Simulating effects of arousal on lane keeping : Are drowsiness and cognitive load opposite ends of a single spectrum ?

Both drowsiness and cognitive load have been demonstrated to significantly affect driving performance. By drowsiness, we here refer to a reduced level of alertness, where alertness is assumed to be governed by circadian cycle sleep homeostasis processes (Borbely and Achermann, 1999). By cognitive load, we refer to the need for cognitive, or executive, control to perform driving and/or non-driving related tasks such as cell phone conversation (Engström et al., in press). One important negative effect of drowsiness on driving can be broadly construed as a less responsive brain reacting more slowly to hazardous stimuli (Lin et al., 2010). This is manifested, for example, as increased lane keeping variability and an increased frequency of relatively large steering corrections (Liu et al., 2009). By contrast, a large number of studies (see reviews in He et al., 2014, and Engström et al., in press) have found that cognitive load, somewhat counterintuitively, reduces lane-keeping variability which is typically accompanied by an increase in small steering corrections (Markkula and Engström, 2006). Several explanations for this improvement effect of cognitive load on lane keeping have been offered in the literature, but none appears to be fully in line with the available evidence (Engström et al., in press). Here we suggest, based on a conceptual model outlined by Engström et al. (in press), that the observed performance effects of drowsiness and cognitive load on lane keeping and steering may be due to a single shared mechanism, that is, neural responsiveness, modulated by cortical arousal, determining the drivers’ sensitivity to lane keeping error. This mechanism is implemented in a computational model and it is investigated, by means of simulation, whether the model can reproduce the empirically observed effects.