Training deep neural density estimators to identify mechanistic models of neural dynamics

eLife, Год журнала: 2020, Номер 9

Опубликована: Сен. 17, 2020

Mechanistic modeling in neuroscience aims to explain observed phenomena terms of underlying causes. However, determining which model parameters agree with complex and stochastic neural data presents a significant challenge. We address this challenge machine learning tool uses deep density estimators—trained using simulations—to carry out Bayesian inference retrieve the full space compatible raw or selected features. Our method is scalable features can rapidly analyze new after initial training. demonstrate power flexibility our approach on receptive fields, ion channels, Hodgkin–Huxley models. also characterize circuit configurations giving rise rhythmic activity crustacean stomatogastric ganglion, use these results derive hypotheses for compensation mechanisms. will help close gap between data-driven theory-driven models dynamics.

Язык: Английский

---

A memristive neuron and its adaptability to external electric field

Chaos An Interdisciplinary Journal of Nonlinear Science, Год журнала: 2023, Номер 33(2)

Опубликована: Фев. 1, 2023

Connecting memristors into any neural circuit can enhance its potential controllability under external physical stimuli. Memristive current along a magnetic flux-controlled memristor estimate the effect of electromagnetic induction on circuits and neurons. Here, charge-controlled is incorporated one branch simple to an electric field. The field energy kept in each component respectively calculated, equivalent dimensionless function H obtained discern firing mode dependence from capacitive, inductive, memristive channels. HM channel occupies highest proportion Hamilton H, neurons present chaotic/periodic modes because large injection field, while bursting spiking behaviors emerge when HL holds maximal H. modified control this neuron accompanying with parameter shift shape deformation resulting accommodation channel. In presence noisy disturbance stochastic resonance induced neuron. Exposed stronger absorb more behave as signal source for shunting, negative new model address main properties biophysical neurons, it further be used explore collective self-organization networks flow disturbance.

Язык: Английский

---

In vivo ephaptic coupling allows memory network formation

Cerebral Cortex, Год журнала: 2023, Номер 33(17), С. 9877 - 9895

Опубликована: Июль 7, 2023

Abstract It is increasingly clear that memories are distributed across multiple brain areas. Such “engram complexes” important features of memory formation and consolidation. Here, we test the hypothesis engram complexes formed in part by bioelectric fields sculpt guide neural activity tie together areas participate complexes. Like conductor an orchestra, influence each musician or neuron orchestrate output, symphony. Our results use theory synergetics, machine learning, data from a spatial delayed saccade task provide evidence for vivo ephaptic coupling representations.

Язык: Английский

---

Direction selectivity in retinal bipolar cell axon terminals

Neuron, Год журнала: 2021, Номер 109(18), С. 2928 - 2942.e8

Опубликована: Авг. 13, 2021

The ability to encode the direction of image motion is fundamental our sense vision. Direction selectivity along four cardinal directions thought originate in direction-selective ganglion cells (DSGCs) because directionally tuned GABAergic suppression by starburst cells. Here, utilizing two-photon glutamate imaging measure synaptic release, we reveal that all arises earlier than expected at bipolar cell outputs. Individual contained distinct populations axon terminal boutons with different preferred directions. We further show this bouton-specific tuning relies on cholinergic excitation from and inhibition wide-field amacrine DSGCs received both aligned inputs untuned among heterogeneously glutamatergic bouton populations. Thus, directional excitatory visual pathway incrementally refined terminals their recipient DSGC dendrites two neurotransmitters co-released

Язык: Английский

---

Ancestral circuits for vertebrate color vision emerge at the first retinal synapse

Science Advances, Год журнала: 2021, Номер 7(42)

Опубликована: Окт. 13, 2021

In vivo recordings of cone photoreceptor outputs in a tetrachromate reveal efficient ancestral strategy for color processing.

Язык: Английский

---

Bayesian inference for biophysical neuron models enables stimulus optimization for retinal neuroprosthetics

eLife, Год журнала: 2020, Номер 9

Опубликована: Окт. 27, 2020

While multicompartment models have long been used to study the biophysics of neurons, it is still challenging infer parameters such from data including uncertainty estimates. Here, we performed Bayesian inference for detailed neuron a photoreceptor and an OFF- ON-cone bipolar cell mouse retina based on two-photon imaging data. We obtained multivariate posterior distributions specifying plausible parameter ranges consistent with allowing identify poorly constrained by To demonstrate potential mechanistic data-driven models, created simulation environment external electrical stimulation optimized stimulus waveforms target cells, current major problem retinal neuroprosthetics.

Язык: Английский

---

Probing and predicting ganglion cell responses to smooth electrical stimulation in healthy and blind mouse retina

Scientific Reports, Год журнала: 2020, Номер 10(1)

Опубликована: Март 23, 2020

Abstract Retinal implants are used to replace lost photoreceptors in blind patients suffering from retinopathies such as retinitis pigmentosa. Patients wearing regain some rudimentary visual function. However, it is severely limited compared normal vision because non-physiological stimulation strategies fail selectively activate different retinal pathways at sufficient spatial and temporal resolution. The development of improved rendered difficult by the large space potential stimuli. Here we systematically explore a subspace stimuli electrically stimulating healthy mouse retina epiretinal configuration using smooth Gaussian white noise delivered high-density CMOS-based microelectrode array. We identify linear filters ganglion cells (RGCs) fitting linear-nonlinear-Poisson (LNP) model. Our stimulus evokes spatially temporally confined spiking responses RGC which accurately predicted LNP Furthermore, find diverse shapes stage model, suggesting preferred electrical RGCs. filter base identified our approach could provide starting point model-guided search for prosthetics.

Язык: Английский

---

Simulation-based inference for efficient identification of generative models in computational connectomics

PLoS Computational Biology, Год журнала: 2023, Номер 19(9), С. e1011406 - e1011406

Опубликована: Сен. 22, 2023

Recent advances in connectomics research enable the acquisition of increasing amounts data about connectivity patterns neurons. How can we use this wealth to efficiently derive and test hypotheses principles underlying these patterns? A common approach is simulate neuronal networks using a hypothesized wiring rule generative model compare resulting synthetic with empirical data. However, most rules have at least some free parameters, identifying parameters that reproduce be challenging as it often requires manual parameter tuning. Here, propose simulation-based Bayesian inference (SBI) address challenge. Rather than optimizing fixed fit data, SBI considers many parametrizations performs identify are compatible It uses simulated from multiple candidate relies on machine learning methods estimate probability distribution (the 'posterior over conditioned data') characterizes all data-compatible parameters. We demonstrate how apply computational by inferring an silico rat barrel cortex, given vivo measurements. identifies wide range show access posterior allows us analyze their relationship, revealing biologically plausible interactions enabling experimentally testable predictions. further applied different spatial scales quantitatively out invalid hypotheses. Our applicable models used connectomics, providing quantitative efficient way constrain

Язык: Английский

---

Controlling morpho-electrophysiological variability of neurons with detailed biophysical models

iScience, Год журнала: 2023, Номер 26(11), С. 108222 - 108222

Опубликована: Окт. 17, 2023

Variability, which is known to be a universal feature among biological units such as neuronal cells, holds significant importance, as, for example, it enables robust encoding of high volume information in circuits and prevents hypersynchronizations. While most computational studies on electrophysiological variability were done with single-compartment neuron models, we instead focus the detailed biophysical models multi-compartmental morphologies. We leverage Markov chain Monte Carlo method generate populations electrical reproducing experimental recordings while being compatible set morphologies faithfully represent specifi morpho-electrical type. demonstrate our approach layer 5 pyramidal cells study particular, find that morphological alone insufficient reproduce variability. Overall, this provides strong statistical basis create neurons controlled

Язык: Английский

---

Avoidance of axonal stimulation with sinusoidal epiretinal stimulation

Journal of Neural Engineering, Год журнала: 2024, Номер 21(2), С. 026036 - 026036

Опубликована: Март 28, 2024

Abstract Objective. Neuromodulation, particularly electrical stimulation, necessitates high spatial resolution to achieve artificial vision with acuity. In epiretinal implants, this is hindered by the undesired activation of distal axons. Here, we investigate focal and axonal retinal ganglion cells (RGCs) in configuration for different sinusoidal stimulation frequencies. Approach. RGC responses at frequencies between 40 100 Hz were tested ex-vivo photoreceptor degenerated (rd10) isolated retinae. Experiments conducted using a high-density CMOS-based microelectrode array, which allows localize cell bodies axons resolution. Main results. We report current charge density thresholds axon 40, 60, 80, an electrode size effective area 0.01 mm 2 . Activation avoided up amplitude 0.23 µ A (corresponding 17.3 C cm −2 ) 0.28 (14.8 60 Hz. The threshold ratio increases from 1.1 1.6 Hz, while frequency, almost no detected intensity range. With use synaptic blockers, demonstrate underlying direct mechanism cells. Finally, high-resolution imaging label-free electrophysiological tracking, extent bundles. Significance. Our results can be exploited define spatially selective strategy avoiding future thereby solving one major limitations vision. may extended other fields neuroprosthetics stimulation.

Язык: Английский

---