Etienne Benoit Roesch
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Etienne Benoit Roesch has a total of 140 co-authors. Here are the top co-authors by number of shared publications:
- Asad Malik
Brain Embodiment Lab, School of Systems Engineering, University of Reading, Reading, UK
9 shared publications - Slawomir Jaroslaw Nasuto
University of Reading
7 shared publications - Alexis Kirke
Interdisciplinary Centre for Computer Music Research, University of Plymouth, Plymouth, PL4 8AA, UK
6 shared publications - Duncan Williams
Acoustics Research Centre, School of Science, Engineering and Environment, University of Salford 1 , Salford M5 4WT
6 shared publications - Ian Daly
University of Essex
5 shared publications James Weaver
Brain Embodiment Laboratory, Biomedical Engineering Section, School of Biological Sciences, University of Reading, Reading, United Kingdom
5 shared publications
Latest Publications
- Open Access
- 25 August 2026
- Published by Public Library of Science (PLoS) in PLOS Mental Health
- Vol. 3 (8) , e0000691
A Bayesian approach for exploring the effects of household income on self-reported mental health measures during the COVID-19 pandemic
The relationship between income and mental health is well-established yet complex, as both can be confounded by age, gender, and other demographic factors. To clarify these dynamics, this study examines the effects of age and income on symptoms of depression and generalised anxiety across female and male genders during the COVID-19 pandemic in the UK. Data from Wave-1 and Wave-6 of the COVID-19 Psychological Research Consortium (C19PRC) study were analysed using Bayesian hierarchical ordered-logistic mediation models. Results showed that older age was linked to less depression and anxiety in both genders. Higher income was linked to better mental health in males in both waves. For females, however, higher income was only weakly linked to more symptoms early in the pandemic, but by Wave‑6, it had become a substantially protective factor for their mental health. Additionally, males reported better mental health than females across most income groups, except at the lowest income group, where both genders experienced comparable distress. Using robust inferential statistical models, our findings explain the nuanced relationships between socioeconomic factors and mental health across genders during a public health crisis. This evidence can inform targeted mental health policies and economic support programs for future emergencies.
- 25 August 2026
- Published by Public Library of Science (PLoS) in PLOS Mental Health
- Vol. 3 (8) , e0000691
Open AccessA Bayesian approach for exploring the effects of household income on self-reported mental health measures during the COVID-19 pandemic
The relationship between income and mental health is well-established yet complex, as both can be confounded by age, gender, and other demographic factors. To clarify these dynamics, this study examines the effects of age and income on symptoms of depression and generalised anxiety across female and male genders during the COVID-19 pandemic in the UK. Data from Wave-1 and Wave-6 of the COVID-19 Psychological Research Consortium (C19PRC) study were analysed using Bayesian hierarchical ordered-logistic mediation models. Results showed that older age was linked to less depression and anxiety in both genders. Higher income was linked to better mental health in males in both waves. For females, however, higher income was only weakly linked to more symptoms early in the pandemic, but by Wave‑6, it had become a substantially protective factor for their mental health. Additionally, males reported better mental health than females across most income groups, except at the lowest income group, where both genders experienced comparable distress. Using robust inferential statistical models, our findings explain the nuanced relationships between socioeconomic factors and mental health across genders during a public health crisis. This evidence can inform targeted mental health policies and economic support programs for future emergencies.
- Open Access
- 18 December 2024
- Published by MDPI AG in Applied Sciences
- Vol. 14 (24) , 11855
We Are Not Equipped to Identify the First Signs of Cyber–Physical Attacks: Emotional Reactions to Cybersecurity Breaches on Domestic Internet of Things Devices
The increasing number of domestic Internet of Things (IoT) devices in our lives leads to numerous benefits, but also comes with an increased risk of cybersecurity breaches. These breaches have psychological consequences for the users. We examined the nature of the psychological impact of cybersecurity breaches on domestic IoT by investigating emotional experiences in a scenario study (Study 1) and a field experiment (Study 2) using the five emotion components of the Component Process Model (CPM) and emotion regulation as a framework. We replicated a three-dimensional structure for emotional experiences found in a previous study, with an addition of an ancillary fourth dimension in the second study. The first dimension represents emotional intensity. The second bipolar dimension describes constructive vs. unconstructive action tendencies. On the third dimension, also bipolar, cognitive and motivational emotion features are opposed to affective emotion features. The fourth dimension, labeled distress symptoms, mainly reflects negative emotions. In Study 2, most of the introduced frequent irregularities on IoT devices were not noticed, and the intensity of emotional reactions and tendencies to react in a constructive way decreased throughout the phases of the experiment. These findings reveal that we are not emotionally equipped to identify potential threats in the cyber world.
- 18 December 2024
- Published by MDPI AG in Applied Sciences
- Vol. 14 (24) , 11855
Open AccessWe Are Not Equipped to Identify the First Signs of Cyber–Physical Attacks: Emotional Reactions to Cybersecurity Breaches on Domestic Internet of Things Devices
The increasing number of domestic Internet of Things (IoT) devices in our lives leads to numerous benefits, but also comes with an increased risk of cybersecurity breaches. These breaches have psychological consequences for the users. We examined the nature of the psychological impact of cybersecurity breaches on domestic IoT by investigating emotional experiences in a scenario study (Study 1) and a field experiment (Study 2) using the five emotion components of the Component Process Model (CPM) and emotion regulation as a framework. We replicated a three-dimensional structure for emotional experiences found in a previous study, with an addition of an ancillary fourth dimension in the second study. The first dimension represents emotional intensity. The second bipolar dimension describes constructive vs. unconstructive action tendencies. On the third dimension, also bipolar, cognitive and motivational emotion features are opposed to affective emotion features. The fourth dimension, labeled distress symptoms, mainly reflects negative emotions. In Study 2, most of the introduced frequent irregularities on IoT devices were not noticed, and the intensity of emotional reactions and tendencies to react in a constructive way decreased throughout the phases of the experiment. These findings reveal that we are not emotionally equipped to identify potential threats in the cyber world.
- Open Access
- 12 December 2024
- Published by Center for Open Science in OSF Preprints
Making interview transcripts open: Preliminary results from a scoping review
Making research data open is recognised as improving both the quality of individual research outputs, and the effectiveness of the perpetuated ‘self-correcting’ goals of research overall. Many funders now mandate open data as a condition of their grants. This can be difficult for qualitative researchers, whose data poses different obstacles to the quantitative data that these policies seem to be aimed at. This pre-print represents the preliminary results of a scoping review into the practical steps that qualitative researchers can take towards opening their transcripts. From each included article, we extracted specific actions that a researcher could take to overcome an identified barrier. From these, we constructed eight themes: confidentiality, consent, misappropriation of data, context, copyright, IRB approval, researcher distress and time and money. We have presented the preliminary findings in the form of a table based ‘tool’, organised via these themes, which we used to develop our own data management plan.
- 12 December 2024
- Published by Center for Open Science in OSF Preprints
Open AccessMaking interview transcripts open: Preliminary results from a scoping review
Making research data open is recognised as improving both the quality of individual research outputs, and the effectiveness of the perpetuated ‘self-correcting’ goals of research overall. Many funders now mandate open data as a condition of their grants. This can be difficult for qualitative researchers, whose data poses different obstacles to the quantitative data that these policies seem to be aimed at. This pre-print represents the preliminary results of a scoping review into the practical steps that qualitative researchers can take towards opening their transcripts. From each included article, we extracted specific actions that a researcher could take to overcome an identified barrier. From these, we constructed eight themes: confidentiality, consent, misappropriation of data, context, copyright, IRB approval, researcher distress and time and money. We have presented the preliminary findings in the form of a table based ‘tool’, organised via these themes, which we used to develop our own data management plan.
- Open Access
- 19 September 2024
- Published by Elsevier in Computers & Security
- Vol. 148
Doing cybersecurity at home: A human-centred approach for mitigating attacks in AI-enabled home devices
- 19 September 2024
- Published by Elsevier in Computers & Security
- Vol. 148
Open AccessDoing cybersecurity at home: A human-centred approach for mitigating attacks in AI-enabled home devices
- Top Cited PapersOpen Access
- 26 June 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (6) , e0305354
A real-world test of artificial intelligence infiltration of a university examinations system: A “Turing Test” case study
The recent rise in artificial intelligence systems, such as ChatGPT, poses a fundamental problem for the educational sector. In universities and schools, many forms of assessment, such as coursework, are completed without invigilation. Therefore, students could hand in work as their own which is in fact completed by AI. Since the COVID pandemic, the sector has additionally accelerated its reliance on unsupervised ‘take home exams’. If students cheat using AI and this is undetected, the integrity of the way in which students are assessed is threatened. We report a rigorous, blind study in which we injected 100% AI written submissions into the examinations system in five undergraduate modules, across all years of study, for a BSc degree in Psychology at a reputable UK university. We found that 94% of our AI submissions were undetected. The grades awarded to our AI submissions were on average half a grade boundary higher than that achieved by real students. Across modules there was an 83.4% chance that the AI submissions on a module would outperform a random selection of the same number of real student submissions.
- 26 June 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (6) , e0305354
Top Cited PapersOpen AccessA real-world test of artificial intelligence infiltration of a university examinations system: A “Turing Test” case study
The recent rise in artificial intelligence systems, such as ChatGPT, poses a fundamental problem for the educational sector. In universities and schools, many forms of assessment, such as coursework, are completed without invigilation. Therefore, students could hand in work as their own which is in fact completed by AI. Since the COVID pandemic, the sector has additionally accelerated its reliance on unsupervised ‘take home exams’. If students cheat using AI and this is undetected, the integrity of the way in which students are assessed is threatened. We report a rigorous, blind study in which we injected 100% AI written submissions into the examinations system in five undergraduate modules, across all years of study, for a BSc degree in Psychology at a reputable UK university. We found that 94% of our AI submissions were undetected. The grades awarded to our AI submissions were on average half a grade boundary higher than that achieved by real students. Across modules there was an 83.4% chance that the AI submissions on a module would outperform a random selection of the same number of real student submissions.
- Open Access
- 14 October 2023
- Published by Center for Open Science in PsyArXiv
A real-world test of artificial intelligence infiltration of a university examinations system: a “Turing Test” case study
The recent rise in artificial intelligence chatbots, such as ChatGPT, poses a fundamental problem for the educational sector. In universities and schools, many forms of assessment, such as coursework, are traditionally completed without supervision. Therefore, students could hand in work as their own which is in fact completed by AI. With the COVID pandemic, the sector has additionally accelerated its reliance on unsupervised ‘take home exams’. If students cheat using AI and this cannot be detected in submissions, this threatens the integrity of the way in which students are assessed. We report a rigorous, blind study in which we injected 100% AI written submissions into the examinations system in five undergraduate modules, across all years of study, for a BSc degree in Psychology at a reputable UK university. In this naturalistic setting, we found that 94% of AI submissions were undetected. The grades awarded to AI submissions were on average half a grade boundary higher than that achieved by real students and across modules there was an 83.4% chance that the AI submissions on a module would outperform a random selection of the same number of real student submissions. We discuss our findings in terms how the educational sector will have to adapt to a “new normal”, which invariably will have to include AI.
- 14 October 2023
- Published by Center for Open Science in PsyArXiv
Open AccessA real-world test of artificial intelligence infiltration of a university examinations system: a “Turing Test” case study
The recent rise in artificial intelligence chatbots, such as ChatGPT, poses a fundamental problem for the educational sector. In universities and schools, many forms of assessment, such as coursework, are traditionally completed without supervision. Therefore, students could hand in work as their own which is in fact completed by AI. With the COVID pandemic, the sector has additionally accelerated its reliance on unsupervised ‘take home exams’. If students cheat using AI and this cannot be detected in submissions, this threatens the integrity of the way in which students are assessed. We report a rigorous, blind study in which we injected 100% AI written submissions into the examinations system in five undergraduate modules, across all years of study, for a BSc degree in Psychology at a reputable UK university. In this naturalistic setting, we found that 94% of AI submissions were undetected. The grades awarded to AI submissions were on average half a grade boundary higher than that achieved by real students and across modules there was an 83.4% chance that the AI submissions on a module would outperform a random selection of the same number of real student submissions. We discuss our findings in terms how the educational sector will have to adapt to a “new normal”, which invariably will have to include AI.
- Open Access
- 12 October 2022
- Published by Center for Open Science in PsyArXiv
A Bayesian Workflow Approach for Testing the Capability of Signal Detection Theory Models
Signal detection theory (SDT) is an important tool for research in psychology and neuroscience. Estimating participants’ sensitivity (d’) and bias (c) from responses to stimuli is at the heart of a wide range of experimental domains. In the present work, we investigate a conventional Bayesian SDT model and incrementally build upon it to improve the model’s capacity to render precise estimates of sensitivity, bias, hit rate, false alarm rate, and correlation between sensitivity and bias. To test these models, we simulate responses from 100 participants to an oddball task with 25 targets and 75 foils. We simulate a high sensitivity and low bias group (d^'≅2±0.5, c≅0.1±0.5), and a low sensitivity and high bias group (d^'≅1±0.5, c≅0.5±0.5). We test three variations of the model, one with fixed parameters, a second with varying parameters, and a third with multivariate varying parameters with Lewandowski-Kurowicka-Joe (LKJ) correlation priors. The models are assessed progressively via a full Bayesian workflow with following steps: prior predictive checks, precision analysis, convergence checks, posterior assessment via similarity measures, posterior predictive checks, and model comparison. Bayesian workflow results provide rich information about priors, convergence, posterior distributions, and models’ predictive capacity. These outcomes exemplify the capacity of a gradual model design Bayesian approach to examine theoretical assumptions and provide more informative model outputs.
- 12 October 2022
- Published by Center for Open Science in PsyArXiv
Open AccessA Bayesian Workflow Approach for Testing the Capability of Signal Detection Theory Models
Signal detection theory (SDT) is an important tool for research in psychology and neuroscience. Estimating participants’ sensitivity (d’) and bias (c) from responses to stimuli is at the heart of a wide range of experimental domains. In the present work, we investigate a conventional Bayesian SDT model and incrementally build upon it to improve the model’s capacity to render precise estimates of sensitivity, bias, hit rate, false alarm rate, and correlation between sensitivity and bias. To test these models, we simulate responses from 100 participants to an oddball task with 25 targets and 75 foils. We simulate a high sensitivity and low bias group (d^'≅2±0.5, c≅0.1±0.5), and a low sensitivity and high bias group (d^'≅1±0.5, c≅0.5±0.5). We test three variations of the model, one with fixed parameters, a second with varying parameters, and a third with multivariate varying parameters with Lewandowski-Kurowicka-Joe (LKJ) correlation priors. The models are assessed progressively via a full Bayesian workflow with following steps: prior predictive checks, precision analysis, convergence checks, posterior assessment via similarity measures, posterior predictive checks, and model comparison. Bayesian workflow results provide rich information about priors, convergence, posterior distributions, and models’ predictive capacity. These outcomes exemplify the capacity of a gradual model design Bayesian approach to examine theoretical assumptions and provide more informative model outputs.
- Open Access
Bayesian models for event-related potentials time-series and electrode correlations estimation
Recent developments on Bayesian inference and prediction for effects on event-related potentials (ERPs) stem from a wide variety of methods, including machine learning, multilevel models, and others. However, few of these approaches make use of clear estimates of the ERP voltage across time and across space (scalp). In the present article, via an iterative process, we propose Gaussian random walk (GRW) models that can estimate voltage uncertainty across time and also provide correlation matrices across electrodes. We apply these models to real ERP data from a P3b paradigm as an example. We discuss results in terms of past and current literature of both ERP estimation and electroencephalography analysis in general.
Open AccessBayesian models for event-related potentials time-series and electrode correlations estimation
Recent developments on Bayesian inference and prediction for effects on event-related potentials (ERPs) stem from a wide variety of methods, including machine learning, multilevel models, and others. However, few of these approaches make use of clear estimates of the ERP voltage across time and across space (scalp). In the present article, via an iterative process, we propose Gaussian random walk (GRW) models that can estimate voltage uncertainty across time and also provide correlation matrices across electrodes. We apply these models to real ERP data from a P3b paradigm as an example. We discuss results in terms of past and current literature of both ERP estimation and electroencephalography analysis in general.
- Open Access
- 29 May 2022
- Published by Elsevier in NeuroImage
- Vol. 258, 119340
Systematic validation of an automated thalamic parcellation technique using anatomical data at 3T.
- 29 May 2022
- Published by Elsevier in NeuroImage
- Vol. 258, 119340
Open AccessSystematic validation of an automated thalamic parcellation technique using anatomical data at 3T.
- Open Access
- 12 May 2021
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 23 (5) , e24879
Emotional Reactions to Cybersecurity Breach Situations: Scenario-Based Survey Study
Background: With the ever-expanding interconnectedness of the internet and especially with the recent development of the Internet of Things, people are increasingly at risk for cybersecurity breaches that can have far-reaching consequences for their personal and professional lives, with psychological and mental health ramifications.Objective: We aimed to identify the dimensional structure of emotion processes triggered by one of the most emblematic scenarios of cybersecurity breach, the hacking of one’s smart security camera, and explore which personality characteristics systematically relate to these emotion dimensions.Methods: A total of 902 participants from the United Kingdom and the Netherlands reported their emotion processes triggered by a cybersecurity breach scenario. Moreover, they reported on their Big Five personality traits, as well as on key indicators for resilient, overcontrolling (internalizing problems), and undercontrolling (aggression) personality types.Results: Principal component analyses revealed a clear 3-dimensional structure of emotion processes: emotional intensity, proactive versus fight/flight reactions, and affective versus cognitive/motivational reactions. Regression analyses revealed that more internalizing problems (β=.33, P<.001), resilience (β=.22, P<.001), and agreeableness (β=.12, P<.001) and less emotional stability (β=–.25, P<.001) have significant predictive value for higher emotional intensity. More internalizing problems (β=.26, P<.001), aggression (β=.25, P<.001), and extraversion (β=.07, P=.01) and less resilience (β=–.19, P<.001), agreeableness (β=–.34, P<.001), consciousness (β=–.19, P<.001), and openness (β=–.22, P<.001) have significant predictive value for comparatively more fight/flight than proactive reactions. Less internalizing problems (β=–.32, P<.001) and more emotional stability (β=.14, P<.001) and aggression (β=.13, P<.001) have significant predictive value for a comparatively higher salience for cognitive/motivational than affective reactions.Conclusions: To adequately describe the emotion processes triggered by a cybersecurity breach, two more dimensions are needed over and above the general negative affectivity dimension. This multidimensional structure is further supported by the differential relationships of the emotion dimensions with personality characteristics. The discovered emotion structure could be used for consistent predictions about who is at risk to develop long-term mental well-being issues due to a cybersecurity breach experience.
- 12 May 2021
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 23 (5) , e24879
Open AccessEmotional Reactions to Cybersecurity Breach Situations: Scenario-Based Survey Study
Background: With the ever-expanding interconnectedness of the internet and especially with the recent development of the Internet of Things, people are increasingly at risk for cybersecurity breaches that can have far-reaching consequences for their personal and professional lives, with psychological and mental health ramifications.Objective: We aimed to identify the dimensional structure of emotion processes triggered by one of the most emblematic scenarios of cybersecurity breach, the hacking of one’s smart security camera, and explore which personality characteristics systematically relate to these emotion dimensions.Methods: A total of 902 participants from the United Kingdom and the Netherlands reported their emotion processes triggered by a cybersecurity breach scenario. Moreover, they reported on their Big Five personality traits, as well as on key indicators for resilient, overcontrolling (internalizing problems), and undercontrolling (aggression) personality types.Results: Principal component analyses revealed a clear 3-dimensional structure of emotion processes: emotional intensity, proactive versus fight/flight reactions, and affective versus cognitive/motivational reactions. Regression analyses revealed that more internalizing problems (β=.33, P<.001), resilience (β=.22, P<.001), and agreeableness (β=.12, P<.001) and less emotional stability (β=–.25, P<.001) have significant predictive value for higher emotional intensity. More internalizing problems (β=.26, P<.001), aggression (β=.25, P<.001), and extraversion (β=.07, P=.01) and less resilience (β=–.19, P<.001), agreeableness (β=–.34, P<.001), consciousness (β=–.19, P<.001), and openness (β=–.22, P<.001) have significant predictive value for comparatively more fight/flight than proactive reactions. Less internalizing problems (β=–.32, P<.001) and more emotional stability (β=.14, P<.001) and aggression (β=.13, P<.001) have significant predictive value for a comparatively higher salience for cognitive/motivational than affective reactions.Conclusions: To adequately describe the emotion processes triggered by a cybersecurity breach, two more dimensions are needed over and above the general negative affectivity dimension. This multidimensional structure is further supported by the differential relationships of the emotion dimensions with personality characteristics. The discovered emotion structure could be used for consistent predictions about who is at risk to develop long-term mental well-being issues due to a cybersecurity breach experience.