Information Systems and Neuroscience

Online purchase environments such as Amazon are beginning to utilize generative artificial intelligence (GenAI)-based review summaries (GARS) as representations that aggregate product information and associated valence from large numbers of product reviews. Although GARS may enable the unprecedented, accurate summarization of large volume of information found within product reviews, potentially assisting users’ decision-making, they are likely to tease out and juxtapose the positive and negative valence prevalent in product reviews concurrently. We examine how such GARS heighten the simultaneous activation of positive and negative dispositions—ambivalence—and theorize how such ambivalence affects users’ attention and purchase intentions in online contexts. We design and conduct an eye-tracking based within-subject laboratory experiment to investigate these effects. Results suggest that the presence of GARS in online purchase environments is likely to increase users’ perceived ambivalence and subsequently impact their cognitive and behavioral outcomes, which are identifiable via eye-tracking techniques.

In the era of ChatGPT and other generative AI tools, white-collar workers are given tremendous potential to simplify everyday tasks. Within vocational psychology, this phenomenon is known as job crafting. We conduct an electroencephalography-based mixed-factorial experiment to explore the underlying mechanisms of how and why the use of generative AI tools can lead to job crafting. Relying on cognitive load theory and resource demand theory, we measure the effects of ChatGPT use and prompt engineering guidance in strategic thinking tasks. We hypothesize that individuals who use ChatGPT without and with prompting examples rely on cognitive offloading to avoid cognitive effort, affecting resource demands. An initial evaluation of our experiment task design provides promising results. We plan our experiment with participants who are familiar with executive assistant tasks. Our expected results contribute to the ongoing discussion of ICT-enabled job crafting and provide empirical-driven explanations of AI-enabled job crafting mechanisms.

In this study, we investigate user interactions on an Artificial Intelligence (AI)-enabled platform, featuring a chatbot designed to augment user engagement in learning how to solve a puzzle. Our platform’s core AI algorithm leverages macro-actions-sequences of moves, not interpretable or usable to users, essentially functioning as a ‘black box.’ To address this, we employed scaffolding design strategies and explainable AI principles to develop an innovative conversational user interface (UI). Utilizing eye-tracking techniques, we collected users’ gaze data to assess their gaze patterns and attention distribution during a problem-solving process. By focusing on gaze metrics, e.g., fixation duration, saccade frequency, area of interest, and heatmap, we found users’ visual attention correlates with scaffolding-enabled UI elements, and thus positively influencing their problem-solving experience. This preliminary study allows us to evaluate the UI's intuitiveness and identify design strategies for user engagement improvement, adaptable to diverse learning styles, and thus potentially enhancing problem-solving efficiency on our AI platform.

This paper investigates how the believability of news headlines is impacted when those news headlines are accompanied by generative AI images. We reports on a series of in-progress experiments. The results suggest that the use of AI images reduces the believability of news headlines, especially when those news headlines are presented by news institutions, rather than named individuals. The negative impact of AI images on believability does not seem to be attenuated by warning labels which state that the images used are AI generated. These findings have implications for online news, and other application domains where believability is important.

The use of Human-AI teams is becoming more common in service contexts. However, customers do not always trust the recommendations of AI agents. We draw on social identity theory to show that trust in human-led teams is higher than in AI-led teams. We use behavioral (questionnaires on trust and compliance) and neuroscientific measures (EEG) to investigate customer perceptions. We demonstrate that individuals invest more cognitive effort when faced with recommendations from AI-led service providers. This is corroborated by higher alpha-band activity in the AI-led team condition. We also show that individuals trust AI-led teams less and are less likely to comply with their recommendations. We contribute to social identity theory and propose that managers adopt different communication styles when they employ human-led versus AI-led teams.

The use of Autonomic Nervous System (ANS) activity measurements provides valuable insight into the physiological responses elicited by cognitive and emotional stimuli during human interactions with Information Systems (IS), thereby furthering the development of NeuroIS research. This paper aims to outline the potential for future research efforts by exploring the applicability of such measurements in the field of interruption science. Extending a previous review of neurophysiological measurements in interruption science, our focus is on unraveling the methodological complexity found in the identified papers. Specifically, we provide an overview of the various ANS activity measurements applied in interruption science research. Furthermore, to lay a foundation for understanding the use of such measurements in this research field, we present an example study per measurement (i.e., eye-related, heart-related, skin- or body-related, muscle-related, or respiratory-related measurements), outlining following factors: research objective, research method, sample size, study population, and research results.

The application of brain imaging methods can provide invaluable insights into the neural substrates of human cognition, emotion, behavior, and performance, thereby facilitating a profound understanding of information processing in the context of Neuro-Information Systems (NeuroIS). In this paper, we aim to pave the way for future research efforts by detailing the applicability of such methods in the research field of interruption science. Building on a previous review of neurophysiological measurements in interruption science, we focus on the methodological aspects of the identified papers. Specifically, we provide an overview of how brain imaging methods can be used in interruption science by providing an overview of the applied brain imaging methods and their measurement purpose. We also describe one example study used in the domains of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) by detailing the following factors: research objective, research method, sample size, study population, and research results.

Hormone measurements can serve as crucial indicators for assessing an individual's physiological state in the context of Neuro-Information Systems (NeuroIS). Understanding hormonal responses to interruptions can provide insights into human cognition, emotion, behavior, and performance. By analyzing hormone levels, researchers can measure the effect of interruptions on an individual's physiological processes, including performance in social interactions or performance in human–computer interaction situations. This nexus of NeuroIS and interruption science provides a unique perspective on how technological interruptions affect the neurophysiological responses of individuals. Building on a previous umbrella review of neurophysiological measurements in interruption science, in the current paper we review completed empirical studies with a focus on the methodological aspects of the hormone measurements in the papers. Specifically, we describe identified studies that applied hormone measurements with saliva or urine samples by detailing the following factors: research objective, research method, sample size, study population, and research results. Finally, to lay a foundation for future research activities based on the paper analysis, this review outlines methodological complexities of hormone measurements at the nexus of interruption science and NeuroIS.

The circulation of fake news on social media platforms is a major concern. This phenomenon is exacerbated by users’ mindless and rapid scrolling through news feeds and the copious amounts of information that users are exposed to on social media platforms. Although judgement is accompanied by a variety of subjective experiences such as fluency, research on fake news on social media has largely strayed away from understanding the role of these experiences on the perception of fake news. To address this gap, we propose a NeuroIS approach to investigate how two manipulations of fluency: (i) time constraint and (ii) information overload, impact users’ belief in fake news on social media. We also aim to uncover the neural mechanisms by which these manipulations impact truth judgements, which may shed light on the mechanisms by which fake news become entrenched and assist in combatting the fake news proliferation we are witnessing today.

In brain-computer interface (BCI) research, electroencephalograms (EEGs) such as the Unicorn Hybrid Black (UHB) have entered the market as low-cost alternatives to other EEG devices. This study has two aims: the first is to assess the suitability of the UHB for BCI research, and the second is to assess the feasibility of a meditation BCI designed to provide users with feedback about mind wandering episodes. A BCI was created using the UHB and corresponding Python API to assess various machine learning algorithms’ classification accuracy of a meditation paradigm that uses self-caught experience sampling to capture mind wandering. Key findings suggest that while the UHB is sufficient to capture relevant brain signals associated with mind wandering, more research is required on appropriate intervention techniques.

This ongoing study is investigating the effects of confirmation bias and document readability on user document relevance judgments in interactive information systems. Preliminary results from a within-subjects eye-tracking experiment suggest a significant interaction between document readability and the level of confirmation bias on user engagement during the relevance evaluation process. Higher readability has been found to enlarge the disparity in reading efforts between relevant and irrelevant documents. Users with higher level of confirmation bias need to engage more attention when evaluating relevance. These initial findings help to elucidate these dynamic relationships and provide insights that could help refine user interface design for more effective and unbiased human information interaction.

Recruiting diverse participants for NeuroIS studies is a major research challenge. The requirement for physical participation in NeuroIS labs poses geographical, physical, and financial barriers to many equity deserving groups, such as older adults, people with disabilities, and patients with mental or physical diseases. As a result, many such equity deserving user groups are excluded from NeuroIS research and are consequently unable to inform the research and development of technologies. This is extremely unfortunate, since their input would inform the designs of technologies that can potentially improve the quality of their lives, if they are afforded an equitable opportunity to partake in NeuroIS research activities. We present a case where these challenges were faced by our research team. We propose two potential solutions to tackle this significant Equity, Diversity, and Inclusion (EDI) challenge: (1) mobile NeuroIS labs; and (2) data collection in the field.

Artificial intelligence-based decision-support systems (AI DSS), powered by complex algorithms, often lack transparency. To tackle this challenge, organizations deploy explainable artificial intelligence (XAI). However, studies reveal that XAI use does not necessarily result in enhanced human-XAI performance. Recently, a call has been made for more AI studies from the user perspective to better understand this phenomenon. Approaches to such studies need advancement, too. Indeed, most existing studies on cognitive mechanisms behind XAI-supported decision-making rely on integration of behavioral data and think-aloud protocols, post-hoc surveys, or interviews. Neurocognitive mechanisms behind XAI-supported decision-making remain a black box. The goal of the current paper is to provide a basis for neurophysiological studies on XAI-supported decision-making in organizational context. For this, we conduct an integrative literature review at the intersection of three domains: XAI from the user perspective; neurophysiological lens on XAI; XAI within the three-stage dual-process model of cognition.

Rehabilitation therapists increasingly utilize virtual reality (VR) to enhance exercise engagement and customization. However, the lack of realistic sensory feedback from traditional VR controllers may impede motor learning. This study aims at assessing the effect of a VR controller adapter on electroencephalography (EEG) neural correlates of motor learning in VR settings. We will collect the data of 30 healthy participants performing virtual pool billiard shots. The results of this study will guide the development of more effective naturalistic VR-based motor tasks in rehabilitation.

In multi-modal learning analytics, one collects biometric data from different sensors, including EEG, eye-tracking, wristbands, and facial expression (through cameras). This paper presents an approach of detection causal relationships between different measurements taken under an experiment of comprehension of business process representations. The results identify differences between high and low performers. Future work will describe additional results from the experiment and see how this insight can be used in supporting process model comprehension and learning from process models, including providing tool-support as a scaffold in the modelling process.

Interest in the use of neurophysiological instruments for real-world studies in the workplace is increasing, also intensified by the simultaneously growing use of various commercial self-tracking technologies. However, the application of neurophysiological tools for real-world workplace research is associated with challenges—an aspect that has received little attention in previous research. This article outlines the key challenges encountered when applying neurophysiological measurements in the workplace, drawing on insights gained in an interdisciplinary research project on digital workplaces. We identify challenges along four main themes: technical tool requirements, data processing and interpretation, tool interaction, and organizational collaboration. Additionally, we discuss how these challenges were addressed within our case. As a contribution, this article offers important considerations and recommendations for the effective application of neurophysiological tools in real-world workplace research.

Understanding the physiological underpinnings of consumer preferences is key to enhancing online decision-making platforms. This research explores whether the attributes traditionally valued in hotel selections also trigger measurable physiological arousal, combining neurophysiological measurements with quantitative analysis to deepen our understanding of consumer behavior in the digital age. This study explores the influence of physiological arousal on online hotel selection decisions, utilizing skin conductance, eye-tracking, and choice-based conjoint analysis to assess the correlation between physiological arousal and attribute importance. Findings of this laboratory experiment reveal a significant positive correlation between arousal levels and the importance of hotel attributes such as review valence, star category, and proximity to town center. Highlighting the impact of cognitive responses on consumer preferences, this research underscores the potential of neurophysiological insights to enhance digital booking platforms, contributing to the NeuroIS field by demonstrating the integration of neurophysiological and quantitative measurements to deepen understanding of consumer behavior.

The rapidly growing capabilities of modern sensor technology combined with AI technologies now make it technically possible to capture biosignals in real time and use them to develop biosignal-adaptive systems. However, as the processing chain of biosignals is complex, the challenge is to ensure transparency and replicability. We consider the pursuit of an open science paradigm to be particularly promising to address this challenge. In 2023, we presented the initial concept of the HUman BIosignal Intelligence Platform (HUBII) at NeuroIS. Over the last year, we have piloted HUBII as part of a capstone semester project in a lecture. We received valuable comments and suggestions for improvement from the pilot users. Thus, we are proposing HUBII 2.0 with an updated architecture following the Open Science paradigm. HUBII 2.0 serves as a central entry point for biosignal intelligence and provides easy discoverability of biosignal pipelines through a systematic classification along the dimensions of human activity, sensor, device and biosignal. Furthermore, it supports referencing of citable pipelines using existing infrastructures such as Hugging Face.

This study examines the Theta/Beta Ratio (TBR) in EEG measurements to infer the cognitive load during decision-making across different attachment styles. Participants’ EEGs were analyzed within the secretary problem framework while they were at rest and making decisions. The secure attachment group demonstrated higher TBR difference medians, suggesting a lower cognitive load. Conversely, the fearfully avoidant group exhibited significantly lower medians, indicating a heightened cognitive load. The anxiously attached group’s TBR varied widely, suggesting inconsistent cognitive engagement, while the avoidant group’s levels were moderate. These results imply that attachment style determines the cognitive load experienced during decision-making, with fearfully avoidant individuals bearing the highest load. This research elucidates the neurobiological links between attachment styles and cognitive processing.

Gaming has become a prevalent entertainment and leisure activity worldwide, particularly among young adults. While gaming can provide enjoyment and social interaction, its impact on well-being remains a concern, especially among the next generation of adults. There is evolving interest in understanding the relationship between emotions, gaming, and emotional well-being. Electroencephalography signals recorded in real-time and integrated with self-report measures can provide a more comprehensive understanding of emotional momentary experiences during gaming. This proposal aims to investigate the relationship between emotions, gaming, and well-being by combining self-report measures with neurophysiological data to explore emotional well-being in college students.

Eye contact is an essential non-verbal cue in human communication. Due to the absence of gaze information in video meetings, however, team members are not able to perceive and anticipate eye contact in this remote setting. To solve this problem, we explore how episodes of mutual gaze can be visualized in video meeting systems using eye tracking technology and report preliminary findings by conducting the first cycle of a design science research project. The preliminary results of our study indicate that gaze-adaptive visualization of mutual gaze increases trust between remote team members in video meetings. However, participants also indicated that the visualization might be distracting. Future studies should therefore investigate the effect on team performance more closely.

Hormone fluctuations due to the menstrual cycle can significantly impact women’s physical and cognitive performance. While athletes already incorporate this in training plans to optimize training outcomes, it is still neglected in knowledge work. To address this gap, we review existing literature on hormone fluctuations and their impact on cognitive performance to design a cycle-sensitive task manager for knowledge workers (this paper). The proposed follow-up design science study should incorporate both the physiological menstrual cycle data as well as tagging tasks based on their required cognitive performance (future paper). The resulting app can then arrange the upcoming tasks in accordance with the cognitive performance strengths of the respective menstrual phase. This will enhance awareness among female employees while allowing an increase in task performance through cycle-sensitive work task management.

Hyper-realistic avatars (HRAs) are digital embodiments of a human, created by capturing and combining that person’s video and vocal likenesses. We study the efficacy of videos delivered by HRAs as a communication channel in business-to-business applications as compared to videos delivered by their human counterparts. An online experiment with 290 participants gathered performance and biometric data across three video treatments. There are no statistically significant differences in information retention, engagement, or trust between respondents who watched the real human versus the HRA. A significant (p = 0.003) positive difference is found in information retention between the HRA-delivered video and the video delivered by the HRA with disclosure. This study lays the foundation for future HRA research directions.

Large language models (LLMs) are transforming the way we work, learn, and access information. As our dependence on these tools grows, it becomes crucial to enhance their accuracy and ensure they align with our ethical standards. The most high-performing language models are currently trained and refined with the help of explicit human feedback. Here we propose a study that investigates the feasibility of implicit human feedback through passive brain-computer interfaces (pBCIs). Two calibration paradigms for moral judgment and error-perception elicitation and detection are described. The obtained classification models will be tested in an application phase with simulated chatbot conversations. If proven successful, pBCIs could provide novel and informative human implicit feedback in the process of LLM development.

Many laboratory experiments use real-effort tasks to increase the external validity of their findings. Real-effort tasks activate emotional reactions that are absent in stated-effort tasks. But there is little evidence whether and to which extent emotional reactions differ between participants, and how they affect effort provision. Since self-reported measures can be sensitive to the experimental context, we use heart rate measurements to investigate how participants feel during the task. We conducted a real-effort experiment with 84 participants and collected heart rate data with Polar H10 Heart Rate Sensors. We applied time series clustering on the heart rate data, focusing on shape-based distance (SBD) and k-shape clustering for analysis. The results demonstrate differences in heart rate patterns among participant clusters, but not in effort provision. This research contributes to a better understanding of emotional reactions during real-effort tasks and offers a novel approach to studying these emotions using heart rate measurements.

Deep fake technology poses a significant threat to the political discourse. Within the realm of information systems (IS) research, the study of fake news on social media has primarily focused on written text, with comparatively little emphasis on deep fake videos. However, the effectiveness of visual content in persuading users of the authenticity of the events presented surpasses that of textual content, a phenomenon underscored by IS research. In this study, we analyze whether the theories of confirmation bias and cognitive dissonance are also applicable to deep fake videos. To this end, we propose a between-subject experiment with measurements of brain activity and heart rate to shed light on the underlying cognitive and affective processes that guide users’ behavior and beliefs. Accordingly, we contribute to the IS literature by employing neurophysiological measurements to analyze human processing of deep fakes, which assists social media platforms in developing effective mitigation strategies.

The information systems (IS) literature has already appreciated that human decisions often deviate from what would be considered ‘rational’. However, the complex role of attention during decision-making has sparked less discussion. Here, we present our ongoing work on a heuristic framework that combines recent computational theoretical accounts from neuroeconomics on an active role of attention in reward anticipation and evidence accumulation with neurobiologically plausible models of attentional control. We discuss how seemingly unrelated behavioral phenomena can arise from a set of attentional processes in the light of this framework. Furthermore, we point out how researchers and practitioners may use our model to design more dynamic digital choice architectures–i.e., using real-time data from eye-tracking or electroencephalography (EEG)–and to develop digital training interventions targeting enduring changes in attentional control to eventually empower individuals to make healthier or more sustainable decisions.

This study addresses the challenges of cognitive overload in video meetings by proposing a cognitive load-adaptive microbreak intervention based on non-invasive recording of biosignals. Through the systematic exploration of the problem space, we derive meta-requirements from literature and semi-structured interviews, laying the foundation for the solution design. We propose a system architecture for a neuro-adaptive video meeting system that automatically suggests microbreaks during video meetings when a predetermined number of participants exhibit elevated cognitive load using EEG. To ensure flexibility and alignment with the ongoing video meeting, we suggest granting meeting participants the option to postpone breaks. The study represents the groundwork for an ongoing design science research project designing cognitive load-adaptive microbreak interventions to improve virtual team meeting performance.

Functional illiterates make up a significant portion of the population and need accessible interfaces. Icons, among other interface elements, are a design consideration that is often used to improve accessibility. However, neurophysiological evidence on their effectiveness in functional illiterates remains scant. In a controlled experiment based on online banking stimuli, this study used EEG and behavioral measures to objectively evaluate the efficaciousness of icons compared to words in literate and functional illiterate subjects. Results show that icons alone might not be sufficient to increase accessibility because their abstract nature can hinder their interpretation.

Smishing, a rising phishing threat via SMS, affects millions of mobile users and businesses annually. It employs psychological persuasion techniques to deceive users into downloading malware and disclosing sensitive information, rendering victims vulnerable to social and financial exploitation. Research on phishing has typically focused on emails and social networks. However, despite its increasing prevalence, smishing remains understudied. In this research-in-progress, we aim to address this gap by building on Cognitive Dissonance Phishing Persuasion Theory and present an experimental study that will explore users’ cognitive and emotional responses to the most prevalent persuasion techniques in practice. We hypothesize that certain phrases (e.g., limited offers) and persuasion techniques (e.g., scarcity) will elicit distinct user responses (e.g., cognitive load, technostress) that render users more vulnerable to smishing. Employing eye-tracking, pupillometry, EEG, and EDA, this study pioneers the investigation of users’ neurophysiological responses to smishing, enhancing our understanding of phishing within the SMS domain.

The impact of AI tools like ChatGPT on cognitive load in knowledge work is not yet fully understood in the evolving field of human-AI interaction. This study aims to explore the cognitive load dynamics arising from AI-assisted tasks, revealing their potential to streamline workflows, but also risking cognitive overload, potentially hindering task performance, learning, and enjoyment. Anchored in cognitive load theory, our research proposes an experiment that leverages wearable EEG technology to empirically investigate the cognitive load fluctuations experienced by individuals engaged in AI-assisted programming tasks. By dissecting the interplay between user engagement and AI assistance, this study seeks to uncover effective patterns of AI tool usage that mitigate cognitive overload. Thereby, we aim to contribute to cognitive load theory by detailing the interactive load dynamics inherent in AI-assisted work.

Existing studies have reported on the potentially harmful effect of Instagram for women’s self esteem and body perception. This study investigates the differences when women browse Instagram posts with images that have been edited to look more conventionally attractive, and posts that include unedited images. Preliminary results from an online experiment (n = 400) suggest that women who view edited images have lower self esteem and higher body image concerns, but only if those women choose to “like” the posts. We describe an ongoing follow-up laboratory study that uses EEG, alongside eye-tracking and galvanic skin response, to analyze the psychological and physiological effects when women view these two types of posts.

Decision support systems that evaluate user decisions have the potential to improve financial decision-making by alerting users to potentially disadvantageous choices. However, the feasibility of such systems, especially in complex decision-making scenarios, remains underexplored. This work in progress aims to investigate to what extend EEG-based decision support systems can be implemented using current technology. In a pilot study, we adapted the Iowa Gambling Task, a well-established decision-making paradigm, and collected 33-channel EEG data from three participants. As a proof of concept, we used a convolutional neural network (EEGNet) to classify positive and negative feedback, achieving subject-dependent binary classification accuracies ranging from 67 to 75%. These findings demonstrate the potential for developing and evaluating decision support systems that detect suboptimal decisions in real-world financial applications.

The finance sector, just like e-commerce, utilizes online platforms (websites or mobile apps) to deliver its services or products, making usability and user experience one of the key concerns of digital banking. Having identified a research gap in using psychophysiological data to understand the determinants of customer satisfaction on digital platforms, this study focuses on predicting factors influencing users’ satisfaction and intention to recommend a banking website using both self-reported and psychophysiological data. With a within-subject study design, we collected data on 100 participants. Our research-in-progress aims to develop a machine learning model capable of predicting real-time user satisfaction and the likelihood of a user recommending a digital banking experience to friends or colleagues. Results showed that psychophysiological metrics improved the prediction of users’ intention to recommend. Similar features such as Phasic EDA, pupil size, time-to-first-mouse-click, k-coefficient, emotional valence, and subjective success were found to be good predictors of both intention to recommend and customer satisfaction.

We propose a panel on neurophysiological approaches for understanding information seeking behavior.