When there is information overload on the visual modality, another system of warnings must be adopted to prevent potential risks—tactile warning systems present a viable alternative. Building on work on design approaches for auditory warning systems that match appropriate warnings to the severity of risk, this thesis presents an approach to design tactile warnings based on perceived urgency. To do this, I use a subjective rating technique. I performed three experiments to demonstrate this approach. Our research approach uses subjective rating technique to evaluate perceived urgency. Three experiments were conducted to design tactile warnings with a tactile interface developed by attaching a grid of tactors on a vest. In Experiment 1 and 2, I evaluated perceived urgency of several warning designs with three important parameters of tactile warnings with subjective rating. In Experiment 3 I examined one warning design in the context of flight simulation. The results of Experiment 1 and 2 showed that participants can discriminate between all levels of perceived urgency from most warning parameters. In Experiment 3, the results showed that selected warning design was correctly mapped with the severity of most events. The findings suggest that tactile warnings based on perceived urgency can be a possible approach, but further studies will be required to evaluate different parameters of tactile warnings.

When there is information overload on the visual modality, another system of warnings must be adopted to prevent potential risks—tactile warning systems present a viable alternative. Building on work on design approaches for auditory warning systems that match appropriate warnings to the severity of risk, this thesis presents an approach to design tactile warnings based on perceived urgency. To do this, I use a subjective rating technique. I performed three experiments to demonstrate this approach. Our research approach uses subjective rating technique to evaluate perceived urgency. Three experiments were conducted to design tactile warnings with a tactile interface developed by attaching a grid of tactors on a vest. In Experiment 1 and 2, I evaluated perceived urgency of several warning designs with three important parameters of tactile warnings with subjective rating. In Experiment 3 I examined one warning design in the context of flight simulation. The results of Experiment 1 and 2 showed that participants can discriminate between all levels of perceived urgency from most warning parameters. In Experiment 3, the results showed that selected warning design was correctly mapped with the severity of most events. The findings suggest that tactile warnings based on perceived urgency can be a possible approach, but further studies will be required to evaluate different parameters of tactile warnings.

Phishing is a form of social engineering, in which attackers attempt to trick victims with e-mails designed to look like legitimate requests (Vishwanath et al., 2011), aiming “to exploit human error or human behaviour with the objective of gaining access to information or services” (European Union Agency for Cybersecurity [ENISA], 2023, p. 7). According to the ENISA Threat Landscape 2023 report, phishing is the most prevalent form of social engineering and predicted to continue posing a significant threat to users (ENISA, 2023). In this, attackers have also already succeeded in circumventing second-factor authentication (2FA) (ENISA, 2023). This thesis aimed to contribute to the ongoing research concerning the mitigation of social engineering attacks by investigating phishing susceptibility and a UX-based mitigation approach in the context of 2FA, which to the best of our knowledge had not been previously researched. Based on prior research in other contexts, stress, attention, elaboration, involvement and 2FA frequency were identified as potentially relevant factors. Under consideration of these factors, a 2FA implementation was designed, combining automated URL verification with verification-basedwarnings. An online study (N = 94) was conducted to investigate how the posited susceptibility factors as well as the UX design can influence users’ decisions to abort or proceed with logging in to a suspicious website. In this, a between-subject study design was used to investigate howmuch of an impact specifically the “opinionatedness” of a warning design, i.e. its “use of visual design cues to promote a recommended course of action” (Felt et al., 2015, p. 2893), can have in the 2FA context compared to an otherwise identical design offering a neutral choice. In the collected sample, involvement had a significant negative effect on the likelihood to proceed with the login, in accordance with its posited influence. In addition, confidence in the decision made was discovered as another potential predictor, also showing a significant negativeeffect on the likelihood to proceed in the collected sample. The observed effect of the opinionated design can be seen to contradict the posited assumption that users would be more likely to go with the promoted action. However, overall, the results can be seen to suggest thattaking susceptibility factors into consideration when designing 2FA implementations might be a promising approach towards phishing mitigation. Further research is needed to validate these indications, due to the insufficient sample size and use of convenience sampling in this thesis.

Neste trabalho, é desenvolvido um estudo sobre o uso de alertas auditivos e hápticos dentro do ambiente veicular. É proposto um sistema de assistência ao motorista para alertá-lo de que uma colisão lateral com outros veículos pode ser iminente, quando a observação do espelho retrovisor é insuficiente ou o motorista está desatento. O sistema é composto por um módulo de ativação e um módulo de alerta. O módulo de ativação detecta a condição de perigo por sensores posicionados ao redor do veículo e na tendência de abandono de faixa. Esse trabalho enfoca o módulo de alerta, para o qual se propõe o uso de som tridimensional juntamente com estímulo de vibração. Um ambiente de simulação é proposto para se reproduzir em laboratório o ambiente em que o motorista atua. Nesse ambiente, o sujeito deve utilizar um jogo de corrida projetado na parede à sua frente usando um volante e pedais para acelerar e frear. O motorista se mantém ocupado na tarefa de conduzir o veículo pela trajetória do jogo enquanto é submetido a situações em que um outro veículo se aproxima pela região lateral-traseira velozmente. O assento utilizado é equipado com os elementos geradores dos estímulos vibratórios e auditivos. No primeiro experimento, estuda-se o comportamento de 12 participantes na situação crítica sem o uso do sistema de alerta e conclui-se que a reação preferencial em tal situação é o esterçamento do volante. No segundo experimento, 23 participantes são submetidos ao mesmo cenário adicionando-se o sistema de alertas auditivo e háptico. São coletados os dados de tempo de reação, esterçamento do volante, pressionamento dos pedais e distância entre os carros. Também são registradas as percepções dos participantes sobre o sistema de alerta utilizado. A análise dos dados indica que o sistema de alerta é elegível para uso e pode auxiliar em termos de tempo de reação, devendo-se, entretanto, revisar as características dos alertas utilizados, especialmente o auditivo, para se reduzir o desconforto e evitar prejuízo significativo no controle do veículo. / In this work, we develop a study about the usage of auditive and haptic alerts in the vehicular environment. It is proposed a driver assistance system to warn the driver that a lateral collision with other vehicle may be imminent, when the rear mirror observation is insufficient or the driver is distracted. The system is composed by an activation module and a warning module. The activation module detects the dangerous condition using sensors placed around the vehicle and on the identification of a lane departure tendency. This work focuses on the warning module, which proposes the usage of three-dimensional sound together with vibration stimulus. A simulation environment is proposed to reproduce in laboratory the environment where the driver actuates. In this environment, the subject is requested to play a driving game projected in the frontal wall using a steering wheel and pedals to accelerate and brake. While the driver conducts the vehicle over the game route, he is submitted to the situation where another vehicle approaches very fastly from behind. The seat is equipped with the vibration and auditory stimuli generator elements. In the first experiment, it is assessed the behavior of 12 subjects in the critical situation without the usage of the warning system and it is concluded that the major reaction in such situation is the steering of the wheel. In the second experiment, 23 subjects are submitted to the same scenario adding the auditory and haptic warning system. The following data are collected: reaction time, wheel steering, pedal pressing and distance between the vehicles. The perception of the participants about the used warning system is registered. The analysis of all data indicates that the system is eligible to be used and is able to help regarding reaction times. Nevertheless, it is still necessary to review the characteristics of the used warnings, specially the auditory, to reduce discomfort and avoid loss of control during driving.

Neste trabalho, é desenvolvido um estudo sobre o uso de alertas auditivos e hápticos dentro do ambiente veicular. É proposto um sistema de assistência ao motorista para alertá-lo de que uma colisão lateral com outros veículos pode ser iminente, quando a observação do espelho retrovisor é insuficiente ou o motorista está desatento. O sistema é composto por um módulo de ativação e um módulo de alerta. O módulo de ativação detecta a condição de perigo por sensores posicionados ao redor do veículo e na tendência de abandono de faixa. Esse trabalho enfoca o módulo de alerta, para o qual se propõe o uso de som tridimensional juntamente com estímulo de vibração. Um ambiente de simulação é proposto para se reproduzir em laboratório o ambiente em que o motorista atua. Nesse ambiente, o sujeito deve utilizar um jogo de corrida projetado na parede à sua frente usando um volante e pedais para acelerar e frear. O motorista se mantém ocupado na tarefa de conduzir o veículo pela trajetória do jogo enquanto é submetido a situações em que um outro veículo se aproxima pela região lateral-traseira velozmente. O assento utilizado é equipado com os elementos geradores dos estímulos vibratórios e auditivos. No primeiro experimento, estuda-se o comportamento de 12 participantes na situação crítica sem o uso do sistema de alerta e conclui-se que a reação preferencial em tal situação é o esterçamento do volante. No segundo experimento, 23 participantes são submetidos ao mesmo cenário adicionando-se o sistema de alertas auditivo e háptico. São coletados os dados de tempo de reação, esterçamento do volante, pressionamento dos pedais e distância entre os carros. Também são registradas as percepções dos participantes sobre o sistema de alerta utilizado. A análise dos dados indica que o sistema de alerta é elegível para uso e pode auxiliar em termos de tempo de reação, devendo-se, entretanto, revisar as características dos alertas utilizados, especialmente o auditivo, para se reduzir o desconforto e evitar prejuízo significativo no controle do veículo. / In this work, we develop a study about the usage of auditive and haptic alerts in the vehicular environment. It is proposed a driver assistance system to warn the driver that a lateral collision with other vehicle may be imminent, when the rear mirror observation is insufficient or the driver is distracted. The system is composed by an activation module and a warning module. The activation module detects the dangerous condition using sensors placed around the vehicle and on the identification of a lane departure tendency. This work focuses on the warning module, which proposes the usage of three-dimensional sound together with vibration stimulus. A simulation environment is proposed to reproduce in laboratory the environment where the driver actuates. In this environment, the subject is requested to play a driving game projected in the frontal wall using a steering wheel and pedals to accelerate and brake. While the driver conducts the vehicle over the game route, he is submitted to the situation where another vehicle approaches very fastly from behind. The seat is equipped with the vibration and auditory stimuli generator elements. In the first experiment, it is assessed the behavior of 12 subjects in the critical situation without the usage of the warning system and it is concluded that the major reaction in such situation is the steering of the wheel. In the second experiment, 23 subjects are submitted to the same scenario adding the auditory and haptic warning system. The following data are collected: reaction time, wheel steering, pedal pressing and distance between the vehicles. The perception of the participants about the used warning system is registered. The analysis of all data indicates that the system is eligible to be used and is able to help regarding reaction times. Nevertheless, it is still necessary to review the characteristics of the used warnings, specially the auditory, to reduce discomfort and avoid loss of control during driving.

Future connected vehicles will be able to warn about hidden dangers already before they are visible for the driver. With sight obstructions as one of the most common factors of accident causation, there is a huge potential to improve traffic safety. However, it is unclear how to design the human-machine-interface of such systems to effectively warn drivers about invisible dangers. Especially the expectation that such warnings will be comparably unreliable lead to conflicting demands on amount and coding of warning information. Earlier work shows that warnings that contain more specific information about a hazard can improve drivers understanding of and responses to warnings but they can also raise processing costs and delay reactions or even distract drivers. Psychological theories as well as related research indicates that augmented reality (AR) has the potential to improve warning effectiveness through optimized coding of additional information. AR warnings can inherently transmit the location of a hazard and – due to the corresponding approach of the referenced display towards the driver – could increases the salience of a warning. The general aim of this work is to understand the human factors of future communication-based collision warnings. Based on a theoretical analysis revealing the most relevant questions within that context, we conducted three driving simulator studies to understand the impact of AR warning design on the effectiveness of unreliable warnings about sight obstructed dangers. To consider not only short-term effectiveness, all studies contained several necessary as well as unnecessary warnings that were analyzed in detail. The first study with 88 participants investigated the benefit of prototypical AR warnings over unspecific warnings of different modalities (visual vs. auditory). Visual AR warnings showed advantages over the other warning designs in gaze and brake reaction times, passing speeds, collision rates and subjective evaluation. Auditory AR warnings did not reveal comparable effects. The second test with 80 participants examined the contribution of different design aspects of visual AR warnings. Adding specific warning symbols or scaling animations to the warnings showed some positive but rather inconsistent effects. In contrast, spatial referencing even of an unspecific warning symbol with AR consistently improved driver’s reactions and evaluations. A third experiment with 36 participants observed the differential effects of the spatial information per se and the coding of the information with AR. The warnings had either no spatial information, symbolically encrypted spatial information or AR encrypted spatial information. A higher amount of information consistently led to stronger brake reactions, higher trust and better subjective evaluation. Additionally, with AR encryption we observed faster fixations as well as brake reactions. The present research emphasizes the importance of specificity for warnings about hidden hazards and the potential of AR especially for in-vehicle warnings of future collision avoidance systems. The systematic analysis of psychological factors of warning design and the corresponding findings on their relative contribution to driver’s behavior might also be transferred to other domains and applications of warning and information design. / In naher Zukunft werden vernetzte Fahrzeuge bereits vor sichtverdeckten Gefahren warnen können, noch bevor diese für den Fahrer sichtbar sind. Da Sichtverdeckungen bei einem Großteil schwerer Verkehrsunfälle eine Rolle spielen, stellt dies ein großes Potenzial zur Erhöhung der Verkehrssicherheit dar. Unklar ist jedoch, wie die Mensch-Maschine- Schnittstelle solcher Systeme gestaltet werden sollte, um Autofahrer möglichst effektiv vor noch nicht sichtbaren Gefahren zu warnen. Insbesondere die Vorhersage, dass solche Systeme nur eine begrenzte Zuverlässigkeit haben werden, führt zu teilweise widersprüchlichen Anforderungen an Informationsmenge und Kodierung der Warnungen. Frühere Arbeiten haben gezeigt, dass Warnungen mit spezifischen Informationen über eine Gefahr einerseits Verständnis und Reaktionen der Fahrer auf die Warnungen verbessern, andererseits aber auch kognitiven Verarbeitungsaufwand und Reaktionszeiten erhöhen und ablenken können. Sowohl kognitionspsychologische Theorien als auch Studien aus unserem Forschungsgebiet deuten darauf hin, dass die Darstellungsprinzipien der erweiterten Realität (AR, für engl. augmented reality) das Potenzial bieten, die Effizienz solcher Warnungen durch eine optimierte Kodierung von Zusatzinformationen zu steigern. AR-Warnungen können inhärent die Position einer Gefahr übermitteln, ohne dass der Fahrer dazu eine abstrakte Repräsentation der Information auf die reale Umwelt übertragen muss. Das grundlegende Ziel der vorliegenden Arbeit besteht darin, die psychologischen Faktoren zukünftiger vernetzter Kollisionswarnungen zu verstehen. Ausgehend von der theoretischen Analyse relevanter psychologischer Theorien wurden wesentliche Implikationen und offene Fragestellungen abgeleitet. Zur Beantwortung dieser Fragen wurden drei Fahrsimulator- Studien durchgeführt, in denen der Einfluss von AR als Darstellungsprinzip auf die Effizienz begrenzt zuverlässiger Warnungen über sichtverdeckte Gefahren untersucht wurden. Um valide Aussagen über die längerfristige Wirksamkeit treffen zu können, wurden in den Versuchen sowohl notwendige als auch unnötige Warnungen betrachtet. Eine erste Studie mit 88 Teilnehmern untersuchte den Mehrwert prototypischer AR-Warnungen unterschiedlicher Modalität (visuell vs. auditiv) gegenüber unspezifischen Warnungen. Visuelle AR-Warnungen zeigten klare Vorteile bezüglich Blick- und Bremsreaktionen, Geschwindigkeiten, Kollisionszahlen und subjektiven Bewertungen. Auditive AR-Warnungen hingegen führten zu einzelnen positiven, jedoch auch einigen negativen Effekten. In der zweiten Studie mit 80 Teilnehmern wurden die Auswirkungen gestalterischer Teilaspekte visueller AR-Warnungen verglichen. Das Hinzufügen von spezifischen Warnsymbolen über Bewegungsrichtung und Typ der Gefahr oder einer Vergrößerungs-Animation mit gleicher zeitlicher Veränderung wie bei der AR-Warnung führte zu einzelnen positiven aber nicht konsistenten Effekten. Im Gegensatz dazu führten räumlich verortete AR-Warnungen erneut sowohl zu schnelleren und stärkeren Fahrerreaktionen als auch zu besseren subjektiven Bewertungen. In der dritten Studie mit 36 Teilnehmern wurden schließlich die individuellen Effekte der räumlichen Information an sich und der Codierung dieser Information mittels AR analysiert. Dazu wurden Warnungen ohne Information über die Position der Gefahr, mit symbolisch kodierter Information, sowie mit AR-kodierter Information verglichen. Der höhere Informationsgehalt führte zu durchgehend stärkeren Bremsungen, höherem Systemvertrauen und besseren subjektiven Bewertungen. Darüber hinaus ermöglichte die AR-Kodierung desselben Informationsgehalts der Warnungen sowohl schnellere Gefahrenentdeckung als auch kürzere Bremsreaktionszeiten. Insgesamt bestätigen die Ergebnisse der vorliegenden Arbeit einen deutlichen Mehrwert von spezifischen Informationen bei Warnungen vor sichtverdeckten Gefahren sowie das hohe Potenzial von AR als Darstellungsprinzip, insbesondere für Warnungen zukünftiger Kollisionsvermeidungssysteme. Die systematische Analyse der bei der Gestaltung von Warnungen relevanten psychologischen Faktoren sowie unsere empirischen Erkenntnisse zu deren relativen Einfluss auf das Nutzerverhalten können zudem auf Warnungen anderer Anwendungen und Domänen übertragen werden, und somit einen generellen Beitrag zur Vermeidung von Unfällen in Mensch-Maschine-Systemen liefern.

Future connected vehicles will be able to warn about hidden dangers already before they are visible for the driver. With sight obstructions as one of the most common factors of accident causation, there is a huge potential to improve traffic safety. However, it is unclear how to design the human-machine-interface of such systems to effectively warn drivers about invisible dangers. Especially the expectation that such warnings will be comparably unreliable lead to conflicting demands on amount and coding of warning information. Earlier work shows that warnings that contain more specific information about a hazard can improve drivers understanding of and responses to warnings but they can also raise processing costs and delay reactions or even distract drivers. Psychological theories as well as related research indicates that augmented reality (AR) has the potential to improve warning effectiveness through optimized coding of additional information. AR warnings can inherently transmit the location of a hazard and – due to the corresponding approach of the referenced display towards the driver – could increases the salience of a warning. The general aim of this work is to understand the human factors of future communication-based collision warnings. Based on a theoretical analysis revealing the most relevant questions within that context, we conducted three driving simulator studies to understand the impact of AR warning design on the effectiveness of unreliable warnings about sight obstructed dangers. To consider not only short-term effectiveness, all studies contained several necessary as well as unnecessary warnings that were analyzed in detail. The first study with 88 participants investigated the benefit of prototypical AR warnings over unspecific warnings of different modalities (visual vs. auditory). Visual AR warnings showed advantages over the other warning designs in gaze and brake reaction times, passing speeds, collision rates and subjective evaluation. Auditory AR warnings did not reveal comparable effects. The second test with 80 participants examined the contribution of different design aspects of visual AR warnings. Adding specific warning symbols or scaling animations to the warnings showed some positive but rather inconsistent effects. In contrast, spatial referencing even of an unspecific warning symbol with AR consistently improved driver’s reactions and evaluations. A third experiment with 36 participants observed the differential effects of the spatial information per se and the coding of the information with AR. The warnings had either no spatial information, symbolically encrypted spatial information or AR encrypted spatial information. A higher amount of information consistently led to stronger brake reactions, higher trust and better subjective evaluation. Additionally, with AR encryption we observed faster fixations as well as brake reactions. The present research emphasizes the importance of specificity for warnings about hidden hazards and the potential of AR especially for in-vehicle warnings of future collision avoidance systems. The systematic analysis of psychological factors of warning design and the corresponding findings on their relative contribution to driver’s behavior might also be transferred to other domains and applications of warning and information design. / In naher Zukunft werden vernetzte Fahrzeuge bereits vor sichtverdeckten Gefahren warnen können, noch bevor diese für den Fahrer sichtbar sind. Da Sichtverdeckungen bei einem Großteil schwerer Verkehrsunfälle eine Rolle spielen, stellt dies ein großes Potenzial zur Erhöhung der Verkehrssicherheit dar. Unklar ist jedoch, wie die Mensch-Maschine- Schnittstelle solcher Systeme gestaltet werden sollte, um Autofahrer möglichst effektiv vor noch nicht sichtbaren Gefahren zu warnen. Insbesondere die Vorhersage, dass solche Systeme nur eine begrenzte Zuverlässigkeit haben werden, führt zu teilweise widersprüchlichen Anforderungen an Informationsmenge und Kodierung der Warnungen. Frühere Arbeiten haben gezeigt, dass Warnungen mit spezifischen Informationen über eine Gefahr einerseits Verständnis und Reaktionen der Fahrer auf die Warnungen verbessern, andererseits aber auch kognitiven Verarbeitungsaufwand und Reaktionszeiten erhöhen und ablenken können. Sowohl kognitionspsychologische Theorien als auch Studien aus unserem Forschungsgebiet deuten darauf hin, dass die Darstellungsprinzipien der erweiterten Realität (AR, für engl. augmented reality) das Potenzial bieten, die Effizienz solcher Warnungen durch eine optimierte Kodierung von Zusatzinformationen zu steigern. AR-Warnungen können inhärent die Position einer Gefahr übermitteln, ohne dass der Fahrer dazu eine abstrakte Repräsentation der Information auf die reale Umwelt übertragen muss. Das grundlegende Ziel der vorliegenden Arbeit besteht darin, die psychologischen Faktoren zukünftiger vernetzter Kollisionswarnungen zu verstehen. Ausgehend von der theoretischen Analyse relevanter psychologischer Theorien wurden wesentliche Implikationen und offene Fragestellungen abgeleitet. Zur Beantwortung dieser Fragen wurden drei Fahrsimulator- Studien durchgeführt, in denen der Einfluss von AR als Darstellungsprinzip auf die Effizienz begrenzt zuverlässiger Warnungen über sichtverdeckte Gefahren untersucht wurden. Um valide Aussagen über die längerfristige Wirksamkeit treffen zu können, wurden in den Versuchen sowohl notwendige als auch unnötige Warnungen betrachtet. Eine erste Studie mit 88 Teilnehmern untersuchte den Mehrwert prototypischer AR-Warnungen unterschiedlicher Modalität (visuell vs. auditiv) gegenüber unspezifischen Warnungen. Visuelle AR-Warnungen zeigten klare Vorteile bezüglich Blick- und Bremsreaktionen, Geschwindigkeiten, Kollisionszahlen und subjektiven Bewertungen. Auditive AR-Warnungen hingegen führten zu einzelnen positiven, jedoch auch einigen negativen Effekten. In der zweiten Studie mit 80 Teilnehmern wurden die Auswirkungen gestalterischer Teilaspekte visueller AR-Warnungen verglichen. Das Hinzufügen von spezifischen Warnsymbolen über Bewegungsrichtung und Typ der Gefahr oder einer Vergrößerungs-Animation mit gleicher zeitlicher Veränderung wie bei der AR-Warnung führte zu einzelnen positiven aber nicht konsistenten Effekten. Im Gegensatz dazu führten räumlich verortete AR-Warnungen erneut sowohl zu schnelleren und stärkeren Fahrerreaktionen als auch zu besseren subjektiven Bewertungen. In der dritten Studie mit 36 Teilnehmern wurden schließlich die individuellen Effekte der räumlichen Information an sich und der Codierung dieser Information mittels AR analysiert. Dazu wurden Warnungen ohne Information über die Position der Gefahr, mit symbolisch kodierter Information, sowie mit AR-kodierter Information verglichen. Der höhere Informationsgehalt führte zu durchgehend stärkeren Bremsungen, höherem Systemvertrauen und besseren subjektiven Bewertungen. Darüber hinaus ermöglichte die AR-Kodierung desselben Informationsgehalts der Warnungen sowohl schnellere Gefahrenentdeckung als auch kürzere Bremsreaktionszeiten. Insgesamt bestätigen die Ergebnisse der vorliegenden Arbeit einen deutlichen Mehrwert von spezifischen Informationen bei Warnungen vor sichtverdeckten Gefahren sowie das hohe Potenzial von AR als Darstellungsprinzip, insbesondere für Warnungen zukünftiger Kollisionsvermeidungssysteme. Die systematische Analyse der bei der Gestaltung von Warnungen relevanten psychologischen Faktoren sowie unsere empirischen Erkenntnisse zu deren relativen Einfluss auf das Nutzerverhalten können zudem auf Warnungen anderer Anwendungen und Domänen übertragen werden, und somit einen generellen Beitrag zur Vermeidung von Unfällen in Mensch-Maschine-Systemen liefern.