Examining Distracted Drivers' Underestimation of Time and Overestimation of Speed
First, it is the case that humans are not all that good at estimating the length of a given interval of time even under the best of circumstances (Zimbardo & Boyd, 1999). Interval length estimation is defined as a subjective evaluation of the duration of a given interval of time. Most research that has focused on time perception either uses: (1) the retrospective paradigm or (2) the prospective paradigm (Sucala, Scheckner & David, 2010). With the retrospective paradigm, study participants are unaware that they will be asked to estimate the length of an interval. With the prospective paradigm, study participants are aware that they will be asked to estimate the duration of an interval while they are exposed to specific experimental stimuli or events. The results from the various studies point to cognitive load as the key factor impacting interval duration estimates for both retrospective and prospective paradigms. In the retrospective paradigm, participants overestimate the temporal interval with increasing task difficulty. Block and Reed (1978) use memory models to explain why this relation exists. Specifically, contextual changes encoded from one interval are used to make inferences about the time that has elapsed during the said interval. Since there are generally more contextual changes during high cognitive load tasks than low cognitive load tasks, the retrospective interval estimations are longer during the high cognitive load tasks. However, the results obtained in the prospective paradigm were just the opposite of those obtained in the retrospective paradigm. Now the participants were found to underestimate the temporal intervals with increasing task difficulty. Researchers explained these results using attentional models. The models assumed that with increasing task difficulty, cognitive resources were being preferentially allocated for task related information processing thereby leaving less room for temporal processing. Of the two paradigms, the prospective paradigm is arguably the one more closely related to the tasks that drivers face when performing in-vehicle tasks. And, disturbingly, it is the one in which the participants underestimate the duration of an interval, especially during high workload tasks. Speed being inversely proportional to time, any underestimation in time translates into an overestimation of speed. Why is this critical? A driver who underestimates time (thinks he/she has only 1.5 seconds when they actually have 2 seconds to complete a maneuver) will necessarily overestimate speed (travel faster relative to if they realized they had 2 seconds available) resulting in maneuvers that can be unsafe (e.g. rear end collisions) With respect to the speed selection process, the drivers' perception of time is known to affect their choice of speed. Example, Consider an urban environment with a four-way, signal-controlled intersection. Consider the yellow phase. A driver can choose to go or stop. The drivers' perception of time (to stop in time for the red light) quantified by their subjective estimation is a predictor of whether the driver will stop or go. As described above, the average drivers' ability to estimate the length of an interval is not good at all thereby biasing his/her decision making in such a situation, what is typically a 'dilemma zone'. Alternatively, consider a driver approaching a midblock crosswalk with a pedestrian approaching from the left side. Will the driver yield? Or will the driver proceed? Again, his/her perception of time affects their decision and this brings us to the first aim in specific. Aim 1 is to understand when the driver is engaged in the primary activity of driving and is also concurrently engaged in a secondary in-vehicle task (different tasks with varying levels of cognitive load), to what extent is his/her ability to perceive time and therefore appropriately select speed compromised (e.g. an extended texting task, a short text messaging task and/or a mock cellphone task)? Do spillover tasks cause drivers to further underestimate the duration of aninterval, thereby leading to riskier maneuvers characterized by poor speed selection, as compared to other similar in-vehicle tasks with mere switching effects? Samuel and Fisher (2015) show that spillover tasks adversely affect reacquisition of situation awareness on the forward roadway when engaged in a task that requires alternation of glances between the inside of the vehicle and the forward roadway. Further, there is what has been referred to as the Zeigarnik effect described as the tendency to feel frustrated if a task is not completed. In the context of driving, taking five short glances to finish a 10 s task will be more frustrating than taking two long glances. Griest-Bousquet and Schiffman (1992) found that participants overestimated the length of the time it took to complete ten tasks in a retrospective paradigm by some considerable amount when they were given twenty tasks, interrupted after completing ten, and asked to estimate the time it took to complete the ten tasks as opposed to being given ten tasks, allowed to complete then tenth task, and then asked to estimate the time it took to complete the ten tasks. Based on the wider set of results from studies of retrospective and prospective estimates of interval length, one would expect the Zeigarnik effect in a prospective task to lead to large underestimations of the duration of a given interval. And this in turn translates to large overestimations in speed choice. Aim 2: And then, the related secondary question is how do secondary in-vehicle tasks that require several short glances performed in conjunction with the primary activity of driving affect a drivers' speed selection as opposed to ones that require few long glances (e.g. Navigation vs Texting).