Dr. Mathias Vukelic

Dr. Dr. Mathias Vukelic


Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 July 2023 in Sensors
Verified
Journal Article
Open access
PDF file
SGSabrina Gado
Sabrina Gado
Katharina LingelbachKatharina Lingelbach
Katharina Lingelbach
MWMaria Wirzberger
Maria Wirzberger
DVDr. Mathias Vukelic
Dr. Mathias Vukelic
Humans’ performance varies due to the mental resources that are available to successfully pursue a task. To monitor users’ current cognitive resources in naturalistic scenarios, it is essential to not only measure demands induced by the task itself but also consider situational and environmental influences. We conducted a multimodal study with 18 participants (nine female, M = 25.9 with SD = 3.8 years). In this study, we recorded respiratory, ocular, cardiac, and brain activity using functional near-infrared spectroscopy (fNIRS) while participants performed an adapted version of the warship commander task with concurrent emotional speech distraction. We tested the feasibility of decoding the experienced mental effort with a multimodal machine learning architecture. The architecture comprised feature engineering, model optimisation, and model selection to combine multimodal measurements in a cross-subject classification. Our approach reduces possible overfitting and reliably distinguishes two different levels of mental effort. These findings contribute to the prediction of different states of mental effort and pave the way toward generalised state monitoring across individuals in realistic applications.

23 November 2023 in Frontiers in Neuroergonomics
Verified
Journal Article
Open access
DVDr. Mathias Vukelic
Dr. Mathias Vukelic
MBMichael Bui
Michael Bui
AVAnna Vorreuther
Anna Vorreuther
KLKatharina Lingelbach
Katharina Lingelbach
Deep reinforcement learning (RL) is used as a strategy to teach robot agents how to autonomously learn complex tasks. While sparsity is a natural way to define a reward in realistic robot scenarios, it provides poor learning signals for the agent, thus making the design of good reward functions challenging. To overcome this challenge learning from human feedback through an implicit brain-computer interface (BCI) is used. We combined a BCI with deep RL for robot training in a 3-D physical realistic simulation environment. In a first study, we compared the feasibility of different electroencephalography (EEG) systems (wet- vs. dry-based electrodes) and its application for automatic classification of perceived errors during a robot task with different machine learning models. In a second study, we compared the performance of the BCI-based deep RL training to feedback explicitly given by participants. Our findings from the first study indicate the use of a high-quality dry-based EEG-system can provide a robust and fast method for automatically assessing robot behavior using a sophisticated convolutional neural network machine learning model. The results of our second study prove that the implicit BCI-based deep RL version in combination with the dry EEG-system can significantly accelerate the learning process in a realistic 3-D robot simulation environment. Performance of the BCI-based trained deep RL model was even comparable to that achieved by the approach with explicit human feedback. Our findings emphasize the usage of BCI-based deep RL methods as a valid alternative in those human-robot applications where no access to cognitive demanding explicit human feedback is available.

31 December 2020 in Brain Sciences
Verified
Journal Article
Open access
PDF file
DVDr. Mathias Vukelic
Dr. Mathias Vukelic
Katharina LingelbachKatharina Lingelbach
Katharina Lingelbach
KPKathrin Pollmann
Kathrin Pollmann
MPMatthias Peissner
Matthias Peissner
Affect monitoring is being discussed as a novel strategy to make adaptive systems more user-oriented. Basic knowledge about oscillatory processes and functional connectivity underlying affect during naturalistic human–computer interactions (HCI) is, however, scarce. This study assessed local oscillatory power entrainment and distributed functional connectivity in a close-to-naturalistic HCI-paradigm. Sixteen participants interacted with a simulated assistance system which deliberately evoked positive (supporting goal-achievement) and negative (impeding goal-achievement) affective reactions. Electroencephalography (EEG) was used to examine the reactivity of the cortical system during the interaction by studying both event-related (de-)synchronization (ERD/ERS) and event-related functional coupling of cortical networks towards system-initiated assistance. Significantly higher α-band and β-band ERD in centro-parietal and parieto-occipital regions and β-band ERD in bi-lateral fronto-central regions were observed during impeding system behavior. Supportive system behavior activated significantly higher γ-band ERS in bi-hemispheric parietal-occipital regions. This was accompanied by functional coupling of remote β-band and γ-band activity in the medial frontal, left fronto-central and parietal regions, respectively. Our findings identify oscillatory signatures of positive and negative affective processes as reactions to system-initiated assistance. The findings contribute to the development of EEG-based neuroadaptive assistance loops by suggesting a non-obtrusive method for monitoring affect in HCI.


27 August 2025
Verified
Book Chapter
Closed access
PDF file
Doris AschenbrennerDoris Aschenbrenner
Doris Aschenbrenner
AHAnett Hübner
Anett Hübner
NTNektaria Tagalidou
Nektaria Tagalidou
MBMichael Bui
Michael Bui
DVDr. Mathias Vukelic
Dr. Mathias Vukelic
NBNatalie Basedow
Natalie Basedow

1 January 2018 in Frontiers in Human Neuroscience
Verified
Journal Article
Open access
FMFeroze T. Malik
Feroze T. Malik
KPKathrin Pollmann
Kathrin Pollmann
MPMatthias Peissner
Matthias Peissner
DVDr. Mathias Vukelic
Dr. Mathias Vukelic
Neuro-adaptive interaction paradigms are being discussed as a strategy to make assisitive technologies more user-oriented by recognizing mental states and adapting assistance accordingly (Zander, Krol, Birbaumer, & Gramann, 2016). Still, so far, little empirical evidence exists about the neuroelectrical makers of emotional user appraisal in realistic scenarios that can be used for real-time classification. We recently presented a new experimental paradigm, AFFINDU, for research in the domain of neuro-adaptive technologies (Pollmann, Ziegler, Peissner, & Vukelić, 2017). We simulated an adaptive system that induces positive and negative affective user states by supporting or impeding the completion of a navigation task. AFFINDU roots in a prominent appraisal model of emotions (Sander, Grandjean, & Scherer, 2005) which states that appraisal is a cognitive mechanism that considers a rapid evaluation, personal judgement and implications of events. In the previous study, we included the AFFINDU system with a simple grid-based menu, through which the user had to navigate in a controlled step-by-step sequence. Participants’ affective reactions to adaptive rearrangements of the menu that supported or impeded the navigation task were recorded using electroencephalography (EEG). The event-related potential (ERP) analysis showed that distinct neuroelectrical markers of affective states can be found that are induced by supportive adaptations (SA) and impeding adaptations (IA). The current study investigates whether similar markers of emotional appraisal can be found in a more realistic, ecologically valid scenario. To do so, we included the AFFINDU system in a more complex interaction task: The user could freely navigate through a whole website with different pages instead of navigating step-by-step through the grid-based menu. The system performed SA and IA that changed the graphical UI as well as graphical changes without any consequence for task achievement (neutral adaptations; NA). Similarly to a study by (Peissner & Edlin-White, 2013), for each adaptation the user could provide feedback to the adaptation (keep, revoke, I don’t care). 24 participants (18 to 42 years, 12 females) were recruited for the 75-minute experimental session (Fig.1). A ten-minutes testing period, during which participants could explore the website and navigation, was followed by six measurement blocks (Fig.1a). In each block, they had to perform three search tasks. They were confronted with 90 adaptations (30 SA, 30 IA and 30 NA) in random order. Each task started with an overview of 16 different apps. Participants had to open one specific app and complete the search tasks by navigating through different website pages (e.g. find the information for the weather forecast for a specific day for a specific city in the weather app). Within the app, they had to narrow their information down by navigating through different website pages (e.g. starting with weather app>selecting the country>selecting the city>selecting the day). The website pages were designed with different UI layout patterns, e.g. grid menu, dynamic tabular menu or scroll menu. These layout patterns were altered by the underlying AFFINDU system to produce SA (make it easier to find target, Fig.1c), IA (make it harder to find target, Fig.1d) and NA (irrelevant for finding target, Fig.1e). Each adaptation was followed by a pop-up asking for participants’ feedback (keep adaptation(“Yes”), revoke adaption(“No”), I don’t care(“Never mind”).The interaction was implemented as a controllable event-based procedure including different time windows (Fig.1f) to enable simultaneous recordings of EEG and near-infrared spectroscopy. Here, we only present results for the EEG analysis. For the EEG analysis we grouped the SA, IA and NA for all blocks together. Stimulus-locked epochs (200 msec before to 1000 msec after the beginning of the adaptation) were created separately for SA, IA and NA. The pre-processing and artefact correction procedure and calculation of signed signed r-square values was done in the same way as in our previous study (Pollmann et al., 2017). For each ERP we conducted a separate multiple dependent sample t-test (SA vs. IA, SA vs. NA and IA vs. NA) on the level of individual electrodes to identify possible spatial differences of adaptation-induced modulations. Here, a cluster-based, non-parametric randomization approach which included correction for multiple comparisons (FieldTrip toolbox; (Oostenveld, Fries, Maris, & Schoffelen, 2011) was performed. Fig.2 shows participants’ overall subjective feedback to the three different adaptations . The results serve as a manipulation check, clearly showing that the IA were disapproved while the SA were approved by the participants. NA resulted in a more or less equal distribution of yes, no, or never mind answers. By studying the ERP results, we found discriminative separability in the most relevant ERP-time windows for emotional appraisal, such as N200, P300 and the late positive potential (LPP). The results from the signed r-square values revealed the following TOIs to be most discriminative between adaptations: N200=190-250 msec, P300=325–410 msec, LPP=575–800 (Fig. 3). The non-parametric randomization test revealed significant differences for the ERP topographical changes among the three different adapations (Fig. 4). Electrode clusters, showing significant differences are indicated by filled black circles. We found that emotional appraisal, approve SA and disapprove IA, significantly distinguishes clusters from the NA at frontal and parietal sites. Furthermore, electrode clusters at frontal and parietal regions also significantly discriminate positive (SA) and negative (IA) appraisal checks within the N200, P300 and LPP-components. These different modulations of ERPs might index different stages of emotional...