Efficient coding explains neural response homeostasis and stimulus-specific adaptation DOI Creative Commons
Edward Young, Yashar Ahmadian

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2023, Номер unknown

Опубликована: Ноя. 1, 2023

Abstract In the absence of adaptation, average firing rate neurons would rise or drop when changes in environment make their preferred stimuli more less prevalent. However, by adjusting responsiveness neurons, adaptation can yield homeostasis and stabilise rates at fixed levels, despite stimulus statistics. sensory cortex, is typically also specific, that reduce to over-represented stimuli, but maintain even increase far from ones. Here, we present a normative explanation grounded efficient coding principle, showing this yields an optimal trade-off between fidelity metabolic cost neural firing. Unlike previous theories, formulate problem computation-agnostic manner, enabling our framework apply periphery. We then general Distributed Distributional Codes, specific computational theory representations serving Bayesian inference. demonstrate how homeostatic coding, combined with such representations, provides for stimulus-specific widely observed across brain, scheme be accomplished divisive normalisation adaptive weights. Further, develop model within framework, fitting it previously published experimental data, quantitatively account measures adaption primary visual cortex.

Язык: Английский

Excitation creates a distributed pattern of cortical suppression due to varied recurrent input DOI Creative Commons
Jonathan F. O’Rawe, Zhishang Zhou, Anna J. Li

и другие.

Neuron, Год журнала: 2023, Номер 111(24), С. 4086 - 4101.e5

Опубликована: Окт. 20, 2023

Язык: Английский

Процитировано

13
Cellular-resolution optogenetics reveals attenuation-by-suppression in visual cortical neurons DOI Creative Commons
Paul K. LaFosse, Zhishang Zhou, Jonathan F. O’Rawe

и другие.

Proceedings of the National Academy of Sciences, Год журнала: 2024, Номер 121(45)

Опубликована: Ноя. 1, 2024

The relationship between neurons' input and spiking output is central to brain computation. Studies in vitro anesthetized animals suggest that nonlinearities emerge cells' input-output (IO; activation) functions as network activity increases, yet how neurons transform inputs vivo has been unclear. Here, we characterize cortical principal activation awake mice using two-photon optogenetics. We deliver fixed at the soma while varies with sensory stimuli. find responses optogenetic are nearly unchanged excited, reflecting a linear response regime above resting point. In contrast, dramatically attenuated by suppression. This attenuation powerful means filter arriving suppressed cells, privileging other excited neurons. These results have two major implications. First, somatic neural accord used recent machine learning systems. Second, IO can inputs-not only do stimuli change outputs, but these changes also affect input, attenuating some leaving others unchanged.

Язык: Английский

Процитировано

4
Ensemble priming via competitive inhibition: local mechanisms of sensory context storage and deviance detection in the neocortical column DOI Creative Commons
Ryan Thorpe, Christopher I. Moore, Stephanie R. Jones

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Янв. 9, 2025

The process by which neocortical neurons and circuits amplify their response to an unexpected change in stimulus, often referred as deviance detection (DD), has long been thought be the product of specialized cell types and/or routing between mesoscopic brain areas. Here, we explore a different theory, whereby DD emerges from local network-level interactions within column. We propose that deviance-driven neural dynamics can emerge through ensembles have fundamental inhibitory motif: competitive inhibition reciprocally connected under modulation feed-forward selective (dis)inhibition. Using this framework, were able simulate variety phenomena pertaining experimentally observed shifts tuning across neurons, time, stimulus history. Anchoring our approach phenomena, used computation modeling two networks vastly levels biophysical detail test hypotheses on emergent robustness underlying connectivity parameters. With number corollary predictions tested future vivo studies, show ensemble priming via (dis)inhibition acts mechanism for sensory context storage does not require input other areas-a novel theoretical paradigm resolves previously confounding aspects encoding predictive processing neocortex.

Язык: Английский

Процитировано

0
Efficient coding explains neural response homeostasis and stimulus-specific adaptation DOI Open Access
Edward Young, Yashar Ahmadian

Опубликована: Март 4, 2025

In the absence of adaptation, average firing rate neurons would rise or drop when changes in environment make their preferred stimuli more less prevalent. However, by adjusting responsiveness neurons, adaptation can yield homeostasis and stabilise rates at fixed levels, despite stimulus statistics. sensory cortex, is typically also specific, that reduce to over-represented stimuli, but maintain even increase far from ones. Here, we present a normative explanation grounded efficient coding principle, showing this yields an optimal trade-off between fidelity metabolic cost neural firing. Unlike previous theories, formulate problem computation-agnostic manner, enabling our framework apply periphery. We then general Distributed Distributional Codes, specific computational theory representations serving Bayesian inference. demonstrate how homeostatic coding, combined with such representations, provides for stimulus-specific widely observed across brain, scheme be accomplished divisive normalisation adaptive weights. Further, develop model within framework, fitting it previously published experimental data, quantitatively account measures adaption primary visual cortex.

Язык: Английский

Процитировано

0
Efficient coding explains neural response homeostasis and stimulus-specific adaptation DOI Open Access
Edward Young, Yashar Ahmadian

Опубликована: Март 4, 2025

In the absence of adaptation, average firing rate neurons would rise or drop when changes in environment make their preferred stimuli more less prevalent. However, by adjusting responsiveness neurons, adaptation can yield homeostasis and stabilise rates at fixed levels, despite stimulus statistics. sensory cortex, is typically also specific, that reduce to over-represented stimuli, but maintain even increase far from ones. Here, we present a normative explanation grounded efficient coding principle, showing this yields an optimal trade-off between fidelity metabolic cost neural firing. Unlike previous theories, formulate problem computation-agnostic manner, enabling our framework apply periphery. We then general Distributed Distributional Codes, specific computational theory representations serving Bayesian inference. demonstrate how homeostatic coding, combined with such representations, provides for stimulus-specific widely observed across brain, scheme be accomplished divisive normalisation adaptive weights. Further, develop model within framework, fitting it previously published experimental data, quantitatively account measures adaption primary visual cortex.

Язык: Английский

Процитировано

0
Impact of Local Connectivity Patterns on Excitatory-Inhibitory Network Dynamics DOI Creative Commons
Yuxiu Shao, David Dahmen, Stefano Recanatesi

и другие.

PRX Life, Год журнала: 2025, Номер 3(2)

Опубликована: Май 20, 2025

Networks of excitatory and inhibitory (EI) neurons form a canonical circuit in the brain. Seminal theoretical results on dynamics such networks are based assumption that synaptic strengths depend type they connect, but otherwise statistically independent. Recent physiology datasets, however, highlight prominence specific connectivity patterns go well beyond what is expected from independent connections. While decades influential research have demonstrated strong role basic EI cell structure, extent to which additional features influence remains be fully determined. Here we examine effects pairwise motifs linear excitatory-inhibitory using an analytical framework approximates terms low-rank structures. This approximation mathematical derivation dominant eigenvalues matrix, it predicts impact responses external inputs their interactions with cell-type structure. Our reveal particular pattern connectivity, namely chain motifs, much stronger eigenmodes than other motifs. In particular, over-representation induces positive eigenvalue inhibition-dominated networks, generates potential instability requires revisiting classical excitation-inhibition balance criteria. Examining inputs, show can own induce paradoxical responses, where increased input leads decrease activity due recurrent feedback. These findings direct implications for interpretation experiments optogenetic perturbations measured used infer dynamical regime cortical circuits. Published by American Physical Society 2025

Язык: Английский

Процитировано

0
Behavioral optogenetics in nonhuman primates; a psychological perspective DOI Creative Commons
Arash Afraz

Current Research in Neurobiology, Год журнала: 2023, Номер 5, С. 100101 - 100101

Опубликована: Янв. 1, 2023

Optogenetics has been a promising and developing technology in systems neuroscience throughout the past decade. It difficult though to reliably establish potential behavioral effects of optogenetic perturbation neural activity nonhuman primates. This poses challenge on future optogenetics humans as concepts need be developed primates first. Here, I briefly summarize viable approaches taken improve primate optogenetics, then focus one approach: improvements measurement behavior. bring examples from visual behavior show how choice method might conceal large effects. will discuss "cortical detection" task detail an example sensitive that can record cortical stimulation with high fidelity. Finally, encouraged by rich scientific landscape ahead invite developers chronically implantable devices designed for simultaneous recording intervention

Язык: Английский

Процитировано

7
Single cell optogenetics reveals attenuation-by-suppression in visual cortical neurons DOI Creative Commons
Paul K. LaFosse, Zhishang Zhou, Jonathan F. O’Rawe

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2023, Номер unknown

Опубликована: Сен. 14, 2023

Abstract The relationship between neurons’ input and spiking output is central to brain computation. Studies in vitro anesthetized animals suggest nonlinearities emerge cells’ input-output (activation) functions as network activity increases, yet how neurons transform inputs vivo has been unclear. Here, we characterize cortical principal activation awake mice using two-photon optogenetics. We deliver fixed at the soma while varies with sensory stimuli. find responses optogenetic are nearly unchanged excited, reflecting a linear response regime above resting point. In contrast, dramatically attenuated by suppression. This attenuation powerful means filter arriving suppressed cells, privileging other excited neurons. These results have two major implications. First, somatic neural accord used recent machine learning systems. Second, IO can — not only do stimuli change outputs, but these changes also affect input, attenuating some leaving others unchanged. Significance statement How their into outputs fundamental building block of Past studies measured (IO) or states. measure intact brain, where ongoing influence input. Using state-of-the-art methods precise near cell body, soma, discover supralinear-to-linear function, contrary previous findings threshold-linear, strongly saturating, power law functions. function shape allows decrease to, filter, they below firing rates, computation term attenuation-by-suppression.

Язык: Английский

Процитировано

6
Brain state transitions primarily impact the spontaneous rate of slow-firing neurons DOI Creative Commons

Bradley Dearnley,

Melissa M. Jones, Martynas Dervinis

и другие.

Cell Reports, Год журнала: 2023, Номер 42(10), С. 113185 - 113185

Опубликована: Сен. 28, 2023

The spontaneous firing of neurons is modulated by brain state. Here, we examine how such modulation impacts the overall distribution rates in neuronal populations neocortical, hippocampal, and thalamic areas across natural pharmacologically driven state transitions. We report that all examined combinations area transition category, structure rate similar, with almost fast-firing experiencing proportionally weak modulation, while slow-firing exhibit high inter-neuron variability magnitude, leading to a stronger on average. further demonstrate this linked left-skewed logarithmic scale recapitulated bivariate log-gamma, but not Gaussian, distributions. Our findings indicate preconfigured log-rate rigid long left tail malleable generic property forebrain circuits.

Язык: Английский

Процитировано

4
Recurrent cortical networks encode natural sensory statistics via sequence filtering DOI Creative Commons
Ciana Deveau, Zhishang Zhou, Paul K. LaFosse

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Фев. 28, 2024

Recurrent neural networks can generate dynamics, but in sensory cortex it has been unclear if any dynamic processing is supported by the dense recurrent excitatory-excitatory network. Here we show a new role for connections mouse visual cortex: they support powerful dynamical computations, filtering sequences of input instead generating sequences. Using two-photon optogenetics, measure responses to natural images and play them back, finding inputs are amplified when played back during correct movie context- preceding sequence corresponds vision. This selectivity depends on network mechanism: earlier patterns produce other local neurons, which interact with later patterns. We confirm this mechanism designing that or suppressed These data suggest cortical perform predictive processing, encoding statistics world input-output transformations.

Язык: Английский

Процитировано

1