Biology subjects

Scarpetta, S.

Publications and source records attributed to Scarpetta, S..

8 recordsLinked to original sources

bioRxiv2026

FUSE: FUsing EEG-MEG in a Shared Embedding via self-supervised learning for BCI

Combining complementary neurophysiological modalities offers a promising strategy for improving motor imagery (MI) brain-computer interfaces (BCIs), but learning shared representations across modalities remains largely unexplored. Here, we propose a two-phase deep learning framework for multimodal EEG-MEG decoding that explicitly decouples representation learning from downstream classification. In the first phase, a convolutional encoder-decoder learns a shared latent representation by predicting the power spectral density (PSD) of EEG and MEG signals directly from time-domain activity, rather than using the conventional objective of reconstructing the input signal. In the second phase, the encoder is frozen and its learned representations are reused, without further adaptation, to perform the classification of the downstream MI-BCI task. The framework was evaluated on simultaneous EEG and MEG recordings from 20 participants. The learned representations consistently outperformed conventional handcrafted spectral features, increasing median classification accuracy from 0.734 to 0.794. The multimodal framework also improved performance over MEG alone (median accuracy from 0.680 to 0.794) and yielded a modest increase over EEG alone (median accuracy from 0.765 to 0.794), providing a more robust decoding strategy than single-modality approaches. Furthermore, the learned latent representations were transferable across participants, with more than half of cross-subject models performing within 0.01 accuracy of their subject-specific counterparts. These findings demonstrate that task-agnostic representation learning can capture physiologically meaningful multimodal neural representations that remain transferable across individuals, offering a promising foundation for more robust and reusable BCI pipelines.

neuroscience↗

bioRxiv2026

Temporal persistence and structural organization of neuronal avalanche dynamics

Brain activity can be understood as a sequence of neuronal avalanches, i.e., transient episodes of coordinated activation that emerge across scales, from individual neurons and local networks to whole-brain dynamics. Avalanches are typically characterized by features such as size, duration, number of active components, and the silent time separating consecutive events. Although these features have been extensively characterized through their marginal distributions, their temporal organization and dependence on the underlying brain architecture remain poorly understood, leaving us without a framework for embedding neuronal avalanches within slower brain dynamics. Here, we analyzed eyes-closed resting-state magnetoencephalography recordings and the corresponding structural connectomes from 30 healthy participants to investigate the dynamics of avalanche sizes and silent times. We found that large avalanches preferentially followed short silent times, whereas small avalanches were more likely to occur after long silent periods. Based on the empirical joint distributions of avalanche size and silent time, we could define four types of events occurring above chance levels (avalanche large or small, preceding pause long or short). Mixed categories, combining a small value of one feature with a large value of the other, occurred more frequently than expected, while same-category events happened less often than chance. Furthermore, consecutive events tended to remain in the same category, a phenomenon referred to as persistence. We next investigated whether a brain region's connectivity profile shapes its propensity to participate in avalanches of different sizes. More strongly connected regions participated most often in small avalanches, whereas weakly connected regions were preferentially recruited during large avalanches. This pattern may reflect the greater sensitivity of highly connected hubs to fluctuations propagating through the network, resulting in frequent but spatially contained events. By contrast, the recruitment of more peripheral regions may require broader and stronger collective activity, occurring only during rarer, large-scale avalanches. In contrast, regional participation showed no clear association with the silent time preceding an avalanche. Together, these findings show that neuronal avalanches are neither temporally independent nor anatomically unconstrained: their sequence retains a memory of preceding events, while structural topology shapes which regions are recruited as avalanches grow. By connecting avalanche dynamics with slower temporal organization and the structural connectome, our results provide a multiscale framework for understanding how transient events are embedded within ongoing brain activity.

neuroscience↗

bioRxiv2026

Modularity-dependent storage of dynamic spiking patterns: bridging micro- and mesoscopic representations

Biological systems rely on asynchronous and temporally overlapping dynamics, allowing for the concurrent activation of multiple processes. This principle is particularly evident in brain function, where cognitive tasks engage distributed, interacting regions rather than sequentially isolated ones. To investigate the mechanisms enabling such coordination, we study a modular spiking neural network composed of leaky integrate-and-fire neurons and governed by spike-timing-dependent plasticity (STDP). Our model stores modular spatiotemporal patterns both at the mesoscopic level (sequences of modules) and at the microscopic level (precise spike timings) and includes a parameter, , which regulates the degree of temporal overlap between modules activations. By tuning , the network transitions from sequential to overlapping regimes, ranging from synfire chain-like dynamics to fully co-activated modules. We investigate how the temporal structure influences the networks capacity to encode and selectively retrieve multiple dynamical patterns, while considering biological constraints such as the cost of long-range connectivity. Our results offer insight into how spatiotemporal coding and network organisation support robust, large-scale memory storage and replay.

neuroscience↗

bioRxiv2025

Symmetry breaking and avalanche shapes in modular neural networks

Modularity is as a key characteristic of structural and functional brain networks across species and spatial scales. We investigate the stochastic Wilson-Cowan model on a modular network in which synaptic strengths differ between intra-module and inter-module connections. The system exhibits a rich phase diagram comprising symmetric (SL and SH) and "broken symmetry" (B1, B2, ...) phases. The phase SL (SH) is characterized by the same low (high) activity in all the modules, while the Bm phases are characterized by a high activity in m modules and low activity in the remaining modules. Between SL and SH, and between SL and B1, there are two lines of critical points, where the system shows a critical behaviour, with power law distributions in the avalanches. Along these lines, avalanche shapes differ systematically: they are symmetric or right-skewed at the SL-SH transition, but can become left-skewed over intermediate durations along the SL-B1 critical line. These results provide a theoretical framework that accounts for both symmetric and left-skewed neural avalanche shapes observed experimentally, linking modular organization to critical brain dynamics.

neuroscience↗

bioRxiv2024

Criticality explains structure-function relationships in the human brain

Healthy brain exhibits a rich dynamical repertoire, with flexible spatiotemporal patterns replays on both microscopic and macroscopic scales. How do fixed structural connections yield a diverse range of dynamic patterns in spontaneous brain activity? We hypothesize that the observed relationship between empirical structure and functional patterns is best explained when the microscopic neuronal dynamics is close to a critical regime. Using a modular Spiking Neuronal Network model based on empirical connectomes, we posit that multiple stored functional patterns can transiently reoccur when the system operates near a critical regime, generating realistic brain dynamics and structural-functional relationships. The connections in the model are chosen as to force the network to learn and propagate suited modular spatiotemporal patterns. To test our hypothesis, we employ magnetoencephalography and tractography data from five healthy individuals. We show that the critical regime of the model is able to generate realistic features, and demonstrate the relevance of near-critical regimes for physiological brain activity.

neuroscience↗

bioRxiv2022

Scale-free avalanches in humans NREM sleep and their relationships with sleep macro and microstructure

Sleep plays a key role in preserving brain function, keeping the brain network in a state that ensures optimal computational capabilities. Empirical evidence indicates that such a state is consistent with criticality, where scale-free neuronal avalanches emerge. However, the relationship between sleep, emergent avalanches, and criticality remains poorly understood. Here we fully characterize the critical behavior of avalanches during sleep, and study their relationship with the sleep macro- and micro-architecture, in particular the cyclic alternating pattern (CAP). We show that avalanche size and duration distributions exhibit robust power laws with exponents approximately equal to -3/2 e -2, respectively. Importantly, we find that sizes scale as a power law of the durations, and that all critical exponents for neuronal avalanches obey robust scaling relations, which are consistent with the mean-field directed percolation universality class. Our analysis demonstrates that avalanche dynamics depends on the position within the NREM-REM cycles, with the avalanche density increasing in the descending phases and decreasing in the ascending phases of sleep cycles. Moreover, we show that, within NREM sleep, avalanche occurrence correlates with CAP activation phases, particularly A1, which are the expression of slow wave sleep propensity and have been proposed to be beneficial for cognitive processes. The results suggest that neuronal avalanches, and thus tuning to criticality, actively contribute to sleep development and play a role in preserving network function. Such findings, alongside characterization of the universality class for avalanches, open new avenues to the investigation of functional role of criticality during sleep with potential clinical application. Significance statementWe fully characterize the critical behavior of neuronal avalanches during sleep, and show that avalanches follow precise scaling laws that are consistent with the mean-field directed percolation universality class. The analysis provides first evidence of a functional relationship between avalanche occurrence, slow-wave sleep dynamics, sleep stage transitions and occurrence of CAP phase A during NREM sleep. Because CAP is considered one of the major guardians of NREM sleep that allows the brain to dynamically react to external perturbation and contributes to the cognitive consolidation processes occurring in sleep, our observations suggest that neuronal avalanches at criticality are associated with flexible response to external inputs and to cognitive processes, a key assumption of the critical brain hypothesis.

neuroscience↗

bioRxiv2022

Power spectrum and critical exponents in the 2D stochastic Wilson Cowan model

The power spectrum of brain activity is composed by peaks at characteristic frequencies superimposed to a background that decays as a power law of the frequency, f-{beta}, with an exponent {beta} close to 1 (pink noise). This exponent is predicted to be connected with the exponent{gamma} related to the scaling of the average size with the duration of avalanches of activity. "Mean field" models of neural dynamics predict exponents {beta} and{gamma} equal or near 2 at criticality (brown noise), including the simple branching model and the fully connected stochastic Wilson Cowan model. We here show that a 2D version of the stochastic Wilson Cowan model, where neuron connections decay exponentially with the distance, is characterized by exponents {beta} and{gamma} markedly different from those of mean field, respectively around 1 and 1.3. The exponents and{tau} of avalanche size and duration distributions, equal to 1.5 and 2 in mean field, decrease respectively to 1.29 {+/-} 0.01 and 1.37 {+/-} 0.01. This seems to suggest the possibility of a different universality class for the model in finite dimension.

neuroscience↗

bioRxiv2022

Alpha rhythm induces attenuation-amplification dynamics in neural activity cascades

The alpha rhythm is a distinctive feature of the awake resting-state of the human brain. Recent evidence suggests that alpha plays an active role in information processing, modulating behavioral and cognitive performance. However, the functional role of alpha oscillations in the resting-state neuronal dynamics remains poorly understood. To address this question, we investigate collective neural activity during resting wake and NREM sleep, a physiologic state with marginal presence of alpha rhythm. We show that, during resting wake, alpha oscillations drive an alternation of attenuation and amplification bouts in neural activity. Our analysis indicates that inhibition is activated in pulses that last a single alpha cycle and gradually suppress neural activity, while excitation is successively enhanced over timescales of a few alpha cycles to amplify neural activity. Furthermore, we show that long-term, intermittent fluctuations in alpha amplitude--known as the "waxing and waning" phenomenon--are associated with an attenuation-amplification mechanism acting over the timescales of several seconds and described by a power law decay of the activity rate in the "waning" phase. Importantly, we do not observe such dynamics during NREM sleep. The results suggest that the alpha rhythm acts as a "pacemaker" for the alternation of inhibition and excitation bouts across multiple timescales, the "waxing and waning" being a long-term control mechanism of cortical excitability. The amplification regime observed beyond the timescales of the individual alpha cycle suggests in turn that alpha oscillations might modulate the intensity of neural activity not only through pulses of inhibition, as proposed in the pulsed inhibition hypothesis, but also by timely enhancing excitation (or dis-inhibition).

neuroscience↗