The visual occlusion technique was originally developed by Senders et al. to assess driver workload and to evaluate the cognitive demands of the roadway. Recently, the technique was adopted as a means to estimate the visual demand of in-vehicle devices (Gelau & Krems, 2004) and has been introduced by the International Standards Organization (ISO) as a standard for assessing visual demand from in-vehicle systems (ISO, 2007). The term visual demand is defined as “the degree of visual activity required to extract information from an object to perform a specific task” (ISO, 2002, p. 4).
The visual occlusion technique stabs to emulate the blinking intervals of visual demand of an IVT in a traffic situation. By blocking (occluding) the visual input from the IVIS tested, either manually or by software, it can represent the time spent looking at the road. The main idea behind the method is to block the vision for an amount of time while performing a task on an IVIS. The blocked vision is intended to imitate the primary task of driving, although this can be criticized by the fact that there is no Visio-spatial processing or any other type of task load present that is typically associated to driving (Monk & Kidd, 2007). Most implementations of the visual occlusion technique that assess driver distraction do not impose additional cognitive demands on participants during the occluded periods.
An advantage of conducting tests using the occlusion technique is the option to examine the IVIS in a static atmosphere. No testing in real traffic situations or simulator environments is needed.
2.1 Measures for Occlusion
To make clear some terms, the following definitions have been used. These have been set by ISO in the draft international standard recommendation for using the occlusion technique (ISO, 2005).The measurements recorded when using the occlusion method are total task time (TTT) for an IVIS task in both unconcluded and occluded conditions. The total shutter open time is calculated (TSOT), total time that vision is not occluded when using an occlusion procedure. A ratio, referred to as the R-value, between the TTT unconcluded and TSOT is calculated and gives an indication of how resumable the task is after interruption (ISO, 2005).
There has been considerable debate within the research community regarding the length of time participants are allowed vision and the duration of occluded periods (Stevens et al., 2004). The ISO standard proposes a shutter closed time of 2 seconds and a shutter open time of 1.5 seconds (ISO, 2005). 1.5 seconds is felt to be a tolerable time for drivers to glance away from the road whilst in motion (Wierwille, 1993), (Weir et al., 2003), (Rockwell, 1988). The ISO standard also recommends that the interaction interface should be revised if the majority of the subjects fail to complete the IVIS task. The ISO standard also states that if the resumability ratio or R-value exceeds 1, it shows that the participants may be having difficulty in resuming the task. It is left for the users of the standard to determine the exact limit for Safety or design alteration (ISO, 2005).
2.2 Occluding Techniques
There are two different ways to determine when vision should be occluded. Either the subject chooses to see the IVIS for a certain time period or the occluded intervals are preset. These are known as voluntary (or user-paced) and involuntary (or system-paced) occlusion respectively. There are several studies testing the optimal times for preset intervals. Most recent studies show that the occluded times most closely representing driving is approximately around 1 or 2 seconds (Stevens, 2004) When applying voluntary occlusion the person testing a system is in control of when the vision and occlusion intervals occur by using a control (Stevens, 2004) . When the person indicates that he/she needs visual input to continue the task a vision interval of preset time is presented or the interval lasts until the task is finished. Voluntary occlusion has not been extensively used. One study used it to distinguish between simple and stylized maps. In the study both voluntary and involuntary occlusion was used and concluded that both techniques distinguished between the complex and simple visual search task (Krems et al., 2000).
The advantage of voluntary occlusion is that the simulated time-sharing aspect may provide a more realistic test as the subject is in control of the viewing frequency. However, the variability of the results when using voluntary occlusion would most likely increase. This would require larger sample sizes and cause more complex assessment (Stevens, 2004).
The more commonly used method in IVIS evaluation is involuntary occlusion. Here the test person is not in control of the vision intervals. The occlusion and vision intervals are continuous and fixed in length also known as system paced occlusion. Only this type of occlusion will be considered in this study in accordance with the ISO recommendations for using the occlusion method in IVIS evaluation (ISO, 2005). A study aiming to determine what involuntary occlusion reveals about time-sharing showed that, even though time lengths were comparable, the subjects chose to divide the tasks differently when driving in traffic than what they were forced to with involuntary occlusion (Tsimhoni, 2003). The advantage of involuntary occlusion is that no secondary task (such as pressing a button and controlling vision intervals) is added and the intervals can be paced using the acceptable time limits for eyes off the road (Stevens, 2004). Of the different ways of occluding, the most frequently used method these days is through software by screen blanking and will be used in our study. The advantages with this method are that it is easy to set the occlusion intervals and it allows for total occlusion. The other method is goggle-like glasses with liquid crystal displays, the displays are either milky white or transparent another method is by occluding the visual information manually. Test results have not seemed to differ between the methods. Subjects have tended to find the spectacles to be slightly uncomfortable and irritable when used for longer periods of time (Stevens, 2004).
2.3 Safety Guidelines and Regulations Based on Occlusion
SAE safety standard recommend that no task should take longer than 15 seconds to perform and no longer than 20 seconds to perform if interrupted (SAE, 2000). TTTUnoccl must therefore signify if the task should be considered safe to use in traffic. This guideline has been heavily criticized as it does not take interruptability into account (Tsimhoni, 2003), (Stevens, 2004).
The Japanese Automobile Manufacturers Association (JAMA) guidelines state that ‘information should no require keeping a close watch on the screen’ (JAMA, 2001). Vision intervals should be 1.5 seconds and 1.0 second occluded interval. A maximum total shutter open time of approximately 7 seconds is recommended, while the maximum total glance duration was 8 seconds (JAMA, 2001), (Tsutomu & Hiroshi, 2002). Since the occlusion method aims to simulate the use of IVIS in a driving situation the total shutter open time should be possible to compare to total glance duration. Also the TTTUnoccl should be comparable with baseline task times and TTTOccl should be comparable with TTTDriving- total task time while driving or while parked (Stevens, 2004).
2.4 Back Ground – Validity and Reliability of Visual Occlusion Technique
Reliability and validity are often confused, but the terms actually describe two completely different concepts, although they are often closely inter-related. Reliability is the consistency of a measure, a measure is said to have a high reliability if it produces consistent results under consistent conditions. For example, measurements of people’s height and weight are often extremely reliable (Carlson, 2009). For the reliance of occlusion test, it is essential that it gives steady results for an IVIS at any time it is applied.
Validity covers the whole experimental concept and establishes whether the results obtained meet all of the requirements of the scientific research method. Reliability does not imply validity; a lack of reliability does place a limit on the overall validity of a test. A test that is not perfectly reliable cannot be perfectly valid. While a reliable test may provide useful valid information, a test that is not reliable cannot possibly be valid (Davidshofer et al., 2005). The validity of occlusion method requires authentication. An occlusion test should have an adequate level of logical validity and must be capable to recognize systems and tasks that can be incompatible for use in a moving automobile. Driving performance with systems that have been accepted or rejected by the occlusion test has to be observed, to set up projecting validity.
A bunch of studies have been made to evaluate the validity of the occlusion technique (Young et al., 2003), (Stevens et al., 2004). There is concern that there has been a lack of synchronization between studies (Young et al., 2003), at the same time individual studies have confirmed that the method has promises. Associations have been found between occlusion measures and a number of measures of driving performance, considering the occlusion trials possibly a valid substitute for on-road and simulator studies (Stevens et al., 2004).
Bonds have been found between TSOT and total glance time in different studies – the total time drivers spend glancing away from the road whilst driving (Niiya, 2000), (Hashimoto & Atsuni, 2001). As compared to total glance time TSOT is much more easily obtained that is why it is very encouraging (Stevens et al., 2004). Recommendations regarding acceptable values for TSOT have been made; however, research suggests that many IVIS seriously exceed these (Stevens et al., 2004). Total glance time has also been shown to correlate well with static task time with full vision (Green, 1999).Many authors have noted that evaluating only total task time cannot predict if an IVIS task is suitable for use while driving (Baumann et al., 2004). Task performance is found to drop considerably when the IVIS task is interrupted. Besides indicating total task time a drastic degradation in lane keeping has typically been shown in simulated driving when interacting with an IVIS requiring too much visual attention (Green & Tsimhoni, 2003).
The occlusion method is aimed for indicating the resumability of IVTs. In the international standard (ISO, 2007), resumability is defined as “ease with which a dialogue can be continued after it is interrupted” and “a dialogue is considered resumable if continuation is without a significant degradation in performance of the completion of the dialogue” (p. 3).
The resumability ratio is suggested to be a measure of how interruptible a task is. A study examining the resumability ratio found that for tasks of similar total task time pointed out that the interruptability of the tasks varied significantly (Fichtenberg, N., 2001). This holds up the view that total task time (TTT) is not sufficient on its own. However, some results seem to indicate that the resumability ratio is very sensitive to TTT but not for task complexity, and had thus been criticized for not indicating task complexity (Fichtenberg, N., 2001). The reason for this is that there is a tendency for short TTT to have high resumability value. The higher the resumability value more unsuitable the task is suggested to be, which the case might not necessarily be when the TTT is very short.
Seeing that TSOT associates well with total glance time it can be inferred that the measure may have same flaws. Therefore, it is proposed that static task time, TSOT and total glance time are not necessarily indicative of the impact of a task on driving performance (Stevens et al., 2004). Care should be taken with the measure as it can be negatively affected by the static task time. The draft ISO standard suggests that the occlusion technique is only suitable for tasks longer than 5 seconds in duration. R has been found to be influenced more by static task time when static task time is short, than by task complexity (Goujon, 2001), (Karlsson & Fichtenberg, 2001).
A particular concern of Applied Ergonomics contributed to the technique was published in 2004 (Gelau & Krems, 2004). In this issue, the occlusion technique was applied to a navigation destination entry task, and compared the results to performance on the same task while driving. They found that whereas the results of occlusion technique did not replicate the visual demands of driving, they did provide an adequate comparison. Similarly, Noy, Lemoine, Klachan, and Burns (2004) found that the task time results using the occlusion technique were similar to those in a simulated driving condition for a range of in-vehicle tasks, including scrolling visual search, static visual search, and radio tuning.
Pettitt, Bayer, and Stevens (2006) carried out a research with the aim to validate the occlusion technique as an in-vehicle device evaluation tool by comparing the occlusion technique and an on-road assessment of two navigation and sound tasks. They found that the resumability ratio R was not related to single glance durations, but was related to the total visual demands of a device. On the whole, they concluded that the trends produced by occlusion technique were similar to the results from the on-road portion of the study. While acknowledging the need for a low-cost and accessible method, and that the prescribed version of the occlusion technique in the ISO standard (ISO, 2007) fulfills that need, it is important to recognize the key aspects of the time-sharing context of driving that is ignored by the visual occlusion method
(Lansdown et al., 2004).
A recent study by Kujala (2012) indicated that the use of Kinetic scrolling for in-vehicle search tasks is not recommended whereas page-by-page swiping seems to go well convincingly better for in-vehicle displays. After the driving, the participants also rated Kinetic as most distracting of the studied methods. Resumability of the interrupted in-vehicle visual search was advocated as the key factor explaining these effects. Due to the unstructured nature of Kinetic scrolling it was suggested to encourage unsystematic search strategies. Generally an additional movement can be seen after the finger is lifted off the screen if the list of options is accelerated with force. Based on how fast the finger was dragged on the screen, the duration, speed and deceleration of the additional movement can vary. By the methods supporting page-by-page scrolling the position of resumption after disruptions should be more effortlessly anticipated. Furthermore, Page swipe supported strategies of changing pages without any visual attention to the device which may explain its superiority in the study.