Biology subjects

Clark, D. A.

Publications and source records attributed to Clark, D. A..

At least 19 recordsLinked to original sources

bioRxiv2026

Heavy-tailed connections support robust neural communication

In the synaptic wiring between neurons, some rare connections are orders of magnitude stronger than most. This heavy-tailed connectivity is ubiquitous across different species, brain regions, and cell types, suggesting that it serves a general purpose for neural function. Here, we show, both analytically and in simulations, that heavy-tailed connections reduce communication errors in the presence of synaptic noise. In minimal models of individual neurons, noisy connections induce errors at a rate that depends only on a simple network statistic, which we refer to as robustness. For a fixed number of incoming synapses, neurons with broader distributions of connection strength are more robust and, therefore, admit fewer errors. Across several connectomes, we find that real neurons are consistently more robust than comparable null networks. In the whole brain of the fruit fly Drosophila, robustness increases with distance from the sensory periphery, such that central neurons are more robust than those involved in sensory processing. Together, these results indicate that heavy-tailed connections may arise universally to fortify the brain against noise in communication.

neuroscience↗

bioRxiv2026

Haplotype and diversity signatures of ultra-soft selective sweeps in HIV-1

Many urgent medical and agricultural challenges are driven by resistance evolution via soft selective sweeps of multiple simultaneous mutations. Standard approaches to detect these mutations involve genome scans for regions with reduced diversity and increased haplotype lengths. However, it is unknown the extent to which those signatures persist as the number of mutations driving resistance grows. Here, we analyzed longitudinal linkage-resolved data from 10 intra-host HIV populations treated with broadly neutralizing antibody 10-1074. We found that HIV escapes 10-1074 with minimal perturbations to diversity and haplotype homozygosity in the region surrounding the sweep in the majority (8/10) of treated individuals. We matched these in vivo escape trajectories to forward simulations and found that adaptive mutations conferring escape must have been present on 20 or more genetic backgrounds to generate these signatures. These "ultra-soft" sweep signatures more closely resemble genetic patterns in a treatment non-responder without an adaptive response to 10-1074 than those of two other trial participants where adaptation occurred via harder selective sweeps. Our results demonstrate that HIV can adapt to a broadly neutralizing antibody treatment while retaining nearly all of its standing genetic diversity and that selection scans dependent on regional diversity and haplotype homozygosity signatures fail in this "ultra-soft" regime.

evolutionary biology↗

bioRxiv2026

A neural circuit for olfactory motion detection

Movement is a defining feature of dynamic environments and a key variable encoded by sensory systems. In vision, neural mechanisms that extract motion direction have been studied extensively across species, revealing circuits that compare signals across space and time. Odor plumes likewise carry motion information in their spatiotemporal structure, which walking fruit flies detect and exploit to improve navigation (Brudner et al., 2025; Kadakia et al., 2022), independent of wind sensing. However, how olfactory circuits compute odor motion remains unknown. Here we identify circuitry in the Drosophila antennal lobe that transforms bilateral odor inputs into direction-selective signals that are propagated to higher brain regions. These direction-selective signals were widespread across most measured olfactory channels, spanning pheromones, food odors, and aversive cues. Directional responses were elicited by odor traces with naturalistic temporal statistics and tuned to inter-antennal delays of {approx}40 ms. Connectomic, physiological, and behav-ioral analyses identified one GABAergic inhibitory neuron in the antennal lobe as a key mediator of direction selectivity. Measurements of dynamics in direction-selective neurons revealed ipsilateral excitation and delayed contralateral inhibition. In a minimal data-driven model, this mechanism was sufficient to generate the observed direction-selective responses. These results establish motion detection as a circuit-level olfactory computation and reveal how an early sensory circuit transforms spatiotemporal chemical signals into a neural representation of odor motion to guide navigation.

neuroscience↗

bioRxiv2026

Dynamic trajectory cues drive sequenced integration in approach detectors

Detecting approaching objects is essential for survival, and studies across animals have identified neurons and circuits selective for objects that grow in size over time, a hallmark of visual approach. However, approach as a physical process generates a rich ensemble of correlated cues governed by geometry and physics, of which size expansion is just one, and how the brain incorporates this broader cue structure to detect approach remains unknown. Here, we show that a visual stimulus without size expansion but with modulated luminance, a cue intrinsic to approach, elicits compelling percepts of approach and retreat in both humans and the fruit fly Drosophila, revealing an unsuspected, cross-species percept of approach based on luminance alone. Using targeted genetic silencing and two-photon calcium imaging in flies, we identify neurons that mediate evasive responses to both expansion-based and luminance-based approach, establishing them as general approach detectors. Guided by the geometry of approach, we show that luminance cues should precede detectable expansion signals during naturalistic approach. Consistent with this, these neurons combine luminance and expansion cues synergistically, but only when luminance changes precede expansion, matching their natural temporal order, thus performing sequenced cue integration. Together, these findings anchor a cross-species perceptual phenomenon to a defined circuit computation and reveal how neural circuits are tuned to the dynamic cue structure of natural events.

neuroscience↗

bioRxiv2024

Visual circuitry for distance estimation in Drosophila

Animals must infer the three-dimensional structure of their environment from two-dimensional images on their retinas. In particular, visual cues like motion parallax and binocular disparity can be used to judge distances to objects. Studies across several animal models have found neural signals that correlate with visual distance, but the causal role of these neurons in distance estimation as well as the range of possible neural properties that can inform distance estimation have remained poorly understood. Here, we developed a novel high-throughput behavioral assay to identify neurons in the Drosophila visual system that are involved in distance estimation during free locomotion. We found that silencing the primary motion detectors in the fly visual system eliminated their ability to perceive distance, consistent with a reliance on motion parallax to judge distance. Through a targeted silencing screen of visual neurons during behavior and through in vivo two-photon microscopy, we identified a visual projection neuron that encodes the parallax signal in the relative motion of foreground and background. Interestingly, it differs from previously identified parallax-tuned neurons in its lack of direction selectivity both to moving bars and to moving backgrounds. This non-canonical tuning is interpretable in the context of parallax signals that the fly would likely encounter during naturalistic walking behavior. Our results demonstrate how both direction selective and non-direction selective feature-detecting neurons can contribute to distance estimation using parallax cues, providing a framework for considering broader classes of parallax-encoding neurons in distance estimation across visual systems.

neuroscience↗

bioRxiv2024

Spatial and morphological organization of mitochondria across a connectome

Neuronal function depends critically on the cell biological organization of mitochondria, which regulate calcium signals and produce energy, among other roles. However, little is known about how mitochondria are organized within the circuits of neurons that make up each brain. To uncover the systematic rules that govern mitochondria shape and position in a connectome, we analyzed the morphological and spatial organization of more than 100,000 mitochondria in over 1,000 visual projection neurons in the Drosophila connectome. We found that mitochondrial shape and size differ systematically between cell types, and are distinct enough between cell types to serve as an identifying fingerprint. Moreover, we derived three quantitative rules that describe how mitochondria are positioned within neurons relative to synapses and other subcellular features: (1) they are positioned with a precision of 2-3 microns; (2) their relative preference for pre- and postsynaptic sites and other subcellular features differs between axons and dendrites; (3) their positions were specialized to different cell types. These organizing rules correlated with functional and anatomical properties of the cells, including visual responses and input connectivity. We also find that, in the flys olfactory associative learning circuits, mitochondria are enriched at presynapses to particular postsynaptic cells by accumulating in functional sub-compartments of axons. Overall, our findings reveal a robust set of organizing principles for mitochondria within and between cells, uncovering cell biology that maps onto the organization of the connectome and adding new dimensions for understanding circuit function in the connectome.

neuroscience↗

bioRxiv2024

Humans use local spectrotemporal correlations to detect rising and falling pitch

To discern speech or appreciate music, the human auditory system detects how pitch increases or decreases over time. However, the algorithms used to detect changes in pitch, or pitch motion, are incompletely understood. Here, using psychophysics, computational modeling, functional neuroimaging, and analysis of recorded speech, we ask if humans can detect pitch motion using computations analogous to those used by the visual system. We adapted stimuli from studies of vision to create novel auditory correlated noise stimuli that elicited robust pitch motion percepts. Crucially, these stimuli are inharmonic and possess no persistent features across frequency or time, but do possess positive or negative local spectrotemporal correlations in intensity. In psychophysical experiments, we found clear evidence that humans can judge pitch direction based only on positive or negative spectrotemporal intensity correlations. The key behavioral result--robust sensitivity to the negative spectrotemporal correlations--is a direct analogue of illusory "reverse-phi" motion in vision, and thus constitutes a new auditory illusion. Our behavioral results and computational modeling led us to hypothesize that human auditory processing may employ pitch direction opponency. fMRI measurements in auditory cortex supported this hypothesis. To link our psychophysical findings to real-world pitch perception, we analyzed recordings of English and Mandarin speech and found that pitch direction was robustly signaled by both positive and negative spectrotemporal correlations, suggesting that sensitivity to both types of correlations confers ecological benefits. Overall, this work reveals how motion detection algorithms sensitive to local correlations are deployed by the central nervous system across disparate modalities (vision and audition) and dimensions (space and frequency).

neuroscience↗

bioRxiv2024

Broken time reversal symmetry in visual motion detection

Our intuition suggests that when a movie is played in reverse, our perception of motion in the reversed movie will be perfectly inverted compared to the original. This intuition is also reflected in many classical theoretical and practical models of motion detection. However, here we demonstrate that this symmetry of motion perception upon time reversal is often broken in real visual systems. In this work, we designed a set of visual stimuli to investigate how stimulus symmetries affect time reversal symmetry breaking in the fruit fly Drosophilas well-studied optomotor rotation behavior. We discovered a suite of new stimuli with a wide variety of different properties that can lead to broken time reversal symmetries in fly behavioral responses. We then trained neural network models to predict the velocity of scenes with both natural and artificial contrast distributions. Training with naturalistic contrast distributions yielded models that break time reversal symmetry, even when the training data was time reversal symmetric. We show analytically and numerically that the breaking of time reversal symmetry in the model responses can arise from contrast asymmetry in the training data, but can also arise from other features of the contrast distribution. Furthermore, shallower neural network models can exhibit stronger symmetry breaking than deeper ones, suggesting that less flexible neural networks promote some forms of time reversal symmetry breaking. Overall, these results reveal a surprising feature of biological motion detectors and suggest that it could arise from constrained optimization in natural environments. SignificanceIn neuroscience, symmetries can tell us about the computations being performed by a circuit. In vision, for instance, one might expect that when a movie is played backward, ones motion percepts should all be reversed. Exact perceptual reversal would indicate a time reversal symmetry, but surprisingly, real visual systems break this symmetry. In this research, we designed visual stimuli to probe different symmetries in motion detection and identify features that lead to symmetry breaking in motion percepts. We discovered that symmetry breaking in motion detection depends strongly on both the detectors architecture and how it is optimized. Interestingly, we find analytically and in simulations that time reversal symmetries are broken in systems optimized to perform with natural inputs.

neuroscience↗

bioRxiv2024

Adaptation to visual sparsity enhances responses to infrequent stimuli

Sensory systems adapt their response properties to the statistics of their inputs. For instance, visual systems adapt to low-order statistics like mean and variance to encode the stimulus efficiently or to facilitate specific downstream computations. However, it remains unclear how other statistical features affect sensory adaptation. Here, we explore how Drosophilas visual motion circuits adapt to stimulus sparsity, a measure of the signals intermittency not captured by low-order statistics alone. Early visual neurons in both ON and OFF pathways alter their responses dramatically with stimulus sparsity, responding positively to both light and dark sparse stimuli but linearly to dense stimuli. These changes extend to downstream ON and OFF direction-selective neurons, which are activated by sparse stimuli of both polarities, but respond with opposite signs to light and dark regions of dense stimuli. Thus, sparse stimuli activate both ON and OFF pathways, recruiting a larger fraction of the circuit and potentially enhancing the salience of infrequent stimuli. Overall, our results reveal visual response properties that increase the fraction of the circuit responding to sparse, infrequent stimuli.

neuroscience↗

bioRxiv2023

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.

neuroscience↗

bioRxiv2023

Drosophila integrates visual evidence and counterevidence in self motion 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 confuse the movement of external objects with genuine self motion. Here, we show that stationary patterns on the retina, which constitute negative evidence against self rotation, are used by the fruit fly Drosophila to suppress inappropriate stabilizing rotational behavior. In silico experiments show that artificial neural networks optimized to distinguish self and world 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 flys motion- and optic flow-detectors. Our results exemplify how the compact brain of the fly incorporates negative evidence to improve heading stability, exploiting geometrical constraints of the visual world.

neuroscience↗

bioRxiv2022

Identifying inputs to visual projection neurons in Drosophila lobula by analyzing connectomic data

Electron microscopy-based connectomes provide important insights into how visual circuitry of fruit fly Drosophila computes various visual features, guiding and complementing behavioral and physiological studies. However, connectomic analyses of lobula, a putative center of object-like feature detection, remains underdeveloped, largely because of incomplete data on the inputs to the brain region. Here, we attempted to map the columnar inputs into the Drosophila lobula neuropil by performing connectivity- and morphology-based clustering on a densely reconstructed connectome dataset. While the dataset mostly lacked visual neuropils other than lobula, which would normally help identify inputs to lobula, our clustering analysis successfully extracted clusters of cells with homogeneous connectivity and morphology, likely representing genuine cell types. We were able to draw a correspondence between the resulting clusters and previously identified cell types, revealing previously undocumented connectivity between lobula input and output neurons. While future, more complete connectomic reconstructions are necessary to verify the results presented here, they can serve as a useful basis for formulating hypotheses on mechanisms of visual feature detection in lobula.

neuroscience↗

bioRxiv2021

A seasonal pulse of ungulate neonates influences space use by carnivores in a multi-predator, multi-prey system

O_LIThe behavioral mechanisms by which predators encounter prey are poorly resolved. In particular, the extent to which predators engage in active search for prey versus incidentally encountering them is unknown. The distinction between search and incidental encounter influences prey population dynamics with active search exerting a stabilizing force on prey populations by alleviating predation pressure on low-density prey and increasing it for high-density prey. C_LIO_LIParturition of many large herbivores occurs during a short and predictable temporal window in which young are highly vulnerable to predation. Our study aims to determine how a suite of carnivores responds to the seasonal pulse of newborn ungulates using contemporaneous GPS locations of four species of predators and two species of prey. C_LIO_LIWe used step-selection functions to assess whether coyotes, cougars, black bears, and bobcats actively searched for parturient females in a low-density population of mule deer and a high-density population of elk. We then assessed whether searching carnivores shifted their habitat use toward areas exhibiting a high probability of encountering neonates. C_LIO_LINone of the four carnivore species encountered parturient mule deer more often than expected by chance suggesting that predation of young resulted from incidental encounters. By contrast, we determined that cougar and male bear movements positioned them in proximity of parturient elk more often than expected by chance which is evidence of searching behavior. Although both male bears and cougars searched for neonates, only male bears used elk parturition habitat in a way that dynamically tracked the phenology of the elk birth pulse suggesting that maximizing encounters with juvenile elk was a motivation when selecting resources. C_LIO_LIOur results support the existence of a stabilizing mechanism to prey populations through active search behavior by predators because carnivores in our study searched for the high-density prey species (elk) but ignored the low-density species (mule deer). We conclude that prey density must be high enough to warrant active search, and that there is high interspecific and intersexual variability in foraging strategies among large mammalian predators and their prey. C_LI

ecology↗

bioRxiv2021

Neural mechanisms to exploit positional geometry for collision avoidance

Visual motion provides rich geometrical cues about the three-dimensional configuration 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.

neuroscience↗

bioRxiv2021

Manipulating neural dynamics to tune motion detection

Neurons integrate excitatory and inhibitory signals to produce their outputs, but the role of input timing in this integration remains poorly understood. Motion detection is a paradigmatic example of this integration, since theories of motion detection rely on different delays in visual signals. These delays allow circuits to compare scenes at different times to calculate the direction and speed of motion. It remains untested how response dynamics of individual cell types drive motion detection and velocity sensitivity. Here, we sped up or slowed down specific neuron types in Drosophilas motion detection circuit by manipulating ion channel expression. Altering the dynamics of individual neurons upstream of motion detectors changed their integrating properties and increased their sensitivity to fast or slow visual motion, exposing distinct roles for dynamics in tuning directional signals. A circuit model constrained by data and anatomy reproduced the observed tuning changes. Together, these results reveal how excitatory and inhibitory dynamics jointly tune a canonical circuit computation.

neuroscience↗

bioRxiv2021

Intelligent Resolution: Integrating Cryo-EM with AI-driven Multi-resolution Simulations to Observe the SARS-CoV-2 Replication-Transcription Machinery in Action

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) replication transcription complex (RTC) is a multi-domain protein responsible for replicating and transcribing the viral mRNA inside a human cell. Attacking RTC function with pharmaceutical compounds is a pathway to treating COVID-19. Conventional tools, e.g., cryo-electron microscopy and all-atom molecular dynamics (AAMD), do not provide sufficiently high resolution or timescale to capture important dynamics of this molecular machine. Consequently, we develop an innovative workflow that bridges the gap between these resolutions, using mesoscale fluctuating finite element analysis (FFEA) continuum simulations and a hierarchy of AI-methods that continually learn and infer features for maintaining consistency between AAMD and FFEA simulations. We leverage a multi-site distributed workflow manager to orchestrate AI, FFEA, and AAMD jobs, providing optimal resource utilization across HPC centers. Our study provides unprecedented access to study the SARS-CoV-2 RTC machinery, while providing general capability for AI-enabled multi-resolution simulations at scale.

biophysics↗

bioRxiv2021

Odor motion sensing enables complex plume navigation

Studies dating back a century (Flugge, 1934) have stressed the critical role of the wind as the primary directional cue in odor plume navigation. Here, we show that Drosophila shape their navigational decisions using a second directional cue - the direction of motion of odors - which they detect from the temporal correlations of the odor signal between their two antennae. Using a high-resolution virtual reality paradigm to deliver spatiotemporally complex fictive odors to freely-walking flies, we demonstrate that such odor direction sensing is computationally equivalent to motion detection algorithms underlying motion detection in vision. Simulations and theoretical analysis of turbulent plumes reveal that odor motion contains valuable directional information absent from the airflow; indeed, this information is used by both Drosophila and virtual agents to navigate naturalistic odor environments. The generality of our findings suggests that odor direction sensing is likely used throughout the animal kingdom, and could significantly improve olfactory robot navigation in harsh chemical environments.

neuroscience↗

bioRxiv2021

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 an anatomically-constrained shallow neural network model and train it 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. The LPLC2-like solutions are favored when a population of units is trained on the task, but not when units are trained in isolation. The trained model 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. These results show that LPLC2 neuron properties and tuning are predicted by optimizing an anatomically-constrained neural network to detect impending collisions.

neuroscience↗