Accurate decoding of lower-limb motion from EEG signals is essential for advancing brain-computer interface (BCI) applications in movement intent recognition and control. This study presents NeuroDyGait, a two-stage, phase-aware EEG-to-gait decoding framework that explicitly models temporal continuity and domain relationships. To address challenges of causal, phase-consistent prediction and cross-subject variability, Stage I learns semantically aligned EEG-motion embeddings via relative contrastive learning with a cross-attention-based metric, while Stage II performs domain relation-aware decoding through dynamic fusion of session-specific heads. Comprehensive experiments on two benchmark datasets (GED and FMD) show substantial gains over baselines, including a recent 2025 model EEG2GAIT. The framework generalizes to unseen subjects and maintains inference latency below 5 ms per window, satisfying real-time BCI requirements. Visualization of learned attention and phase-specific cortical saliency maps further reveals interpretable neural correlates of gait phases. Future extensions will target rehabilitation populations and multimodal integration.

Brain-computer interfaces (BCIs) depend on the reliable decoding of brain activity, yet key rhythms like alpha and mu are not spectrally static and can shift with cognitive and motor demands. Here, we investigated within-session changes in instantaneous alpha/mu frequency and magnitude during motor-related BCI calibration using an oscillator-tracking framework based on an extended Kalman filter (EKF). We applied this method to four public EEG datasets spanning motor execution and imagery tasks. Across all datasets, we observed consistent increases in mu instantaneous frequency and magnitude over central sensorimotor regions, indicative of motor engagement and possible training-related neuroplasticity. In contrast, posterior and surrounding cortical areas often showed alpha slowing, suggestive of declining vigilance or cognitive fatigue, or alternatively, resource reallocation via inhibition of task-irrelevant regions. These opposing spatial trends underscore the functional heterogeneity of alpha-band activity across the cortex. Our results highlight the potential of real-time frequency tracking not only to improve decoding accuracy but also to monitor neurophysiological state changes and guide adaptive adjustments in BCI calibration paradigms.

Human cognition involves intricate neural processes for error perception and correction. These processes are crucial in error-monitoring processes such as feedback, learning, control, and decision-making. We present a complete workflow using explainable artificial intelligence (XAI) to guide the feature extraction of electroencephalographic (EEG) signals in a classification task with error-related brain responses. The identification of relevant channels for classification problems has practical relevance, as a dense electrode setup reduces the usability of brain-computer interfaces (BCIs). Specialists can inspect and select seemingly important sensors based on knowledge of brain regions and neural patterns. However, reduced configurations often harm accurate model performance. The contribution of this work lies in demonstrating that XAI can be used to inform the extraction of relevant temporal and spatial information with a strong connection to machine learning model sensitivity. We employed a local and a global XAI method to i) evaluate consistency with expert knowledge, ii) identify relevant time points for asynchronous error decoding, and iii) systematically reduce the error decoding EEG setup. This advances the integration of XAI in neuroscience, thus contributing to the design of practical BCIs.

Event-related potential-based Brain-Computer Interfaces are becoming widely popular due to their ability to send commands quickly with high accuracy. However, the stationary characteristics of electroencephalographic signals, coupled with their low signal-to-noise ratio, lead to variations in amplitude, time period, and latency in the patterns of event-related potentials across different trials, sessions, days and subjects. Conventional feature extraction and machine learning algorithms are not designed to handle these differences, requiring the development of methods that can address these variations. Here, we propose a novel lightweight deep neural network for event-related potential classification, consisting of three modules. In this model, we have a spatio-temporal module that learns local features simultaneously across channels and time points. Following this, there's a component extractor module comprising depthwise convolutions, inspired by mixed depthwise convolutions, to capture the event-related potential characteristics with different temporal durations. Lastly, an advanced temporal layer addresses event-related potential shape and scale variations using deformable convolutions. We conducted experiments on event-related potential detection in a subject-independent scenario using one error-related negativity potential dataset and three perturbation-evoked potential datasets. Comparisons were made with established methods including two conventional machine learning algorithms and three well-known deep learning architectures, demonstrating that our model outperformed them in terms of classification accuracy and parameter efficiency. In our analysis, we aimed to understand the model's performance using gradient-weighted class activation mapping and t-distributed stochastic neighbor embedding. These methods facilitated the visualization and interpretation of our model's effectiveness, providing insights into its relationship with the neuroscientific characteristics of event-related potentials.

Error perception is known to elicit distinct brain patterns, which can be used to improve the usability of systems facilitating human-computer interactions, such as brain-computer interfaces. This requires a high-accuracy detection of erroneous events, e.g., misinterpretations of the user’s intention by the interface, to allow for suitable reactions of the system. In this work, we concentrate on steering-based navigation tasks. We present a combined electroencephalography-virtual reality (VR) study investigating different approaches for error detection and simultaneously exploring the corrective human behavior to erroneous events in a VR flight simulation. We could classify different errors allowing us to analyze neural signatures of unexpected changes in the VR. Moreover, the presented models could detect errors faster than participants naturally responded to them. This work could contribute to developing adaptive VR applications that exclusively rely on the user’s physiological information.

When persons interact with the environment and experience or witness an error (e.g. an unexpected event), a specific brain pattern, known as error-related potential (ErrP) can be observed in the electroencephalographic signals (EEG). Virtual Reality (VR) technology enables users to interact with computer-generated simulated environments and to provide multi-modal sensory feedback. Using VR systems can, however, be error-prone. In this paper, we investigate the presence of ErrPs when Virtual Reality users face 3 types of visualization errors: (Te) tracking errors when manipulating virtual objects, (Fe) feedback errors, and (Be) background anomalies. We conducted an experiment in which 15 participants were exposed to the 3 types of errors while performing a center-out pick and place task in virtual reality. The results showed that tracking errors generate error-related potentials, the other types of errors did not generate such discernible patterns. In addition, we show that it is possible to detect the ErrPs generated by tracking losses in single trial, with an accuracy of 85%. This constitutes a first step towards the automatic detection of error-related potentials in VR applications, paving the way to the design of adaptive and self-corrective VR/AR applications by exploiting information directly from the user’s brain.

Affective brain-computer interface (aBCI) introduces personal affective factors into human-computer interactions, which could potentially enrich the user's experience during the interaction with a computer. However, affective neural patterns are volatile even within the same subject. To maintain satisfactory emotion recognition accuracy, the state-of-the-art aBCIs mainly tailor the classifier to the subject-of-interest and require frequent re-calibrations for the classifier. In this paper, we demonstrate that the recognition accuracy of aBCIs deteriorates when re-calibration is ruled out during the long-term usage for the same subject. Then, we propose a stable feature selection method to choose the most stable affective features, for mitigating the accuracy deterioration to a lesser extent and maximizing the aBCI performance in the long run. We validate our method on a dataset comprising six subjects' EEG data collected during two sessions per day for each subject for eight consecutive days.

Spectral band power features are one of the most widely used features in the studies of electroencephalogram (EEG)-based emotion recognition. The power spectral density of EEG signals is partitioned into different bands such as delta, theta, alpha and beta band etc. Though based on neuroscientific findings, the partition of frequency bands is somewhat on an ad-hoc basis, and the definition of frequency ranges of the bands of interest can vary between studies. On the other hand, it is also arguable that one definition of power bands could perform equally well on all subjects. In this paper, we propose to use autoencoder to automatically learn from each subject the salient frequency components from power spectral density estimated as periodogram by Fast Fourier Transform (FFT). We propose a network architecture especially for EEG feature extraction, one that adopts hidden unit clustering with added pooling neuron per cluster. The classification accuracy with features extracted by our proposed method is benchmarked against that with standard power features. Experimental results show that our proposed feature extraction method achieves accuracy ranging from 44% to 59% for three-emotion classification. We also see a 4-20% accuracy improvement over standard band power features.

Emotion recognition is an integral part of affective computing. An affective brain-computer-interface (BCI) can benefit the user in a number of applications. In most existing studies, EEG (electroencephalograph)-based emotion recognition is explored in a classificatory manner. In this manner, human emotions are discretized by a set of emotion labels. However, human emotions are more of a continuous phenomenon than discrete. A regressive approach is more suited for continuous emotion recognition. Few studies have looked into a regressive approach. In this study, we investigate a portfolio of EEG features including fractal dimension, statistics and band power. Support vector regression (SVR) is employed in this study to estimate subject's valence level by means of different features under two evaluation schemes. In the first scheme, a SVR is constructed with full training resources, whereas in the second scheme, a SVR only receives minimal training resources. MAE (mean absolute error) averages of 0.74 and 1.45 can be achieved under the first and the second scheme, respectively, by fractal feature. The advantages of a regressive approach over classificatory approach lie in continuous emotion recognition and the possibility to reduce training resources to minimal level.

Operating Brain-Computer Interfaces (BCIs) that are based on the detection of changes in oscillatory non-invasive electroencephalogram (EEG) typically involves learning. Commonly the learning process is distributed between the user (reliable EEG pattern generation) and the machine (robust EEG pattern detection). Standard training approaches, however, typically do not allow users to gain meaningful levels of control. A better understanding of brain functioning or the use of sophisticated machine learning are ways to enhance control. Rethinking training paradigms is another option. In this paper, we enhance our game-based training approach by adding competitive elements. Winning is a powerful motivator that increases user engagement of the typically boring BCI training experience. We report on a user with cerebral palsy who successfully gained BCI control and played the classical Tic-Tac-Toe game against his caregiver.

Non-invasive brain-computer interfaces (BCI) aim to assist severely motor impaired persons in their daily life routine, however only a few BCIs have made it out of the laboratory. To foster further development, the Cybathlon, an international multi-discipline tournament, has been founded. One of the disciplines is the BCI-Race, where end users control avatars in a virtual race game by their thoughts. The game supports 4 different commands which accelerate the avatar and increase the chance to win. So far, no gold standard procedure has been established on how to enable, train and individualize multi-class BCI control for users. In this work, we present a 4-stage procedure to closely tailor a multi-class BCI to an end user who will participate in the Cybathlon. In stage I, we test for basic BCI-capability, in stage II we evaluate the most suitable mental tasks for the user and in stage III, we test user compliance while perceiving feedback. Finally in stage IV, the user is playing the competition game. Our procedure provides a promising way to guide users from first contact with BCI technology to actually play a videogame by thoughts. We demonstrate the feasibility of our procedure at the pilot of the GRAZ-BCI racing team MIRAGE91. We believe that an evidence based procedure, maybe similar to the one presented in this work, is a necessity to introduce BCI technology in the daily life of potential end users.

Mutual brain-machine co-adaptation is the most common approach used to gain control over spontaneous electroencephalogram (EEG) based brain-computer interfaces (BCIs). Co-adaptation means the concurrent or alternating use of machine learning and the brain's reinforcement learning mechanisms. Results from the literature, however, suggest that current implementations of this approach does not lead to desired results ("BCI inefficiency"). In this paper, we propose an alternative strategy that implements some recommendations from educational psychology and instructional design. We present a jigsaw puzzle game for Android devices developed to train the BCI skill in individuals with cerebral palsy (CP). Preliminary results of a supporting study in four CP users suggest high user acceptance. Three out of the four users achieved better than chance accuracy in arranging pieces to form the puzzle.

In their early days, brain-computer interfaces (BCIs) were only considered as control channel for end users with severe motor impairments such as people in the locked-in state. But, thanks to the multidisciplinary progress achieved over the last decade, the range of BCI applications has been substantially enlarged. Indeed, today BCI technology cannot only translate brain signals directly into control signals, but also can combine such kind of artificial output with a natural muscle-based output. Thus, the integration of multiple biological signals for real-time interaction holds the promise to enhance a much larger population than originally thought end users with preserved residual functions who could benefit from new generations of assistive technologies. A BCI system that combines a BCI with other physiological or technical signals is known as hybrid BCI (hBCI). In this work, we review the work of a large scale integrated project funded by the European commission which was dedicated to develop practical hybrid BCIs and introduce them in various fields of applications. This article presents an hBCI framework, which was used in studies with nonimpaired as well as end users with motor impairments.

The articles in this special issue focus on brain-computer interfacing. The papers are dedicated to this growing and diversifying research enterprise, and features important review articles as well as some important current examples of research in this area. The field of brain-computer interface (BCI) research began to develop about 25 years ago and transformed from initially isolated demonstrations by a few groups into a large scientific enterprise that is currently producing hundreds of peer-reviewed articles and several dedicated conferences and workshops each year. This level of productivity is reflective of the large and continually growing enthusiasm by the scientific community, funding agencies, and the public.

For individuals with high spinal cord injury (SCI), restoring missing grasping function is a high priority. Neuroprostheses based on functional electrical stimulation (FES) can partly compensate the loss of upper extremity function in people suffering from tetraplegia. With noninvasive, multichannel neuroprostheses a pinch and power grasp can be accomplished for everyday use. Hybrid systems combining FES with active orthoses hold promise for restoring a completely lost arm function. Novel control interfaces are needed to make full use of the many degrees of freedom of complex hybrid neuroprostheses. Motor imagery (MI)-based brain-computer interfaces (BCIs) are an emerging technology that may serve as a valuable adjunct to traditional control interfaces for neuroprosthetic control. Shared control and context-specific autonomy are most effective for reducing the users' workload. The modularity of upper extremity neuroprostheses as well as their associated control interfaces enable customization of the systems to adapt to the impairment and needs of each individual end user. This work provides an overview of the application of noninvasive hybrid BCI-controlled upper extremity neuroprostheses in individuals with high SCI with a strong focus on the results from the European Integrated Project Tools for Brain-Computer Interaction and will describe the challenges and promises for the future.

We investigated whether listener-assisted scanning, an alternative communication method for persons with severe motor and visual impairments but preserved cognitive skills, could be used for spelling with EEG. To that end spoken letters were presented sequentially, and the participants made selections by performing motor execution/imagery or a cognitive task. The motor task was a brisk dorsiflexion of both feet, and the cognitive task was related to working memory and perception of human voice. The motor imagery task yielded the most promising results with respect to letter selection accuracy, albeit with a large variation in individual performance. The cognitive task yielded significant ( p = 0.05) albeit moderate results. Closer inspection of grand average ERPs for the cognitive task revealed task-related modulation of a late negative component, which is novel in the auditory BCI literature. Guidelines for further development are presented.

We investigate whether an electroencephalography technique could be used for yes/no communication with auditory scanning. To be usable by the target group, i.e., minimally conscious individuals, such a brain-computer interface (BCI) has to be very simple and robust. This leads to the concept of a single-switch BCI (ssBCI). With an ssBCI it is possible to reliably detect one certain, individually trained, brain pattern of the individual, and use it to control all kinds of applications using yes/no responses. A total of 10 healthy volunteers (20-27 years) participated in an initial cue-based session with a motor imagery (MI) task after brisk passive feet/hand movement. Four of them reached MI classification accuracies above 70% and, thus, fulfilled the inclusion criterion for participation in the 2nd session. In the 2nd session, MI was used to communicate yes/no answers to a series of questions in an auditory scanning mode. Two of the three participants of the 2nd session were able to reliably communicate their intent with 90% or above correct and 0% false responses. This work showed, for the 1st time, the use of a ssBCI based on passive and imagined movements for communication in auditory scanning mode.

The use of statistical models and statistical inference for characterizing the interplay between brain structures and human behavior (functional brain mapping) is common in neuroscience. Statistical methods, however, require the availability of sufficiently large data sets. As a result, experimental paradigms used to collect behavioral trials from individuals are data centered and not user centered. This means that experimental paradigms are tuned to collect as many trials as possible, are generally rather demanding, and are not always motivating or engaging for individuals. Subject cooperation and their compliance with the task may decrease over time. Whenever possible, paradigms are designed to control for factors such as fatigue, attention, and motivation. In this paper, we propose the use of the Kinect motion tracking sensor (Microsoft, Inc., Redmond, WA, USA) in a game-based paradigm for noninvasive electroencephalogram (EEG)-based functional motor mapping. Results from an experimental study with able-bodied subjects playing a virtual ball game suggest that the Kinect sensor is useful for isolating specific movements during the interaction with the game, and that the computed EEG patterns for hand and feet movements are in agreement with results described in the literature.

Brain-computer interface (BCI) systems are not often used as input devices for modern games, due largely to their low bandwidth. However, BCIs can become a useful input modality when adapting the dynamics of the brain-game interaction, as well as combining them with devices based on other physiological signal to make BCIs more powerful and flexible. We introduce the Graz BCI Game Controller (GBGC) and describe how techniques such as context dependence, dwell timers and other intelligent software tools were implemented in a new system to control the Massive Multiplayer Online Role Playing Game World of Warcraft (WoW).