arXiv subjects

Marius Klug

Publications and source records attributed to Marius Klug.

2 recordsLinked to original sources

arXiv2026

Decoding Workload and Agreement From EEG During Spoken Dialogue With Conversational AI

Passive brain-computer interfaces offer a potential source of implicit feedback for alignment of large language models, but most mental state decoding has been done in controlled tasks. This paper investigates whether established EEG classifiers for mental workload and implicit agreement can be transferred to spoken human-AI dialogue. We introduce two conversational paradigms - a Spelling Bee task and a sentence completion task- and an end-to-end pipeline for transcribing, annotating, and aligning word-level conversational events with continuous EEG classifier output. In a pilot study, workload decoding showed interpretable trends during spoken interaction, supporting cross-paradigm transfer. For implicit agreement, we demonstrate continuous application and precise temporal alignment to conversational events, while identifying limitations related to construct transfer and asynchronous application of event-based classifiers. Overall, the results establish feasibility and constraints for integrating passive BCI signals into conversational AI systems.

cs.HC↗

arXiv2025

Neuroadaptive Haptics: Comparing Reinforcement Learning from Explicit Ratings and Neural Signals for Adaptive XR Systems

Neuroadaptive haptics offers a path to more immersive extended reality (XR) experiences by dynamically tuning multisensory feedback to user preferences. We present a neuroadaptive haptics system that adapts XR feedback through reinforcement learning (RL) from explicit user ratings and brain-decoded neural signals. In a user study, participants interacted with virtual objects in VR while Electroencephalography (EEG) data were recorded. An RL agent adjusted haptic feedback based either on explicit ratings or on outputs from a neural decoder. Results show that the RL agent's performance was comparable across feedback sources, suggesting that implicit neural feedback can effectively guide personalization without requiring active user input. The EEG-based neural decoder achieved a mean F1 score of 0.8, supporting reliable classification of user experience. These findings demonstrate the feasibility of combining brain-computer interfaces (BCI) and RL to autonomously adapt XR interactions, reducing cognitive load and enhancing immersion.

cs.HC↗