Neuromodulation in Small Networks

Williams, Alex H.; Hamood, Albert W.; Marder, Eve

doi:10.1007/978-1-0716-1006-0_26

Cited by 2 publications

(2 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Neuromodulators are ubiquitous control signals for causing networks to switch between states (Gutierrez & Marder, 2014; Marder et al, 2014). While they have been extensively analyzed in small CPGs (Golowasch, 2019; Williams et al, 2022), they also play a key role in setting network learning rates (Doya, 2002; Jepma et al, 2016; Rutledge et al, 2009). One of the key mechanisms by which neuromodulators affect neuronal activity is by modulating ion channels, which can be done by channel insertion or by phosphorylation (Nadim & Bucher, 2014).…”

Section: Discussionmentioning

confidence: 99%

Mechanisms and implications of high depolarization baseline offsets in conductance-based neuronal models

Kumar,

Shahul,

Bhalla

2024

Preprint

View full text Add to dashboard Cite

Somatic current-step injection is a routinely used protocol to characterize neurons and measure their electrophysiological properties. A signature feature of the responses of many neuronal types is an elevated baseline of action potential firing from the resting membrane potential, the depolarization baseline level offset (DBLO). We find that four key factors together account for high DBLO: Liquid Junction Potential correction, subthreshold impedance amplitude profile, fast potassium delayed rectifier kinetics, and appropriate transient sodium current kinetics. We show that simple mechanisms for DBLO, such as Ohmic depolarization due to current input, fail to explain the effect, and many sophisticated conductance-based models also do not correctly manifest DBLO. Neither low pass filtering effect of membrane nor high reversal potential of potassium channels are able to explain high DBLO. Using stochastic parameter search in conductance-based models of CA1 pyramidal neurons, we explore cellular morphology configurations and channel kinetics. Multi-compartment models which matched experimental subthreshold impedance amplitude profiles had higher DBLO than single compartment models. DBLO was further increased in models that exhibited rapid deactivation time-constants in the dominant potassium conductance. We also saw that certain transient sodium current kinetics resulted in higher DBLO than others. We emphasize that correct expression of DBLO in conductance-based models is important to make quantitative predictions about the levels of ion channels in a neuron and also to correctly predict mechanisms underlying cellular function, such as the role of persistent sodium in imparting firing bistability to a neuron.

show abstract

Section: Discussionmentioning

confidence: 99%

Mechanisms and implications of high depolarization baseline offsets in conductance-based neuronal models

Kumar,

Shahul,

Bhalla

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Our work builds on a body of literature dating back to the 1980s, when pioneering computational neuroscientists added neuromodulation to small biophysical circuit models (for reviews, see [7][8][9]). These models consist of coupled differential equations whose biophysical parameters are carefully specified to simulate biologicallyaccurate spiking activity.…”

Section: Modeling Neuromodulatory Signalsmentioning

confidence: 99%

Structured flexibility in recurrent neural networks via neuromodulation

Costacurta,

Bhandarkar,

Zoltowski

et al. 2024

Preprint

View full text Add to dashboard Cite

The goal of theoretical neuroscience is to develop models that help us better understand biological intelligence. Such models range broadly in complexity and biological detail. For example, task-optimized recurrent neural networks (RNNs) have generated hypotheses about how the brain may perform various computations, but these models typically assume a fixed weight matrix representing the synaptic connectivity between neurons. From decades of neuroscience research, we know that synaptic weights are constantly changing, controlled in part by chemicals such as neuromodulators. In this work we explore the computational implications of synaptic gain scaling, a form of neuromodulation, using task-optimized low-rank RNNs. In our neuromodulated RNN (NM-RNN) model, a neuromodulatory subnetwork outputs a low-dimensional neuromodulatory signal that dynamically scales the low-rank recurrent weights of an output-generating RNN. In empirical experiments, we find that the structured flexibility in the NM-RNN allows it to both train and generalize with a higher degree of accuracy than low-rank RNNs on a set of canonical tasks. Additionally, via theoretical analyses we show how neuromodulatory gain scaling endows networks with gating mechanisms commonly found in artificial RNNs. We end by analyzing the low-rank dynamics of trained NM-RNNs, to show how task computations are distributed.

show abstract

Neuromodulation in Small Networks

Cited by 2 publications

References 61 publications

Mechanisms and implications of high depolarization baseline offsets in conductance-based neuronal models

Mechanisms and implications of high depolarization baseline offsets in conductance-based neuronal models

Structured flexibility in recurrent neural networks via neuromodulation

Contact Info

Product

Resources

About