Abstract
In the study of brain connectivity, an accessible and convenient way to unveil local functional structures is to infer the time trace of synaptic conductances received by a neuron by using exclusively information about its membrane potential (or voltage). Mathematically speaking, it constitutes a challenging inverse problem: it consists in inferring time-dependent parameters (synaptic conductances) departing from the solutions of a dynamical system that models the neuron’s membrane voltage. Several solutions have been proposed to perform these estimations when the neuron fluctuates mildly within the subthreshold regime, but very few methods exist for the spiking regime as large amplitude oscillations (revealing the activation of complex nonlinear dynamics) hinder the adaptability of subthreshold-based computational strategies (mostly linear). In a previous work, we presented a mathematical proof-of-concept that exploits the analytical knowledge of the period function of the model. Inspired by the relevance of the period function, in this paper we generalize it by providing a computational strategy that can potentially adapt to a variety of models as well as to experimental data. We base our proposal on the frequency versus synaptic conductance curve (f−gsyn), derived from an analytical study of a base model, to infer the actual synaptic conductance from the interspike intervals of the recorded voltage trace. Our results show that, when the conductances do not change abruptly on a time-scale smaller than the mean interspike interval, the time course of the synaptic conductances is well estimated. When no base model can be cast to the data, our strategy can be applied provided that a suitable f−gsyn table can be experimentally constructed. Altogether, this work opens new avenues to unveil local brain connectivity in spiking (nonlinear) regimes.
Highlights
Estimating non-measurable connectivity parameters from experimental measurements is an important challenge to understand brain function
As we know that, by construction, the actual membrane potential follows the pyramidal neuron model, we consider this model as the base model, so we can test how powerful the estimation procedure is if we know the exact neural model
We found the exponential integrate-and-fire model (EIF) to be a convenient base model: on the one hand, its oscillation period can be well approximated by means of the semi-analytical formula (6); on the other, there are examples in the literature showing that its f − I curve can fit f − I curves of neural data
Summary
Estimating non-measurable connectivity parameters from experimental measurements is an important challenge to understand brain function. Brain activity is mainly driven by the intrinsic dynamics of neurons and their mutual connections, which are essentially regulated by synapses. A thorough knowledge of individual membrane potentials and synaptic terms is, crucial to understand how the brain operates. We are far from this utopia: current recording methods can only retrieve partial information, such as the time evolution of membrane potential of one or a few neurons. Other quantities that shape the connectivity, like synaptic conductances, are impossible to be directly recorded in experiments.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
