Non-Sinusoidal Activity Can Produce Cross-Frequency Coupling in Cortical Signals in the Absence of Functional Interaction between Neural Sources.

Edden M Gerber,Andrew Ward,Leon Y Deouell,Boaz Sadeh,Robert T Knight

doi:10.1371/journal.pone.0167351

Abstract

The analysis of cross-frequency coupling (CFC) has become popular in studies involving intracranial and scalp EEG recordings in humans. It has been argued that some cases where CFC is mathematically present may not reflect an interaction of two distinct yet functionally coupled neural sources with different frequencies. Here we provide two empirical examples from intracranial recordings where CFC can be shown to be driven by the shape of a periodic waveform rather than by a functional interaction between distinct sources. Using simulations, we also present a generalized and realistic scenario where such coupling may arise. This scenario, which we term waveform-dependent CFC, arises when sharp waveforms (e.g., cortical potentials) occur throughout parts of the data, in particular if they occur rhythmically. Since the waveforms contain both low- and high-frequency components, these components can be inherently phase-aligned as long as the waveforms are spaced with appropriate intervals. We submit that such behavior of the data, which seems to be present in various cortical signals, cannot be interpreted as reflecting functional modulation between distinct neural sources without additional evidence. In addition, we show that even low amplitude periodic potentials that cannot be readily observed or controlled for, are sufficient for significant CFC to occur.

Highlights

Phase-amplitude cross-frequency coupling (CFC) refers to a dependence between the phase of a slow frequency ("frequency-for-phase") and the power of higher-frequency activity ("frequency-for-amplitude") recorded from the brain [1]–[3]
To examine how phase-amplitude coupling is generated in signals merely through the introduction of semi-periodic potentials, we generated 8 Gaussian spike trains corresponding to doi:10.1371/journal.pone.0167351.g001
We examine what happens to the phase of the slow oscillation contained in the background signal following the addition of the spike trains: Fig 3 shows what happens when a series of 10Hz periodic potentials with a magnitude of 1.5STDs of the background scalp EEG signal is injected into the background signal which already features a prominent 10Hz oscillation due to physiological alpha rhythms recorded over the occipital scalp

Summary

Introduction

Phase-amplitude cross-frequency coupling (CFC) refers to a dependence between the phase of a slow frequency ("frequency-for-phase") and the power of higher-frequency activity ("frequency-for-amplitude") recorded from the brain [1]–[3]. A recent theoretical account by Aru et al posits that any non-stationary process in the signal can affect both the phase of a low-frequency component and the amplitude of a high-frequency component, generating spectral dependencies which would be interpreted as CFC, even though they are driven by a single source [12]. Kramer and colleagues argued that under certain circumstances sharp edges in the data may result in coupling in a wide range of frequencies-for-amplitude [13]. This type of coupling has not been demonstrated until very recently In this report we present real-world examples of intracranial EEG recordings where CFC is likely to be caused by the temporal characteristics of a single process rather than by an interaction between two processes, and demonstrate how such a scenario can realistically be manifested in any EEG signal

Methods

Results

Conclusion