Level I: Non-interactive, Computer Graphic and/or Digital Video Visuals, No Motion
This level of simulation is exemplified by desktop systems with a computer monitor showing computer graphics or high-resolution video of a road scene, using standard controls (wheel, pedals), or sometimes only a mouse or touchscreen to allow the subject to respond to what is displayed on the monitor. The performance measures that can be obtained using such low-cost methods are strictly “part-task”─i.e., one stage or component of driver information processing assumed to underlie safe vehicle control is isolated for study, for example perceptual vigilance as a function of hours without sleep, or hazard detection latency while performing a secondary (divided attention) task.
As one example of such research, a vigilance study conducted by Mills et al. (2001) used a computerized task to study the effects of stimulants (dextroamphetamine at 10 mg), sedatives (a short half-life benzodiazepine, alprazolam at 0.5 mg), and fatigue, on single- and divided-attention responses in different parts of the visual field. Subjects included 18 healthy volunteers age 19 to 37 (no older people were tested). For all subjects, blood samples were taken at predose and at 12 postdose intervals (ranging from 0.25 hours to 12 hours) to determine alprazolam or dextroamphetamine plasma concentrations. Maximum concentrations of alprazolam occurred between 0.75 and 4 hours from dosing, with an average 1.78 hours. Maximum concentrations of dextroamphetamine concentration occurred between 1.5 and 4 hours, with an average of 2.78 hours.
The driving-related tasks in this research included a perceptual task in the center of the computer display, combined with identification of a critical (octagon shape) target presented at varying positions and eccentricities at the periphery of the display. Test stimuli were presented as described below using a 17-inch monitor and a keyboard spacebar and arrow keys were used to record subjects’ responses.
For the central task, subjects were presented with a 7.3 cm by 5.6 cm box in the center of the monitor divided by a vertical double line that simulated a pavement centerline marking. Subjects saw a combination of “headlights” (two white lights) and “taillights” (two red lights) on each trial. One stimulus configuration—white lights on the left side of the line (representing an oncoming vehicle in the adjacent lane) and red lights on the right side of the line (representing a leading vehicle)—was designated as “correct.” Subjects were instructed to only respond to the central task, by pressing the space bar on the computer’s keyboard, if the display was “correct.”
The peripheral task randomly presented 12 shapes for each trial, one of which could be an octagon. The stimulus display duration varied from 1 to 3 seconds. Each critical trial presented the octagon at each of the 12 positions (three levels of eccentricity on four radials), for single- and multiple-response displays.
Behavioral measures included the speed and accuracy of responses to the peripheral targets (at each of three angles of eccentricity) without the divided-attention requirement; the speed and accuracy of responses for the central task and the peripheral target identifications (at each of the three visual angles); and a composite score that calculated as a single linear combination of all measures. The composite score was the measure most sensitive to peak drug effects.
Results of the study showed that spatial “tunneling” was produced with both sedatives and stimulants, however tunneling was unique to each drug. With sedatives, tunneling was shown by decrements in both the single-task and divided-task scores, as the displays became more distant from the center. Stimulant-induced tunneling was characterized by improvements in only the divided-attention task displays near the center of the screen, with little or no changes at the outer edges. The tunneling brought on by depressants coincided with rising and peak blood levels.
The demonstrated impairment from a low dose of the sedative was similar to that observed in a study conducted by Mills, Parkman, and Spruill (1996), with significant deficits in visual scanning and divided attention at peak blood concentrations, which increased as stimuli were presented at increasing eccentricities. The computerized assessment in this research was thus sensitive to deficits caused by the medications that are important to the task of driving—the ability to spot potential hazards in the periphery and to divide attention.
One additional example of simulation methods at this level is provided by the work of Ball, Owsley, and colleagues at the Edward R. Roybal Center for Research in Applied Gerontology at the University of Alabama at Birmingham (UAB). The Roybal Center’s simulator includes a car cab (front seat only) mounted on a fixed platform, with three screens (affording a roughly 120° field of view) displaying projection videos of different driving scenarios filmed (from a driver’s eye perspective) to capture situations and maneuvers of particular interest to researchers (e.g., intersection negotiation). This approach (video) produces a high degree of realism in the driving scene stimuli, but limits the interactivity of the system to slight adjustments in the perceived travel speed of the driver’s vehicle in response to pressure on the gas and brake pedals; this is why the UAB system is classified as ‘Level 1’ for the present discussion.
The UAB researchers have used this simulator since 1998, primarily in studies of aging and visual attention/visual information processing involving the “useful field of view” construct. Many hundreds of older subjects have participated in this research program; results have been widely reported, in peer-reviewed journals. Without discussing the results of any specific study, this review draws attention to the UAB researchers’ experience with simulator sickness among older subjects. In their initial applications of this driving simulator, between one-quarter and one-third of older subjects became ill or ‘queasy’ to the extent that data collection could not proceed. However, after limiting subjects’ exposure to only those driving scenarios involving straight-ahead movement—no horizontal or vertical curves—the rate of simulator sickness was reduced to under 5 percent.44