Field-theoretic approach to compartmental neuronal networks: impact of dendritic calcium spike-dependent bursting

Audrey Teasley, Gabriel Koch Ocker · 2025

Read Field-theoretic approach to compartmental neuronal networks: impact of dendritic calcium spike-dependent bursting by Audrey Teasley, Gabriel Koch Ocker (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-08405.txt.

Introduction

Audrey O'Brien Teasley and Gabriel Koch Ocker build a field theory for networks of neurons with parts, in a 2025 paper. Real neurons are spread out cells whose zones can spike in unlike ways, and inputs land on set zones. Classic population theories shrink each neuron to a point with one spike kind, missing all that.

Their approach handles compartment neurons with several spike types. They apply it to thick tufted layer five cells, whose top branches fire calcium spikes when two signals coincide. In weak coupling the network maps exactly onto a marked point process. Phase maps then show stable zones with silent or active branches, with or without inhibition, and waves between them.

Catalog record

CREATORAudrey Teasley, Gabriel Koch Ocker
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE85 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-08405

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Field-theoretic approach to compartmental neuronal networks: impact of dendritic calcium spike-dependent bursting Preprint: APS/123-QED Audrey Teasley, Gabriel Koch Ocker Abstract Neurons are spatially extended cells; different parts of a neuron have specific voltage dynamics. Important types of neurons even generate different spikes in different parts of the cell. Neurons’ inputs are also often spatially compartmentalized, with different sources targeting different locations on the cell. Classic mean-field theories for neural population activity, however, rely on point-neuron models with at most one type of spike. Here, we develop a statistical field-theoretic approach to understanding collective activity in networks of compartmental neurons, including those generating multiple types of spikes. We use this to examine simple models of networks with thick-tufted layer 5 pyramidal cells, which generate calcium spikes in their apical dendrite when dendritic depolarization coincides with a back-propagating somatic action potential. In the weakly-coupled regime, we uncover an exact mean-field limit for these networks that maps them to a marked point process. We use this mean-field limit to compare the impact of compartmentalized recurrent excitatory and inhibitory connectivity on the equilibrium phase diagram. This exposes regions of metastability between various activity states, including activity with silent vs active dendrites, with and without inhibitory activity, and oscillations. I Introduction The collective activity of neuronal networks underlies sensory perception, motor planning and execution, learning and memory, and other cognitive functions. Mean-field theories for neuronal population activity explain collective activity using a few order parameters. Classical mean-field theories exhibit a wide variety of dynamics Coombes 2010 . Individual neurons also exhibit a broad range of nonlinear dynamics. The most famous of these may be the sodium-potassium (Na + /K + ) action potential, generated near the cell body, that drives neurotransmitter release to allow fast intercellular communication Eccles 1935 ; Hodgkin and Huxley 1952 . Furthermore, neurons are spatially extended cells. Cortical neurons have richly branched dendritic trees, on which they receive synaptic inputs. Neurons can also generate action potentials in their dendrites Larkum et al. 2022 . For example, consider thick-tufted pyramidal cells, a main class of cortical long-range projection neurons. These neurons have a cell body in layer 5 of the cortex and an apical dendrite that extends up to layer 1 (Fig. 1 a). The apical dendrites of these neurons generate calcium spikes Larkum et al. 1999 . These trigger bursts of somatic action potentials, giving the neuron a parallel output train that can be separately modulated by learning and attention Naud and Sprekeler 2018 ; Friedenberger et al. 2023 . A variety of other important neuron types also generate calcium spikes, including thalamic relay neurons, cerebellar Purkinje cells, and layer 2/3 and hippocampal pyramidal cells Brumberg et al. 2000 ; Schwartzkroin and Slawsky 1977 ; Wong et al. 1979 ; Magee and Carruth 1999 ; Larkum et al. 2007 ; Jahnsen and Llinás 1984 ; Destexhe et al. 1998 ; Eccles et al. 1966 ; Davie et al. 2008 . Different types of synaptic input are also spatially localized on pyramidal cells. In the neocortex, long-range feedforward and feedback projections preferentially target specific layers Felleman and Van Essen Feb ; Harris et al. 2019 . The major classes of cortical inhibitory neurons also preferentially target specific compartments. Somatostatin-positive (SOM) Martinotti cells preferentially synapse onto pyramidal cells’ apical dendrites, while parvalbumin-positive (PV) basket cells synapse onto cell bodies Rudy et al. 2011 . Together, these experimental observations beg the question: how does the spatial compartmentalization of synaptic connectivity interact with somatic and dendritic spiking to shape collective activity in cortical networks? Due to their complexity, large networks of compartmental neurons are generally studied through careful and laborious simulation, e.g., Traub et al. 2005 ; Markram 2006 ; Markram et al. 2015 ; Billeh et al. 2020 . Using simulation to survey the parameter combinations underlying healthy or pathological network dynamics becomes intractable as the complexity of the model grows. Classical mean-field theories of neural activity, on the other hand, rely on single-compartment models that either neglect spiking or model only the somatic Na + /K + spike Grossberg 1969 ; Amari 1971 ; Amari 1972 ; Wilson and Cowan 1972 ; Sompolinsky et al. 1988 ; Ohira and Cowan 1993 ; Ginzburg and Sompolinsky 1994 ; van Vreeswijk and Sompolinsky 1998 ; Brunel 2000 ; Gerstner 1995 ; Renart et al. 2010 ; Montbrió et al. 2015 . Here, we propose a statistical field-theoretic approach for networks of compartmental neurons with stochastic spike emission. We construct the joint density functional of the subthreshold voltages of, and action potentials generated by, each compartment of each neuron in a network. We apply this to networks with a simple model of dendritic calcium spike-dependent bursting in pyramidal cells. The large- N mean-field limit of this density functional exposes exact predictions for the rate and fluctuations of somatic and dendritic action potentials. We use these to study how the compartmentalization of excitatory and inhibitory connectivity controls activity in recurrent networks. II Approach and model We divide each neuron i into a set C_{i} of isopotential spatial compartments. The membrane voltage of compartment c in neuron i is v_{i}^{c}(t) . Rather than describing the nonlinear membrane potential dynamics that generate spikes explicitly, we model them as point processes. Each neuron has a set of point processes associated to it; \dot{a}_{i}^{d}(t) is the point process describing spikes of type d in neuron i , also called a spike train. The intensities of these point processes model the underlying generating mechanisms. For example, the classic Na + /K + action potential generated in the somatic compartment ( c=S ) is typically modeled with rectified power-law voltage-intensity transfer functions: \displaystyle\dot{a}_{i}^{S}(t)\sim \displaystyle\mathcal{PP}\big(f(v_{i}^{S}(t))\big), (1) \displaystyle f(v_{i}^{S}(t))= \displaystyle\lfloor v_{i}^{S}(t)-\theta\rfloor_{+}^{p}, where \theta is a spike threshold and p a power that typically ranges between 1 and 5 in cortical cells Paliwal et al. 2025 . Other choices for the voltage-intensity transfer function f are admissible. We only require \lim_{v\rightarrow\infty}f(v)=\infty so that if the somatic voltage blows up, the neuron emits an action potential with probability one. Here, we focus on a model of thick-tufted layer 5 pyramidal cells with somatic ( S ) and apical dendritic ( D ) compartments (Fig. 1 a). The somatic compartment generates Na + /K + action potentials with intensity f(v_{i}^{S}(t)) (Eq. 1 ). In thick-tufted layer 5 pyramidal cells, dendritic calcium spikes are reliably triggered by the backpropagation of a somatic action potential into a depolarized apical dendrite Larkum et al. 1999 . Following Naud & Sprekeler Naud and Sprekeler 2018 , we model this as \begin{aligned} \dot{a}_{i}^{D}(t)\sim\mathcal{PP}\big(\dot{a}_{i}^{S}(t)\,g(v_{i}^{D}(t)\big)\end{aligned}. (2) where v_{i}^{D} is the dendritic voltage and the dendritic transfer function g(v_{i}^{D}(t)) determines the probability that a somatic Na + /K + spike triggers a dendritic calcium spike. We require that 0\leq g\leq 1 ; it is a probability. To focus on the role of compartmentalized action potential types, we here assume linear sub-threshold voltage dynamics within each compartment and linear (current-based) synaptic coupling so that the membrane voltage of compartment c in neuron i obeys \left(\partial_{t}+1\right)v_{i}^{c}=E_{i}^{c}+\sum_{j=1}^{N}\sum_{d\in C_{j}}J_{ij}^{cd}\ast\dot{a}_{j}^{d}, (3) where E_{i}^{c} is an effective reversal potential, which may also depend on external inputs, and the synaptic filter J_{ij}^{cd}(s) models the impact of an action potential of type d in

Read the complete text