Brain synchronizability, a false friend

Abstract

Synchronization plays a fundamental role in healthy cognitive and motor function. However, how synchronization depends on the interplay between local dynamics, coupling and topology and how prone to synchronization a network with given topological organization is are still poorly understood issues. To investigate the synchronizability of both anatomical and functional brain networks various studies resorted to the Master Stability Function (MSF) formalism, an elegant tool which allows analysing the stability of synchronous states in a dynamical system consisting of many coupled oscillators. Here, we argue that brain dynamics does not fulfil the formal criteria under which synchronizability is usually quantified and, perhaps more importantly, what this measure itself quantifies refers to a global dynamical condition that never holds in the brain (not even in the most pathological conditions), and therefore no neurophysiological conclusions should be drawn based on it. We discuss the meaning of synchronizability and its applicability to neuroscience and propose alternative ways to quantify brain networks synchronization.

Optimization of epilepsy surgery through virtual resections on individual structural brain networks

“…where it is used to model the spread of disease in a population, modeled as a contact network (Kermack and McKendrick, 1927;Pastor-Satorras et al, 2015). Here, we applied this epidemiologic model to the propagation of seizure activity in a brain network.…”

“…To quantify the speed of the propagation of activity we calculated the fraction of active nodes at time step 10 (It=10) (supplementary figure S1 C). The fraction of active nodes can be plotted over time and will give a characteristic curve (specifically: a hyperbolic secant) when the spreading rate λ = beta / gamma is above the epidemic threshold of λc (Pastor-Satorras et al, 2015).…”

“…The authors calculated synchronizability on functional networks derived from electrocorticography (ECoG) measurements and found that the change in synchronizability after virtually removing the resection area predicted surgical outcome, although this result was not significant after multiple comparisons corrections. However, synchronizability is defined for structural networks and requires several assumptions that are not fulfilled in brain data, such as identical dynamical units (Papo and Buldú, 2019). Its application to functional brain networks leaves the interpretation unclear and without theoretical foundation.…”

“…Here we introduce a personalized computer model, based upon individual structural brain networks, to guide epilepsy surgery and improve surgical outcome. The dynamic model is derived from the field of epidemics and has the advantage that it is mathematically well-understood, simple and only has two free parameters (Kermack and McKendrick, 1927;Pastor-Satorras et al, 2015). The model captures one aspect of the epileptic brain that is relevant for surgery, namely the propagation of seizures over the network, rather than the underlying biology.…”

Excess entropies reveal higher organization levels in developing neuron cultures

“…The latter may be the case if specific simplifying features, such as criticality [1], synchronization [2,3], or corresponding architectural constraints [4] are present. It has been argued that biological neural networks host such beneficial properties [5][6][7][8], but whenever this fails to dominate the behavior [9], such an assessment turns into a difficult task. Choosing neural networks as the showcase, we demonstrate how symbolic dynamicsfounded 'excess entropies' 1) measure how much simulations differ from the target processes, 2) uncover model inadequacies and 3) provide guidelines for model set-up and improvement.…”