DISCUSIÓN DE RESULTADOS

Summarizing the results of experiments 3 and 4, driving performance was signifi- cantly effected by list length and searching period and the interaction between both factors, as well as by the interaction between velocity and searching period. Direction of curvature and curve radius had a significant effect on driving performance in ex- periment 4 only. Furthermore, statistical analysis revealed significant interactions be- tween curve radius and searching period and between velocity and curve radius in experiment 4 and between direction of curvature and searching period in experiment 3. Error rate was significantly effected by curve radius in experiment 3. Significant effects on misses, however, did not occur in any of the experiments. More generally, task completion times were significantly effected by list length in both experiments, as well as by velocity in experiment 3. Ratings of self-estimated effort and secondary task interference revealed significant effects of velocity and list length in both experi- ments. As no other variables were altered, the differences between experiment 3 and 4 are mainly due to the decrease in reaching distance of the secondary task controls. In the following sections each of the effects of driving speed, list length, direction of curvature, curve radius and searching period on lane deviation, error rate, misses and task completion times, as well as on subjective experiences shall be discussed.

The effect of driving speed. Driving performance was relatively unaffected by speed variations in both experiments. This was also true for task completion times where speed had only a significant effect in experiment 4 on task completion times of lists with four entries. The conclusion is that an increase of overall workload caused by

changes in velocity during these experiments was not high enough to require an ex- tension of task completion times. Nevertheless, fast trials were in both experiments rated to be significantly more effortful and the lists as more distracting than in the slow trials, reflecting a rise in subjectively assessed mental workload. Furthermore, there were significant differences in subjective ratings between experiments 3 and 4 concerning self-estimated effort, secondary task interference and attention allocation: While ratings of secondary task interference were significantly higher in both speed conditions in experiment 3, self-estimated effort was higher for the slow condition in experiment 4, indicating that task demands dropped under an optimal level. In addi- tion, speed became a relevant factor when interacting with other factors, such as searching period or curve radius. Interestingly, there was also a significant effect of control location on velocity concerning error rates, revealing that participants made more errors during the slow condition of experiment 4 while error rates were equal between both experimental settings during the fast condition. This could either be due to different reactions of the controls or would be another indication for the hy- pothesis that performance decreases under very low task demands as well, due to a suboptimal level of arousal.

The effect of list length. Driving performance was best for two entries and lowest for eight entries in both of the experiments. But, as displayed in figure 4.4, both for rather low and rather high task demands, increased control proximity had an increasing ef- fect on driving performance.

0,0 0,5 1,0 1,5 2,0 2,5 3,0 2 4 8 entries Mea n D e v iati on (p ix el ) exp3 exp4

Figure 4.4 Effect of list length on lane deviation (dual task minus single task values) – ex- periment 4

This finding corresponds with those by Kantowitz et al. (1984) where performance was worse and workload was higher for a four-choice task than for a two-choice task in the context of simulated flight.

Task completion times were slightly longer in experiment 3 for all of the list lengths. Furthermore, the more entries in a list, the more effortful the trial was rated and the more interference was attributed to the lists. But neither of the differences in experi- ments 3 and 4 concerning effects of list length on primary and secondary task per- formance or subjective ratings became significant.

The effect of direction of curvature. The type of road segment also influenced driv- ing performance significantly. But while in experiment 3 driving performance de- creased rather linear across straight sections, left turns and right turns respectively, increased control proximity in experiment 4 apparently had a clear facilitating effect on straight sections as well as a slight enhancing effect on right turns, although the effect did not become significant (see figure 4.5).

0,0 0,5 1,0 1,5 2,0 2,5

straight left turn right turn

Mean D e v iat ion (p ix el ) exp3 exp4

Figure 4.5 Effects of direction of curvature on lane deviation (dual task minus single task values) – experiments 3 and 4

The effect of curve radius. The effect of curve radius on driving performance was significant in experiment 4 only. Here again, only during the low demanding condition of moderate curves the change in control location did have a facilitating effect on driving performance, which did not differ significantly from experiment 3 though. The number of extra keystrokes was relatively unaffected by variations of curve radius throughout both of the experiments, which might be explained by an overall decrease in task demands.

The effect of searching period. Again, the current status of the search had the greatest effect on driving performance in both of the experiments. Mean lane devia- tion was smallest during the first 20% of the search, higher between 40% and 60% of the search and worst during the last 20% of the search. But differences between the two experimental settings were only marginal with a slight increase during the last fifth of the search, not reaching level of significance, though. For the first fifth of the search, performance was also very close to (or even better than) performance in the single task situation. As it can generally be assumed that human performance is most reliable under a moderate workload level (Kantowitz & Casper, 1988), one ex- planation for this result might be that task demands in the single task situation al- ready dropped below an optimal level, thus resulting in a situation where participants were in a state of mental underload.

Conclusions. Although participants were instructed to not compromise the driving task while performing a display-related secondary task they did so nevertheless, causing variations in driving performance measures. Adding a secondary task signifi- cantly degraded driving performance in both experiments. List length as well as cur- vature and searching period had still significant decreasing effects on driving per- formance. Thus, increasing the proximity of secondary task controls, enabling drivers to leave their hands on the steering-wheel, could obviously not eliminate the per- formance impairing effects of list length, curvature and searching period. Contrary, experimental results revealed only a small advantage of locating the controls on the steering-wheel, which might partly be explained by the fact that the joystick in ex- periment 3 was rather big and therefore easy to find without having to look at it. Fur- thermore, a distance of 15cm is still rather low compared to some distances of sec- ondary task controls drivers will experience in the car. There were also no other in- struments located around the control to become confused with as it would possibly be the case in an in-vehicle environment.

As a consequence of these findings, stimuli presentation of primary and secondary task were converged on one single monitor in the next experiment to see whether this would lead to fewer decrements in driving performance. It was supposed that participants need less time to switch attention forth and back between primary and secondary task when stimuli are very close to each other, an assumption which should especially apply for the last fifth of the search, where participants have to concentrate more on the secondary task in order to identify the target item.