provides a comprehensive blueprint necessary for a detailed understanding of network dynamics underlying sleep, flexible navigation, and state-dependent action selection.

Insect locomotion: Flies show you how to stay on course

Walking animals are faced with making a trade-off between maintaining a stable posture and gait and pursuing other goals such as keeping a straight path. A new study on exploratory walking in flies provides a sophisticated quantitative account of this behavioural problem, with some intriguing discoveries.

Fast tuning of posture control by visual feedback underlies gaze stabilization in walking Drosophila

Locomotion requires a balance between mechanical stability and movement flexibility to achieve behavioral goals despite noisy neuromuscular systems, but rarely is it considered how this balance is orchestrated. We combined virtual reality tools with quantitative analysis of behavior to examine how Drosophila uses self-generated visual information (reafferent visual feedback) to control gaze during exploratory walking. We found that flies execute distinct motor programs coordinated across the body to maximize gaze stability. However, the presence of inherent variability in leg placement relative to the body jeopardizes fine control of gaze due to posture-stabilizing adjustments that lead to unintended changes in course direction. Surprisingly, whereas visual feedback is dispensable for head-body coordination, we found that self-generated visual signals tune postural reflexes to rapidly prevent turns rather than to promote compensatory rotations, a long-standing idea for visually guided course control. Together, these findings support a model in which visual feedback orchestrates the interplay between posture and gaze stability in a manner that is both goal dependent and motor-context specific.

Decentralized control and local information for robust and adaptive decentralized Deep Reinforcement Learning

Decentralization is a central characteristic of biological motor control that allows for fast responses relying on local sensory information. In contrast, the current trend of Deep Reinforcement Learning (DRL) based approaches to motor control follows a centralized paradigm using a single, holistic controller that has to untangle the whole input information space. This motivates to ask whether decentralization as seen in biological control architectures might also be beneficial for embodied sensori-motor control systems when using DRL. To answer this question, we provide an analysis and comparison of eight control architectures for adaptive locomotion that were derived for a four-legged agent, but with their degree of decentralization varying systematically between the extremes of fully centralized and fully decentralized. Our comparison shows that learning speed is significantly enhanced in distributed architectures-while still reaching the same high performance level of centralized architectures-due to smaller search spaces and local costs providing more focused information for learning. Second, we find an increased robustness of the learning process in the decentralized cases-it is less demanding to hyperparameter selection and less prone to becoming trapped in poor local minima. Finally, when examining generalization to uneven terrains-not used during training-we find best performance for an intermediate architecture that is decentralized, but integrates only local information from both neighboring legs. Together, these findings demonstrate beneficial effects of distributing control into decentralized units and relying on local information. This appears as a promising approach towards more robust DRL and better generalization towards adaptive behavior.

Anipose: A toolkit for robust markerless 3D pose estimation

Quantifying movement is critical for understanding animal behavior. Advances in computer vision now enable markerless tracking from 2D video, but most animals move in 3D. Here, we introduce Anipose, an open-source toolkit for robust markerless 3D pose estimation. Anipose is built on the 2D tracking method DeepLabCut, so users can expand their existing experimental setups to obtain accurate 3D tracking. It consists of four components: (1) a 3D calibration module, (2) filters to resolve 2D tracking errors, (3) a triangulation module that integrates temporal and spatial regularization, and (4) a pipeline to structure processing of large numbers of videos. We evaluate Anipose on a calibration board as well as mice, flies, and humans. By analyzing 3D leg kinematics tracked with Anipose, we identify a key role for joint rotation in motor control of fly walking. To help users get started with 3D tracking, we provide tutorials and documentation at http://anipose.org/.

Predicting individual neuron responses with anatomically constrained task optimization

Artificial neural networks trained to solve sensory tasks can develop statistical representations that match those in biological circuits. However, it remains unclear whether they can reproduce properties of individual neurons. Here, we investigated how artificial networks predict individual neuron properties in the visual motion circuits of the fruit fly Drosophila. We trained anatomically constrained networks to predict movement in natural scenes, solving the same inference problem as fly motion detectors. Units in the artificial networks adopted many properties of analogous individual neurons, even though they were not explicitly trained to match these properties. Among these properties was the split into ON and OFF motion detectors, which is not predicted by classical motion detection models. The match between model and neurons was closest when models were trained to be robust to noise. These results demonstrate how anatomical, task, and noise constraints can explain properties of individual neurons in a small neural network.

Unsupervised quantification of naturalistic animal behaviors for gaining insight into the brain

Neural computation has evolved to optimize the behaviors that enable our survival. Although much previous work in neuroscience has focused on constrained task behaviors, recent advances in computer vision are fueling a trend toward the study of naturalistic behaviors. Automated tracking of fine-scale behaviors is generating rich datasets for animal models including rodents, fruit flies, zebrafish, and worms. However, extracting meaning from these large and complex data often requires sophisticated computational techniques. Here we review the latest methods and modeling approaches providing new insights into the brain from behavior. We focus on unsupervised methods for identifying stereotyped behaviors and for resolving details of the structure and dynamics of behavioral sequences.

Tardigrades exhibit robust interlimb coordination across walking speeds and terrains

Tardigrades must negotiate heterogeneous, fluctuating environments and accordingly utilize locomotive strategies capable of dealing with variable terrain. We analyze the kinematics and interleg coordination of freely walking tardigrades (species: Hypsibius exemplaris). We find that tardigrade walking replicates several key features of walking in insects despite disparities in size, skeleton, and habitat. To test the effect of environmental changes on tardigrade locomotor control circuits we measure kinematics and interleg coordination during walking on two substrates of different stiffnesses. We find that the phase offset between contralateral leg pairs is flexible, while ipsilateral coordination is preserved across environmental conditions. This mirrors similar results in insects and crustaceans. We propose that these functional similarities in walking coordination between tardigrades and arthropods is either due to a generalized locomotor control circuit common to panarthropods or to independent convergence onto an optimal strategy for robust multilegged control in small animals with simple circuitry. Our results highlight the value of tardigrades as a comparative system toward understanding the mechanisms-neural and/or mechanical-underlying coordination in panarthropod locomotion.

LiftPose3D, a deep learning-based approach for transforming two-dimensional to three-dimensional poses in laboratory animals

Markerless three-dimensional (3D) pose estimation has become an indispensable tool for kinematic studies of laboratory animals. Most current methods recover 3D poses by multi-view triangulation of deep network-based two-dimensional (2D) pose estimates. However, triangulation requires multiple synchronized cameras and elaborate calibration protocols that hinder its widespread adoption in laboratory studies. Here we describe LiftPose3D, a deep network-based method that overcomes these barriers by reconstructing 3D poses from a single 2D camera view. We illustrate LiftPose3D's versatility by applying it to multiple experimental systems using flies, mice, rats and macaques, and in circumstances where 3D triangulation is impractical or impossible. Our framework achieves accurate lifting for stereotypical and nonstereotypical behaviors from different camera angles. Thus, LiftPose3D permits high-quality 3D pose estimation in the absence of complex camera arrays and tedious calibration procedures and despite occluded body parts in freely behaving animals.

Universal Features in Panarthropod Inter-Limb Coordination during Forward Walking

Terrestrial animals must often negotiate heterogeneous, varying environments. Accordingly, their locomotive strategies must adapt to a wide range of terrain, as well as to a range of speeds to accomplish different behavioral goals. Studies in Drosophila have found that inter-leg coordination patterns (ICPs) vary smoothly with walking speed, rather than switching between distinct gaits as in vertebrates (e.g., horses transitioning between trotting and galloping). Such a continuum of stepping patterns implies that separate neural controllers are not necessary for each observed ICP. Furthermore, the spectrum of Drosophila stepping patterns includes all canonical coordination patterns observed during forward walking in insects. This raises the exciting possibility that the controller in Drosophila is common to all insects, and perhaps more generally to panarthropod walkers. Here, we survey and collate data on leg kinematics and inter-leg coordination relationships during forward walking in a range of arthropod species, as well as include data from a recent behavioral investigation into the tardigrade Hypsibius exemplaris. Using this comparative dataset, we point to several functional and morphological features that are shared among panarthropods. The goal of the framework presented in this review is to emphasize the importance of comparative functional and morphological analyses in understanding the origins and diversification of walking in Panarthropoda. Introduction.

Measuring and modeling the motor system with machine learning

The utility of machine learning in understanding the motor system is promising a revolution in how to collect, measure, and analyze data. The field of movement science already elegantly incorporates theory and engineering principles to guide experimental work, and in this review we discuss the growing use of machine learning: from pose estimation, kinematic analyses, dimensionality reduction, and closed-loop feedback, to its use in understanding neural correlates and untangling sensorimotor systems. We also give our perspective on new avenues, where markerless motion capture combined with biomechanical modeling and neural networks could be a new platform for hypothesis-driven research.

From Motor-Output to Connectivity: An In-Depth Study of in-vitro Rhythmic Patterns in the Cockroach Periplaneta americana

The cockroach is an established model in the study of locomotion control. While previous work has offered important insights into the interplay among brain commands, thoracic central pattern generators, and the sensory feedback that shapes their motor output, there remains a need for a detailed description of the central pattern generators' motor output and their underlying connectivity scheme. To this end, we monitored pilocarpine-induced activity of levator and depressor motoneurons in two types of novel in-vitro cockroach preparations: isolated thoracic ganglia and a whole-chain preparation comprising the thoracic ganglia and the subesophageal ganglion. Our data analyses focused on the motoneuron firing patterns and the coordination among motoneuron types in the network. The burstiness and rhythmicity of the motoneurons were monitored, and phase relations, coherence, coupling strength, and frequency-dependent variability were analyzed. These parameters were all measured and compared among network units both within each preparation and among the preparations. Here, we report differences among the isolated ganglia, including asymmetries in phase and coupling strength, which indicate that they are wired to serve different functions. We also describe the intrinsic default gait and a frequency-dependent coordination. The depressor motoneurons showed mostly similar characteristics throughout the network regardless of interganglia connectivity; whereas the characteristics of the levator motoneurons activity were mostly ganglion-dependent, and influenced by the presence of interganglia connectivity. Asymmetries were also found between the anterior and posterior homolog parts of the thoracic network, as well as between ascending and descending connections. Our analyses further discover a frequency-dependent inversion of the interganglia coordination from alternations between ipsilateral homolog oscillators to simultaneous activity. We present a detailed scheme of the network couplings, formulate coupling rules, and review a previously suggested model of connectivity in light of our new findings. Our data support the notion that the inter-hemiganglia coordination derives from the levator networks and their coupling with local depressor interneurons. Our findings also support a dominant role of the metathoracic ganglion and its ascending output in governing the anterior ganglia motor output during locomotion in the behaving animal.

Reconstruction of motor control circuits in adult Drosophila using automated transmission electron microscopy

To investigate circuit mechanisms underlying locomotor behavior, we used serial-section electron microscopy (EM) to acquire a synapse-resolution dataset containing the ventral nerve cord (VNC) of an adult female Drosophila melanogaster. To generate this dataset, we developed GridTape, a technology that combines automated serial-section collection with automated high-throughput transmission EM. Using this dataset, we studied neuronal networks that control leg and wing movements by reconstructing all 507 motor neurons that control the limbs. We show that a specific class of leg sensory neurons synapses directly onto motor neurons with the largest-caliber axons on both sides of the body, representing a unique pathway for fast limb control. We provide open access to the dataset and reconstructions registered to a standard atlas to permit matching of cells between EM and light microscopy data. We also provide GridTape instrumentation designs and software to make large-scale EM more accessible and affordable to the scientific community.

Continuous Whole-Body 3D Kinematic Recordings across the Rodent Behavioral Repertoire

In mammalian animal models, high-resolution kinematic tracking is restricted to brief sessions in constrained environments, limiting our ability to probe naturalistic behaviors and their neural underpinnings. To address this, we developed CAPTURE (Continuous Appendicular and Postural Tracking Using Retroreflector Embedding), a behavioral monitoring system that combines motion capture and deep learning to continuously track the 3D kinematics of a rat's head, trunk, and limbs for week-long timescales in freely behaving animals. CAPTURE realizes 10- to 100-fold gains in precision and robustness compared with existing convolutional network approaches to behavioral tracking. We demonstrate CAPTURE's ability to comprehensively profile the kinematics and sequential organization of natural rodent behavior, its variation across individuals, and its perturbation by drugs and disease, including identifying perseverative grooming states in a rat model of fragile X syndrome. CAPTURE significantly expands the range of behaviors and contexts that can be quantitatively investigated, opening the door to a new understanding of natural behavior and its neural basis.

Drosophila uses a tripod gait across all walking speeds, and the geometry of the tripod is important for speed control

Changes in walking speed are characterized by changes in both the animal's gait and the mechanics of its interaction with the ground. Here we study these changes in walking Drosophila. We measured the fly's center of mass movement with high spatial resolution and the position of its footprints. Flies predominantly employ a modified tripod gait that only changes marginally with speed. The mechanics of a tripod gait can be approximated with a simple model - angular and radial spring-loaded inverted pendulum (ARSLIP) - which is characterized by two springs of an effective leg that become stiffer as the speed increases. Surprisingly, the change in the stiffness of the spring is mediated by the change in tripod shape rather than a change in stiffness of individual legs. The effect of tripod shape on mechanics can also explain the large variation in kinematics among insects, and ARSLIP can model these variations.