Brains contain networks of interconnected neurons, so knowing the network architecture is essential for understanding brain function. We therefore mapped the synaptic-resolution connectome of an insect brain (Drosophila larva) with rich behavior, including learning, value-computation, and action-selection, comprising 3,013 neurons and 544,000 synapses. We characterized neuron-types, hubs, feedforward and feedback pathways, and cross-hemisphere and brain-nerve cord interactions. We found pervasive multisensory and interhemispheric integration, highly recurrent architecture, abundant feedback from descending neurons, and multiple novel circuit motifs. The brain’s most recurrent circuits comprised the input and output neurons of the learning center. Some structural features, including multilayer shortcuts and nested recurrent loops, resembled powerful machine learning architectures. The identified brain architecture provides a basis for future experimental and theoretical studies of neural circuits.

To accurately track self-location, animals need to integrate their movements through space. In amniotes, representations of self-location have been found in regions such as the hippocampus. It is unknown whether more ancient brain regions contain such representations and by which pathways they may drive locomotion. Fish displaced by water currents must prevent uncontrolled drift to potentially dangerous areas. We found that larval zebrafish track such movements and can later swim back to their earlier location. Whole-brain functional imaging revealed the circuit enabling this process of positional homeostasis. Position-encoding brainstem neurons integrate optic flow, then bias future swimming to correct for past displacements by modulating inferior olive and cerebellar activity. Manipulation of position-encoding or olivary neurons abolished positional homeostasis or evoked behavior as if animals had experienced positional shifts. These results reveal a multiregional hindbrain circuit in vertebrates for optic flow integration, memory of self-location, and its neural pathway to behavior.Competing Interest StatementThe authors have declared no competing interest.

The central amygdala (CEA) has been richly studied for interpreting function and behavior according to specific cell types and circuits. Such work has typically defined molecular cell types by classical inhibitory marker genes; consequently, whether marker-gene-defined cell types exhaustively cover the CEA and co-vary with connectivity remains unresolved. Here, we combined single-cell RNA sequencing, multiplexed fluorescent in situ hybridization, immunohistochemistry, and long-range projection mapping to derive a “bottom-up” understanding of CEA cell types. In doing so, we identify two major cell types, encompassing one-third of all CEA neurons, that have gone unresolved in previous studies. In spatially mapping these novel types, we identify a non-canonical CEA subdomain associated with Nr2f2 expression and uncover an Isl1-expressing medial cell type that accounts for many long-range CEA projections. Our results reveal new CEA organizational principles across cell types and spatial scales and provide a framework for future work examining cell-type-specific behavior and function.

Cells form networks in animal tissues through synaptic, chemical, and adhesive links. Invertebrate muscle cells often connect to other cells through desmosomes, adhesive junctions anchored by intermediate filaments. To study desmosomal networks, we skeletonised 853 muscle cells and their desmosomal partners in volume electron microscopy data covering an entire larva of the annelid . Muscle cells adhere to each other, to epithelial, glial, ciliated, and bristle-producing cells and to the basal lamina, forming a desmosomal connectome of over 2000 cells. The aciculae - chitin rods that form an endoskeleton in the segmental appendages - are highly connected hubs in this network. This agrees with the many degrees of freedom of their movement, as revealed by video microscopy. Mapping motoneuron synapses to the desmosomal connectome allowed us to infer the extent of tissue influenced by motoneurons. Our work shows how cellular-level maps of synaptic and adherent force networks can elucidate body mechanics.

Many animals rely on vision to navigate through their environment. The pattern of changes in the visual scene induced by self-motion is the optic flow1, which is first estimated in local patches by directionally selective (DS) neurons2–4. But how should the arrays of DS neurons, each responsive to motion in a preferred direction at a specific retinal position, be organized to support robust decoding of optic flow by downstream circuits? Understanding this global organization is challenging because it requires mapping fine, local features of neurons across the animal’s field of view3. In Drosophila, the asymmetric dendrites of the T4 and T5 DS neurons establish their preferred direction, making it possible to predict DS responses from anatomy4,5. Here we report that the preferred directions of fly DS neurons vary at different retinal positions and show that this spatial variation is established by the anatomy of the compound eye. To estimate the preferred directions across the visual field, we reconstructed hundreds of T4 neurons in a full brain EM volume6 and discovered unexpectedly stereotypical dendritic arborizations that are independent of location. We then used whole-head μCT scans to map the viewing directions of all compound eye facets and found a non-uniform sampling of visual space that explains the spatial variation in preferred directions. Our findings show that the organization of preferred directions in the fly is largely determined by the compound eye, exposing an intimate and unexpected connection between the peripheral structure of the eye, functional properties of neurons deep in the brain, and the control of body movements.

Animals flexibly switch between different actions by changing neural activity patterns for motor control. Courting Drosophila melanogaster males produce two different acoustic signals, pulse and sine song, each of which can be promoted by artificial activation of distinct neurons. However, how the activity of these neurons implements flexible song production is unknown. Here, we developed an assay to record neuronal calcium signals in the ventral nerve cord, which contains the song motor circuit, in singing flies. We found that sine-promoting neurons, but not pulse-promoting neurons, show strong activation during sine song. In contrast, both pulse- and sine-promoting neurons are active during pulse song. Furthermore, population calcium imaging in the song circuit suggests that sine song involves activation of a subset of neurons that are also active during pulse song. Thus, differential activation of overlapping, rather than distinct, neural populations underlies flexible motor actions during acoustic communication in D. melanogaster.

While insects such as are flying, aerodynamic instabilities require that they make millisecond time scale adjustments to their wing motion to stay aloft and on course. These stabilization reflexes can be modeled as a proportional-integral (PI) controller; however, it is unclear how such control might be instantiated in insects at the level of muscles and neurons. Here, we show that the b1 and b2 motor units-prominent components of the fly's steering muscle system-modulate specific elements of the PI controller: the angular displacement (integral) and angular velocity (proportional), respectively. Moreover, these effects are observed only during the stabilization of pitch. Our results provide evidence for an organizational principle in which each muscle contributes to a specific functional role in flight control, a finding that highlights the power of using top-down behavioral modeling to guide bottom-up cellular manipulation studies.

During their lifetime, animals must adapt their behavior to survive in changing environments. This ability requires the nervous system to adjust through dynamic expression of neurotransmitters and receptors but also through growth, spatial reorganization and connectivity while integrating external stimuli. For instance, despite having a fixed neuronal cell lineage, the nematode Caenorhabditis elegans’ nervous system remains plastic throughout its development. Here, we focus on a specific example of nervous system plasticity, the C. elegans dauer exit decision. Under unfavorable conditions, larvae will enter the non-feeding and non-reproductive dauer stage and adapt their behavior to cope with a new environment. Upon improved conditions, this stress resistant developmental stage is actively reversed to resume reproductive development. However, how different environmental stimuli regulate the exit decision mechanism and thereby drive the larva’s behavioral change is unknown. To fill this gap, we developed a new open hardware method for long-term imaging (12h) of C. elegans larvae. We identified dauer-specific behavioral motifs and characterized the behavioral trajectory of dauer exit in different environments to identify key decision points. Combining long-term behavioral imaging with transcriptomics, we find that bacterial ingestion triggers a change in neuropeptide gene expression to establish post-dauer behavior. Taken together, we show how a developing nervous system can robustly integrate environmental changes, activate a developmental switch and adapt the organism’s behavior to a new environment.

Learning in deep neural networks is known to depend critically on the knowledge embedded in the initial network weights. However, few theoretical results have precisely linked prior knowledge to learning dynamics. Here we derive exact solutions to the dynamics of learning with rich prior knowledge in deep linear networks by generalising Fukumizu's matrix Riccati solution \citep{fukumizu1998effect}. We obtain explicit expressions for the evolving network function, hidden representational similarity, and neural tangent kernel over training for a broad class of initialisations and tasks. The expressions reveal a class of task-independent initialisations that radically alter learning dynamics from slow non-linear dynamics to fast exponential trajectories while converging to a global optimum with identical representational similarity, dissociating learning trajectories from the structure of initial internal representations. We characterise how network weights dynamically align with task structure, rigorously justifying why previous solutions successfully described learning from small initial weights without incorporating their fine-scale structure. Finally, we discuss the implications of these findings for continual learning, reversal learning and learning of structured knowledge. Taken together, our results provide a mathematical toolkit for understanding the impact of prior knowledge on deep learning.

Animals learn trajectories to rewards in both spatial, navigational contexts and relational, non-navigational contexts. Synchronous reactivation of hippocampal activity is thought to be critical for recall and evaluation of trajectories for learning. Do hippocampal representations differentially contribute to experience-dependent learning of trajectories across spatial and relational contexts? In this study, we trained mice to navigate to a hidden target in a physical arena or manipulate a joystick to a virtual target to collect delayed rewards. In a navigational context, calcium imaging in freely moving mice revealed that synchronous CA1 reactivation was retrospective and important for evaluation of prior navigational trajectories. In a non-navigational context, reactivation was prospective and important for initiation of joystick trajectories, even in the same animals trained in both contexts. Adaptation of trajectories to a new target was well-explained by a common learning algorithm in which hippocampal activity makes dissociable contributions to reinforcement learning computations depending upon spatial context.