Guillaume A. Rousselet

The paper evaluates cluster-based permutation tests for estimating EEG/MEG onsets, showing they have positive bias but perform better than other methods in terms of bias and variability.

This study investigates whether prediction error explains predictability effects on the N1 component during picture-word verification, finding results opposite to a simple predictive coding account.

The paper shows that resampling large neuroimaging datasets with replacement introduces bias in error estimates for brain-behavior associations, suggesting that meta-analyses should resample only up to 10% of the full sample size to avoid these biases.

The paper introduces a method to reduce outlier influence in EEG data using principal component projection weights in linear models, improving statistical power and controlling error rates.

The paper challenges the use of mean and median for skewed distributions like reaction times, advocating for quantile estimation instead.

The paper investigates how word predictability and sentence plausibility affect the N400 brain response during language comprehension using a large-scale study.

The paper investigates how ageing delays neural processing of face features, showing a 40 ms delay in occipital-temporal pathway activity that correlates with slower behavioural responses.

This paper reports a multi-lab failure to replicate the original finding of probabilistic phonological pre-activation in language comprehension, showing no significant article-N400 cloze effect but a significant noun-N400 effect.

The paper provides guidelines for using modern statistical methods in neuroscience to improve power and control Type I error rates.

The paper introduces graphical methods and tools to better compare two groups in neuroscience, offering more detailed insights than traditional bar graphs and t-tests.

The paper introduces a statistical framework for neuroimaging data analysis using mutual information estimated via a Gaussian copula, validated on EEG and MEG data with applications to various signal features.

The paper shows the N170 ERP reflects cross-hemispheric transfer of visual features, not just local processing.