Yarden Cohen

Monitoring the health and well-being of research animals is essential for both ethical and scientific purposes. In songbirds, body weight is one of the main indicators of their overall condition, yet traditional weighing methods can be intrusive and stress-inducing, which could decrease their song rate. We developed an automated system for monitoring the weight of multiple birds in longitudinal neuroscience experiments, which often include birds tethered to data acquisition systems. Building on previous models that were mostly designed for weighing birds outdoors and in large housing cages, our design serves as a perch for the bird to stand on (a perch-scale ) and meets the needs of neuroscience experiments by (i) minimizing cable entanglement to safely accommodate tethered birds, (ii) supporting long-term monitoring of individually-housed birds, and (iii) linking multiple devices into a unified control unit that oversees setup, calibration, and data acquisition. We deployed the system in the cages of six canaries for ten days and validated its accuracy against daily manual weighing. The precision and continuous monitoring of the perch-scale allowed observing physiological patterns such as overnight weight loss. Our system detected 22 sequences of overnight perching in five different birds, showing an average decrease of ≈ 4% of the bird’s body-weight overnight. We also found that daily weight estimates, derived from perch-scale data, were within the range of daily weight fluctuations (5–10%), as they deviated by less than 5% on average when compared to the manual weights. These results validate the device’s sensitivity for detecting subtle and health-related changes. By eliminating the need for manual handling of birds, this system offers a non-invasive, hands-free approach that reduces stress and improves the accuracy of health assessments. Future applications could integrate additional health metrics to provide a more comprehensive understanding of animal welfare in neurophysiology and behavioral studies.

Many complex behaviors involve sequences of basic motor or vocal elements governed by syntactic rules, which facilitate flexible and adaptive actions. Songbirds that crystallize their repertoire of vocal syllables and transitions allow researchers to build probabilistic models of syntax rules, providing insight into underlying neural mechanisms. However, in many complex behaviors, syntax rules change over time, such as during learning or in response to new environmental and social contexts. In this study, we investigated the songs of canaries, a seasonal songbird species. Canaries learn a repertoire of 30-50 syllable types, produce syllables in repeat phrases, and organize these phrases into sequences with long-range syntactic dependencies. Since canaries are known to adapt their repertoire annually, we recorded their songs during the spring mating season and examined the syntactic properties that determine syllable sequencing. Over days and weeks, we observed changes in syllable usage rates, syllable numbers within phrases, phrase positions in songs, and in the long-range dependencies of phrase transitions. Acoustic features of syllables were also found to shift alongside these syntactic changes. Quantifying the variability of these properties revealed that the observed changes were not random. Most birds exhibited a clear trend of moving between order and disorder in their song’s syntactic and acoustic features. Interestingly, this trend varied across individuals; some birds increased their stereotypy and decreased variability across days, while others adopted more disordered and variable song structures. These findings establish canaries as a valuable animal model for studying the neural mechanisms underlying syntax rules in complex motor sequences, their plasticity in social and environmental adaptation, and in implementing individual-specific strategies.

Monitoring the health and well-being of research animals is essential for both ethical and scientific purposes. In songbirds, body weight is one of the main indicators for their overall condition, yet traditional weighing methods can be intrusive and stress-inducing, which could decrease their song rate. We developed a novel, automated system designed to continuously monitor the weight of untethered and tethered birds without disrupting their natural behavior in neuroscience experiments. We used the system to track weight fluctuations in six canaries over several weeks, revealing physiological patterns such as overnight weight loss, with one bird losing approximately 5.17% of its body weight during a 9.5-hour period of inactivity. Our system’s high sensitivity detected weight changes below 1% of body mass, validating its reliability for long-term studies. Control experiments confirmed that weight fluctuations observed were physiological rather than due to equipment deviations. By eliminating the need for manual handling, this system offers a non-invasive, hands-free approach that reduces stress and improves the accuracy of health assessments. This study demonstrates the system’s potential for expanding research on how environmental factors, diet, and other variables influence bird physiology and behavior. Future applications could integrate additional health metrics, providing a more comprehensive understanding of animal welfare in neuroscience and behavioral studies.

Research on animal acoustic communication is being revolutionized by deep learning. In this talk we present vak, a framework that allows researchers in this area to easily benchmark deep neural network models and apply them to their own data. We'll demonstrate how research groups are using vak through examples with TweetyNet, a model that automates annotation of birdsong by segmenting spectrograms. Then we'll show how adopting Lightning as a backend in version 1.0 has allowed us to incorporate more models and features, building on the foundation we put in place with help from the scientific Python stack.

Songbirds provide a powerful model system for studying sensory-motor learning. However, many analyses of birdsong require time-consuming, manual annotation of its elements, called syllables. Automated methods for annotation have been proposed, but these methods assume that audio can be cleanly segmented into syllables, or they require carefully tuning multiple statistical models. Here, we present TweetyNet: a single neural network model that learns how to segment spectrograms of birdsong into annotated syllables. We show that TweetyNet mitigates limitations of methods that rely on segmented audio. We also show that TweetyNet performs well across multiple individuals from two species of songbirds, Bengalese finches and canaries. Lastly, we demonstrate that using TweetyNet we can accurately annotate very large datasets containing multiple days of song, and that these predicted annotations replicate key findings from behavioral studies. In addition, we provide open-source software to assist other researchers, and a large dataset of annotated canary song that can serve as a benchmark. We conclude that TweetyNet makes it possible to address a wide range of new questions about birdsong.

Miniaturized microscopes for head-mounted fluorescence imaging are powerful tools for visualizing neural activity during naturalistic behaviors, but the restricted field of view of first-generation ‘miniscopes’ limits the size of neural populations accessible for imaging. Here we describe a novel miniaturized mesoscope offering cellular-resolution imaging over areas spanning several millimeters in freely moving mice. This system enables comprehensive visualization of activity across entire brain regions or interactions across areas.

Learning to execute precise, yet complex, motor actions through practice is a trait shared by most organisms. Here we develop a novel experimental approach for the comprehensive investigation and characterization of the learning dynamics of practiced motion. Following the dynamical systems framework, we consider a high-dimensional behavioral space in which a trial-by-trial sequence of motor outputs defines a trajectory that converges to a fixed point - the desired motor output. In this scenario, details of the internal dynamics and the trial-by-trial learning mechanism cannot be disentangled from behavioral noise for nonlinear systems or even well estimated for linear systems with many parameters. To overcome this problem, we introduce a novel approach: the sporadic application of systematic target perturbations that span the behavioral space and allow us to estimate the linearized dynamics in the vicinity of the fixed point. The steady-state Lyapunov equation then allows us to identify the noise covariance. We illustrate the method by analyzing sequence-generating neural networks with either intrinsic or extrinsic noise, at time resolutions that span from spike timing to spiking rates. We demonstrate the utility of our approach in experimentally plausible and realizable settings and show that this method can fully characterize the linearized between-trials learning dynamics as well as extract meaningful internal properties of the unknown mechanism that drives the motor output within each trial. We then illustrate how the approach can be extended to nonlinear learning dynamics through a flexible choice of the basis and magnitude of perturbations. Significance statement Movement control ties brain activity to measurable external actions in real time, providing a useful tool for both neuroscientists interested in the emergence of stable behavior and biomedical engineers interested in the design of neural prosthesis and brain-machine interfaces. We approach the question of motor skill learning by introducing artificial errors through a novel perturbative scheme amenable to analytic examination in the linearized regime close to the desired behavior. Numerical simulations then demonstrate how to probe the learning dynamics in both linear and nonlinear systems. These findings stress the usefulness of analyzing responses to deliberately induced errors and the importance of properly designing such perturbation experiments. Our approach provides a novel generic tool for monitoring the acquisition of motor skills.

Songbirds have long been studied as a model system of sensory-motor learning. Many analyses of birdsong require time-consuming manual annotation of the individual elements of song, known as syllables or notes. Here we describe the first automated algorithm for birdsong annotation that is applicable to complex song such as canary song. We developed a neural network architecture, “TweetyNet”, that is trained with a small amount of hand-labeled data using supervised learning methods. We first show TweetyNet achieves significantly lower error on Bengalese finch song than a similar method, using less training data, and maintains low error rates across days. Applied to canary song, TweetyNet achieves fully automated annotation of canary song, accurately capturing the complex statistical structure previously discovered in a manually annotated dataset. We conclude that TweetyNet will make it possible to ask a wide range of new questions focused on complex songs where manual annotation was impractical.

Coordinated skills such as speech or dance involve sequences of actions that follow syntactic rules in which transitions between elements depend on the identities and order of past actions. Canary songs consist of repeated syllables called phrases, and the ordering of these phrases follows long-range rules¹ in which the choice of what to sing depends on the song structure many seconds prior. The neural substrates that support these long-range correlations are unknown. Here, using miniature head-mounted microscopes and cell-type-specific genetic tools, we observed neural activity in the premotor nucleus HVC^2,3,4 as canaries explored various phrase sequences in their repertoire. We identified neurons that encode past transitions, extending over four phrases and spanning up to four seconds and forty syllables. These neurons preferentially encode past actions rather than future actions, can reflect more than one song history, and are active mostly during the rare phrases that involve history-dependent transitions in song. These findings demonstrate that the dynamics of HVC include ‘hidden states’ that are not reflected in ongoing behaviour but rather carry information about prior actions. These states provide a possible substrate for the control of syntax transitions governed by long-range rules.