Aligned and oblique dynamics in recurrent neural networks DOI Creative Commons
Friedrich Schuessler, Francesca Mastrogiuseppe, Srdjan Ostojic

et al.

eLife, Journal Year: 2024, Volume and Issue: 13

Published: Nov. 27, 2024

The relation between neural activity and behaviorally relevant variables is at the heart of neuroscience research. When strong, this termed a representation. There increasing evidence, however, for partial dissociations in an area external variables. While many explanations have been proposed, theoretical framework relationship internal lacking. Here, we utilize recurrent networks (RNNs) to explore question when how dynamics network’s output are related from geometrical point view. We find that training RNNs can lead two dynamical regimes: either be aligned with directions generate variables, or oblique them. show choice readout weight magnitude before serve as control knob regimes, similar recent findings feedforward networks. These regimes functionally distinct. Oblique more heterogeneous suppress noise their directions. They furthermore robust perturbations along Crucially, regime specific (but not feedforward) networks, arising stability considerations. Finally, tendencies toward dissociated recordings. Altogether, our results open new perspective interpreting by relating network output.

Language: Английский

Brain-wide neural activity underlying memory-guided movement DOI Creative Commons
Susu Chen, Yi Liu,

Ziyue Wang

et al.

Cell, Journal Year: 2024, Volume and Issue: 187(3), P. 676 - 691.e16

Published: Feb. 1, 2024

Behavior relies on activity in structured neural circuits that are distributed across the brain, but most experiments probe neurons a single area at time. Using multiple Neuropixels probes, we recorded from multi-regional loops connected to anterior lateral motor cortex (ALM), circuit node mediating memory-guided directional licking. Neurons encoding sensory stimuli, choices, and actions were brain. However, choice coding was concentrated ALM subcortical areas receiving input an ALM-dependent manner. Diverse orofacial movements encoded hindbrain; midbrain; and, lesser extent, forebrain. Choice signals first detected midbrain, followed by thalamus other brain areas. At movement initiation, choice-selective collapsed new patterns driving specific actions. Our provide foundation for models of decision-making initiation.

Language: Английский

Citations

40
Separating cognitive and motor processes in the behaving mouse DOI
Munib A. Hasnain, Jaclyn E Birnbaum,

Juan Luis Ugarte Nunez

et al.

Nature Neuroscience, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 4, 2025

Language: Английский

Citations

1
Large scale, simultaneous, chronic neural recordings from multiple brain areas DOI Creative Commons
Maxwell D Melin,

Anup Khanal,

Marvin Vasquez

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Dec. 23, 2023

Abstract Understanding how brain activity is related to animal behavior requires measuring multi-area interactions on multiple timescales. However, methods perform chronic, simultaneous recordings of neural from many areas are lacking. Here, we introduce a novel approach for independent chronic probe implantation that enables flexible, interrogation regions during head restrained or freely moving behavior. The approach, called indie ( in dependent d ovetail implants e lectrophysiology), repeated retrieval and reimplantation probes. can be combined with other modalities such as skull clearing cortex wide access optogenetics optic fibers. Using this implanted 6 probes chronically one hemisphere the mouse brain. implant lightweight, allows flexible targeting different angles, offers enhanced stability. Our broadens applications recording while retaining its main advantages over acute (superior stability, longitudinal monitoring interrogations) provides an appealing venue study processes not accessible by methods, substrate learning across areas.

Language: Английский

Citations

14
Orexin population activity precisely reflects net body movement across behavioral and metabolic states DOI Open Access

Alexander L Tesmer,

Paulius Viskaitis,

Dane Donegan

et al.

Published: Feb. 7, 2025

Tracking net body movement in real time may enable the brain to estimate ongoing demands and thus better orchestrate muscle tone, energy balance, arousal. To identify neural populations specializing tracking movement, here we compared self-initiated movement-related activity across genetically-defined subcortical neurons mouse brain, including dopaminergic, glutamatergic, noradrenergic, key peptidergic neurons. We show that hypothalamic orexin/hypocretin-producing (HONs) are exceptionally precise movement-trackers, encoding multiple classified behaviors with a high degree of precision, independent head acceleration. This was so precise, video analysis reliably served as low-cost biometric for HON population activity. The internal nutritional states, occurred communication bandwidth distinct from blood glucose. At projection targets, orexin/hypocretin peptide outputs correlated projection-specific manner, indicating functional heterogeneity outputs. Finally, found not encoded same extent other related arousal or energy. These findings indicate orchestrators metabolism finely tuned encode constituting bridge multiplexing motor resources.

Language: Английский

Citations

0
Orexin population activity precisely reflects net body movement across behavioral and metabolic states DOI Open Access

Alexander L Tesmer,

Paulius Viskaitis,

Dane Donegan

et al.

Published: Feb. 7, 2025

Tracking net body movement in real time may enable the brain to estimate ongoing demands and thus better orchestrate muscle tone, energy balance, arousal. To identify neural populations specializing tracking movement, here we compared self-initiated movement-related activity across genetically-defined subcortical neurons mouse brain, including dopaminergic, glutamatergic, noradrenergic, key peptidergic neurons. We show that hypothalamic orexin/hypocretin-producing (HONs) are exceptionally precise movement-trackers, encoding multiple classified behaviors with a high degree of precision, independent head acceleration. This was so precise, video analysis reliably served as low-cost biometric for HON population activity. The internal nutritional states, occurred communication bandwidth distinct from blood glucose. At projection targets, orexin/hypocretin peptide outputs correlated projection-specific manner, indicating functional heterogeneity outputs. Finally, found not encoded same extent other related arousal or energy. These findings indicate orchestrators metabolism finely tuned encode constituting bridge multiplexing motor resources.

Language: Английский

Citations

0
Pan-cortical 2-photon mesoscopic imaging and neurobehavioral alignment in awake, behaving mice DOI Creative Commons
Evan D. Vickers, David A. McCormick

eLife, Journal Year: 2024, Volume and Issue: 13

Published: Jan. 23, 2024

The flow of neural activity across the neocortex during active sensory discrimination is constrained by task-specific cognitive demands, movements, and internal states. During behavior, brain appears to sample from a broad repertoire activation motifs. Understanding how these patterns local global are selected in relation both spontaneous task-dependent behavior requires in-depth study densely sampled at single neuron resolution large regions cortex. In significant advance toward this goal, we developed procedures record mesoscale 2-photon Ca2+ imaging data two novel vivo preparations that, between them, allow for simultaneous access nearly all 0f mouse dorsal lateral neocortex. As proof principle, aligned with behavioral primitives high-level motifs reveal existence populations neurons that coordinated their cortical areas changes movement and/or arousal. methods detail here facilitate identification exploration widespread, spatially heterogeneous ensembles whose related diverse aspects behavior.

Language: Английский

Citations

3
Aligned and oblique dynamics in recurrent neural networks DOI Creative Commons
Friedrich Schuessler, Francesca Mastrogiuseppe, Srdjan Ostojic

et al.

eLife, Journal Year: 2024, Volume and Issue: 13

Published: March 13, 2024

The relation between neural activity and behaviorally relevant variables is at the heart of neuroscience research. When strong, this termed a representation. There increasing evidence, however, for partial dissociations in an area external variables. While many explanations have been proposed, theoretical framework relationship internal lacking. Here, we utilize recurrent networks (RNNs) to explore question when how dynamics network’s output are related from geometrical point view. We find that training RNNs can lead two dynamical regimes: either be aligned with directions generate variables, or oblique them. show choice readout weight magnitude before serve as control knob regimes, similar recent findings feedforward networks. These regimes functionally distinct. Oblique more heterogeneous suppress noise their directions. They furthermore robust perturbations along Crucially, regime specific (but not feedforward) networks, arising stability considerations. Finally, tendencies toward dissociated recordings. Altogether, our results open new perspective interpreting by relating network output.

Language: Английский

Citations

3
Separating cognitive and motor processes in the behaving mouse DOI Creative Commons
Munib A. Hasnain, Jaclyn E Birnbaum,

Juan Luis Ugarte Nunez

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Aug. 24, 2023

The cognitive processes supporting complex animal behavior are closely associated with ubiquitous movements responsible for our posture, facial expressions, ability to actively sample sensory environments, and other critical processes. These strongly related neural activity across much of the brain often highly correlated ongoing processes, making it challenging dissociate dynamics that support from those movements. In such cases, a issue is whether separable movements, or if they driven by common mechanisms. Here, we demonstrate how separability motor can be assessed, and, when separable, each component isolated. We establish novel two-context behavioral task in mice involves multiple show commonly observed taken contaminated When components isolated using approach subspace decomposition, find exhibit distinct dynamical trajectories. Further, properly accounting movement revealed largely separate populations cells encode variables, contrast "mixed selectivity" reported. Accurately isolating particular will essential developing conceptual computational models circuit function evaluating cell types which circuits composed.

Language: Английский

Citations

5
Pan-cortical 2-photon mesoscopic imaging and neurobehavioral alignment in awake, behaving mice DOI Open Access
Evan D. Vickers, David A. McCormick

Published: March 28, 2024

The flow of neural activity across the neocortex during active sensory discrimination is constrained by task-specific cognitive demands, movements, and internal states. During behavior, brain appears to sample from a broad repertoire activation motifs. Understanding how these patterns local global are selected in relation both spontaneous task-dependent behavior requires in-depth study densely sampled at single neuron resolution large regions cortex. In significant advance toward this goal, we developed procedures record mesoscale 2-photon Ca 2+ imaging data two novel vivo preparations that, between them, allow simultaneous access nearly all mouse dorsal lateral neocortex. As proof principle, aligned with behavioral primitives high-level motifs reveal existence populations neurons that coordinated their cortical areas changes movement and/or arousal. methods detail here facilitate identification exploration widespread, spatially heterogeneous ensembles whose related diverse aspects behavior.

Language: Английский

Citations

1
A method to find temporal structure of neuronal coactivity patterns with across-trial correlations DOI
Duho Sihn, Soyoung Chae, Sung-Phil Kim

et al.

Journal of Neuroscience Methods, Journal Year: 2024, Volume and Issue: 408, P. 110172 - 110172

Published: May 22, 2024

Language: Английский

Citations

1