Is realistic neuronal modeling realistic?

Abstract

Scientific models are abstractions that aim to explain natural phenomena. A successful model shows how a complex phenomenon arises from relatively simple principles while preserving major physical or biological rules and predicting novel experiments. A model should not be a facsimile of reality; it is an aid for understanding it. Contrary to this basic premise, with the 21st century has come a surge in computational efforts to model biological processes in great detail. Here we discuss the oxymoronic, realistic modeling of single neurons. This rapidly advancing field is driven by the discovery that some neurons don't merely sum their inputs and fire if the sum exceeds some threshold. Thus researchers have asked what are the computational abilities of single neurons and attempted to give answers using realistic models. We briefly review the state of the art of compartmental modeling highlighting recent progress and intrinsic flaws. We then attempt to address two fundamental questions. Practically, can we realistically model single neurons? Philosophically, should we realistically model single neurons? We use layer 5 neocortical pyramidal neurons as a test case to examine these issues. We subject three publically available models of layer 5 pyramidal neurons to three simple computational challenges. Based on their performance and a partial survey of published models, we conclude that current compartmental models are ad hoc, unrealistic models functioning poorly once they are stretched beyond the specific problems for which they were designed. We then attempt to plot possible paths for generating realistic single neuron models.

Multimodal determinants of phase-locked dynamics across deep-superficial hippocampal sublayers during theta oscillations

“…Our simulations were based on a realistic full-scale model of the CA1 microcircuit able to autonomously generate theta oscillations 31 . To constrain the parametric space further (>16 free parameters per cell), we adopted genetic algorithms (GA) 32,33 to identify values for passive, active and synaptic conductances (dubbed gene factors) resulting in realistic behavior in a given morphology. For instance, more than 4000 intrinsic parameters fitted experimental somatic and dendritic responses to current pulses ( Fig.2B; black dots), providing individuals (i.e.…”

“…To preserve the biologically-inspired somatodendritic distribution of the different ionic channels, gene factors multiply maximal conductances in a given cell, independent on the compartment. Gene factors for ionic channel conductance were chosen to fit the neuron intrinsic properties [32][33][34][35] . Gene factors of synaptic conductance of the CA3 inputs and their associated feedforward inhibitory inputs (Axo, Bis, CCK, PV and SCA) were selected to target experimental data 34 .…”

“…SOM represents a predictive model that needs to be validated. Thus, for each mean element of the matrix we represented the time-frequency components using wavelet analysis and used information from the similarity matrix to identify theta-nested spectral components tSC1 (<20Hz centered at the theta peak), tSC2 (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) Hz, at the falling phase), tSC3 (50-60 Hz; at the theta peak) and tSC4 (70-90 Hz at the peak), as validated recently 39 . Cycles from adjacent validated elements were combined together and used for subsequent phase-locked firing analysis.…”

Inferring neuronal ionic conductances from membrane potentials using CNNs

“…The main challenge in fitting neuronal models to the waveforms is that multiple sets of parameters may lead to the similar neuronal output [25,6], which is caused by model non-linearity and/or correlation between the parameters. We are aiming to find the solutions a recorded neuron is using.…”

“…However, determining these parameters empirically is technically challenging, especially across complete neuronal arbors [3,4,5]. This poses a major challenge to simulate neurons using computational methods [6]. A common approach to overcome this challenge is to use neuronal models with ion channel densities described by adjustable parameters [7,8,9,10] .…”