Currently, many new in-vehicle information and communication systems are entering our cars. They range from media players to navigation systems to speech based systems which allow the use of the vehicles functions by speech. While these systems are intended to facilitate the driving task, some research shows they also may also have the potential to distract drivers (e.g. Santos et al., 2005). In order to evaluate to which degree such systems are suitable for the while driving, many evaluation methods have been proposed. While most of these evaluation methods allow the measurement of visual-manual distraction (e.g. eye tracking), cognitive distraction is much more complicated to measure. However, as more systems become multi-modal or purely
speech-based, it is getting even more important to measure the impact of cognitive distractions. Recent studies (Merat & Jamson, 2008) explored a promising method to evaluate cognitive workload: detection response tasks (DRTs). Such detection response tasks rely on at least a dual task setting, where the impairment in a secondary task (the detection response task) is an indication of the workload imposed by the primary task. In this study three types of DRTs were evaluated: peripheral detection response task (PDRT), auditory detection response task (ADRT) and tactile detection response task (TDRT). In order to evaluate the sensitivity of each of these DRTs, cognitive tasks like the n-back task (Mehler et al., 2009) and a counting task were deployed in two levels of difficulty. Additionally, these cognitive tasks were presented in a visual, auditory, as well as pure cognitive way to clarify if any interactions between the different modalities of the DRTs and the presentation modes of the cognitive tasks exist. Results revealed significant differences between high and low levels of cognitive workload for all three types of the DRT variants evaluating the reaction time. However, a closer examination of the results showed that the PDRT is not adequately sensitive to measure increased cognitive workload on the counting task if the dependent measure is the hit rate. It is concluded that all three DRT variants are a sensitive measurement technique to assess cognitive workload. More research is needed to validate these findings on the use of real world tasks. Furthermore it has to be proven if it is possible to apply one of the DRT variants to a tertiary design (driving + test task + DRT), as this would increase the ecological validity of the method.