Strayer et al. in this volume show that increases in cognitive workload caused by drivers' involvement in distracting activities that allow them to keep their eyes on the road lead to decrements in indices of safe driving performance. Although there is agreement that in-vehicle tasks that require drivers to take their eyes off the road increase crash risk, there is mounting controversy about whether in-vehicle tasks that do not require drivers to take their eyes off the forward roadway increase crash risk—thus the conundrum: How can there be an abundance of cognitively distracting activities and controversy about whether such activities increase crash risk?
A quiet revolution is in progress in the field of human factors. This revolution is broadly based, finding a home not only in such traditional areas as workplace layout and industrial ergonomics but also in more recent areas, such as humancomputer interaction and cognitive engineering. The revolution affects the way research questions are posed and the types of answers that are offered. No longer is a good or better solution adequate. Rather, the focus has turned to the best or optimal solution. In this article the research in optimal performance engineering is selectively reviewed. It is argued that increasing the number and range of attempts to engineer optimal performance will lead not only to safer and more efficient designs but also to an increase over the long run in the applicability and accessibility of laboratory research.
Objective: The objective was to identify key cognitive processes that are impaired when drivers divert attention from driving. Background: Driver distraction is increasingly recognized as a significant source of injuries and fatalities on the roadway. Method/Results: A "SPIDER" model is developed that identifies key cognitive processes that are impaired when drivers divert attention from driving. SPIDER is an acronym standing for scanning, predicting, identifying, decision making, and executing a response. Conclusion: When drivers engage in secondary activities unrelated to the task of driving, SPIDER-related processes are impaired, situation awareness is degraded, and the ability to safely operate a motor vehicle may be compromised. Application: The pattern of interference helps to illuminate the sources of driver distraction and may help guide the integration of new technology into the automobile.
Sound localization has been studied extensively. Curiously, although much is known about factors that affect errors, little is known about factors that influence response time. Three experiments were performed in an attempt to identify the separate influence of each of several different factors. All trials used a single broadband noise signal emanating from one of a subset of six loudspeakers equally spaced around the participant in the azimuthal plane. Both the size of the subset and the balance of relative probabilities from speaker to speaker were altered to evaluate the relationship between information content and the dependent variable, choice reaction time. Choice reaction time was found to be related to the information content of the sound stimulus in all cases. It was also found to be related to the presence of pairs of speakers that were symmetrically opposed in front of and behind the participant. Models of choice reaction time in a sound localization task have clear implications for practice. For example, they suggest that multiple auditory collision warnings may endanger drivers.
A new model of the visual search process is developed which can improve the design of large symbol sets such as those used by nuclear power plant personnel, air traffic controllers, and battlefield troops. An experiment was conducted to determine whether the new, componential model or an already existing, discriminability model better explains visual search behavior. The results were consistent with the componential model. We show how to use the componential model to help automate selection of the optimal symbol set (i.e., the symbol set that minimizes the average time to find a target).
Office computer users view well over a billion displays in a given year. The savings of only a fraction of a second in the time it takes users to process each display can potentially lead to enormous time and cost savings. In recent research investigators have shown that on average subjects are quicker to find a target option in a highlighted display than in a display without highlighting. Paradoxically, in related research other investigators have shown that subjects are slower to find a target in a highlighted display than in a display without highlighting. In an attempt to resolve this paradox, an additional set of experiments was performed. Results from these experiments suggest that in order to determine whether highlighting will be of benefit, one must know the type of highlighting, the level of highlighting validity, and the probability that subjects attend first to the highlighted options.
Objective: This study aimed (a) to determine whether older drivers looked less often for potential threats while turning than younger drivers and (b) to compare the effectiveness of active and passive training on older drivers' performance and evaluation of their driving skills in intersections. Background: Age-related declines in vision, physical abilities, psychomotor coordination, and cognition combine to make it less likely that older drivers will look for potential threats during a turn. Research suggests that active training should be an effective means of improving older drivers' performance and self-awareness. Method: In Experiment 1, younger and older participants drove a series of virtual intersection scenarios, were shown video replays, and were provided feedback. In Experiment 2, older drivers were assigned to one of three cohorts: active simulator training, passive classroom training, or no training. Pre- and posttraining simulator and field drives assessed training effectiveness. Results: In Experiment 1, older drivers looked less often during turns than younger drivers. Customized feedback was successful in altering drivers' perception of their abilities. In Experiment 2, active training increased a driver's probability of looking for a threat during a turn by nearly 100% in both posttraining simulator and field drives. Those receiving passive training or no training showed no improvement. Conclusion: Compared with passive training, active training is a more effective strategy for increasing older drivers' likelihood of looking for threats during a turn. Application: The results of this research can guide the development of programs that could reduce intersection crashes among older drivers.
Typically, detailed quantitative and computer models of human operators performing real world tasks cannot easily be developed. We propose a technique that more easily allows for that development. We propose that when a cognitive task analysis has been carried out, a computer simulation model useful for approximations of task completion time is often within reach. The first step is to construct an activity network or order-of-processing diagram from the task analysis. Second, activity durations are found in the literature or approximated through multidimensional scaling. Finally, equations are written for calculating task completion time, or a program is written for simulations to estimate this time. Resulting models can be useful for optimizing system design. The approach is illustrated with an activity network by W. D. Gray, B. E. John, and M. E. Atwood (1993) for a telephone operator task. Simulations demonstrate the feasibility of using multidimensional scaling to obtain approximate activity durations. The approach is also illustrated with an order-of-processing diagram representing drivers reading roadside message displays. We point out that if a more detailed picture of unobservable mental processes in a task is needed, techniques have been developed for this through analysis of response times. Actual or potential applications of this research include system design, human-computer interaction, message comprehension, and simulation of information-processing tasks.
Objective: Evaluation of the effects of a PC-based training program on risk perception in a driving simulator. Background: Novice drivers have a fatality rate some eight times higher than that of the most experienced group of drivers, primarily because of the novice driver's inability to predict ahead of time the risks that will appear in the roadway. Current driver education programs, at least those in the United States, do not emphasize the teaching of risk awareness skills to novice drivers. Method: APC-based risk awareness and perception training program was developed and evaluated. The training involved using plan (top-down) views of 10 risky scenarios that helped novice drivers identify where potential risks were located and what information should be attended. Both the 24 trained novice drivers and 24 untrained novice drivers were evaluated on an advanced driving simulator. The eye movements of both groups of drivers were measured. The evaluation on the driving simulator included both scenarios used in the training and others not used in training. Results: The set of trained novice drivers were almost twice as likely as untrained drivers to fixate appropriately either on the regions where potential risks might appear or on signs that warned of potentially risky situations ahead, both for the scenarios they had encountered in training and for novel scenarios. Application: The PC training program developed, which is portable and can be widely used, has great promise in improving risk perception for novice drivers on the road.
Twenty women were asked to generate forces using a dynamometer that were consistent with one of three different work-rest schedules (a low-, medium-, and high-force schedule). Each work-rest schedule consisted of 6 identical blocks of 10 work-rest cycles. Each of the 10 work-rest cycles lasted 1 min. The first work-rest cycle in each block consisted of a 6--s maximal voluntary contraction and a 54--s rest. The remaining 9 work-rest cycles in each block consisted of a submaximal contraction and a rest period. The desired force of the submaximal contraction, the length of this contraction, and the duration of the rest period remained constant within schedules but varied across schedules. The amount of physiological work was kept constant among schedules. The fatigue that developed in the medium-force schedule was significantly lower than that developed in either the low- or high-force schedule. A model was developed that predicted the amount of fatiguable strength at the beginning and end of each contraction of a work-rest cycle. When fit to the results from the experiment, the model explained 94% of the variance. The model can be used to predict the work-rest schedule that minimizes fatigue in a given repetitive job, thereby potentially increasing productivity and reducing the incidence of cumulative trauma disorders.
Increasingly users must navigate through a hierarchy of menus in order to access the various functions available on a computer. As the number of functions proliferates and the menu hierarchies grow more complex, the time it takes users to access any given function has become unacceptably slow. Specifically, given a set of functions, recent studies have shown how to identify the structure of the complete, homogeneous hierarchy that minimizes the average time to access these functions. In this article a general method for finding the optimal hierarchy is developed. The method does not confine the search for an optimal hierarchy to the set of complete, homogeneous hierarchies; nor, given suitable input, does the method select as optimal a hierarchy that cannot be easily navigated. It is important to identify the optimal hierarchy because it can reduce the average access time considerably, in some cases (at least in theory) by almost 100%.
Objective: The aim of this study was to illustrate how a consideration of glance sequences to in-vehicle tasks and their associated distributions can be informative. Background: The rapid growth in the number of nomadic technologies and in-vehicle devices has the potential to create complex, visually intensive tasks for drivers that may incur long in-vehicle glances. Such glances place drivers at increased risk of a motor vehicle crash. Method: We used eye-glance data from a study of distraction training programs to examine the change in glance duration distributions across consecutive glances during the performance of various in-vehicle tasks. Results: The sequential analysis across trained and untrained drivers showed that the proportion of late-sequence glances longer than a 2-s threshold among untrained drivers was almost double the number of such glances for the trained drivers, that the third and later glances were particularly problematic, and that training reduced the proportion of early- and later-sequence glances. Conclusion: Examining how the duration of off-road glances varies as a function of their order in a sequence of glances and the visual demands of the task can offer important insights into the change in the distracting potential of in-vehicle tasks across glances and the effects of training. Application: The sequential analysis of in-vehicle glance data can be useful for researchers and practitioners and has implications for the development and evaluation of training programs as well as for task and interface design.
Automobile drivers were recently found to be risk averse when choosing among routes that had an average travel time shorter than the certain travel time of a route considered as a reference. Conversely, drivers were found to be risk seeking when choosing among routes that had an average travel time longer than the certain travel time of the reference route. In a driving simulation study in which the reference route had a range of travel times, this pattern was replicated when the reference range was smaller than the ranges of the available routes. However, the pattern was reversed when the reference range was larger than the ranges of the available routes. We recently proposed a simple heuristic model that fit the relatively complex data quite well. Actual or potential applications of this research include the design of variable message signs and of route choice support systems.
In two experiments, participants chose between staying on a main route with a certain travel time and diverting to an alternative route that could take a range of travel times. In the first experiment, travel time information was displayed on a sheet of paper to participants seated at a desk. In the second experiment, the same information was displayed in a virtual environment through which participants drove. Overall, participants were risk-averse when the average travel time along the alternative route was shorter than the certain travel time of the main route but risk-seeking when the average travel time of the alternative route was longer than the certain travel time along the main route. In the second experiment, in which cognitive load was higher, participants simplified their decision-making strategies. A simple probabilistic model describes the risk-taking behavior and the load effects. Actual or potential applications of this research include the development of efficient travel time information systems for drivers.
