RESNA 26th International Annual Confence
Reaction time data have traditionally been used to draw conclusions about the cognitive complexities of a given motor task since such processes cannot be directly observed. More complex motor activities have been shown to require more time for selection and preparation of a response and this increase in cognitive processing is reflected in a longer reaction time interval. This study quantifies the importance of including reaction time data in investigations comparing performance of computer input devices. Analysis of reaction time data provides rehabilitation specialists with clinically valuable information beyond what can be ascertained from measures of speed and accuracy alone.
Studies comparing the efficacy of computer input devices assist rehabilitation therapists in making device decisions for individual users. Speed of performance, captured as movement time, is a common measure for evaluating the effectiveness of a computer input device (1). Movement time has been used extensively in the disability research as a means of documenting performance differences between input devices in a variety of tasks. Although movement time data verifies that performance differences exist, researchers are left to speculate as to the cause of these differences.
Reaction time (RT) is defined as the interval of time from the presentation of a stimulus to the initiation of a response. RT is widely used as a method for studying motor performance (2) but has received only limited attention in studies comparing users' performance with computer input controls. Because reaction time is presumed to reflect the mental processes that result in movement, the differentiation of RT from movement time has the potential to provide information about why performance differences exist between devices.
A previous study comparing the performance of a head-operated device and an arm-hand operated device indicated that the head-operated device performed significantly faster in both a target acquisition task and a text-typing task (3). The purpose of this paper was to analyze the reaction time data collected from the target acquisition task for its clinical relevance in explaining why performance differences emerged.
Twenty-four typical young adults served as subjects. Participants included five males and 19 females with an average age of 21 years, 9 months. Twenty-three subjects were right-handed; one was left-handed. The head-operated input device was the Tracker 2000? (4). The hand/arm-operated device was the IntelliKeys? keyboard (5). To complete the experimental task, a cursor keys overlay was designed for the keyboard, which allowed horizontal, vertical and diagonal cursor movements. The operational features for the devices were set identically.
Participants were tested individually over two sessions (part of a larger study). Order of presentation of devices was randomly determined and counterbalanced. Positioning relative to the computer was standardized across subjects. Circular targets representing three widths (0.5 cm, 1 cm and 1.5 cm), three distances (6.2 cm, 3.1 cm and 1.55 cm) and eight radial directions were used (5, 50, 95, 140, 185, 230, 275 and 340 degrees). Subjects completed 6 target acquisition blocks for each device. A block consisted of 72 randomly presented trials representing each target size, target distance and radial combination (3 x 3 x 8) for a total of 432 trials per device.
To begin a trial, the subject moved the cursor arrow tip toward the home target positioned at the center of the screen. When the cursor arrow tip crossed the boundary of the home target, the home target turned yellow, marking the beginning of the fore period, which varied randomly from 0.5 to 2 seconds. The fore period was followed by the `go' signal: an auditory beep coupled with a change in color of the home target to green and the simultaneous appearance of the capture target at a random size, distance and angle. The subject then moved the cursor to the target as rapidly as possible and stopped. To acquire a target, the tip of the cursor arrow had to be maintained inside for 500 milliseconds.
After each trial, the test software automatically recorded pertinent including reaction time. For this study, reaction time was defined as the time elapsed between the appearance of the `go' signal and the initiation of cursor movement. The beginning of cursor movement was defined as the moment at which the cursor moved 10 pixels.
The experiment was analyzed as a mixed factorial design. Results for the reaction time data (RT) are presented in Figure 1 and Table 1. RT was significantly faster for the head device compared to the hand/arm device and was affected by all variables in the model. Significant interactions were stratified by device type to produce tests of simple effects.
Results indicated that RT was 61 ms faster in session two than in session one for the hand/arm device. However, RT was not significantly affected by session when the head device was used. RT was significantly affected by target distance when the head device was used; as distance to the target increased, mean RT significantly decreased. When the hand/arm device was used, RT was significantly affected by target width and target angle; RT increased as the width of the target decreased. With respect to the affect of angle on RT for the hand/arm device, post hoc analysis did not indicate any specific pattern for the results.
The finding that RT was significantly faster for the head device than for the hand/arm device suggests differences in the complexity of motor control for the two devices. The head device may have offered an anatomical advantage over the hand/arm device in that the stimulus and the response were more naturally linked; subjects moved the head in conjunction with the eyes to map the target or letter location. When the hand/arm device was used decisions about where to move were compounded by decisions about how to move since subjects had to consider the cursor arrow combinations needed to acquire a given stimulus; thus requiring greater planning. Additionally, the required translation from the horizontal cursor keys to the vertical computer screen may have lowered stimulus-response compatibility for the hand/arm device resulting in increased RT.
Difficulty of an aiming task has been shown to vary as a function of target distance and target width. For this study, RT was differentially affected by distance and width for the two devices. For example, when the hand/arm device was used RT was significantly affected by target width but not distance. This finding suggests that planning for accuracy was more difficult than planning for distance when the hand/arm device was used. One possible explanation is that it was more difficult to home in on a target using the hand/arm device so more time was required to plan for end-point accuracy.
When the head device was used, RT was significantly affected by target distance but not width. The fact that target width was not a significant factor suggests that it was easier to be precise with the head device and so less time had to be devoted to planning for accuracy. Previous research has demonstrated that as the distance to a target increases, RT increases. However, for the head device, RT decreased as the distance to the target increased. One possible explanation is that control for the head device may operate in a closed-loop system with adjustments being made as movement is carried out, so less prior planning is necessary.
These results highlight the value of collecting RT data in comparative device studies. The RT results suggested that the head device improved task performance by maximizing stimulus-response compatibility and reducing the time needed for response programming.