Biology subjects

International Brain Laboratory,

Publications and source records attributed to International Brain Laboratory,.

12 recordsLinked to original sources

bioRxiv2026

Serotonin induces DOWN states across the anesthetized mouse forebrain

Serotonergic neurons in the dorsal raphe nucleus (DRN) project extensively throughout the forebrain, yet their influence on global network states remains controversial. Serotonin is implicated in regulating slow-oscillations, so-called UP and DOWN states, which occur during sleep and light anesthesia. While optogenetic fMRI suggests that serotonin (5-HT) suppresses brain-wide activity, classical electrical stimulation studies report the induction of cortical UP states. To resolve this discrepancy, we combined optogenetic 5-HT stimulation with large-scale Neuropixel recordings across the forebrain of lightly anesthetized mice. We demonstrate that selective 5-HT release consistently induces DOWN states across the cortex and striatum, while leaving midbrain dynamics largely unaffected. A multi-area computational model constrained by biologically plausible connectivity, indicates that in certain regions the transitions arises from network-level synchronization rather than direct DRN input.

neuroscience↗

bioRxiv2026

Lightning Pose 3D: an uncertainty-aware framework for data-efficient multi-view animal pose estimation

Multi-view pose estimation is essential for quantifying animal behavior in scientific research, yet current methods struggle to achieve accurate tracking with limited labeled data and suffer from poor uncertainty estimates. We address these challenges with a flexible framework that can operate with or without camera calibration, combining novel training and post-processing techniques with an uncertainty-aware pseudo-labeling distillation procedure. Our multi-view model processes all camera views jointly using a pretrained vision transformer backbone, and a simulated occlusion technique encourages the model to learn robust cross-view correspondences without requiring camera parameters. When camera parameters are available, 3D data augmentations and a triangulation-based loss further encourage geometric consistency. We extend the Ensemble Kalman Smoother (EKS) post-processor to the nonlinear case, leveraging camera geometry, and introduce a variance inflation technique that detects cross-view inconsistencies and corrects overconfident predictions. We validate our approach on five datasets spanning three species (fly, mouse, bird), including a multi-animal dataset with two visually distinct individuals; the proposed pipeline consistently outperforms existing methods across datasets. We demonstrate how these improvements translate to downstream scientific analyses using data from the International Brain Laboratory, showing improved unsupervised behavioral clustering and neural decoding of paw kinematics with just 200 labeled frames. To facilitate adoption, we developed a browser-based, cloud-compatible user interface that supports the full life cycle of multi-view pose estimation, from labeling and model training to post-processing with EKS and diagnostic visualizations.

neuroscience↗

bioRxiv2025

How learned expectations shape brain-wide responses

Expectations stemming from prior knowledge have long been known to influence stimulus processing and decision-making. Empowered by brain-wide recordings during a two-choice sensory decision-making task from the International Brain Laboratory1, we sought to identify the form of expectation-based modulations and how they influence ongoing computations. To discern where, when, and how prior knowledge about the stimulus - known to be stored in the brain over the inter-trial period2 - influences brain activity over the arc from stimulus processing to choice selection and action generation, we disentangled the encoding of the highly correlated stimulus, choice, and prior variables as sensory processors, choice/integration processors, and movement generators. We examined which brain regions were modulated by prior expectations and at which point in the task, and characterized whether those modulations take the form of biasing initial activations or response gains. We found that the activities of choice/integrators and movement generators were influenced by prior expectation, while purely stimulus processors were not significantly modulated by prior expectation. In addition, we found that expectation-based modulation takes the form of both initial activity bias and gain modulation in integration/choice computations and in movement generators. We further find that the brain-wide recordings reveal an emergent simplicity: A few-parameter mechanistic neural circuit model captures the population dynamics of stimulus response, integration/choice processing, and action initiation, as well as their modulations by the prior. After fitting to physiology, the models behavior predicts the psychometric and chronometric behavior of mice with no additional parameters. It reveals that the larger bias and gain modulations of movement initiators relative to integration/choice in the data and model is inherited by amplifying integrator/choice inputs rather than being an additional expectation-based effect. Collectively, our results characterize in fine spatiotemporal detail and brain-wide breadth the mechanisms and loci where prior expectations influence cognition, while the novel synthesis between large-scale electrophysiology and mechanistic modeling reveals an organizing simplicity behind perceptual decision making across the brain.

neuroscience↗

bioRxiv2025

Age-related changes in behavioral and neural variability in a decision-making task

Age-related cognitive decline in learning and decision-making may arise from increased variability of neural responses. Here, we investigated how ageing affects behavioural and neural variability by recording >18,000 neurons across 16 brain regions (including cortex, hippocampus, thalamus, midbrain, and basal ganglia) in younger and older mice performing a visual decision-making task. Older mice showed more variable response times, reproducing a common finding in human ageing studies. Ageing globally increased firing rates and post-stimulus neural variability (quantified using the Fano factor), and decreased variability quenching-the reduction in neural variability upon stimulus presentation. Older animals showed higher overall firing rates across areas of visual and motor cortex, striatum, midbrain, and hippocampus, but lower firing rates in thalamic areas. Age-related attenuation in stimulus-induced variability quenching was most prominent in visual and motor cortex, striatum, and thalamic area. These findings show how large-scale neural recordings can help uncover regional specificity of ageing effects in single neurons, ultimately improving our understanding of the neural basis of age-related cognitive decline.

neuroscience↗

bioRxiv2025

Spatially distributed and regionally unbound cellular resolution brain-wide processing loops in mice

Until recently, brain activity could be examined either globally, through regional averaging, or locally, at cellular resolution--but not both at once. As a result, regions were characterized as functionally homogeneous units while single neurons were often characterized as heterogeneously and randomly tuned. Here, we overcome this long-standing divide by leveraging the unprecedented density of International Brain Laboratory electrophysiological recordings in mice performing a sensorimotor decision-making task. Computationally integrating these recordings and ordering neurons by functional similarity reveals directly to the eye the full temporal arc of computations unfurling over the phases of a trial and across the brain at cellular and millisecond resolution, providing a dynamical map of the architecture of brain-wide computation. We report that one of the largest shares of brain-wide variance is explained not by task-related sensory responses, integration, or decision related signals, but by neurons encoding internally generated, trial-timed broadcast sequences that carry contextual information and are biased toward hippocampal and entorhinal involvement. We find that individual neurons reliably specialize in interpretable functions, yet neurons with similar functional roles are widely dispersed across regions. Each region contains cells spanning nearly the full repertoire of response types, rendering functional tuning largely unpredictable from cytoarchitectural boundaries or spatial location alone. In particular, even primary visual cortex contains a broad diversity of response types beyond stimulus-locked responses. The dominant discernible organizational distinction is instead between cortical and subcortical processing. Subcortical neurons carry the bulk of internal computations in expert mice, including integration, decision-making, and movement initiation. These patterns support a model of broadcast rather than staged computation, in which signals are widely shared and transformed rather than serially processed. Finally, we find that despite widespread regional mixing, representations at the single neuron level are not randomly mixed. Neurons combine specific features in highly correlated groupings, contradicting the hypothesis of random mixed selectivity at the cellular level. Together, our findings suggest a revised architecture of brain computation: typically interpretable and non-random selectivity at the level of single neurons, coupled to continual brain-wide broadcast rather than staged processing, in which internally structured activity dominates neural dynamics and scaffolds cognition.

neuroscience↗

bioRxiv2025

Serotonin modulates neural dynamics in a subspace orthogonal to the choice space

Serotonin (5-HT) is a central neuromodulator which is implicated in, amongst other functions, cognitive flexibility. 5-HT is released from the dorsal raphe nucleus (DRN) throughout nearly the entire forebrain. Little is known, however, about how serotonin affects downstream populations of neurons and how this modulation might support its cognitive functions. Here, we optogenetically stimulated serotonergic neurons in the DRN while recording large parts of the brain with Neuropixels during quiet wakefulness and performance of a perceptual decision-making task. During quiet wakefulness, 5-HT stimulation induced a rapid switch in internal state, as indicated by dilation of the pupil, suppression of hippocampal sharp wave ripples, and increased exploratory behaviors, such as whisking. To elucidate the brain-wide effect of serotonin release we performed acute Neuropixel recordings in seven target locations, a total of 7,478 neurons were recorded across 17 mice. We found that 5-HT stimulation significantly modulated neural activity in all the recorded brain regions, both during quiet wakefulness and task performance. During task performance, however, we observed no change in behavior when stimulating 5-HT. We found that the 5-HT modulation of high-dimensional neural dynamics is confined to a subspace which is orthogonal relative to the choice axis. These observations describe a possible mechanism for the induction of state-dependent stimulus representations, suggesting a neural basis for neuromodulatory effects on brain-wide circuits to flexible decision-making.

neuroscience↗

bioRxiv2024

Exploiting correlations across trials and behavioral sessions to improve neural decoding

Traditional neural decoders model the relationship between neural activity and behavior within individual trials of a single experimental session, neglecting correlations across trials and sessions. However, animals exhibit similar neural activities when performing the same behavioral task, and their behaviors are influenced by past experiences from previous trials. To exploit these informative correlations in large datasets, we introduce two complementary models: a multi-session reduced-rank regression model that shares similar behaviorally-relevant statistical structure in neural activity across sessions to improve decoding, and a multi-session state-space model that shares similar behavioral statistical structure across trials and sessions. Applied across 433 sessions spanning 270 brain regions in the International Brain Laboratory public mouse Neuropixels dataset, our decoders demonstrate improved decoding accuracy for four distinct behaviors compared to traditional approaches. These results generalize across additional datasets, species, and behavioral tasks. Unlike existing deep learning approaches, our models are interpretable and efficient, uncovering low-dimensional representations that predict animal decisions, quantifying single-neuron contributions to decoding behaviors, and identifying different activation timescales of neural activity across the brain. Code: https://github.com/yzhang511/neural_decoding.

neuroscience↗

bioRxiv2024

Identifying the factors governing internal state switches during nonstationary sensory decision-making

Recent work has revealed that mice do not rely on a stable strategy during perceptual decision-making, but switch between multiple strategies within a single session [1, 2]. However, this switching behavior has not yet been characterized in non-stationary environments, and the factors that govern switching remain unknown. Here we address these questions using an internal state model with input-driven transitions. Our approach relies on a hidden Markov model (HMM) with two sets of per-state generalized linear models (GLMs): a set of Bernoulli GLMs for modeling the animals state- and stimulus-dependent choice on each trial, and a multinomial GLM for modeling input-dependent transitions between states. We used this model to analyze a dataset from the International Brain Laboratory (IBL), in which mice performed a binary decision-making task with non-stationary stimulus statistics. We found that mouse behavior in this task was accurately described by a four-state model. This model contained two "engaged" states, in which performance was good despite slight left and right biases, and two "disengaged" states, where performance was low and exhibited with larger left and right biases, respectively. Our analyses revealed that mice preferentially used left-bias strategies during left-bias stimulus blocks, and right-bias strategies during right-bias stimulus blocks, meaning that they could achieve reasonably high performance even in disengaged states simply by biasing choice toward the side with greater prior probability. Our model showed that past choices and past stimuli predicted transitions between left- and right-bias states, while past rewards predicted transitions between engaged and disengaged states. In particular, greater past reward predicted transition to disengaged states, suggesting that disengagement may be associated with satiety.

neuroscience↗

bioRxiv2023

Dissecting the Complexities of Learning With Infinite Hidden Markov Models

Learning to exploit the contingencies of a complex experiment is not an easy task for animals. Individuals learn in an idiosyncratic manner, revising their approaches multiple times as they are shaped, or shape themselves, and potentially end up with different strategies. Their long-run learning curves are therefore a tantalizing target for the sort of individualized quantitative characterizations that sophisticated modelling can provide. However, any such model requires a flexible and extensible structure which can capture radically new behaviours as well as slow changes in existing ones. To this end, we suggest a dynamic input-output infinite hidden semi-Markov model, whose latent states are associated with specific components of behaviour. This model includes an infinite number of potential states and so has the capacity to describe substantially new behaviours by unearthing extra states; while dynamics in the model allow it to capture more modest adaptations to existing behaviours. We individually fit the model to data collected from more than 100 mice as they learned a contrast detection task over tens of sessions and around fifteen thousand trials each. Despite large individual differences, we found that most animals progressed through three major stages of learning, the transitions between which were marked by distinct additions to task understanding. We furthermore showed that marked changes in behaviour are much more likely to occur at the very beginning of sessions, i.e. after a period of rest, and that response biases in earlier stages are not predictive of biases later on in this task.

animal behavior and cognition↗

bioRxiv2023

Pinpoint: trajectory planning for multi-probe electrophysiology and injections in an interactive web-based 3D environment

Targeting deep brain structures during electrophysiology and injections requires intensive training and expertise. Even with experience, researchers often cant be certain that a probe is placed precisely in a target location and this complexity scales with the number of simultaneous probes used in an experiment. Here, we present Pinpoint, open-source software that allows for interactive exploration of stereotaxic insertion plans. Once an insertion plan is created, Pinpoint allows users to save these online and share them with collaborators. 3D modeling tools allow users to explore their insertions alongside rig and implant hardware and ensure plans are physically possible. Probes in Pinpoint can be linked to electronic micro-manipulators allowing real-time visualization of current brain region targets alongside neural data. In addition, Pinpoint can control manipulators to automate and parallelize the insertion process. Compared to previously available software, Pinpoints easy access through web browsers, extensive features, and real-time experiment integration enable more efficient and reproducible recordings.

neuroscience↗

bioRxiv2023

Brain-wide representations of prior information in mouse decision-making

The neural representations of prior information about the state of the world are poorly understood. To investigate them, we examined brain-wide Neuropixels recordings and widefield calcium imaging collected by the International Brain Laboratory. Mice were trained to indicate the location of a visual grating stimulus, which appeared on the left or right with prior probability alternating between 0.2 and 0.8 in blocks of variable length. We found that mice estimate this prior probability and thereby improve their decision accuracy. Furthermore, we report that this subjective prior is encoded in at least 20% to 30% of brain regions which, remarkably, span all levels of processing, from early sensory areas (LGd, VISp) to motor regions (MOs, MOp, GRN) and high level cortical regions (ACAd, ORBvl). This widespread representation of the prior is consistent with a neural model of Bayesian inference involving loops between areas, as opposed to a model in which the prior is incorporated only in decision-making areas. This study offers the first brain-wide perspective on prior encoding at cellular resolution, underscoring the importance of using large scale recordings on a single standardized task.

neuroscience↗

bioRxiv2020

Reverse-engineering Recurrent Neural Network solutions to a hierarchical inference task for mice

We study how recurrent neural networks (RNNs) solve a hierarchical inference task involving two latent variables and disparate timescales separated by 1-2 orders of magnitude. The task is of interest to the International Brain Laboratory, a global collaboration of experimental and theoretical neuroscientists studying how the mammalian brain generates behavior. We make four discoveries. First, RNNs learn behavior that is quantitatively similar to ideal Bayesian baselines. Second, RNNs perform inference by learning a two-dimensional subspace defining beliefs about the latent variables. Third, the geometry of RNN dynamics reflects an induced coupling between the two separate inference processes necessary to solve the task. Fourth, we perform model compression through a novel form of knowledge distillation on hidden representations - Representations and Dynamics Distillation (RADD)- to reduce the RNN dynamics to a low-dimensional, highly interpretable model. This technique promises a useful tool for interpretability of high dimensional nonlinear dynamical systems. Altogether, this work yields predictions to guide exploration and analysis of mouse neural data and circuity.

neuroscience↗