Biology subjects

Chavarriaga, R.

Publications and source records attributed to Chavarriaga, R..

8 recordsLinked to original sources

bioRxiv2025

Prediction, Syntax and Semantic Grounding in the Brain and Large Language Models

Language comprehension involves continuous prediction of upcoming words, with syntactic structure and semantic meaning intertwined in the human brain. To date, few studies have used combined magnetoencephalography (MEG) and electroen-cephalography (EEG) measurements to investigate how syntactic processing, predictive coding, and semantic grounding interact in real time. Here we present the first combined MEG-EEG investigation of syntactic processing and semantic grounding under naturalistic conditions. Twenty-nine healthy participants listened to a German audio book while their neural responses were recorded. Event-related fields and event-related potentials for four word classes - nouns, verbs, adjectives, and proper nouns - showed highly reproducible, characteristic spatio-temporal signatures, including significant pre-onset activity for nouns, suggesting enhanced predictability of this word class. Source-space analyses revealed pronounced activation in the pre- and post-central gyri for nouns, suggesting a deeper semantic grounding of nouns in e.g. sensory experiences than verbs. To further investigate predictive mechanisms, we analyzed the hidden representations of the large language model Llama. By comparing the transformer-based representations to neural responses, we explored the relationship between computational language models and human brain activity, offering new insights into syntactic and semantic prediction. These findings highlight the power of simultaneous MEG-EEG recordings in unraveling the predictive, syntactic, and semantic mechanisms that underlie the comprehension of natural language.

neuroscience↗

bioRxiv2022

EEG error-related potentials encode magnitude of errors and individual perceptual thresholds

Error-related potentials (ErrP) are a prominent electroencephalogram (EEG) correlate of performance monitoring, and so crucial for learning and adapting our behavior. Although there exists an agreement that ErrP signal awareness to errors, it remains poorly understood whether they encode further information. Here we report an experiment with sixteen participants during three recording sessions in which occasional visuomotor rotations of varying magnitude occurred during a cursor reaching task. We designed a brain-computer interface (BCI) to detect ErrP in single trials that provided real-time feedback to participants by changing the color of the cursor upon ErrP detection. The individual ErrP-BCI decoders exhibited good transfer across recording sessions and scalability over the varying magnitude of errors. Our results indicate that ErrPs encode not only the conscious perception of errors, but also their magnitude, in their amplitude and latency. Furthermore, a non-linear relationship between the ErrP-BCI output and the magnitude of errors predicts individual perceptual thresholds to detect rotations. The uncovered relationship is consistent with non-human primate studies, which found a similar relationship between the size of errors and simple spike activity of Purkinje cells, and we conjecture a cerebellar contribution to ErrP. Our experimental setup and findings open new avenues to probe and extend current theories of performance monitoring, which are based on response conflict tasks, by incorporating continuous human-interaction tasks as well as analysis of the ErrP complex as a whole rather than individual peaks.

neuroscience↗

bioRxiv2022

EEG signature of breaks in embodiment in VR

The brain mechanism of embodiment in a virtual body has grown a scientific interest recently, with a particular focus on providing optimal virtual reality (VR) experiences. Disruptions from an embodied state to a less- or non-embodied state, denominated Breaks in Embodiment (BiE), are however rarely studied despite their importance for designing interactions in VR. Here we use electroencephalography (EEG) to monitor the brains reaction to a BiE, and investigate how this reaction depends on previous embodiment conditions. The experimental protocol consisted of two sequential steps; an induction step where participants were either embodied or non-embodied in an avatar, and a monitoring step where, in some cases, participants saw the avatars hand move while their hand remained still. Our results show the occurrence of error-related potentials linked to observation of the BiE event in the monitoring step. Importantly, this EEG signature shows amplified potentials following the non-embodied condition, which is indicative of an accumulation of errors across steps. These results provide neurophysiological indications on how progressive disruptions impact the expectation of embodiment for a virtual body.

neuroscience↗

bioRxiv2021

Stopping vs Resting state during motor imagery paradigm

Current non-invasive Brain Machine interfaces commonly rely on the decoding of sustained motor imagery activity (MI). This approach enables a user to control brain-actuated devices by triggering predetermined motor actions. One major drawback of such strategy is that users are not trained to stop their actions. Indeed, the termination process involved in BMI is poorly understood with most of the studies assuming that the end of an MI action is similar to the resting state. Here we hypothesize that the process of stopping MI (MI termination) and resting state are two different processes that should be decoded independently due to the exhibition of different neural pattens. We compared the detection of both states transitions of an imagined movement, i.e. rest-to-movement (onset) and movement-to-rest (offset). Our results shows that both decoders show significant differences in term of performances and latency (N=17 Subjects) with the offset decoder able to detect faster and better MI termination. While studying this difference, we found that the offset decoder is primarily based on the use of features in Beta band which appears earlier. Based on this finding, we also proposed a Random Forrest based decoder which enable to distinguish three classes (MI, MI termination and REST).

neuroscience↗

bioRxiv2021

Mobile brain/body imaging of landmark-based navigation with high-density EEG

Coupling behavioral measures and brain imaging in naturalistic, ecological conditions is key to comprehend the neural bases of spatial navigation. This highly-integrative function encompasses sensorimotor, cognitive, and executive processes that jointly mediate active exploration and spatial learning. However, most neuroimaging approaches in humans are based on static, motion constrained paradigms and they do not account for all these processes, in particular multisensory integration. Following the Mobile Brain/Body Imaging approach, we aimed to explore the cortical correlates of landmark-based navigation in actively behaving young adults, solving a Y-maze task in immersive virtual reality. EEG analysis identified a set of brain areas matching state-of-the-art brain imaging literature of landmark-based navigation. Spatial behavior in mobile conditions additionally involved sensorimotor areas related to motor execution and proprioception usually overlooked in static fMRI paradigms. Expectedly, we located a cortical source in or near the posterior cingulate, in line with the engagement of the retrosplenial complex in spatial reorientation. Consistent with its role in visuo-spatial processing and coding, we observed an alpha power desynchronization while participants gathered visual information. We also hypothesized behavior-dependent modulations of the cortical signal during navigation. Despite finding few differences between the encoding and retrieval phases of the task, we identified transient time-frequency patterns attributed, for instance, to attentional demand, as reflected in the alpha/gamma range, or memory workload in the delta/theta range. We confirmed that combining mobile high-density EEG and biometric measures can help unravel the brain structures and the neural modulations subtending ecological landmark-based navigation.

neuroscience↗

bioRxiv2019

Beta-driven closed-loop deep brain stimulation can compromise human motor behavior in Parkinson’s Disease

Closed-loop or adaptive deep brain stimulation (DBS) for Parkinsons Disease (PD) has shown comparable clinical improvements to continuous stimulation, yet with less stimulation times and side effects. In this form of control, stimulation is driven by pathological beta oscillations recorded from the subthalamic nucleus, which have been shown to correlate with PD motor symptoms. An important consideration is that beta activity is itself modulated during volitional movements, yet it is unknown the impact that these volitional modulations may have on the efficacy of closed-loop systems. Here, three PD patients performed a functional reaching task during closed-loop stimulation while we measured their motor behavior. Our results show that closed-loop stimulation can alter motor performance at distinct movement intervals. Of particular relevance, closed-loop DBS compromised behavior during the returning period by increasing the amount of submovements executed, and in turn delayed movement termination. Following these findings, we hypothesize that the use of machine learning decoding different movement intervals to fully switch off the stimulator may be beneficial, and present here an exemplary approach decoding the initiation of the movement returning interval above chance level. These findings highlight the importance of evaluating these systems during functional tasks, and the need of extracting more robust biomarkers encoding ongoing symptoms or tasks execution intervals.

neuroscience↗

bioRxiv2019

Real-time Detection of Driver’s Movement Intention in Response to Traffic Lights

Movements are preceded by certain brain states that can be captured through various neuroimaging techniques. Brain-Computer Interfaces can be designed to detect the movement intention brain state during driving, which could be beneficial in improving the interaction between a smart car and its driver, by providing assistance in-line with the drivers intention. In this paper, we present an Electroencephalogram based decoder of such brain states preceding movements performed in response to traffic lights in two experiments: in a car simulator and a real car. The results of both experiments (N=10: car simulator, N=8: real car) confirm the presence of anticipatory Slow Cortical Potentials in response to traffic lights for accelerating and braking actions. Single-trial classification performance exhibits an Area Under the Curve (AUC) of 0.71{+/-}0.14 for accelerating and 0.75{+/-}0.13 for braking. The AUC for the real car experiment are 0.63{+/-}0.07 and 0.64{+/-}0.13 for accelerating and braking respectively. Moreover, we evaluated the performance of real-time decoding of the intention to brake during online experiments only in the car simulator, yielding an average accuracy of 0.64{+/-}0.1. This paper confirm the existence of the anticipatory slow cortical potentials and the feasibility of single-trial detection these potentials in driving scenarios.

neuroscience↗

bioRxiv2018

Differential contributions of subthalamic beta rhythms and neural noise to Parkinson motor symptoms

BackgroundExcessive beta oscillatory activity in the subthalamic nucleus (STN) is linked to Parkinsons disease and associated motor symptoms. However, the relationship between beta activity and motor symptoms has been inconsistent, which may influence the efficacy of closed-loop deep brain stimulation.\n\nHypothesisWe hypothesized that this variability is due to the degree of neural noise in STN recordings. Recent evidence has shown that neural noise is influenced by multiple factors, such as development, aging and disease, and could confound measures of beta activity. In this work, we propose a model that disentangles beta oscillatory activity and neural noise in the STN power spectrum.\n\nMethodsWe investigated the impact of neural noise on estimations of beta activity and motor symptoms from data recorded bilaterally from the subthalamic nuclei of thirteen Parkinsonian patients.\n\nResultsResults showed that the relationship between beta oscillatory amplitude and motor symptoms (bradykinesia and rigidity) significantly improved when neural noise was removed from the estimation of beta activity.\n\nConclusionThese findings emphasize the importance of modeling neural components independently for understanding physiological processes associated with Parkinsons disease, and identifying better biomarkers for characterizing symptom severity. Subsequently, we predict that our findings can have a direct application for closed-loop deep brain stimulation on Parkinsons Disease.

neuroscience↗