Ericka Rovira

Understanding reliance on automation: effects of error type, error distribution, age and experience

An obstacle detection task supported by ‘imperfect’ automation was used with the goal of understanding the effects of automation error types and age on automation reliance. Sixty younger and sixty older adults interacted with a multi-task simulation of an agricultural vehicle (i.e. a virtual harvesting combine). The simulator included an obstacle detection task and a fully manual tracking task. A micro-level analysis provided insight into the way reliance patterns change over time. The results indicated that there are distinct patterns of reliance that develop as a function of error type. A prevalence of automation false alarms led participants to under-rely on the automation during alarm states while over-relying on it during non-alarm states. Conversely, a prevalence of automation misses led participants to over-rely on automated alarms and under-rely on the automation during non-alarm states. Older adults adjusted their behaviour according to the characteristics of the automation similar to younger adults, although it took them longer to do so. The results of this study suggest that the relationship between automation reliability and reliance depends on the prevalence of specific errors and on the state of the system. Understanding the effects of automation detection criterion settings on human–automation interaction can help designers of automated systems to make predictions about human behaviour and system performance as a function of the characteristics of the automation.

EFFECTS OF UNRELIABLE AUTOMATION ON DECISION MAKING IN COMMAND AND CONTROL

The effectiveness of automated decision aids used by human operators in command and control systems may depend not only on automation reliability, but also on the type (stage) and level the automated support provides. Automation can be applied to information acquisition, information integration and analysis, decision choice selection, or action implementation (Parasuraman, Sheridan, & Wickens, 2000). The present study examined the effects of variations in the stage of automation support on performance in a “Sensor to Shooter” targeting simulation of command and control. Independent variables included the type and level of automation support (complete listing, priority listing, top choices, and recommendation of decision choice) and the reliability of the automation (60% and 80%). Dependent variables included accuracy and reaction time of target engagement decisions. Compared to manual performance, reliable automation did not affect the accuracy of target engagement decisions but did significantly reduce decision times. When the automation was unreliable, under the higher reliability condition (80%) there was a greater cost in accuracy performance for higher levels of automation aiding (priority listing, top choice, and recommendation) than at a lower level (complete listing). The results support the view that automation unreliability has a greater performance cost for decision automation than for information automation. This performance cost generalizes across a number of different forms of decision-aiding.

Effects of Information and Decision Automation on Multi-Task Performance

Automation purported to assist human operators may itself be an additional source of complexity and uncertainty. Because high reliability cannot always be assured, imperfect automation can add to uncertainty and thereby degrade performance. The present study examined the relative benefits and costs of information and decision automation and investigated the effects of uncertainty resulting from automation unreliability during multiple task performance. Subjects were either provided with status information (“information” automation) or a recommendation for action (“decision” automation) for the system monitoring sub-task of the Multi-Attribute Task Battery (MAT). Two levels of automation reliability were compared. The detrimental effect of unreliable automation—a decrease in the detection rate of malfunctions—was greater for automation of higher reliability, a result consistent with previous findings of automation-related complacency. This effect of automation unreliability was also greater for decision than for information automation.

The Influences of Social Networks on Phishing Vulnerability

Phishing is a form of electronic deception in which an attacker tries to cause the recipient to do something or disclose data that they likely would not normally do by mimicking a trustworthy entity. These attacks have been increasing at an alarming rate and can cause damages in the form of identity theft, financial losses, and compromised security for organizations and governmental institutions. Additionally, phishing attacks have become very sophisticated and even more successful because of the lack of vigilance by computer users. Successful phishes have particularly strong implications for military populations, and have the potential to threaten national security. In an attempt to reduce the overall success rate of a phishing attack, this paper applies the foundations of social network analysis to identify how social network structures among a military company of future US Army officers. are most influential in reducing the spread of a phish. This experimental study collected empirical and survey data in an effort to analyze the flow of information and influence of people in phishing awareness within an organization.

Displaying Contextual Information Reduces the Costs of Imperfect Decision Automation in Rapid Retasking of ISR Assets

Objective: The impact of a decision support tool designed to embed contextual mission factors was investigated. Contextual information may enable operators to infer the appropriateness of data underlying the automation’s algorithm. Background: Research has shown the costs of imperfect automation are more detrimental than perfectly reliable automation when operators are provided with decision support tools. Operators may trust and rely on the automation more appropriately if they understand the automation’s algorithm. The need to develop decision support tools that are understandable to the operator provides the rationale for the current experiment. Method: A total of 17 participants performed a simulated rapid retasking of intelligence, surveillance, and reconnaissance (ISR) assets task with manual, decision automation, or contextual decision automation differing in two levels of task demand: low or high. Automation reliability was set at 80%, resulting in participants experiencing a mixture of reliable and automation failure trials. Dependent variables included ISR coverage and response time of replanning routes. Results: Reliable automation significantly improved ISR coverage when compared with manual performance. Although performance suffered under imperfect automation, contextual decision automation helped to reduce some of the decrements in performance. Conclusion: Contextual information helps overcome the costs of imperfect decision automation. Application: Designers may mitigate some of the performance decrements experienced with imperfect automation by providing operators with interfaces that display contextual information, that is, the state of factors that affect the reliability of the automation’s recommendation.

Does the domain of technology impact user trust? Investigating trust in automation across different consumer-oriented domains in young adults, military, and older adults

ABSTRACT Trust has been shown to be a determinant of automation usage and reliance. Thus, understanding the factors that affect trust in automation has been a focus of much research. Despite the increased appearance of automation in consumer-oriented domains, the majority of research examining human-automation trust has occurred in highly specialised domains (e.g. flight management, military) and with specific user groups. We investigated trust in technology across three different groups (young adults, military, and older adults), four domains (consumer electronics, banking, transportation, and health), two stages of automation (information and decision automation), and two levels of automation reliability (low and high). Our findings suggest that trust varies on an interaction of domain of technology, reliability, stage, and user group.

Empirical Benefits of Training to Phishing Susceptibility

Social engineering continues to be the most worrisome vulnerability to organizational networks, data, and services. The most successful form of social engineering is the practice of phishing. In the last several years, a multitude of phishing variations have been defined including pharming, spear phishing, and whaling. While each has a specific reason for its success, they all rely on a user failing to exercise due diligence and responsibility. In this paper, we report on a recent phishing experiments where the effects of training were evaluated as well as gathering demographic data to explore the susceptibility of given groups.

Adaptive Change in the Type of Automation Support Reduces the Cost of Imperfect Decision Aids in a Simulated Battlefield Engagement Task

Automation that is meant to aid the human operator may actually be detrimental to performance, particularly if faulty decision recommendations are provided (decision automation), as opposed to prioritized or integrated information advisories that are incorrect (information automation). Because automation can be imperfect, operator over-reliance on decision automation can degrade performance. The present study examined whether temporary adaptive changes in the type and level of automation—-between decision and information automation, or between decision automation and manual performance—–could mitigate the cost of automation imperfection in a combat engagement selection task. Twelve participants were provided with two types of automation (decision and information) and also performed the task manually. In three conditions, the type of automation was alternated during performance of the task over three blocks of trials. In all three conditions, decision automation was provided in the first and third blocks of the task, with the middle block requiring the use of decision automation, information automation, or manual performance. The accuracy of engagement decisions improved in the third block with decision automation when it was preceded by a temporary adaptive change to information automation. No such improvement occurred when decision automation was used throughout the task or when the adaptive change involved a temporary return to manual performance. This suggests that providing the user with short periods of information automation can help mitigate some of the costs of imperfect decision automation by keeping the operator in the decision-making loop.

Effects of individual differences in working memory on performance and trust with various degrees of automation

ABSTRACT Previous studies showed performance benefits with correct automation, but performance costs when the automation was incorrect (i.e. provided an incorrect course of action), particularly as degrees of automation increased. Automation researchers have examined individual differences, but have not investigated the relationship between working memory and performance with various degrees of automation that is both correct and incorrect. In the current study, working memory ability interacted with automation reliability and degree of automation. Higher degrees of correct automation helped performance while higher degrees of incorrect automation worsened performance, especially for those with lower working memory. Lower working memory was also associated with more trust in automation. Results illustrate the interaction between degree of automation and individual differences in working memory on performance with automation that is correct and automation that fails.

Trust in Computers and Robots: The Uses and Boundaries of the Analogy to Interpersonal Trust

Trust is a complex concept having many meanings and hinting at many variables, and is not a single concept, or state, or continuum. Panelists will briefly argue their stances concerning concepts of trust in automation, and whether (or to what extent) our understanding of trust in automation should be addressed by analogy to interpersonal trust. There is considerable divergence of opinion on these matters, and on the question of whether it is possible for robots to engage in trustworthy relations with humans.