Randomly connected networks generate emergent selectivity and predict decoding properties of large populations of neurons

Sederberg, Audrey; Nemenman, Ilya

doi:10.1371/journal.pcbi.1007875

Cited by 18 publications

(12 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The evolution of neuronal networks in vitro was characterized by coherent maturation of synaptic connectivity and ECM expression, partially resembling neuronal circuit development in vivo [14]. Synaptic connectivity of neuronal networks in vitro established randomly, mimicking the organization of cortical networks [3,61]. During cultivation, synapse density increased until maturity was reached after 21 days in vitro (DIV) (Fig.…”

Section: Cell Culturesmentioning

confidence: 89%

Inhibitory control in neuronal networks relies on the extracellular matrix integrity

Dzyubenko¹,

Fleischer²,

Manrique-Castano³

et al. 2021

Cell. Mol. Life Sci.

View full text Add to dashboard Cite

Inhibitory control is essential for the regulation of neuronal network activity, where excitatory and inhibitory synapses can act synergistically, reciprocally, and antagonistically. Sustained excitation-inhibition (E-I) balance, therefore, relies on the orchestrated adjustment of excitatory and inhibitory synaptic strength. While growing evidence indicates that the brain’s extracellular matrix (ECM) is a crucial regulator of excitatory synapse plasticity, it remains unclear whether and how the ECM contributes to inhibitory control in neuronal networks. Here we studied the simultaneous changes in excitatory and inhibitory connectivity after ECM depletion. We demonstrate that the ECM supports the maintenance of E-I balance by retaining inhibitory connectivity. Quantification of synapses and super-resolution microscopy showed that depletion of the ECM in mature neuronal networks preferentially decreases the density of inhibitory synapses and the size of individual inhibitory postsynaptic scaffolds. The reduction of inhibitory synapse density is partially compensated by the homeostatically increasing synaptic strength via the reduction of presynaptic GABAB receptors, as indicated by patch-clamp measurements and GABAB receptor expression quantifications. However, both spiking and bursting activity in neuronal networks is increased after ECM depletion, as indicated by multi-electrode recordings. With computational modelling, we determined that ECM depletion reduces the inhibitory connectivity to an extent that the inhibitory synapse scaling does not fully compensate for the reduced inhibitory synapse density. Our results indicate that the brain’s ECM preserves the balanced state of neuronal networks by supporting inhibitory control via inhibitory synapse stabilization, which expands the current understanding of brain activity regulation. Graphic abstract

show abstract

Section: Cell Culturesmentioning

confidence: 89%

Inhibitory control in neuronal networks relies on the extracellular matrix integrity

Dzyubenko¹,

Fleischer²,

Manrique-Castano³

et al. 2021

Cell. Mol. Life Sci.

View full text Add to dashboard Cite

show abstract

“…they point to the same decision. We found that this task could be implemented by a unit-rank, single population network similar to perceptual decision-making (see also [Sederberg and Nemenman, 2020]), the only difference being that the two modalities correspond to two independent input patterns ( Supplementary Fig. S8).…”

Section: Implications For Structure In Neural Selectivitymentioning

confidence: 91%

The role of population structure in computations through neural dynamics

Dubreuil

Valente

Beirán

et al. 2020

Preprint

View full text Add to dashboard Cite

AbstractNeural computations are currently investigated using two competing approaches: sorting neurons into functional classes, or examining the low-dimensional dynamics of collective activity. Whether and how these two aspects interact to shape computations is currently unclear. Using a novel approach to extract computational mechanisms from networks trained with machine-learning tools on neuroscience tasks, here we show that the dimensionality of the dynamics and cell-class structure play fundamentally complementary roles. While various tasks can be implemented by increasing the dimensionality in networks consisting of a single global population, flexible input-output mappings instead required networks to be organized into several sub-populations. Our analyses revealed that the subpopulation structure enabled flexible computations through a mechanism based on gain-controlled modulations that flexibly shape the dynamical landscape of collective dynamics. Our results lead to task-specific predictions for the structure of neural selectivity and inactivation experiments.

show abstract

“…In the mean-field model, we assume that choice selectivity of inhibitory neurons arises from specific connections from choice-selective excitatory neurons (Σ EI in our model). While it is possible that choice selectivity could arise from external inputs to interneurons 34 or even from random connections between excitatory and inhibitory neurons 35 , most circuit models assume stimulus information is exclusively provided by inputs to excitatory neurons. Inhibitory choice-selectivity also emerged in our RNNs trained to perform 2AFC task 25 .…”

Section: Discussionmentioning

confidence: 99%

Two roles for choice selective inhibition in decision-making circuits

Roach

Churchland

Engel

2022

Preprint

View full text Add to dashboard Cite

During perceptual decision-making, the firing rates of cortical neurons reflect upcoming choices. Recent work showed that excitatory and inhibitory neurons are equally selective for choice. However, functional consequences of inhibitory choice selectivity in decision-making circuits are unknown. We developed a circuit model of decision-making which accounts for the specificity of inputs to and outputs from inhibitory neurons. We found that selective inhibition expands the space of circuits supporting decision-making, allowing for weaker or stronger recurrent excitation when connected in a competitive or feedback motif. The specificity of inhibitory outputs sets the trade-off between speed and accuracy of decisions by altering the attractor dynamics in the circuit. Recurrent neural networks trained to make decisions display the same dependence on inhibitory specificity and the strength of recurrent excitation. Our results reveal two concurrent roles for selective inhibition in decision-making circuits: stabilizing strongly connected excitatory populations and maximizing competition between oppositely selective populations.

show abstract

Randomly connected networks generate emergent selectivity and predict decoding properties of large populations of neurons

Cited by 18 publications

References 37 publications

Inhibitory control in neuronal networks relies on the extracellular matrix integrity

Inhibitory control in neuronal networks relies on the extracellular matrix integrity

The role of population structure in computations through neural dynamics

Two roles for choice selective inhibition in decision-making circuits

Contact Info

Product

Resources

About