This project developed a stand-alone training module intended to accelerate the process of perceptual expertise for young, novice drivers. The approach is based on learning technologies that have proven successful in developing expertise in other fields of application. The module was designed to require no special equipment (other than internet access and a monitor of sufficient size and resolution so that images are clear) and be able to be self-administered through a client-server architecture. Expert driving, as in other domains of expertise, is based on a process of rapid and unconscious recognition and response under time pressure. Higher order visual perceptual skills are the essence of this expertise. Experienced drivers are able to quickly recognize emerging situations or potential hidden threats without effortful cognition and thus avoid putting themselves in a serious conflict situation. Novice teen drivers require an extended period to acquire this skill. Current driver training approaches successfully impart procedural knowledge but not the fluency associated with expertise. However, there are advanced learning technologies in the fields of perceptual learning and adaptive learning that are specifically designed to accelerate this process of expertise. Successful applications have been seen in diverse domains, such as medicine/surgery, aviation, and mathematics. These methods have produced rapid and enduring advances with improved fluency, automaticity, and insight. This project developed a training module to deploy advanced learning technologies to accelerate teen driver perceptual learning. A perceptual/adaptive learning module (PALM) was constructed using a client-server architecture, accessible on a computer through a standard web browser. The fully functional system was exercised and piloted for usability with a sample of six teen drivers holding provisional licenses. Formal evaluation of the training benefits of the module was not included within the scope of this project. The module incorporated six distinct learning categories: (1) anticipating that another vehicle on the road may move into your path; (2) anticipating that a vehicle directly ahead may slow severely or stop; (3) anticipating that someone or something off the road might move into the road; (4) recognizing that something about the road ahead may force you to change your planned path; (5) recognizing where something significant may be obscured from view; and (6) recognizing the presence of an emergency vehicle in the area that might cause conflicts. Ten distinct scenarios were developed as exemplars within each of these learning categories. A driving simulator was used to generate each of the 60 scenarios, as video clips of short drives (25-90 seconds in duration). Each video clip terminated with a “crash” resulting from a lack of driver reaction to the emerging threat situation. These video clips provided the basis for the PALM procedure. Three classes of training trials occurred in the course of training: “Watch” trials, “Respond” trials, and “No Event” trials. During “Watch” trials, the user watches the video, which stops just before the hazard fully manifests itself. A set of six multiple choice options (corresponding to the six learning categories) appears in place of the video and the user must indicate the type of hazard that they should first be able to anticipate. In “Respond” trials, the user presses the space bar during the video when they recognize a potential hazard. They then see the same set of choice options as for the “Watch” trials. During “No Event” trials, the “Respond” procedure is in effect but no hazard event occurs during the video. The correct behavior is to not press the space bar. Feedback is given for incorrect responses for all trial types. The user progresses through the module based on response speed and accuracy, with the spacing/sequencing algorithm indicating that a given learning category should be presented on a given trial. Problems from each category are repeated as needed until mastery criteria for that category are met. The trainee receives a Progress Report on speed and accuracy after each block of 12 trials and a final screen once all categories are mastered. The usability of the PALM was assessed through testing with six 18-19 year old participants who all held Maryland provisional driving licenses. Each participant was scheduled for up to three one-hour-long sessions, held at Westat laboratories. The training was essentially self-administered by the participant. A Westat researcher observed and took notes, provided assistance if needed, and administered a set of questions at the end of each session. Three participants met the mastery criteria within two sessions and the other three completed early in the third session; average completion time was 99.7 minutes. Testing indicated good usability, user engagement, and subjective sense of benefit. Performance accuracy improved substantially across all learning categories. As desired, the module appeared easy to implement and use, highly engaging, and efficiently accomplished. Participants subjectively felt that their hazard awareness and safe driving skill were improved. Problems or dislikes were few and minor. All participants felt they would recommend the training to others. The participants learned to recognize general conditions of driving threats (distinguishing learning categories) and perceptual skills to anticipate specific emerging, and often subtle, potential hazard events in very diverse scenarios. Thus this pilot exercise suggests that the module has performed as desired and has substantial promise as a practical and efficient means of promoting expertise in perceptual skills for novice drivers. Formal post-training assessment of the PALM will be required to objectively demonstrate meaningful and sustained benefits in driving performance. (Author/publisher)
Abstract