Biology subjects

Couzin, I. D.

Publications and source records attributed to Couzin, I. D..

At least 19 recordsLinked to original sources

bioRxiv2026

Hebbian learning generates individuality in spatial decision-making

Individuals exposed to similar environments often develop consistent behavioral differences, yet how learning generates individuality remains unclear. We study agents whose neural connections develop through unsupervised Hebbian learning during pursuit of a moving target. All agents begin with identical networks without recurrent connections and neurons tuned to world-fixed directions. The target coactivates neurons with overlapping directional receptive fields while leaving more distant neurons relatively inactive. Hebbian learning converts this activity pattern into local excitation and longer-range inhibition. Stochastic differences in target motion and neural activity during learning produce individual-specific departures from each network's average center-surround profile. We then freeze the connections and repeatedly test agents with two equally attractive targets, a task absent from training. Individual agents tend to spend more time near one target than the other, with preferences differing among agents but remaining repeatable across trials. Replacing each network with its average connectivity profile strongly reduces preference strength and leaves little repeatability. Rearranging which connections are stronger or weaker than average, while preserving the amount of variation and average profile, retains repeatable preferences and improves tracking in a separate moving-target test. However, preferences confer no general advantage when targets differ in value. Under standard sensory and inhibitory settings, repeatable preferences and accurate choices coexist only at intermediate neural noise levels tested. A matched egocentric model, with tuning relative to the agent's heading, shows no detectable target-side repeatability. Hebbian learning can therefore generate individual spatial preferences, whose expression depends on how the network represents space.

neuroscience↗

bioRxiv2026

Using semi-folded arenas to observe spontaneous emergence of collective trails in whole colonies of termites

Collective movement in social organisms emerges from local interactions and can generate large-scale spatial patterns of ecological relevance. In termites, trail formation is a well-known collective phenomenon, yet reproducing and recording its emergence under controlled laboratory conditions using whole colonies remains challenging. Existing laboratory approaches often rely on confined arenas or manually assembled subgroups, which can restrict movement and limit observation of colony-level dynamics. Here, we present a semi-folded arena designed for whole-colony observation of termite movement under controlled conditions. We developed a circular semi-folded arena that remained continuously connected to an intact nest and allowed individuals to move across a central observation surface while recirculating through a folded peripheral section. Using whole colonies of the Neotropical termite Constrictotermes cyphergaster, we recorded exploratory activity under baseline conditions, in the absence of added food or water. High-resolution video recordings were analysed using automated movement extraction to recover trajectories and visualise collective trail structure. Within the first 6 min of activity, collective trail structure was observed in 15 of the 16 colonies analysed. Under these conditions, the semi-folded setup captured early collective trail structure, visible as convergence of cumulative trajectories along shared routes radiating from the arena entrance region. Automated movement extraction was compatible with dense whole-colony recordings and yielded large quantities of positional data during the initial observation interval. Descriptive trajectory-based outputs, including speed distributions for workers and soldiers, showed that the recordings were suitable for recovery of fine-scale movement information. Repeatedly used routes were also often marked by visible dark traces on the paper lining by the end of the observations, providing a qualitative record of cumulative route use. The semi-folded arena provides a practical method for recording whole-colony termite movement under laboratory conditions while maintaining continuous nest access and avoiding manual transfer of individuals during trials. Rather than replacing conventional arena designs, this approach offers an additional methodological option for studying emergent movement patterns in species for which whole-colony observation is feasible. More broadly, it expands the experimental toolkit available for investigating colony-scale spatial organisation under controlled conditions.

ecology↗

bioRxiv2025

FERAL: A Video-Understanding System for Direct Video-to-Behavior Mapping

Quantifying animal behavior often requires segmenting continuous actions into discrete, interpretable states, yet most automated pipelines infer actions from keypoint dynamics and are limited by keypoint tracking quality. Here we present FERAL (Feature Extraction for Recognition of Animal Locomotion), a supervised video-understanding toolkit that maps raw video directly to frame-level behavioral labels, bypassing the keypoint-extraction and pose-classification stages of conventional pipelines. Across benchmarks, FERAL matches or exceeds state-of-the-art pose- and video-based baselines. On the CalMS21 mouse social-interaction benchmark, it exceeds Googles VideoPrism using only a quarter of the training data. FERAL generalizes across species (apes, zebras, mice, ants, flies, and nematodes), recording conditions, and levels of organization, from single animals to social interactions and colony-scale collective behavior. Released as a user-friendly, open-source package, it integrates with existing analysis pipelines. By enabling scalable, species-agnostic behavioral quantification directly from raw video, FERAL broadens the experimental paradigms available for studying animal behavior.

animal behavior and cognition↗

bioRxiv2025

Allocentric Flocking

Understanding how group-level dynamics arise from individual interactions re- mains a core challenge in collective behavior research. Traditional models assume that animals follow simple behavioral rules, like explicitly aligning with neighbors, yet experimental support for such interactions is often lacking. Here we consider a model grounded in the neurobiological principles underlying animals navigational circuits, particularly the fact that animals encode their headings, and also bearings to objects (e.g., other individuals) in their environment, via a world-centered--allocentric--neural coding. We compare this to an egocentric representation, where bearings are encoded with respect to the arbitrary heading of the animal. An allocentric framework, as op- posed to an egocentric one, is shown to enable effective tracking of dynamically moving targets. Moreover, we demonstrate that when individuals themselves act as sensory inputs to each other, that sophisticated, coherent collective motion can emerge di- rectly from navigational circuits (and thus, may readily evolve in nature), without requiring explicit alignment, or additional rules of interaction.

ecology↗

bioRxiv2024

Spatiotemporal dynamics of locomotor decisions in Drosophila melanogaster

Decision-making in animals often involves choosing actions while navigating the environment, a process markedly different from static decision paradigms commonly studied in laboratory settings. Even in decision-making assays in which animals can freely locomote, decision outcomes are often interpreted as happening at single points in space and single moments in time, a simplification that potentially glosses over important spatiotemporal dynamics. We investigated locomotor decision-making in Drosophila melanogaster in Y-shaped mazes, measuring the extent to which their future choices could be predicted through space and time. We demonstrate that turn-decisions can be reliably predicted from flies locomotor dynamics, with distinct predictability phases emerging as flies progress through maze regions. We show that these predictability dynamics are not merely the result of maze geometry or wall-following tendencies, but instead reflect the capacity of flies to move in ways that depend on sustained locomotor signatures, suggesting an active, working memory-like process. Additionally, we demonstrate that fly mutants known to have sensory and information-processing deficits exhibit altered spatial predictability patterns, highlighting the role of visual, mechanosensory, and dopaminergic signaling in locomotor decision-making. Finally, highlighting the broad applicability of our analyses, we generalize our findings to other species and tasks. We show that human participants in a virtual Y-maze exhibited similar decision predictability dynamics as flies. This study advances our understanding of decision-making processes, emphasizing the importance of spatial and temporal dynamics of locomotor behavior in the lead-up to discrete choice outcomes.

animal behavior and cognition↗

bioRxiv2023

Honey bee drones are synchronously hyperactive inside the nest

1Eusocial insects operate as an integrated collective with tasks allocated among individuals. This applies also to reproduction, through coordinated mating flights between male and female reproductives. While in some species male sexuals take only a single mating flight and never return, in the Western honey bee, Apis mellifera, the male sexuals (drones) live in the colony throughout their lives. Prior research has focused almost exclusively on drone behavior outside of the nest (mating flights), while ignoring the majority of their life, which is spent inside the nest. To understand the in-nest behavior of drones across their lives, we used the BeesBook tracking system to track 192 individually-marked drones continuously for over 20 days, to examine how drones moved and spent time in the nest. In agreement with previous work, we find that drones spend most of their time immobile at the nest periphery. However, we also observe that drones have periods of in-nest hyperactivity, during which they become the most active individuals in the entire colony. This in-nest hyperactivity develops in drones after age 7 days, occurs daily in the afternoon, and coincides with drones taking outside trips. We find strong synchronization across the drones in the start/end of activity, such that the drones in the colony exhibit a "shared activation period". The duration of the shared activation period depends on the weather; when conditions are suitable for mating flights, the activation period is extended. At the individual-level, we see that the activation order changes from day to day, suggesting that both the external influence of weather conditions, as well as exchange of social information, influences individual activation. Using an accumulation-to-threshold model of drone activation, we show that simulations using social information match experimental observations. These results provide new insight into the in-nest behavior of drones, and how their behavior reflects their role as the male gametes of the colony.

animal behavior and cognition↗

bioRxiv2022

Multi-animal behavioral tracking and environmental reconstruction using drones and computer vision in the wild

O_LIMethods for collecting animal behavior data in natural environments, such as direct observation and bio-logging, are typically limited in spatiotemporal resolution, the number of animals that can be observed, and information about animals social and physical environments. C_LIO_LIVideo imagery can capture rich information about animals and their environments, but image-based approaches are often impractical due to the challenges of processing large and complex multi-image datasets and transforming resulting data, such as animals locations, into geographic coordinates. C_LIO_LIWe demonstrate a new system for studying behavior in the wild that uses drone-recorded videos and computer vision approaches to automatically track the location and body posture of free-roaming animals in georeferenced coordinates with high spatiotemporal resolution embedded in contemporaneous 3D landscape models of the surrounding area. C_LIO_LIWe provide two worked examples in which we apply this approach to videos of gelada monkeys and multiple species of group-living African ungulates. We demonstrate how to track multiple animals simultaneously, classify individuals by species and age-sex class, estimate individuals body postures (poses), and extract environmental features, including topography of the landscape and animal trails. C_LIO_LIBy quantifying animal movement and posture, while simultaneously reconstructing a detailed 3D model of the landscape, our approach opens the door to studying the sensory ecology and decision-making of animals within their natural physical and social environments. C_LI

animal behavior and cognition↗

bioRxiv2021

Both Prey and Predator Features Determine Predation Risk and Survival of Schooling Prey

Predation is one of the main evolutionary drivers of social grouping. While it is well appreciated that predation risk is likely not shared equally among individuals within groups, its detailed quantification has remained difficult due to the speed of attacks and the highly-dynamic nature of collective prey response. Here, using high-resolution tracking of solitary predators (Northern pike) hunting schooling fish (golden shiners), we not only provide detailed insights into predator decision-making but show which key spatial and kinematic features of predator and prey influence individuals risk to be targeted and survive attacks. Pike tended to stealthily approach the largest groups, and were often already inside the school when launching their attack, making prey in this frontal "strike zone" the most vulnerable to be targeted. From the preys perspective, those fish in central locations, but relatively far from, and less aligned with, neighbours, were most likely to be targeted. While the majority of attacks (70%) were successful, targeted individuals that did manage to avoid capture exhibited a higher maximum acceleration response just before the attack and were further away from the pike s head. Our results highlight the crucial interplay between predators attack strategy and response of prey in determining predation risk in mobile animal groups.

animal behavior and cognition↗

bioRxiv2021

The geometry of decision-making

Choosing among spatially-distributed options is a central challenge for animals, from deciding among alternative potential food sources or refuges, to choosing with whom to associate. Using an integrated theoretical and experimental approach (employing immersive virtual reality), we consider the interplay between movement and vectorial integration during decision-making regarding two, or more, options in space. In computational models of this process we reveal the occurrence of spontaneous and abrupt "critical" transitions (associated with specific geometrical relationships) whereby organisms spontaneously switch from averaging vectorial information among, to suddenly excluding one, among the remaining options. This bifurcation process repeats until only one option--the one ultimately selected--remains. Thus we predict that the brain repeatedly breaks multi-choice decisions into a series of binary decisions in space-time. Experiments with fruit flies, desert locusts, and larval zebrafish reveal that they exhibit these same bifurcations, demonstrating that across taxa and ecological context, there exist fundamental geometric principles that are essential to explain how, and why, animals move the way they do.

animal behavior and cognition↗

bioRxiv2021

The dominant axes of lifetime behavioral variation in honey bees

Insect colonies are decentralized systems that employ task allocation, whereby individuals undertake different roles to fulfil colony needs, such as honey bee "nurses", "nest workers", and "foragers". However, the extent to which individuals can be well-classified by discrete "roles", how they change behavior from day-to-day, over entire lifetimes, and with environmental conditions, is poorly understood. Using long-term automated tracking of over 4,200 individually-identified bees Apis mellifera, we use behavioral metrics to quantify and compare behavior. We show that individuals exhibit behavioral variation along two dominant axes that represent nest substrate use and movement within the nest. Across lifetimes, we find that individuals differ in foraging onset, and that certain bees exhibit lifelong consistencies in their movement patterns. Furthermore, we examine a period of sudden nectar availability where the honey stores tripled over 6 days, and see that the colony exhibits a distributed shift in activity that did not require a large-scale colony reorganization. Our quantitative approach shows how collective units differ over days and lifetimes, and how sources of variation and variability contribute to the colonys robust yet flexible response.

animal behavior and cognition↗

bioRxiv2021

Collective detection based on visual information in animal groups

The spatio-temporal distribution of individuals within a group (i.e its internal structure) plays a defining role in how individuals interact with their environment, make decisions, and transmit information via social interactions. Group-living organisms across taxa, including many species of fish, birds, ungulates, and insects, use vision as the predominant modality to coordinate their collective behavior. Despite this importance, there have been few quantitative studies examining visual detection capabilities of individuals within groups. We investigate key principles underlying individual, and collective, visual detection of stimuli (which could include cryptic predators, potential food items, etc.) and how this relates to the internal structure of groups. While the individual and collective detection principles are generally applicable, we employ a model experimental system of schooling golden shiner fish (Notemigonus crysoleucas) to relate theory directly to empirical data, using computational reconstruction of the visual fields of all individuals to do so. Our integrative approach allows us to reveal how the external visual information available to each group member depends on the number of individuals in the group, the position within the group, and the location of the external visually-detectable stimulus. We find that in small groups, individuals have detection capability in nearly all directions, while in large groups, occlusion by neighbors causes detection capability to vary with position within the group. We then formulate a simple, and generally applicable, model that captures how visual detection properties emerge due to geometric scaling of the space occupied by the group and occlusion caused by neighbors. We employ these insights to discuss principles that extend beyond our specific system, such as how collective detection depends on individual body shape, and the size and structure of the group.

animal behavior and cognition↗

bioRxiv2020

TRex, a fast multi-animal tracking system with markerless identification, 2D body posture estimation and visual field reconstruction

Automated visual tracking of animals is rapidly becoming an indispensable tool for the study of behavior. It offers a quantitative methodology by which organisms sensing and decision-making can be studied in a wide range of ecological contexts. Despite this, existing solutions tend to be challenging to deploy in practice, especially when considering long and/or high-resolution video-streams. Here, we present TRex, a fast and easy-to-use solution for tracking a large number of individuals simultaneously using background-subtraction with real-time (60Hz) tracking performance for up to approximately 256 individuals and estimates 2D visual-fields, outlines, and head/rear of bilateral animals, both in open and closed-loop contexts. Additionally, TRex offers highly-accurate, deep-learning-based visual identification of up to approximately 100 unmarked individuals, where it is between 2.5-46.7 times faster, and requires 2-10 times less memory, than comparable software (with relative performance increasing for more organisms/longer videos) and provides interactive data-exploration within an intuitive, platform-independent graphical user-interface.

animal behavior and cognition↗

bioRxiv2020

VAE-SNE: a deep generative model for simultaneous dimensionality reduction and clustering

Scientific datasets are growing rapidly in scale and complexity. Consequently, the task of understanding these data to answer scientific questions increasingly requires the use of compression algorithms that reduce dimensionality by combining correlated features and cluster similar observations to summarize large datasets. Here we introduce a method for both dimension reduction and clustering called VAE-SNE (variational autoencoder stochastic neighbor embedding). Our model combines elements from deep learning, probabilistic inference, and manifold learning to produce interpretable compressed representations while also readily scaling to tens-of-millions of observations. Unlike existing methods, VAE-SNE simultaneously compresses high-dimensional data and automatically learns a distribution of clusters within the data -- without the need to manually select the number of clusters. This naturally creates a multi-scale representation, which makes it straightforward to generate coarse-grained descriptions for large subsets of related observations and select specific regions of interest for further analysis. VAE-SNE can also quickly and easily embed new samples, detect outliers, and can be optimized with small batches of data, which makes it possible to compress datasets that are otherwise too large to fit into memory. We evaluate VAE-SNE as a general purpose method for dimensionality reduction by applying it to multiple real-world datasets and by comparing its performance with existing methods for dimensionality reduction. We find that VAE-SNE produces high-quality compressed representations with results that are on par with existing nonlinear dimensionality reduction algorithms. As a practical example, we demonstrate how the cluster distribution learned by VAE-SNE can be used for unsupervised action recognition to detect and classify repeated motifs of stereotyped behavior in high-dimensional timeseries data. Finally, we also introduce variants of VAE-SNE for embedding data in polar (spherical) coordinates and for embedding image data from raw pixels. VAE-SNE is a robust, feature-rich, and scalable method with broad applicability to a range of datasets in the life sciences and beyond.

animal behavior and cognition↗

bioRxiv2020

Social networks predict the life and death of honey bees

In complex societies, individuals roles are reflected by interactions with other conspecifics. Honey bees (Apis mellifera) generally change tasks as they age, but developmental trajectories of individuals can vary drastically due to physiological and environmental factors. We introduce a succinct descriptor of an individuals social network that can be obtained without interfering with the colony. This network age accurately predicts task allocation, survival, activity patterns, and future behavior. We analyze developmental trajectories of multiple cohorts of individuals in a natural setting and identify distinct developmental pathways and critical life changes. Our findings suggest a high stability in task allocation on an individual level. We show that our method is versatile and can extract different properties from social networks, opening up a broad range of future studies. Our approach highlights the relationship of social interactions and individual traits, and provides a scalable technique for understanding how complex social systems function.

animal behavior and cognition↗

bioRxiv2019

Schistocephalus parasite infection alters sticklebacks' movement ability and thereby shapes social interactions

Parasitism is ubiquitous in the animal kingdom. Although many fundamental aspects of host-parasite relationships have been unravelled, few studies have systematically investigated how parasites affect organismal movement. Here we combine behavioural experiments of Schistocephalus solidus infected sticklebacks with individual-based simulations to understand how parasitism affects individual movement ability and how this in turn influences social interaction patterns. Detailed movement tracking revealed that infected fish swam slower, accelerated slower, turned more slowly, and tended to be more predictable in their movements than did non-infected fish. Importantly, the strength of these effects increased with increasing parasite load (% of body weight), with the behaviour of more heavily infected fish being more impaired. When grouped, pairs of infected fish moved more slowly, were less cohesive, less aligned, and less coordinated than healthy pairs. Mixed pairs exhibited intermediate behaviours and were primarily led by the non-infected fish. These social patterns emerged naturally in model simulations of self-organised groups composed of individuals with different speeds and turning tendency, consistent with changes in mobility and manoeuvrability due to infection. Together, our results demonstrate how infection with a complex life cycle parasite affects the movement ability of individuals and how this in turn shapes social interactions, providing important mechanistic insights into the effects of parasites on host movement dynamics.

animal behavior and cognition↗

bioRxiv2019

Fast and robust animal pose estimation

Quantitative behavioral measurements are important for answering questions across scientific disciplines--from neuroscience to ecology. State-of-the-art deep-learning methods offer major advances in data quality and detail by allowing researchers to automatically estimate locations of an animals body parts directly from images or videos. However, currently-available animal pose estimation methods have limitations in speed and robustness. Here we introduce a new easy-to-use software toolkit, DeepPoseKit, that addresses these problems using an eZcient multi-scale deep-learning model, called Stacked DenseNet, and a fast GPU-based peak-detection algorithm for estimating keypoint locations with subpixel precision. These advances improve processing speed >2x with no loss in accuracy compared to currently-available methods. We demonstrate the versatility of our methods with multiple challenging animal pose estimation tasks in laboratory and field settings--including groups of interacting individuals. Our work reduces barriers to using advanced tools for measuring behavior and has broad applicability across the behavioral sciences.

animal behavior and cognition↗

bioRxiv2018

Counteracting estimation bias and social influence to improve the wisdom of crowds

Aggregating multiple non-expert opinions into a collective estimate can improve accuracy across many contexts. However, two sources of error can diminish collective wisdom: individual estimation biases and information sharing between individuals. Here we measure individual biases and social influence rules in multiple experiments involving hundreds of individuals performing a classic numerosity estimation task. We first investigate how existing aggregation methods, such as calculating the arithmetic mean or the median, are influenced by these sources of error. We show that the mean tends to overestimate, and the median underestimate, the true value for a wide range of numerosities. Quantifying estimation bias, and mapping individual bias to collective bias, allows us to develop and validate three new aggregation measures that effectively counter sources of collective estimation error. In addition, we present results from a further experiment that quantifies the social influence rules that individuals employ when incorporating personal estimates with social information. We show that the corrected mean is remarkably robust to social influence, retaining high accuracy in the presence or absence of social influence, across numerosities, and across different methods for averaging social information. Utilizing knowledge of estimation biases and social influence rules may therefore be an inexpensive and general strategy to improve the wisdom of crowds.

animal behavior and cognition↗

bioRxiv2017

Mobility can promote the evolution of cooperation via emergent self-assortment dynamics

The evolution of costly cooperation, where cooperators pay a personal cost to benefit others, requires that cooperators interact more frequently with other cooperators. This condition, called positive assortment, is known to occur in spatially-structured viscous populations, where individuals typically have low mobility and limited dispersal. However many social organisms across taxa, from cells and bacteria, to birds, fish and ungulates, are mobile, and live in populations with considerable inter-group mixing. In the absence of information regarding others traits or conditional strategies, such mixing may inhibit assortment and limit the potential for cooperation to evolve. Here we employ spatially-explicit individual-based evolutionary simulations to incorporate costs and benefits of two coevolving costly traits: cooperative and local cohesive tendencies. We demonstrate that, despite possessing no information about others traits or payoffs, mobility (via self-propulsion or environmental forcing) facilitates assortment of cooperators via a dynamically evolving difference in the cohesive tendencies of cooperators and defectors. We show analytically that this assortment can also be viewed in a multilevel selection framework, where selection for cooperation among emergent groups can overcome selection against cooperators within the groups. As a result of these dynamics, we find an oscillatory pattern of cooperation and defection that maintains cooperation even in the absence of well known mechanisms such as kin interactions, reciprocity, local dispersal or conditional strategies that require information on others strategies or payoffs. Our results offer insights into differential adhesion based mechanisms for positive assortment and reveal the possibility of cooperative aggregations in dynamic fission-fusion populations.\n\nAuthor SummaryCooperation among animals is ubiquitous. In a cooperative interaction, the cooperator confers a benefit to its partner at a personal cost. How does natural selection favour such a costly behaviour? Classical theories argue that cooperative interactions among genetic relatives, reciprocal cooperators, or among individuals within groups in viscous population structures are necessary to maintain cooperation. However, many organisms are mobile, and live in dynamic (fission-fusion) groups that constantly merge and split. In such populations, the above mechanisms may be inadequate to explain cooperation. Here, we develop a minimal model that explicitly accounts for mobility and cohesion among organisms. We find that mobility can support cooperation via emergent dynamic groups, even in the absence of previously known mechanisms. Our results may offer insights into the evolution of cooperation in animals that live in fission fusion groups, such as birds, fish or mammals, or microbes living in turbulent media, such as in oceans or in the bloodstreams of animal hosts.

evolutionary biology↗