Animals continuously extract and evaluate diverse sensory information from the environment to guide behavior. Yet, how neural circuits integrate multiple, potentially conflicting, inputs remains poorly understood. Here, we use larval zebrafish to address this question, leveraging their robust optomotor response to coherent random dot motion and phototaxis towards light. We demonstrate that animals employ an additive behavioral algorithm of three visual features: motion coherence, luminance level, and changes in luminance. Using brain-wide two-photon imaging, we identify the loci of these computations, with the anterior hindbrain emerging as a multifeature integration hub. Through single-cell neurotransmitter and morphological analyses of functionally identified neurons, we characterize potential connections within and across computational nodes. These experiments reveal three parallel and converging computational pathways, matching our behavioral results. Our study provides a mechanistic brain-wide account of how a vertebrate brain integrates multiple features to drive sensorimotor decisions, bridging behavioral algorithms with their neural implementation.

Animals stabilize their heads to ensure stable sensory input and effective motor coordination. Head stabilization in response to vestibular stimuli is mediated by the vestibulo-collic reflex (VCR). While the VCR has been characterized in tetrapods, it remains unknown whether fish, which lack an anatomical neck, employ head-stabilization behavior. Here, we demonstrate that larval zebrafish exhibit head-stabilization behavior: during pitch tilts, they adjust their head orientation relative to the body by rostral body flexion. The rostral body flexes ventrally during head-up posture, whereas it flexes dorsally during head-down posture. These flexions partially compensate for head pitch changes, contributing to head stabilization. We identify the neural circuits and muscles responsible for these flexions. Both dorsal and ventral flexions are mediated by the same vestibular nucleus, but neural signals were transmitted through distinct pathways, either involving or bypassing a class of reticulospinal neurons. The dorsal and ventral flexions are produced by specialized dorsal and ventral muscles in the rostral body, respectively. The neural circuits underlying these body flexions share similarities with those underlying the mammalian VCR. Together, our results suggest that a head-stabilization system may have existed in ancestral fish and may have provided an evolutionary foundation upon which the tetrapod VCR emerged.

Across animal taxa, nest-building behavior is performed using a generalizable and flexible action sequence. In order to accomplish its goal (a stable nest), the brain appears to compare intermediate steps in the process to stored neural representations that are reminiscent of cognitive templates. Deviations from these templates drive progress, preferences, and corrections in the execution of this behavior. Here, we investigated the stereotypy vs. plasticity of nest building by the cichlid Lamprologus ocellatus, a fish species that manipulates abandoned snail shells to build shelters for breeding and protection. We find that nest building is composed of a sequence of behaviors that are tied together by a series of stimulus-response loops, allowing for restarts and shortcuts as the behavioral program unfolds. The attraction to a shell object is innate, as is the final appearance of the nest. The behavior of an inexperienced animal is initially uncoordinated but is fine-tuned by repeated building opportunities. Shells need to conform to rigid geometric criteria in order to be acceptable as a potential home. Nest building is accompanied by focused neural activity in brain regions homologous to the mammalian hippocampus and neocortex. In conclusion, we have uncovered the constraints and flexibility of an instinctive, goal-directed behavior, which appears to employ cognitive template matching. Video abstract

Neuronal phenotypic traits such as morphology, connectivity and function are dictated, to a large extent, by a specific combination of differentially expressed genes. Clusters of neurons in transcriptomic space correspond to distinct cell types and in some cases-for example, Caenorhabditis elegans neurons1 and retinal ganglion cells2-4-have been shown to share morphology and function. The zebrafish optic tectum is composed of a spatial array of neurons that transforms visual inputs into motor outputs. Although the visuotopic map is continuous, subregions of the tectum are functionally specialized5,6. Here, to uncover the cell-type architecture of the tectum, we transcriptionally profiled its neurons, revealing more than 60 cell types that are organized in distinct anatomical layers. We measured the visual responses of thousands of tectal neurons by two-photon calcium imaging and matched them with their transcriptional profiles. Furthermore, we characterized the morphologies of transcriptionally identified neurons using specific transgenic lines. Notably, we found that neurons that are transcriptionally similar can diverge in shape, connectivity and visual responses. Incorporating the spatial coordinates of neurons within the tectal volume revealed functionally and morphologically defined anatomical subclusters within individual transcriptomic clusters. Our findings demonstrate that extrinsic, position-dependent factors expand the phenotypic repertoire of genetically similar neurons.

As animals adapt to new situations, neuromodulation is a potent way to alter behavior, yet mechanisms by which neuromodulatory nuclei compute during behavior are underexplored. The serotonergic raphe supports motor learning in larval zebrafish by visually detecting distance traveled during swims, encoding action effectiveness, and modulating motor vigor. We tracked the raphe's input-output computations at millisecond timescales using voltage and neurotransmitter imaging and found that swimming opens a gate for visual input to cause spiking in serotonergic neurons, enabling the encoding of action outcomes and filtering out learning-irrelevant visual signals. Specifically, swim commands initially inhibited serotonergic neurons via γ-aminobutyric acid (GABA). Immediately after, membrane voltage increased via post-inhibitory rebound, allowing swim-induced visual motion to evoke firing through glutamate, triggering serotonin release to modulate future motor vigor. Ablating local GABAergic neurons impaired raphe coding and motor learning. Thus, serotonergic neuromodulation arises from action-outcome coincidence detection within the raphe.

Animal behavior is adapted to the sensory environment in which it evolved, while also being constrained by physical limits, evolutionary history, and developmental trajectories. The hunting behavior of larval zebrafish (Danio rerio), a cyprinid native to streams in Eastern India, has been well characterized. However, it is unknown if the complement and sequence of movements employed during prey capture by zebrafish is universal across freshwater teleosts. Here, we explore the syntax of prey capture behavior in larval fish belonging to the clade Percomorpha, whose last common ancestor with cyprinids lived ~240 mya. We compared the behavior of four cichlid species from Lake Tanganyika endemic to deep benthic parts of the lake (Lepidiolamprologus attenuatus, Lamprologus ocellatus, and Neolamprologus multifasciatus) or inhabiting rivers (Astatotilapia burtoni) with that of medaka (Oryzias latipes), a fish found in rice paddies in East Asia. Using high-speed videography and neural networks, we tracked eye movements and extracted swim kinematics during hunting from larvae of these five species. Notably, we found that the repertoire of hunting movements of cichlids is broader than that of zebrafish, but shares basic features, such as eye convergence, positioning of prey centrally in the binocular visual field, and discrete prey capture bouts, including two kinds of capture strikes. In contrast, medaka swim continuously, track the prey monocularly without eye convergence, and position prey laterally before capturing them with a side swing. This configuration of kinematic motifs suggests that medaka may judge distance to prey predominantly by motion parallax, while cichlids and zebrafish may mainly use binocular visual cues. Together, our study documents the diversification of locomotor and oculomotor adaptations among hunting teleost larvae.

Animals continuously extract and evaluate diverse sensory information from the environment to guide behavior. Yet, how neural circuits integrate multiple, potentially conflicting, inputs during decision-making remains poorly understood. Here, we use larval zebrafish to address this question, leveraging their robust optomotor response to coherent random dot motion and phototaxis towards light. We demonstrate that animals employ an additive behavioral algorithm of three visual features: motion coherence, luminance level, and changes in luminance. Using brain-wide two-photon imaging, we identify the loci of these computations, with the anterior hindbrain emerging as a multifeature sensory integration hub. Through single-cell neurotransmitter and morphological analyses of functionally identified neurons, we characterize potential connections within and across computational nodes. These experiments reveal three parallel and converging pathways, matching our behavioral results. Our study provides a mechanistic brain-wide account of how a vertebrate brain integrates multiple features to drive sensorimotor decisions, bridging the algorithmic bases of behavior and its neural implementation.

Brood care relies on interactions between parents and offspring. Emergence of nestlings from their nest has been hypothesized to rely on the readout by the parent of the maturational state of the young. Theoretical considerations predict a conflict: parents should push for early emergence, if possible, to reduce care demands and maximize the number of reproductive cycles, whereas offspring should delay leaving to maximize resource allocation and protection by the parents. We tested this prediction in Lamprologus ocellatus, a shell-dwelling cichlid from Lake Tanganyika. We developed a laboratory paradigm to investigate the factors influencing emergence from the shell and found that mothers ensure their young stay inside the nest until 9 days after egg laying. Emergence coincides with an inversion of larval phototactic tendency from dark-seeking to light-seeking behavior on day 9. When we experimentally created a timing conflict by introducing older larvae to a foster mother, the mother resisted the (subjectively) premature emergence of her adopted fry. Removing the mother did not alter the larval intrinsic schedule, provided fresh water was supplied inside the shell. These findings suggest that, in L. ocellatus brood care, maternal and offspring behavior is normally synchronized by independent timing mechanisms. Our findings highlight the intricate coordination of parental and offspring behavior, offering insights into the evolutionary pressures shaping brood care in cichlids and challenging the traditional view of parent-offspring conflict over emergence timing.

Many vertebrates, including fish, amphibians, and reptiles, dynamically adjust their body color to the perceived brightness of the ambient background. This response takes tens of minutes and involves the aggregation or dispersion of pigment granules within the melanophores of the skin, resulting in the pale or dark appearance of the animal, respectively.1 In teleosts, ambient light detection depends on the retina, which transmits the signal to the hypothalamus, leading to the secretion of peptide hormones via the pituitary gland into the bloodstream. Melanin-concentrating hormone (MCH) is released upon light stimulation, resulting in the blanching of the skin. Alpha-melanocyte-stimulating hormone (α-MSH, encoded by the pomca gene and produced in the hypothalamus and pituitary) has the opposite effect: light inhibits its secretion, leading to melanin dispersion in the skin.1,2,3,4,5 Because "dark appearance" is an easy-to-score phenotype in larval zebrafish, defective visual background adaptation (VBA) has been used in genetic screens as a proxy for retinal defects,6,7 but the responsible neuroendocrine circuit remained elusive. Here, we identified the molecular and cellular components underlying this response. We found that intrinsically photosensitive retinal ganglion cells (ipRGCs), expressing the homeobox transcription factor Onecut1, project to the neuropil region of the preoptic area, where their axons overlap with the dendrites of pmchl-expressing hypothalamic neurons. Signals from these ipRGCs increase the transcription of pmchl, one of two genes encoding MCH isoforms, and repress pomca. Ablation of either onecut1-positive ipRGCs or pmchl-expressing hypothalamic neurons prevents the fish larva from adapting to a bright background.

Since the advent of optogenetics, great progress has been made in developing tools to modulate and detect cellular activity using light. We present a two-component optogenetic silencing tool, RoCK (rhodopsin cyclase/K+ channel), which pairs the rhodopsin-guanylyl cyclase CaRhGC with customized SthK K+ channels that are engineered to open selectively upon guanosine 3',5'-monophosphate (cGMP) binding. By enhancing the cGMP sensitivity and open probability of SthK mutants, we obtained four channel variants suited for different levels of cGMP concentration. CaRhGC's membrane-bound nature enables localized cGMP production, and the lack of dark activity reduces the risk for off-target effects. Optimized RoCK effectively modulated cellular activity in mouse hippocampal neurons, in acute hippocampal slices, and in rabbit cardiomyocytes. In zebrafish, RoCK silenced motor neurons in vivo, suppressing the characteristic coiling behavior of embryos, thus highlighting its potential for behavioral studies. In summary, RoCK expands our optogenetic toolkit threefold for fast cGMP production, fast cGMP sensing, and K+-based cell silencing.

The brain is spatially organized into subdivisions, nuclei and areas, which often correspond to functional and developmental units. A segmentation of brain regions in the form of a consensus atlas facilitates mechanistic studies and is a prerequisite for sharing information among neuroanatomists. Gene expression patterns objectively delineate boundaries between brain regions and provide information about their developmental and evolutionary histories. To generate a detailed molecular map of the larval zebrafish diencephalon, we took advantage of the Max Planck Zebrafish Brain (mapzebrain) atlas, which aligns hundreds of transcript and transgene expression patterns in a shared coordinate system. Inspection and co-visualization of close to 50 marker genes have allowed us to resolve the tripartite prosomeric scaffold of the diencephalon at unprecedented resolution. This approach clarified the genoarchitectonic partitioning of the alar diencephalon into pretectum (alar part of prosomere P1), thalamus (alar part of prosomere P2, with habenula and pineal complex), and prethalamus (alar part of prosomere P3). We further identified the region of the nucleus of the medial longitudinal fasciculus, as well as the posterior and anterior parts of the posterior tuberculum, as molecularly distinct basal parts of prosomeres 1, 2, and 3, respectively. Some of the markers examined allowed us to locate glutamatergic, GABAergic, dopaminergic, serotoninergic, and various neuropeptidergic domains in the larval zebrafish diencephalon. Our molecular neuroanatomical approach has thus (1) yielded an objective and internally consistent interpretation of the prosomere boundaries within the zebrafish forebrain; has (2) produced a list of markers, which in sparse combinations label the subdivisions of the diencephalon; and is (3) setting the stage for further functional and developmental studies in this vertebrate brain.

Vertebrates rely on rod photoreceptors for vision in low-light conditions. Mammals have a specialized downstream circuit for rod signaling called the primary rod pathway, which comprises specific cell types and wiring patterns that are thought to be unique to this lineage. Thus, it has been long assumed that the primary rod pathway evolved in mammals. Here, we challenge this view by demonstrating that the mammalian primary rod pathway is conserved in zebrafish, which diverged from extant mammals ~400 million years ago. Using single-cell RNA-sequencing, we identified two bipolar cell (BC) types in zebrafish that are related to mammalian rod BCs (RBCs) of the primary rod pathway. By combining electrophysiology, histology, and ultrastructural reconstruction of the zebrafish RBCs, we found that, like mammalian RBCs, both zebrafish RBC types connect with all rods in their dendritic territory, and provide output largely onto amacrine cells. The wiring pattern of the amacrine cells post-synaptic to one RBC type is strikingly similar to that of mammalian RBCs, suggesting that the cell types and circuit design of the primary rod pathway have emerged before the divergence of teleost fish and amniotes. The second RBC type, which forms separate pathways, is either lost in mammals or emerged in fish.

The vertebrate serotonergic system plays a critical role in modulating adaptive behavior. Yet, it has been challenging to unravel the downstream targets and the effects of serotonin on ongoing neural dynamics due to its widespread innervation and the complex nature of receptor signaling. Here, we show that the serotonergic system controls brain-wide neural dynamics in a spatially dualistic manner, global and compartmentalized, during motor adaptation behavior in zebrafish. Larval zebrafish adapt the vigor of tail motions depending on environmental drag force during visual pursuit behavior in a serotonin-dependent manner. Whole-brain imaging of serotonin release and systematic spatial mapping of serotonin receptors showed highly compartmentalized patterns that span multiple brain areas. Interestingly, whole-brain neural activity imaging combined with the perturbation of tph2+ raphe serotonin neurons revealed dualistic modulation of neural activity depending on behavioral encoding: global suppression of locomotor networks and the compartmentalized enhancement of midbrain sensory networks, both of which synergistically enabled motor adaptation. The compartmentalized modulation resulted from local serotonin release and receptor expression, while the global effect was due to modulation of a key network hub that broadcasts behavioral state signals. Our results reveal how the serotonergic system interacts with brain-wide neural dynamics through its parallel interactions and provide a conceptual framework for understanding the neural mechanisms of widespread serotonergic behavioral control. ### Competing Interest Statement The authors have declared no competing interest.

Animals constantly need to judge the valence of an object in their environment: is it potential food or a threat? The brain makes fundamental decisions on the appropriate behavioral strategy by integrating external information from sensory organs and internal signals related to physiological needs. For example, a hungry animal may take more risks than a satiated one when deciding to approach or avoid an object. Using a proteomic profiling approach, we identified the Calmodulin-interacting peptide Pcp4a as a key regulator of foraging-related decisions. Food intake reduced abundance of protein and mRNA of pcp4a via dopamine D2-like receptor-mediated repression of adenylate cyclase. Accordingly, deleting the pcp4a gene made zebrafish larvae more risk averse in a binary decision assay. Strikingly, neurons in the tectum became less responsive to prey-like visual stimuli in pcp4a mutants, thus biasing the behavior toward avoidance. This study pinpoints a molecular mechanism modulating behavioral choice according to internal state.

Recordings of the physiological history of cells provide insights into biological processes, yet obtaining such recordings is a challenge. To address this, we introduce a method to record transient cellular events for later analysis. We designed proteins that become labeled in the presence of both a specific cellular activity and a fluorescent substrate. The recording period is set by the presence of the substrate, whereas the cellular activity controls the degree of the labeling. The use of distinguishable substrates enabled the recording of successive periods of activity. We recorded protein-protein interactions, G protein–coupled receptor activation, and increases in intracellular calcium. Recordings of elevated calcium levels allowed selections of cells from heterogeneous populations for transcriptomic analysis and tracking of neuronal activities in flies and zebrafish.

Neuronal phenotypic traits such as morphology, connectivity, and function are dictated, to a large extent, by a specific combination of differentially expressed genes. Clusters of neurons in transcriptomic space correspond to distinct cell types and in some cases (e. g., C. elegans neurons[1][1] and retinal ganglion cells[2][2]–[4][3]) have been shown to share morphology and function. The zebrafish optic tectum is composed of a spatial array of neurons that transforms visual inputs into motor outputs. While the visuotopic map is continuous, subregions of the tectum are functionally specialized[5][4],[6][5]. To uncover the cell-type architecture of the tectum, we transcriptionally profiled its neurons, revealing more than 60 cell types that are organized in distinct anatomical layers. We then measured the visual responses of thousands of tectal neurons by two-photon calcium imaging and matched them with their transcriptional profile. Furthermore, we characterized the morphologies of transcriptionally identified neurons using specific transgenic lines. Surprisingly, we found that neurons that are transcriptionally similar can diverge functionally and morphologically. Incorporating the spatial coordinates of neurons within the tectal volume revealed functionally and morphologically defined anatomical subclusters within individual transcriptomic clusters. Our findings demonstrate that extrinsic, position-dependent factors expand the phenotypic repertoire of genetically similar neurons. ### Competing Interest Statement The authors have declared no competing interest. [1]: #ref-1 [2]: #ref-2 [3]: #ref-4 [4]: #ref-5 [5]: #ref-6

The zebrafish visual system has become a paradigmatic preparation for behavioral and systems neuroscience. Around 40 types of retinal ganglion cells (RGCs) serve as matched filters for stimulus features, including light, optic flow, prey, and objects on a collision course. RGCs distribute their signals via axon collaterals to 12 retinorecipient areas in forebrain and midbrain. The major visuomotor hub, the optic tectum, harbors nine RGC input layers that combine information on multiple features. The retinotopic map in the tectum is locally adapted to visual scene statistics and visual subfield-specific behavioral demands. Tectal projections to premotor centers are topographically organized according to behavioral commands. The known connectivity in more than 20 processing streams allows us to dissect the cellular basis of elementary perceptual and cognitive functions. Visually evoked responses, such as prey capture or loom avoidance, are controlled by dedicated multistation pathways that-at least in the larva-resemble labeled lines. This architecture serves the neuronal code's purpose of driving adaptive behavior.