No abstract yet

The focus of this talk is not about my research in AI or Robotics but my own journey on trying to do research and understand intelligence in a rapidly evolving research landscape. I will trace my path from conducting early-stage research during graduate school, to working on practical solutions within a startup environment, and finally to my current role where I participate in more structured research at a major tech company. Through these varied experiences, I will provide different perspectives on research and talk about how my core beliefs on intelligence have changed and sometimes even been compromised. There are no lessons to be learned from my stories, but hopefully they will be entertaining.

Digital platforms generate unprecedented traces of human behaviour, offering new methodological approaches to understanding collective action, polarisation, and social dynamics. Through analysis of millions of digital traces across multiple studies, we demonstrate how online behaviours predict offline action: Brexit-related tribal discourse responds to real-world events, machine learning models achieve 80% accuracy in predicting real-world protest attendance from digital signals, and social validation through "likes" emerges as a key driver of mobilization. Extending this approach to conspiracy narratives reveals how digital traces illuminate psychological mechanisms of belief and community formation. Longitudinal analysis of YouTube conspiracy content demonstrates how narratives systematically address existential, epistemic, and social needs, while examination of alt-tech platforms shows how emotions of anger, contempt, and disgust correlate with violence-legitimating discourse, with significant differences between narratives associated with offline violence versus peaceful communities. This work establishes digital traces as both methodological innovation and theoretical lens, demonstrating that computational social science can illuminate fundamental questions about polarisation, mobilisation, and collective behaviour across contexts from electoral politics to conspiracy communities.

Synthetic face datasets are increasingly used to overcome the limitations of real-world biometric data, including privacy concerns, demographic imbalance, and high collection costs. However, many existing methods lack fine-grained control over identity attributes and fail to produce paired, identity-consistent images under structured capture conditions. In this talk, I will present FLUXSynID, a framework for generating high-resolution synthetic face datasets with user-defined identity attribute distributions and paired document-style and trusted live capture images. The dataset generated using FLUXSynID shows improved alignment with real-world identity distributions and greater diversity compared to prior work. I will also discuss how FLUXSynID’s dataset and generation tools can support research in face recognition and morphing attack detection (MAD), enhancing model robustness in both academic and practical applications.

We developed a novel paradigm measuring implicit identity recognition using Fast Periodic Visual Stimulation (FPVS) with EEG among 16 students and 12 police officers with normal face processing abilities. Participants' neural responses to a 1-Hz tagged oddball identity embedded within a 6-Hz image stream revealed implicit recognition with high-quality mugshots but not CCTV-like images, suggesting optimal resolution requirements. Our findings extend previous research by demonstrating that even unfamiliar identities can elicit robust neural recognition signatures through brief, repeated passive exposure. This approach offers potential for objective validation of face processing abilities in forensic applications, including assessment of facial examiners, Super-Recognisers, and eyewitnesses, potentially overcoming limitations of traditional behavioral assessment methods.

Recent advances in artificial intelligence have led to new forms of misinformation, including highly realistic “deepfake” synthetic media. We conduct three experiments to investigate how and why retail investors react to deepfake financial news. Results from the first two experiments provide evidence that investors use a “realism heuristic,” responding more intensely to audio and video deepfakes as their perceptual realism increases. In the third experiment, we introduce an intervention to prompt analytical thinking, varying whether participants make analytical judgments about credibility or intuitive investment judgments. When making intuitive investment judgments, investors are strongly influenced by both more and less realistic deepfakes. When making analytical credibility judgments, investors are able to discern the non-credibility of less realistic deepfakes but struggle with more realistic deepfakes. Thus, while analytical thinking can reduce the impact of less realistic deepfakes, highly realistic deepfakes are able to overcome this analytical scrutiny. Our results suggest that deepfake financial news poses novel threats to investors.

Fast periodic visual stimulation (FPVS) has emerged as a promising tool for assessing cognitive function in individuals with dementia. This technique leverages electroencephalography (EEG) to measure brain responses to rapidly presented visual stimuli, offering a non-invasive and objective method for evaluating a range of cognitive functions. Unlike traditional cognitive assessments, FPVS does not rely on behavioural responses, making it particularly suitable for individuals with cognitive impairment. In this talk I will highlight a series of studies that have demonstrated its ability to detect subtle deficits in recognition memory, visual processing and attention in dementia patients using EEG in the lab, at home and in clinic. The method is quick, cost-effective, and scalable, utilizing widely available EEG technology. FPVS holds significant potential as a functional biomarker for early diagnosis and monitoring of dementia, paving the way for timely interventions and improved patient outcomes.

Facial expressions are inherently dynamic, and our visual system is sensitive to subtle changes in their temporal sequence. However, researchers often use dynamic morphs of photographs—simplified, linear representations of motion—to study the neural correlates of dynamic face perception. To explore the brain's sensitivity to natural facial motion, we constructed a novel dynamic face database using generative neural networks, trained on a verified set of video-recorded emotional expressions. The resulting deepfakes, consciously indistinguishable from videos, enabled us to separate biological motion from photorealistic form. Results showed that conventional dynamic morphs elicit distinct responses in the brain compared to videos and photos, suggesting they violate expectations (n400) and have reduced social salience (late positive potential). This suggests that dynamic morphs misrepresent facial dynamism, resulting in misleading insights about the neural and behavioural correlates of face perception. Deepfakes and videos elicited largely similar neural responses, suggesting they could be used as a proxy for real faces in vision research, where video recordings cannot be experimentally manipulated. And yet, despite being consciously undetectable as fake, deepfakes elicited an expectation violation response in the brain. This points to a neural sensitivity to naturalistic facial motion, beyond conscious awareness. Despite some differences in neural responses, the realism and manipulability of deepfakes make them a valuable asset for research where videos are unfeasible. Using these stimuli, we proposed a novel marker for the conscious perception of naturalistic facial motion – Frontal delta activity – which was elevated for videos and deepfakes, but not for photos or dynamic morphs.

Vulnerability to distraction varies across the general population and significantly affects one’s capacity to stay focused on and successfully complete the task at hand, whether at school, on the road, or at work. In this talk, I will begin by discussing how distractibility is typically assessed in the literature and introduce our innovative ERP approach to measuring it. Since distractibility is a cardinal symptom of ADHD, I will introduce its most widely used paper-and-pencil screening tool for the general population as external validation. Following that, I will present the Load Theory of Attention and explain how we used perceptual load to test the reliability of our neural marker of distractibility. Finally, I will highlight potential future applications of this marker in clinical and educational settings.

Do You Know Your Blood Glucose Level? You Probably Should! A single measurement is not enough to truly understand your metabolic health. Blood glucose levels fluctuate dynamically, and meaningful insights require continuous monitoring over time. But glucose is just one example. Many other molecular concentrations in the body are not static. Their variations are influenced by individual physiology and overall health. PhenoSign, a Swiss MedTech startup, is on a mission to become the leader in real-time molecular analysis of complex fluids, supporting clinical decision-making and life sciences applications. By providing real-time, in-situ molecular insights, we aim to advance medicine and transform life sciences research. This talk will provide an overview of PhenoSign’s journey since its inception in 2022—our achievements, challenges, and the strategic roadmap we are executing to shape the future of real-time molecular diagnostics.

Lapses in attention are ubiquitous and, unfortunately, the cause of many tragic accidents. One potential solution may be to develop assistance systems which can use objective, physiological signals to monitor attention levels and predict a lapse in attention before it occurs. As it stands, it is unclear which physiological signals are the most reliable markers of inattention, and even less is known about how reliably they will work in a more naturalistic setting. My project aims to address these questions across two experiments: a lab-based experiment and a more ‘real-world’ experiment. In this talk I will present the findings from my lab experiment, in which we combined EEG and pupillometry to detect markers of inattention during two computerised sustained attention tasks. I will then present the methods for my second, more ‘naturalistic’ experiment in which we use the same methods (EEG and pupillometry) to examine whether these markers can still be extracted from noisier data.

Generative AI is fundamentally transforming the software development industry by improving processes such as software testing, bug detection, bug fixes, and developer productivity. This talk explores how AI-driven techniques, particularly large language models (LLMs), are being utilized to generate realistic test scenarios, automate bug detection and repair, and streamline development workflows. As these technologies evolve, they promise to improve software quality and efficiency significantly. The discussion will cover key methodologies, challenges, and the future impact of generative AI on the software development lifecycle, offering a comprehensive overview of its revolutionary potential in the industry.

Many situations rely on the accurate identification of people with whom we are unfamiliar. For example, security at airports or in police investigations require the identification of individuals from photo-ID. Yet, the identification of unfamiliar faces is error prone, even for practitioners who routinely perform this task. Indeed, even training protocols often yield no discernible improvement. The challenge of unfamiliar face identification is often thought of as a perceptual problem; however, this assumption ignores the potential role of decision-making and its contributing factors (e.g., criterion placement). In this talk, I am going to present a series of experiments that investigate the role of decision-making in face identification.

Recent research has seen many behavioral comparisons between humans and deep neural networks (DNNs) in the domain of image classification. Often, comparison studies focus on the end-result of the learning process by measuring and comparing the similarities in the representations of object categories once they have been formed. However, the process of how these representations emerge—that is, the behavioral changes and intermediate stages observed during the acquisition—is less often directly and empirically compared. In this talk, I'm going to report a detailed investigation of the learning dynamics in human observers and various classic and state-of-the-art DNNs. We develop a constrained supervised learning environment to align learning-relevant conditions such as starting point, input modality, available input data and the feedback provided. Across the whole learning process we evaluate and compare how well learned representations can be generalized to previously unseen test data. Comparisons across the entire learning process indicate that DNNs demonstrate a level of data efficiency comparable to human learners, challenging some prevailing assumptions in the field. However, our results also reveal representational differences: while DNNs' learning is characterized by a pronounced generalisation lag, humans appear to immediately acquire generalizable representations without a preliminary phase of learning training set-specific information that is only later transferred to novel data.

Within the last decade, Deep Artificial Neural Networks (DNNs) have emerged as powerful computer vision systems that match or exceed human performance on many benchmark tasks such as image classification. But whether current DNNs are suitable computational models of the human visual system remains an open question: While DNNs have proven to be capable of predicting neural activations in primate visual cortex, psychophysical experiments have shown behavioral differences between DNNs and human subjects, as quantified by error consistency. Error consistency is typically measured by briefly presenting natural or corrupted images to human subjects and asking them to perform an n-way classification task under time pressure. But for how long should stimuli ideally be presented to guarantee a fair comparison with DNNs? Here we investigate the influence of presentation time on error consistency, to test the hypothesis that higher-level processing drives behavioral differences. We systematically vary presentation times of backward-masked stimuli from 8.3ms to 266ms and measure human performance and reaction times on natural, lowpass-filtered and noisy images. Our experiment constitutes a fine-grained analysis of human image classification under both image corruptions and time pressure, showing that even drastically time-constrained humans who are exposed to the stimuli for only two frames, i.e. 16.6ms, can still solve our 8-way classification task with success rates way above chance. We also find that human-to-human error consistency is already stable at 16.6ms.

Three studies conducted in the context of PhD work (UNIL) aimed at proving evidence to address the question of potential gender differences in trait anxiety and executive control biases on behavioral efficacy. In scope were male and female non-clinical samples of adult young age that performed non-emotional tasks assessing basic attentional functioning (Attention Network Test – Interactions, ANT-I), sustained attention (Test of Variables of Attention, TOVA), and visual recognition abilities (Object in Location Recognition Task, OLRT). Results confirmed the intricate nature of the relationship between gender and health trait anxiety through the lens of their impact on processing efficacy in males and females. The possibility of a gendered theory in trait anxiety biases is discussed.

I my talk I will give an overview of recent research on deception and concealed information detection. I will start with a short introduction on the problems and shortcomings of traditional deception detection tools and why those still prevail in many recent approaches (e.g., in AI-based deception detection). I want to argue for the importance of more fundamental deception research and give some examples for insights gained therefrom. In the second part of the talk, I will introduce the Concealed Information Test (CIT), a promising paradigm for research and applied contexts to investigate whether someone actually recognizes information that they do not want to reveal. The CIT is based on solid scientific theory and produces large effects sizes in laboratory studies with a number of different measures (e.g., behavioral, psychophysiological, and neural measures). I will highlight some challenges a forensic application of the CIT still faces and how scientific research could assist in overcoming those.

Emotion is defined as a rapid psychological process involving experiential, expressive and physiological responses. These emerge following an appraisal process that involves cognitive evaluations of the environment assessing its relevance, implication, coping potential, and normative significance. It has been suggested that changes in appraisal processes lead to changes in the resulting emotional nature. Simultaneously, it was demonstrated that personality can be seen as a predisposition to feel more frequently certain emotions, but the personality-appraisal-emotional response chain is rarely fully investigated. The present project thus sought to investigate the extent to which personality traits influence certain appraisals, which in turn influence the subsequent emotional reactions via a systematic analysis of the link between personality traits of different current models, specific appraisals, and emotional response patterns at the experiential, expressive, and physiological levels. Major results include the coherence of emotion components clustering, and the centrality of the pleasantness, coping potential and consequences appraisals, in context; and the differentiated mediating role of cognitive appraisal in the relation between personality and the intensity and duration of an emotional state, and autonomic arousal, such as Extraversion-pleasantness-experience, and Neuroticism-powerlessness-arousal. Elucidating these relationships deepens our understanding of individual differences in emotional reactivity and spot routes of action on appraisal processes to modify upcoming adverse emotional responses, with a broader societal impact on clinical and non-clinical populations.

Understanding and potentially reversing memory decline necessitates a comprehensive examination of memory's evolution throughout life. Traditional memory assessments, however, suffer from a lack of comparability across different age groups due to the diverse nature of the tests employed. Addressing this gap, our study introduces a novel, ACT-R model-based memory assessment designed to provide a consistent metric for evaluating memory function across a lifespan, from 5 to 85-year-olds. This approach allows for direct comparison across various tasks and materials tailored to specific age groups. Our findings reveal a pronounced U-shaped trajectory of long-term memory function, with performance at age 5 mirroring those observed in elderly individuals with impairments, highlighting critical periods of memory development and decline. Leveraging this unified assessment method, we further investigate the therapeutic potential of rs-fMRI-guided TBS targeting area 8AV in individuals with early-onset Alzheimer’s Disease—a region implicated in memory deterioration and mood disturbances in this population. This research not only advances our understanding of memory's lifespan dynamics but also opens new avenues for targeted interventions in Alzheimer’s Disease, marking a significant step forward in the quest to mitigate memory decay.

In 2014, the US National Research Council called for the development of new lineup technologies to increase eyewitness identification accuracy (National Research Council, 2014). In a police lineup, a suspect is presented alongside multiple individuals known to be innocent who resemble the suspect in physical appearance know as fillers. A correct identification decision by an eyewitness can lead to a guilty suspect being convicted or an innocent suspect being exonerated from suspicion. An incorrect decision can result in the perpetrator remaining at large, or even a wrongful conviction of a mistakenly identified person. Incorrect decisions carry considerable human and financial costs, so it is essential to develop and enact lineup procedures that maximise discrimination accuracy, or the witness’ ability to distinguish guilty from innocent suspects. This talk focuses on new technology and innovation in the field of eyewitness identification. We will focus on the interactive lineup, which is a procedure that we developed based on research and theory from the basic science literature on face perception and recognition. The interactive lineup enables witnesses to actively explore and dynamically view the lineup members. The procedure has been shown to maximize discrimination accuracy, which is the witness’ ability to discriminate guilty from innocent suspects. The talk will conclude by reflecting on emerging technological frontiers and research opportunities.