Stimulus Dependence of Local Field Potential Spectra: Experiment versus Theory

Francesca Barbieri, Alberto Mazzoni, Nikos K. Logothetis, Stefano Panzeri, Nicolas BrunelJournal of Neuroscience 2014 · DOI

Abstract

The local field potential (LFP) captures different neural processes, including integrative synaptic dynamics that cannot be observed by measuring only the spiking activity of small populations. Therefore, investigating how LFP power is modulated by external stimuli can offer important insights into sensory neural representations. However, gaining such insight requires developing data-driven computational models that can identify and disambiguate the neural contributions to the LFP. Here, we investigated how networks of excitatory and inhibitory integrate-and-fire neurons responding to time-dependent inputs can be used to interpret sensory modulations of LFP spectra. We computed analytically from such models the LFP spectra and the information that they convey about input and used these analytical expressions to fit the model to LFPs recorded in V1 of anesthetized macaques (Macaca mulatta) during the presentation of color movies. Our expressions explain 60%-98% of the variance of the LFP spectrum shape and its dependency upon movie scenes and we achieved this with realistic values for the best-fit parameters. In particular, synaptic best-fit parameters were compatible with experimental measurements and the predictions of firing rates, based only on the fit of LFP data, correlated with the multiunit spike rate recorded from the same location. Moreover, the parameters characterizing the input to the network across different movie scenes correlated with cross-scene changes of several image features. Our findings suggest that analytical descriptions of spiking neuron networks may become a crucial tool for the interpretation of field recordings.

Introduction

In systems neuroscience, researchers typically record two types of signals: discrete action potentials ("spikes") from individual neurons and Local Field Potentials (LFPs), which represent the low-frequency, continuous electrical activity of the surrounding tissue. While spikes offer high resolution for small populations, LFPs capture the integrated synaptic activity of thousands of neurons. However, LFPs are notoriously difficult to interpret because they are an ambiguous mixture of diverse neural processes. Traditionally, "inverting" the LFP to understand underlying network dynamics—such as the balance of excitation and inhibition—has required computationally intensive simulations. This study presents a rigorous mathematical framework to bridge network theory and experimental LFP data.

Methods and Modeling

The researchers developed analytical, closed-form equations for the LFP power spectrum, moving beyond black-box simulations. The model utilizes a recurrent network of Leaky Integrate-and-Fire (LIF) neurons, comprising both excitatory (E) and inhibitory (I) populations.

The network is driven by time-dependent external inputs modeled as an Ornstein-Uhlenbeck (OU) process, mimicking the statistics of natural visual signals. The LFP is modeled as a proxy: the sum of the absolute values of all synaptic currents (AMPA and GABA) arriving at excitatory neurons. By applying linear response theory, the authors calculated how finite-size fluctuations in a network of $N$ neurons create specific resonances in the LFP.

Simulation of network dynamics showing external input, E/I neuron activity, firing rates, and the resulting LFP proxy.

To validate the model, electrophysiological recordings (LFP and Multi-Unit Activity) were taken from the primary visual cortex (V1) of three anesthetized rhesus macaques (Macaca mulatta) during the presentation of naturalistic color movies. These movies were segmented into 100 distinct 2-second scenes, from which visual features like luminance, contrast, and orientation were extracted.

Key Insights and Results

The analytical approach revealed that the LFP power spectrum is a "two-part story." Low frequencies are primarily driven by external stimuli, whereas high-frequency "gamma" peaks (40–100 Hz) emerge from internal network resonances—specifically the interplay between E and I populations. Notably, while inhibitory neurons are often considered "invisible" in LFPs due to their morphology, this model demonstrates that their influence is clearly visible in the shaping of network dynamics.

The model demonstrated exceptional explanatory power, accounting for 60% to 98% of the variance in LFP spectrum shapes across different movie scenes. Furthermore, the "best-fit" parameters—such as synaptic time constants—were highly consistent with known physiological values. Crucially, firing rates predicted solely from LFP data correlated significantly (average $r \approx 0.5$) with actual recorded Multi-Unit Activity (MUA), proving that the LFP contains sufficient information to reconstruct population spiking behavior. Finally, network input parameters correlated strongly with image features, particularly temporal contrast.

Discussion and Future Directions

This work suggests a shift toward "quantitative spectroscopy" of the brain, where LFPs serve as a readable map of network function. However, limitations remain. The current model assumes a spatially simple "point" network, neglecting the complex layered architecture of V1. Additionally, as the data were collected under anesthesia, future research must validate these findings in awake, behaving animals where brain dynamics, such as gamma oscillations, are more coherent. Addressing potential "spike leakage" into high-frequency LFP bands could also further refine the model's accuracy.

Summary updated

Abstract

The local field potential (LFP) captures different neural processes, including integrative synaptic dynamics that cannot be observed by measuring only the spiking activity of small populations. Therefore, investigating how LFP power is modulated by external stimuli can offer important insights into sensory neural representations. However, gaining such insight requires developing data-driven computational models that can identify and disambiguate the neural contributions to the LFP. Here, we investigated how networks of excitatory and inhibitory integrate-and-fire neurons responding to time-dependent inputs can be used to interpret sensory modulations of LFP spectra. We computed analytically from such models the LFP spectra and the information that they convey about input and used these analytical expressions to fit the model to LFPs recorded in V1 of anesthetized macaques (Macaca mulatta) during the presentation of color movies. Our expressions explain 60%-98% of the variance of the LFP spectrum shape and its dependency upon movie scenes and we achieved this with realistic values for the best-fit parameters. In particular, synaptic best-fit parameters were compatible with experimental measurements and the predictions of firing rates, based only on the fit of LFP data, correlated with the multiunit spike rate recorded from the same location. Moreover, the parameters characterizing the input to the network across different movie scenes correlated with cross-scene changes of several image features. Our findings suggest that analytical descriptions of spiking neuron networks may become a crucial tool for the interpretation of field recordings.
BibTeX
@article{barbieri2014stimulus,
  author = {Francesca Barbieri and Alberto Mazzoni and Nikos K. Logothetis and Stefano Panzeri and Nicolas Brunel},
  title = {{Stimulus Dependence of Local Field Potential Spectra: Experiment versus Theory}},
  journal = {Journal of Neuroscience},
  year = {2014},
  doi = {10.1523/jneurosci.5365-13.2014},
}
Figure