Organizational Unit:

School of Psychology

Permanent Link

https://hdl.handle.net/1853/70756

Parent Organization

Organizational Unit

College of Sciences

Includes Organization(s)

Organizational Unit

Work Science Center

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/1133

Full item page

Publication Search Results

Now showing 1 - 10 of 24

Directability Through AI Customization: The Effect of Choice on Trust and Acceptance in Highly Automated Vehicles

(Georgia Institute of Technology, 2023-12-05) Scott-Sharoni, Sidney Tammie

People feel apprehensive about using or relying on highly automated vehicles (American Automotive Association, 2019). One method of assuaging fears involves providing explanations for the system’s behaviors using a Human-Machine Interface (HMI). However, understanding the amount of information for optimal human-automation interaction can prove difficult due to differences in individuals’ preferences, experiences, and needs. An underexplored method that may account for these discrepancies involves providing users with choices or customization. The Coactive Design Approach suggests that including directability, or the power to influence a system’s actions, may improve how users interact with systems (Johnson et al., 2014). The following study investigated how customization affordances and modified vehicle aspect of a Level 4 automated vehicle affected trust and acceptance. One hundred twenty participants experienced one highly automated simulator drive, during which they engaged in a visually demanding game. A MANOVA assessed the interaction of and main effects of customization availability and modified vehicle aspect on trust and acceptance. While participants who customized had higher average trust and acceptance in the automated vehicle than participants who did not customize, only the main effect of vehicle aspect significantly impacted the multivariate dimension of trust and acceptance in the automated vehicle. That is, modifications to the vehicle impacted users regardless of whether they chose the modification. The game score and subjective trust did significantly correlate to a small, positive extent, indicating that higher trust in a system may improve non-driving related task performance. Future research should continue to investigate the role of choice in the interaction between individuals and highly automated systems to understand the psychological impacts of directability.
Perceived Relational Risk and Perceived Situational Risk: Scale Development

(Georgia Institute of Technology, 2020-11-05) Stuck, Rachel E.

Interactions with technology are a significant part of daily life, both at home and at work. Understanding how to support successful human-technology interaction is essential for Engineering Psychology. Perceived relational and situational risk are key components to understanding interactions with technologies including adoption, trust, and use. However, perceived risk was only recently separated into these two distinct types: relational and situational. In addition, prior measures of perceived risk focus on hazards, not interactions with technology or automation. The goal of this dissertation was to develop and validate scales of perceived relational risk and perceived situational risk. These scales built on previous work exploring perceived risk and incorporated scale items related to affect, probability, severity, and domains. Evaluations of internal reliability, construct validity, and test-retest reliability were conducted for both scales. The items for both scales had excellent internal reliability, acceptable test-retest reliability, and support for construct validity. After determining the validity of the items, items were selected to create the final scales. These scales allow future researchers to rigorously and accurately study how perceived relational risk and perceived situational risk affect with trust, each other, and technology use.
Development and validation of the situational trust scale for automated driving (STS-AD)

(Georgia Institute of Technology, 2020-05-26) Holthausen, Brittany Elise

Trust in automation is currently operationalized with general measures that are either self-report or behavioral in nature. However, a recent review of the literature suggests that there should be a more specific approach to trust in automation as different types of trust are influenced by different factors (Hoff & Bashir, 2015). This work is the development and validation of a measure of situational trust for the automated driving context: The Situational Trust Scale – Automated Driving (STS-AD). The first validation study showed that situational trust is a separable construct from general trust in automation and that it can capture a range of responses as seen in the difference between scores after watching a near automation failure video and non-failure videos. The second study aimed to test the STS-AD in a mid-fidelity driving simulator. Participants drove two routes: low automation (automated lane keeping only) high automation (adaptive cruise control with automated lane keeping). The results of the second study provided further support for situational trust as a distinct construct, provided insight into the factorial structure of the scale, and pointed towards a distinction between advanced driver assistance systems (ADAS) and automated driving systems (ADS). The STS-AD will revolutionize the way that trust in automation is conceptualized and operationalized. This measure opens the door to a more nuanced approach to trust in automation measurement that will inform not only how drivers interact with automated systems; but, can impact how we understand human-automation interaction as a whole.
Measuring the influence of automation on situation awareness in highly automated vehicles

(Georgia Institute of Technology, 2020-04-28) Becerra, Zoe Marie

Higher levels of automation, such as adaptive cruise control (ACC) and automated lane keeping (ALK), are becoming more and more common in vehicles. With the inclusion of these automated features, the role of the driver is shifting from an active, operator role to a passive, supervisory role. As drivers enter this transition, it is critical they understand how the automation is performing and remain aware of the roadway environment. Situation awareness (SA) is the understanding of what is going on around you. Previous research has shown how a driver’s SA is impacted by many factors including: age, driving experience, distraction, and secondary task engagement. Little work has explored the direct influence of level of automation on SA or how best to measure SA in an automated vehicle. To address these issues, this study examined how SA changes as a function of level of automation in the driving domain using three measures of SA. Participants completed two twenty-minute simulated drives with two levels of automation: low automation (ALK only); and high automation (ALK and ACC). The order of the drives were counterbalanced. Throughout the drives, the Situation Present Assessment Method (SPAM) and secondary task engagement were used to measure SA. SPAM is a query-based measure in which questions about the situation are periodically presented; the situation remains present and the participant continues to perform the task. Secondary task engagement was measured by the total time voluntarily spent playing a game of Tetris, a visuospatial task. After each drive, participants completed the SART questionnaire to subjectively measure their perceived SA. Additionally, the NASA-TLX and a Trust in Automation Scale were administered after each drive to measure subjective workload and trust. Results showed between the three administered measures of SA, query-based measures (SPAM) and subjective measures (SART) were more sensitive compared to performance measures (secondary task engagement). Further, there was evidence to suggest a combination of query-based and subjective measures is best to assess SA in the automated driving context. Concerning the impact of automation level on SA, high automation systems supported higher SA compared to low automation systems. The results also indicated the patterns of SA were different in the low and high automation drives. There were no significant changes in the pattern of SA during the low automation drive. However, the results suggested a quadratic trend best described the pattern of SA in the high automation drive. These insights will provide guidance to develop better standardized measures of SA for future research. In addition, these findings can inform the design of interventions to support driver SA, especially in low automated vehicles.
Impact of action-object congruency on the integration of auditory and visual stimuli in extended reality

(Georgia Institute of Technology, 2020-03-24) May, Keenan Russell

Extended Reality (XR) systems are currently of interest to both academic and commercial communities. XR systems may involve interacting with many objects in three-dimensional space. The usability of such systems could be improved by playing sounds that are perceptually integrated with visual representations of objects. In the multisensory integration process, humans take into account various types of crossmodal congruency to determine whether auditory and visual stimuli should be bound into unified percepts. In XR environments, spatial and temporal congruency may be unreliable. As such, the present research expands on associative congruency, which refers to content congruency effects that are acquired via perceptual learning in response to exposure to co-occurrent stimuli or features. A new type of associative congruency is proposed called action-object congruency. Research in ecological sound perception has identified a number of features of objects and actions that humans can discern based on the sounds produced by sound-producing events. Since humans can infer such information through sound, this information should also inform the integration of auditory and visual stimuli. When perceiving a realistic depiction of a sound-producing event such as a strike, scrape or rub, integration should be more likely to occur if a concurrently-presented sound is congruent with the objects and action that are seen. These effects should occur even if the visual objects and the sound are novel and unrecognizable, as long as relevant features can be ascertained visually and via sound. To evaluate this, the temporal and spatial ventriloquism illusions were utilized to assess the impact of action congruency and object congruency on multisensory integration. Visual depictions of interacting objects were displayed in virtual reality, and congruent or incongruent sounds were played over speakers. In two types of trials, participants either localized the sounds via pointing, or judged whether the sounds and visual events were simultaneous. Action-object congruent visual and auditory pairings led to greater localization biasing and higher rates of perceived simultaneity, reflecting stronger integration of stimuli. Action and object congruency were both impactful, but action congruency had a larger effect. The effects of action and object congruency were additive, providing support for the linear summation model of congruency type combination. These results suggest that action-object congruency can be used to better understand how humans conduct multisensory integration as well as to improve MSI in future XR environments.
Measuring the effects of display design and individual differences on the utilization of multi-stream sonifications

(Georgia Institute of Technology, 2019-07-30) Schuett, Jonathan Henry

Previous work in the auditory display community has discussed the impact of both display design and individual listener differences on how successfully listeners can use a sonification. This dissertation extends past findings and explores the effects of display and individual differences on listeners’ ability to utilize a sonification for an analytical listening task when multiple variables are presented simultaneously. This is considered a more complicated task and pushes listeners’ perceptual abilities, but is necessary when wanting to use sonifications to display more detailed information about a dataset. The study used a two by two between- subjects approach to measure the effects of display design and domain mapping. Acoustic parameters were assigned to either the weather or the health domain, and these mappings were either created by an expert sound designer or arbitrarily assigned. The acoustic parameters were originally selected for the weather domain, so those display conditions were expected to result in higher listener accuracy. Results showed that the expert mapped weather sonification led to higher mean listener accuracy than the arbitrarily mapped health display when listeners did not have time to practice, however with less than an hour of practice the significant main effects of design and domain mapping went away and mean accuracy scores increased to a similar level. This dissertation introduces two models for predicting listener accuracy scores, the first model uses musical sophistication and self-reported motivation scores to predict listener accuracy on the task before practice. The second model uses musical sophistication, self-reported motivation, and listening discrimination scores to predict listener accuracy on the sonification task after practice.
Understanding the misunderstanding: Why confidence intervals are poorly understood and evaluating proposed solutions across sensory modalities

(Georgia Institute of Technology, 2019-05-23) Batterman, Jared M.

The error bar representation of a confidence interval is the most ubiquitous display of uncertainty in statistical analysis. However, despite this, error bars are poorly understood even by seasoned scientists and researchers across disciplines. The root of this misunderstanding is not certain, but researchers have posited several hypotheses ranging from the structure of the display itself, to how it is presented in the classroom. Studies have thus far been either incomplete or inconclusive, leading some to call for the elimination of the use of error bars entirely. However, research into statistics education (suggesting the error bar representation may not even be taught in contemporary classrooms) demonstrates that a lack of exposure in an educational likely contributes to the displays’ poor level of discernment. Furthermore, promising research into auditory uncertainty displays suggests that sound may be a better medium to present this information than vision. The studies conducted in this dissertation demonstrate that not only are auditory displays viable alternatives to visual representations of confidence intervals (i.e., error bars), but when used together can improve overall performance. Furthermore, the presence of visual displays was shown to enhance performance on questions involving confidence intervals, but not on other topics, suggesting that confidence intervals may hold a unique status when it comes how they are affected by alternative display types. Limited training was also shown to improve performance, suggesting that if confidence interval displays were taught in classrooms, overall performance could improve and lead students to understand this topic just as well as other statistical concepts. The success of the auditory representation could also lead to the creation of cheaper, more universally designed, materials to help convey uncertainty to all students, regardless of sensory ability.
Investigating age-related differences in spatial presence formation and maintenance in virtual reality

(Georgia Institute of Technology, 2019-04-02) McGlynn, Sean A.

Virtual reality has numerous applications with the potential to support physical, cognitive, and socio-emotional well-being across a range of users. The effectiveness of these applications in achieving desirable outcomes (e.g., transfer of training, enjoyment, treatment efficacy) has been shown to depend on the extent that the user experiences a sense of being physically located in the virtual environment. This 'sense of being' is termed spatial presence. Research on this concept has primarily focused on the effect that the objective immersiveness of the system (e.g., screen resolution, field of view, audio quality) has on the level of spatial presence that users experience in the virtual environment. The goal of this dissertation was to better understand the components of the full spatial presence process (i.e., formation and maintenance), validate measurement methods for capturing within-experience changes in spatial presence formation and maintenance, changes in spatial presence levels over time, and the cognitive abilities that influence spatial presence formation and maintenance. 25 younger and 25 older adults participated in virtual reality experiences over the course of three days. Age was used as a proxy for changes in cognitive abilities. Additionally, measures of specific attentional abilities were administered as well as existing and novel measures of spatial presence during and after the virtual reality. The primary findings of this dissertation are as follows: 1) In general, there was little evidence of age-related or time-related differences in spatial presence, 2) Presence formation occurred rapidly, 3) Participants experienced high levels of spatial presence, 4) Participants maintained spatial presence in the virtual environment for the majority of their sessions, 5) Disturbances in presence were easily recovered from, 6) Methods of measuring within-experience fluctuations in presence were validated, with some methodological caveats. These findings are informative to spatial presence theory, future research, and measurement and have practical contributions for designers of virtual reality applications, experiences, and systems.
Understanding automation handoff impacts on workload and trust when mitigated by reliability displays

(Georgia Institute of Technology, 2018-05-10) Noah, Brittany Elise

Current commercial vehicles are beginning to include automated features such as adaptive cruise control and automated lane keeping. This is a first step towards full vehicle automation which is predicted to be possible within the next five years. As automated features are integrated into vehicles, the driver must know how to properly interact with and trust these systems. A key element of drivers interacting and relying on these systems is the handover of control between the vehicle and driver. This handover, occurring during times of automation error, will be a critical point of high workload for drivers when driving a partially or fully automated vehicle. If the driver is aware of the system’s performance and can appropriately calibrate his or her trust, then these instances of handover may become less stressful and easier to complete successfully. This study explored the driving performance, trust, visual scanning behaviors, perceived workload, and objective workload for handover scenarios. There were four between-subjects display conditions: (1) no display; and reliability displays using (2) quantitative information (percentage of reliability); (3) qualitative information (direct representation of a number); and (4) representational information (abstract representation of a number). Participants completed two drives. The first drive aided in familiarization with the automated lane keeping system. In the second drive, the handover drive, participants experienced an automation failure resulting in transition of control from automated to manual. Results from this study showed that there was a difference in subjective experience between the baseline and handover drive due to experiencing an automation failure. Participants in the no display condition were more affected by the automation failure, greatly decreasing their overall trust in the automated lane keeping system. Participants with reliability displays were able to appropriately calibrate their trust to system performance and were less impacted by the automation failure, experiencing a slight, statistically insignificant, decrease in trust. These findings will impact the implementation and design of automation reliability displays and shows that drivers with reliability displays are less impacted by automation failure than those without reliability displays.
The effect of experience on the use of multimodal displays in a multitasking interaction

(Georgia Institute of Technology, 2018-01-03) Gable, Thomas Matthew

Theories and previous work suggest that performance while multitasking can benefit from the use of displays that employ multiple modalities. Studies often show benefits of these multimodal displays but not to the extent that theories of multimodal task-sharing might suggest. However, it is often the case that the studies investigating this effect give users at least one type of display that they are not accustomed to, often an auditory display, and compare their performance on these novel displays to a visual display, with which most people are familiar. This leaves a question open regarding the effects of longer-term experience with these multimodal displays. The current study investigated the effect of practice with multimodal displays, comparing two multimodal displays to a standard visuals-only display. Over the course of four sessions, participants practiced a list-searching secondary task on one of three display types (two auditory plus visual displays, and one visual-only display) while performing a visual-manual task. Measures of search-task and primary task performance along with workload, visual behaviors, and perceived performance were collected. Results of the study support previous work with regard to more visual time on the primary task for those using multimodal displays, and show that perceived helpfulness increased over time for those using the multimodal displays. However, the results also point to practice effects taking place almost equally across the conditions, which suggest that initial task-sharing behaviors seen with well-designed multimodal displays may not benefit as much from practice as hypothesized, or may require additional time to take hold. The results of the research are discussed regarding their use in research and applying multimodal displays in the real world as well as in how these results fit with theories of multimodal task-sharing.