ABSTRACT
Neurons in the brain are submerged into oscillating extracellular Local Field Potential (LFP) created by
synchronized synaptic currents. The dynamics of these oscillations is one of the principal characteristics
of the brain activity at all levels: from the individual neurons’ spiking to the activity of networks that underlie
high-level cognitive processes. However, our interpretation of the LFP structure and functions depend on
the techniques that we use for data analyses. The oscillatory nature of LFP motivates using Fourier
methods, which have dominated LFP research for decades and currently constitute the only systematic
framework for understanding the “brain rhythms.” Yet these methods poorly handle two fundamental
attributes of biological signals: noise and non-stationarity, and may therefore obscure the structure of the
LFP data and its physiological meaning. We have recently adapted a powerful technique that previously
applied to studying complex physical signals (e.g., gravitational waves, magnetic resonances, etc.) for
nuanced analysis of the LFP oscillations. By applying this method, we discovered that hippocampal and
cortical LFPs recorded in rodents consist of a few frequency-modulated waves, which we call Oscillons.
We hypothesize that these objects represent the actual, physical structure of the brain waves and hence
may hold keys to better understanding of the circuit mechanisms of learning and memory. Another
principal feature of our method is an impartial marker of the noise component, which allows us to identify
and remove the “noise shell” from the signal and then to investigate not only the noise itself, but also the
interplays between the noise and the regular, oscillatory part of the signal, their interactions with neuronal
spiking, etc.
Since Alzheimer’s Disease (AD) is characterized by alterations in both the oscillatory and stochastic
activity in the hippocampal network, the quest of better understanding of AD-induced pathologies fits
ideally the strengths of our approach. Our goal is to use it for studying the circuit mechanisms of AD and
to learn to manipulate the network activity through our methodology.