Driven to distraction

	Sign In Sign-Up

Driven to distraction: in search of better ways to present Route Guidance Information

return

Driven to distraction: in search of better ways to present Route Guidance Information / P G Jackson
In: Paper presented at the PACTS (Parliamentary Advisory Committee for Transport Safety) Conference: Transport Telematics and Road Safety: How many stand to benefit ?.- October 1996.

University of London Centre for Transport Studies

Driven to distraction: in search of better ways to present Route Guidance Information

P G Jackson

Centre for Transport Studies

Imperial College, London

1 INTRODUCTION

After many years in development, in-vehicle route guidance and information systems (IVRGIS) are now available to the general public. Ergonomists and human factors specialists have already played a major part in the development of such systems, with advances being made in interface design, and in understanding the effects upon driver behaviour of diffferent types of guidance information (e.g. Bonsall, 1992). Research such as that of Verwey, 1989, 1993; Wierwille, 1993; and Dingus et al., 1993 has highlighted the potential additional workload associated with screen-based route guidance and information, prompting the development of an auditory interface to convey many instructions. Research supports this development: well-designed verbal guidance is easier than words and symbols for drivers to process (Verwey, .1993), and is actually preferred by drivers (Streeter et al., 1985; Burnett and Joyner, 1996). Moreover, the auditory resource is under-utilised in the driving environment, whereas the visual resource is of central importance to driving - with 30% of all accidents being caused by visual distraction (Staughton and Storie, 1977).

To date, however, little research has considered the possible distraction which might be caused by attending to audible route guidance instructions. It has been assumed that, because the driving task utilises different cognitive resources to listening to audible instructions, there will he no interference when the two tasks are carried out simultaneously. This paper reports on a series of experiments which, have considered this issue. For brevity, the paper does not describe the experiments in any great detail, but briefly describes the method used, summarises the major findings and then discusses the implications of these results. A fuller description of the methodology can be found in Jackson (1995), while Jackson (1996a, 1996b) describe the main results in detail.

2 THEORETICAL BACKGROUND

Approaches to the study of the impact of route guidance upon driver behaviour include: the use of instrumented vehicles in real-world trials; vehicle simulators; laboratory studies using mock-ups of the driving situation; computer simulations which enable researchers to study drivers responses to information; and stated preference surveys which study individual preferences to various aspects of IVRGIS. Real-world studies, whilst having strong validity, are problematic in that driving ability is a highly variable skill, being dependent upon a range of external factors beyond the control of the driver and the experimenter. It is thus difficult to compare the performance of different drivers in real-world situations, where the driving environment and the demands it places upon the individual can never be exactly the same. Car simulators, although costly, can provide response data in a variety of mock-driving situations. However, the extent to which they accurately simulate the driving experience, and the extent to which an individual's behaviour in a simulator can be compared with that in a real driving environment are open to debate.

It is common within cognitive psychology, when studying unseen processes such as memory, to observe effects upon a related cognitive skill, and from this, to infer the possible effects upon the skill of interest. Rather than studying the effects of listening to route guidance upon driving performance directly, the present study uses experiments which consider the effects upon a process known as cognitive mapping.

2.1 Cognitive maps and the introduction of IVRGIS

When we are asked for directions to a particular destination, in the absence of external assistance, we make use of mental representations of previously acquired spatial knowledge to create a suitable route. These internal representations are known as 'cognitive maps'. Cognitive mapping is thus the process by which we acquire, encode, store, retrieve and decode information about our spatial surroundings (Downs and Stea, 1973). Cognitive mapping progresses from a basic knowledge of landmarks to knowledge of the routes which connect these landmarks, and with sufficient experience, the ability to integrate the information gained from several routes into a survey representation of a wider area as a whole. The knowledge we have of our home neighbourhood is usually of this form, allowing us to estimate the direction of unseen locations and to create new routes between places which we have not travelled.

Spatial problem solving behaviours such as wayfinding and navigation, and all travel behaviour, are dependent upon cognitive maps (Arthur and Passini, 1992). On the basis of this information, we make decisions regarding routes, modes of transport, departure times etc. So, for those concerned with travel behaviour and human behaviour generally, the effect that additional information might have upon cognitive maps and cognitive mapping is clearly a matter of some interest. Indeed, transportation researchers have hypothesised that

"the potential for expanding and enriching cognitive maps through ATIS (advanced traveller information systems) may be substantial...." Schofer et al., (1993).

In-vehicle route guidance and information systems (IVRGIS) aim to improve the efficiency of our transport networks by making drivers more aware of the potential route options available to them. By enabling travellers to make informed route choice decisions it is hypothesised that traffic load will be spread more equally across the road network, and the transport system as a whole. In addition, IVRGIS are seen to be capable of reducing the wasted mileage which results from using inefficient and/or incorrect routes.

2.2 From declarative to procedural knowledge

With regard to the development of cognitive skills such as driving and cognitive mapping, Anderson (1982) distinguishes between 2 major stages in development: a declarative stage, involving the interpretation of facts about the skill, and a procedural stage, in which these facts are combined into a set of procedures enabling performance of the skill. With regard to learning a new route, for example, our knowledge of the route is initially in the form of declarative knowledge: a list of objective facts and instructions which, when combined in sequence, enable us to reach a destination. Hence, our initial cognitive maps of a novel environment are based around a series of directions and landmarks at which to make turns. To travel the route we need to recall these instructions in the correct sequence. With more experience declarative knowledge becomes increasingly proceduralised: knowledge of the route becomes more and more automated, and eventually we are able to navigate the route with minimal conscious efffort.

Cognitive mapping involves spatial ability skills which are also important in driving. Drivers are frequently called upon to estimate distances and directions. They must be able to quickly and

2.1 Cognitive maps and the introduction of IVRGIS

When we are asked for directions to a particular destination, in the absence of external assistance, we make use of mental representations of previously acquired spatial knowledge to create a suitable route. These internal representations are known as 'cognitive maps'. Cognitive mapping is thus the process by which we acquire, encode, store, retrieve and decode information about our spatial surroundings (Downs and Stea, 1973). Cognitive mapping progresses from a basic knowledge of landmarks to knowledge of the routes which connect these landmarks, and with sufficient experience, the ability to integrate the information gained from several routes into a survey representation of a wider area as a whole. The knowledge we have of our home neighbourhood is usually of this form, allowing us to estimate tile direction of unseen locations and to create new routes between places which we have not travel led.

"the potential for expanding and enriching cognitive maps through ATIS (advanced traveller information systems) may be substantial...." Schofer et al., (1993).

2.2 From declarative to procedural knowledge

Cognitive mapping involves spatial ability skills which are also important in driving. Drivers are frequently called upon to estimate distances and directions. They must be able to quickly and accurately estimate the width of gaps, the speed of oncoming traffic, and maintain distances between self and other vehicles. In addition, in unfamiliar surroundings, drivers have to scan the-numerous roadside signs for information relevant to them. This process of actively searching out information, processing it and making judgements as to its relevance, develops with experience. The search process also places an additional demand upon available cognitive resources. For newly qualified drivers this additional demand is more problematic because their driving ability is still in the form of declarative knowledge. With extensive experience more of the driving task becomes proceduralised, requiring minimal conscious effort. Once the driving task has become largely proceduralised, the individual is able to actively search out information with relative ease. However, in the first years after passing the driving test, wayfinding whilst driving can cause new drivers problems in terms of switching attention between tasks.

3 RESEARCH QUESTIONS AND PROCEDURE

This research addresses 3 central questions:

Are all drivers equally capable of processing the information contained in audible route guidance instructions?

Does attending to audible route guidance have any impact upon workload?

Are the guidance instructions currently in use the most optimal instructions?

In order to address these questions, the research has used an innovative approach, enabling a comparison of the effects of different types of information upon groups of individuals ranging in age from 17 to 86.

3.1 Procedure

The experiments used video footage to provide a standardised test environment which was the same for all those who saw it. Video was chosen because it gives excellent experimental control, while conveying a realistic image of the environment. Three interconnected routes through an area of West London were filmed (see figure 1). Although each route on its own only provides partial information about the area, the spatial knowledge acquired from viewing all three routes, if combined, should be quite complete. The experiments proceeded as follows. Participants were shown the film of the first route. Having seen the route they performed a series of tasks, designed to test their route knowledge (see Jackson, 1995 for a description of these tasks). Once they had completed the tasks for the first route, they were shown the video of the second route, and then carried out the tasks again. This same process was then repeated for the third route. Having seen all three routes they then carried out two more tasks which tested their ability to integrate the information acquired from the three separate routes into a 'survey' representation of the area as a whole. This was tested using two tasks: the cross route pointing task, which involves participants estimating directions from landmarks on one route to landmarks on the other two routes (which requires the integration of information acquired from two separate routes); and drawing a sketch map of the study area.

Figure 1: The study area showing routes and landmarks

The experiments were carried out in two phases. Phase I concentrated upon identifying group differences between groups in their ability to form cognitive maps of an unfamiliar area. As a result of this phase a second series of experiments was conducted which attempted to improve the effectiveness of the route guidance instructions.

A total of 133 people (67 female, 66 male) took part in phase I of the experiment. A fuller description of the results of phase I can be found in Jackson (1995). Participants all saw the same films, but in phase I were assigned to one of 4 experimental conditions which differed in the information they heard:

Group l: see the videos with no additional information;

Group 2: hear spoken route guidance instructions;

Group 3: hear a news radio broadcast;

Group 4: hear radio, with route guidance information overriding this as required.

Participants were assigned to one of these conditions on the basis of age (18-21, 22-54, 55 and over) gender and driving ability (driver/non-driver). The following section summarises the results of phase I.

Figure 1: The study area showing routes and landmarks

Group 1: see the videos with no additional information:

Group 2: hear spoken route guidance instructions;

Group 3: hear a news radio broadcast;

Group 4: hear radio, with route guidance information overriding this as required.

4 RESULTS

Full results of phase I can be found in Jackson (1996b). The following is a summary of the main findings of this phase of the research.

4.1 Group differences in cognitive mapping ability

The main indicator of acquired spatial knowledge is performance of the cross route pointing task, as this requires subjects to integrate the information acquired from the three separate routes into a configurational representation of the study area as a whole. Analysis of variance (ANOVA) carried out on the cross route pointing task data showed that the main variables age, driver status and sex all had a significant effect upon cognitive mapping ability. Of these main variables, age was the most significant, and large differences were found between the three age groups. Whilst age had been expected to play a part in cognitive mapping ability, the strength of the significant effect of driver status was less expected. However, on all tasks drivers performed better than non-drivers.

Further analysis showed that there was a significant interaction between the 3 main variables. Age was found to be the most significant predictor of performance, with male and female drivers in the youngest age group performing best of all groups. Gender was not significant however, with the two groups performing equally well at both the cross route pointing and map sketching tasks. Driving ability was still significant - both groups performed significantly better than their non-driving peers.

In summary, age clearly has a major impact upon cognitive mapping ability, with both accuracy and quantity of spatial knowledge acquired declining with age. Whilst this effect had been predicted by reviews of previous research, the scale of the decline in cognitive mapping ability from the 18-21 year old subjects to those only a few years older was not anticipated. Drivers performed significantly better than non-drivers in all age groups, although the effect of driving capability was less apparent in the youngest and oldest age groups, where effects of age were more significant. Although ANOVA of the main effect variables showed gender differences in cognitive mapping ability, when driving ability was taken into account, no gender differences were found.

4.2 The effects of listening to route guidance information upon cognitive mapping ability.

The second main aim of phase I of the research project was to study the impact of attending to audible route guidance instructions upon ability to acquire a cognitive map of an unfamiliar area. The experiment required that all subjects were naïve with respect to the study area prior to the commencement of the experiment, and so it was not possible for the same subjects to view the area under different information conditions. Consequently, the experiment was designed to enable a comparison of the performance of similar subjects across the different information conditions. In the middle age band, two groups, matched in terms of gender and driving status, watched the videos under different information conditions. The four levels of information were collapsed into two groups: those hearing salient information (route guidance) and those hearing non-salient information (radio) or nothing. Subjects' performances showed that drivers and non-drivers respond differently to additional information, as do young and old subjects.

The results suggest that, for certain groups, listening to route guidance instructions has an impact upon performance of a primary task. The wayfinding performance of subjects who heard salient information (groups II and IV) was inferior to that of subjects who heard non-salient information. This was most evident in the cross route pointing task. The two groups of male drivers in the 22-54 age group had a mean pointing error of 43 degrees in the non-salient condition and 68 degrees in the salient information condition. Similarly, the two groups of female drivers in this age group had a mean error of 49 degrees in the non-salient condition and 65 degrees in the salient information condition.

5 PHASE II

Having identified that age and ability to drive were significant factors in determining participants' ability to acquire a cognitive map of the study area, phase II focused exclusively upon drivers aged 35 and under. The groups were kept as homogenous as possible in order that the main difference between them would be the information heard whilst watching the videos. Hence, phase II involved 80 participants, aged from 18 to 35, in four groups of 20 (10 male and 10 female). All were students at Imperial College and all held a full driving licence.

The differential performance of similar subjects across the information conditions, and the reduced accuracy evidenced by those subjects who watched the videos in the salient information conditions, led to a reappraisal of the route guidance instructions presented to subjects It was apparent that the route guidance instructions used in phase I, (based upon authentic guidance instructions currently in use) were not assisting in the development of a cognitive map of the area but, on the contrary, were having a negative impact upon this process. Consequently, phase II was designed with the central aim of improving the quality of these instructions.

The four groups saw the same videos watched by the original groups in phase I. They differed only in what they heard in addition to the video footage. The groups in phase II were assigned to one of the following information conditions:

Group 1: Watched the videos in silence;

Group 2: Heard complete route guidance instructions;

Group 3: Heard landmark information followed by directions (eg. "At the garage, turn left");

Group 4: Heard directions followed by landmark information (eg. "Turn left at the garage").

Groups 3 and 4 therefore heard instructions linking the turn information to landmarks occurring in the vicinity of that turn. Previous environmental cognition research has shown that landmarks are most important in learning routes and unfamiliar areas. Hence, conditions 3 and 4 were created to test whether it is possible to strengthen the cognitive maps people acquire, simply by incorporating the landmarks which occur in the vicinity of choice points. By linking the direction of a turn to a landmark an association is made which might otherwise have been overlooked. In order to check whether order of presentation has any effect, group 4 presents the same information but in the opposite order to group 3.

evident in the cross route pointing task. The two groups of male drivers in the 22-54 age group had a mean pointing error of 43 degrees in the non-salient condition and 68 degrees in the salient information condition. Similarly, the two groups of female drivers in this age group had a mean error of 49 degrees in the non-salient condition and 65 degrees in the salient information condition.

5 PHASE II

Group 1: Watched the videos in silence;

Group 2: Heard complete route guidance instructions;

Group 3: Heard landmark information followed by directions (eg. "At the garage, turn left");

Group 4: Heard directions followed by landmark information (eg. "Turn left at the garage").

5.1 Summary of Phase II results

The results of the cross route pointing task in phase I showed that age, sex and ability to drive all had a significant effect upon performance: with younger subjects performing better than older; male subjects being more accurate than female; and drivers performing better than non-drivers. In phase II the cross route pointing task produced much less dramatic effects. Age was again found to be significant (p <0.007), with the group mean for subjects aged 18-21 being slightly more accurate than that of the 22-and-over age group. Information group also had a significant effect upon cross route pointing scores (p<0.05). Those who heard landmarks followed by directions were least accurate, while those who heard directions followed by landmarks were the most accurate.

An analysis of the mean pointing error scores of the four information groups on the single route pointing tasks showed that those who heard directions followed by landmarks were the only group whose performance improved across the three routes. A comparison of the groups' subsequent performances at the cross route pointing task showed that this group had the lowest mean pointing error score at this task. Figure 2 shows the mean direction estimation performances of the four information groups for the three routes and for the cross route pointing task (XRT). The results offer support for the view that improved performance across the three single route tasks is evidence of learning. They also support the view that good configurational knowledge is dependent upon good route knowledge

Figure 2: Direction estimating accuracy by information condition.

6 DISCUSSION

6.1 Improving the quality of information

With regard to the provision of route guidance information, one of the central issues to arise from the results of phase I was whether the poor performance of those who heard full guidance resulted from their reduced role in wayfinding, or whether it was due to quantity of information being processed. Cognitive mapping research has established that self-navigation assists in the development of spatial knowledge of an unfamiliar area, whereas directed guidance leads to poor cognitive maps. If poor performance was due to reduced wayfinding role, then any form of route guidance would have a similar effect, and so changing the instructions would have little effect upon performance. If, on the other hand, it was due to an inability to process the guidance information, then changing the instructions could have an effect. There are two potential processing difficulties: either the quantity of information causes problems of overload; or, the type of information causes problems of comprehension.

Phase II attempted to improve upon the findings of phase I by using guidance instructions which made explicit reference to what was being seen. These amended instructions reduced the quantity of information to be processed, as well as changing the type of information heard. If the performance of groups who heard these instructions was poorer than that of those who heard nothing, but not as poor as those who heard full guidance, then it would lead one to conclude that the quantity of information being processed was having an impact. If this were the case, then reducing the quantity of information would reduce the problem, but any form of guidance would place an additional load and thus would have a detrimental effect upon performance. If type of information was causing problems, then different types of information could improve performance, potentially beyond that of those hearing nothing. However, if the instructions were having a negative effect because they were reducing the subject's role in the wayfinding task, then amending the instructions would have little impact upon performance: all groups hearing instructions would perform worse than those who heard nothing.

The results of phase II suggest that this was not the case: those who heard amended instructions performed significantly better than both those who heard full guidance and those who heard nothing. This implies that the reason for the impaired performance of those hearing full guidance was the type and quantity of information provided. It is also possible that the amended instructions, unlike full guidance, did not reduce the wayfinding role, perhaps even enhancing it by prompting the viewer to make links between turns and landmarks which might not have been made in the absence of guidance. In all of this the role of landmarks appears to be crucial.

6.1.1 The effects of different types and quantities of information

The nature and direction of this effect, however, was not consistent across all tasks. On certain tasks effects were clearly due to the type of information present in the instructions. On the distance estimation task, for example, those who heard full guidance instructions were given distance-to-turn information (e.g. "50 metres to left turn") which some subjects used to assist in estimating total route distance. In other tasks performance seemed to be related to the quantity of information present in the instructions. The most consistent finding, however, was evident in the pointing tasks. Both on the single route and cross route pointing tasks, the direction estimates of those who heard amended

6 DISCUSSION

6.1 Improving the quality of information

6.1.1 The effects of different types and quantities of information

The nature and direction of this effect, however, was not consistent across all tasks. On certain tasks effects were clearly due to the type of information present in the instructions. On the distance estimation task, for example, those who heard full guidance instructions were given distance-to-turn, information (e.g. "50 metres to left turn") which some subjects used to assist in estimating total route distance. In other tasks performance seemed to be related to the quantity of information present in the instructions. The most consistent finding, however, was evident in the pointing tasks. Both on the single route and cross route pointing tasks, the direction estimates of those who heard amended instructions linking directions to landmarks (e.g. "Turn left at the church") were significantly more accurate than other groups. The size of the effect was especially noticeable on the pointing task for the final route viewed, the green route. On this route the mean group error for direction estimates of subjects hearing directions followed by landmarks was only 27.78 degrees, 12 degrees lower than that of those hearing full guidance and 18 degrees lower than the group mean of those who heard no guidance. The cross route pointing task showed the same effect, but differences between group mean error scores were much smaller. This result supports the theory that linking direction information with visible landmarks enabled a stronger representation of each route to be acquired. This enhanced route knowledge enabled a more accurate configurational representation to be created.

6.1.2 The importance of order of presentation of guidance information

The results for the pointing tasks also showed up an unexpected difference between the two versions of amended instructions. Both groups heard precisely the same information, simply presented in two different orders. Consequently, little difference had been anticipated in the performances of the two groups. However, while the mean error scores for those in the directions-landmarks group were much lower than other groups, the mean error scores for those who heard landmarks-directions were noticeably higher. Presentation order of information would thus appear to be of great importance, to some extent more important than the information actually presented. At first this result seems confusing: why should people hearing instructions of the form "Turn left at the garage" perform so much better than those hearing instructions of the form "At the garage turn left" ? An information processing approach to the problem offers one possible explanation by considering the cognitive requirements of dealing with each instruction.

In both scenarios two separate pieces of equally important information have to be processed: the direction that we have to turn in order to follow the correct route; and the junction (landmark) at which we have to carry out this action. Separately, the two pieces of information are useless: it is only when they are integrated that we are able to make use of them. In order for this integration process to take place, the first part of the instruction must be received, processed and stored in working memory while the second piece of information is similarly operated upon. Clearly, if the first piece of information is easy to process, then cognitive resources are made available for the processing of the second piece of information. Conversely, if the first piece of information is ambiguous, or requires some degree of interpretation (and thus additional processing) before it can be stored in working memory, then the cognitive resources available to deal with the second piece of information will be reduced. Indeed, one could argue that, if there is ambiguity in the first piece of information the second piece of information may not be attended to, and so could be missed. From this perspective it is quite clear that those in the direction-landmark group have a distinct advantage. The first part of each instruction delivered consists of a two-word phrase: either 'turn left' or 'turn right'. This information is easily processed, and because it is short, is easily stored in working memory while the second part of the instruction is being attended to and processed. Integration of the two pieces of information can thus begin immediately the second piece of information is processed. In the reverse situation, the first piece of information to be processed is much longer, in the present experiments up to 7 words long as opposed to a maximum of two. The information is also more open to interpretation, and may require visual search to pinpoint the location of the landmark being referred to (although, by definition landmarks should be visually distinctive and easily recognised). As this information is being processed the second piece of information is being delivered. While this is short and easy to process, cognitive resources may still be attending to and processing the initial part of the instruction, thus leaving less attention available to deal with the turn information. In short, the link between the two vital pieces of information is weak, resulting in potential storage and retrieval problems. Given that making estimates of directions between landmarks on different routes depends upon the integration of separately acquired information into a configurational whole, this result supports the view that guidance instructions linking directions to landmarks can enhance the formation of cognitive maps.

6.2 Group differences

6.2.1 The effects of driving experience

Phase I identified group differences due to age and ability to drive. Analysis of the phase II results also revealed differences between groups. Most noticeable of these was the emergence of an effect due to driving experience. Analysis of variance showed that, when the sample was partitioned according to number of years' driving experience, subjects with limited driving experience were less competent at certain tasks. The single route direction estimates of those who had been driving for a year or less were significantly less accurate than those who had been driving for even 2-3 years. In addition, those who had been driving for a year or less recalled and correctly located significantly fewer landmarks than more experienced groups.

Driving experience clearly enhances waytinding ability. However, until the majority of the component tasks of driving have been proceduralised and can be achieved with little conscious effort, wayfinding skills cannot develop, as available cognitive resources are directed towards safely controlling the vehicle. For newly qualified drivers, driving ability is still predominantly in the form of declarative knowledge, thus requiring more conscious effort, and so leaving less spare mental capacity for the development of wayfinding skills. This explains the poor performance of drivers with only 1 years' driving experience relative to those with just 2 or 3 years driving experience. The use of IVRGIS during the first few years of post qualification driving could be problematic, preventing the normal development of wayfinding skills (see figure 3).

6.2.2 The effects of age

Cognitive mapping requires the temporary storage of one piece of information whilst processing and integrating additional information. This process depends upon the use of working memory (Baddeley, 1986). The amount of working memory available to us declines with age. Consequently, the ability of elderly drivers to store information whilst processing further information is impaired, resulting in poor wayfinding and navigating performance.

7 CONCLUSIONS

The results of this research lead to a number of conclusions and recommendations for the future design of in-vehicle route guidance and information systems which incorporate an auditory interface.

Contrary to previous research, all drivers are not the same with regard to their ability to process route guidance instructions. Two groups in particular, drivers who have only recently passed their driving test, and elderly drivers, may experience difficulties in their ability to process additional information whilst safely controlling their vehicle.

Designers of IVRGIS should take account of the differing information processing capabilities of different age and driving ability groups by making different levels of information is weak resulting in potential storage and retrieval problems. Given that making estimates of directions between landmarks on different routes depends upon the integration of separately acquired information into a configurational this result supports the view that guidance instructions linking directions to landmarks can enhance the formation of cognitive maps.

6.2 Group differences

6.2.1 The effects of driving experience

Phase I identified group differences due to age and ability to drive. Analysis of the phase II results also revealed differences between groups. Most noticeable of these was the emergence of an effect due to driving experience. Analysis of variance showed that when the sample was partitioned according to number of years' driving experience. subjects with limited driving experience were less competent at certain tasks. The single route direction estimates of those who had been driving for a year or less were significantly less accurate than those who had been driving for even 2-3 years. In addition those who had been driving for a year or less recalled and correctly located significantly fewer landmarks than more experienced groups.

Driving experience clearly enhances wayfinding ability. However until the majority of the component tasks of driving have been proceduralised and can be achieved with little conscious effort wayfinding skills cannot develop as available cognitive resources are directed towards safely controlling the vehicle. For newly qualified drivers driving ability is still predominantly in the form of declarative knowledge thus requiring more conscious effort and so leaving less spare mental capacity for the development of wayfinding skills. This explains the poor performance of drivers with only 1 years' driving experience relative to those with just 2 or 3 years driving experience. The use of IVRGIS during the first few years of post qualification driving could be problematic preventing the normal development of wayfinding skills (see figure 3).

6.2.2 The effects of age

Cognitive mapping requires the temporary storage of one piece of information whilst processing and integrating additional information. This process depends upon the use of working memory (Baddeley 1986). The amount of working memory available to us declines with age. Consequently the ability of elderly drivers to store information whilst processing further information is impaired resulting in poor wayfinding and navigating performance.

7 CONCLUSIONS

Contrary to previous research all drivers are not the same with regard to their ability to process route guidance instructions. Two groups in particular drivers who have only recently passed their driving test and elderly drivers may experience difficulties in their ability to process additional information whilst safely controlling their vehicle.

Designers of IVRGIS should take account of the differing information processing capabilities of different age and driving ability groups by making different levels of information available. Ideally, drivers should be able to tailor route guidance to suit their particular needs, the purpose of their trip, and their personal comfort level.

Contrary to the predictions of transportation researchers, the audible route guidance instructions currently in use may actually inhibit the development of cognitive maps. Drivers using IVRGIS may not become familiar with the route options available to them, rather, the reliance upon an IVRGIS may prevent the normal development of a mental representation of the area.

To be most effective, route guidance should he expressed in terms of what the driver actually sees at decision points such as road junctions. By linking turn information to salient features and landmarks in the immediate vicinity, cognitive maps of the area might he strengthened.

Newly qualified drivers have not yet acquired good wayfinding skills. It is only once they have mastered the driving task, and this has become predominantly proceduralised, that wayfinding skills really develop. Using an IVRGIS system during the first few years of driving may impact upon the normal development of these skills. causing more, rather than less, wasted mileage due to inefficient route selection.

Figure 3 The development of wayfinding skills with and without lVRGIS

Acknowledgements

This research initiated under support from the Engineering and Physical Sciences Research Council and received additional funding from the University of London Central Research Fund. I would like to thank Gary Burnett and Peter Burns of HUSAT for their helpful advice and suggestions, Mike Williams and Julia Clarke of Vidit for filming and editing the videos, and my supervisor, John Polak. The views expressed, however, are the responsibility of the author alone.

8 REFERENCES

Anderson. J.R. (1982) Acquisition of cognitive skill, Psychological Review, 89 (4) 369-406.

Arthur, P. and R. Passini (1992) Wayfinding: People, Signs and Architecture, McGraw-Hill. New York, NY.

Baddeley, A. (1986) Working Memory Oxford University Press, Oxford.

Bonsall, P.W. (1992) The influence of route guidance advice on route choice in urban networks, Transportation, 19 (1) 1 -23.

Burnett, G.E and S.M. Joyner (1996) Vehicle Navigation Systems: getting it right from the driver's perspective. Journal of Navigation, 49 (2) 174-177.

Dingus. T.A. and M.C. Hulse (1993) Some human factors design issues and recommendations for automobile navigation information systems, Transportation Research C, 1: (2) 119-131.

Downs, R.M and D. Stea (Eds.), (1973) Image and Environment: Cognitive mapping and Spatial Behavior, Aldine, Chicago.

Jackson, P.G. (1995) Using video to simulate wayfinding and the effects of route guidance information: description of methodology, University of London Centre for Transport Studies Working paper ISSN 0967-6309.

Jackson, P.G. (1996a) How will route guidance information affect cognitive maps? Journal of Navigation, 49 (2) 178- 186.

Jackson, P.G. (1996b) Group differences in ability to use verbal route guidance and navigation instructions, in S.A. Robertson (Ed.) Contemporary Ergonomics 1996, Taylor and Francis, London.

Staughton, G.C., Storie, V.J. (1977) Methodology of an in-depth accident investigation survey, Report LR 762, Transport Road Research Laboratory Crowthorne Berks

Streeter, L.A.. Vitello, D., and Wonsiewicz, S. (1985) How to tell people where to go: comparing navigational aids, International Journal of Man/Machine Studies, 22 (4) 549-562.

Verwey, W.B. (1989) Simple in-car route guidance information from another perspective: modality versus coding, Conference record of papers presented at the First Vehicle Navigation and Information Systems Conference, Toronto, Ontario, Canada September 11-13, 1989.

Verwey, W B. (1993) Further evidence for benefits of verbal route guidance instructions over symbolic spatial guidance instructions, Proceedings of the 4th IEEE-IEE Vehicle Navigation and Information Systems Conference, 1993, Ottawa, Ontario.

Wierwille, W.W. (1993) Demands on driver resources associated with introducing advanced technology into the vehicle, Transportation Research C, 1 (2) 133-142.