Jason da Silva Castanheira

This paper introduces an autoencoder framework to derive individualized neurophysiological profiles from resting-state MEG data without participant labels, achieving high accuracy in differentiation and prediction.

The study examines how neurophysiological traits evolve across the lifespan using MEG data from 1,000 individuals, finding age-related shifts in brain regions associated with individuality and genetic influences.

This study shows that detrending neural spectra conflates rhythmic and arrhythmic brain activity, while specparam provides accurate separation of these components.

This study investigates the genetic basis of neurophysiological variability using twin and unrelated individuals, finding significant genetic influence on brain activity traits linked to neurotransmission and cortical gene expression patterns.

This paper introduces a Bayesian model-selection method to automatically determine the number of spectral peaks in neural power spectra, improving reproducibility and robustness.

The paper uses EEG, machine learning, and multivariate statistics to identify neural features predicting hypnotic susceptibility, finding that aperiodic EEG activity at rest is the strongest predictor.

This study investigates the variability of fMRI BOLD signals in the human brain, showing it reflects multi-scale brain organization rather than noise.

This study investigates how exogenous and endogenous attention differentially and overlap in shaping perceptual facilitation during visual processing using EEG and machine learning.

This study investigates how exogenous attention interferes with endogenous attention through alpha-band neural activity, using a moderated mediation model to show the mediating role of alpha power in attentional interference.

This paper introduces SPRiNT, a method to time-resolve aperiodic and periodic neural activity, showing its effectiveness on synthetic and empirical data.

The paper investigates how endogenous attention affects exogenous attention processing and shows that alpha-band brain activity patterns change when both types of attention are engaged.

This study determines the minimal data duration needed for robust spectral estimates in resting-state MEG, showing that 30-120 seconds of data are sufficient across different recording conditions and populations.

This study identifies individuals using resting-state MEG data, showing high identifiability rates and robustness over time.