Active intrinsic conductances in recurrent networks allow for long-lasting transients and sustained activity with realistic firing rates as well as robust plasticity

2021 ◽  
Author(s):  
Tuba Aksoy ◽  
Harel Z. Shouval
Keyword(s):  
Recurrent Networks ◽  
Firing Rates ◽  
Sustained Activity

scholarly journals Dynamic population coding and its relationship to working memory

2018 ◽  
Vol 120 (5) ◽  
pp. 2260-2268 ◽  
Author(s):  
Ethan M. Meyers
Keyword(s):  
Working Memory ◽  
Population Coding ◽  
Future Research ◽  
Neural Basis ◽  
Firing Rates ◽  
Sustained Activity ◽  
Early Results ◽  
Population Codes ◽  
Dynamic Population ◽  
Insight Into

For over 45 years, neuroscientists have conducted experiments aimed at understanding the neural basis of working memory. Early results examining individual neurons highlighted that information is stored in working memory in persistent sustained activity where neurons maintained elevated firing rates over extended periods of time. However, more recent work has emphasized that information is often stored in working memory in dynamic population codes, where different neurons contain information at different periods in time. In this paper, I review findings that show that both sustained activity as well as dynamic codes are present in the prefrontal cortex and other regions during memory delay periods. I also review work showing that dynamic codes are capable of supporting working memory and that such dynamic codes could easily be “readout” by downstream regions. Finally, I discuss why dynamic codes could be useful for enabling animals to solve tasks that involve working memory. Although additional work is still needed to know definitively whether dynamic coding is critical for working memory, the findings reviewed here give insight into how different codes could contribute to working memory, which should be useful for guiding future research.

scholarly journals Synaptic Plasticity in Correlated Balanced Networks

2020 ◽  
Author(s):  
Alan Eric Akil ◽  
Robert Rosenbaum ◽  
Krešimir Josić
Keyword(s):  
Neural Networks ◽  
Synaptic Plasticity ◽  
General Theory ◽  
Recurrent Networks ◽  
Weight Changes ◽  
Initial State ◽  
Firing Rates ◽  
Plausible Mechanism ◽  
Induced Changes ◽  
Balanced Networks

AbstractThe dynamics of local cortical networks are irregular, but correlated. Dynamic excitatory– inhibitory balance is a plausible mechanism that generates such irregular activity, but it remains unclear how balance is achieved and maintained in plastic neural networks. In particular, it is not fully understood how plasticity induced changes in the network affect balance, and in turn, how correlated, balanced activity impacts learning. How does the dynamics of balanced networks change under different plasticity rules? How does correlated spiking activity in recurrent networks change the evolution of weights, their eventual magnitude, and structure across the network? To address these questions, we develop a general theory of plasticity in balanced networks. We show that balance can be attained and maintained under plasticity induced weight changes. We find that correlations in the input mildly, but significantly affect the evolution of synaptic weights. Under certain plasticity rules, we find an emergence of correlations between firing rates and synaptic weights. Under these rules, synaptic weights converge to a stable manifold in weight space with their final configuration dependent on the initial state of the network. Lastly, we show that our framework can also describe the dynamics of plastic balanced networks when subsets of neurons receive targeted optogenetic input.

scholarly journals Cyclic transitions between higher order motifs underlie sustained activity in asynchronous sparse recurrent networks

2019 ◽  
Author(s):  
Kyle Bojanek ◽  
Yuqing Zhu ◽  
Jason MacLean
Keyword(s):  
Neural Network ◽  
Higher Order ◽  
Network Activity ◽  
Synaptic Connectivity ◽  
Functional Networks ◽  
Dynamical Regime ◽  
Spiking Neural Network ◽  
Recurrent Networks ◽  
Sustained Activity ◽  
Spiking Activity

AbstractMany studies have demonstrated the prominence of higher-order patterns in excitatory synaptic connectivity as well as activity in neocortex. Surveyed as a whole, these results suggest that there may be an essential role for higher-order patterns in neocortical function. In order to stably propagate signal within and between regions of neocortex, the most basic - yet nontrivial - function which neocortical circuitry must satisfy is the ability to maintain stable spiking activity over time. Here we algorithmically construct spiking neural network models comprised of 5000 neurons using topological statistics from neocortex and a set of objective functions that identify networks which produce naturalistic low-rate, asynchronous, and critical activity. We find that the same network topology can exhibit either sustained activity under one set of initial membrane voltages or truncated activity under a different set. Yet these two outcomes are not readily differentiated by rate or criticality. By summarizing the statistical dependencies in the pairwise activity of neurons as directed weighted functional networks, we examined the transient manifestations of higher-order motifs in the functional networks across time. We find that stereotyped low variance cyclic transitions between three isomorphic triangle motifs, quantified as a Markov process, are required for sustained activity. If the network fails to engage the dynamical regime characterized by a recurring stable pattern of motif dominance, spiking activity ceased. Motif cycling generalized across manipulations of synaptic weights and across topologies, demonstrating the robustness of this dynamical regime for sustained spiking in critical asynchronous network activity. Our results point to the necessity of higher-order patterns amongst excitatory connections for sustaining activity in sparse recurrent networks. They also provide a possible explanation as to why such excitatory synaptic connectivity and activity patterns have been prominently reported in neocortex.Author summaryHere we address two questions. First, it remains unclear how activity propagates stably through a network since neurons are leaky and connectivity is sparse and weak. Second, higher order patterns abound in neocortex, hinting at potential functional relevance for their presence. Several lines of evidence suggest that higher-order network interactions may be instrumental for spike propagation. For example, excitatory synaptic connectivity shows a prevalence of local neuronal cliques and patterns, and propagating activity in vivo displays elevated clustering dominated by specific triplet motifs. In this study we demonstrate a mechanistic link between activity propagation and higher-order motifs at the level of individual neurons and across networks. We algorithmically build spiking neural network (SNN) models to mirror the topological and dynamical statistics of neocortex. Using a combination of graph theory, information theory, and probabilistic tools, we show that higher order coordination of synapses is necessary for sustaining activity. Coordination takes the form of cyclic transitions between specific triangle motifs. The results of our model are consistent with numerous experimental observations in neuroscience, and their generalizability to other weakly and sparsely connected networks is predicted.

scholarly journals When do correlations increase with firing rates in recurrent networks?

2017 ◽  
Vol 13 (4) ◽  
pp. e1005506 ◽  
Author(s):  
Andrea K. Barreiro ◽  
Cheng Ly
Keyword(s):  
Recurrent Networks ◽  
Firing Rates

scholarly journals Exact Solutions for Rate and Synchrony in Recurrent Networks of Coincidence Detectors

2008 ◽  
Vol 20 (11) ◽  
pp. 2637-2661 ◽  
Author(s):  
Shawn Mikula ◽  
Ernst Niebur
Keyword(s):  
Analytical Solutions ◽  
Single Neuron ◽  
State Transition ◽  
Transition Probabilities ◽  
Population Coding ◽  
Recurrent Networks ◽  
Firing Rates ◽  
Finite State ◽  
Cross Correlations ◽  
The Relationship

We provide analytical solutions for mean firing rates and cross-correlations of coincidence detector neurons in recurrent networks with excitatory or inhibitory connectivity, with rate-modulated steady-state spiking inputs. We use discrete-time finite-state Markov chains to represent network state transition probabilities, which are subsequently used to derive exact analytical solutions for mean firing rates and cross-correlations. As illustrated in several examples, the method can be used for modeling cortical microcircuits and clarifying single-neuron and population coding mechanisms. We also demonstrate that increasing firing rates do not necessarily translate into increasing cross-correlations, though our results do support the contention that firing rates and cross-correlations are likely to be coupled. Our analytical solutions underscore the complexity of the relationship between firing rates and cross-correlations.

scholarly journals Neural Integrator: A Sandpile Model

2008 ◽  
Vol 20 (10) ◽  
pp. 2379-2417 ◽  
Author(s):  
Maxim Nikitchenko ◽  
Alexei Koulakov
Keyword(s):  
Nmda Receptor ◽  
Angular Velocity ◽  
Positive Feedback ◽  
Neuronal Response ◽  
Eye Position ◽  
Recurrent Networks ◽  
Neural Integrator ◽  
Intrinsic Properties ◽  
Firing Rates ◽  
Spontaneous Noise

We investigated a model for the neural integrator based on hysteretic units connected by positive feedback. Hysteresis is assumed to emerge from the intrinsic properties of the cells. We consider the recurrent networks containing either bistable or multistable neurons. We apply our analysis to the oculomotor velocity-to-position neural integrator that calculates eye positions using the inputs that carry information about eye angular velocity. By analyzing this system in the parameter space, we show the following. The direction of hysteresis in the neuronal response may be reversed for the system with recurrent connections compared to the case of unconnected neurons. Thus, for the NMDA receptor-based bistability, the firing rates after ON saccades may be higher than after OFF saccades for the same eye position. The reversal of hysteresis occurs in this model only when the size of hysteresis differs from neuron to neuron. We also relate the macroscopic leak time constant of the integrator to the rate of microscopic spontaneous noise-driven transitions in the hysteretic units. Finally, we investigate the conditions under which the hysteretic integrator may have no threshold for integration.

Gated Recurrent Networks for Video Super Resolution

2021 ◽  
Author(s):  
Santiago Lopez-Tapia ◽  
Alice Lucas ◽  
Rafael Molina ◽  
Aggelos K. Katsaggelos
Keyword(s):  
Super Resolution ◽  
Recurrent Networks

scholarly journals Differentiable Mask for Pruning Convolutional and Recurrent Networks

2020 ◽  
Author(s):  
Ramchalam Kinattinkara Ramakrishnan ◽  
Eyyub Sari ◽  
Vahid Partovi Nia
Keyword(s):  
Recurrent Networks

scholarly journals A deep convolutional visual encoding model of neuronal responses in the LGN

2021 ◽  
Vol 8 (1) ◽  
Author(s):  
Eslam Mounier ◽  
Bassem Abdullah ◽  
Hani Mahdi ◽  
Seif Eldawlatly
Keyword(s):  
Firing Rate ◽  
Visual Information ◽  
Visual Stimulation ◽  
Neuronal Population ◽  
Neuronal Firing ◽  
Electrode Arrays ◽  
Deep Convolutional Neural Networks ◽  
Firing Rates ◽  
Spatiotemporal Representation

AbstractThe Lateral Geniculate Nucleus (LGN) represents one of the major processing sites along the visual pathway. Despite its crucial role in processing visual information and its utility as one target for recently developed visual prostheses, it is much less studied compared to the retina and the visual cortex. In this paper, we introduce a deep learning encoder to predict LGN neuronal firing in response to different visual stimulation patterns. The encoder comprises a deep Convolutional Neural Network (CNN) that incorporates visual stimulus spatiotemporal representation in addition to LGN neuronal firing history to predict the response of LGN neurons. Extracellular activity was recorded in vivo using multi-electrode arrays from single units in the LGN in 12 anesthetized rats with a total neuronal population of 150 units. Neural activity was recorded in response to single-pixel, checkerboard and geometrical shapes visual stimulation patterns. Extracted firing rates and the corresponding stimulation patterns were used to train the model. The performance of the model was assessed using different testing data sets and different firing rate windows. An overall mean correlation coefficient between the actual and the predicted firing rates of 0.57 and 0.7 was achieved for the 10 ms and the 50 ms firing rate windows, respectively. Results demonstrate that the model is robust to variability in the spatiotemporal properties of the recorded neurons outperforming other examined models including the state-of-the-art Generalized Linear Model (GLM). The results indicate the potential of deep convolutional neural networks as viable models of LGN firing.

Sign in / Sign up

Export Citation Format

Share Document