Brain-wide electrical dynamics encode an appetitive socioemotional state

Mague, Stephen D.; Talbot, Austin; Blount, Cameron; Duffney, Lara J.; Walder-Christensen, Kathryn K.; Adamson, Elise; Bey, Alexandra L.; Ndubuizu, Nkemdilim; Thomas, Gwenaёlle E.; Hughes, Duncan B.; Sinha, Saurabh; Fink, Alexandra M.; Gallagher, Neil M.; Fisher, Rachel L.; Jiang, Yong‐Hui; Carlson, David; Dzirasa, Kafui

doi:10.1101/2020.07.01.181347

Cited by 6 publications

(11 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All models were given a latent space dimension of 20. These parameter choices were made to match common settings where supervised generative models have been used [9,8]. Details on training and implementation are given in Supplemental Section 10.…”

Section: Resultsmentioning

confidence: 99%

“…We show empirically that this approach is flawed in our applications since the posterior distribution is inconsistent. Fortunately, the objective of (7) can easily be adapted to handle the missing data as…”

Section: B Incorporating Missing Data With Sosmentioning

confidence: 99%

“…D EVELOPING interpretable and explainable generative models has long been an integral area of machine learning and Bayesian modeling [1,2,3,4]. Generative models have great scientific utility in developing testable hypotheses and designing gold-standard causal experiments [5,6,7]. An interpretable relationship between the generative model representation and the observed covariates provides insight as to how to develop causal scientific experiments [7,8].…”

Section: Introductionmentioning

confidence: 99%

“…Many scientific applications require that the learned representation also be predictive of an auxiliary variable. In the neuroscience research motivating this work, this is often a behavioral outcome [9], a genetic phenotype [10], or the presence of some disorder or disability [11]. Obtaining a predictive latent representation allows for researchers to identify patterns relevant to this auxiliary variable and establish causality through experimental modification [12].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Supervising the Decoder of Variational Autoencoders to Improve Scientific Utility

Tu¹,

Talbot

Gallagher³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Probabilistic generative models are attractive for scientific modeling because their inferred parameters can be used to generate hypotheses and design experiments. This requires that the learned model provide an accurate representation of the input data and yield a latent space that effectively predicts outcomes relevant to the scientific question. Supervised Variational Autoencoders (SVAEs) have previously been used for this purpose, where a carefully designed decoder can be used as an interpretable generative model while the supervised objective ensures a predictive latent representation. Unfortunately, the supervised objective forces the encoder to learn a biased approximation to the generative posterior distribution, which renders the generative parameters unreliable when used in scientific models. This issue has remained undetected as reconstruction losses commonly used to evaluate model performance do not detect bias in the encoder. We address this previously-unreported issue by developing a second order supervision framework (SOS-VAE) that influences the decoder to induce a predictive latent representation. This ensures that the associated encoder maintains a reliable generative interpretation. We extend this technique to allow the user to trade-off some bias in the generative parameters for improved predictive performance, acting as an intermediate option between SVAEs and our new SOS-VAE. We also use this methodology to address missing data issues that often arise when combining recordings from multiple scientific experiments. We demonstrate the effectiveness of these developments using synthetic data and electrophysiological recordings with an emphasis on how our learned representations can be used to design scientific experiments.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: B Incorporating Missing Data With Sosmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Supervising the Decoder of Variational Autoencoders to Improve Scientific Utility

Tu¹,

Talbot

Gallagher³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Theories about mPFC function emphasize that it integrates sensory and limbic information to exibly guide behavior based on task rules 20,21 . mPFC circuitry has also been broadly implicated in social cognition [22][23][24] , social behaviors 25,26 , social defeat 27,28 , and social dominance 10,11,14 . These conceptual frameworks, led us to hypothesize that mPFC neurons encode social rank and are part of top-down circuits to guide behavior based on social rank.To determine if mPFC encodes social rank we needed to measure social dominance in an assay to facilitate statistical comparison.…”

mentioning

confidence: 99%

A cortical-hypothalamic circuit decodes social rank and promotes dominance behavior

Padilla-Coreano¹,

Batra²,

Patarino³

et al. 2020

Preprint

View full text Add to dashboard Cite

How do we know our social rank? Most social species, from insects to humans, self-organize into social dominance hierarchies (1–4). The establishment of social ranks serves to decrease aggression, conserve energy, and maximize survival for the entire group (5–8). Despite dominance behaviors being critical for successful interactions and ultimately, survival, we have only begun to learn how the brain represents social rank (9–12) and guides behavior based on this representation. The medial prefrontal cortex (mPFC) has been implicated in the expression of social dominance in rodents (10,11), and in social rank learning in humans (13,14). Yet precisely how the mPFC encodes rank and which circuits mediate this computation is not known. We developed a trial-based social competition assay in which mice compete for rewards, as well as a computer vision tool to track multiple, unmarked animals. With the development of a deep learning computer vision tool (AlphaTracker) and wireless electrophysiology recording devices, we have established a novel platform to facilitate quantitative examination of how the brain gives rise to social behaviors. We describe nine behavioral states during social competition that were accurately decoded from mPFC ensemble activity using a hidden Markov model combined with generalized linear models (HMM-GLM). Population dynamics in the mPFC were predictive of social rank and competitive success. This population-level rank representation translated into differences in the individual cell responses to task-relevant events across ranks. Finally, we demonstrate that mPFC cells that project to the lateral hypothalamus contribute to the prediction of social rank and promote dominance behavior during the reward competition. Thus, we reveal a cortico-hypothalamic circuit by which mPFC exerts top-down modulation of social dominance.

show abstract

Modular Electrode Array for Multi‐site Extracellular Recordings from Brains of Freely Moving Rodents

2022

View full text Add to dashboard Cite

Multi-site extracellular recordings from awake, freely moving rodents are an insightful technique that allows deduction of the dynamics of neural activity within a network of brain regions. Multiple advances in the design and materials of recording setups are available in the literature. However, most of these designs require several skill sets to assemble the electrodes and are expensive.Here, we explain in detail a custom design to build a multi-site (16 sites) electrode array (EA) and record extracellular electrical signals (local field potential and multi-unit spiking activity) at variable depths in freely behaving rodents. This EA weighs ∼3.0 g and costs less than $30. It provides mesoscopic neural activity maps (at millimeter scale) at low spatial resolution, thus enabling the experimenting group to further target specific regions with more expensive high-density probes at the resolution of an individual neuron. The article outlines the processes of building and implanting the array and recording neural activity during a behavior task. We also highlight the limitations of our design and the necessary steps to troubleshoot common issues faced during the initial implementation of the protocols. Finally, we explain the specific data one would obtain while using the probes during social interactions between rodents.

show abstract

Brain-wide electrical dynamics encode an appetitive socioemotional state

Cited by 6 publications

References 46 publications

Supervising the Decoder of Variational Autoencoders to Improve Scientific Utility

Supervising the Decoder of Variational Autoencoders to Improve Scientific Utility

A cortical-hypothalamic circuit decodes social rank and promotes dominance behavior

Modular Electrode Array for Multi‐site Extracellular Recordings from Brains of Freely Moving Rodents

Contact Info

Product

Resources

About