A Cognitive Approach to Improving Young Novice Drivers' Risk Management

Prasannah Prabhakharan

BSc (Hons)

A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Aviation

Faculty of Science

December 2012

ORIGINALITY STATEMENT

„I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.‟

„I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

AUTHENTICITY STATEMENT

„I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.

iii

Abstract

Every day, just over 1,000 young people under the age of 25 years lose their lives as a result of road crashes around the world. One theme that has gained significant attention to address this „young novice driver problem‟ is that its resolution does not lie solely in developing the physical skill of driving but through the development of the cognitive skills necessary to drive. As such, the present research aimed to investigate the utility of various training methods to improve the cognitive skills of young novice drivers. Research from the aviation industry has demonstrated that a cognitive training method, termed episodic training, can improve pilots‟ risk management in a simulated environment. Drawing from this research, experiment 1 aimed to investigate whether episodic training could produce similar improvements in motorists‟ risk management behaviour, namely in the area of speeding. The results revealed that episodic training was an effective method to reduce young novice drivers‟ tendency to speed in a simulated driving environment. Experiment 2 aimed to examine the impact of episodic training on drivers‟ cognitive resources, with the introduction of a secondary task. The results revealed that implementing a speed management strategy through episodic training was successful in isolation; however, when performed in conjunction with a secondary task, there was a trade-off in terms of how cognitive resources were allocated. This result prompted experiment 3 to explore the cognitive underpinnings of how young novice drivers distributed cognitive resources when performing a dual-task and whether it was possible to train how these resources were allocated. Cognitive resource allocation was calculated by assessing performance on a dual visual and auditory computer task. The results from this experiment revealed that iv

individuals opted to evenly distribute cognitive resources in the dual-task exercise rather than allocate based on the demand characteristics of the task. The results also revealed that cognitive resource allocation can be trained by providing explicit feedback about performance. The results of these experiments contribute to the existing road safety literature and support the need to provide more targeted training for young novice drivers to promote the acquisition of the cognitive skills of driving. These results and their implications for road safety are discussed.

Acknowledgements

When embarking on this PhD, I never imagined the great journey I was about to travel.

However, I did not go on this journey alone and I would like to acknowledge all those that have travelled with me; those for shorter periods and especially those that have come with me all the way.

First and foremost, I would like to thank my supervisor, Dr. Brett Molesworth for his exceptional guidance and mentorship. Brett has provided me with a boundless source of encouragement and support and has helped cultivate my research potential. It has been a pleasure working under his supervision and words cannot truly express my appreciation and gratitude for his unparalleled supervision.

I would also like to thank my co-supervisor, Dr. Julie Hatfield, for her refreshing insight towards the development of this research as well as her supervisory support. Many thanks to Prof. Jason Middleton, Prof. Ann Williamson and Prof. Michael Regan for their academic insight ensuring the quality and integrity of the research. I thank my university colleagues Dr. Naomi Dunn, Amy Chung, Jerome Favand, as well as the staff within the

School of Aviation who have provided me with a stimulating, encouraging and friendly environment for my research to flourish.

I owe my most sincere gratitude to Matthew Vella and Gerard Lewis as well as a countless list of close friends who have been there to nurture and encourage my successes and to overcome more trying times. Their support has allowed me to truly appreciate the value of friendship when embarking on such an endeavour.

Finally, I would not be where I am today without the endless support, encouragement and love I have received from my mother, Thilageswary Prabhakharan and my brother, Prakash Prabhakharan. I will forever be indebted to both of you for nurturing my full potential, both as an academic and as a person.

vii

Table of Contents

Abstract ...... iii Acknowledgements ...... vi Table of Contents ...... viii List of Tables ...... xiii List of Figures ...... xiv

Chapter 1: Introduction ...... 1 1.1 The Young Novice Driver Problem ...... 2 1.2 Inexperience vs. Immaturity ...... 5 1.3 Factors Contributing to Crash Risk ...... 8 1.4 Thesis Overview ...... 9

Chapter 2: Risk Management in Young Novice Drivers ...... 11 2.1 Risk and Hazards ...... 11 2.2 Models of Risk ...... 13 2.3 Hazard Perception ...... 13 2.4 Risk Perception ...... 16 2.5 Self-Assessed Driving Ability ...... 17 2.6 Risk Acceptance ...... 19 2.7 Risk Propensity ...... 20 2.8 Experiences of Risk ...... 21 2.9 Risk, Hazards, Errors and Violations ...... 23 2.10 Training and Development to Improve Risk Management ...... 25 2.10.1 Reducing Risk Behaviours through Script Modification...... 26 2.11 Summary ...... 32

Chapter 3: Driver Training and Education for Novice Drivers ...... 33 3.1 Traditional Driver Training and Education Programs ...... 33 3.2 Graduated Driver Licensing System: Components and Structure ...... 34 3.2.1 Age...... 34 3.2.2 Minimum Number of Driving Hours...... 35 3.2.3 Restrictions throughout the GDL...... 35 3.3 Effectiveness of Graduated Driver Licensing Systems ...... 37 3.3.1 Effectiveness of the GDL Learner Phase...... 38 3.3.2 Effectiveness of the GDL Provisional Phase...... 38 3.4 Effectiveness of Supplementary Programs ...... 40 3.5 Failures of Current Driver Training and Education ...... 43 3.6 Future Direction of Driver Training and Education ...... 45 3.7 Summary ...... 46

viii

Chapter 4: Acquisition of Cognitive Skills of Driving ...... 48 4.1 Model of Information Processing ...... 48 4.1.1 Sensory and Perceptual Systems...... 49 4.2 Attentional Skill ...... 51 4.2.1 Selective Attention...... 53 4.2.2 The SEEV model of Selective Attention...... 55 4.3 Memory Systems ...... 56 4.4 Working Memory Systems ...... 57 4.4.1 Central Executive...... 59 4.4.2 Phonological Loop...... 59 4.4.3 Visuo-Spatial Sketchpad...... 60 4.4.4 Episodic Buffer...... 60 4.5 Long-Term Memory Systems ...... 61 4.5.1 Declarative Memory: Episodic Memory and Semantic Memory...... 61 4.5.2 Procedural Memory and Perceptual Representation Systems...... 63 4.6 Working Memory vs. Long-Term Memory...... 64 4.7 Decision-Making ...... 66 4.7.1 Issues with Decision-Making...... 67 4.8 Multiple Resource Theory and Multitasking ...... 68 4.9 Cognition and Driving ...... 70 4.9.1 Working Memory, Situational Awareness and Driving...... 70 4.9.2 Driver Distraction and Attention ...... 72 4.9.3 Training of Cognitive Driving Skills ...... 73 4.9 Summary ...... 80

Chapter 5: Experiment 1 – Episodic Training to Reduce Risk Behaviours in Young Novice Drivers ...... 82 5.1 Participants ...... 84 5.2 Design ...... 85 5.3 Materials and Apparatus ...... 87 5.3.1 Hardware...... 87 5.3.2 Software...... 89 5.3.3 Cases Examples and Consequences (see Appendix D)...... 92 5.3.4 Wisconsin Card Sorting Test...... 93 5.3.5 Questionnaires and Scales...... 93 5.3.6 Post-Drive Questionnaires...... 97 5.4 Procedure ...... 98 5.5 Results ...... 101 5.5.1 Questionnaires and Scales...... 101 5.5.2 Group 1 and Group 2 Multiple Choice Questions...... 101 5.5.3 5 km Practice Drive...... 102 5.5.4 21 km Test Drive...... 102 5.5.5 Post-Drive Questionnaire...... 107 5.6. Discussion ...... 113 ix

5.6.1 Non Significant Result on 80 km/h Zone (Percentage of Speeding)...... 116 5.6.2 Alternative Explanations...... 116 5.6.3 Results of the Semantic Differential Items...... 118 5.6.4 Perceived Amount of Speeding...... 119 5.6.5 Perceived Effectiveness and Generalisation of Training...... 121 5.6.6 Future Research into Episodic Training...... 121 5.6.7 The Cognitive Driver...... 122 5.7 Conclusion ...... 123

Chapter 6: Experiment 2 – Role of Cognitive Resource Allocation in the Implementation of a Modified Driving Behaviour ...... 125 6.1 Participants ...... 127 6.2 Design ...... 127 6.3 Materials and Apparatus ...... 128 6.3.1 Hardware...... 128 6.3.2 Software...... 129 6.3.3 Questionnaires and Scales...... 130 6.3.4 Mental Arithmetic Task...... 130 6.3.5 Post-Drive Questionnaires...... 131 6.4 Procedure ...... 132 6.5 Results ...... 134 6.5.1 Questionnaires and Scales...... 134 6.5.2 10 km Drive Task...... 134 6.5.3 Mental Arithmetic Task...... 136 6.5.4 Post-Drive Questionnaire...... 137 6.6 Discussion ...... 143 6.6.1 Cognitive Mechanisms...... 144 6.6.2 Alternate Explanation...... 146 6.6.3 Post-Drive Questionnaire...... 147 6.6.4 Cognitive Integration...... 148 6.6.5 Part-Task vs. Whole-Task Training...... 148 6.6.6 Implications...... 149 6.6.7 Limitations and Future Research...... 150 6.7 Conclusion ...... 152

Chapter 7: Experiment 3 – Cognitive Resource Allocation to a Novel Dual-Task in Young Novice Drivers...... 153 7.1 Participants ...... 157 7.2 Design ...... 158 7.3 Apparatus and Stimulus ...... 159 7.3.1 Hardware & Software...... 159 7.3.2 Stimuli...... 159 7.3.3 Inquisit Files – Dual-Task...... 163 7.3.4 Questionnaires...... 164 x

7.3.5 Cognitive Resource Allocation Calculation and Macro File...... 164 7.4 Procedure ...... 167 7.5 Results ...... 170 7.5.1 Demographics...... 170 7.5.2 Cognitive Resource Allocation Results...... 171 7.6 Discussion ...... 173 7.6.1 Naturalistic Learning...... 174 7.6.2 Training of CRA...... 177 7.6.3 Cognitive Mechanism of CRA...... 178 7.6.4 Strategy Development and its Application...... 178 7.6.5 Case-Specific vs. Case-General Strategies...... 180 7.6.6 Motivation...... 181 7.6.7 Lower or Higher Processing to Implement CRA...... 181 7.6.8 Limitations...... 182 7.6.9 Future Research...... 186 7.6.10 Implications...... 187

Chapter 8: General Discussion ...... 190 8.1 Aim and Rationale of the Study ...... 190 8.2 Generalisation of Results ...... 193 8.3 Overview of the Limitations of the Study ...... 195 8.4 Implications for Road Industry ...... 196 8.4.1 Failures of GDL...... 196 8.4.2 Application of Episodic Training...... 197 8.4.3 Compartmentalising Training Programs...... 197 8.4.4 Assessment of Driver Skill...... 198 8.4.5 Training of Cognitive Driving Task...... 199 8.5 Driving and Cognitive Load ...... 201 8.6 Theoretical Implications ...... 202 8.7 Future Research ...... 204 8.7.1 Exploring Episodic Training...... 204 8.7.2 The Gap between Cognition and Behaviour...... 205 8.7.3 Feature Identification Training...... 205 8.7.4 Hierarchical Training...... 206 8.8 Conclusion ...... 207

References ...... 210

Appendices ...... 255 Appendix A – Experiment 1: 5 km practice track (STISIM script) ...... 255 Appendix B – Experiment 1: 10.5 km training track (STISIM script) ...... 263 Appendix C – Experiment 1: 21 km test track (STISIM script) ...... 267 Appendix D – Experiment 1: Cases Examples and Consequences ...... 273 Appendix E – Experiment 1: Participant Information Statement ...... 275 xi

Appendix F – Experiment 1: Participant Script ...... 277 Appendix G – Experiment 1: Percentage of distance speeding (PSY output files) .... 280 Appendix H – Experiment 1: Number of zone violations (PSY output files) ...... 290 Appendix I – Experiment 2: 10 km track (STISIM script) ...... 300 Appendix J – Experiment 2: Visual aid for Mental Arithmetic Task ...... 304 Appendix K – Experiment 2: Participant Information Statement ...... 305 Appendix L – Experiment 2: Participant Script ...... 307 Appendix M – Experiment 3: Participant Information Statement ...... 310 Appendix N – Experiment 3: Visual aid for the dual-task ...... 313 Appendix O – Experiment 3: Participant Script ...... 314

xii

List of Tables

Table 1. Summary of Design of Exp 1 ...... 86

Table 2. Scores on the „Speed‟ Semantic Differential Questions – Exp 1 ...... 108

Table 3. Scores on the „Safe Driving‟ Semantic Differential Questions – Exp 1 ...... 109

Table 4. Proportion of 'Yes' Responses to “Do you recall exceeding the speed limit” ...... 109

Table 5. Number of Times Participants Recalled Exceeding The Speed Limit ...... 110

Table 6. Correlation Between Overall Percentage of Speeding Compared to Number

of Times Participants Recalled Exceeding The Speed Limit In Each Group ...... 111

Table 7. Responses to “Have You Been in a Similar Driving Situation” ...... 112

Table 8. Factorial Design of Exp 2 ...... 126

Table 9. Summary of Design of Exp 2 ...... 128

Table 10. Scores on the „Speed‟ Semantic Differential Questions – Exp 2 ...... 138

Table 11. Scores on the „Safe Driving‟ Semantic Differential Questions – Exp 2 ...... 139

Table 12. Responses to "Do You Recall Exceeding The Speed Limit?" ...... 140

Table 13. Summary of Design of Exp 3 ...... 158

Table 14. Letter Pairs Used for the Visual Stimuli ...... 160

Table 15 Signal Detection Theory Outcomes ...... 165

xiii

List of Figures

Figure 1. Model of driving behaviour in response to potential hazards...... 14

Figure 2. Crash rate by licence status and months of licensure ...... 39

Figure 3: A schematic diagram of human cognitive architecture ...... 49

Figure 4. A schematic diagram of working memory ...... 58

Figure 5. The participant's side of the driving simulator ...... 88

Figure 6: Image of the StiSim simulated environment...... 90

Figure 7. Total percentage of speeding in all zones distributed across group – Exp 1...... 104

Figure 8. Percentage of speeding per zone distributed across group – Exp 1...... 105

Figure 9. Overall frequency of zone violations distributed across group – Exp 1 ...... 106

Figure 10. Frequency of violations per zone distributed across group – Exp 1 ...... 107

Figure 11. Mean ratings of the effectiveness of training – Exp 1 ...... 113

Figure 12. Percentage of speeding (by distance) in week 2 – Exp 2...... 136

Figure 13. Percentage of Correct Responses in the Mental Arithmetic Task ...... 137

Figure 14. Rating of the effectiveness of the training (simulated environment) – Exp 2 .. 141

Figure 15. Rating of the effectiveness of the training (real environment) – Exp 2...... 142

Figure 16. Difficulty rating – Exp 2 ...... 142

Figure 17: An example of the visual target stimuli – Exp 3 ...... 161

Figure 18: Cognitive resources allocation to visual task in week 2 – Exp 3 ...... 172

Figure 19. GDE framework...... 206

xiv

Chapter 1: Introduction

Every year, more than 1.2 million people lose their lives as a result of road crashes (WHO,

2009). In 2012, Australia recorded 1,310 road fatalities; equivalent to 5.75 per 100,000 population (BITRE, 2013). However, this figure is rather modest when compared to the

2010 figures of countries such as Korea, with 5,505 deaths (11.26 per 100,000), Italy with

4,090 deaths (6.78 per 100,000) and the USA, which recorded 32,788 road deaths (10.65 per 100,000; International Traffic Safety Data Analysis Group, 2012).

Fatalities from road crashes only tell part of a story. A disturbing picture emerges when injuries and non-fatal crashes are reviewed, which included permanent injury and disability. At a global level, approximately 20–50 million people suffer non-fatal injuries from motor vehicle accidents every year (WHO, 2007). In the one year period between

2006–2007, an estimated 53,553 people in Australia (approximately 253 per 100,000) were seriously injured due to „land transport crashes‟ with 9243 of these injuries (approximately

155 per 100,000) judged to be life threatening (Henley & Harrison, 2009). The number of injuries in Australia, however, is overshadowed by the 2.49 million people (approximately

820 per 100,000) in the USA that are injured in a motor vehicle crash each year (NHTSA,

2008).

Road crashes also have a notable economic impact. A study by Connelly &

Supangan (2006) assessed that the annual cost of road crashes for Australia in 2003 was approximately AUD $17 billion. More recently the Australian Government conservatively estimates the cost at AUD $27 billion (BITRE, 2009). Further, the global economic cost of 1

road crashes is said to be approximately US $518 billion (AUD 495 billion; WHO, 2009).

Needless to say, road crashes still remain a major cause of preventable injury and death which come at a significant economic burden.

1.1 The Young Novice Driver Problem

Just over 1,000 young people under the age of 25 years lose their lives every day as a result of road crashes around the world, with thousands more injured or disabled (WHO, 2007).

World Health Organization reports that road-related fatalities are deemed the leading cause of death of people between the ages of 15–29 with approximately 30% of all those killed and injured being children and young people under the age of 25 (WHO, 2011). In

Australia, just over 50% of all seriously injured persons in road crashes (52,066 people) are under the age of 30. Given the vulnerabilities of young people, it seems imperative that road safety professionals, including researchers, attempt to address this young novice driver problem.

The problem can firstly be characterised by the disproportionate number of crashes relative to the size of the population. In Australia, 16–24 years olds represent about 15% of the driving population but account for 35% of all fatal crashes and around 50% of injury crashes (Macdonald, 1994). The American Academy of Pediatrics (2006) reports similar disproportions; the 12 million adolescent drivers in the USA only represent approximately

6% of the total driver population but account for approximately 14% of the fatal crashes.

This same disproportionate number of fatalities and injuries relative to the population is a global phenomenon and has been documented in many other countries such as Canada, the 2

UK, The Netherlands, France, Germany, Denmark and Sweden (Engstrom, Gregersen,

Hernetkoski, Keskinen, & Nyberg, 2003). Many improvements in licensing and road safety practices have produced marked reductions in the number of young novice driver deaths

(Chen et al., 2010), but the disproportion still remains (BITRE, 2013).

The second characteristic is the significantly high crash risk of provisional drivers in the initial months of solo driving. Provisional (or post-licensed) drivers have one of the highest crash rates of any driving population; between 20–33 times higher in the first year of driving than supervised driving during the learner period (Forsyth, Maycock, & Sexton,

1995; Gregersen, Nyberg, & Berg, 2003). This extreme crash risk is said to decrease rapidly during the first few months of licensure (De Craen, Twisk, Hagenzieker, Elffers, &

Brookhuis, 2008) or in the first 5,000 kilometres of driving (De Craen, Twisk,

Hagenzieker, Elffers, & Brookhuis, 2011). Sagberg (2000) suggests that there is a 50% reduction in crashes with Norwegian drivers after the first eight months of provisional licensing and similar results have also been found in Sweden (Gregersen, Berg, Dahlstedt, et al., 2000; Gregersen et al., 2000) and in Canada (reduction of 41% in the first seven months; Mayhew, Fields, & Simpson, 2000). McCartt, Shabanova and Leaf (2003) closely examined the declining crash risk of recently licensed drivers and found that the crash rate

(per 100 licensed drivers) was 5.9 in the first months of licensure but decreased to 3.4 in the second month. Within 12 months the rate was between 1.3 and 3.0; a 51%–78% reduction.

In sum, the young novice driver problem is characterised by two major components (the disproportionate number of crashes relative to the population and the initial increased crash risk) which need to be addressed in training and education interventions. It should be noted,

however, that these two components are likely to be interrelated and are not viewed as two distinct factors contributing to the problem.

The magnitude of the young novice driver problem is highlighted through their crash risk. Williams and Wells (1995) suggest that drivers between the ages of 16–19 have a crash risk four times higher than that of older drivers. Further, Williams and Wells (1995) state that 16 year olds have a crash risk three times higher than that of 18 year olds and 10 times higher than 35–39 year olds. In another more comprehensive assessment of international crash risk of young novice drivers in 2006, it was suggested that drivers under the age of 25 are twice as likely than other drivers to be killed in road crashes in Australia,

Canada, Ireland, New Zealand, Poland, Portugal, Spain and the USA, three time more likely in Austria, Great Britain, Belgium, Denmark, Finland, France and The Netherlands and four times more likely in Germany (OECD, 2006).

As part of a strategy to reduce young novice driver fatalities, a comprehensive understanding of the risks associated with driving is necessary (Hodgdon, Bragg, & Finn,

1981). It has been identified that young novice drivers do not necessarily drive any further than older drivers (Goldstein, 1972) but are exposed to riskier types of driving such as night-time driving or with multiple distractions. Whilst older drivers are better able to deal with the increased risk of a crash in certain driving situations, young novice drivers are less able to do so due to their lack of experience (Crettenden & Drummond, 1994). Wasielewski and colleagues investigated the mechanisms of young novice driver risk and identified that they are more likely to have smaller headway distance and to engage in speeding behaviours, both of which are described as risk-taking behaviours (Evans & Wasielewski,

1983; Wasielewski, 1984). However, Macdonald (1994) states that these risk behaviours may not necessarily only be due to active risk-taking but may also result from risk and decision-making errors. That is, incorrect assessment and judgement may cause a risk behaviour; erroneous behaviours. An example of this may be when an individual does not actively monitor their speed and as such, exceeds the speed limit. The error in this example is a failure to effectively manage speed. Importantly, whilst the behaviour may not have been intentional (or a violation), it still results in an increase in their crash risk irrespective.

1.2 Inexperience vs. Immaturity

Two prominent factors which contribute to the young novice driver problem are the driver‟s age (or developmental maturity) and their level of driving expertise (experience).

Whilst both have been identified as risk factors, there is much debate as to which of these factors is more of a primary contributor to the problem (McCartt, Mayhew, Braitman,

Ferguson, & Simpson, 2009; Williams, 1999). Expertise is typically measured by assessing an individual‟s driving hours, licence type or training and/or education. Age on the other hand is measured in years as well as through brain development.. Neuroscientific research suggests that higher-order cognition known as executive functioning, such as selective attention, decision-making, voluntary response inhibition, working memory, multitasking and problem solving, are not fully matured during the teenage years and continues to develop during adolescence (Blakemore & Choudhury, 2006; Hooper, Luciana, Conklin, &

Yarger, 2004; Leon-Carrion, Garcia-Orza, & Perez-Santamaria, 2004; Luciana, Conklin,

Cooper, & Yarger, 2005; Luna, 2004; Luna, Garver, Urban, Lazar, & Sweeney, 2004). 5

The major problem with most research aimed at separating these two factors is that they are highly correlated. That is, age and experience generally increase concurrently during the driver‟s licensure (McCartt et al., 2009). Mayhew, Simpson and Pak (2003) found that the initial inflated crash risk was lower for older novice drivers than younger novice drivers. A report by Williams (1999) stated that 16 year-olds had almost three times the crash risk of older teenage drivers, and compared to middle-aged drivers, this crash risk was exaggerated tenfold. A review by McCartt et al. (2009) suggests that teenage drivers

(16–19) have a higher crash risk than older drivers, even when length of licensure and differences in driving exposure is controlled for. These studies seem to indicate an independent contribution of age to the young novice driver problem. However, in the same review, McCartt et al. (2009) states that, when exposure patterns are controlled for, there are also strong „protective‟ effects due to length of licensure (p. 217). They go on to state that in the 11 studies reviewed to assess age versus experience effects, all but one suggested that the effect of experience was the leading crash risk determinate. Whilst the literature seems to be unclear as to which of these two factors is more of a contributor to crash ri sk, it is widely accepted that whilst inexperience is a risk factor at any age, the combination of inexperience coupled with young age significantly multiplies crash risk (Mayhew,

Simpson, & Pak, 2003; Williams, 1999, 2006).

Another factor which is interrelated with inexperience and immaturity is the engagement in risk behaviours by young novice drivers. Specifically, risk behaviours which stem from a lack of experience are likely to be due to errors (Catchpole, 2005; Hatfield &

Fernandes, 2009; Jonah, 1986), which are defined as “unintentional deviations from

normative performance” (De Winter, Wieringa, Kuipers, Mulder & Mulder, 2007, p. 139).

However, risk behaviours as a result of immaturity are likely to stem from a wilful decision to commit and engage in violations, which are defined as “intentional deviations from normative performance” (De Winter et al., 2007, p. 139). There is a plethora of research to suggest that crash risk is significantly higher in young novice drivers due to their violations

(Catchpole, 2005; Hatfield & Fernandes, 2009; Jonah, 1986; Simon & Corbett, 1996;

Williams, 2003). Reasons for these types of violations include, but are not limited to: a need to seek sensation (Hatfield & Fernandes, 2009), peer influences based on subjective norms as well as individuals‟ attitudes to driving (Engstrom et al., 2003; Parker, Manstead,

Stradling, & Reason, 1992). These ideas are supported by Parker et al., (1992), who found that young drivers reported to experience higher peer pressure to commit road violations

(speeding, drink driving, dangerous overtaking etc.) than older drivers. Roberti (2007) also states that scores on Zuckerman‟s sensation seeking scale correlated with active risk-taking behaviour in drivers and self-reported speeding behaviours.

These studies suggest that engagement in risk behaviours in young novice drivers is likely to be a combination of erroneous behaviours due to their inexperience and risk-taking behaviours due to their immaturity. Given that it can be difficult to distinguish whether behaviours are produced by an error or a violation (or both), the collective term „risk behaviours‟ will be used.

1.3 Factors Contributing to Crash Risk

Much research has been dedicated towards identifying factors that contribute to road crashes, in an attempt to reduce injury and fatalities on Australian roads. The World Health

Organization identifies five key risk factors which contribute to road causalities: speeding, drink driving, failure to wear motorcycle helmets, failure to use seat belts, and failure to use child restraints (WHO, 2011).

Three main factors have been identified as the leading causes of road crashes in

Australia: drink driving, fatigue and speeding (ABS, 2007). In NSW, between 1990 and

2010, there has been a decrease in the percentage of road crashes attributed to drink driving

– from 7.1% to 3.8%. In contrast, the percentage of road crashes attributed to driver fatigue appears to have reached a plateau – from 7.3% to 7.7%. However, speeding as a contributing factor has increased – from 13.4% to 16.8% (Transport for NSW, 2010).

Further, these percentages are likely to be conservative estimates of their contribution to road crashes as it would not be possible to determine their relative contribution in all crashes.

Based on these statistics, it is evident that not only is speeding the biggest contributing factor to road casualties but also seems to be a factor that has shown no reduction in its contribution to fatality rates over the years. Given that research suggests that young novice drivers engage in speeding behaviours (Catchpole, 2005), this will be the primary risk behaviour investigated in the present research. It should be noted that the term

„speeding behaviour‟ henceforth is in reference to drivers who engage in driving speeds which are either in violation of speed limits or are excessive for the road conditions.

1.4 Thesis Overview

The present research conducted as part of this thesis aimed to ultimately achieve one outcome; to facilitate in the understanding and development of cognitive training which would contribute to reducing the young novice driver problem. The research draws from previous research in the road, aviation and psychology domains, with the primary focus on reducing drivers‟ risk behaviours, specifically speeding.

The present chapter defines and highlights the young novice driver problem and key factors which contribute to it. Specifically, a leading contributing factor to the young novice driver problem is inexperience and immaturity (i.e. their engagement in risk behaviours). Chapter 2 explores risk and hazard management problems that affect driving behaviours and focuses on how these skills develop in young novice drivers over time, with experience or through training. Chapter 3 considers current driver training and education and how these methods facilitate drivers‟ skill acquisition. This chapter will also highlight some of the areas of driver training and education that seem to be underdeveloped, specifically in driver cognition. Chapter 4 further explores the human cognitive architecture and how these systems are utilised by drivers. This chapter highlights the important role of driver cognition in how young novice drivers model their behaviour.

Following this, Chapter 5 presents experiment 1 (Prabhakharan & Molesworth,

2011) which extends research in the aviation domain to develop and evaluate a cognitive training method (termed episodic training) to reduce speeding behaviour in young novice drivers. Chapter 6 presents experiment 2 (Prabhakharan, Molesworth & Hatfield, 2012) 9

which aimed to explore how cognitive resources were utilised when implementing a speed management strategy produced by episodic training. Chapter 7 presents experiment 3 which aimed to not only explore how cognitive resources were allocated towards a novel dual- task, but also investigated the potential to train the allocation of these (cognitive) resources.

The final chapter, Chapter 8 discusses the implications of the results from the three empirical studies in respect to road safety (both applied and theoretical).

Chapter 2: Risk Management in Young Novice Drivers

All human activity involves some degree of risk and intuitively, risk-taking has the potential to increase the probability of an unfavourable outcome (Hunter, 2002). As mentioned previously, one of the major contributing factors to the young novice driver problem is their engagement in risk behaviours (Catchpole, 2005; Deery, 1999; Hatfield &

Fernandes, 2009; Jonah, 1986; Simon & Corbett, 1996; Williams, 2003). Risk as a holistic construct in young novice drivers is a complex interplay of attitudes, motivations, perceptions and decision-making (Deery, 1999) and each of these subclasses of risk are discussed below in further detail.

2.1 Risk and Hazards

Breakwell (2007) states that risk is defined under two dimensions: probability and effect. In terms of probability, risk refers to the likelihood of a specific negative event as a result of exposure to a hazard. In terms of effect, risk refers to the extent of the detriment associated with the negative event – that is, the scale or severity of the consequences. Breakwell

(2007) argues that whilst these two definitions are quite distinct and should accordingly have separate words to define them, common usage has led to risk being used for both. This common usage means that risk is used inaccurately in a number of contexts. An example would be „There is a risk of getting caught for speeding this weekend‟. This could be interpreted as the probability of getting caught for speeding and/or could refer to the

consequences associated with speeding this weekend (e.g. fine or loss of licence).

Breakwell (2007) concludes that whilst the two dimensions are distinct, most often they are examined in conjunction with each other. As such, the author also feels that it is unnecessary to distinguish the difference and will refer to risk collectively in both these dimensions.

Breakwell (2007) also highlights that there is often a common misinterpretation of risk and hazards. A hazard is defined as “anything (animate, inanimate; natural or human product) that has the potential to cause harm (to people or their environment)” (Breakwell,

2007, p. 2). Whilst this is evidently a different construct to risk, the terms „hazard‟ and

„risk‟ are also sometimes used synonymously. An example of this would be the „risk‟ of smoking. Breakwell (2007) correctly iterates that the act of smoking is a hazard. The probability of negative outcomes or the effects of smoking are the risk. In the context of this thesis, speeding is a deemed a „hazard‟; the probability of negative outcome or the magnitude of the effect is the „risk‟ associated with speeding. This distinction is important when evaluating risk and hazard management, and as such these definitions will be used strictly throughout this thesis.

Deery (1999) explains that risk can also vary significantly depending on the relative context from which it is viewed. That is, a drivers‟ subjective experience of risk can differ from the objective risk of a situation. Brown and Groeger (1988) state that objective risk is

“the ratio between some measure of adverse consequences of events and some measure of exposure to conditions under which those consequences are possible” (p. 586). For example, considering the number of crashes compared to the size of the young novice

driver population, it can be determined that young novice drivers have an objectively higher risk of crash involvement than older drivers.

Subjective risk is distinct from objective risk and is commonly referred to as risk perception (Brown & Groeger, 1988; Deery, 1999). This is an individual‟s ability to quantify and qualify the objective risk. In theory, the subjective and objective risk of a situation should be harmonious; problems can occur when they are not. An understanding of this distinction is identified as critical when evaluating risk.

2.2 Models of Risk

Many models exist which map human behaviour and the underlying mechanisms that govern them. One such example is the theory of planned behaviour, initially proposed by

Ajzen and Fishbein, (1980) which models how components such as attitudes, motivations, subjective norms and intentions produce behaviour/s. Along the same vein, Deery, (1999) proposed a driver-specific model accounting for production of risk behaviours through hazard and risk perception as well as subjective and objective assessments of risk and skill

(Figure 1). The present chapter will draw largely from the constructs described in this model and will be explored in further detail.

2.3 Hazard Perception

According to Deery (1999), one of the first factors that mediate safe driving behaviour is hazard perception. Hazard perception (in the context of driving) relates to detecting and

identifying potential hazards in the driving environment through effective scanning of the road, evaluation of other drivers‟ location in traffic, and predicting traffic behaviour

(Brown & Groeger, 1988). Many reviews of literature exist which examine hazard

Figure 1. Model of processes underlying driving behaviour in response to potential hazards (adapted from Deery, 1999).

perception as a safety-critical driving skill (Benda & Hoyos, 1983; Crundall & Underwood,

1998; Deery, 1999; Drummond, 1989; Ferguson, 2003; Horswill & McKenna, 2004; Isler,

Starkey, & Williamson, 2009; Mayhew & Simpson, 1995; McKenna & Horswill, 1999;

Senserrick & Haworth, 2005; Soliday, 1975; Soliday & Allen, 1972; Underwood, Crundall,

& Chapman, 2002).

Relevant literature clearly identifies a marked difference in hazard perception in young novice drivers compared to older experienced drivers (Deery, 1999; Mayhew &

Simpson, 1995). Novice drivers display a smaller range of horizontal scanning, utilise peripheral vision less efficiently whilst driving, look closer and fixate on vehicles in front, are poor at identifying distant hazards, check mirrors less frequently, glance at and fixate on fewer objects and less frequently, and focus more on stationary objects compared to more experienced drivers (Benda & Hoyos, 1983; Deery, 1999; Drummond, 1989; Soliday,

1975; Soliday & Allen, 1972). A study by Oude Egberink, Lourens and van der Molen

(1986) also found that younger drivers detected child pedestrians and cyclists significantly less often and drove more dangerously than older drivers.

There is extensive research to suggest that young novice drivers generally fail to perceive situations in their entirety or holistically. Instead, novices tend to focus (or fixate) on smaller components of the larger picture (Deery, 1999; Milech, Glencross, & Hartley,

1989). This apparent „tunnelling‟ effect is not only seen in driving, but is a phenomenon in many other domains such as in chess, radiology and physics when novice performance is compared to expert performance (Boshuizen & Schmidt, 1992; Chi, Feltovich, & Glaser,

1981; Larkin, McDermott, Simon, & Simon, 1980).

Benda and Hoyos (1983) suggest that failures in hazard perception in the context of complex driving scenarios result from a “large information flow per time unit [which] requires high perceptive and cognitive selectivity and constant vigilance” (p. 8). Thus, hazard perception may be lacking in young novice drivers, in part, due to an underdeveloped strategy to manage the information processing demands of driving.

According to Drummond (1989), poor hazard perception increases the probability of crashes due to errors. Deery (1999) argues that more experienced drivers are able to process

and integrate larger quantities of information to make better holistic hazard judgements than less experienced drivers. The development of better hazard perception skills is a clear marker of growing expertise, and critical to young novice driver safety (Drummond, 1989;

Brown & Groeger, 1988).

2.4 Risk Perception

At its most basic level, the perception of risk is about risk estimation; how big is the risk

(Breakwell, 2007)? Brown and Groeger (1988) expand on this idea by stating that drivers‟ risk perception is determined by the interplay between the input of information regarding the perceived hazard of traffic situations (i.e. hazard perception; see 2.3 Hazard Perception) and the individual‟s perceived ability to prevent road crashes, given a perceived hazard; their self-assessed driving ability (s ee 2.5 Self-Assessed Driving Ability).

Vernick et al. (1999) argue that an inaccurate perception of one or both of these components of risk perception (hazard perception and self-assessed driving ability) is a leading cause of fatalities in young novice drivers. Many studies have tried to differentiate between the two factors and their relative contribution to risk perception but this has proven difficult (Brown & Copeman, 1975; Deery, 1999; Drummond, 1989; Groeger & Brown,

1989). This is largely because an individual‟s overall perception of risk is reduced compared to the objective risk of the traffic situation when the individual accounts for their ability to manage the risk through their driving ability.

In the injury domain, research supports the theory that the perceived probability of injury and the severity of potential consequences mediates risk perception (Lowrance, 16

1980; Slovic, Fischhoff, & Lichtenstein, 1979), with some research suggesting the perceived probability of injury to be a stronger contributing factor of the two, to overall risk perception (Wogalter, Young, Brelsford, & Barlow, 1999).

Perceived risk also appears to be negatively correlated with self-reported crash records (Deery, 1999). Quimby (1988) found that drivers‟ subjective ratings of risk were negatively correlated with their self-reported number of crashes and that this was most prevalent in young novice drivers. This negative relationship between risk perception and crash and violation records has also been shown in experimental studies (Adams, 1968;

Currie, 1969; Pelz & Krupat, 1974; Spicer, 1964). Pelz and Krupat (1974) found that drivers who had previously been involved in at least one crash (i.e. crash record; no violations), as well as drivers with a crash record as well as violations, perceived hazards less, perceived less accurately the dangerous features of a traffic scenario and reacted to hazards in a more abrupt manner, compared to drivers with no crashes or violation record

(Pelz & Krupat, 1974).

In sum, risk perception undoubtedly plays a significant role in driver safety. The literature in the field seems to agree that since risk perception is subjective, there is a large potential for the objective risk of a driving scenario to be inaccurately perceived (Brown &

Copeman, 1975; Deery, 1999).

2.5 Self-Assessed Driving Ability

Accurate assessment of one‟s ability is essential for any skill-based activity. In the driving domain, a discrepancy between objective driving ability and self-assessed driving ability 17

can have significant safety impacts. Vernick et al. (1999) suggest that an inaccurate appraisal of one‟s assessed driving ability greatly increases crash risk.

One major factor which contributes to an inaccurate appraisal of self-assessed driver ability is overconfidence (also referred to as optimism bias; Deery, 1999). Whilst most drivers are able to accurately assess the overall risks of driving within the driver population, most drivers disassociate from the population, believing the same objective risk does not apply to them personally (Cairney, 1982; DeJoy, 1992; Groeger & Brown, 1989;

Holland, 1993; Job, 1990). In particular, young novice drivers consider themselves as more skilful and safer than the average driver and this notion is well supported in the literature

(Brown & Groeger, 1988; DeJoy, 1989, 1992; Deery, 1999; Delhomme, 1991; Drummond,

1989; Engstrom et al., 2003; Gregersen, 1993, 1996b; Guppy, 1993; Matthews & Moran,

1986; McCormick, Walkey, & Green, 1986; McKenna, Stanier, & Lewis, 1991; Senserrick

& Haworth, 2005; Spolander, 1982; Svenson, 1981). Matthews and Moran (1986) found that whilst young novice drivers were aware that they were at a greater risk of a crash compared to older more experienced drivers, they rated their own risk as being lower than their peers. McKenna et al. (1991) investigated whether this overestimation of skill resulted from high ratings of themselves or low ratings of other drivers. They found the former to be true, confirming that the problem is as a result of an overestimation of self-assessed driving rather than an underestimation of others.

Brown (1982) suggests that young novice drivers‟ overconfidence stems from the misinterpretation that their rapid acquisition of the physical skills of driving is an indication that they are highly skilled drivers in all capacities. Through this inaccurate perception of

their ability, young novice drivers can wrongly assume they are capable of handling more demanding traffic situations than their skill and experience allows. The process by which the demands of driving and an individual‟s self-assessed ability are evaluated is termed

„calibration‟ (Fuller, 2000; Kuiken & Twisk, 2001; Petzoldt, Weiß, Krems, & Bannert,

2011). Kuiken and Twisk (2001) state that calibration involves the driver being “actively engaged in assessing what the driving task requires in terms of actions or the avoidance of actions, and the potential difficulties involved” (p. 14). Whilst over time and with greater experience, novice drivers are able to acquire a more accurate representation of their driving ability, there is an inflated crash risk during this period (Vernick et al., 1999).

Hence, from a training perspective, it seems intuitive that the aim should be to provide individuals with a more accurate representation of their own driving ability, to facilitate better risk judgements.

2.6 Risk Acceptance

Another determinant of risk behaviours is risk acceptance, which is defined as a “perceived risk or risk threshold which a driver is willing to accept” (Deery, 1999, p. 226). However, it should be noted that this willingness to accept a risk is strongly determined by how the risk is perceived. Thus, inaccurate risk acceptance could result from poor risk perception.

Risk acceptance is a balance of the risk against potential benefit of the risk behaviour (Hunter, 2002; Sokolowska & Pohorille, 2000). It should be noted that, whilst in some cases the risk behaviour may lead to benefits (e.g. engaging in speeding to reach a destination quicker), the actual behaviour itself may also be perceived as a potential benefit 19

(e.g. the „thrill‟ of speeding). Young novice drivers‟ tendency to engage in risky driving behaviours may be seen as an indirect measure of risk acceptance (Deery, 1999).

Behaviours such as unsafe following distances (Evan & Wasielewski, 1983), accepting narrower gaps when entering traffic and running yellow lights (Koneci, Ebbesen, &

Koneci, 1976), and tendencies to speed more often (Wasielewski., 1984) could be interpreted as high levels of risk acceptance. But, as mentioned earlier, this problem could also stem from inaccurate risk perception.

It is also important to note that unlike risk perception or self-assessed driving ability

(which arise from perceptions and judgements), risk acceptance is the first of the factors in

Deery‟s (1999) model which is a motivational factor. That is, the individual actively chooses to engage in a behaviour that they themselves identify as risky – risk-taking behaviour. Hatfield and Fernandes (2009) note that inexperience and errors in driving do not account for all risky driving; a notion supported in the literature (Catchpole, 2005;

Jonah, 1986). From a training perspective, attitudes and motivations seem to play an important role in risk management and these factors need to be considered when developing training to improve risk management in drivers, particularly young novice drivers.

2.7 Risk Propensity

Risk propensity is defined as a positive attitude towards taking recognised risks (Hatfield &

Fernandes, 2009; Rohrmann, 2004). In incorporating risk propensity into Deery‟s (1999)

model, it seems that risk propensity is an attitude, which would feed into risk acceptance.

That is, a positive attitude towards risk is likely to allow for a greater acceptance of risk.

Hatfield and colleagues (Hatfield and Fernandes, 2009; Hatfield, Fernandes, Faunce

& Job, 2008) directly assessed risk propensity by employing self-report questionnaires designed to measure attitudes to risk and found that compared to older drivers, young drivers had lower risk aversion, higher risk propensity and stronger motives for risky driving (speeding and drink driving). Similarly, De Pelsmacker and Janssens (2007) found a significant positive relationship on self-reported attitudes towards speed, predicted intention to speed and subsequent speeding behaviour. Whilst more research is needed in the area, the research by Hatfield and colleagues (Hatfield & Fernandes, 2009; Hatfield et al., 2008) suggest that risk propensity and attitudes towards risk behaviours could be a contributing factor in young novice drivers‟ crash risk.

2.8 Experiences of Risk

There has been little research in the area of how experiences of risk affect risk perception.

Lima, Barnett and Vala (2005) found that the perception of risk (to technological hazards) had a negative relationship to the amount of technology an individual was exposed to. That is, exposure to a hazard can affect future sensitivity to it (Richardson, Sorenson, and

Soderstrom, 1987; Benthin, Slovic, & Severson, 1993; Zuckerman, 1979). Breakwell

(2007) interpreted this correlation to suggest that habitual experiences of hazards reduce their perceived risk whereas a novel experience of a hazard enhances its perceived risk. A study by Twigger-Ross and Breakwell (1999) suggests that this desensitisation effect may 21

only occur when exposure to a hazard is voluntary. Their research suggests that involuntary exposure to hazards is associated with a greater perception of risk and this occurs as a result of the involuntary hazard exposure being perceived as uncontrollable and unfamiliar

(Twigger-Ross and Breakwell, 1999). Thus, how risk is experienced can have major implications for how it is perceived.

Experience can also affect risk perception through the information that exposure offers. Specifically, Breakwell (2007) states that exposure and experience can provide new information about particular events. An example could be that exposure to an old road that has recently been resealed may make the road less dangerous and easier to drive on and, as such, the perceived risk of driving on that road may be reduced. New information as a result of an experience can be affirmative of what is expected or can be novel and unexpected. It can also be actively sought out or it can be passively provided to an individual. Other factors such as the source of the information (in terms of trustworthiness or expertise) can also affect risk perception. Suffice to say that the types of information extracted as a result of an experience can vastly impact risk perception in a number of complex ways.

Breakwell (2007) suggests that one key factor in changing one‟s risk estimate depends on the subjective significance of the experience to the individual – the salience of the information. From a training perspective, these results seem to suggest that, in order to increase an individual‟s risk perception of a hazard, the information provided must be novel and be significant to the individual.

2.9 Risk, Hazards, Errors and Violations

Breakwell (2007) states that errors can lead to the production of hazards, and driving errors are no exception to this. Williams (1988) elaborated on factors which produce errors

(adapted from Breakwell, 2007) which include:

 high workload,

 inadequate knowledge, ability or experience,

 poor design of the technologies that have to be used,

 inadequate supervision or training,

 stressful environment,

 mental state (fatigue, preoccupation anxiety distraction, etc.), and

 change (in routine, circumstance, environment etc.).

As these factors increase or combine, the number and frequency of errors is also likely to increase (Breakwell, 2007; Deitz & Thoms, 1991; Grasso, Rothschild, Jordan, &

Jayaram, 2005). Due to their inexperience with driving, this is particularly problematic for young novice drivers, where the summated weight of these factors would have major safety implications for error-free driving.

Violations on the other hand are seen as a form of risk-taking as it is by definition an active decision to do something that is known to be unacceptable (Breakwell, 2007).

Breakwell (2007) identifies three types of violations:

 deliberate violations (excessive speeding or illegal U-turns),

 habitual violations (changing lanes without indicating), and

 situational imposed violations (going into another lane to avoid a collision) 23

Reason, Parker and Free (1994) also classify violations in a similar way: routine violations (similar to habitual violations) and optimising violations (similar to deliberate violations as well as situational imposed violations). Reason et al. (1994) point out that optimising violations do not necessarily occur to optimise the situation but can also be a means of achieving a goal. An example highlighted by Breakwell (2000) could be that a young novice driver speeds not only to optimise the situation by reaching their destination quicker but may also be a means to test their vehicle‟s performance and their driving skill.

Breakwell (2007) points out how counter-intuitive it is for deliberate and habitual violations to occur (given they are known to be unacceptable behaviours) but states that it may be due to these violations becoming normative. That is, if a violation occurs without consequence, it is deductively valid to assume that subsequent violations will also be without consequence. An example of this would be continuing to engage in speeding behaviour given that previous episodes of speeding have not resulted in a crash/es or infringement/s. Breakwell (2007) also suggests that violations can be an appropriate behaviour, particularly if rules being violated are poor or if benefits of the violation substantially outweigh the cost. From a training perspective, the goal should be to provide sufficient training to ensure that errors are minimised. This can be achieved by reducing exposure to hazardous situations (e.g. restricting blood alcohol limits or passenger restrictions) or by providing adequate training and education to ensure sufficient knowledge and skills for safe driving. In terms of training to reducing violations rates, it seems appropriate that training should focus on highlighting the consequences of aberrant behaviour and/or highlighting the limited positive outcomes.

One of the clearest demonstrations of the relationship between errors, violations and risk behaviours in drivers is widely implemented Manchester Driver Behaviour

Questionnaires (DBQ; Davey, Wishart, Freeman & Watson, 2007; Lajunen & Summala,

2003; Reason et al., 1990). This questionnaire has been extensively used in the road safety literature to demonstrate the association with the likelihood of being involved in an accident (Dobson et al., 1999; Mesken, Lajunen & Summala, 2002; Parker et al., 1995;

Reason et al., 1990).

2.10 Training and Development to Improve Risk Management

Experience and maturity are vital in the development of hazard and risk management skills.

By understanding the processes by which these skills are developed, it may be possible to design targeted training experiences which fast-track their development in young novice drivers.

The present chapter has identified a number of key factors which would facilitate better risk management training; the primary aim of the present research. In terms of providing better appraisal of self-assessed driving ability, Brown (1982) states that training should aim to address the problem of overconfidence in young novice drivers. Similarly,

Watts and Quimby (1980) state that driver training should focus on improving the alignment of subjectively perceived risk with the actual objective risk to facilitate better risk perception, rather than largely focusing on physical driving skills. Brown and

Copeman (1975) also argue that driver training should target young novice drivers with experiences designed to improve appraisal of their own abilities and their limitations. 25

In terms of providing better risk management by facilitating accurate hazard perception, the present chapter has clearly highlighted the issue of „tunnelling‟, or a failure to perceive the driving scene holistically (Benda & Hoyos, 1983; Deery, 1999; Drummond,

1989; Milech, Glencross, & Hartley, 1989). Whilst this skill develops with experience, it seems possible that training interventions could attempt to facilitate in this learning. Brown and Groeger (1988) also state that the primary mechanism behind effective hazard detection

(a necessary skill of hazard perception) is through the development of efficient search strategies skills. As novice drivers become more experienced with the traffic environment, they learn to anticipate potentially hazardous situations. This type of knowledge is said to be stored in cognitive structures known as schemas which are cognitive constructs that organise information to provide a database of knowledge within a specific domain and allow individuals to form expectations about how an object interacts with the world (Deery,

1999; Eysenk & Keane, 2005; Milech, Glencross, & Hartley, 1989; Sweller, 1994). There is research to suggest that modifying these cognitive constructs through training can produce significant improvements to risk management (Molesworth, Bennett, & Kehoe,

2011; Molesworth, Wiggins, & O‟Hare, 2003; 2006).

2.10.1 Reducing Risk Behaviours through Script Modification.

Schemas can be further broken down into constitute sub-structures known as „frames‟ and

„scripts‟. Frames are said to be “knowledge structures relating to some aspect of the world

(e.g. a restaurant) containing fixed structural information (e.g. has floors and tables) and slots for variable information (e.g. materials from which it is made)” (Eysenk & Keane,

2005, p. 384). Scripts (or event schemas) are said to contain “knowledge about events, and 26

consequences of events” (Eysenk & Keane, 2005, p. 383). Schank and Abelson (1977) provide the example of a restaurant script, which contains information about all the usual expectations which occur when having a meal at a restaurant. Based on this example, it can be assumed that the formation of this schema would occur at the outset of the restaurant experience; whether that be through their own experience, through the experience of others or acquiring knowledge about it through another means. This would include: how the waiter should behave, how to order, what the restaurant layout should include etc.

Importantly, if this expectation is violated, appropriate action is taken. For example, if the waiter fails to provide menus, the guest will try to catch their attention to obtain one.

In Bartlett‟s (1932) schema theory, he states that more often than not, the precise material about an event is forgotten, but the underlying schema is not. As a result, stories and events can become distorted when they become incorporated as part of a larger schema.

Termed a „rationalisation error‟, Sulin and Dooling (1974) found that when one group was given a story about a fictitious character named „Gerald Marin‟ and his political ambition and dictator attitudes, and a second group was given the same story but with the name

„Adolf Hitler‟ as the subject, individuals were likely to incorrectly identify that the story also contained the sentence „He hated Jews‟. That is, the schematic knowledge about Adolf

Hitler infiltrated and distorted the memory of the story and produced false recollections.

It seems apparent that the schemas are useful to provide basic outlines of knowledge and expectations within a given domain. That is, they are not detailed knowledge structures which include refined detail about every event encountered. The

usefulness of this type of system is it allows for a multitude of specific case examples to be incorporated into the one schematic network.

Schank and Abelson (1977) added to Bartlett‟s theory by further explaining the dynamic between how typical versus atypical situations are categorised into schemas. They proposed that firstly, new information pertaining to a schema is combined with existing structures within that domain. These episodes are either identified as being typical

(consistent with existing structures) or atypical (inconsistent with existing structures).

Those that are defined as atypical episodes are tagged separately to the overall schema. One important feature is that these atypical episodes are better recognised than typical episodes as they are easier to discriminate from other episodes within the schema. That is, information that is atypical is more salient to the individual than information that is typical.

Finally, long-term recall is better for typical episodes than atypical episodes as time passes, as recall is largely based on schema knowledge (Schank & Abelson, 1977).

It should be noted however, that as the number of atypical episodes increases, the schema would naturally begin to incorporate them into a typical expectation. Further, a significant change to the foundation of the schema may result in episodes previously categorised as not being part of the schema to now be incorporated (and vice versa). For example, if a child categorised all larger animals with four legs as „horses‟, explaining to the child that those with stripes were called „zebras‟ would modify the schema to exclude all animals with stripes. Similarly, if the child was told that „tigers‟ are also a type of cat, this previously uncategorised item is also incorporated into the „cat schema‟.

The use of schema knowledge can also be applied to driving. Through experience, individuals build a large inventory of driving episodes that are incorporated into a driving schema, specifically a driving script. Based on this script, drivers are able to formulate a typical driving situation to not only model their behaviour but use existing knowledge about hazards and risks associated to anticipate dangerous and hazardous situations.

However, according to Roy and Chi (2005), if this knowledge structure is faulty in any way

(e.g. inaccurate, incomplete, or non-existent), then appropriate behaviour is not likely to result. That is, just as the child can mistakenly incorporate a zebra into the horse schema, if a driver mistakenly incorporates particular behaviours, such as speeding, into a „normal driving‟ script, their behaviour may reflect this. Importantly, with reference to the child example, the child will continue to categorise the zebra into the horse schema unless they are given information to correct this. This logic can be applied to road safety as well. Given that risk is an intangible psychological construct, speeding behaviour can be used as an outcome measure. That is, a method to improve young novice driver risk management, such as speed management, may come from maximising their learning potential through refining the acquisition of appropriate script knowledge.

Taking on the idea of modifying scripts to improve risk management, Molesworth and colleagues (Molesworth et al., 2011; Molesworth et al., 2003, 2006) explored the utility of training to reduce pilots‟ risk behaviours. Molesworth et al. (2003) state that most training to reduce risk behaviours in aviation is focused on hazard identification and awareness through the use of crash statistics and case examples in an attempt to highlight potential hazards and the risks associated. Whilst the effectiveness of these programs is not

fully established, Molesworth et al. (2003) argue that it is likely to be limited. One potential criticism they put forward is that exposing individuals to this type of training may actually be detrimental to risk management as the individual dissociates from the population having not encountered similar outcomes given similar risk behaviours. This in turn may support an individual‟s „self-fulfilling‟ prophecy, where the individual overestimates their ability and thus underestimates the risk associated (Molesworth et al., 2003); a theme also present in young novice drivers (Brown & Groeger, 1988; DeJoy, 1989, 1992; Delhomme, 1991;

Drummond, 1989; Engstrom et al., 2003; Gregersen, 1993, 1996b; Guppy, 1993; Matthews

& Moran, 1986; McCormick et al., 1986; McKenna et al., 1991; Senserrick & Haworth,

2005; Spolander, 1982; Svenson, 1981).

Molesworth et al. (2003) suggest that training would be more effective if it were to personalise the risk involved for particular activities and this idea is supported in the literature (DeJoy, 1992). Through a series of studies, Molesworth et al. (2003) examined the effectiveness of methods to train better risk management in pilots, and the role of feedback in correcting inappropriate risk behaviours. There is a vast amount of literature on the usefulness and effectiveness of feedback during training interventions (Adcroft, 2011;

Archer, 2010; Hattie & Timperley, 2007; Kamp, Dolmas, Berkel & Schmidt, 2013;

Komaki, Heinzmann & Lawson, 1980; Phye, 1989; 1991; Porte, Xeroulis, Reznick &

Dubrowski, 2007; Schmidt & Wulf, 1997). Adcroft (2011) suggests that feedback during training works by firstly demonstrating what is „good‟ about the behaviour and secondly by helping the individual close the gap between their current behaviour and the more appropriate or desired behaviour (Archer, 2010; Adcroft, 2011; Kamp et al., 2013).

In applying this theory about feedback during training, Molesworth et al. (2003)‟s study divided qualified pilots in to one of three training groups. The first group were asked to read three crash case examples of a pilot who engaged in low-level flight; the second group watched a case example video describing a pilot who crashed while engaging in low- level flight; and the third group completed a simulated flight where they were required to count the number of cars which passed in front of a house over a 10 minute period.

Following their flight, participants in the third group were given personalised feedback regarding their risky flying; specifically, the minimum altitude to which they descended during the flight. According to the visual flight rules, pilots are not permitted to descend below 500 feet over non-populated areas. Following a one week interval, all pilots completed a simulated flight, which involved reading a number on the deck of an oil tanker.

The pilots who received the simulated flight episode plus feedback descended to an average altitude of 657 feet, approximately 220 feet higher than participants from the other two groups. The results demonstrate that there is a significant improvement in risk management in pilots that received a simulated flight followed by personalised feedback relative to the other types of training. In a subsequent experiment, Molesworth et al. (2006), found that risky low flying behaviour was significantly reduced in pilots that completed a simulated flight and received personalised feedback regarding their risky flying, compared to co- pilots who were merely present during the same flight. Molesworth et al. (2006) attributed the success of their training (later termed episodic training; Prabhakharan & Molesworth,

2011) to the way the pilot‟s „normal flying‟ script is amended through the information the individual received about their own risk behaviour. This in turn facilitates the modification

of the existing script to model future flying behaviour. Further, due to the personalisation

(and subsequent salience) of the information, this training is likely to have produced a deeper level of processing which is said to facilitate storage in long-term memory (Craik,

2002; Craik & Lockhart, 1972). The extension and significance of these studies to the present thesis is deferred until its application in experiment 1 and 2 (chapter 5 and 6).

2.11 Summary

Risk is inherent in all activities and driving is no exception to this. For young novice drivers, risk presents itself in two forms. The first occurs as a result of inexperience and a lack of refinement of driver skills leading to risk behaviours produced through errors. The second occurs as a result of making decisions to commit violation through active risk- taking. These risk behaviours are considered to be a primary mechanism contributing to the young novice driver problem (Evans & Wasielewski, 1983; Wasielewski, 1984). It seems that reducing risk behaviours should be at the forefront of young novice driver training. The following chapter will review current driver training and education systems in an attempt assess the effectiveness of such systems in reducing crash risk. Further, the chapter will aim to identify ways to improve these systems through alternative training and education interventions.

Chapter 3: Driver Training and Education for Novice Drivers

3.1 Traditional Driver Training and Education Programs

The idea of driver training was introduced as a concept in the early 1900s (Mayhew et al.,

2000). After more than a century, innovation and implementation of road safety strategies have produced marked changes in how individuals are trained and licensed and there is significant variation in licensing practices across the globe. A basic framework which currently exists is to implement learning phase/s (which usually consists of both theory and training) before full licensure (Engstrom et al., 2003).

A comprehensive review by Gregersen (1999) compared licensing systems and identified four general types of systems. The single phase licensing system consists of a single test phase comprising of a written test and driving test (theoretical and practical driver education and training). Once these tests are passed, the driver is fully licensed and deemed at an equivalent level to other fully licensed road users (Engstrom et al., 2003). The single phase licensing system with probationary licensing is an extension of the single phased system with the additional requirement of a probationary period of licensure prior to full licensure (Engstrom et al., 2003). Important to note is that during the probationary period, no further training or testing is mandated and there are limited restrictions imposed on the licencee. Two phase systems add another level to single phase licensing systems

(with probationary licensing). Theoretical and/or practical education and training is required during the provisional stage although there are no tests to assess whether any new

knowledge or skills have been acquired (Engstrom et al., 2003). The most widely used licensing system is the three phase licensing system, commonly referred to as the

Graduated Driver Licensing (GDL) System, which further builds on two phase systems by having three stages of licensure: a learner‟s stage, a provisional stage and full licensure.

This type of system is implemented widely across the USA, Canada, New Zealand,

Australia as well as many other European countries (Engstrom et al., 2003). GDL systems are said to allow for the acquisition of the necessary skills and knowledge over an extended period of time, usually 2–5 years (Senserrick & Haworth, 2005). In addition, GDL imposes restrictions on the exposure to high-risk practices, which are gradually lifted over the course of their licensure (Williams, 1999; Williams & Shults, 2010). A more detailed consideration of GDL is provided below.

3.2 Graduated Driver Licensing System: Components and Structure

3.2.1 Age.

It seems intuitive that licensing age needs to be appropriately regulated in any GDL system to ensure that individuals are not only mentally capable of dealing with the demands of driving but also at an appropriate level of maturity to fully appreciate road safety as a construct.

There is a plethora of research to suggest there is a significant positive safety effect from delaying licensure (Begg & Langley, 2009; McCartt, Teoh, Fields, Braitman, &

Hellinga, 2010; Williams, 2009; Williams & Shults, 2010). A recent study by McCartt et

al. (2010) suggests that licensing age can have a significant impact on a driver‟s fatal crash risk. In an evaluation of GDL licensing law in the USA, they found that a delay of six months for drivers between the ages of 15 and 17 reduced their fatal crash risk by seven per cent. Further, a delay of 12 months of licensure demonstrated a reduction of 13 per cent of fatal crash risk. This reduction in crash risk has been demonstrated worldwide (Agent et al.,

1998; Mayhew, Simpson, Williams, & Desmond, 2003; McCartt et al., 2010; Ulmer,

Ferguson, Williams, & Preusser, 2001).

3.2.2 Minimum Number of Driving Hours.

There has been much debate as to whether there is a strong relationship between minimum driving hours and crash risk. Gregersen (1997) found that gaining 120 hours of driving experience compared to 40 hours, lowered crash risk in provisional drivers by 40% in

Sweden. However, the evaluation of USA GDL systems by McCartt et al. (2010) suggests that neither lengthening the minimum holding period of a novice‟s license, nor increasing minimum practice time, has any significant effects on fatal crash risk in drivers aged 15 -

17.

3.2.3 Restrictions throughout the GDL.

In addition to the duration of the licensing process, additional restrictions are typically implemented throughout GDL systems to minimise crash risk. There is well-substantiated evidence to support the effectiveness of these restrictions on young novice driver safety.

Passenger and Night-Time Driving Restrictions

Restricting the number of passengers and times of day of driving during the first few months of probationary licensure has been widely demonstrated to play a critical role in reducing crash risk in young novice drivers (Chaudhary, Williams, & Nissen, 2007;

Chen, Baker, & Li, 2000; Cooper, Gillen, & Atkins, 2005; Doherty, Andrey, & MacGregor,

1998; Ferguson, Teoh & McCartt, 2007; IIHS, 2009; Masten & Hagge, 2004; McCartt et al., 2010; Preusser, Ferguson, & Williams, 1998a, 1998b; Rice, Peek-Asa, & Kraus, 2004;

Senserrick, Kallan, & Winston, 2007; Ulmer, Williams, & Preusser, 1997; Williams, 2003;

Williams, Ferguson, & McCartt, 2007; Williams, Ferguson, & Wells, 2005; Williams &

Shults, 2010).

Night-time driving restriction also seems to play an important role in the reduction of young novice driver crash risk. According to McCartt et al. (2010), night-time driving restrictions beginning at 1 am, reduced crash rates by nine per cent (compared to no restriction) and this reduction grew to 18 per cent when the restriction began at 9 pm.

Mobile Phone Restrictions

GDL systems also place restrictions on use of mobile (or cell) telephones. Mobile telephones are a significant hazard to driver safety, not only through mobile conversation but also with texting (McEvoy et al., 2005). Studies using simulations as well as instrumented vehicles have demonstrated that the use of mobile phones impairs driver performance on measures such as reaction time, speeding and situational awareness

(McEvoy et al., 2005). Further, it has been demonstrated that cognitive impairments occur irrespective of whether drivers use hand-free devices or not (McEvoy et al., 2005). The

impact of this impairment was found to be as high as a four-fold increase of crash risk during a brief call and this effect was not influenced by gender, age or availability of hand- free device (McEvoy et al., 2005; Redelmeier & Tibshirani, 1997).

3.3 Effectiveness of Graduated Driver Licensing Systems

The effectiveness of GDL programs is a topic of much discussion in the literature (Fohr,

Layde, & Guse, 2005; Foss, Feaganes, & Rodgman, 2001; Hallmark, Veneziano, Falb,

Pawlovich, & Witt, 2008; Mayhew et al., 2003; Neyens & Boyle, 2008; Shope & Molnar,

2004; Ulmer, Preusser, Williams, Ferguson, & Farmer, 2000).

As a whole, GDL has repeatedly been demonstrated to reduce crash rates, typically between 10 -30% (Shope & Molnar, 2003; Simpson, 2003; McCartt et al., 2010; Williams,

2006). According to Williams (2006), this is largely due to two major factors. Firstly, the natural delay of licensure allows for maturation as well as an increase in experience.

Secondly, GDL is effective through the implementation of restrictions such as night-time and passenger restrictions (Williams, 2006) which are well documented crash risk factors in the young novice driver population (Lin & Fearn, 2003; Masten & Hagge, 2004; Williams

& Shults, 2010).

Based on a regression analysis of the fatal crash rates in the USA from 1996 to

2007, McCartt et al. (2010) found that overall, GDL and most of its components had strong longitudinal effectiveness in reducing fatal crash risk for novice drivers between the ages of

15 to 17. Langley, Wagenaar and Begg (1995) also found significant reductions in crash rates following the introduction of GDL in New Zealand, and this was most prominent for 37

drivers between the ages of 15-19. Australia is no exception to the significant reductions in young novice driver crash risk as a result of implementing GDL (Senserrick & Whelan,

2003).

3.3.1 Effectiveness of the GDL Learner Phase.

In Australia, prior to learners‟ licensure, all drivers must pass a road law knowledge test.

Once passed, the learner phase of the licensing process is primarily aimed at providing learner drivers with most basic vehicle-handling skills that allows them to operate a motor vehicle (McKnight, 1992; Senserrick & Haworth, 2005). Drivers undergo this type of basic driver training and it is seen as providing the foundation for further development. Learner drivers have the lowest crash risk of any other driver group (Senserrick & Haworth, 2005;

Williams, Preusser, & Ferguson, 1997). Mayhew (2003) also found that over a 24 month learner period, crash rates for learner drivers remained unchanged (see Figure 2). Thus, it seems the learner phase of the GDL does not contribute to the inflated crash risk of young novice drivers. However, this is intuitively expected. Learners under supervision are relatively unlikely to display risk behaviours through errors in driving because they are supplementarily managed by their supervisor. In addition, learner drivers are unlikely to engage in risk-taking behaviours for fear of being reprimanded. It seems far more likely for drivers to engage in risk behaviours post-licensing, either through errors or violations.

3.3.2 Effectiveness of the GDL Provisional Phase.

Figure 2 illustrates the substitutionally higher crash rates of provisional drivers in the initial few months of solo driving (Mayhew et al., 2003) which characterises the young novice

driver problem. Whilst there was a significant and consistent decrease in the first seven months (41% reduction), the rate of decline slows in subsequent months, with a 24 month comparison revealing an overall crash rate reduction of 60% from initial licensure.

Figure 2. Crash rate by licence status and months of licensure (adapted from Mayhew et al. 2003).

The literature seems to suggest that licensing systems such as GDL fail to adequately address this increased crash risk in the young novice driver population in the first few months of solo driving. Based on this assessment, it seems there is still room for improvement in addressing the increased crash risk of young novice drivers. Specifically, two major problems still remain. Firstly, the initial few months of unsupervised driving has an inflated high crash risk. Secondly, worldwide there continues to be a disproportionate

number of young novice driver fatalities compared to the size of the population. These problems may be addressed through the introduction of supplementary training programs.

3.4 Effectiveness of Supplementary Programs

Supplementary programs have long been on offer to provide training in addition to that incorporated within licensing systems, and sometimes intended as post-licensing training.

The administration of current training programs loosely fall into two categories: school based programs and commercial programs.

School-Based Programs

Several school-based programs have been implemented (and studied) through various licensing phases in an attempt to reduce crash risk of post-licensed drivers.

However, several international reviews of literature have demonstrated that there is no clear evidence to suggest that in-school driver education and training programs reduce crash risk of their participants post-licensing (Christie, 2001; Christie & Harrison, 2003; Engstrom et al., 2003; Ferguson, 2003; Mayhew & Simpson, 1996; Senserrick & Haworth, 2005;

Struckman-Johnson, Lund, Williams, & Osborne, 1989; Vernick et al., 1999; Williams,

2006; Woolley, 2000). One of the largest and most comprehensive studies carried out to assess the effectiveness of in-school driver training programs was the DeKalb County project (Hatakka, Keskinen, Gregersen, & Glad, 1999; Mayhew & Simpson, 1996;

Mayhew, Simpson, Williams, & Ferguson, 1998). In this study, which spanned over two decades, a group of 16,000 pre-licensed high school students were randomly assigned to one of three experimental groups. The first group was taught from an existing 72 hour Safe 40

Performance Curriculum (SPC) which included 32 hours of classroom education, 16 hours of driving simulation, 16 hours of driving range training, three hours of practice of evasive manoeuvres and three hours of on-road instruction during the day (20 minutes at night).

The second group underwent the Pre-Driver Licensing (PDL) Curriculum and received the minimum training required to pass a license rest and included a 20 hours classroom education, driving range and simulator training as well as one hour of on-road behind the wheel training complemented by supervised driving with parents. The third group was a control group which received no other formal training other than regular supervisory driving training they were expected to receive from their parents or professional instructors

(Hatakka et al., 1999).

Initial evaluations of the program suggest that there were fewer crashes per licensed driver in the SPC and PDL conditions compared to controls (Stock, Weaver, Ray, Brink, &

Sadoff, 1983). However, based on further evaluation, this was later rejected by Lund,

Williams and Zodar (1986) who suggested there was no significant decrease in crash risk as a result of either of the two training interventions and that in fact, as a result of drivers becoming licensed sooner, the intervention actually led to an increase in crash rates. This was also supported by Mayhew and Simpson (1996) in their review of the project.

Given that the DeKalb County project has widely been considered “the most large, well designed and ambitious effort to evaluate the effectiveness of formal instruction to date” (Mayhew & Simpson, 1996, p. 33), it seems that school-based programs may not be an effective method to reducing crash risk (Hatakka et al., 1999; Mayhew & Simpson,

1996; Mayhew et al., 1998).

Commercial Courses

In addition to school-based programs, commercial courses, designed to provide more advanced and targeted driver skill training, are another method employed to reduce crash risk. In essence, these supplementary training programs are designed to enhance and accelerate the acquisition of skill necessary for driving, primarily focusing on physical skill.

Countries such as Norway and Finland, as part of the learner phase, mandate this type of training to promote safety skills such as skid control and vehicle control on icy roads.

Similar supplementary training programs have been assessed around the globe

(Holubowycz & McLean, 1980; Payne, Brownlea, & Hall, 1984; Sowerbutts, 1975).

However, evaluations of these training programs have found that there was no significant benefit following attendance on these course and, in some cases crash rates for novice drivers actually increased (Engstrom et al., 2003; Gregersen, 1996b; Gregersen & Bjurulf,

1995; Katila, Keskinen, Hatakka, & Laapotti, 2004).

The Department of Transport: London, developed a training program known as the

Pass Plus Scheme to enable young drivers to gain confidence and experience in a range of driving environments (minimum duration of program 6 hours; Edwards, 2005). However, the preliminary results seem to suggest that there was no difference between those that did and did not take part in the training scheme based on self-reported learning experiences, attitudes and behaviours (Edwards, 2005). Nonetheless, the development of such programs demonstrates the need for innovative solutions to address the young driver problem.

Katila, Keskinen and Hatakka (1996) suggest that advanced skill training focused on physical skill may increase confidence more than is warranted. Thus, rather than

providing drivers with skills to cope with unexpected emergency situations as intended, advanced skill training may result in drivers failing to avoid hazardous conditions. This notion is supported in the literature (Keskinen, 1996; Keskinen, Hatakka, Katila, &

Laapotti, 1992; Lynam & Twisk, 1995; Mayhew & Simpson, 1996; Woolley, 2000).

The research literature seems to suggest that there is no clear evidence to support that additional training or education programs (i.e. advanced driver training) have lasting safety implications which reduce the crash risk or traffic violations post-licensing (Christie,

1996, 2001; Ferguson, 2003; Haworth, Kowaldo, & Tingvall, 2000; Langford, 1999;

Mayhew & Simpson, 1995, 1996; Mayhew, Simpson, Williams, & Ferguson, 1996;

McKnight, 1992; Williams & Ferguson, 2004). Based on the evidence derived from these studies, there is a compelling argument not to include such programs (Katila et al., 1996;

Katila et al., 2004; Keskinen, 1996; Keskinen et al., 1992).

3.5 Failures of Current Driver Training and Education

In an international review of literature of effectiveness of driver training in reducing crash risk, Christie (2001) put forward a series of compelling points as to why the conventional ideologies of driver training and education are outdated.

Firstly, Christie (2001) states that advocating driver training and education to improve skill and knowledge as a solution to the young novice driver problem assumes that there is an initial deficiency in basic skill and/or knowledge which is a major contributor to crash risk. Christie (2001) continues by proposing that the risk behaviours, particularly of young novice drivers, have far less to do with physical skill and/or knowledge but are more 43

due to motivations and higher-order cognitive skills, and this idea is well supported in the literature (Catchpole, Cairney, & Macdonald, 1994; Engstrom et al., 2003; McKnight &

Resnick, 1993; Senserrick and Haworth, 2005). In a general sense, higher-order cognitive skills refer to skills such as question asking, critical thinking, systematic/lateral thinking, decision making, problem solving, evaluative thinking, and knowledge transfer (Zoller &

Pushkin, 2007). Application of these skills to driving facilitate hazard and risk perception, self-calibration (ability to moderate tasks based on one‟s ability), attentional control, time sharing (sharing limited attentional resources between driving tasks) and situational awareness (Christie, 2001; Senserrick & Haworth, 2005). In contrast, lower-order cognitive skills are a basic recall or application of memorised information to familiar situations or applying set procedures or rules to repetitious exercises (Zoller & Pushkin, 2007).

Christie (2001) concludes that a failure of many driver training programs is that they focus on developing crash avoidance and emergency driving situations. The problem with this type of training is said to be that the average crash resulting in injury or fatality is likely to be a rare or unforseen event. It seems logical that the driver is unlikely to remember and/or apply their training in these unforeseen circumstances, largely due to forgetting the training coupled with a lack of practice; a concept which is not unique to driving (Christie, 2001). This issue becomes compounded when drivers actively engage in risk-taking behaviours.

Engstrom et al. (2003) state that whilst there seems to be conclusively increased safety benefit from supplementary training and education programs, the reasons for this may not be directly linked to the ineffectiveness of the programs themselves. Firstly, they

note (and as mentioned earlier) that studies that examined driver training in exchange for a time discount (i.e. reduce amount of time under supervision; Boase & Tasca, 1998;

Mayhew & Simpson, 2002) were confounded in their results. That is, the reduction in time as a result of the driver training and the skills gained was not equivalent to the 3–4 months of experience the individuals were discounted. Engstrom et al. (2003) also argue that these types of training and education programs are typically too short (no more than a couple of days) to produce marked long lasting driver safety behaviours – given that this is not expected in other areas of education and training (Engstrom et al., 2003).

Another issue which has been recently identified stems from the assumption that providing driver training and education necessarily produces on-road safety benefits, which may not always be true. Lund et al. (1986) state that programs such as the DeKalb program should only be seen as a method to largely increase driver skill, and fall short of being a comprehensive program that reduces crash involvement. It is argued that effective driver training should not only aim to develop skills but also promote safety behaviour; a result which is supported in the literature (Beanland, Goode, Salmon, & Lenne, 2013; Isler,

Starkey, & Sheppard, 2011).

3.6 Future Direction of Driver Training and Education

The failure of driving training and educations systems, such as GDL, classroom based education programs or advanced driver training courses, seems to be the result of the heavy emphasis on physical skill-based training. Williams (2006) suggests that systems and training such as these largely fail to address key factors important to the safety of young 45

novice drivers, where current systems aim “to control exposure to risky situations, not to change driver attitudes” (p. i5). Hatakka et al. (1999), in an extensive review of driver training literature state the one problem with driver training is that “adequate psychomotor skills and physiological function are not enough for good and safe performance as a driver”

(p. 13).

McCartt et al. (2009) state that whilst driving training and education may help accelerate better basic physical driving skills, this approach is “far less successful in developing the more complex skills and judgements needed to drive safely across a wide range of potentially hazardous, and often rare, situations” (p. 218). Whilst it has been suggested that the physical handling of a motor vehicle as well as the learning of traffic laws can be acquired with 15 hours of driving experience (Hall & West, 1996), Deery

(1999) states that young novice drivers have limited experience in the higher-order cognitive skills necessary to drive safely within the traffic environment.

The literature seems to demonstrate that the future of driver training is not to further develop the motor handling skills in young novice drivers. The aspect of driver training which is largely underdeveloped as a component of training and seems to be lacking in this population is the higher-order cognitive skills necessary to drive.

3.7 Summary

The present chapter summarises current training and education that has been offered to young novice drivers, either through the course of their licensure or as supplementary training. The GDL system is arguably one of the greatest improvements to road safety in 46

modern times; a major contributor to reductions in global crashes rates. However, despite its success, the two core characteristics which make up the young novice driver problem still remains, namely, the disproportionate number of young novice driver crashes relative to the population and the inflated crash risk in the initial few months of licensure.

One major flaw which is clearly identified in the literature is the lack of emphasis on the training of higher-order cognitive skills. Whilst these skills develop over time with greater experience, it comes at a significant cost through the increased crash risk of young novice drivers in their initial months of solo driving. Resolving this increased crash risk should be the primary aim of all novice driver training systems as well as supplementary programs.

It seems imperative to have a comprehensive understanding of not only what cognitive systems are involved in driving, but how they are acquired and how they interact with each other. This understanding would set the foundations for developing training programs to assist young novice drivers acquire these skills quicker and more efficiently.

Chapter 4: Acquisition of Cognitive Skills of Driving

As early as the 1930s (Forbes, 1939), the view that road safety was limited to the physical skill of driving was beginning to change. Specifically, research began to develop the idea that road safety had a much greater complexity than just a physical skill-based behaviour and was beginning to explore the psychological aspects of crash involvement (Drummond,

1989). It now seems intuitive that cognition plays a vital role in road safety. The following chapter aims to explore how cognitive systems are acquired and utilised, and their application to drivers. Importantly, an understanding of cognition the chapter aims to investigate how these cognitive skills can be maximised to promote road safety. Whilst many perspectives and approaches exist to examine the cognitive domain, the present chapter will examine cognition from a systems perceptive. That is, the chapter will aim to isolate and define important cognitive systems and identify how these systems are proposed to work together.

4.1 Model of Information Processing

A concept that is fundamental to models of information processing and attention is that cognitive resources are limited. The introduction of multiple tasks induces high cognitive workload. Cognitive overload occurs where sufficient resources cannot be allocated to manage all tasks effectively. In addition, there are limitations of both the information which can be processed and the rate at which it can be performed (information processed per time

unit; Shinar, 2007). One of the most well-established models of information processing has

been proposed by Wickens (1992), as seen in Figure 3, and this chapter will examine the

working components of this model.

Attentional Resources

Receptors Response Stimuli Decision Response Perception and execution response STSS selection

Working memory

Long-term memory

Feedback Figure 3: A schematic diagram of human cognitive architecture proposed by Wickens (1992; adapted from Shinar, 2007).

4.1.1 Sensory and Perceptual Systems.

According to Wickens (1992), individuals perceive the world largely through the raw

information provided through the sensory systems. With a constant stream of information

which floods the senses, the individual is designed to sift through and decipher what is

relevant. This initially occurs in the short-term sensory stores (STSS; Wickens, 1992). As

STSSs decay extremely quickly (between 0.5- 2 seconds), relevant and salient information

must be extracted efficiently (Sperling, 1960; Treisman, 1964). The system that determines which information is perceived as relevant is for the most part referred to as attention (see

4.2 Attentional Skill).

Information that is not attended to in the brief time it is stored in STSSs is lost. That is, all transient unattended events never enter consciousness and are essentially as though they never happened (Shinar, 2007). However, whilst attention implies the active processing of information, perception is solely to detect a stimulus and not an all-or- nothing process. Moreover, information is processed to varying degrees (e.g. shallow or deep processing of information) and, as such, the human operator can be „aware‟ of them at varying levels of consciousness. Shinar (2007) explains that much of the routine driving

(presumably in experienced drivers) occurs with minimal level of conscious awareness and that the operator is barely aware of it, despite attending to and responding appropriately to stimuli. Many studies have demonstrated this effect; whilst drivers perceive (and sometimes even respond to) road traffic situations, they are unable to explicitly recall what it was

(Martens, 2000; Milosevic & Gajic, 1986; Naatanen & Summala, 1976; Shinar & Drory,

1983). This type of utilisation of cognitive systems is appropriate for the driving domain because, once appropriate behaviours have been executed, there is no need for the information that triggered them to remain in memory stores as it is no longer relevant. In fact, keeping it in may impact negatively on future behaviour as it unnecessarily consumes the limited memory stores.

4.2 Attentional Skill

As mentioned above, attentional skills are required to attend to the multitude of information streaming in from sensory inputs. Drummond (1989) states that attentional skills are the foundation of effective driving. However, it has been stated that attention should not be viewed as a one-dimensional system but rather is influenced by a number of factors. Zaidel,

Paalberg and Shinar (1979) list four basic dimensions to attention: Intensity – the total effort invested towards a target; Distribution – allocation strategy towards multiple targets;

Regularity – persistence and consistency of attentional behaviour; and Mode of control – active or passive. Drummond (1989) states the improvement of these four dimensions through experience results in refined attentional skills necessary for driving.

In terms of attentional errors, Gopher and Kahneman (1971) identify two types: errors of omission or errors of intrusion. Errors of omission result from a temporary lapse in attention or failures to report targets, whereas errors of intrusion result from wrongly attending to (and reporting) non–targets. Avolio, Kroeck and Panek (1985) expanded on this research by using both visual and auditory stimuli and refined the classification of errors in attention. In addition to omission and intrusion errors, they also identified switching errors, which are due to a failure to switch from a non-target to a target. Avolio et al. (1985) found that, compared to volunteer drivers with „no crash‟ history, those with

„one or more crashes‟ were significantly more likely to make errors. In particular, the crash group were more likely to make errors of omission in the visual modality. Switching errors correlated the strongest with crash involvement in both modalities. Avolio et al. (1985) suggested that attentional switching (or the management of switching errors) may be a

necessary attentional skill for safe driving. Further, they suggest that poor novice driver performance may be due to switching errors or as a result of an inability to manage the controlled cognitive processes necessary for effective switching.

Schneider and Shiffrin (1977) define controlled cognitive processes as a temporary activation of sequences, which require active attention, are resource dependent and are within the control of the individual. In contrast, automatic cognitive processes are defined as pre-existing sequences (through learning), which are activated by appropriate stimuli, and proceed automatically, without the necessity for conscious control or the demand for attentional resources (Schneider & Shiffrin, 1977). Schneider and Shiffrin (1977) demonstrated that controlled processing could only occur serially whereas automatic processing could occur in parallel. In addition, their study suggests that attention can be

„attracted‟ should an automatic process require active controlled processing of stimuli.

Stress on the limited cognitive resources can be minimised by controlled processes becoming more automated (Schneider & Shiffrin, 1977) through training or by experience of a particular cognitive process. This allows for the resources, which were initially consumed by the controlled processes, to become available to allocate towards other processes (Schneider & Shiffrin, 1977).

Over time and with greater experience, attentional control becomes more automatic.

Drummond (1989) states one outcome of experience in relation to attention is the use of knowledge of stimulus expectancy. That is, the distribution of attention is “knowledge driven and purposeful” (Drummond, 1989, p. 30) and young novice drivers must learn how to allocate attentional resources through the use of acquired knowledge about stimuli

expectancy. An example of this would be automatically scanning crossroads for hazards such as approaching vehicles. That is, the individual (with appropriate experience) is likely to have knowledge that approaching crossroads can produce hazards such as approaching vehicles and, through the use of this knowledge, attentional systems can automatically scan for them. For all drivers, this skill would become particularly useful in the domain of hazard identification and perception to provide knowledge-based foresight into managing risk (see Chapter 2: Risk Management in Young Novice Drivers). However, without experience and inventory of knowledge (stored in schemas), novice drivers are unlikely to have an automated attentional system; resulting in resource-dependent controlled allocation of attention.

4.2.1 Selective Attention.

Schneider and Shiffrin (1977) define selective attention as “the control of information processing so that a sensory input is perceived or remembered better in one situation than another according to the desires of the subject” (p. 4). This system is vital to the active processing of information due to the limited capacity of working memory systems. In particular, selective attention only becomes of critical use when there are multiple stimuli competing for the same cognitive system and the subject is forced to select how attention is distributed. Schneider and Shiffrin (1977) define reductions in performance which occur due to the misallocation of attention as „selective attention deficit‟. Schneider and Shiffrin

(1977) further break down selective attention deficits into divided attention deficits and focused attention deficits. These deficits follow closely with the errors described in Avolio et al. (1985) and Gopher and Kahneman (1971). Divided attention deficits occur when there 53

is a reduction in performance because the correct stimuli are not given the necessary attention (Schneider & Shiffrin, 1977), which are likely due to errors of omission and switching errors (Avolio et al., 1985; Gopher & Kahneman, 1971). Focused attention deficits are when the individual allocates appropriate attentional resources to the correct stimuli, but fails to ignore competing stimuli; errors of omission in conjunction with errors of intrusion.

Schneider and Shiffrin (1977) suggest that selective attention systems exist due to the limited capacity of cognitive systems. That is, if cognitive systems could process all information without limitation, there would be no need for a division of attention or selection judgements to occur. They propose that, therefore, any selective attention deficits in performance necessarily implies a capacity limitation; an idea well supported in the literature (Driver, 2001; Schneider & Shiffrin, 1977; Shiffrin, 1976).

Many studies have demonstrated the relationship between selective attentional skill and crash involvement (Gopher & Kahneman, 1971; Garrison, 2011; Kahneman, Ben-Ishai,

& Lotan, 1973; Salvucci, 2002; Trick, Enns, Mills, & Vavrik, 2004; Utter, 2001). For example, in the aviation domain, Gopher and Kahneman (1971) found a significant relationship between the number of errors on an auditory selective attention test and the objective level of pilot flying skill. Specifically, pilots who had been independently selected

(through assessment of flying capability) to pilot high-performance planes had fewer performance errors on the selective attention tasks than those who had been selected to pilot slower propeller planes. Further, the selective attention test was validated as a predictive

measure for flying aptitude. Kahneman et al. (1973) also showed that errors on a selective attention task were correlated with the number of crashes in bus drivers.

4.2.2 The SEEV model of Selective Attention.

The SEEV model of selective attention (Wickens, Helleberg, Goh, Xu & Horrey, 2001) attempts to deconstruct the factors which contribute to how attention is allocated and prioritised. This model consists of four factors: Saliency, Effort, Efficiency and Value

(Wicken & Horrey, 2009). Saliency refers to the ability for an object or event to „capture‟ attention (Wicken & Horrey, 2009). Effort refers to a component which discourages scanning between two locations that are further apart (Wicken & Horrey, 2009). That is, objects that are further apart require greater effort to scan than objects that are closer together. Efficiency refers to the tendency to scan locations which have a “higher information bandwidth frequency” (Wicken & Horrey, 2009, p. 61). That is, locations that are most likely to have the greatest amount of information. Value refers to the tendency to attend to locations which are of more value to the task. Importantly, Saliency and Effort are considered to be bottom-up influences on attention; driven by the environment. However,

Efficiency and Value are considered top-down influences; driven by an individual‟s prior knowledge and effortful allocation of resources.

These factors are expected to significantly differ between young novices and older, more experienced drivers in a number of ways. Firstly, whilst both groups are considered to be able to allocate attention using both the bottom-up influences (Saliency and Effort), young novice drivers would be less likely to inhibit these influences towards inappropriate stimuli. This idea is based on well-established literature which demonstrates poor inhibitory 55

control in young people (Blakemore & Choudhury, 2006; Keskinen, Hatakka, Katila,

Lapotti, & Peraaho, 1999). Secondly, given that top-down influences require experience and knowledge of where to assign appropriate expectations and values, it intuitively stands to reason that young novice drivers would be more susceptible to top-down errors than older, more experienced drivers.

The combination of these four factors are said to be able to effectively account for and predict the allocation of attentional resources as well as task prioritisation. In addition, this model has successfully been validated in aviation and in road research as an effective tool for determining how selective attention resources are allocated (Horrey, Wickens &

Consalus, 2006; Wickens et al., 2008; Wicken, Goh, Helleburg, Horrey, & Talleur, 2003;

Wicken & Horrey, 2009).

4.3 Memory Systems

Memory is one of the cornerstones of cognitive science. As defined by Schneider and

Shiffrin (1977), it can be conceived to be “a large and permanent collection of nodes that become complexly and increasingly interassociated and interrelated through learning” (p.

2). This conceptualisation of memory, referred to as a connectionist network model, is well supported in the literature (Rumelhart, McClelland, & Group, 1986; Schneider & Shiffrin,

1977). The model focuses on the representation of information clusters into „nodes‟, and it is stated that all cognition occurs through the activation and interactions of these nodes. As is widely known, memory systems fall largely into two categories: short-term memory systems (STM), now more comprehensively referred to as working memory (WM) systems, and long-term memory systems (LTM; Shinar, 2007). These two systems in 56

conjunction with sensory and perceptual stores (see 4.1.1 Sensory and Perceptual Systems) are said to form a multi-store model of memory (Eysenk & Keane, 2005). These three stores of information are categorised as such because they significantly differ in terms of temporal duration and storage capacity as well as other factors (Eysenk & Keane, 2005).

Whilst the presentation of the model is simplified in this chapter, it provides a useful conceptualisation to understand memory systems and their application to road safety.

In the context of the cognitive model proposed by Wickens (1992), whilst most perceptual information in the driving domain is responded to with minimal cognitive effort, some information (namely, information that has been actively attended to) requires further processing. This is particularly important under emergency or high hazard scenarios where not only is it important to effectively manage the potentially increased rate of information, but also manage and minimise errors. According to Wickens (1992), the management of information is executed through the use of WM and LTM systems.

4.4 Working Memory Systems

Since initially proposed by William James (1890), much research has gone into further understanding the functioning of WM (Atkinson & Shiffrin, 1968, 1971; Baddeley, 1986;

Baddeley & Hitch, 1974; Schneider & Shiffrin, 1977; Shiffrin, 1999). The standard model of STM (Nairne, 2002) states that it is a made up of a collective set of activated information stores in memory. Nairne (2002) continues by stating that information represented in STM is an activation of permanent knowledge, which resides in LTM (Nairne, 2002). Further,

Nairne (2002) states that access to this information is immediate and effortless but that the 57

information is fragile in nature and can be quickly lost. The prevention of this „decay of activation‟ is achieved by rehearsal (Nairne, 2002).

However, this model conceptualised STM as a unitary system under the multi-store model; one system governing all of its workings. Baddeley and Hitch (1974) redefined

STM to a much more complex system known as working memory (WM) which was not just a unitary memory resource, but rather a multicomponent cognitive system (Baddeley,

2000). Many breakthroughs in research have led to WM being conceptualised as having four separate subsystems: central executive (CE), phonological loop (PL), visuo-spatial sketchpad (VSS) and an episodic buffer (EB; Baddeley, 2000; see Figure 4). Each of these subsystems will be explored below with a particular focus on how they can be operationalised in the driving domain.

Visuo-spatial sketch pad (VSS)

Central Executive (CE) Long-Term Sensory Attention Episodic Memory Input Memory Buffer (LTM) (EB)

Phonological Loop (PL)

Articulatory Control

Phonological Store

Figure 4. A schematic diagram of working memory proposed by Baddeley (2000).

4.4.1 Central Executive.

The central executive (CE), as its name would suggest, is the control room for WM. Its role in WM is to deal with the immediate information processing of active nodes of information.

The CE is capable of completing various problem-solving tasks without the limitation of modality. To achieve this role, the CE also utilises and integrates information from both the

PL and VSS, which are slave subsystems (Eysenk & Keane, 2005). Importantly, the CE deals with high cognitive demand tasks and is limited in its processing capacity. In this regard, it has similarities to attentional systems (Eysenk & Keane, 2005). In addition, the

CE also has access to LTM stores which it can also „activate‟ for its purpose (Baddeley,

1996). In addition, Smith and Jonides (1999) suggest that the CE is in charge of switching attention between tasks, planning of subtasks to achieve goals, selective attention, updating and checking the contents of WM and coding time and place features of information.

However, despite all the research investigating the CE, there is much about its function which remains unclear. Firstly, whilst it has been established that its capacity is limited, measuring the actual limits has proven difficult. Whilst it has also been deemed modality free, the exact constraints of this function also remain unclear. Whilst evidence suggests that the CE is a unitary system of working memory, this too remains uncertain

(Eysenk & Keane, 2005). As stated by Baddeley and Hitch (2000, p. 129), “the CE is the least well understood component of the Baddeley and Hitch model”.

4.4.2 Phonological Loop.

The phonological loop (PL) is a slave system within WM. The role of the PL is to deal with spoken and written material. According to Baddeley (1990), the PL also has subsystems of 59

its own; a passive phonological store and an articulatory control process. The phonological store acts as an inner ear and stores speech-based information. Spoken words are directly stored into this system (Eysenk & Keane, 2005). The articulatory control process is responsible for circulating information to ensure that it remains „active‟ in the loop. The articulatory control process also serves the function of converting written material into an articulatory code for the PL to utilise (Baddeley & Hitch, 1974).

4.4.3 Visuo-Spatial Sketchpad.

The Visuo-Spatial Sketchpad (VSS) is used as a temporary storage and manipulation of visual and spatial information (Eysenk & Keane, 2005). Logie (1995) suggests that the VSS also has subcomponents, which are utilised for its operation: the visual cache and the inner scribe. The visual cache is used to store information about the visual form and colour of stimuli whilst the inner scribe deals with movement and spatial information, rehearses information in the visual cache and is responsible for transfer of information to the CE

(Eysenk & Keane, 2005).

4.4.4 Episodic Buffer.

The notion of an episodic buffer is a more recent addition to Baddeley‟s initial model.

Baddeley (2000) introduced this new feature into WM because, whilst the PL and VSS were deemed to process and store specific information, the CE was only defined as having the capacity to process general information; the introduction of the episodic buffer allowed for a storage component for CE. Baddeley and Wilson (2002) state that the role of the episodic buffer is to mediate and store information from the PL and VSS for the CE to

process. Further, the episodic buffer is designed to “integrate information from a range of sources into a single complex structure” (Eysenk & Keane, 2005, p. 204).

4.5 Long-Term Memory Systems

According to Schacter, Wagner and Buckner (2000), in addition to working memory, humans also have four other memory systems: episodic memory, semantic memory, perceptual representation systems and procedural memory. These four systems collectively form Long-term Memory (LTM). As its name would suggest, LTM is a permanent, passive repository of information (Schneider & Shiffrin, 1977). In essence, LTM functions as a database of information at the disposal of the operator. This section elaborates on the functions and mechanisms of the four systems and how they relate to driving.

4.5.1 Declarative Memory: Episodic Memory and Semantic Memory.

According to Schacter and Tulving (1994), as well as earlier work by Tulving (1972), episodic memory and semantic memory form two separate memory systems. Eysenk and

Keane (2005), whilst acknowledging this distinction, still combine these two systems into the category of declarative memory. The main argument for combination of these two systems is that they are both involved in explicit memory. Explicit memory is the memory system utilised for tasks which require conscious activation of memories (Eysenk & Keane,

2005). One of the main arguments against the combination of these two systems came from

Tulving (1972). He argues that episodic memory refers to the storage and retrieval of specific events or episodes occurring in a particular place or at a particular time. In contrast,

he argues that semantic memory is analogous to a “mental thesaurus, organised knowledge a person possesses about the words and other verbal symbols, their meaning and referents, about relations among them, and about rules, formulas and algorithms for the manipulation of these symbols, concepts and relations” (Tulving, 1972, p. 386).

Whilst the debate continues and episodic and semantic memory have many similarities and differences, Eysenk and Keane (2005) state that both systems often have to work in parallel. In the context of the driving domain, remembering driving episodes becomes an invaluable learning tool. The modelling of future behaviour can be derived from previous driving episodes to ensure safe driving practices. In addition, the abstraction of semantic information is also integral to safe driving. Examples include the use of information to interpret road traffic signs and road markings. In the driving domain, these two systems are likely to work dependently. In particular, these two systems are vital for hazard perception. For example, to identify a driver who is swerving in their lane would draw from previous episodes (or knowledge about episodic examples) of where this behaviour has been observed (and their outcomes) but would also draw on semantic knowledge to interpret why the driver is behaving in such a manner (e.g. distracted, fatigued or intoxicated). From an applied perspective in the driving domain, semantic knowledge is largely acquired through education programs but also through experience.

Episodic memory is largely acquired through personal experience, although knowledge of other people‟s experience can be incorporated into one‟s episodic memory (Eysenk &

Keane, 2005).

4.5.2 Procedural Memory and Perceptual Representation Systems.

Similar to the episodic and semantic memory systems, procedural memory and perceptual representation system (PRS) have been viewed as two integrated systems of non- declarative memory; though this is still debated (Polderack, Selco, Field, & Cohen, 1999;

Schacter et al., 2000). Unlike the declarative systems mentioned above, both Procedural learning and PRS are both considered to be learned implicitly; that is out of conscious awareness (Eysenk & Keane, 2005). Schacter et al. (2000) state that procedural memory refers to the learning of motor and cognitive skills. Skill in this context is defined as performance improvements resulting from practice to stimuli within a domain (Polderack et al., 1999). PRS is said to be a “collection of domain-specific modules that operate on perceptual information about the form and structure of words and objects” (Schacter et al.,

2000, pp. 635–636). Schacter et al. (2000) point out that the main interest in this system is the role it plays in identifying an object as a result of a specific prior encounter with the object – also known as repetition priming effect. That is, the processing occurs faster and more easily on a second and successive presentation of a stimulus. The advantage of this system is it reduces the workload on processing systems and allows for easy detection of previous stimuli and events.

In terms of the driving domain, procedural memory intuitively deals with not only the physical skill required to handle the vehicle but also the cognitive skills of driving, such as visual scanning techniques and hazard perception skills. The complementary PRS deals with the processing of familiar objects in the driving scene. Examples include traffic lights, road signs, and lane markings, to name a few. This would allow for faster detection of

stimuli, which becomes of particular use in detecting road hazards. From an applied point of view, the learning and acquisition of skill in these two systems is solely through practice for the procedural memory (through personal experience) and through repetition for the

PRS.

4.6 Working Memory vs. Long-Term Memory

Shinar (2007) points out the several noteworthy differences between WM and LTM and how these features work together to form a functional overall memory system. Firstly, both systems vastly differ in terms of their storage capacity. WM is extremely limited. The limit capacity of WM is exemplified in the famous work by Miller (1956) wh o states that WM has a capacity to hold seven ± two „chunks‟ of information. That is, five to nine units of information can be stored for short amounts of time. Whilst Miller‟s (1956) research has been heavily contested (Baddeley, 2000; Cowan, 2001), the fact still remains that the WM has capacity limits (Shinar, 2007). In stark contrast, according to Shinar (2007), LTM is essentially limitless in its storage capacity and can continue to accrue new pieces of information without forgetting old ones. In the driving domain, the WM processes all immediate and new information being present to the driver and this is facilitated by accessing the database of information stored within the LTM, which is used to formulate cognitive and behavioural outputs.

Another distinguishing difference between the two systems is the nature of the information stored. Information that is processed in WM is largely visual or auditory in nature, compared to information that enters LTM which is largely semantic or conceptual. 64

A clear example of this is when driving to a destination you may correctly respond to all situations in the environment, but there is no deeper level of processing; you may not store all information in memory. However, the LTM system would have been accessed to follow the route you took to reach your destination.

Another difference between these systems, which stems from the previous distinction, is the way in which memories decay. Information in WM can in theory remain there indefinitely, as long as it isn‟t bumped off by another piece of information or does not naturally decay over time (Shinar, 2007). One means of preventing decay is through rehearsal (Shinar, 2007). Importantly, once the necessary outcome has been achieved, most of the time there is no necessity to maintain the information in WM. In contrast, information in LTM is virtually permanent. That is, memories in LTM do not decay in the same sense as in WM. Instead they are prone to becoming „lost‟ (Shinar, 2007). Shinar

(2007) points out through the use of analogy that, whilst books in a library will always remain there, books can be misplaced and essentially lost in the library forever. That is, whilst the physical memory still remains, access via retrieval pathways may not be available or may become lost. In addition, memories which have strong salience or are frequently used are less likely to be susceptible to retrieval failures.

In the context of driving, WM plays an important role in the selection and assessment of all stimuli within the driving environment. However, as mentioned earlier, if the information has been attended to and no action is necessary, the information is lost as it no longer serves any purpose. In terms of LTM, once learned, driving skill becomes a skill

that cannot be easily unlearned. Further, as mentioned above, strongly salient memories as well as frequently used memories are less likely to become lost in the LTM library.

The final distinction defined by Shinar (2007) looks at the retrieval of information from these systems. Whilst retrieval of information in WM is immediate, the retrieval in

LTM is dependent on the efficiency of a search strategy employed. This intuitively makes sense when the library analogy of memory is drawn upon once again; it is much easier and quicker to find an item in a library containing seven ± two „chunks‟ of information compared to a library of theoretically an infinite number of items. It is also intuitive as to how these apply in the driving domain as well; WM has easy access to current information of the driving environment which requires active real-time processing whilst the retrieval of information from LTM (to enrich decision-making) is usually of no pressing immediacy.

4.7 Decision-Making

The next phase in Wickens‟ (1992) model is the utilisation of the information extracted from memory systems to produce outcomes; the process of decision-making. Decision- making in a general and ideal sense is defined as the implementation of a particular response, given the assessment of the values and utility of all (perceived) possible outcomes

(Drummond, 1989). There are a number of processes involved in effective decision- making, at differing levels of analysis (Triggs, 1981). At a general level, decision-making not only involves skill-based factors such as applying heuristics (rule-based techniques) and strategies, but also involves complex motivational inputs (Drummond, 1989).

4.7.1 Issues with Decision-Making.

Drummond (1989) argues that, when it comes to decision-making, the “output can only be as good as the inputs allows” (p. 31). That is, decision-making failures may be due to poor information acquisition, processing or storage, resulting in decisions which lack depth. It seems that increasing the quality and quantity of information an individual receives (or providing training to facilitate this) will intuitively increase the efficiency and quality of decision-making.

Another issue which arises is the dynamic nature of decision-making. As stated by

Drummond (1989), given the complexity of driving tasks and the skills required, there is likely to be large variation in decisions between drivers. As such, similar to hazard perception (see 2.3 Hazard perception), the nature of decision-making highlights the importance of schemas and other cognitive models that drivers adopt to cope with the task demands of driving, and these are intuitively likely to be improved with experience

(Drummond, 1989). In this regard, decision-making skill seems likely to be heavily influenced by the level of experience of novice drivers, though the natural positive relationship between decision-making skills with age (and, as such, brain development and maturation) makes this conclusion unclear (McCartt et al., 2009).

Within the driving domain, empirical research examining decision-making largely does not address it as a function of experience. Most studies in the field look at decisions at points of increased risk such as emergency situations where unrefined decision-making skill can have significant consequences. It has been stated that in emergency driving situations, an inability to formulate an appropriate response plays an important role in

ineffective decision-making (Bathurst, 1980; Drummond, 1989; Malaterre, Ferrandez,

Fleury, & Lechner, 1988). Further, a study by Malaterre et al. (1988) suggested that novice drivers, due to their lack of experience, rely on heuristic (rule) based responses. For example, whilst a novice driver may state that one of the initial responses to a driving hazard is to brake, this response is not generated by drawing from experience but is instead drawn from the individual‟s schema based on general knowledge and rules. That is, the decision-making is not necessarily a judgement based on the available information but is derived based on the most appropriate response within the individual‟s schema. Naturally, as the individual gains experience, the schema absorbs experiential information to incorporate into the existing structure, thus providing a richer inventory of experiences to draw from. This highlights the issues associated with decision-making and the availability of appropriate responses, limited by the individual‟s driving experience; the richer the inventory of experience, the more refined the decision-making skill.

4.8 Multiple Resource Theory and Multitasking

Multitasking refers to the perceptual-motor and cognitive processes involved in performing two or more tasks simultaneously. Over the last 60 years, there has been a vast amount of research interest investigating the cognitive architecture of how simultaneous processing occurs, both from theoretical and pragmatic perspectives (Bahrick, Noble, & Fitts, 1954;

Bahrick & Shelly, 1958; Briggs, Peters, & Fisher, 1972; Kahneman, 1973; Kantowitz &

Knight, 1976; Wickens, 1976; 1980; 2005; 2008; Wicken & Colcombe, 2007; Salvucci &

Taatgen, 2008). Kahneman (1973) introduced the idea that when two or more tasks compete for the same (limited) resource, there is likely to be an impact on performance on some or all tasks. Extensive research by Wickens and colleagues (Horrey & Wickens,

2003; Horrey et al., 2006; Isreal, Chesney, Wickens, & Donchin, 1980; Wickens, 1976;

1980; 1984; 2005; 2008; Wicken & Colcombe, 2007) led to the development of a unified theory known as Multiple Resource Theory (MRT), which attempted to account for how cognitive resources were operationalised when multitasking. Specifically, MRT states that tasks require differing amounts and different types of cognitive resources and the level to which they compete for the same resources determines how well they can be processed simultaneously. Wickens (2008) states that these resources vary on 4 dimensions: stage of processing (perceiving, interpreting or responding), how the information is coded (spatially or verbally), modality of information (visual or auditory), and for visual information, whether it is focal or ambient (centre of vision or in the periphery, respectively).

In a computational model of MRT, Wickens and colleagues (Horrey & Wickens,

2003; Wickens, 2002; 2005; 2008; Wickens, Dixon, & Ambinder, 2006) suggest that how time is shared between tasks is determined by three components: how demanding each task is (whether it can be completed automatically or requires active processing), the degree of resource overlap, and resource allocation policy (task prioritisation). Wickens (2008) also highlights how mental workload impacts this computational model. Specifically, he states that of the three components of the model (demand, resource overlap and resource allocation), mental overload most strongly relates to demand. Wickens (2008) states that when demand is low and there is a „residual capacity‟, there is no substantial value to the

model. However, when mental capacity is exceeded (i.e. there is cognitive overload), the model becomes valuable in assessing where performance failures occurred and how these failures can be mitigated.

The application of the computational model of MRT in driving is said to account or help explain multitasking in driving situations (Horrey & Wickens, 2003). In assessing the performance of driving tasks, Horrey and Wickens (2003) found that the MRT model was able to successfully predict the maintenance of lane keeping and responding to unexpected hazards in a simulated driving environment.

4.9 Cognition and Driving

4.9.1 Working Memory, Situational Awareness and Driving.

Much research has been directed towards understanding how the aforementioned cognitive systems are utilised to manage the cognitive demands of driving. Given the importance of immediate information processing whilst driving, it is important to understand how working memory systems are operationalised whilst driving. This ability to actively or passively receive, process and understand one‟s surroundings is commonly referred to as situational awareness (SA; Endsley, 1995, 2000; Gugerty, Brooks, & Treadaway, 2004;

Johannsdottir & Herdman, 2010), and the role of SA, along with working memory and driving ability is well documented (Gugerty & Tirre, 2000; Gugerty, Rakauskas, & Brooks,

2004; Salvucci, 2001; 2005; Salvucci & Taatgen, 2008).

Endsley (1995) divided SA into three levels of processing: (1) perception of elements, (2) comprehension and (3) processing of the elements. Johannsdottir and

Herdman (2010) suggest that perception of elements, such as perceiving the onset of brake lights ahead, can be automatic or controlled requiring selective attention systems to detect relevant information. They state that selective attention systems at this level are likely to particularly draw from CE to act as a supervisor of information processing and to collate information from the driving environment to determine its usefulness to the operator

(Johannsdottir & Herdman, 2010). In addition, there is research to suggest that the CE within working memory would play a critical role in managing cognitive overload (Cowen,

Morey, AuBuchon, Zwilling, & Gilchrist, 2010; Harbluk, Noy, Trbovich, & Eizenman,

2007). Harbluk et al., (2007) found that when the central executive was engaged in processing a secondary task, drivers narrowed their scanning view and scanned less efficiently.

The latter two level of Endsley (1995)‟s model of SA refers to the further processing of selected information. Johannsdottir and Herdman (2010) state that this information is received, maintained and updated in both the VSS and PL and further processing would occur through the use of an episodic buffer and long-term memory.

Previous research in both road and in aviation suggests that the VSS and PL are utilised to code and store different components of SA (Aretz, 1991; Barshi & Healy, 2002;

Garden, Cornoldi, & Logie, 2002; Logie, Baddeley, Mane, Donchin, & Sheptak, 1989).

Specifically, Aretz (1991) found that SA in pilots was largely supported by VSS but under higher cognitive workload, both the VSS and PL were operationalised. This research

supports the idea that the PL is utilised as a support system when there is high SA workload, beyond the capability of the VSS alone.

More recent research into working memory with motorist suggests that the role of the VSS may largely be to process information in the forward view and that the role of the

PL may be to largely process information in the rear view (Johannsdottir & Herdman,

2010).

4.9.2 Driver Distraction and Attention

As mentioned previously, the appropriate selection and division of attention is paramount to safe driving. Within the driving domain, poor selection or division of attention or when attention is diverted away from primary driving tasks in favour of a secondary task, is commonly referred to as driver distraction (Wickens & Horrey, 2009). There is a vast amount of research literature which demonstrates the relationship between distraction and driver safety (Alm & Nilsson, 1995; Geary & Wiley, 1991; Harbluk et al., 2007; Horberry,

Anderson, Regan, Triggs, & Brown, 2006; Jenness, Lattanzio, O'Toole, & Taylor, 2002;

McKnight & McKnight, 1993; Regan, Lee, & Young, 2009; Salvucci, 2002; Srinivasan &

Jovanis, 1997; Wickens & Horrey, 2009; Young, Regan, & Hammer, 2003; Young &

Regan, 2007). Wickens and Horrey (2009) suggest that the major safety implication through driver distraction is a failure to respond to unexpected hazards. They state that this occurs through a phenomenon known as attentional blindness, where an individual is essentially „blind‟ as a result of poor allocation of attention. This blindness is categorisesd as Inattentional Blindness (IB) and Change Blindness (CB; Wickens & Horrey, 2009). IB 72

through driver distraction is stated to occur when an individual fails to „see‟ an object, even when the object is in clear view; a „look but did not see‟ phenomenon (Wickens & Horrey,

2009). An example would be when an individual fails to see a stop sign at an intersection as a result of being distracted. CB through driver distraction, however, is defined as a failure to see unattended changes. An example of this would be failing to notice a change in traffic conditions on the road ahead.

Wickens & Horrey, (2009) notes that whilst these types of errors can occur when an individual is not distracted, undistracted individuals are much quicker to detect and respond to hazards and are less likely to miss hazards; an idea supported in the literature

(McCarley, Vais, Pringle, Kramer, Irwin & Strayer, 2004; Richard, Wright, Ee, Prime,

Shimizu, & Vavrik, 2002).

4.9.3 Training of Cognitive Driving Skills

As this chapter illustrates, there are a number of cognitive systems that need to be developed in order to effectively manage the task demands of driving. How these systems are operationalised provides the foundation for the development of driving-critical cognitive skills. The acquisition of these cognitive skills of driving is of paramount importance to reducing crash involvement (Mayhew & Simpson, 1995). Thus, an intuitive aim would be to fast-track the learning of these cognitive skills; to facilitate the development of expertise.

Whilst limited, there have been attempts to facilitate the acquisition of cognitive driving skills through training. These training programs largely aim to provide individuals 73

with episodic memories of high-risk driving scenarios by exposing them to hazards in a controlled environment. The logic is that an individual will actively process features of the environment and hazards during training (through attentional and WM systems) and in the event that a similar situation presents itself, the recognition of these features (stored in

LTM) will allow for the application of appropriate behaviours (Shinar, 2007). These programs are also assumed to facilitate in more fluent decision-making, as the same information has been previously processed during training. In addition, these programs provide individuals with safety-critical knowledge (such as statistics or implications of unsafe behaviours) which can be stored in LTM and drawn on to enrich WM and decision- making processes (Shinar, 2007).

The most prominent of these is the insight training program developed by

Gregersen and colleagues (Gregersen, 1995, 1996a, 1997; Gregersen & Bjurulf, 1995) in

Sweden. The core idea of insight training is to facilitate the acquisition of physical and cognitive driver skill whilst ensuring that the driver does not become overconfident with their new ability (Gregersen, 1996a). Insight training typically involved (but was not limited to) tasks such as estimating safe distances, avoiding a simulated animal, models demonstrating the head from crash test dummies, roll-over simulation as well as self- diagnosis of strengths and weakness, to name a few. In addition, there is usually a focus on raising awareness of the safety benefits of seat belts, speed management and vehicle headway through modification of driver attitude and motivation. Insight training is considered a cognitive program because it not only provides drivers with a multitude of information to later draw upon (through LTM), but provides them with the training and

experience of how to effectively utilise the cognitive systems at their disposal. There has been a plethora of research which supports insight training around the world (Catchpole et al., 1994; Goldenbeld & Hatakka, 1999; Gregersen, 1995, 1996a; Gregersen & Bjurulf,

1995; Hatakka et al., 1999; Horneman, 1993; Lonero, 1999; Lynam, 1996; Lynam &

Twisk, 1995; Mayhew & Simpson, 1996; McKenna & Crick, 1994; Nyberg & Engstrom,

1999; Peräaho, Keskinen, & Hatakka, 2003; Senserrick & Swinburne, 2001; Twisk, 1995) but evaluations of the effectiveness of these programs have produced mixed results. Nyberg and Engstrom (1999) suggest that whilst training yields positive attitudinal changes in aspects such seat belt use, there was no change in attitude to factors such as vehicle in front, speed or road conditions compared to control with no training. The training administered by

Goldenbeld and Hatakka (1999) also had limited success and suggests that overall, participants felt they had better control of the vehicle after training than was objectively measured; an overestimation of skill.

Whilst the effectiveness of the aforementioned training programs is unclear, others have produced more promising results. A Finnish program introduced into a novice drivers probationary phase consisted of a four-hour training session with an instructor and dealt with risk avoidance and speed control, six to 48 months after licensure (Keskinen, Hatakka,

Katila, Laapotti, & Peraaho, 1999). There were no significant reductions in crashes one year post-training but there were significant reductions in crashes on slippery roads and night-time driving four years post-training (Senserrick & Haworth, 2005). Whilst the results coincided with an overall reduction in crash rates in Finland at the time, Keskinen et al. (1999) suggest that this new approach facilitated a reduction in novice driver crashes.

A driver training program, which used insight training principles, was also evaluated in Australia (Senserrick & Swinburne, 2001). The one-day program involved classroom theory as well as off-road practical training. Evaluations based on pre and post- training as a well as a 10-12 week follow up suggested a positive shift in attitudes and self- reported behaviours (Senserrick & Haworth, 2005).

These results suggest that whilst training and similar methods attempt to address the cognitive skills deficiencies in current licensing systems, its effectiveness is unclear and requires much further research development before application. Specifically, whilst these studies show some changes in attitudes and motivations, which can improve decision- making, these do not necessarily imply long-lasting changes in driver behaviour.

Another method to develop the cognitive skills of driving is through the use of PC- based training programs. These types of programs aim to provide the same type of cognitive skill training as on-road programs but with the addition of a number of benefits.

Firstly, given that they are presented in a virtual environment, they allow for a greater exposure to higher risk hazards whilst at the same time reducing the actual risk of harm

(Petzoldt et al., 2011; Wallace, Haworth, Regan, 2005). Secondly, virtual environments allow for training to be more varied and diverse. Specifically, the ability to control when a scenario commences, when it ends, as well the ability to pause them not only allows from training to be targeted towards improving specific skills but also allows for these programs to be supported by providing real-time feedback and allowing for self-reflection (Petzoldt et al., 2011; Wallace et al., 2005). Lastly, these types of programs are much more effective in terms of cost of operation as well as time (Petzoldt et al., 2011; 2013; Wallace et al., 2005).

A recent evaluation of a PC-based training program by Petzoldt et al. (2011), whereby participants completed a video clips of traffic scenes and were intermittently asked questions about potential hazards, found that more experienced drivers achieved higher scores on the training tasks than novice drivers, suggesting that their experience aided in better scores. This was taken as support for and validation of cognitive training through driver education methods.

Two major training products developed and validated as training programs to develop a range of higher-order cognitive skills were the Driver ZED (Fisher et al., 2002) and DriveSmart program (Regan, Triggs, & Godley, 2000). The Driver ZED program

(developed by the AAA Foundation for Traffic safety) was designed to train risk perception skills in young novice drivers using PC driving techniques. The training consisted of presenting simulated drive scenario clips on a fixed-base driving simulator and asking participants to complete one of four training modes. In the „scan‟ mode, at the end of a scenario participants were asked to answer a question about the drive to assess if they were paying attention (e.g. „Was there a vehicle approaching in the rear view mirror?‟). In the

„spot‟ mode, the scenario would stop on the last frame and participants were asked to click each risky element in the scene. In the „act‟ mode, the participants were asked mid-way through the scenario what action they would take (e.g. would you speed up or slow down at the coming intersection?). In the „drive‟ mode, the driver needed to click the scene when they would take action to potentially avoid a collision (Fisher et al., 2002). These modes are likely to train drivers on how to better operationalise attention by drawing from working memory systems to facilitate decision-making processes. The efficacy of the program was

evaluated on young novice drivers that received the training compared to novices and experienced drivers that had not received the training and was evaluated in a simulated environment approximately one week post-training. The results revealed that young novice drivers who received the training correctly slowed down more frequently (and less abruptly), drove slower when approaching signed potential hazards (e.g. pedestrian crossings) and let up off the accelerator quicker compared to untrained young drivers. In some cases, they even performed comparably to experienced drivers (Fisher et al., 2002).

The DriveSmart program was a more comprehensive program in that it aimed to train four cognitive skills which were identified as key factors in reducing crash involvement: hazard perception, attentional control, time sharing and calibration. The efficacy of the DriveSmart program was assessed immediately after, and four weeks after training. At both time points, trainees showed significantly better hazard perception and attentional control compared to control groups. In addition, drivers‟ self-assessment of their own behaviour suggested that the training did not inflate confidence in their driving ability

(Regan et al., 2000; Senserrick & Haworth, 2005).Whilst the effectiveness of this training is noteworthy in its own right, its foundations took an interesting „back to basics‟ approach focusing on the core cognitive systems which develop driving skills. Regan et al. (2000) developed components of DriveSmart based on the work of Gopher (1992) and a training method known as Variable Priority (VP) training. VP training at its core was designed to focus on training the allocation of attentional resources towards tasks in a specified manner.

In one experiment conducted by Gopher (1992), participants were asked to perform both a visual tracking task as well as a letter typing task simultaneously and were asked to

split their attention towards the two tasks in varying proportions (across the 50-three minute trials). On different trials, they were asked to allocate attentional resources in differing proportion towards one task (e.g. 25%, 35%, 50%, 65% and 75%) and the remaining proportion to the other. Groups were given on-line feedback about the desired level of performance and how they actually performed. Two control groups were also used to compare VP groups‟ performance. Both control groups were instructed to perform equally on both tasks (i.e. 50/50), but one group was given on-line feedback whilst the other was not. The results indicated that all three groups improved on performance over time in the

50/50 allocation condition but the VP trained group showed much steeper rates of improvement over time and overall performance was better on both tasks compared to the controls. A training similar to VP training was also assessed by Kramer, Larish and Strayer

(1995), who specifically looked at performance differences between younger (18-29 years) and older adults (60-74 years). Interestingly, both groups showed the same performance improvements over time but there was no age related difference between groups.

From an applied perspective, research has also demonstrated that VP training has significant transfer effects to similar tasks (Gopher, Weil, & Bareket, 1994; Hart & Battiste,

1992; Shebilske, Regan, Arthur, & Jordon, 1992). Specifically, Gopher et al. (1994) demonstrated that providing learner pilots with 10 hours worth of VP training on a „space fortress‟ game led to significant increases in instructor ratings of pilots‟ flight performance

(compared to pilots that did not receive training).

Applying the research, Regan and colleagues (Regan, Deery, & Triggs, 1998;

Regan et al., 1998; Shebilske et al., 1992) demonstrated the effectiveness of this type of

training on young novice drivers. Specifically, Regan et al. (1998) assessed the effects of

VP training (compared to control; no training) in drivers of varying experience. They assessed performance on a 2 km driving simulation where participants were asked to 1) manage their speed according to the signs, 2) complete a numeric calculation task, and 3) complete a comprehension task where they were required to answer true/false questions about its content. The results found that VP training was effective in enhancing attentional control in novice drivers and enhanced their ability to “detect, perceive and respond to potential traffic hazards” (Regan, Deery, & Triggs, 1998, p. 1456).

Given these results, it seems intuitive that further research should be dedicated towards investigating how the cognitive mechanisms behind VP training facilitate performance enhancements in driver cognition.

4.9 Summary

This chapter presented an overview of the current state of knowledge of human cognition in an attempt to understand how these systems apply to driving. It is evident that driver cognition is a complex interplay of attentional, memory and decision-making systems, which work as a collective to produce appropriate responses and behaviours. The chapter also highlights where potential problems may occur when drivers become distracted or when cognitive resources are poorly allocated or managed. Based on this understanding, the natural progression of research would aim to begin to investigate how to both maximise the potential of each of these systems in isolation, in addition to their collaboration and interconnectivity which is vital to execute all cognitive tasks. The training research 80

presented above demonstrates that there is value in further investigating the training of fundamental cognitive skills and allowing these skill and performance enhancements to transfer to other domains, specifically in road safety.

Chapter 5: Experiment 1 – Episodic Training to Reduce Risk Behaviours in Young

Novice Drivers

Based on the literature reviewed, it seems that current driver training strategies, such as the

GDL or supplementary training programs, do not adequately address the young novice driver problem. This is further complicated by young novice drivers‟ engagement in risk behaviours, both erroneous behaviours and through violations (Catchpole, 2005; Hatfield &

Fernandes, 2009; Jonah, 1986; Simon & Corbett, 1996; Williams, 2003) which increase their crash risk.

Much research has gone into evaluating why such programs do not sufficiently facilitate crash risk reductions. One major theme that resonates throughout the literature is there is not enough emphasis in these programs to develop the higher-order skills necessary to drive safely (Catchpole et al., 1994; Christie, 2001; Deery, 1999; Engstrom et al., 2003;

Hatakka et al., 1999; McCartt et al., 2009; McKnight & Resnick, 1993). Specifically, a number of key factors are identified that may address the failures of the current systems.

With all training, it seems that efforts should be made to reduce overconfidence of one‟s skill following training in terms of one‟s self-assessed driving ability (Brown, 1982). This could be addressed by training that provides an accurate appraisal of their abilities and their limitations and aims to align subjective risk of a situation with the objective risk (Brown &

Copeman, 1975; DeJoy, 1992; Watts & Quimby, 1980).

In the aviation domain, Molesworth and colleagues (Molesworth et al., 2011;

Molesworth et al., 2003, 2006) demonstrated that these key failures could be addressed to 82

facilitate better risk management in pilots. Specifically, they suggest that risk behaviours can be reduced through the modification of an individual‟s script knowledge. The present research aims to employ similar methods to Molesworth et al. (2003) to examine the utility of a series of risk management training interventions designed to promote better risk management behaviours in young novice drivers.

Molesworth et al. (2003) assessed risk management based on errors/violations of low flying altitude as their dependent measure of a risk behaviour. Transfering to the road domain, the present research assessed risk behaviours in terms of speeding in a simulated drive. This decision was based on statistics which suggest that speeding is a major contributing factor to road fatalies (ABS, 2007; Transport for NSW, 2010; WHO, 2011).

Training consisted of three types of intervention. Group 1 (Case) received written case examples of other individuals whose speeding behaviour resulted in a crash. Group 2

(Case + Rule) received the same case examples as well as details about the rules violated and the consequences of their behaviours. Whilst not directly analogous, these two groups were deemed to be a similar presentation method to those that received an electronic newsletter and video presentation of case examples in Molesworth et al.‟s (2003) research.

The analogy between them was that all case examples were of the risk behaviours (and consequences) of other individuals.

Group 3 (Episodic) involved a simulated drive where, at the conclusion individuals were provided with personalised feedback about the speeding-relevant rules violated in the simulator and the potential consequences of their behaviour. This group was analgous to the simulated flight group in Molesworth et al.‟s (2003) research.

Based on Molesworth et al. (2003), groups 1 and 2 would be expected to reduce their speeding based on inferences made about consquences when others engage in risk behaviours. These consequences are either implied (group 1) or explicitly mentioned (group

2). Group 3 would be expected to reduce their speeding as they recognise the risk behaviour that they themselves have exhibited which highlights the potential disparity between their subjective assessment of their risk behaviour with the potential objective risk of the situation. Further it provides individuals with an accurate appraisal of their level of skill and discourages overconfidence.

Similar to Molesworth et al.‟s (2003) research, it is hypothesised that when compared to control (group 4) who receive a task unrelated to driving, individuals who recieved personalised feedback regarding their behaviour (group 3) will engage in significantly less speeding than groups that received case examples (group 1) or case examples in conjunction with rules violated and consequences (group 2). Given that these behavioural changes are assumed to be as a result of changes in scripts, it is hypothesised that behavioural changes will also be accompaied by attutidinal changes as well as risk and hazard estimates.

5.1 Participants

Fifty-eight participants (24 female) were recruited for the research and randomly allocated to one of four groups. Nine had their learner‟s permit, four had their provisional 1 licence,

30 had their provisional 2 licence, ten had their full license and five had an international driving permit allowing them to drive on NSW roads. All participants had normal or 84

corrected to normal vision. Participants were primarily recruited from the Bachelor of

Aviation degrees at the University of New South Wales. This method involved addressing students during formal classroom sessions, briefly informing them about the research and obtaining contact details of those students interested in participating. A supplementary recruitment method involved the use of the social networking website, Facebook, to advertise the experiment. This method involved creating a „group‟, which any Facebook member could join to participate, but was largely directed at the primary researcher‟s friends list.

Participants were required to be between the ages of 18-25 (inclusive; M = 21.17,

SD = 1.73) and had an average of 50.09 hours of supervised driving experience (SD =

24.74) and 29.21 hours of unsupervised driving (SD = 22.25). Participants drove, on average, 7.31 hours a week (SD = 7.46). Participants were neither reimbursed nor received course credit for their contribution. The research was approved in advance by the Human

Research Ethics Advisory (HREA) Panel of UNSW.

5.2 Design

A one-way factorial design was implemented, which consisted of four training interventions: case-based training (group 1), case + rule-based training (group 2), episodic drive + feedback training (group 3) and control (group 4). Each of the training groups were designed as isolated programs and variables were not systematically manipulated between groups. As such, the experiment was purposely designed to only meaningfully compare each of the training programs to control. In addition, the use of contrasts analysis opposed 85

to an ANOVA maximises power to detect an effect (Tabachnick & Fidell, 2007). The experiment was conducted in a simulated environment over two weeks; week 1 as a training session and week 2 test session. The main dependent variable used to assess performance was percentage of speed exceedance as a function of distance and frequency of zone violations during test. These variables were selected as being the most directly relevant measures of speeding behaviour. Secondary to this, changes in attitude were also assessed using scores on questionnaires and scales (from training to test sessions). Table 1 shows the intervention which each group received; the components of which are described in further detail below (see 5.3 Materials and Apparatus).

Table 1

Summary of Design of Experiment 1

Training session (Week 1)

Group 1 Questionnaires Case + Multiple Choice

Group 2 Questionnaires Case + Multiple Choice  Rules

Group 3 Questionnaires 10.5 km Drive + Personalised Feedback

Group 4 Questionnaires Control (Card Sorting Task)

Test session (Week 2)

Group 1

Group 2 Post-Drive 5 km Practice Drive 21 km Test Drive Group 3 Questionnaire

Group 4

5.3 Materials and Apparatus

5.3.1 Hardware.

All tasks were performed on a HP Compaq (dc7800) PC, running Microsoft Windows XP

Professional with Service Pack 3. Hardware of the PC included an Intel Core 2 Duo

(E8400) 3.00 GHz processor, 4GB of RAM, a 512MB Nvidia GeForce 7300 LE graphics card as well as a SoundMAX HD Integrated sound card. The experiment room was divided by a partition to separate the researcher‟s side from the participant‟s side. On the researcher‟s side was the aforementioned PC, a Dell 19-inch flat panel LCD monitor

(E197FPf) and a standard QWERTY keyboard and optical mouse. The participant‟s side

(see Figure 5) consisted of an Acer 27-inch wide screen LCD monitor (B273HU), a

Logitech G25 Racing Wheel set (E-UP15; the steering wheel, shifter module, accelerator, and brake; clutch pedal was removed) and a Dell 2.1 sound system (Zylux Acoustics

Corporation; Model A525). The participant‟s side also had a driver-side car seat out of a

2002 Mazda 626 which did not include a seatbelt.

Figure 5. The participant's side of the driving simulator

As per the requirements of the driving simulation software, a dual-view setup of the monitors was employed. Both monitors were connected to the aforementioned graphics card, which had one DVI output and VGA output. As defined by the software, the Dell monitor was connected through the VGA port and was set as the primary display and the

Acer monitor was connected through the DVI port and was set as the secondary display.

This setup was critical to ensure that the simulation was presented to the participants‟ side and that the summary page was present to the researchers‟ side. The resolution of the both displays was at 1280 x 1024 at a refresh rate of 60Hz.

5.3.2 Software.

All tasks throughout the experiment, other than driving simulations were presented through

Inquisit 3 (Build 3.0.3.2) by Millisecond Software. Fonts throughout the experiment were black in colour and present on a neutral grey background. All driving simulations throughout the experiment were run through STISIM Drive™ driving simulator (Build

2.08.04) by Systems Technology Inc.

Simulated Driving Vehicle

As seen in Figure 6, the driving simulator was a low-fidelity simulator with a basic instrument cluster which consisted of an analog speedometer and tachometer, as well as a digital odometer (to one decimal place). The bonnet of the vehicle was blue in colour and there was also a rear-view mirror displayed.

Driving Simulation Tracks

In total, there were three tracks designed for the experiment: a 5 km practice drive,

10.5 km training drive and a 21 km test drive.

5 km Practice Track (Appendix A for STISIM Script)

The 5 km practice drive was designed for participants to familiarise themselves with the driving simulation in terms of the physical manipulation of the vehicle, the visual scene and the lack of inertial feedback experienced travelling at differing speed zone types, namely 40 km/h, 60 km/h and 80 km/h.

The drive contained four curves (indicated by a posted „curve‟ sign as seen on NSW roads) which were between –90 and 90 degrees each, two right hand turns (R) and two left hand turns (L), presented as L- R- L- R approximately 900 m apart. There were six speed zones in total (indicated by a posted speed sign as seen on NSW roads), one 40 km/h, three

60 km/h and two 80 km/h. There were also vehicles driving in the opposite direction as well as parked cars on both sides of the road. A pseudo random (unpredictable to the individual participant but consistent between participants) allocation of buildings and trees was also present.

Figure 6: Image of the StiSim simulated environment.

10.5 km Training Track (Appendix B for STISIM Script)

Participants allocated to the episodic drive condition (group 3) took part in the 10.5 km training drive. Participants were instructed that their task was to deliver packages to businesses and residents along a predefined route as quickly as possible without breaking

NSW road rules. Speed zones throughout the course were defined to fit the objective of training task and, as such, buildings and/or houses were only present at delivery zones.

There were a total of five delivery zones and these were the only sections of the drive which were 40 km/h zones. This use of a 40 km/h zone is consistent with New South Wales speed zoning guidelines (Road and Traffic Authority, 2011), which allows for 40 km/h zones where there is high pedestrian activity (i.e. outside schools). There were four 60 km/h zones and four 80 km/h zones which were pseudo randomly placed between delivery zones.

The purpose of having different speed zones was largely to ensure that participants were engaged with their speed management task. In addition, it allowed for the assessment of differences in speeding behaviour dependent on speed zone. The rationale for this is based on the notion that individuals exhibit different speeding behaviour and have different attitudes towards speeding depending on speed zones (Archer, Fotheringham, Symmons &

Corben, 2008; Department of Infrastructure and Transport, 2011).

The drive consisted of four curves to break up the straight road. In addition to simple L and R turns, an „S‟ curve was used which was sequenced as either left then right hand turns or vice versa (LR and RL respectively). The curves were presented as R- RL - L-

RL and were spaced approximately 2 km apart (four in total).

To increase ecological validity, between delivery points, trees were randomly selected and placed (by STISIM) to increase fidelity, with a maximum of 200 trees displayed at any one time. Other motorists were pseudo randomly presented as oncoming traffic (unpredictable to the individual participant but consistent between participants). No motorists were travelling in the same direction (to ensure that choice of speed was not influenced by other drivers) and no parked vehicles were present during the drive.

21 km Test Track (Appendix C for STISIM Script)

All participants drove the same test track in the second week of the experiment. The test drive task was twice the distance of the episodic drive condition that group 3 completed in week 1. The drive consisted of 26 speed zones; 10 were 40 km/h zones, 10 were 60 km/h zones and six were 80 km/h zones. There were eight curves in the drive which were presented in the sequence R- LR- L- RL- R- LR- L- RL. There were 10 delivery points with a similar configuration of buildings to the 10.5 km test drive. Variables such as tree distribution and density, as well as other motorist and parked vehicles, were identical to the

10.5 km drive.

5.3.3 Cases Examples and Consequences (see Appendix D).

Three fictional cases (presented through Inquisit 3) were used as case examples of motor vehicle crashes in which speeding was an identified causal factor. These cases were presented to the case-based training condition (group 1) and the case + rule-based training condition (group 2). All cases named a driver of unspecified age, travelling on a named 92

Sydney road in exceedance of the specified speed limit. The cases also described in detail how the crash occurred and the nature of the injuries (and, in one case, fatality) either to the driver, a passenger or a pedestrian. The cases were presented in great detail to evoke strong imagery, which was aimed to parallel the way information and content is presented typically in anti-speeding campaigns. All participants who read the cases received four multiple choice questions immediately after reading each case, to ensure participants read the case examples thoroughly.

In addition to the three case examples, participants in group 2 received information

(under NSW Legislation as of July 2009) about specified speed limit, the speed exceedance over the speed limit, the fine they would have received given their licence type and demerit point penalty, as well as the maximum penalty if the matter was to be heard in court.

5.3.4 Wisconsin Card Sorting Test.

The Wisconsin Card Sorting Test is a PC-based task which was designed as a way to measure flexibility in thinking (Berg, 1948). Participants in the control group (group 4) were asked to click one of four card options on the screen, based on a learned rule, which changed dependent on the participants‟ responses. The task was administered through

Inquisit 3. The task was of no theoretical relevance to the research but was employed as a control task to occupy a comparable amount of experimental time.

5.3.5 Questionnaires and Scales.

The following questionnaires were presented to participants, in this order, through Inquisit

3. The rationale for using these questionnaires and scales was based on previous research

which has demonstrated significant relationships between risk attributes and driver performance (Hatfield et al., 2008; Lawton, Parker, Manstead, & Stradling, 1997; Owsley,

McGwin & McNeal, 2003). These questionnaires and scales served to assess whether all groups were equal on critical variables prior to training.

 Crash History and Infringements (Hatfield et al., 2008)

 Perceived Risk of Being Caught for Speeding (Hatfield et al., 2008)

 Semantic Differential Item Measures of Speeding-Related Attitudes (Greenwald,

McGhee, & Schartz, 1998; Hatfield et al., 2008)

 Driver Behaviour Questionnaire (DBQ; Lawton, Parker, Manstead, & Stradling,

1997).

 Montag & Comrey Driving Internality and Externality Scales (Montag & Comrey,

1987)

 Zuckerman‟s Sensation Seeking Scale Form V (SSS-V; Zuckerman, Eysenck, &

Eysenck, 1978).

To ensure all questions were answered, Inquisit would not allow participants to proceed to the next page unless all questions were answered. The only exception to this rule was with rating scale questions, where the default marker placement could also be taken to be the answer.

General Demographics

Participants responded to questions regarding their gender, age, type of driver‟s licence, number of hours driving experience with and without supervision, average number

of hours spent driving per week, and total number of hours driving.

Crash History and Infringements (Hatfield, et al., 2008)

Participants indicated if they had been involved any motor vehicle crashes as a driver, irrespective of fault, as well as if they had ever received an infringement notice for speeding.

Perceived Risk of Being Caught for Speeding (Hatfield et al., 2008)

Participants rated their chances of being detected when exceeding the speed limit by

“police on the side of the road with a radar”, “an automatic fixed speed camera” and

“police in a moving patrol vehicle with a radar”. Responses were made on a 7-point scale ranging from “Extremely Likely (1)” to “Extremely Unlikely (7)” with a “50/50 (4)” response in the middle and unlabelled markers at points (2), (3), (5) and (6). Higher scores reflected a lower perceived risk of being caught for speeding.

Semantic Differential Item Measures of Speeding-Related Attitudes (Greenwald, McGhee,

& Schartz, 1998; Hatfield et al., 2008).

Participants completed a set of five semantic differential item scales regarding speeding as used by Hatfield et al. (2008). These 7-point scales were anchored at each end by polar-opposite adjective pairs: “beautiful–ugly”, “good–bad”, “pleasant–unpleasant”,

“honest–dishonest”, and “nice–awful”, with five numerically labelled markers for intermediate scores. On each scale, participants indicated one of the seven responses where

positive attitudes towards speeding were negatively weighted (–3, –2 & –1) and negative scores towards speeding were positively weighted (1, 2 & 3). Participants also completed the same five semantic differential items for „safe driving‟ which were scored in the same manner as the speeding items. Scores of each participant were averaged across the five items (separately for speeding and safe driving). Overall positive scores on this measure indicated a more negative attitude towards speeding/safe driving and the opposite for negative values.

Driver Behaviour Questionnaire (DBQ; Lawton, Parker, Manstead, & Stradling, 1997).

The extended Driver Behaviour Questionnaire contained 27 items which were divided in four categories: lapses in attention, errors, as well as aggressive and „ordinary‟ violations items. This is a widely used scale and has been validated over many studies

(Parker, Reason, Manstead, & Stradling, 1995; Reason, Manstead, Stradling, Baxter, &

Campbell, 1990). Participants indicated the frequency at which they exhibited each behaviour whilst driving on a six-point scale: “Never” (1), “Hardly Ever” (2),

“Occasionally” (3), “Quite Often” (4), “Frequently” (5), “Always” (6). Higher scores on this measure indicated more frequent aberrant driving behaviour.

Montag & Comrey Driving Internality and Externality Scales (Montag & Comrey, 1987)

This 30-item scale consisted of 15 driving internality (DI) questions and 15 driving externality (DE) questions. Participants were asked to express their degree of agreement or disagreement on the following rating scale: “Disagree very much” (-3), “Disagree quite a

bit” (-2), “Disagree Some” (-1), “Agree a Little” (1), “Agree Quite a Bit” (2), “Agree Very

Much” (3). Scores on the externality items and internality items were averaged and then subtracted from one another (DE – DI) to produce on overall score on this measure. If the scores were positive, it indicated a more internal attribution as the causes of crashes, and the opposite was true if negative.

Zuckerman‟s Sensation Seeking Scale Form V (SSS-V) (Zuckerman, Eysenck, & Eysenck,

1978).

This scale consisted of the original 40 forced-choice items which assessed four factors: thrill and adventure seeking (TAS), experience seeking (ES), disinhibition (DIS) and boredom susceptibility (BS). Higher scores on this measure indicated greater desire for sensation.

5.3.6 Post-Drive Questionnaires.

The post-drive questionnaires were administered after the second week drive. Two of the same scales used in week 1 – „Perceived risk of being caught for speeding‟ (Hatfield et al.,

2008) and „Semantic differential item measures of speeding-related attitudes‟ (Greenwald et al., 1998; Hatfield et al., 2008) – were administered again to assess whether training in week one had produced change in perceptions and attitudes.

In addition, participants were asked whether they thought they had exceeded the speed limit in the simulation (Yes/No response) in week 2, and if so, how many times. The question was designed to assess how many times they were aware of their speeding behaviour (i.e. intentional violations or recognition of error). Participants were also asked 97

whether they recalled participating in a similar task to the simulated task they had just completed (Yes/No response). This question was directed to assess whether participants, who felt they had been in a similar situation, drove differently to those who had not. As a secondary purpose, it was also implemented to assess whether participants in the episodic drive condition were aware of the similarities between the tasks over the two weeks. The last two questions explicitly asked participants to subjectively assess the effectiveness of the training in week 1 in reducing their speeding behaviour, both in this simulation in week

2 as well as in real driving situations since their training in the previous week. Responses were made on a 7-point scale ranging from “Not at all Effective (1)” to “Extremely

Effective (7)” with five unlabelled markers at points (2), (3), (4), (5) and (6), with higher scores indicating greater perceived effectiveness.

5.4 Procedure

Participants initially read an information statement (Appendix E) regarding the nature of the experiment. The statement outlined that the experiment was to be conducted in two one- hour sessions spaced approximately one week apart. The description of the experiment outlined that in the first week they would complete a series of questionnaires, and may participate in a driving simulation. In the second week, they would complete a driving simulation as well as a post -drive questionnaire. The potential risks were also expressed as psychological distress from operating the simulator as well as physical symptoms such as motion sickness, dizziness, fatigue, and/or nausea. The limited direct personal benefits of their participation, as well as measures employed to ensure confidentiality and anonymity, 98

was also explained. Once both consent forms were signed, participants were randomly allocated to one of four conditions and commenced the experiment with a battery of questionnaires. Participants were informed prior to commencement that it was important to answer honestly and from their own point of view, and were also instructed to ask for clarification if they did not understand any sections.

Following the questionnaires, participants underwent one of the four training conditions in the first week. Participants in group 1 and 2 were informed that they would read a number of case examples and would answer a series of multiple choice questions following each case. The case-based training (group 1) received their three case examples.

After reading the cases, they completed four multiple choice questions. In the case + rule- based training (group 2), participants were given the same three cases with the addition of factual information (after the multiple choice section) about the road rules.

The episodic drive condition (group 3) did not participate in a case reading task.

Instead, they completed the 10.5 km parcel delivery drive on the driving simulator.

Participants were given a mock scenario where they were new employees of a newspaper company (see Appendix F for Participant Script). Their task was to drive a delivery vehicle to deliver newspapers to businesses and residents as quickly as possible without violating

NSW road rules. Importantly, it was not highlighted that speeding behaviour was the primary focus of the task. The drive took approximately 10 minutes. Finally, to ensure that participants did not experience any adverse symptoms as a result of the simulator, all participants were asked whether they felt any physical or psychological symptoms at the conclusion of the drive; no participants reported any such symptoms.

Following the drive, participants were given immediate personalised feedback about their drive by the researcher (see Appendix F under „Feedback‟). Information included total delivery time, number of speed exceedance, maximum speed exceedance, licence-specific fine for speeding, licence-specific demerit point penalties for speeding, as well as information about the potential harm they could have inflicted to themselves, their passengers and pedestrians as a result of their speeding behaviour.

The fourth group (control) completed a PC-based version of the Wisconsin Card

Sorting Test (Berg, 1948) presented through Inquisit 3.

Approximately one week post-training, all participants returned for the test week.

This week was the same for all participants irrespective of group allocation and was designed to assess the effects of the training in week 1. All participants initially completed the 5 km practice drive on the simulator. This was to reduce the novelty of the driving simulator to those who had not experienced it (groups 1, 2 and 4) and to ensure that the episodic drive condition (group 3) did not have an advantage in terms of their ability to physically handle the vehicle. Once this drive was complete, all participants were asked whether they felt any psychological or physical symptoms as a result of driving the simulator. If any symptoms were reported, the experiment was terminated. No symptoms were reported, so all participants went on to commence the 21 km test on the simulator. The test drive involved the same delivery task experienced by the episodic drive condition

(group 3) in week 1. Once the drive was complete, all participants were again asked if they felt any psychological or physical symptoms. Finally, all participants completed a post- drive questionnaire through Inquisit 3.

100

5.5 Results

5.5.1 Questionnaires and Scales.

Recall the primary purpose of the questionnaires and scales was to ensure no differences between groups. Hence, a series of one-way ANOVAs were conducted on all pre-training ratio and interval data with α set at .05. A repeated measures analysis was not appropriate for these variables given that these scales and questionnaires was only measured once (pre- training). This included age, perceived risk of being caught speeding, number of hours driving with and without supervision, average number of hours spent driving a week, semantic differential items for both speeding and safe driving, DBQ, internality-externality scale as well as Zuckerman‟s SSS-V. No statistically significant differences in any of the variables were present across groups, prior to training; largest F, F(1, 54) = .70, p = .56.

A series of Chi Square tests were also conducted to analyse the non-parametric variables; gender, licence, infringement and crash, with α set at .05. No statistically significant differences in any of the variables were present across groups; largest Χ2, Χ2 (3,

N = 58) = 3.37, p = .36. Combined, these results suggest that the groups did not differ prior to training.

5.5.2 Group 1 and Group 2 Multiple Choice Questions.

Each participant was assessed individually and was deemed to have read the cases sufficiently if they answered above chance in two or more of the cases. The average participant in the case and case + rule groups answered 93.33% and 91.67% (respectively) of questions correctly. Further analysis failed to reveal a statistically significant difference

101

between group 1 and group 2 in terms of competency, t(27) = 3.66, p = .72. This suggests that both groups were equally able to retain and recall similar knowledge about the case examples.

5.5.3 5 km Practice Drive.

The 5 km practice drive in week 2 was designed to assess whether the groups differed in terms of handling characteristics of the driving simulator and to provide all participants with exposure to the simulator prior to the test task. Three measures were used to assess their ability to drive the vehicle: number of collisions, number of centreline crossings and number of road excursions. These variables were chosen because they were deemed to be most reflective of an individual‟s ability to physically manipulate the vehicle. A one-way

ANOVA with Bonferroni adjusted α set at .05/3 = .017 failed to reveal a statistically significant difference between groups based on these three variables largest F, F(1, 54) =

1.05, p = .39. These results indicate that the groups were the same in terms of their ability to drive the vehicle and thus any difference in driving behaviour was deemed to be attributed to the training.

5.5.4 21 km Test Drive.

The 21 km test drive was the main task used to assess the effects of training from the first week. Two dependent variables were assessed: percentage of distance speeding and frequency of zone violations.

102

Percentage of Distance Speeding (see Appendix G for PSY Output Files)

This variable was calculated by STISIM and was presented as a value on each participant‟s output data file (.plb extension; Over speed limit [% Distance]). Percentage over the speed limit was also presented as a function of time; however, this value was not used, as time to complete the track varied across participants, whereas the total distance did not. Values were also manually extracted from the STISIM data file (viewed through

Inquisit 3) to validate the summated output provided by STISIM and were recorded on an

Excel spread sheet. Analysis of the data was conducted through the statistical package PSY developed by Bird (2004). The program allowed for the analysis of the percentage of speeding using a series of Bonferroni adjusted planned comparisons. The comparisons of interest were limited to the three that directly addressed the research questions about the efficacy of each type of training – each of the training groups compared to control. Figure 7 illustrates the total percentage of speeding across all speed zones for each of the four groups. With α set at .017 (Bonferroni adjusted .05/3), result failed to reveal a statistically significant differences between neither group 1 nor group 2 compared to control (group 4;

F(1, 54) = .02, p = .89 and F(1, 54) = .20, p = .66 respectively). However, group 3 demonstrated a lower overall percentage of speeding compared to control, and this result was statistically significant (group 4; F(1, 54) = 8.68, p = .005). These results suggests that a training program involving a driving episode and feedback is most effective in reducing young novice driver speeding behaviour of the three types of training assessed.

103

30 25 20

15 * 10 Total Percentageof Speeding Speeding (by distance) 5 0 Group 1 Group 2 Group 3 Group 4

Figure 7. Total percentage of speeding (by distance) in all speed zones distributed across group.

Since the test drive contained various zones, further analysis aimed to investigate whether the overall results were repeated in each zones. Figure 8 illustrates the percentage of speeding per zone distributed across group. Results failed to reveal any statistically significant reduction in speeding in any of three speed zones when group 1 or group 2 compared to control (largest F, F(1, 54) = 2.57, p = .12). Results however revealed a statistically significant reduction in speeding in the 40 km/h and 60 km/h zones, when group 3 was compared to control (group 4; F(1, 54) = 6.87, p = .01 and F(1, 54) = 6.27, p =

.02) respectively, but not in the 80 km/h zone, although results were approaching statistical significance (F(1, 54) = 3.34, p = .07). The results indicate that the lack of effectiveness of the group 1 and 2 training was not dependent on speed zone, compared to control. The preliminary interpretation of the results for group 3 seems to indicate a difference in the percentage of speeding was dependent on speed zone type. That is, the training may be more effective in lower speed zones than higher speed zones.

104

20 40km/h

15 60km/h 80km/h

10 (by distance) (by *

PercentageSpeeding of 5 *

0 Group 1 Group 2 Group 3 Group 4

Figure 8. Percentage of speeding (by distance) per zone distributed across group. The sum of the percentages across the 3 zones equals to the overall speeding percentage.

Number of Zone Violations (see Appendix H for PSY Output Files)

The data was initially extracted from each participant‟s raw STISIM data file (.plb extension). Whilst STISIM had calculated scores for the number of violations a participant committed, the value was grossly misrepresentative. This was because STISIM categorised violations without regard for the duration or severity of the speed exceedance. So a participant who exceeded the speed limit by 0.01 km/h for 0.3 seconds was recorded the same as a participant who exceeded the speed limit continuously by 100 km/h for 3 minutes.

To account for this, violations were defined by the speed zones. Of the 26 speed zones (ten 40 km/h zones, ten 60 km/h zones and six 80 km/h zones) each participant received a dichotomised „yes‟ or „no‟ per speed zone, if they exceeded the speed limit at all or not. The number of „yes‟ responses were then summed per speed zone type so that there

105

were four final scores for each participant: one overall (the sum of 26 zones) and one per speed zone type.

A series of Bonferroni planned comparisons were conducted through PSY, comparing the three training groups to control, both overall and per speed zone type. Figure

9 illustrates the total frequency of zone violations across groups. With α set at .017

(Bonferroni adjusted .05/3), group 1 or group 2 did not differ in terms of the overall frequency of zones violated when compared to control (group 4; F(1, 54) = .40, p = .53 and

F(1, 54) = .07, p = .80 respectively). However, there was a statistically significant difference when group 3 was compared to control (group 4; F(1, 54) = 10.77, p = .002).

Consistent with the percentage of speeding data, the results indicated that group 3 training was the most effective method in reducing speeding behaviour in young drivers, compared to control.

20 * Overall 15

5 Number Number Zone of Violations

0 Group 1 Group 2 Group 3 Group 4

Figure 9. Overall frequency of zone violations distributed across group.

106

Further analysis was performed to assess whether the overall effect was zone specific. Figure 10 illustrates no reduction in the frequency of zone violations in any speed zones for group 1 and group 2, compared to control (group 4); largest F, F(1, 54) = 1.11, p

= .30. However, the results revealed that group 3 training reduced the frequency of zone violations in all three speed zone types – 40 km/h, 60 km/h and 80 km/h zones – compared to control (group 4), and this result was statistically significant; smallest F(1, 54) = 6.57, p

= .01. These results suggest that compared to control, group 3 committed zone violations significantly less frequently. Groups 1 and 2 did not show any difference compared to control.

8 * * 40km/h 6 60km/h

4 * 80km/h

2 Number Number Zone of Violations

0 Group 1 Group 2 Group 3 Group 4

Figure 10. Frequency of violations per zone distributed across group. N.B: scores out of 10 for 40 km/h and 60 km/h zones and out of 6 for 80 km/h zones.

5.5.5 Post-Drive Questionnaire.

Analysis of „perceived risk of being caught for speeding‟ and „semantic differential item measures of speeding-related attitudes‟ were analysed using a mixed repeated measure

107

ANOVA (α set at .05) with two factors; one between subjects factor (group allocation) and one repeated measures factor (pre- and post-training scores). The results of a repeated measures analysis did not show a statistically significant difference between week 1 and week 2 in participants‟ perceived risk of being caught for speeding ( F(1, 54) = .40, p =

.53).

The semantic differential items for speeding demonstrated a main effect for time

(F(1, 54) = 4.89, p = .03) with attitudes to speeding becoming more negative post-training

(i.e. that speeding was bad). There was no group effect (F(3, 54) = 3.63, p = .54) or interaction (F(3, 54) = .51, p = .68). That is, all groups changed their attitude to speeding equally (see Table 2).

The semantic differential items for safe driving showed no overall change in participants‟ attitude to safe driving post-training (F(1, 54) = .24 , p = .63). That is, their attitude towards safe driving was not influenced positively or negatively. There was also no group effect (F(3, 54) = .30, p = .83) or interaction (F(3, 54) = .62, p = .61; see Table 3).

Table 2 Average Score on the „Speed‟ Semantic Differential Questions – Experiment 1 Week 1 SD Week 2 SD

Group 1 -0.093 1.134 0.373 1.077

Group 2 0.414 0.962 0.800 1.223

Group 3 0.280 1.090 0.760 1.093

Group 4 0.400 1.148 0.414 0.850

Note: positive scores on this measure indicated a more negative attitude towards speeding and the opposite for negative values.

108

Table 3

Average Score on the „Safe Driving‟ Semantic Differential Questions – Experiment 1

Week 1 SD Week 2 SD

Group 1 -1.520 0.941 -1.400 0.845

Group 2 -1.429 0.811 -1.686 0.767

Group 3 -1.507 0.623 -1.533 0.683

Group 4 -1.686 0.907 -1.714 0.803

Note: positive scores on this measure indicated a more negative attitude towards safe driving and the opposite for negative values.

A Chi Square analysis was conducted for the dichotomously scored question „Do you recall exceeding the speed limit‟ post-drive. The results failed to reveal any statistical difference between groups; Χ2 (3, N = 45) = .78, p = .86 (see Table 4). A univariate ANOVA was conducted to assess „How many times do you recall exceeding the speed limit‟. This too failed to reveal a statistical difference between groups (F(1, 54) = .78 , p = .51; see Table

5). These results suggest that participants in all groups were equally aware (or unaware) that they exceeded the speed limit.

Table 4

Proportion of 'Yes' Responses to “Do you recall exceeding the speed limit”

Group Allocation Percentage 'Yes'

Group 1 86.67%

Group 2 85.71%

Group 3 60.00%

Group 4 78.57%

109

Table 5

Average Number of Times Participants Recalled Exceeding The Speed Limit

Group Allocation Average Number of Exceedances Recalled

Group 1 4.92

Group 2 3.92

Group 3 5.11

Group 4 6.09

Further analysis was conducted to assess the relationship between actual speeding measures

(percentage of speeding and frequency of zone violations variables) compared to perceived speeding measures. Given that all participants exhibited speeding behaviour to varying degrees, a two-tailed Pearson‟s product-moment correlations analysis, with α set at .05, was conducted comparing responses to „Do you recall exceeding the speed limit‟ (dichotomous yes/no response) with percentage of speeding as well as frequency of zone violations. The results failed to reveal a statistical difference in both speeding measures for those that did and did not recall speeding (largest r, r(56) = .14, p = .29). A two-tailed Pearson‟s product- moment correlation was then conducted (α set at .05) between the number of times participants recalled exceeding the speed limit compared to actual speeding measures. The results revealed a moderate, positive correlation between perceived speeding and overall percentage of speeding, r(43) = .35, p = .02. In contrast, the results failed to reveal a statistically significant correlation between perceived speeding and frequency of zone violations, r(43) = .26, p = .09.

110

Given these results and the significant differences between training interventions, further analyses were conducted to assess whether the overall significant relationship between perceived speeding and overall percentage of speeding was significant in all groups. As seen in Table 6, results revealed that there was only a strong positive correlation between perceived speeding and overall percentage of speeding for group 1, r(11) =

.70, p = .008; no other correlations were revealed to be statistically significant for the other groups (largest r, r(9) = .36, p = .27). This result revealed that the statistical significance of the overall relationship between perceived speeding and percentage of speeding was due to the strong correlation of group 1 and thus was group specific.

Table 6

Correlation Between Overall Percentage of Speeding Compared to Number of Times Participants Recalled Exceeding The Speed Limit In Each Group

Group 1 Group 2 Group 3 Group 4

Pearson‟s r .702** .048 .082 .363 Overall % of p – value .008 .882 .834 .273 Speeding N 13 12 9 11

Note: ** Correlation is statistically significant at the 0.01 level (2-tailed).

In order to determine whether individuals recalled being in a similar driving environment, a Chi Square analysis was conducted for the questions „recall being in a similar driving situation, real or simulated‟. Results failed to reveal a statistically significant difference between groups; Χ2 (3, N = 44) = .36, p = .95. This result suggests that all

111

participants, irrespective of group allocation, felt as though they had experienced a similar situation, as identified by themselves and this is confirmed by the proportion of „yes‟ responses, as seen in Table 7.

Table 7

Proportion of 'Yes' Responses to “Have You Been in a Similar Driving Situation”

Group Allocation Percentage 'Yes'

Group 1 66.67%

Group 2 85.71%

Group 3 80.00%

Group 4 71.43%

The final two questions assessed participants‟ perceived effectiveness of the training. A univariate ANOVA with α set at .05 revealed that there was a statistically significant effect between groups in rating the effectiveness in a simulated environment

(F(3, 54) = 16.60, p < .001). A post-hoc analysis revealed that the result of overall perceived effectiveness of training in a simulated environment was attributed to group 3 when compared to control (group 4; p < .05). However, a univariate ANOVA to assess the effectiveness of training in real driving situations failed to reveal a statistically significant difference (F(3, 54) = 2.04, p = .12; see Figure 11).

112

7.00 * Effect in 6.00 Sim Driving 5.00 Effect in 4.00 Real Driving 3.00

Effectiveness Rating 2.00

1.00

0.00 Group 1 Group 2 Group 3 Group 4

Figure 11. Mean ratings of the effectiveness of training in simulated driving and in real driving on a 7 point scale, where 1 indicated „Not at all Effective‟ and 7 indicated „Extremely Effective‟.

5.6. Discussion

The aim of the present research was to examine the utility of a series of training methods to curb speeding behaviour in young novice drivers. The results suggest that providing individuals with case examples of speeding behaviour and their consequences was not effective in reducing speeding behaviour, regardless of whether the cases included information about the road rules violated and their legal ramifications. Conversely, the research suggests that providing individuals with a driving episode plus feedback was effective in reducing speeding behaviour in young drivers, in a simulated environment.

Specifically, when a simulated driving episode is followed by feedback regarding participants‟ speeding behaviour, there was a significant reduction in overall percentage of speeding, and frequency of zone violations, when tested one week later.

113

This training method was drawn from the aviation industry where similar methods have been found to improve pilots‟ risk management behaviour in the form of compliance with safety rules (Molesworth et al., 2011; Molesworth, Tsang, & Kehoe, 2011;

Molesworth et al., 2003, 2006). These results suggest that the effectiveness of this training is not domain specific and in this experiment demonstrates its application in a road context.

According to Bednall and Kehoe (2011) and Molesworth et al. (2011), the success of this type of training is due to the active processing of each „experience‟ that facilitates the creation, addition, correction, and/or refinement of existing knowledge structures known as scripts. The results are consistent with this notion that speeding behaviour in young novice drivers may be in part due to faulty scripts which are used to model future behaviour.

Young novice drivers may develop a „normal driving‟ script which incorporates the notion that „exceeding the speed limit does not have undesirable consequences‟. As a result, speed exceedance would not be considered deviant behaviour in the context of this faulty script. The problem may be further exacerbated if the young novice drivers rationalise their behaviour as superior to the average driver, since they have not experienced any undesirable consequences associated with such behaviour (DeJoy, 1992). This leads to two major problems for modelling future driving behaviour.

The first problem is that the limitation of the „normal driving‟ script in terms of speeding is potentially boundless. Specifically, if the script can be modified based on new experiences, and new experiences test the boundaries of the old scripts (such as greater speed exceedance), then modification of the script opens the metaphorical flood gates as to

114

what constitutes a „normal‟ speed. However, in the event of a negative experience, it is expected that the script will be rewritten or corrected to better align with what is deemed safe practice. This is expected to occur based on the individual‟s reflection and appraisal of the negative experience. According to DeJoy (1992), if the individual does not take personal responsibility for the negative event (i.e. the crash), it is likely that their behaviour

(modelled by their scripts) will remain unchanged.

The second problem, which stems from the first, is that since the speeding behaviour is considered „typical‟ (i.e. normal), there is little or no cognitive processing, as each experience lacks novelty. As a result, without active processing of the experience, there is no modification to existing knowledge structures/scripts as each experience is in align with the script (Bednall & Kehoe, 2011; Molesworth et al., 2011). At worst, this problem is self-propagating, maintaining faulty and unhealthy self-beliefs, simply because there is little information in the script to state otherwise.

The effectiveness of this episodic training is attributed to the way it addresses these two problems. The feedback provided following a simulation highlights to the individual a potential problem with their „normal driving‟ script. Specifically, this post-drive feedback quantifies the objectively deviant behaviour, forcing the individual to take personal responsibility for their behaviour. The novelty of information through the feedback also makes the driving episode a salient and „atypical‟ episode which is distinct from the

„normal driving‟ script, thereby further increasing the likelihood of actively processing the information. When the driver experiences a similar situation in the future, they are now able to draw on the modified script to guide and direct more appropriate behaviour.

115

5.6.1 Non Significant Result on 80 km/h Zone (Percentage of Speeding).

Whilst there was an overall reduction in the percentage of speeding in group 3, post-hoc analysis did not show a statistically significant reduction in 80 km/h zones. While it could be argued this result was approaching statistical significance (p = .07), the fact that it was not could be attributed to relatively large sample variances rather than a failure of the training. Further, evidence to suggest that training was not zone specific is seen where group 3 showed reduction in the frequency of zone violations in all speed zones.

However, community surveys conducted by the Department of Infrastructure and

Transport (2011) examining attitudes towards speeding found that there are attitudinal differences depending on speed zones. Specifically, when participants are asked “how fast should people be allowed to drive without being booked for speeding?” in 60 km/h and 100 km/h zones (Department of Infrastructure and Transport, 2011, p. 39 & p.43 respectively),

85% of participants considered 5 km/h or less to be acceptable in 60 km/h zones but only

55% of participants considered 5 km/h or less to acceptable in the 100 km/h zones

(Department of Infrastructure and Transport, 2011). These surveys suggest that attitudes towards speeding are not absolute, but may be relative to the speed zones.

These two competing explanations clearly warrant future research, which should aim to further assess if there is any speed zone effects from this type of training.

5.6.2 Alternative Explanations.

Whilst the results are consistent with script modification, other interpretations could be put forward. It is possible that the feedback received by group 3 highlighted speeding as the research focus and, as such, increased attention to speeding and/or motivation to not speed. 116

This interpretation could be seen as a demand characteristic resulting from a researcher bias; reducing their speeding behaviours because they felt it is the result the researcher was expecting.

Whilst this interpretation does not support the idea of script modification, it is seen as a positive effect of the training. This interpretation would suggest that the training results in a self-generated recognition of the need to modify speeding behaviour and motivation to act upon it. That is, receiving feedback alone is not sufficient to produce a behavioural change; the individual must acknowledge the relevance of the information and modify their behaviour from self-generated motivations. Future research should aim to distinguish the primary mechanisms for the effectiveness of episodic training.

Another explanation for the results could conclude that the behavioural change was produced by a context-based retrieval of memory since the driving episode group (group 3) participated in an identical task (parcel delivery) and a similar drive, under the same experimental conditions in the week prior. This line of argument could claim that the results were merely a classically conditioned learning effect. That is, the unconditioned response (UR; negative feelings towards speeding for the episodic training condition) may have been elicited by the environmental context and/or the driving scenario (CS - conditioned stimuli). However, this explanation is also considered to be a point in favour of such training. The abstraction of similarities and differences from past experiences (or cases) is integral to the application of appropriate scripts (Alba & Hasher, 1983). The success of the episodic drive training is attributed not only to the individual‟s ability to abstract new information from the first week, but to identify appropriate future scenarios

117

where this information or modified behaviour can be applied. In the present experiment, similarity between training and test was intentional to ensure that feature identification was obvious. It should be noted, however, that whilst some association between the CS and UR may be necessary for appropriate application of scripts, it does not sufficiently account for all the working components of speeding behaviour. That is, speeding behaviour is not merely elicited by one or more CS, but is a complex cognitive interplay of motivations, risk factors, attitudes and environmental factors. Of interest is whether the identification of common features occurs at a conscious level, and, if so, whether feature identification should be incorporated into future training.

5.6.3 Results of the Semantic Differential Items.

Results from the post-drive questionnaire provided insight into attitudes and awareness of speeding and the training methods employed. Of interest is that results of the semantic differential items for speeding participants in each group had an overall more negative attitude post-training. However, no significant changes in attitudes were observed for the safe driving items. The results suggest that the two components, (speeding and safe driving) are separate in their cognitive representations in knowledge structures. That is, a change in attitude to speeding does not necessarily imply a directionally opposite change in attitude to safe driving. This may due to the fact that speeding is not necessarily the only component involved in „unsafe driving‟ and as such was not sufficient to produce a polar change in attitude. If, however, these results are reflective of how attitudes change, the results suggest that training drivers to drive safely and training drivers not to speed may have mutually

118

exclusive components, particularly in cognitive representations. Future research should aim to investigate and further substantiate this possibility.

5.6.4 Perceived Amount of Speeding.

Groups did not differ in terms of the number of times they perceived themselves to exceed the speed limit. This result could be explained in a number of ways. It could be argued that the results indicate that all individuals have a poor recollection of their own speeding behaviour. This is consistent with the aforementioned notion that little to no active cognitive processing occurs in familiar driving situations, albeit simulated. However, when the relationship between perceived speeding and actual speeding measures were observed, a statistically significant correlation existed for group 1 (see Table 6) in the overall percentage of speeding measure. These results suggest that whilst all groups perceived to recall speeding similarly, this perception was only accurately recalled in group 1. This poor recollection explanation is also inconsistent with the results of the perceived effectiveness of training, where individuals in group 3 expressed greater effectiveness of the training in reducing their own speeding behaviour. That is, whilst individuals recognised that as a result of training, their speeding behaviour was reduced, they were unable to accurately quantify this reduction.

One proposed explanation for group 1‟s accurate recollection of their speeding as well as group 3‟s poor recollection of their speeding could be explained by the availability of cognitive resources. Specifically, it could be argued that since group 1 had very little task demand, they had sufficient availability of cognitive resources to process a greater amount of information from the driving scenario. As such, they were able to accurately 119

store and retrieve information regarding their speeding behaviour, post-drive. This explanation also holds true conversely for group 3. That is, given that a significant proportion of cognitive resources were directed towards managing the driving task (i.e. speed management), less important information (such as number of times speeding) is either not collected and/or not stored in memory. This explanation could poses a very concerning problem for young novice driver safety where it is implied that when cognitive resources are allocated to one task, other potentially important information may not be collected and/or stored. These results are consistent with the notion that young novice drivers lack insight into their own driving behaviour and skill (Cairney, 1982; DeJoy, 1992;

Groeger & Brown, 1989; Holland, 1993; Job, 1990; McKenna et al.,1991). That is, this lack of insight may stem from a lack of availability of cognitive resources necessary to evaluate their driving behaviour and skill. However, given that group 2 and group 4 also had very little task demand, this explanation may not completely account for why they were unable to accurately recall their speeding similar to group 1.

Another possible explanation for the statistical non-significance may be due to the failure of participants to fully understand the question. It is also possible that individuals‟ perceived speed exceedance may be due to large variability in the participants‟ definition of speeding. For example, whilst one individual may have a liberal definition of speeding (10 km/h or more), another may have a much more conservative definition (1 km/h or more).

This explanation would mean that individuals are aware of their speeding behaviour but, due to variability of definitions, it is not captured in the data. This is of particular concern if

120

individuals are using the same definitions to manage their own speed in real driving situations.

Future research is essential to investigate the role of cognitive resource allocation in driving training, particularly looking at how the introduction of new tasks affects the cognitive processing of other existing tasks.

5.6.5 Perceived Effectiveness and Generalisation of Training.

The results of the questions which asked the individuals to assess the effectiveness of the training revealed that participants in group 3 recognised the significant impact of the training in terms of modifying their speeding behaviour in the second week‟s drive.

However, these participants also identified that the effectiveness was not transferred to real driving situations. Groups 1, 2 and 4 all recognised that their training method was not an effective method to reduce their speeding behaviour, in both simulation and actual driving on the road, which was supported by their behaviour in week 2. These results preliminarily suggest that such a training intervention may have limited transfer into real world application. However, it is possible that the individual is unaware of the effectiveness of the training. That is, significant behavioural changes may have occurred outside of their awareness. Further testing is necessary to investigate both the idea of transfer of training and objective effectiveness of this training.

5.6.6 Future Research into Episodic Training.

The malleability of the training also remains untested. Specifically, what are the integral characteristics of this type of training that facilitate the behavioural change? Some

121

characteristics identified could include the driving episode (which contained a number of speed zone changes) but also the feedback and the specific information provided such as their speeding or the potential consequences. In addition, whilst other dependent variables such as minimum/maximum/standard deviation of speed throughout the drive were not collected, these variables may have further been able to validate the results of episodic training. Future research should aim to include such dependent measures to provide further objective data to support the present results. Further, as mentioned earlier, it is possible that the effects are as a result of a researcher bias. Further testing is needed to identify which of these characteristics are necessary to produce this training effect.

The longevity of the training effect is also unknown, given that this experiment only examined a temporal gap of one week. Research in the aviation domain has suggested that training methods similar to episodic training not only produce better risk management immediately but produce effects that pertain six months post-training (Molesworth &

Wiggins, 2007). Future research should investigate methods that will sustain long term behavioural change and ways to prevent relapse.

5.6.7 The Cognitive Driver.

The results of the present research suggest that the behavioural changes produced by episodic training occur as a result of cognitive modification of an individual‟s script. In a review of the driver training literature, Senserrick and Haworth (2005) highlight the promising results found from training methods which involve developing driver cognition.

While the present research appears to support these findings, there appears to be a limited understanding of how cognitive constructs interact in the driving domain, specifically how 122

the retrieval and application of scripts interact with other cognitive tasks associated with driving. For example, while the application of a „normal driving‟ script can reduce speeding behaviour, its impact on other driving tasks is unclear. Future research should aim to evaluate the cognitive demands of behaviour modifications training such as episodic training.

5.7 Conclusion

The findings of the present research reflect positively on an applied training method to improve young novice drivers‟ speed management. The results suggest that providing individuals with case examples of other drivers‟ speeding behaviour has no impact in terms of reducing their own speeding behaviour. These cases examples are also no more effective in reducing speeding if the cases also highlight the rules violated and the potential consequences of the speeding behaviour. However, the results revealed that a training method in which individuals are placed in a simulated drive and provided with personalised feedback regarding their speeding behaviours reduced subsequent speeding. The effectiveness of this „episodic training‟ is attributed to way it changes cognitive structures known as „scripts‟, which are used to model driver behaviour. From an operational perspective, these results suggest driver training programs may benefit from providing novices with a more accurate appraisal of their performance. Although the experiment was conducted in a simulated environment, the results show promise for the future of driver training in terms of reducing speeding behaviour. While further research is required to

123

investigate the underlying cognitive mechanisms governing the changes in behaviour, the prognosis for episodic driver training appears positive.

124

Chapter 6: Experiment 2 – Role of Cognitive Resource Allocation in the

Implementation of a Modified Driving Behaviour

One of the results from experiment 1 suggested that drivers who received episodic training were able to reduce their speed but were no more able to recall how many times they exceeded the speed limit than participants in the control group, who did not receive training. Given that they recognised the effectiveness of the training, the result gave rise to the idea that their poor recollection of their speeding may have been an indication of the cognitive stress imposed on their limited cognitive resources as a result of actively having to modify their behaviour and manage their speed.

The second experiment aimed to explore this idea by investigating how cognitively demanding it is to implement a newly modified speed management strategy. That is, with a modified „normal driving‟ script, does the implementation of this behavioural change divert cognitive resources away from other tasks associated with driving? Research suggests that in the context of driver training, the implementation of a new or modified task, such as a speed management strategy, should initially be highly cognitively demanding (Paas, van

Gog, & Sweller, 2010; Sweller, 1988, 1994). In situations where this new or modified task is performed in isolation, the cognitive impact of these demands should be minimal, as cognitive resources can be wholly directed towards performing, monitoring and correcting the new task. However, if this task is performed in conjunction with others, there may be a negative impact, specifically decrements in performance on one or both tasks. Although the importance of cognitive resource diversion to driving has been recognised in the literature 125

(Horberry et al., 2006; Young et al., 2003), there is a need for further research into how drivers prioritise and allocate cognitive resources towards tasks. The experiment aimed to investigate this by introducing a secondary task to be performed in conjunction with a driving task by drivers that had received episodic training and those that had not. It is important to note that whilst the present research refers to a „secondary task‟, it is not intended as a distractor but as another task that is to be performed in conjunction with the drive task. Further, whilst termed a „secondary task‟, this not a reflection of how the tasks should be prioritised, but merely that it is in addition to the first task (i.e. a second task).

A factorial design (as seen in Table 8) was implemented to ensure that the impact of episodic training, the secondary task, as well as their interaction, could be assessed.

Table 8

Factorial Design of Experiment 2

No Episodic Training Episodic Training

No Secondary Task Group 1 Group 2

Secondary Task Group 3 Group 4

It is hypothesised that, as in experiment 1, speed management would be improved following receiving episodic training in isolation, i.e. reductions in speeding behaviour.

Speed management is hypothesised to be impaired with the introduction of a secondary task in isolation, i.e. increases in speeding behaviour. This hypothesis is based on the idea that the introduction of a secondary task diverts cognitive resources away from speed

126

management, and, as such, is less effectively managed. In terms of the interaction, it is hypothesised that with the introduction of a secondary task, speed management performance will be impaired, and this impairment will occur irrespective of whether participants receive episodic training or not.

6.1 Participants

Fifty-nine participants were recruited for the research through the same methods as experiment 1 (see 5.1 Participants). Of these, 16 had their learner‟s permit, 16 had their provisional 1 licence, 22 had their provisional 2 licence and five had their full licence. All participants had normal or corrected to normal vision. The average age of the participants was 20.78 years (SD = 1.99) and all were within the age range of 18-25 years (inclusive).

Participants averaged 68.37 (SD = 35.72) hours of supervised driving experience and 25.12

(SD = 23.96) hours of unsupervised driving. Participants reported driving an average of

8.59 (SD = 13.94) hours per week. Participants were reimbursed with a $40 bookshop voucher from the Australian Booksellers Association for their time. The research was approved in advance by the UNSW Human Research Ethics Advisory Panel.

6.2 Design

As seen in Table 9, the experiment comprised a 2 x 2 factorial design. The only dependent variable, which also appeared in experiment 1 was the percentage of drivers‟ speed

127

exceedance as a function of distance. Table 9 shows the tasks that each group received which will be described in more detail below.

Table 9

Summary of Design of Experiment 2

Pre Session 1 Session 1 Session 2 Post-Session 2

Group 1 Drive Questionnaires Test Drive Group 2 Drive + Episodic Training Post-Drive + Secondary Group 3 Drive Questionnaire Task Test Drive + Secondary Task Group 4 Drive + Episodic Training

6.3 Materials and Apparatus

6.3.1 Hardware.

Experiment 2 employed the same apparatus as experiment 1 (see 5.3.1 Hardware) in addition to a secondary laptop computer which was used to play secondary tasks‟ audio files, independent to the main PC. A Dell Inspiron 13z laptop, running Windows 7 Home

Premium 64-bit, included an Intel Core 2 SU4100 1.30 GHz processor, 4GB of RAM, an

Intel Graphics Media Accelerator X4500MHD (SU4100) as well as Intel High Definition

Audio 2.0. A set of stereo Logitech speakers (Model Number: S-0264B; 120V, 60Hz,

0.5A) were connected to the laptop.

Configuration of the room and previous hardware was also identical to experiment

1. The Dell laptop was placed on the researcher‟s side and the Logitech speakers were

128

placed on the participant‟s side, next to the Dell speakers. Both speakers where placed behind the participant‟s side monitor and were out of sight.

6.3.2 Software.

Software used in this experiment was identical to experiment 1 (see 5.3.2 Software). The audio files on the Dell laptop were played through Windows Media Player.

Driving Simulation Track

In contrast to experiment 1, the present experiment only contained one drive track, a

10 km drive. The decision to not have a practice drive was largely based on the results from experiment 1, which showed that participants did not vary significantly in terms of their ability to handle the vehicle with little practice. Given that the experiment 1 and 2 samples were recruited from the same population, their ability to drive the vehicle was assumed to be the same.

Participants completed the same 10 km drive in both weeks. The purpose of this was to ensure that any training effects reduced by the individuals‟ inability to recognise the similarity of the track, and as such, the applicability of the training when tested.

10 km Track (Appendix I for STISIM Script)

The drive consisted of four curves, largely inserted to reduce the monotony of the straight road. They were presented as R- RL - L- RL and were spaced approximately 2 km apart. As in experiment 1, the drive consisted of a courier-style package delivery task in which there were a total of five delivery zones. There were a total of 18 speed zones, six for 129

each zone type: 40 km/h, 60 km/h and 80 km/h. Unlike the drive in experiment 1, speed zones started independently of the delivery zones. That is, delivery zones were no longer always 40 km/h zones. The rationale for this was to firstly ensure that there were an equal number of the three speed zone types and, secondly to allow for greater randomisation of speed zone presentations. All other aspects such as tree density and distribution, traffic, and parked vehicles were identical to experiment 1.

6.3.3 Questionnaires and Scales.

All scales and questionnaires completed by participants as well as the scoring was identical to experiment 1 (see 5.3.5 Questionnaires). As in experiment 1, the primary purpose for the questionnaires and scales was to assess whether all groups were equal on critical variables prior to training.

6.3.4 Mental Arithmetic Task.

The mental arithmetic task was employed as a secondary task. Previous research has used similar tasks in the study on cognitive processing and driver distraction (Geary & Wiley,

1991; Harbluk et al., 2007). In this task, participants listened to an audio file where a series of 4-digit numbers were presented through the Logitech speakers via the Dell Laptop.

There were a total of 50 4-digit numbers, which were pseudo randomly sequenced (same for all participants but unpredictable to the individual). Participants were instructed that the task required them to add „1‟ to each of the 4 digits and verbally read out the new 4-digit number. For example, if the participant heard a 4-digit series of „4 6 2 1‟, the correct answer would read „5 7 3 2‟. Numbers from 1 through to 8 were used; 0 and 9 were not

130

used to avoid any confusion. Participants were also informed that they had 10 seconds to derive the solution once the 4-digit number was read (which also was the inter-stimulus interval). If they forgot any of the numbers, participants were asked to replace the forgotten number with the word „blank‟ and to continue with the remaining numbers. A visual aid was also used when individuals received task instructions (see Appendix J) to assist in the comprehension of the instructions.

The audio files were wave files, which were approximately 15 seconds in length with the 4-digit number being read in the first 5 seconds. The final 10 seconds of each audio file was silent and allowed for the participant to derive the solution. In total, the task took 15 seconds x 50 items + 10 seconds of introductory silence added to the first audio file; a total of 12 minutes and 40 seconds.

A response was deemed correct if the participant correctly answered with all five numbers in the sequence. All incorrect numbers, numbers incomplete within the time limit and numbers which were not recalled were considered to be errors. Any errors in the sequence were counted as an error for that number sequence as a whole. Each participant completed 50 items in week 1 and in week 2 completed as many as possible until the driving simulation had concluded but only the first 50 items were analysed.

6.3.5 Post-Drive Questionnaires.

The Post-Drive Questionnaires were identical to the questionnaire in experiment 1 (see

5.3.6 Post-Drive Questionnaires) with the addition of three extra questions. The first additional question asked participants to rate how difficult they found the drive in the 131

second week. This question was designed to assess whether individuals could accurately appraise task demands. Participants rated the perceived difficulty on a 7-point scale ranging from “Not at all difficult (1)” to “Extremely difficult (7)” with unlabelled markers at points

(2) through to (6).

6.4 Procedure

Participants initially read an information statement (Appendix K) regarding the nature of the experiment. The statement outlined that the experiment was to be conducted in two one- hour sessions spaced approximately one week apart. The description of the experiment outlined that in the first week they would complete a series of questionnaires and scales, and that they would participate in a driving simulation. In the second week they would complete a driving simulation as well as a Post-Drive Questionnaire. The potential risks were also expressed (as outlined in 5.4 Procedure). Participants were also informed that they would receive a $40 gift voucher from the Australia Bookshop Association at the completion of the experiment in the second week.

Once consent forms were signed, participants commenced the experiment with a battery of questionnaires. Participants were given the same instruction as experiment 1 in terms of how to complete the questionnaire. Following the questionnaires, participants took part in the mental arithmetic task (secondary task) in isolation. The purpose of completing this task in isolation was to get a baseline measure of task performance in all groups. Whilst it was not critical for all groups to participate in this task in week 1, it was administered to ensure that all groups participated in the experiment for a comparable amount of time and 132

to ensure that groups did not differ in any other way than the independent variables manipulated.

Following the mental arithmetic task, all participants underwent the 10 km drive on the simulator. The drive took approximately 10 minutes. Once this drive was complete, all participants were asked whether they felt any psychological or physical symptom as a result of driving the simulator. No participants reported any effects. For participants in group 1 and group 3, the conclusion of this simulation was the end of the tasks for the first session.

Group 2 and group 4 received the episodic training identical to group 3 in experiment 1. Participants were given immediate personalised feedback about their drive by the researcher (see Appendix L under „Feedback‟; details outlined in 5.4 Procedure). At the end of the session, all participants were thanked for their time and a mutually suitable time was arranged for the following week.

Approximately one week after training, all participants returned for the testing week. All groups took part in an identical driving simulation to week 1. Group 3 and group

4 were asked to complete the secondary task while driving, specifically the same mental arithmetic task completed in week 1. Group 1 and 2 completed the simulation with no secondary task. Finally, all participants completed a Post-Drive Questionnaire through

Inquisit 3.

133

6.5 Results

6.5.1 Questionnaires and Scales.

A series of one-way ANOVAs were conducted on all ratio and interval data with α set at

.05. This included age, perceived risk of being caught speeding, number of hours driving with and without supervision, average number of hours spent driving a week, semantic differential items for both speeding and safe driving, DBQ, internality-externality scale as well as Zuckerman‟s SSS-V. With all assumptions met, including homogeneity of variance, no statistically significant differences were evident in any of the variables across group, prior to training; largest F, F(1, 55) = 1.34, p = .27.

A series of Chi Square tests were also conducted to analyse the non-parametric variables – gender, licence type, infringement and crash – with α set at .05. The results failed to reveal any statistically significant differences in any of the variables across groups; largest Χ2, Χ2 (3, N = 59) = 2.55, p = .47. Combined, these results indicate that all groups were equal prior to training on these variables.

6.5.2 10 km Drive Task.

A 2 (episodic training) x 2 (secondary task) factorial ANOVA was conducted on the percentage of drivers‟ speed exceedance as a function of distance on the week 2 drive. As illustrated in Figure 12, the results revealed that participants that received episodic training

(Groups 2 and 4) exhibited less speeding behaviour compared to those participants that did not (Groups 1 and 3), and this result was statistically significant; F(1, 55) = 56.91, p < .001.

There was also a statistically significant main effect for secondary task; F(1, 55) = 4.52, p =

134

.04. That is, participants that completed the secondary task during the drive (Groups 3 and

4), exhibited less speeding behaviour than those that did not receive the secondary task

(Groups 1 and 2). In addition, there was a statistically significant interaction between the variables; F(1, 55) = 8.06, p = .01. As illustrated in Figure 12, these results suggest that receiving episodic training can dramatically reduce speeding behaviour in young novice drivers. Whilst the introduction of a secondary task also seems to reduce speeding behaviour (compared to not receiving a secondary task), it does not seem to impact speeding in those that receive episodic training as well.

Given the significance and nature of the interaction, four simple effect analyses were conducted to further assess the relationship between factors as suggested by Field

(2009). With a Bonferroni adjusted α of .05/4 = .013, the simple effect of the secondary task was statistically significant when participants did not receive episodic training (group 1 vs. 3; F(1, 55) = 12.54, p = .001), but was not statistically significant when participants did receive episodic training (group 2 vs. 4; F(1, 55) = .25, p = .62). The simple effect of episodic training revealed that in both comparisons there was a statistically significant difference in speeding behaviour (smallest F, F(1, 55) = 11.27, p = .001), irrespective of whether they received the secondary task (group 3 vs. 4), or not (group 1 vs. 2).

135

No Secondary Task 50 Group 1 Secondary Task

30 Group 3

10 Group 4 PercentageSpeedingof (ByDistance) Group 2 0 No Episodic Training Episodic Training

Figure 12. Percentage of speeding (by distance) in week 2.

6.5.3 Mental Arithmetic Task.

The first analysis was conducted to assess whether there was any difference between groups in terms of their ability to perform the secondary task in isolation. An overall one- way ANOVA was deemed appropriate as there was no experimental manipulation in week

1. Results comparing error rates on the task failed to reveal any differences between groups in terms of their errors made; F (1, 55) = .41, p = .74 (see Figure 13).

Following this, an analysis was conducted to assess how receiving episodic training impacted improvement on the secondary task from week 1 to week 2. Two paired sample t- tests were conducted with group 3 and group 4 to determine whether any changes in performance from week 1 to week 2 were influenced by episodic training in week 1. This was deemed appropriate over a mixed factorial analysis as the only two comparisons of 136

interest were within group differences over time. The results revealed that there was a statistically significant improvement in group 3‟s performance on the secondary task from week one to week two (t(14) = -2.19, p = .05) but there was no statistically significant improvement between weeks for group 4 (t(14) = -.11, p = .91; see Figure 13).

0.90 *

0.85

0.80

0.75

0.70

0.65 PecentageCorrectof Responses 0.60 Group 3 Group 4 Week 1 0.75 0.78 Week 2 0.84 0.78

Figure 13 . Percentage of Correct Responses in the Mental Arithmetic Task for Groups 3 and 4 Across Week 1 and Week 2

6.5.4 Post-Drive Questionnaire.

Analysis of „perceived risk of being caught for speeding‟ and „semantic differential item measures of speeding-related attitudes and safe-driving attitudes‟ were conducted identically to experiment 1 (see 5.3.6 Post-Drive Questionnaires). No overall change from week 1 to week 2 was observed in participants‟ „perceived risk of getting caught for speeding‟ (F(1, 55) = .97, p = .33) and there was no effect between groups; F(1, 55) = .41,

137

p = .75) There was also no interaction effects between time and group allocation; F(1, 55) =

1.10, p = .36.

Results for the semantic differential item for speeding showed that overall, there was a change in participants‟ attitude towards speeding over time (from week 1 to week 2;

F(1, 55) = 5.04, p = .03). However, the results failed to reveal a statistically significant difference between groups (F(1, 55) = .95, p = .42) nor was there an interaction effect; F (1,

55) =1.21, p = .32. That is, all groups changed their attitude to speeding equally over time.

Further inspection of the group means (see Table 10) indicated that attitudes to speeding post-training had become more negative (i.e. that speeding was bad).

The results for the semantic differential item for safe driving showed no overall change in their attitude to safe driving over time (from week 1 to week 2; F(1, 55) = 2.23, p

= .14), no overall effect between groups ( F(1, 55) = .53, p = .66), nor any interaction effects between group allocation over time (F(1, 55) = 1.71, p = .18; see Table 11).

Table 10

Average Score on the „Speed‟ Semantic Differential Questions – Experiment 2

Week 1 SD Week 2 SD

Group 1 0.480 1.090 0.893 0.916

Group 2 0.600 1.197 0.629 1.087

Group 3 1.173 1.249 1.227 1.093

Group 4 0.533 1.168 1.040 1.117

Note: positive scores on this measure indicated a more negative attitude towards speeding and the opposite for negative values

138

Table 11

Average Score on the „Safe Driving‟ Semantic Differential Questions – Experiment 2

Week 1 SD Week 2 SD

Group 1 -1.800 0.891 -1.867 0.810

Group 2 -1.800 0.967 -1.614 0.857

Group 3 -1.573 1.176 -1.893 0.740

Group 4 -1.280 0.861 -1.680 0.649

Note: positive scores on this measure indicated a more negative attitude towards safe driving and the opposite for negative values

A Chi Square analysis for the question “Do you recall exceeding the speed limit?” failed to reveal differences between groups (Χ2 (3, N = 59) = 6.80, p = .08), though this approached statistical significance. Inspection of the column averages for the proportion of

„yes‟ responses (see Table 12) to this question suggests that those that did not receive episodic training were more likely to report that they exceeded the speed limit. The row averages for the proportion of „yes‟ responses suggests that those that completed the secondary task were more likely to report speeding than those that did not.

A Kruskal-Wallis analysis was conducted for the question “How many times do you recall exceeding the speed limit?”. This method was deemed appropriate because while the data was in a ratio scale of measurement, homogeneity of variance was rejected (Levene‟s

Test; F(3, 45) = 11.61, p < .001). Results revealed that there were no statistically significant differences between groups in terms of the recollection of how many times they exceeded the speed limit; Χ2 (3, N = 49) = 2.87, p = .41.

139

Table 12

Proportion of "Yes" Responses to "Do You Recall Exceeding The Speed Limit?"

No Episodic Training Episodic Training Row Average

No Secondary Task 80% 64% 72%

Secondary Task 100% 87% 93%

Column Average 90% 75%

A Chi Square analysis was also conducted for whether participants “recall being in a similar driving situation, real or simulated”. Results for this question failed to reveal a statistically significant difference between individuals‟ ability to recall being in a similar situation (Χ2 (3, N = 59) = 2.87, p = .41). This result suggests that all participants, irrespective of group allocation, recognised that they had been in a similar situation before.

Participants were also explicitly asked to rate how effective they found the training to be, both in the simulated drive and in real driving situations. Homogeneity of variance was not met for the question asking about the effectiveness of the training in the second week simulation (Levene‟s Test; F(3, 55) = 4.06, p = .01), but groups were of moderate size and sufficiently equivalent so that the analysis was still considered valid (Tabachnick

& Fidell, 2007). A 2 x 2 factorial ANOVA conducted, with α set at .05, revealed that those that received episodic training considered the training to be more effective in a simulated environment that those that did not receive episodic training; F(1, 55) = 36.73 , p < .001. In contrast, results failed to reveal any statistically significant differences between those that received the secondary task compared to those that did not; F(1, 55) = .34 , p = .52. There was also no statistically significant interaction between the two factors; F(1, 55) = .03 , p =

140

.86 (see Figure 14).

A 2 x 2 factorial ANOVA was also conducted to assess the reported effectiveness of training in the real world. No main or interaction effects were statistically significant; largest F, F(1, 55) = 2.45 , p = .12 (see Figure 15).

Following this, participants were asked to rate how difficult they found the second week drive task. A 2 x 2 factorial ANOVA conducted with α set at .05 showed that there was no statistically significant differences in difficulty ratings in those that received episodic training compared to those that did not; F(1, 58) = .50 , p = .48. In contrast, those that received the secondary task rated the second week drive as more difficult than those that did not; F(1, 58) = 19.29 , p < .001. There was also no statistically significant interaction between the two factors; F(1, 58) = .50 , p = .48 (see Figure 16).

No Secondary Task 8 Secondary Task 7

1 Rateings of Effectiveness RateingsEffectiveness oftraining of 0 No Episodic Training Episodic Training

Figure 14. Rating of the effectiveness of the training in a simulated environment on a 7 point scale from “Not at all effective (1)” to “Extremely effective (7)”.

141

No Secondary Task

6 Secondary Task 5

1 Ratings Effectivenessof Training of

0 No Episodic Training Episodic Training

Figure 15. Rating of the effectiveness of the training in a real driving environment on a 7 point scale from “Not at all effective (1)” to “Extremely effective (7)”.

7 No Secondary Task Secondary Task 6

3 Difficulty Rating 2

0 No Episodic Training Episodic Training

Figure 16. Difficulty rating of experiment 2 on a 7 point scale from “Not at all difficult (1)” to “Extremely difficult (7)”.

142

6.6 Discussion

The aim of the present research was to investigate the cognitive mechanisms that underpin the modified behaviour seen in the episodic training condition in experiment 1.

Specifically, when an individual has received episodic training (drive + personalised feedback), what cognitive pressures does this impose on the limited cognitive resources and how do young novice drivers prioritise and allocate resources given this stress? The present research investigated the strain on cognitive resources when involved in actively managing speeding behaviour as a result of training. Experiment 2 replicated the findings of experiment 1 and supported the hypothesis of the research that episodic training successfully reduced speeding behaviour compared to receiving no training. Performing a secondary task during the test drive was also found to reduce speeding behaviour compared to no secondary task, rejecting the hypothesis which predicted impairment in speed management. However, the interaction between the two factors indicated that the secondary task only impacted those groups that did not receive the episodic training; those that did receive the training were unaffected by the secondary task. The hypothesis that a secondary task would impede the beneficial effects of the episodic training was not supported.

Participants in group 4 (episodic training + secondary task) reduced their speed significantly more than group 3 (No episodic training + secondary task), but this came at a trade-off. Participants in group 3 performed better in the secondary task compared to week

1, whereas group 4 showed no change. These results illustrate that group 4, as a result of the episodic training had placed a greater emphasis on speed management in week 2 compared to group 3, who appeared to place a greater emphasis on the secondary task.

143

6.6.1 Cognitive Mechanisms.

Whilst both the episodic training and the secondary task statistically reduced speeding behaviour, the cognitive mechanisms behind their effect are theorised to be distinctively different. The results of the research suggest that the success of the episodic training is due to a greater allocation of cognitive resources to the drive task (i.e. speed management in the test drive) as opposed to the secondary task, and this is likely to be due to the individual assigning task priorities as such. As a result, the reductions in speeding are due to conscious, effortful monitoring and resources-dependent modification of speeding behaviour. Similar reductions in speeding were observed in group 4 because speed management remained the primary task despite group 4 also performing a secondary task simultaneously with the drive. Consistent with this explanation, participants in group 4 did not improve in their performance on the secondary task during week 2. Group 3 (who did not receive episodic training in week 1) showed performance improvements on the secondary task.

The reduction in speeding due to the secondary task, however, can be attributed to the limited availability of cognitive resources. That is, given two tasks competing for limited cognitive resources, with one task arguably higher in risk compared to the other, it is conceivable that participants sought to maximise their performance by reducing the cognitive demands of the driving task. This was achieved by reducing the speed at which they travelled in order to allow for sufficient cognitive resources to complete the secondary task. In other words, participants engaged in a compensatory behaviour, namely reducing the potential risk of an accident while at the same time maximising their overall

144

performance on both tasks. Given that group 3 had lower speeding than group 1 it is likely that without directive, participants allocated the greater portion of available resources to the secondary task, because they identified this task as the highest priority. One possible explanation for this allocation priority is that the participants identified this task as the most difficult and thus requiring more cognitive effort to satisfactorily complete. Another is that the participants perceived this to be more important to the aim of the experiment.

Once the two tasks commence, the individual recognises the cognitive stress and their inability to manage both tasks simultaneously. As a result, rather than changing the priority of tasks or sacrificing one task‟s performance for the performance on the other, the individual finds a cognitive strategy to satisfy both tasks. Rather than monitoring their speeding near the speed limit (for example, 5 km/h below), they may have opted to reduce their speed further (for example, 10 km/h below), in order to eliminate the cognitive load of this task on overall performance. Hence, this lack of closer monitoring of speed frees up cognitive resources to be directed towards the secondary task. This type of cognitive strategy is well supported in the driver distraction literature, where reductions in speed are said to be as a result of less attention toward the driving task as well as a compensatory mechanism to reduce crash risk (Horberry et al., 2006; Young & Regan, 2007; Young et al.,

2003). Further, it has been shown that in-vehicle distractions, such as mobile phone use, produce similar reductions in speeding (Horberry et al., 2006; Jenness, et al., 2002;

Srinivasan & Jovanis, 1997; Young & Regan, 2007).

This explanation is further supported by the results in the Post-Drive Questionnaire asking participants to rate the subjective difficulty of the tasks where participants who

145

received the secondary task rated the task more difficult than those that did not. The results suggest that individuals recognise the limitations of their cognitive resources whilst driving and completing a secondary task simultaneously.

The results indicate that group 3 neglected the drive task for the secondary task, whilst group 4 neglected the secondary task for the drive task. The results neatly highlight that both tasks cannot be effectively performed or improved simultaneously without further training. As such, the individual allocated cognitive resources based on subjectively assessed task priority. This type of task prioritisation is supported in the driver distraction literature (Young et al., 2003). Importantly, this type of allocation produces significant variation in terms of speed management. That is, how an individual decides to allocate cognitive resources to tasks whilst driving can have potentially significant safety impacts.

6.6.2 Alternate Explanation.

One alternate explanation for the results draws from the driver fatigue literature.

Specifically, when assessing the groups that did not receive episodic training (group 1 and group 3), results revealed that there was significantly less speeding in groups that received the secondary task (group 3) compared to groups that did not (group 1). Whilst it has been interpreted that this reduction is due to recognition of cognitive load and subsequent cognitive trade-off (see 6.6.1 Cognitive Mechanisms), it could alternatively be argued that reductions in speeding (in group 3) are due to a „reengagement‟ in the task. That is, poorer speed management in group 1 could be viewed as a by-product of fatigue due to monotony of the driving task. As such, drivers in group 1 may „disengage‟ from the task, leading to poor management of speed; an idea supported by the literature (Hockey, Wastell, & Sauer, 146

1998; Matthews & Desmond, 2002; Williamson, Lombardi, Folkard, Stutts, Courtney, &

Connor, 2011). The secondary task could be viewed as a means to „reengage‟ the participant to actively monitor not only their secondary task performance but also to better manage their speed.

This explanation could have been substantiated with pre and post-drive fatigue measures, but fatigue-related driver performance was beyond the scope of the experiment.

It should be noted that, irrespective of the mechanisms which caused the reductions in speeding (cognitive task trade-off or reengagement to task), the role of cognitive resources and their allocation towards tasks still largely determines how they are performed. Future research should aim to investigate the role of fatigue on speed management in young novice drivers, particularly under high cognitive workload.

6.6.3 Post-Drive Questionnaire.

Similar to experiment 1, the Post-Drive Questionnaire again revealed interesting results with regards to the semantic differential (see 5.5.5 Post-Drive Questionnaire). The results seem to add weight to the previous findings that attitudinal changes to speeding do not necessarily imply correlated changes to safe driving attitudes; the two components may be semantically more disassociated than assumed.

Results were similar to experiment 1 with regards to perceived effectiveness of training in a simulated environment as well as in the real world. These results further add weight to the results from experiment 1 suggesting that episodic training may not translate to driving scenarios in reality. Further research should aim to investigate the potential for simulator training to change driving behaviour in the real world. In addition, research could 147

look into whether the effectiveness of episodic training applies when training occurs in reality.

Results for participant‟s rating of their perceived difficultly found groups that received the secondary task (group 3 and group 4) compared to those that did not (group 1 and group 2) were approaching statistical significance. The results seem to suggest that young novice drivers have a good awareness of the cognitive stress they are under.

6.6.4 Cognitive Integration.

The research suggests that, by extension in a driving context, individuals may benefit from learning driving tasks with sufficient cognitive resources available to fully process the task demands. Importantly, the cognitive demands of learning are not only brought about by each task‟s complexity, but by the integration of these tasks simultaneously. That is, not only does an individual have to learn how to cognitively manage each task and its cognitive demand, but must also learn how to cognitively integrate multiple tasks together without sacrificing performance. When multiple complex novel tasks are present at the same time, integration of these tasks consumes cognitive resources to such an extent that task performance decreases. This is again supported in the driver distraction literature (Horberry et al., 2006; Young et al., 2003).

6.6.5 Part-Task vs. Whole-Task Training.

The question which stems from this research is how cognitive training in the driving domain should be delivered. Specifically, should the learning of the cognitive skills of driving be taught in components or parts (Part-Task Training) which can be bought together

148

or taught as a whole (Whole-task Training) all at once. There is much debate in the motor skill training literature as to which of these two types of training produce the most beneficial training outcomes (Fontana, Mazzardo, Furtado Jr, & Gallagher, 2009; Teague,

Gittelman, & Park, 1994). Naylor and Briggs (1963) proposed that skill acquisition is based on the complexity of elements of a task (the cognitive demand) and the organisation of the tasks. They suggested that part-task training should be employed when complexity is high and when organisation is low and propose whole-task training when the converse is true.

This proposal was substantiated in meta-analysis conducted by Fontana et al. (2009).

Whilst Naylor and Briggs (1963) proposal focuses on the training of physical skill, it is expected that these same training techniques also apply to the training of cognitive skills.

One question which needs to be addressed with future research with regards to these training programs is how „complex‟ and „organised‟ are the cognitive tasks of driving?

6.6.6 Implications.

The primary implications of the present research substantiate the cognitive stress that is induced when applying speed management behaviours produced by episodic training. The research suggests that when young novice drivers are given multiple tasks, how they prioritise tasks and allocate cognitive resources can have a significant impact on performance and safety.

The results have implications for current driver licencing procedures which could benefit from more clear structure in terms of the learning of the cognitive tasks necessary to drive. Specifically, drivers may benefit from better directions as to how to allocate cognitive resources towards driving tasks in the initial phases of learning. The results from 149

the present research imply that, without directive, young novice drivers are left to assign cognitive resources as they see appropriate and given their lack of experience, this is unlikely to be the best utilisation of their limited cognitive resources.

Currently in Australia, licensing procedures mandate no formal learning structure for driver training, physical or cognitive. The results of this research suggest that without a formal learning structure, the demands of learning how to drive (both physical and cognitive) as well as cognitive integration of multiple tasks may be overwhelming.

6.6.7 Limitations and Future Research.

Many avenues exist for future research exist which may also attempt to address some of the limitations of the present research. From the results of this research, it is unclear whether these effects of episodic training will be seen in real driving environments. Future research should investigate whether simulator-based training can be used for driver training or whether energy should be directed towards applying episodic training to real driving scenarios.

Research which extends from the present experiment could also aim to look at how perceived task difficulty is assessed by individuals, particularly during the initial assessment of how to divide cognitive resources to multiple tasks. This type of research would shed light on factors which may contribute to the misallocation or mismanagement of cognitive resources. The extension of such research could assist in identifying specific training strategies.

Future research should also aim to explore the intricacies of this type of training.

Whilst episodic training is viewed as a holistic training program, it is acknowledged as a 150

limitation of the research that multiple components could have individually or collectively produced the results. Specifically, not only did participants in the episodic training conditions receive immediate personalised feedback regarding their performance, they were also exposed to two sessions of the driving simulation. Therefore, future research should tease out the effect of these components.

Research could also systematically investigate how variations in workload affect driver behaviour. In addition, assessing mental load using other dependent measures such as eye tracking data could shed more light into the workings of this type of training and the cognitive mechanisms behind its success.

An exciting avenue of research could aim to investigate whether it is possible to train individuals to allocate finite quantities of cognitive resources to specific tasks.

Specifically, can cognitive resources be taught to be allocated in an arbitrarily defined way?

The terms „cognitive resources‟ and „cognitive mechanisms‟ are used quite loosely but largely refer to the collective implementation of both lower and higher-order cognitive skills. The specific mechanisms and the interaction between long-term memory (LTM), working memory (WM), attentional resources and decision-making skills are not fully understood, and go beyond the scope of this experiment. It is hypothesised however, that the speed management strategy produced by the episodic training is intuitively stored in

LTM where it has cognitive ties to the individual‟s schema of driving. The application of this strategy however, is hypothesised to be implemented through a complex interaction of

WM and attentional resources, both divided and selective. Future research should aim to

151

investigate these hypotheses to strengthen the literature as to the exact cognitive workings behind episodic training. This theme is deferred to the aims of experiment 3.

6.7 Conclusion

The findings of the present research highlight the cognitive implications for implementing a speed management strategy produced by episodic training. Specifically, when a secondary task is introduced, cognitive resources are allocated depending on the individual assessment of task priorities. From an operational perspective, the results identify the cognitive stress imposed by episodic training and the subsequent performance decrements. The results suggest that young novice drivers, without directive, are likely to allocate cognitive resources inappropriately towards driving tasks. From a theoretical perspective, the results open up an avenue of research which can aim to dissect the intricacies of driver training from a cognitive perspective. The following chapter will aim to explore the workings of human cognition and its application to road safety.

152

Chapter 7: Experiment 3 – Cognitive Resource Allocation to a Novel Dual-Task in

Young Novice Drivers.

The genesis of the third experiment drew from the results of the second and was directed into the field of cognitive psychology. Specifically, it was noted in experiment 2 that drivers prioritised tasks based on subjective assessment and this led to decrements in performance in some tasks due to a cognitive trade-off or task prioritisation.

Given the lack of formal cognitive training in current driver licensing systems, it could be argued that young novice drivers are required to learn how to allocate their already stretched cognitive resources as they deem appropriate to the demands of the driving tasks.

However, their allocation is likely to be inappropriate, given their lack of experience in managing the cognitive demands of driving. This notion is also in line with SEEV model of selective attention where young novice drivers are likely to be more susceptible to top- down influences due to a lack of knowledge and experience (Wickens & Horrey, 2009; See

4.2.2 The SEEV model of Selective Attention).

It seems intuitive that the progression of this idea is to ultimately provide young novice drivers with training as to how to allocate cognitive resources towards the multitude of driving tasks. In essence, the aim would be to fast track the development of task- appropriate resource allocation in young novice drivers.

Research into Variable Priority training (Gopher, 1992, 1996; Gopher &

Kahneman, 1971; Gopher et al., 1994; Regan, Deery, & Triggs, 1998; Regan, Deery,

Triggs, et al., 1998; Regan et al., 2000; Shebilske et al., 1992) provides a good foundation 153

for this training approach to be investigated. Specifically, Gopher, (1992) demonstrated that the ability to control how attentional resources were allocated towards simple computer- based tasks could be facilitated through training. Further, Regan, Deery, and Triggs (1998) demonstrated that this same type of training could improve attention control in drivers in a simulated driving environment. Thus, these training methods are said to facilitate better overall control of how attentional resources are distributed. However, what remains untested was the potential for this type of training to be used to train the allocation of specific amounts of cognitive resources towards tasks. The present experiment aims to go beyond the scope of VP training to assess the potential to train the allocation of cognitive resources.

One point of clarification is in regards to the terminology used to describe the cognitive mechanisms proposed to be involved in these types of training. VP training is repeatedly referred to as training attentional systems but this classification neglects a number of cognitive systems described in Chapter 4 (STSS, WM, LTM and decision- making skills), which play a vital role in the performance of these tasks. That is, this type of training actually provides individuals with the skill to integrate the complex workings of all of these cognitive resources. Given this distinction, the present experiment will refer to such training as Cognitive Resource Allocation (CRA) training.

The experiment was designed to answer two research questions. The first aimed to investigate how individuals allocated cognitive resources during a dual-task exercise given differing task difficulties. Specifically, the experiment aimed to assess how well young novice drivers could assess task difficulty and subsequently apply this knowledge. This

154

question was designed to reflect current driver training procedures, where individuals are left to assess the difficulty of driving tasks and subsequently allocated resources based upon this implicitly learned information; a „naturalistic learning‟ approach. The second research question aimed to determine whether it was possible to provide a training intervention to facilitate in task specific allocation of cognitive resources. That is, to assess whether targeted training could be provided to teach an individual the best way to distribute cognitive resources to manage task performance. This question was designed to further investigate a solution to the cognitive trade-off effects seen in the results of experiment 2 by implementing a feedback-based training approach.

In an attempt to neatly assess the intricacies of how cognitive resources were distributed, the present experiment moved away from driving simulations and assessed the effects of training in a simplified computer-based dual-task. Specifically, participants took part in a purpose-designed dual visual and auditory task simultaneously. The use of a separate visual and auditory task is well document in the multitasking literature to ensure that tasks do not compete for the same resources and that there is minimal resource overlap

(See section 4.8 Multiple Resource Theory and Multitasking; Horrey & Wickens, 2003;

Jaeggi, Buschkuehl, Perrig & Meier, 2010; Richard et al., 2002; Wickens, 2002; 2005;

2008; Wickens et al., 2006). The task was modelled based upon the dual n-back task by

Jaeggi et al. (2010) where individuals were required to respond to a visual and auditory task. The rationale for not adopting Jaeggi et al. (2010)‟s n-back task specifically was because performance depended on the use of working memory resources to remember previous trials. As such, the task was modified to have a visual and auditory task that could

155

be completed within the length of the trial, without having to remember previous trials. The task was also designed to be significantly difficult to complete simultaneously. The rationale for using such a dual-task was to employ a task which would theoretically consume 100% of cognitive resources, making it necessary to distribute resources appropriately to perform both tasks simultaneously. In addition, the modalities of the tasks were different (visual and auditory) to ensure that both stimuli could be perceived simultaneously without any interference or competition from the other task. Any learning that occurred in the training session (week 1) was assessed in the test session (week 2) where they all performed the same dual-task.

The research questions were addressed by employing a 2 x 2 factorial design; manipulating training and task difficulty (Table 13. ) Training consisted of a participants receiving feedback to amend their performance after each block. Specifically, participants receive explicit feedback after each block of the dual-task regarding their performance to allocate significantly more resources towards the visual task (75% of their total cognitive resources) than the auditory (25% of their total cognitive resources). The rationale for this was that with this feedback, participants would actively coordinate cognitive systems and redirect cognitive resources towards the necessary tasks. That is, through repetition and practice, it is expected that participants would learn and store a strategy in memory that contains information about how to process the dual-task simultaneously as well as information about how cognitive systems were utilised and operationalised. This type of training continues along the same vein of the two previous experiments where individuals who are provided with explicit feedback are able to amend their behaviour based on active

156

cognitive processing of the task requirements. During the test session, participants in this condition are hypothesised to draw from their memory of the training session and allocate cognitive resources towards the dual-task in a similar way, a theoretical 75/25 split. Task difficulty was manipulated by having an „easy‟ and „hard‟ version of the visual task; the auditory task was the same between groups. Whilst the version of the task differed depending on group allocation in Week 1, all participants performed the easy version of the task in Week 2. This ensured that any group differences in performance in Week 2 could be casually attributed to the training in Week 1. It is hypothesised that without any training intervention in Week 1, individuals will distribute their cognitive resources to maximise performance on both tasks equally and the same allocation strategy will be applied in Week

2. Whilst participants that complete the „easy‟ version of the task in Week 1 are expected to split their resources equally (i.e. 50/50 split), participants that complete the „hard‟ version of the task in Week 1 are expected to have allocated significantly more resources towards the visual task to achieve the same performance. Thus it is hypothesised that in Week 2, these participants will incorrectly „over-allocate‟ resources towards the visual task thus producing significantly better performance on the visual task.

7.1 Participants

Sixty participants were recruited through the same two methods as employed in the previous experiments (see 5.1 Participants). All participants were young novice drivers

(consistent with previous experiments) within the age range of 18-25 years (inclusive; average age 21.42 years, SD = 2.27). Of these, two had their learner‟s permit, seven had 157

their provisional 1 licence, 22 had their provisional 2 licence, 20 had their full licence and nine had an international driving permit allowing them to drive on NSW roads. All participants had normal or corrected to normal vision. All participants were reimbursed with a $40 bookshop voucher from the Australian Booksellers Association for their time.

The research was approved in advance by the UNSW Human Research Ethics Advisory

Panel.

7.2 Design

As previous mentioned, a 2 x 2 factorial design was employed with training containing two levels (CRA training and No CRA training) and task difficulty also containing two levels

(easy and hard; Table 13). The dependant variable was performance on the visual task alone in week 2 (CRATotal % for the visual task; See section 7.3.5 Cognitive Resource

Allocation Calculation and Macro File).

Table 13

Summary of Design of Experiment 3

Easy Visual Task (Week 1) Hard Visual Task (Week 1)

No CRATraining (No Feedback) Group 1 Group 2

CRA Training (Feedback) Group 3 Group 4

158

7.3 Apparatus and Stimulus

7.3.1 Hardware & Software.

A single PC was used running Windows XP Premium 64bit, with an Intel Core 2 SU4100

1.30 GHz processor, 4GB of RAM, an Intel Graphics Media Accelerator X4500MHD

(SU4100), Intel High Definition Audio 2.0 as well as standard keyboard and optical mouse.

A set of stereo Sennheiser headphones (Model Number: S-0264B; 120V, 60Hz, 0.5A) were connected to the PC and all auditory stimuli were played through these. A dual monitor set up was employed with one primary monitor on the participant‟s side and a secondary monitor on the researcher‟s side. The researcher‟s table was positioned perpendicularly behind, and out of sight to the participant‟s table. There was no partition separating the participant and the researcher. In terms of software, all stimuli throughout the experiment were presented through Inquisit 3 (version 3.0.2.0).

7.3.2 Stimuli.

Visual Letter Matrix

The visual stimulus was a 20 x 20 square matrix of letters (400 letters). The letters used were in 12 point font size (approximately fitting into a 16 x 16 pixel square) and in upper case, presented in Microsoft‟s „Calibri‟ font in a neutral grey colour (present on a black background). The stimuli were created using Microsoft Windows Paint (Version 6.1, Build

7600). The matrix of letters was presented in the middle of the 800 x 600 pixel screen. In some matrices, all 400 letters were the same. In other matrices, either one, two or three

„main target letters‟ were pseudo randomly (same between participants but unpredictable)

159

placed in the letter matrix. Letter combinations where chosen based on physical similarity to the main target letter. This presentation of the stimulus was in accordance with the

Feature Identification Theory by Treisman and Gelade (1980), where only one feature (the letters in the matrix) were varied in each stimulus; the letters were all the same colour, neutral grey. This ensured that there was no need for the integration of multiple features.

Table 14 shows the 18 letters which were paired as either main target letters (different from majority of the matrix) or non-target letter (same as majority of the matrix). In total there were 72 stimuli created (54 with targets – 9 x 8 letter combinations, and 18 without – all 18 letters individually; see Figure 17 for an example of target stimuli).

Table 14

Letter Pairs Used for the Visual Stimuli

1. E & F

2. J & I

3. M & N

4. P & R

5. O & Q

6. U & V

7. X & Y

8. C & G

9. B & D

Note: both letters in each row were used as targets and non-targets

160

The stimuli were pilot tested for detection difficulty; a critical component of the group differences. All 72 visual targets created were tested on four participants (including the primary researcher). Each of the stimuli were presented through Inquisit 3 (See 7.3.1

Hardware & Software) to the participant for a maximum of five seconds and participants were instructed to press the spacebar as quickly as possible if they detected any letters in the matrix array that were different from the rest; if they deemed the matrix to be uniform, they were asked to wait for the stimulus to clear automatically before the next stimulus appeared. The task did not necessitate that they identify where the target/s appeared in the matrix, but merely to state their presence. Each stimulus was present three times (216 presentations).

Figure 17: An example of the visual stimuli used where within an array of non-target „O‟ there are two target „Q‟.

161

After inspecting reaction times of the pilot data, a criterion was defined to distinguish between the 54 target stimuli as to which were easy, moderate and hard to detect in the matrix. The criterion for the easy stimuli was an average detection time of less than 1,250 milliseconds and for the hard stimuli greater than 2,500 milliseconds. Target stimuli which were on average detected in between these times were considered of moderate difficulty. Target difficulty was determined if more than three participants agreed on the difficulty level. Based on this criterion, 12 easy and 12 hard stimuli were selected as the final stimuli. In addition, six stimuli were assessed as being moderately difficult. The remainder of the stimuli were not used for the experiment.

The division of the visual stimuli into the „easy‟ and „hard‟ categories then featured in the main experiment. Moreover, the easy version employed the 12 „easy‟ stimuli as well as the six moderate stimuli (18 stimuli in total). The „hard‟ version employed the 12 hard stimuli in addition to the six moderate stimuli. The moderate stimuli were included in both versions not only as a common anchor but to ensure that the tasks were not too easy or hard. In addition, the 18 non-target stimuli were included in both easy and hard versions of the task (36 stimuli in both conditions).

Auditory Dual Numbers

The auditory stimuli consisted of all numbers from one through to nine. The audio of the numbers being spoken were sampled from Google Translate (Google Inc, 2011) by using the „Listen‟ function after typing each number as a word in English. The audio was recorded internally through an Apple iMac running Mac OS X Lion (version 10.7.3) and

162

was processed using the software GarageBand 2011 (version 6.0.5). Each number was then paired with every other number and saved as individual files (9 x 8 = 72 audio files). For each file, one number was played through the left channel of the headphone and the other through the right channel of the headphone. Both numbers were synchronised to play at the exact same time in each channel for precisely 1.5 seconds.

7.3.3 Inquisit Files – Dual-Task.

For the visual component of the dual-task exercises (visual and auditory task), in both conditions („easy‟ and „hard‟), Inquisit randomly drew from a pool of stimuli (36 visual and

72 auditory) to produce a single block of the task, which consisted of 40 pairings of the visual and auditory stimuli producing 40 trials. Each trial consisted of one visual stimulus and one auditory stimulus presented simultaneously for 1.5 seconds. Each block went for one minute (1.5s x 40 trials). Prior to each block, participants were provided a warning that the task was about to commence with a three second countdown. At the conclusion of each block an additional notice was provided asking them to switch the monitor off.

Participants completed either 10, 15 or 20 blocks of the task. This varied between participants based on performance. Specifically, when it was observed by the researcher that the participant‟s performance reached a plateau or the participant appeared notably fatigued, the experiment ceased. Performance plateau was defined as consistent performance for three consecutive blocks. Terminating the experiment at these points was deemed appropriate as it was evident that further task performance would not produce meaningful data.

163

7.3.4 Questionnaires.

A questionnaire also featured in the experiment designed to obtain background information about participants, such as age, gender, highest level of education and type of driver‟s licence. In addition, participants were asked how many hours they slept on average per night as well as how many hours they had slept the previous night. Participants were also asked if they had had any amount of stimulant (e.g. caffeine) greater than their usual intake.

The last three questions were aimed at assessing whether a participant‟s performance on the task could be attributed to or at least influenced by these factors. These three questions were also repeated in the second week.

Similar to previous experiments, the main purpose of obtaining demographic information was to ensure groups were equal on identified variables prior to the commencement of the experiment. Following the questionnaires, participants completed the computer task.

7.3.5 Cognitive Resource Allocation Calculation and Macro File.

Cognitive Resource Allocation (CRA) was assessed based on the creation of a formula theorised to measure CRA as a result of performance. The formula was based on Signal

Detection theory, which was designed as a way to measure detectability of targets from non-targets. As seen in Table 15, Signal Detection Theory allows for four possible outcomes based on whether a target was present or not (Target vs. Non-Target) and whether the target was identified or not as Present or Absent.

164

Table 15

Signal Detection Theory Outcomes

„Present‟ „Absent‟

Target Hit Miss

Non-Target False Alarm Correct Rejections

The formula employed for this is presented in Equation 1 and was employed separately for the visual and auditory task. The formula was designed in such a way to distinguish effortful correct behaviour compared to random key presses and was achieved by placing the correct behaviours (Hits and CRs) as the numerators and the incorrect behaviours (FAs and Misses) as the denominators. This allowed for correct behaviours to be standardised over incorrect behaviours; the more correct behaviours over incorrect behaviours, the greater the CRATASK value (where CRATASK refers to either CRAVISUAL or

CRAAUDITORY). The rationale for using this formula was based on the assumption that when a greater number of correct behaviours were exhibited with less incorrect behaviour, it was theorised to suggest that a greater proportion of cognitive resources were allocated towards that task.

Equation 1: Formula used to calculate CRA for each task (CRAVISUAL and CRAAUDITORY)

165

The CRA TOTAL % was then calculated as the proportion of CRATASK over the sum of the

CRA towards both tasks (see Equation 2). This percentage value for each stimulus type

(CRAVISUAL and CRAAUDITORY) was theorised to represent the proportion of the total cognitive resources allocated toward performance on that task; total summed to 100%.

Equation 2: Formula used to calculate percentage of CRA toward each stimulus

For ease and speed of calculation, a Microsoft Excel Macro application was employed to calculate task resource allocation based on the above formulae. This was calculated by inputting raw participant data files produced by Inquisit into an Excel spreadsheet and running a macro application to arrange and analyse the data. Based on this measure, participants‟ cognitive resource allocation was inferred.

Use of a macro allowed for participants‟ performance to be assessed immediately.

For group 1 (explicit), the macro file produced was used to provide immediate feedback for the participants. For groups 2 (implicit) and group 3 (control), the macro file was used for the researcher‟s observation of the participant‟s performance. Whilst these participants did not receive any feedback about their performance, this real-time observation allowed the researcher to assess fatigue-related performance detriments as well as when performance reached a plateau.

166

7.4 Procedure

Participants initially read an information statement (see Appendix M) regarding the nature of the experiment. The statement outlined that the experiment was to be conducted over two one-hour sessions spaced approximately one week apart, and outlined that in both sessions participants would be asked to complete a short questionnaire followed by a PC- based task. The potential risks were also identified as including experiences of „cognitive fatigue‟ as well as physical symptoms such as motion sickness, dizziness, fatigue and/or nausea due to the rapid changes of visual stimuli. Participants were also informed that they would receive a $40 gift voucher from the Australian Bookshop Association at the completion of the experiment in the second week. Once consent forms were signed, participants commenced the experiment with the demographic questionnaire (see 7.3.4

Questionnaires for details).

The task was identical for all participants, but unbeknown to them was the level of difficulty of each task. Participants were instructed that they would take part in a series of one minute computer task blocks. In each block, they received a number of visual and auditory stimuli simultaneously and their task was to respond to both items correctly. The visual component of the task required participants to press “A” on the keyboard if they noticed any letters were different from the rest of the matrix. If all the letters in the matrix were perceived to be the same, they were not required to press any key. The auditory component of the task required participants to press “L” if they identified the two numbers they heard as both odd or both even. If the auditory stimuli contained one odd and one even number, they were not required to press any keys. All participants were also given a visual

167

example of both stimuli and had the opportunity to ask questions about the dual-task (see

Appendix N for visual example). Whilst these instructions were identical between the groups, further information provided to the participants distinguished the groups.

Participants were randomly allocated into one of four groups (see Table 13). Those in the CRA training conditions (g roup 3 and 4) were provided with instructions regarding the task, as well as being informed that they would receive verbal feedback from the researcher regarding their performance after each block. Specifically, the feedback was based on a real-time calculation of their performance on the previous block and was aimed at highlighting where their performance was lacking (either the visual or auditory component of the dual task). The feedback was presented in an identical format between participants to ensure their change in behaviour was explicitly due to the feedback and not the way in which the feedback was delivered (see Appendix O under „Feedback‟).

Specifically, participants were only told which task they needed to improve in performance; no negative feedback was ever provided. The rationale behind employing this type of feedback was based on the assumption that 100% of cognitive resources were being operationalised. As a result, an increment of CRA towards one task is expected to naturally produce a decrement towards the other.

Assessment of CRA performance was calculated immediately after the block was complete using the CRA macro. The scores were said to reflect the division of cognitive resources allocated to the task. In the case of group 1 (easy task), feedback was given in such a way to encourage participants to allocate a theoretical 75% of cognitive resources towards the visual task and the remaining 25% towards the auditory task. This allocation

168

was based on a similar distribution used in Gopher (1992).

In addition to the basic task instructions, participants that received the hard version of the task (group 2 and 4) were informed that each block had a significant number of targets for both the visual and auditory component, and that if they felt they were not responding enough in the block to one or both stimuli, they were likely to be missing the targets. This instruction was given to facilitate an internal motivation to allocate cognitive resources towards the component they perceived to be more difficult or that required greater resources to perform adequately (though perceived difficulty remained untested).

These conditions were designed to produce a natural learning of CRA based on task difficulty. It was expected that individuals would recognise that they were not performing adequately on the visual task (due to its difficulty) and thus actively allocate more resources to improving performance on this task. For group 4, this coupled with feedback was expected to provide a more accurate appraisal of task difficulty and thus, better resources allocation compared to control.

All participants were informed that the tasks could get quite repetitive but were told the more poorly they performed, the longer the first session would take. Participants were explicitly told to stay as motivated as possible to conclude the first session as quickly as possible. At the conclusion of the session, participants were thanked for their time and a mutually suitable time was arranged for the following week.

The second week commenced with the two questions regarding previous night‟s sleep and stimulant intake. Following this, participants were informed that they would perform 10 blocks of the same task performed in the previous week and were reminded of

169

the task instructions. In fact, the task was the same for all groups; they all completed the easy version of the dual-task. Importantly, no instructions were provided to any groups regarding how they should perform or how they should allocate resources nor was any feedback given. Following the 10 blocks, participants were debriefed, provided their reimbursement and thanked for their time; this concluded the experiment.

7.5 Results

7.5.1 Demographics.

In order to determine whether participants were equal prior to commencing the experiment on the demographic variables, a series of one-way ANOVAs were conducted on all ratio and interval data (i.e. age, average hours of sleep and hours slept last night) with α set at

.05. With all assumptions met, including homogeneity of variance, no statistically significant differences were evident in any of the variables across groups; largest F, F(1,

56) = 1.34, p = .27.

A series of Chi Square tests were also conducted to analyse the non-parametric data

(gender, level of education, licence type and excess stimulant intake) with α set at .05. The results revealed no statistically significant differences in any of the variables across groups; largest Χ2, Χ2 (3, N = 59) = 2.55, p = .47. Combined, these results indicate that all groups were equal prior to the commencement of the experiment on these critical variables.

170

7.5.2 Cognitive Resource Allocation Results.

Assessment of the implicit condition (group 2) compared to control (group 3) was designed to address the first aim as to how cognitive resources were allocated in a dual-task exercise when difficulty was manipulated unbeknown to the participant. The second aim was assessed by examining performance between the explicit group (group 1) compared to control (group 3).

Analysis

A 2 x 2 MANOVA were conducted to analyse the data. As previously mentioned, the dependent measures for all the CRA analyses examined performance on the visual task alone in week 2. That is, the calculated CRATotal % (see Equation 2) for the visual task. It was of no significance which of the two modalities (visual or auditory) was used for the analysis as they were complementary to each other (summed to 100%). That is, the choice as to which modality to use in the analysis had no impact on the results; significance would be identical because the sum of the values for both tasks was equal to 100.

In total, two sets of both DVs were assessed to examine differences between conditions on the first block alone and the average of blocks two through to ten. Results were assessed in this manner in order to ensure that if there were any initial experimental effects, it would be identified and not washed out from examining the entire data set in isolation. This could occur if the effect was weak or not consolidated sufficiently in memory. Consolidation in this context refers to the process by which memories are stored

171

into LTM (Dudai, 2002; 2004). If the storage of CRA is strongly consolidated in memory, these initial effects were expected to continue.

Results

0.75

0.70 0.65 0.60 0.55 0.50 0.45 0.40 CRA towardsthe Visual Task 0.35 Easy Hard No Feedback (Block 1) Feedback (Block 1) No Feedback (Blocks 2-10) Feedback (Blocks 2-10)

Figure 18: Cognitive resources allocation to visual task in week 2 (block 1 and blocks 2-10).

Figure 18 illustrates the results of the CRA analysis. In analysing the first block alone (grey lines, solid and broken, in Figure 18), results revealed that there was a significant main effect for CRA training, F(1, 59) = 28.47, p < .001. That is, participants that received CRA training (broken grey line in Figure 18) allocated significantly greater cognitive resources towards the visual task than participants in that did not receive CRA training (46.17% and

63.93% respectively; solid grey line in Figure 18). The results failed to reveal a significant main effect for task difficulty, F (1, 59) = 3.64, p = .06). That is, task difficulty did not affect participant‟s allocation of resources towards the visual task (52.2% for easy and

172

58.90% for hard). The results also failed to reveal a significant interaction effect between the two factors (CRA training and Task Difficulty), F (1, 59) = .005, p = .95.

Analysis of the average performance across of blocks 2-10 also revealed a main effect for CRA training (black lines, solid and broken, in Figure 18), F(1, 59) = 25.15, p <

.001), however failed to reveal a significant main effect for difficulty (F(1, 59) = 1.39, p =

.24) or an interaction effect (F(1, 59) = .009, p = .92). These results indicate that performance was similar between block 1 and blocks 2-10.

7.6 Discussion

The present research aimed to address two research questions. The first aimed to investigate whether individuals could allocate appropriate cognitive resources based on self-assessed task difficulty. The second research question aimed to investigate whether it was possible to train how cognitive resources were allocated towards tasks. In summary, the results indicated that the participants were poor at allocating resources based on task difficulty.

The interpretation of these results is deferred for later discussion (see 7.6.1 Naturalistic

Learning).

The results also indicated that the participants that received CRA training performed significantly different to those that did not receive training. Specifically, participants in the

CRA training condition allocated significantly greater resources towards the visual task compared to those that did not receive CRA training. This effect was statistically significant from the first block of the experiment and continued through to the final block (average of blocks 2-10). The results suggest that based on training in week 1, information regarding 173

how cognitive resources are allocated towards the task/s is stored in memory and this information was later operationalised to aid in task performance in week 2. In reference to the research question, the results suggest that it is possible to train the allocation of cognitive resources.

7.6.1 Naturalistic Learning.

In reference to the first research question, which aimed to investigate whether individuals were able to allocate cognitive resources appropriately, a number of possible explanations exist for the results of the experiment. Recall the results illustrated that there was no significant difference between resource allocation between group 1 and 2 (easy and hard version of task; no CRA training). Further the results indicated that cognitive resources were distributed approximately evenly towards the two tasks (53% and 56% towards the visual task for groups 1 and 2 respectively). Whilst the amount of cognitive resources allocated towards the easy condition (group 1) is not surprising, it is an unexpected result that individuals did not allocate greater resources towards the visual task in the hard condition (group 2).

The most obvious explanation for this result is that the participants in the hard condition (group 2) did not correctly appraise the difficulty of the task in week 1, and thus did not sufficiently allocate cognitive resources to meet the task demands in the following week. This would be despite receiving instruction that if they felt they were underperforming on either of the two tasks, to attempt to allocate more cognitive resources to the task that was lacking. With an inaccurate appraisal of the task difficulty in week 1, performance on the second week is expected to be comparable to group 1. 174

Another explanation relates to individuals‟ memory and the application of CRA information from memory. That is, even if individual recognised and allocated appropriate cognitive resources towards the tasks in week 1, it is possible that this information was either not stored in memory systems to be operationalised in week 2, or the information was appropriately stored but not drawn upon to model behaviour in week 2. With regards to the latter, whilst it is possible that participants were unable to recognise that the tasks were similar, this is deemed an unlikely explanation given the identical surface features of the tasks in both weeks.

Based on the results and the possible explanations, it is apparent that further testing is required to holistically understand the results. Whilst the mechanism behind why cognitive resources were poorly allocated is unclear, the behaviour produced is not. That is, irrespective of how the failure occurred, the hard condition (group 2) did not appropriately allocate sufficient cognitive resources towards the dual-task in week 2.

In the context of driving, the difficulty manipulation was designed to represent the naturalistic learning environment that is commonly implemented on Australian roads.

Specifically, young novice drivers in most licensing systems are not explicitly trained as to how cognitive resources should be operationalised towards the multitude of driving tasks; they are expected to learn this skill over time with experience. The results of the easy condition (group 1) suggest that participants allocated cognitive resources in such a manner that both tasks were performed equally as well, a phenomenon that could be described as an

„equalisation‟ of resource allocation as opposed to prioritisation. In other words, participants aimed to maximise performance (and subsequently minimise error) on both

175

tasks equally. The extension of this is the same strategy would apply irrespective of number of tasks; if there were three equal tasks, cognitive resources would be split into three, though this was evidently not tested. This type of performance seems intuitive as a default strategy; without directive or any reason for prioritisation of either of the two tasks, the individual aims to maximise performance. In a road safety context, the implications of these results suggest that without directive, young novice drivers would aim to maximise performance on all driving tasks through an equalisation strategy.

The results of the hard condition (group 2) suggest that, even when task difficulty is increased in one task over another, young novice drivers are either poor at identifying this increase in task difficulty or are poor at allocating available resources effectively.

Whichever the explanation, the end result is the same and may be a contributing factor to the young novice driver problem. The implication of this is that young novice drivers are forced to allocate cognitive resources as they see fit. This instinctively has major safety implications, as some driving tasks require more cognitive resources than others and empowering young inexperienced drivers to determine which tasks require prioritisation could prove detrimental. Whilst it is acknowledged that the present research was conducted in a simplistic environment and did not involve driving, the results, by extension suggest that driver training which promotes self-assessed allocation of cognitive resources towards tasks may yield little to no differences compared to no training. However this remains untested, and is hence an area for future research.

176

7.6.2 Training of CRA.

The second research question aimed to investigate if it was possible to train the allocation of cognitive resources. The results indicated that those that received CRA training (group 3 and 4) successfully demonstrated that it was possible to train individuals to allocate cognitive resources in a particular manner, compared to those that did not (group 1 and 2).

By extension, these results add weight to the notion that information about how cognitive resources are operationalised and allocated is stored in individuals‟ memory systems.

The success of the CRA training can be attributed to the accurate appraisal of the task prioritisation. This was thought to have occurred through the feedback they received, thereby allowing them to correctly appraise which tasks require greater CRA (to meet task requirements). This type of corrected appraisal of task prioritisation facilitated by explicit feedback is therefore assumed to be an integral component to the training. This training in turn allows for this information to be stored in memory systems, a theme which resonates throughout the present series of experiments. Specifically, conscious and effortful changes in how resources are allocated seem to be a core component to the success of training CRA.

The active processing would also increase the salience of the processing occurring; another factor which is likely to facilitate retention of CRA knowledge.

In a road safety context, the results provide a potential solution to the aforementioned issues as seen in the implicit group. Specifically, this type of training has the potential to train young novice drivers in how to allocate cognitive resources in a specified manner to cognitively manage the task demands of driving. Based on this result, it seems that there is a necessity to specifically identify the physical and cognitive skills

177

drivers require to safely use the roads as opposed to training methods designed to deliver a broad holistic range of basic driver skills. This idea refers back to the idea of part-task training vs. whole-task training (see 6.6.5 Part-Task vs. Whole-Task Training).

7.6.3 Cognitive Mechanism of CRA.

In an attempt to understand the cognitive mechanisms of CRA, it is important to dissect the theory underpinning it. Based on the results of the CRA training conditions, it can be assumed that information about CRA is stored in LTM systems. This assumption is derived from the fact that the memory of the training lasted more than a week. The literature clearly states that this is well past the capability of WM, which is said to decays within 2 minutes

(Shinar, 2007). However, the relationship between memory and other cognitive mechanisms underpins the success of the CRA training remains unknown. Whilst investigating these goes beyond the scope of the present research, it should be noted that their execution requires both lower-order and higher-order cognitive processing, the importance of which is discussed below (see 7.6.7 Lower or Higher Processing to

Implement CRA).

7.6.4 Strategy Development and its Application.

Strategy development is integral to successfully complete these tasks, given their complexity and this idea is well supported in the literature (Geary & Wiley, 1991; Horberry et al., 2006; Young et al., 2003). Over a number of blocks, it appears individuals develop a strategy, which is either aimed at maximising performance or to meet task requirements. As mentioned earlier, the results seem to indicate that the control condition developed a

178

strategy to maximise performance on both the auditory and visual tasks. The results suggest that the strategy employed aimed to maintain a „cognitive equilibrium‟. That is, without any other directive, the strategy employed seems to maintain equal resource allocation, and thus, performance on all tasks. Based on the results of the implicit condition, it appears that a similar strategy was adopted (i.e. to maximise performance), even with a hard task.

The success of CRA training is evident in the way individuals perform and, by natural extension the allocation of their cognitive resources. The feedback provided after each block not only seems to facilitate the change of strategy necessary to meet task requirements but also highlights the importance, and salience of the cognitive mechanisms utilised to meet the task requirements. In addition, the feedback and repetition loop of the task allows for an accurate subjective appraisal of the task‟s objective requirements and difficulty. As such, information about CRA is necessary to effectively implement the strategy learned in week 1.

The most interesting finding comes from the results of week 2. Whilst the strategy for completing the task may still be the same, the individual was not obligated to allocate resources in a similar fashion to week 1. The results, however, demonstrated that rather than reverting to a „cognitive equilibrium‟ (i.e. to maximise performance), individuals who received CRA training modelled behaviours similar to week 1. This could only occur if information about how cognitive resources were operationalised in week 1 is stored in

LTM.

179

7.6.5 Case-Specific vs. Case-General Strategies.

Cognitive strategies employed are said to be largely rule-based (Newell & Simon, 1972).

That is, the information stored about them is not case-specific but rather a general instruction that can be applied in the future to contextually similar tasks (as identified by the individual). The allocation of cognitive resources however, is hypothesised to be case- specific. The assessment of how cognitive resources need to be divided intuitively occurs within the first few trials of any given task. It is assumed that the integration of a case non- specific strategy into memory systems is done with relative ease because the information is generalised and simplistic. However, the integration of information regarding CRA into memory systems is assumed to be case specific.

This explanation neatly accounts for the results of the experimental conditions.

Specifically, whilst those that did not receive CRA training may only have learned a case general cognitive strategy from their week 1 training, those that did received CRA training may have recognised the cognitive demands of the specific tasks. As such, information about how the task was cognitively managed would be stored as a case specific episodic memory. That is, those that received CRA training would not only store information about the general strategy implemented but also store information specific to the task demands.

This recognition of the increased task demand is likely to be encouraged through the feedback provided which helps identify the importance of how the cognitive resources are allocated towards successfully completing the task. This increased salience is assumed to be sufficient for this information to be incorporated into episodic memory systems. Whilst

180

this explanation and assumptions remain untested, this avenue of research could shed light into the potential of transferring CRA training.

7.6.6 Motivation.

Another key factor that is assumed to play a role is the individual‟s motivation towards performance on the tasks. Specifically, the feedback not only provides individuals with corrective information to amend their behaviour but also provides a motivation and direction for improvement. The repeated explicit feedback could possibly serve as a motivational tool for participants to remain engaged in the task. It is possible that motivational factors also play an integral ro le in the salience, and thus storage of knowledge about CRA in the memory of the previous episode.

Participant that did not receive CRA training however, were required to rely on internal motivations to complete the task. Further, their subjective appraisal of their task performance could serve as an implicit feedback mechanism to facilitate future task performance. However, without directive or explicit feedback, the individual has no objective assessment of their performance. Whilst future research should aim to separate motivational factors and the effects of CRA training, this task may prove to be difficult.

Further, it may be necessary for these two factors to be working together to produce the effects demonstrated by the CRA training conditions.

7.6.7 Lower or Higher Processing to Implement CRA.

As mentioned earlier, the task requires the use of lower as well as higher-order cognitive systems. What remains unclear is whether the storage and retrieval of information

181

regarding the allocation of cognitive resources requires higher-order processing or whether this information is accessed through lower-order systems. That is, is the implementation of

CRA training effortful and cognitively taxing or is its processing automated, fluent and resource independent? From an applied training perspective, this question needs further investigation to assess how susceptible this type of training is to manipulation, post- training. Training which requires effortful controlled processing is hypothesised to be more susceptible to other cognitive influence than if the implementation of CRA training is more automated. This hypothesis is supported by Raz, Kirsch, Pollard and Nitkin-Kaner (2006) who state that once a process is automated, it is largely unintentionally initiated and cannot be prevented or stopped.

7.6.8 Limitations.

A few limitations were identified in this research. One limitation which presents itself with all research in the area of cognitive science is the disconnect between the behaviour and the cognitive mechanisms behind it. In particular, for this experiment, error rates were measured by examining performance on the two tasks. The number of errors was assumed to be a reflection of how cognitive resources were allocated. That is, the greater the allocation of cognitive resources, the fewer the errors. But this assumption may not be an accurate representation of how cognitive resources are actually allocated. One way to validate and further strengthen this assumption would be to have other dependent variables that could measure workload in another manner, for example through the use of eye tracking technology and assessing search strategies. However, this limitation will never be completely resolved since cognitive research will always have a gap between the 182

observable behaviour and the cognitive mechanisms behind them. That is, whilst it is theorised that observable behavioural changes are caused by an unobservable cognitive change, it is possible that there are other unobservable constructs that may cause or affect the behaviours produced. This limitation also extends to the formula used to calculate CRA.

That is, the formula was based on the assumption that 100% of cognitive resources were operationalised throughout the task. Whilst it is unlikely that in practice 100% of resources are allocated by an individual at all times, for the purposes of the model and formula, it was a necessary foundational assumption to be used as a frame of reference to explain the behavioural outputs measured during test.

Another criticism, which could be raised about the design of the experiment, could focus on the two tasks used in this experiment. Specifically, the difficulty of the visual task in comparison to the auditory task. It could be argued that the difficulty of these tasks was not equal as there are different cognitive systems used which do not require the same level of cognitive processing. This would be a problem in all conditions if the task was geared towards directing better performance towards the visual task over the auditory. A simple resolution to eliminate this as a variable would be to double the experiment sample and to counterbalance which of the two tasks was given greater allocation during training.

Analysis would reveal whether resource allocation was due to the training or whether task difficulty played a significant part in the effects seen in this experiment. However, this was deemed to be an unnecessary addition of groups and resources, as outlined below.

Components of the aforementioned criticism were an intended part of the design of the experiment. Firstly, the tasks were intentionally selected to be delivered through the two

183

modalities to make sure that perceptual information systems did not overlap. In addition, it ensured that the tasks did not compete for the same resources. That is, the scanning of the visual scene did not affect the individual‟s ability to hear the numbers in the auditory task.

This type of resource competition is well established in the dual and multi-tasking literature

(Salvucci & Taatgen, 2008; Wickens, Lee, Liu, & Becker, 2004).

The criticism regarding the equality of task difficulty was also considered to be of little theoretical significance. Firstly, it was identified that unless the two tasks utilised the same perceptual system (i.e. both tasks were visual or both were auditory), there would be no objective way to control for or assess task equality a priori. In addition, utilising the same perceptual system would bring forth the aforementioned problem of resource competition.

Secondly, it was considered to be of no real theoretical importance if the tasks were not equal. The author felt that the core components necessary for the tasks were that the individual made an accurate assessment of the difficulty of each task and allocated resources appropriately. Irrespective of which of the tasks was more difficult, the individual must make the same assessment and allocate resources towards the task appropriately.

Thirdly, even if the tasks were equated and utilised different perceptual systems, the experiment could not accommodate for individual differences. Specifically, individuals may be naturally better at processing visual information over auditory information (or vice versa) irrespective of task difficulty. Even if tasks were equated, a natural information processing bias between participants could create significant variances within the results.

As mentioned, irrespective of all of these factors, the core component necessary to

184

complete the tasks was to an accurate assessment of the task difficulty and to allocate cognitive resources appropriately to reach the desired level of performance.

Finally, even with all the above mentioned factors possibly contributing to pre- existing task difficulty bias, participants were randomly allocated and exposed to the same level of difficulty, within groups in week 1 and the same level of perceptual processing between groups in week 2. This undoubtedly attributes any differences between groups to the training provided in week 1 and not due to a task difficulty.

Another possible criticism of the experimental design and the conclusions drawn stems from the way in which the experiment was conducted. All participants were informed during recruitment and again at the time of giving consent that the experiment was conducted over two weeks. As a result, participants may have had an added motivation to remember the task and/or strategies used as it may have seemed intuitive that the following week‟s session was similar, or at least related in nature. The problem this may present is that the results from the experiment may have been due to an external motivation from the participants and not due to the internal experimental manipulations.

The author acknowledges the potential influence of this and its potential impact on the experimental results. One way to eliminate this problem would be to deceive the participants by informing them that the second week had no carry-over relation to the first week, prior to the commencement of the experiment. This would ensure that any learning that occurred was due to the motivational factors controlled within the experiment (i.e. the conditions provided) and not due to the external motivations to remember the task.

However, as previously mentioned, since all participants were randomly allocated and all

185

participants were informed equally of the second week‟s testing, the impact of this variable is said to be at best secondary to the primary manipulation for the conditions.

7.6.9 Future Research.

The results of the present research leave a plethora of unanswered question to be explored.

As with all research of this theoretical nature, assessing the same problem in a number of different ways would further validate and give weight to the present results. One method of change worth investigating could attempt to see if the same training effects could be replicated in other perceptual systems, such as tactile systems or other cognitive systems such as working memory, were trained in a similar manner.

The longevity and strength of this training effect would also need to be investigated in future research. Specifically, does this type of training produce lasting effects in the way in which cognitive resources are allocated once an individual receives training? Stemming from this, an area which was not investigated in this research was what would happen once the two tasks were well practiced. It is hypothesised that due to the reduction in cognitive stress the task imposed, residual cognitive resources would become available. Exploring this concept leads to the possibility of sequentially adding more tasks, which can be managed by these residual cognitive resources. The possibility for this cycle to continue and for multiple tasks to be added seems only impeded by information processing limits.

Another important question future research would need to investigate is whether the effects demonstrated by CRA training condition could be seen with more than two tasks. Based on the results and conclusions drawn from the present research, there seems to 186

be reason enough to assume that this would be possible. This question would be of specific interest to a multitude of high-risk industries such as road, rail, aviation and medical professions. Specifically, to be able to train an individual to be able to intuitively allocate resources to high priority tasks, without necessarily sacrificing performance on other subtasks, could prove invaluable to safety and in turn training.

An exciting area of research could also focus on untangling the specific cognitive mechanisms that not only contribute to the success of CRA training. Specifically, this research could examine how these systems are not only operationalised, but also how this information is stored in memory. Unlocking this could lead to the refinement of training methods and has the potential to dramatically change the way training is viewed as a construct. Moreover, this avenue of research has the potential to quantify training at a basic information processing level, and fundamentally redefine what it means to provide training and to have training in a particular domain.

From a training perspective for young novice drivers, it is also of interest to untangle whether the distribution of cognitive resources towards tasks is because, this is the strategy individuals implement by default, or whether it was due to poor assessment of task difficulty. Future research should aim to explore the processing strategies and capacities of young novice drivers.

7.6.10 Implications.

The results of the present research also have the potential to have significant implications in a number of other capacities. At its core, the results of the CRA training conditions suggest that individuals can be trained to allocate cognitive resources in a particular manner 187

towards two tasks. The greater extension of this research would aim to operationalise this type of training in real-world training scenarios.

The genesis of this research was aimed at investigating how young novice drivers allocated cognitive resources towards tasks – that is, to ultimately provide young novice drivers with a quantified method of learning how to allocate cognitive resources towards driving tasks. The results of the research seem to suggest that young novice drivers adopt a default option to maximise performance on all tasks. Whilst over time these skills are refined through experience and tasks of greater priority are correctly appraised, the process of „working it out‟ can have significant consequences in terms of road safety.

Solely dependent on future research in this area, this training would ideally focus on providing drivers with CRA training aimed to cement the fundamental cognitive components of driving. Once these components have been sufficiently practiced and residual cognitive resources are available, further tasks can be added in order to facilitate in the learning of a „second tier‟ of training. This type of training promotes a more sequential structured approach to the training of cognitive driving skill.

Whilst the training was initially conceptualised for the road industry, its applications and the scope of its training potential are far beyond this industry alone. As mentioned, at its core the training has to potential to „wire‟ cognitive processing resources in a particular manner, allowing for greater resource direction towards more important task/s, but not at the expense of neglecting other task/s of less importance. Further, such training has shown significant promise in terms of transferability of skills in aviation and in the military (Gopher, 1996; Gopher et al., 1994; Hart & Battiste, 1992; Shebilske et al.,

188

1992). Thus, this idea could prove invaluable for many other complex safety-orientated industries such as rail as well as medicine.

Finally, this type of training could not only facilitate the quantified learning of

CRA, but could also be used as an assessment tool. That is, it may be possible to specify guidelines or criteria of specific cognitive performance necessary to perform particular tasks. Based on performance, individuals could be assessed as to whether they meet the specified guidelines or criteria. This idea again promotes the idea of redefining training and assessment as constructs in an attempt to quantify them.

189

Chapter 8: General Discussion

8.1 Aim and Rationale of the Study

The primary aim of the present study was to facilitate the development of safe driving practices in young novice drivers by providing better training and education. One theme that has only recently gained significant attention is that resolving the young novice driver problem does not lie solely in developing their physical skill through training (Drummond,

1989; Engstrom et al., 2003; Senserrick & Haworth, 2005). Instead, there seems to be a shift in paradigm where greater focus is on driver cognition, such as decision-making, risk and hazard perception as well as risk management. While many of these skills are taught in other high-hazard industries such as aviation, the effectiveness of training such skills in the road domain is yet to be fully explored. Therefore, the main focus of the present research aimed to specifically develop training and education interventions to facilitate the acquisition of the cognitive skills of driving.

Another factor identified as contributing to the young novice driver problem was the disconnect between the education components of driver licensing and the skill component

(Drummond, 1989; Engstrom et al., 2003; Senserrick & Haworth, 2005). That is, despite acquiring the knowledge required to obtain a driver‟s licence and demonstrating the skills required to handle a motor vehicle, current training places little emphasis on developing insight into the influences necessary for safe driving such as risk perception and recognition of the limits to driving skills (i.e. effective decision-making). Research shows that young novice drivers do not believe anything negative will happen to them (i.e. motor vehicle

190

crash) or that they are not typical of the drivers who end up a statistic (DeJoy, 1992; Ivers et al., 2009).

Within the aviation industry, Molesworth and colleagues (Molesworth et al., 2011;

Molesworth et al., 2003, 2006) had demonstrated repeated success employing a training method to improve pilots‟ risk management. Termed episodic training, they employed a training method which actively engaged pilots in the task and provided them with personalised feedback about their performance; they improved their risk management skills in terms of the minimum altitude pilots descended during a simulated low flying exercise.

The success of this technique is said to be attributed to changes in cognitive knowledge structures known as scripts, which also model driver behaviour.

This was the foundation of the first experiment described in this thesis; to investigate whether the same type of training technique could produce similar results in the road environment to facilitate the management of risk in young novice drivers. The results of this experiment revealed the most superior training method in reducing young novice drivers‟ tendency to speed was an episodic training method. The results revealed that providing individuals with case examples of the dangerous speeding behaviour or providing the same cases in conjunction with information about the violations and consequences was not effective in reducing speeding behaviour. These results are consistent with those from the aviation industry in terms of improving pilots‟ risk management. The results demonstrate the successful application of episodic training within the road environment; addressing the primary aim of the study. The results of experiment 1 however left open the question of how this type of training could be incorporated into a driver‟s existing cognitive

191

functioning. Specifically, would the introduction of this type of behavioural change come at a cognitively taxing cost? The question is of importance from a more global driver training perspective because it investigated the impact of implementing training when cognitive load is high.

As such, the second experiment was designed to examine the impact of an episodic training method on individuals‟ cognitive resources. The experiment implemented a factorial design with two factors: episodic training and a secondary task. The results of this experiment revealed that implementing a speed management strategy produced by episodic training was successful in isolation; however, when performed in conjunction with the secondary task, there was a trade-off in terms of overall performance. From a cognitive perspective, these results indicate that the implementation of episodic training comes at a cognitive cost to performance on other tasks. These cognitive resources consumed are crucial for the safe and successful implementation of the skills and knowledge acquired from training and, without controlling for this impact, individuals are vulnerable to performance detriments. This result adds further insight into the cognitive mechanisms behind the success of episodic training in a bid to validate its application to the road environment. As such, this experiment addresses the core aim of the present study; to develop training and education methods to improve young novice driver risk management.

The results of experiment 2 seemed to highlight that cognitive resources play a vital role in the initial phases of learning. Specifically, how young novice drivers learn how to utilise cognitive resources to successfully and safely manage all the necessary tasks of driving remained unknown. This notion promoted the third experiment which moved to

192

further explore how young novice drivers assessed task difficulty when allocating . Further, the research aimed to investigate whether it was possible to train the allocation of cognitive resources in young novice drivers. The results revealed that when one task is more difficult that the other in week 1, individuals did not apply this knowledge in week 2 but instead opted to evenly distribute cognitive resources in the dual-task exercise rather than allocate based on the demand characteristics of the task. This result alone illustrates that without training, individuals are poor at accurately prioritising tasks, but instead seek to evenly distribute their limited cognitive resources between tasks in an attempt to maintain a

„cognitive equilibrium‟. The results of the research also revealed that cognitive resource allocation can be trained by providing explicit feedback about performance, termed CRA training. That is, young novice drivers were able to be taught how to allocate cognitive resources based on an arbitrary criterion defined by the researcher. The results of this experiment explore avenues of cognitive training using a „back to basics‟ approach and are in line with the core aims of the present study; to develop training to improve young novice driver safety.

8.2 Generalisation of Results

The results from the present study add to the research literature of young novice driver training. Moreover, the research contributes to the ever-growing need for a better understanding of driver cognition. The study demonstrates that young novice drivers‟ decision-making and risk management behaviours could be modified through the use of episodic training. 193

The research however, suggests that the modification and implementation of a speed management behaviour can have a significant impact in the way that cognitive resources are allocated towards driving tasks. Specifically, if cognitive resources are inappropriately allocated whilst driving, safety-critical driving tasks that require greater resources may be neglected whilst other more trivial tasks may be over-resourced. Current driver training programs allow for this to occur where young novice drivers are left to allocate cognitive resources as they deem appropriate and, given their lack for experience, this is unlikely to be the best utilisation of their resources. Importantly through targeted training, these detriments in cognitive performance can be managed, as illustrated in experiment 3. It is proposed that such training (developing cognitive resources allocation) has the potential to „fast-track‟ driver expertise.

From an applied perspective, these results suggest that improvements in young drivers‟ speed management behaviour are possible with the correct training. Importantly, the training needs to be not only focused, but compartmentalised to prevent cognitive overload, and allow appropriate acquisition of the desired skill. This type of training approach is used widely in complex environments such as aviation and medicine

(Helmreich, 2000). In isolation the results question the efficacy of the current GDL systems where there seems to be insufficient development of specific skill sets, particularly in the cognitive domain.

194

8.3 Overview of the Limitations of the Study

All research has inherent limitations and this study was no exception. One of the most significant limitations in the first two experiments was the use of a driving simulation.

There is intuitively a disparity between the effects seen in a simulated drive compared to real driving situations. But the benefit in terms of the near elimination of any risk as well as the cost effectiveness and ease of use makes driving simulators a clear winner on smaller- scale studies such as the present. Further, the results demonstrated by Molesworth and colleagues (Molesworth et al., 2011; Molesworth et al., 2003, 2006) also looked at pilots‟ risk management behaviour in a simulated environment. The question of real interest is whether this disparity is significant. That is, would the differences between a simulated environment and a real world driving environment be sufficient to produce different training effects through observable behaviours? There is a plethora of research that supports the use of simulations and regard it as a valid and safe way of measuring performance under experimental conditions (Dorn & Barker, 2005; Godley, Triggs, & Fildes, 2002;

Kaptein, Theeuves, & Van der Horst, 1996; Reed & Green, 1999).

Another major limitation of the research stems from an issue present in all cognitive research. Due to the nature of cognitive science, only observed behaviours can be measured. Whilst models of cognition attempt to account for how and why these behavioural outputs occur, the fact remains these are only models and until science can bridge this gap it will remain a limitation of all cognitive research. Nonetheless, all three experiments in this study use the most current understandings of human cognition to create

195

experimental hypotheses. The results, as such, are a reflection of the interpretation of what each behavioural output represents in terms of the individuals‟ cognition. Whilst the significance behind the behavioural differences seen in the experiments is unquestionable, the interpretation of their meaning is more open to alternate explanations. The author has attempted to holistically and objectively assess the results and provides the best interpretation that fits the data. However, it is acknowledged that another explanation or interpretation may exist. This limitation can also be clearly improved through further research in the area, which shall be discussed.

8.4 Implications for Road Industry

8.4.1 Failures of GDL.

The implications of the present study address the fundamental problem with current young novice driver education and training programs and are encapsulated by the conclusions drawn by McCartt et al. (2009). They express that whilst systems such as the GDL acknowledge the importance of driver experience in the initial learner phases, these efforts are clearly not enough to reduce young driver fatality rates. Whilst driver education and training seems to “help to accelerate the benefits of experience by developing better skills in novice drivers…there is no evidence these programs reduce crashes” (McCartt et al.,

2009, p. 218; Williams & Ferguson, 2004).

At its core, driver training and education facilitates the acquisition of basic motor control skills. However, currently these methods are “far less successful in developing the

196

more complex skills and judgments needed to drive safely across a wide range of potentially hazardous, and often rare, situations” (McCartt et al., 2009, p. 218).

McCartt et al. (2009) continues by stating that “there remains no proven effective alternative to the learning that occurs incrementally when unsupervised driving begins...” and that the “largest gains from experience occur soon after licensure, and the benefits continue to accrue even after several additional years” (p. 218). The present study has focused on addressing this issue by looking at alternative measures to improve the complex skills necessary to drive. That is, this study demonstrates that facilitating the training and development of young novice drivers‟ cognitive skills may be effective in promoting road safety behaviours of this population, ultimately reducing their crash risk.

8.4.2 Application of Episodic Training.

Episodic training has shown promise in road as well as aviation research and has significant practical implications for road safety. The principal workings of episodic training suggest that, through use of personalised feedback of the faults of a driving episode the individual engages in, faulty script knowledge can be modified to facilitate appropriate scripts to model future driving behaviour. Whilst in the present study the focus was on speed management, there is no reason to suggest that this type of training could not facilitate the correction of other risk behaviours, such as running red lights or narrow gap acceptance.

8.4.3 Compartmentalising Training Programs.

The results of the second experiment suggest that the implementation of training programs, which require significant cognitive resources to operationalise, have the potential to be

197

detrimental towards performance on other tasks. It is proposed that based on the results of this experiment, cognitive overload is a leading contributing factor to young novice driver crash and fatality risk. Further, current driver training and education programs do not in any capacity directly address the cognitive demands of driving, through training or education.

To promote and to incorporate this idea into current GDL systems requires a complete re-modelling where greater emphasis is placed on the development of not only basic driving skills but higher-order cognitive skills as well. Further, as mentioned in the discussion of experiment 2 (see 6.6 Discussion), there should be greater structure to how young novice drivers are required to learn the task necessary to drive. GDL systems should identify and train the core components of driving. This should occur under low cognitive load, allowing for these foundations to be well established from a procedural perspective as well as cognitive resource perspective. Once this phase of learning has been achieved, the

GDL system can introduce increasingly difficult tasks into the existing cognitive architecture already created by the individual.

8.4.4 Assessment of Driver Skill.

Another major failure of current GDL systems identified is the assessment of young novice driver skill when graduating through the GDL phases. Fundamentally, it seems questionable as to whether these assessments are reliable or indeed valid as assessments of an individual‟s capacity to drive safely. Specifically, as with all training and education, the focus should be on the acquisition of skill and knowledge and not on how to pass driving assessments. This potential false positive assessment of driving ability can have major road safety implications and is deemed by the author as being another factor which may 198

contribute to young novice driver crash and fatality risk. In fact, providing an individual with a licence when they do not possess the appropriate skill can actually facilitate young novice drivers‟ overconfidence in their ability; that they are better and more skilful then they actually are.

In line with the proposed re-model of the current GDL systems, assessment of a novice driver‟s ability should not only assess their ability to execute the basic behavioural tasks necessary to drive (such as vehicle handling, speed management, mirror/s checking) but should also assess the acquisition of appropriate cognitive skills. Further, assessment should also determine whether these tasks are well practiced enough that residual cognitive resources exist. The availability of residual resources is not only vital for the introduction of new tasks (such as higher-order decision-making skills) but also allows the execution of correct behaviours under high cognitive stress or emergency situations.

8.4.5 Training of Cognitive Driving Task.

The results of the third experiment have the greatest implications for road safety. In line with the proposed re-model of the GDL system, the results of the third experiment introduce the potential to quantitatively train the cognitive skills of driving. This would need to occur in the same manner as described above; establishing foundational cognitive skills and allowing them to become automated or conducted with relative ease before introducing new tasks. This again allows for the operationalisation of residual cognitive resources to be directed towards the management of the new task, without impeding performance on previously learned tasks. An example to demonstrate how the structure of this type of system would work would be to first train individuals to perform constant 199

hazard assessments whilst maintaining a safe headway. In this situation, a predefined distribution would be calculated as to how cognitive resources should be allocated (e.g.

60% towards hazard assessment and 40% towards headway management). Following satisfactory level of performance of these two tasks, the level of cognitive stress is likely to be reduced and residual resources are likely to be available for use (e.g. 16% towards hazard assessment, 24% towards headway management and 60% residual resources). Note that the consumed resources in this example are still maintained at a 60-40 split, but rather than the split consuming 100% of resources, the same 60-40 split is distribute across only

40% of the total available resources, leaving 60% residually. This 60% would allow for the introduction of a new task – for example, speed management skills. Once again, the goal is to ensure that sufficient cognitive resources are available before introducing a new task.

The results of experiment 3 also open up the possibility for this type of performance correction training to be used as an assessment tool. Specifically, looking at

CRA towards tasks may be used to assess whether resources are being directed appropriately.

This proposed re-model advocates the use of part-task training explored previously

(see 6.6.5 Part-Task vs. Whole-Task Training). Whilst much more research is needed before this training type can be more strongly advocated, the results of the third experiment add weight to part-task training, specifically for driver cognition.

In sum, the GDL system requires greater structure to direct young novice drivers as well as driver trainers on the most effective way to learn the necessary skills for safe driving. These skills should be taught from the ground up, starting with the basic core

200

component and developing and assessing these skills as the individual progresses throughout the learning phase. In addition, these tasks should be taught only when it is deemed that the individual has sufficient cognitive resources to do so. Lastly, GDL systems should also implement assessment tools to not only assess behavioural expertise in driving, but cognitive expertise as well.

8.5 Driving and Cognitive Load

In essence, the results of experiment 3 suggest the near endless potential for the integration of new tasks in the driving domain. Admittedly there is a limit to the processing capacity, but at its core, the results support the notion that with the availability of residual cognitive resources, the introduction and integration of a new task should be achievable (provided it does not exceed cognitive resource limits). It should be noted that as with all new tasks, the introduction will initially consume a large amount of resources but with practice and experience, the individual can learn to incorporate these tasks into existing cognitive workings, thus reducing cognitive load.

However, this notion questions the real dangers of cognitive „distraction‟ on the roads. Specifically, detriments in performance from the use of mobiles phones whilst driving is said to be largely due to the cognitive demands of the tasks (Burns, Parkes,

Burton, & Smith, 2002; Haigney, Taylor, & Westerman, 2000; Reed & Green, 1999).

Importantly, even highly experienced (and skilled) drivers, who are also highly skilled mobile phone users, can have significant cognitive overload issues resulting from problems with trying to integrate these two tasks simultaneously. Whilst there is no doubt that driving 201

whilst talking on a mobile telephone is cognitively taxing, the results of experiment 3 suggest that it could be possible that with training and practice, this can be performed at a level with minimal detriment to safe driving performance. This notion has also been suggested by Richard et al., (2 002) and there has also been previous research which suggests that driver behaviours such as lane maintenance and braking are not significantly affected by telephone use (Briem & Hedman, 1995; Brookhuis, de Vries, & de Waard,

1991) Considering the widespread use of mobile telephones and the increasing use of mobile technology in motor vehicles, it would seem to be a more prudent and sensible strategy to teach motorists to perform these tasks simultaneously than to continue with the blanket outlawing of mobile phone use whilst driving, which is likely to result in high crash rate from illegal practices. Whether science or road authorities are willing to have this discussion remains unknown; however, it can be certain that such a suggestion would be highly contentious.

8.6 Theoretical Implications

The present study not only has major practical applications, but also has vast theoretical implications and the potential for future significant research. Firstly, episodic training has been demonstrated (in simulated environments) to be an effective method in improving risk management in both aviation and road contexts. This opens the possibility for research to investigate the potential for this type of training to be assessed in many other high risk industries, where increases in risk behaviours can have catastrophic consequences. This type of training can be achieved by creating a scenario where an individual‟s risk behaviour 202

is highlighted to them through the use of explicit personalised feedback. It is theorised that based on the first experiment, individuals will modify (and correct) their existing scripts of that risk behaviour (in the right conditions). Industries that could benefit from this type of training include, but are not limited to, medicine, defence, business and finance sectors.

The results of the second experiment also have quite important implications for the training as a construct. The results highlight that without the availability of residual cognitive resources, training in any domain may not be beneficial and may in fact produce performance decrements. The results add to the idea of there being a cognitive trade-off when an individual is pushed to their limits. The research interestingly demonstrates that irrespective of whether a task is well practiced or newly learned, performance is largely dependent on how the individual chooses to utilise available cognitive resources. For training as a construct, the results suggest that it is not only sufficient to introduce a new task into an existing framework but to also highlight its relevance and importance in context to ensure that appropriate cognitive resources are directed towards not only performing the task but incorporating it into existing cognitive architecture.

The results of the third experiment suggest that when given two novel tasks, individuals aim to maximise performance on both tasks by splitting their resources evenly.

This strategy seems to pertain even when one task is objectively more difficult than the other. The results suggest that novice drivers, who lack experiences to draw from, may attempt to perform all tasks as well as they can by distributing resources evenly. This would be despite some tasks requiring more cognitive resources to perform whilst other tasks are

„over-resourced‟.

203

As a solution to this however, the results from the third experiment also suggests that it is possible to train an individual to allocate cognitive resources in a particular manner. As mentioned in the discussion of experiment 3 (see 7.6 Discussion), this type of cognitive training has the potential to redefine the current definitions of what it means to be

„trained‟ within a domain. Specifically, rather than only focusing on the behavioural outputs, cognitive output could also be trained to provide a comprehensive training program.

8.7 Future Research

8.7.1 Exploring Episodic Training.

One of the major areas of research, which still has many unanswered questions, is episodic training. As mentioned, its applicability in real-world scenarios remains untested in both aviation and in road industries. This step is vital for the future viability of this type of training in real-world applications. Following from this, it is also vital to explore the malleability of the training scheme. As mentioned earlier, it would be of great value to assess whether this type of training is effective in reducing risk behaviours not only in the road environment but in other industries where risk behaviours can have significant outcomes. It is also of benefit to explore which aspects of the training contribute to the change in risk behaviours. Specifically, dissecting the training method into its parts would allow for a comprehensive assessment of the components which could be attributed to its

204

success. This would further allow for not only the refinement of training method but also the theory behind its success.

8.7.2 The Gap between Cognition and Behaviour.

As mentioned previously (7.6.8 Limitations), there is an inherent gap between an observed behaviour and the cognitive working behind it. Naturally, future research should aim to better understand this relationship. One possible method is to include other dependent variables to test the same cognitive mechanism. This would add validity to explain the behaviour. In addition, testing the same variables in different contexts and in other domains may further contribute to the cognitive theory.

8.7.3 Feature Identification Training.

One idea raised as an alternate explanation for the effectiveness of episodic training suggested that a critical factor was to ensure that features of the training drive and test drive were similar to ensure that individuals recognised that what was learned in week 1 could be applied in week 2. As discussed earlier, one way to facilitate the abstraction of features could be to provide targeted training. Specifically, by explicitly highlighting critical features of a driving episode, an individual may be able to better recognise the similar features in future driving episodes and apply learned skills and knowledge more efficiently and accurately. Further, this type of training could allow for feature identification to become more quickly processed and facilitate appropriate behaviours becoming more automated.

205

8.7.4 Hierarchical Training.

One exciting avenue of research, which is pivotal to the proposed remodelling of the GDL system, is to systematically determine all driving related tasks and qualify and quantify how these tasks should be learned. Specifically, foundational tasks should be taught first and these should be built upon when appropriate. Importantly, in line with the undercurrent of the study, there should be training in both safety driving behaviours as well as the training of cognitive driving tasks as well. This concept is not original; primary, secondary and tertiary school education is based on building upon established foundations of knowledge.

Further, these types of models have been presented in the driving domain also. The GDE framework, proposed by Keskinen (1996), provides a very basic outline of how driver training should be structured (see Figure 19; adapted from Hatakka, Keskinen, Gregersen,

Glad, & Hernetkoski, 2002). Specifically, Hatakka et al. (1999) suggest that along each of these tiers, training can be addressed through knowledge and skill, reducing risk factors and self-evaluation.

Figure 19. GDE framework proposed by Keskinen, 1996 (adapted from Hatakka et al., 2002).

206

Establishing this type of a hierarchical structured approach to learning not only provides a uniform learning environment for all novice drivers, but also allows for an assessment to be made as to where an individual is in terms of their progression through the learning process.

8.8 Conclusion

Despite all efforts, the young novice driver problem still remains; whilst representing approximately 10% of the driver population, young drivers aged 15–24 years account for over 27% of the driver fatalities (WHO, 2007). According to Australian road authorities, young novice drivers‟ failure to comply with the road rules, namely speeding, is one of the leading contributing factors attributable to this high fatality rate (RTA, 2010; VicRoads,

2010a, b). Within Australia, present initiatives to curb young drivers‟ tendency to speed can be largely divided into two categories: driver education and driver training. Commonly implemented throughout the world is GDL system, which allows for the acquisition of the necessary skills and knowledge to be gained over an extended period of time, usually 2–5 years (Senserrick & Haworth, 2005). In addition, the system imposes restrictions on high- risk practices, which are gradually lifted over the course of their licensure (Williams, 1999;

Williams & Shults, 2010). Whilst the effectiveness of the GDL is unquestionable in terms of reducing young novice driver fatality rates (McCartt et al., 2009), a disproportionate number of fatalities compared to the population clearly indicates that the core problem is not yet resolved or currently being fully addressed. 207

The present study was conducted in an attempt to provide further insight into methods that could provide solutions on how to reduce young novice driver crash risk.

Currently, driver training systems such as the GDL provide drivers with the basic skills necessary to drive but rely largely on experience for drivers to develop the complex higher- order skills necessary to drive. By its very nature, acquiring these skills requires practice and time, both of which occur under high cognitive stress for young novice drivers. This intuitively promotes dangerous driving conditions as well as less than ideal learning condition.

This notion places a direct responsibility on licensing bodies throughout the world to provide young novice drivers with appropriate training and, importantly, the best learning environments to facilitate the acquisition of the necessary skills for safe driving.

Further, young novice driver crashes should not been seen as a sole failure of the driver, but also a failure of licensing authorities to provide the appropriate behavioural and cognitive framework to facilitate driver training and education.

In 2012, 1,310 road users died on Australian roads (BITRE, 2013). Even though this only constitutes 5.75 deaths per 100,000 population, it raises the question of what constitutes an acceptable number of deaths due to road crashes. Whilst crashes and subsequent injury are unavoidable with a human operator, it seems achievable to strive towards a zero fatality rate. Pivotal to achieving this idealistic goal is a thorough and comprehensive training program for drivers, which provides them with all the behavioural and cognitive skills to minimise risk, and maximise safety on the road. The present research

208

aims to place a step in the right direction in promoting training methods for young novice drivers that provide them with these vital skills, which could ultimately save their lives.

209

References

ABS. (2007). Statistical Trends, NSW: Australian Bureau of Statistics.

Adams, J. L. (1968). Measurement of hazard judgement by a stimulus accretion technique.

Safety Research and Education Project. New York, NY: Colombia University

Teachers College.

Adcroft, A. (2011). The mythology of feedback. Higher Education Research &

Development, 30, 405-419.

Agent, K. R., Steenbergen, L., Pigman, J. G., Kidd, P. S., McCoy, C., & Pollack, S. H.

(1998). Impact of the partial graduated license program in Kentucky on teen motor

vehicle crashes. Transportation Research Record, 1779 (pp. 54–61). Washington,

DC: Transportation Research Board.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.

Englewood Cliffs, NJ: Prentice-Hall.

Alba, J. W., & Hasher, L. (1983). Is memory schematic? Psychological Bulletin, 93, 203-

231.

American Academy of Pediatrics. (2006). The Teen Driver. Pediatrics, 118, 2570-2581.

Archer, J. C. (2010). State of the science in health professional education: Effective

feedback. Medical Education, 44, 101-108.

Archer, J., Fotheringham, N., Symmons, M., & Corben, B. (2008). The impact of lowered

speed limits in urban and metropolitan areas. Melbourne, Australia: Monash

University Accident Research Centre.

Aretz, A. J. (1991). The design of electronic map display. Human Factors, 33, 85-101. 210

Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its

control processes. In K. W. Spence & J. T. Spence (Eds.), The psychology of

learning and motivation (Vol. 2). London, UK: Academic Press.

Atkinson, R. C., & Shiffrin, R. M. (1971). The control of short term memory. Scientific

American, 225, 82-90.

Australian Transport Safety Bureau. (2004). Young people and road crashes. Canberra,

Australia: Australian Transport Safety Bureau.

Avolio, B. J., Kroeck, K. G., & Panek, P. E. (1985). Individual Differences in Information

Processing Ability as a Predictor of Motor Vehicle Accidents,. Human Factors, 27,

577-587.

Baddeley, A. D. (1986). Working Memory. Oxford, UK: Clarendon Press.

Baddeley, A. D. (1990). Human Memory: Theory and Practice. Hove, UK: Psychology

Press.

Baddeley, A. D. (1996). Exploring the Central Executive. The Quarterly Journal of

Experimental Psychology, 49A, 5-28.

Baddeley, A. D. (2000). The episodic buffer: A new component of working memory?

Trends in Cognitive Science, 4, 417-423.

Baddeley, A. D., & Hitch, G. J. (1974). Working memory. In G. H. Bower (Ed.), The

psychology of learning and motivation (Vol. 8). London, UK: Academic Press.

Baddeley, A. D., & Hitch, G. J. (2000). Development of Working Memory: Should the

Pascual-Leone and the Baddeley and Hitch Models Be Merged? Journal of

Experimental Child Psychology, 77, 128-137.

211

Baddeley, A. D., & Wilson, B. (2002). Prose recall and amnesia: Implications for the

structure of working memory. Neuropsychologia, 40, 1737-1743.

Bahrick, H. P., Noble, M., & Fitts, P. M. (1954). Extra task performance as a measure of

learning a primary task. Journal of Experimental Psychology, 48, 298-302

Bahrick, H. P., & Shelly, C. (1958). Time sharing as an index of automization. Journal of

Experimental Psychology, 56, 288-293.

Barshi, I., & Healy, A. F. (2002). The effects of mental representation on performance in a

navigation task. Memory & Cognition, 30, 1189-1203.

Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology.

Cambridge, UK: Cambridge University Press.

Bathurst, J. R. (1980). Accident Avoidance Skill Training. Washington, DC: U.S.

Department of Transportation National Highway Traffic Safety Administration.

Beanland, V., Goode, N., Salmon, P. M., & Lenne, M. G. (2013). Is there a case for driver

training? A review of the efficacy of pre- and post-licence driver training. Safety

Science, 51, 127-137.

Bednall, T. C., & Kehoe, E. J. (2011). Effects of self-regulatory instructional aids on self-

directed study. Instructional Science, 39, 205-226.

Begg, D., & Langley, J. (2009). A critical examination of the arguments against raising the

car driver licensing age in New Zealand. Traffic Injury Prevention, 10, 1-8.

Benda, H. V., & Hoyos, C. G. (1983). Estimating hazards in traffic situations. Accident

Analysis & Prevention, 15, 1-10.

212

Benthin, A., Slovic, P., & Severson, H. (1993). A psychometric study of adolescent risk

perception. Journal of Adolescence, 16, 153-168.

Berg, E. A. (1948). A simple objective technique for measuring flexibility in thinking.

Journal of General Psychology, 39, 15-22.

Bird, K. D. (2004). Analysis of Variance via Confidence Intervals. London, UK: Sage

Publications.

Blakemore, S.-J., & Choudhury, S. (2006). Development of the adolescent brain:

implications for executive function and social cognition. Journal of Child

Psychology and Psychiatry, 47, 296-312.

Boase, P., & Tasca, L. (1998). Graduated licensing system evaluation: Interim report 1998.

Toronto, Canada: Ministry of Transportation of Ontario.

Boshuizen, H. P. A., & Schmidt, H. G. (1992). On the role of biomedical knowledge in

clinical reasoning by experts, intermediates and novices. Cognitive Science, 16,

153-184.

Breakwell, G. M. (2000). Risk communication: Factors affecting impact. British Medical

Bulletin, 56, 110-120.

Breakwell, G. M. (2007). The Psychology of Risk. New York, NY: Cambridge University

Press.

Briem, V., & Hedman, L.R. (1995). Behavioural effects of mobile telephone use during

simulated driving. Ergonomics, 38, 2536-2562.

Briggs, G., Peters, G., & Fisher, R. (1972). On the locus of the divided attention effects.

Perception & Psychophysics, 11, 315-320.

213

Brookhuis, K. A., de Vries, G., & de Waard, D. (1991). The effects of mobile telephoning

on driving performance. Accident, Analysis and Prevention, 23, 309-316.

Brown, I. D. (1982). Exposure and experience are a confounded nuisance in research on

driver behaviour. Accident Analysis & Prevention, 14, 345-352.

Brown, I. D., & Copeman, A. K. (1975). Drivers‟ attitudes to the seriousness of traffic

offences considered in relation to the design of sanctions. Accident Analysis &

Prevention, 7, 15-26.

Brown, I. D., & Groeger, J. A. (1988). Risk perception and decision making during the

transition between novice and experienced driver status. Ergonomics, 31, 585-597.

Bureau of Infrastructure Transport and Regional Economics (BITRE). (2009). Cost of road

crashes in Australia 2006. Canberra, Australia: Department of Infrastructure and

Transport.

Bureau of Infrastructure Transport and Regional Economics (BITRE). (2013). Road deaths

Australia –2012 Statistical Summary. Canberra, Australia: Department of

Infrastructure and Transport.

Burns, P. C., Parkes, A., Burton, S., Smith, R. K., & Burch, D. (2002). How dangerous is

driving with a mobile phone? Benchmarking the impairment to alcohol (TRL

Report 547). Crowthorne, UK: Transport Research Laboratory.

Cairney, P. T. (1982). An exploratory study of risk estimates of driving situations.

Australian Road Research Board Proceedings, 11, 233-240.

Catchpole, J. (2005). Learning to take risks: The influence of age and experience on risky

driving (Research Report ARR362). Melbourne, Australia: ARRB Group.

214

Catchpole, J., Cairney, P., & Macdonald, W. (1994). Why are young drivers over-

represented in traffic accidents? Melbourne, Australia: ARRB Group.

Chaudhary, N. K., Williams, A. F., & Nissen, W. (2007). Evaluation and compliance of

passenger restrictions in a graduated driver licensing program. Washington, DC:

National Highway Traffic Safety Administration.

Chen, H. Y., Senserrick, T., Chang, H. Y., Ivers, R. Q., Martiniuk, A. L. C., Boufous, S., &

Norton, R. (2010). Road crash trends for young drivers in New South Wales,

Australia from 1997 to 2007. Traffic Injury Prevention, 11, 8-15.

Chen, L.-H., Baker, S. P., & Li, G. (2000). Carrying passengers as a risk factor for crashes

fatal to 16- and 17-year-old drivers. Journal of the American Medical Association,

283, 1578-1582.

Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Catergorization and representation and

physics problems by experts and novices. Cognitive Science, 5, 121-152.

Christie, R. (1996). Effects of zero blood alcohol content laws on novice driver casualty

accidents in Victoria. Paper presented at the Road Safety Research and Enforcement

Conference, Sydney, Australia.

Christie, R. (2001). The effectiveness of driver training as a road safety measure.

Melbourne, Australia: Royal Automobile Club of Victoria.

Christie, R., & Harrison, W. (2003). Driver training and education programs of the future.

Melbourne, Australia: Royal Automobile Club of Victoria.

Connelly, L. B., & Supangan, R. (2006). The economic cost of road traffic crashes:

Australia, states and territories. Accident Analysis & Prevention, 38, 1087-1093.

215

Cooper, D., Gillen, D., & Atkins, F. (2005). Measuring the impact of passenger restrictions

on new teenage drivers. Accident Analysis & Prevention, 37, 19-23.

Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of

mental storage capacity. Behavioural and Brain Science, 24, 87-114.

Cowen, N., Morey, C. C., AuBuchon, A. M., Zwilling, C. E., & Gilchrist, A. L. (2010).

Seven-year-olds allocate attention like adults unless working memory is overloaded.

Developmental Science, 13, 120-133.

Craik, F. I. M. (2002). Levels of processing: Past, present and future? Memory, 10, 305-

318.

Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: a framework for memory

research. Journal of Verbal Learning and Verbal Behaviour 11, 671-684.

Crettenden, A., & Drummond, A. E. (1994). The Young Driver Problem vs the Young

Problem Driver – A Review and Crash Data Analysis. Melbourne, Australia:

Monash University Accident Research Centre.

Crundall, D., & Underwood, G. (1998). Effects of experience and processing demands on

visual information acquisition in drivers. Ergonomics, 41, 448-458.

Currie, L. (1969). The perception of danger in a stimulated driving task. Ergonomics, 12,

841-849.

Davey, J., Wishart, D., Freeman, J., & Watson, B. (2007) An application of the Driver

Behaviour Questionnaire in an Australian organisational fleet setting.

Transportation Research Part F: Traffic Psychology and Behaviour 10, 11-21.

216

De Craen, S., Twisk, D. A. M., Hagenzieker, M. P., Elffers, H., & Brookhuis, K. A. (2008).

The development of a method to measure speed adaption to traffic complexity:

Idenfitying novice, unsafe, and overconfident drivers. Accident Analysis &

Prevention, 40, 1524-1530.

De Craen, S., Twisk, D. A. M., Hagenzieker, M. P., Elffers, H., & Brookhuis, K. A. (2011).

Do young novice drivers overestimate their driving skills more than experienced

drivers? Different methods lead to different conclusions. Accident Analysis &

Prevention, 43, 1660-1665.

De Pelsmacker, P., & Janssens,W. (2007). The effect of norms, attitudes and habits on

speeding behaviour: Scale development and model building and estimation.

Accident Analysis & Prevention, 39, 6-15.

De Winter, J. C. F., Wieringa, P. A., Kuipers, J., Mulder, J. A., & Mulder, M. (2007)

Violations and errors during simulation-based driver training. Ergonomics, 50, 138-

158.

Deery, H. A. (1999). Hazard and Risk Perception among Young Novice Drivers. Journal of

Safety Research, 30, 225-236.

Deitz, S. R. E., & Thoms, W. E. E. (1991). Pilots, Personality and Performance. New

York, NY: Quorum Books.

DeJoy, D. M. (1989). An attribution theory perspective on alcohol-impaired driving. Health

Education Quarterly, 16, 359-372.

DeJoy, D. M. (1992). An examination of gender differences in traffic accident risk

perception. Accident Analysis & Prevention, 24, 237-246.

217

Delhomme, P. (1991). Comparing one‟s driving with others‟: Assessment of abilities and

frequency of offences – Evidence for a superior conformity of self-bias? Accident

Analysis & Prevention, 23, 493-508.

Department of Infrastructure and Transport. (2011). Community Attitudes to Road Safety –

2011 Survey Report. Canberra, Australia: Author.

Dobson, A., Brown, W., Ball, J., Powers, J., & McFadden, M. (1999). Women drivers‟

behaviour, socio-demographic characteristics and accidents. Accident Analysis and

Prevention, 31, 525-535.

Doherty, S. T., Andrey, J. C., & MacGregor, C. M. (1998). The situational risk of young

drivers: The influence of passengers, time of day and day of week on accident rates.

Accident Analysis & Prevention, 30, 45-52.

Dorn, L., & Barker, D. (2005). The effects of driver training on simulated driving

performance. Accident Analysis & Prevention, 37, 63-69.

Driver, J., (2001). A selective review of selective attention research from the past century.

British Journal of Psychology, 92, 53-78.

Drummond, A. E. (1989). An Overview of Novice Driver Performance Issues. A Literature

Review. Melbourne, Australia: Monash University Accident Research Centre.

Dudai,Y. (2002). Memory from A to Z. Keywords, Concepts and Beyond. Oxford, UK:

Oxford University Press.

Dudai, Y. (2004). The Neurobiology of Consolidations, Or, How Stable is the Engram?

Annual Review of Psychology, 55, 51-86.

Edwards, I., (2005). An evaluation of the enhanced Pass Plus scheme operated by

218

Kirklees Metropolitan Council. In Department of Transport: London, (Eds.), Behavioural

Research in Road Safety 2005: Fifteenth Seminar. Wetherby, UK: DfT

Publications.

Endsley, M. R. (1995). Toward a theory of situation awareness in dynamic systems. Human

Factors, 37, 32-64.

Endsley, M. R. (2000). Theoretical underpinnings of situation awareness: A critical review.

In M. R. Endsley & D. J. Garland, (Eds.), Situation awareness analysis and

measurement. Mahwah, NJ: Erlbaum.

Engstrom, I., Gregersen, N. P., Hernetkoski, K., Keskinen, E., & Nyberg, A. (2003).

Literature Review: Young novice drivers, driver education and training. VTI

rapport 491A. Linkoping, Sweden: Swedish National Road and Transport Research

Institute.

Evan, L., & Wasielewski, P. (1983). Risky driving related to driver and vehicle

characteristics. Accident Analysis & Prevention, 15, 121-136.

Eysenk, M. W., & Keane, M. T. (2005). Cognitive Psychology: A Student's Handbook (5th

ed.). East Sussex, UK: Psychology Press.

Ferguson, S. A. (2003). Other high-risk factors for young drivers – how graduated licensing

does, doesn't, or could address them. Journal of Safety Research, 34, 71-77.

Ferguson, S. A., Teoh, E. R., & McCartt, A. T. (2007). Progress in teenage crash risk

during the last decade. Journal of Safety Research, 38, 137-145.

Field, A. (2009). Discovering Statistics Using SPSS. London, UK: SAGE Publications.

219

Fisher, B. L., Laurie, Glaser, Connerney, K., Pollatsek, A., Duffy, C., & Brock, J. (2002).

Use of a fixed-base driving simulator to evaluate the effects of experience and PC-

based risk awareness training on drivers' decisions. Human Factors, 44, 287-302.

Fohr, S. A., Layde, P. M., & Guse, C. E. (2005). Graduated driver licensing in Wisconsin:

Does it create safer drivers? Wisconsin Medical Journal, 104, 31-36.

Fontana, F. E., Mazzardo, O., Furtado Jr, O., & Gallagher, J. E. (2009). Whole and Part

Practice: A meta-analysis. Perceptual & Motor Skills, 109, 517-530.

Forbes, T. W. (1939). The normal automobile driver as a traffic problem. Journal of

General Psychology, 20, 471-474.

Forsyth, E., Maycock, G., & Sexton, B. (1995). Cohort study of learner and novice drivers:

Part 3, accidents, offences and driving experience in the first three years of driving.

Crowthorne, UK: Transport Research Laboratory.

Foss, R. D., Feaganes, J. R., & Rodgman, E. A. (2001). Initial effects of graduated

licensing on 16-year-old driver crashes in North Carolina. Journal of American

Medical Association, 286, 1588-1592.

Foss, R. D., Goodwin, A. H., McCartt, A. T., & Hellinga, L. A. (2009). Short-term effects

of a teenage driver cell phone restriction. Accident Analysis & Prevention, 41, 419-

424.

Fuller, R. (2000). The Task-Capability Interface model of the driving process. Recherche

Transports Securité, 66, 47-59.

Garden, S., Cornoldi, C., & Logie, R. H. (2002). Visuo-spatial working memory in

navigation. Applied Cognitive Psychology, 16, 35-50.

220

Garrison, T. (2011). Allocating visual attention: how relevance to driving impacts attention

when drivers are distracted. Proceedings of the Sixth International Driving

Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design.

Lake Tahoe, CA.

Geary, D. C., & Wiley, J. G. (1991). Cognitive addition: Strategy choice and speed-of-

processing differences in young and elderly adults. Psychology and Aging, 6, 474-

483.

Godley, S. T., Triggs, T. J., & Fildes, B. N. (2002). Driving simulator validation for speed

research. Accident Analysis & Prevention, 24, 589-600.

Goldenbeld, C., & Hatakka, M. (1999). Voluntary advanced driver training. In S. Siegrist

(Ed.), Driver training, testing and licensing – towards theory-based management of

young drivers‟ injury risk in road traffic: Results of EU-Project GADGET, work

package 3 (pp. 166-184). Berne, Switzerland: Schweizererische Beratungsstelle fur

Unfaliverhutung (BFU).

Goldstein, L. G. (1972). Youthful drivers as a special safety problem. Accident Analysis &

Prevention, 4, 153-189.

Google Inc. (2011). Google Translate, Retrieved on Decemeber , 2010 from

http://translate.google.com.au

Gopher, D. (1992). The skill of attentional control: Aquisition and execution of attention

strategies. In D. Meyer & S. Kornblum (Eds.), Attention and Performance XIV:

Synergies in Experimental Psychology, Artificial Intelligence and Cognitive

Neuroscience. Cambridge, MA: The MIT Press.

221

Gopher, D. (1996). Attentional control: Exploration of the work of an executive controller.

Cognitive Brain Research, 5, 23-38.

Gopher, D., & Kahneman, D. (1971). Individual differences in attention and the prediction

of flight criteria. Perceptual & Motor Skills, 33, 1335-1342.

Gopher, D., Weil, M., & Bareket, T. (1994). Transfer of skill from a computer game trainer

to flight. Human Factors, 36, 1-19.

Grasso, B. C., Rothschild, J. M., Jordan, C. W., & Jayaram, G. (2005). What is the measure

of a safe hospital? Medication errors missed by risk management, clinical staff and

surveyors. Journal of Psychiatric Practice, 11, 268-273.

Greenwald, A. G., McGhee, D. E., & Schartz, J. L. K. (1998). Measuring Individual

Differences in Implicit Cognition: The implicit Association Test. Journal of

Personality and Social Psychology, 74(6), 1464-1480.

Gregersen, N. P. (1993). Driver education in the Nordic countries: Different systems, but

research is common. Nordic Road & Transport Research, 3, 39-41.

Gregersen, N. P. (1995). What should be taught? Basic vehicle control skills or higher

order skills? Paper presented at the New to the Road: Reducing the Risks for Young

Motorists. Proceedings of the First Annual International Conference of the Youth

Enhancement Service, Los Angeles, CA: University of California.

Gregersen, N. P. (1996a). Young car drivers: Why are they over represented in car

accidents? How can driver training improve their situation? Linkoping, Sweden:

Swedish National Road and Transport Research Institute.

222

Gregersen, N. P. (1996b). Young drivers' overestimation of their own skill – an experiment

on the relation between training strategy and skill. Accident Analysis & Prevention,

28, 243-250.

Gregersen, N. P. (1997). Evaluation of 16-years age limit for driver training: First report.

Linkoping, Sweden: Swedish National Road & Transport Research Institute.

Gregersen, N. P. (1999). Overall licensing systems. In S. Siegrist (Ed.), Driver training,

testing and licensing – towards theory based management of young drivers' injury

risk in road traffic. Results of EU-Project GADGET, work package 3. Berne,

Switzerland: Schweizererische Beratungsstelle fur Unfaliverhutung (BFU).

Gregersen, N. P., Berg, H.-Y., Dahlstedt, S., Engstrom, I., Nolen, S., Nyberg, A., Nygaard,

B., & Rimmo, P. A. (2000). Evaluation of the 16-year limit for drievr training –

Final report. Linkoping, Sweden: Swedish National Road and Transport Research

Institute.

Gregersen, N. P., Berg, H.-Y., Engstrom, I., Nolen, S., Nyberg, A., & Rimmo, P. A. (2000).

Sixteen years age limit for learner drivers in Sweden – an evaluation of safety

effects. Accident Analysis & Prevention, 32, 25-35.

Gregersen, N. P., & Bjurulf, P. (1995). Young Novice Drivers: Towards a model of their

accident involvement. Accident Analysis & Prevention, 28, 229-241.

Gregersen, N. P., Nyberg, A., & Berg, H.-Y. (2003). Accident involvement among learner

drivers – An analysis of the consequences of supervised practice. Accident Analysis

& Prevention, 35, 725-730.

223

Groeger, J. A., & Brown, I. D. (1989). Assessing One's Own and Others' Driving Ability:

Influences of Sex, Ages and Experience. Accident Analysis & Prevention, 21, 155-

168.

Gugerty, L., Brooks, J., & Treadaway, C. (2004). Individual differences in situation

awareness for transportation tasks. In. S. Banbury & S. Tremblay (Eds.), Cognitive

approach to situation awareness: Theory, measures and applications. London:

Ashgate.

Gugerty, L., & Tirre, W. C. (2000). Individual differences in situation awareness. In M. R.

Endsley & D. J. Garland (Eds.), Situation awareness analysis and measurement.

Mahwah, NJ: Erlbaum.

Gugerty, L., Rakauskas, M., & Brooks, J. (2004). Effect of remote and in-person verbal

interactions on verbalization rates and attention to dynamic spatial scenes. Accident

Analysis & Prevention, 36, 1029-1043

Guppy, A. (1993). Subjective probability of accident and apprehension in relation to self-

other bias, age, and reported behavior. Accident Analysis & Prevention, 25, 375-

382.

Haigney, D. E., Taylor, R. G., & Westerman, S. J. (2000). Concurrent mobile (cellular)

phone use and driving performance: task demand characteristics and compensatory

processes. Transport Research Part F, 3, 113-121.

Hall, J., & West, R. (1996). The role of formal instruction and informal practice in learning

to drive. Ergonomics, 39, 693-706.

224

Hallmark, S. L., Veneziano, D. A., Falb, S., Pawlovich, M., & Witt, D. (2008). Evaluation

of Iowa‟s graduated driver‟s licensing program. Accident Analysis & Prevention,

40, 1401-1405.

Harbluk, J. L., Noy, Y. I., Trbovich, P. L., & Eizenman, M. (2007). An on-road assessment

of cognitive distraction: Impacts on drivers' visual behaviour and braking

performance. Accident Analysis & Prevention, 39, 372-379.

Hart, S. G., & Battiste, V. (1992). Flight test of a video game trainer. Paper presented at the

36th Annual Meeting of the Human Factors Society, Santa Monica, CA.

Hatakka, M., Keskinen, E., Gregersen, N. P., & Glad, A. (1999). Theories and aims of

education and training measures. In S. Siegrist (Ed.), Driver training, testing and

licensing – towards theory based management of young drivers' injury risk in road

traffic. Results of EU-Project GADGET, Work Package 3. Berne, Switerland:

Schweizererische Beratungsstelle fur Unfaliverhutung (BFU).

Hatakka, M., Keskinen, E., Gregersen, N. P., Glad, A., & Hernetkoski, K. (2002). From

control of the vehicle to personal self-control: Broadening the perspectives to driver

education. Transport Research Part F, 5, 201-215.

Hatfield, J., & Fernandes, R. (2009). The role of risk-propensity in the risky driving of

younger drivers. Accident Analysis & Prevention, 41, 25-35.

Hatfield, J., Fernandes, R., Faunce, G., & Job, R. F. S. (2008). An implicit non-self-report

measure of attitudes to speeding: Development and validation. Accident Analysis &

Prevention, 40, 616-627.

225

Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational

Research, 77, 81-112.

Haworth, N., Kowaldo, N., & Tingvall, C. (2000). Evaluation of pre-driver education

program. Melbourne, Australia: Monash University Accident Research Centre.

Helmreich, R. L. (2000). On error management: lessons from aviation. British Medical

Journal, 320, 781-785.

Henley, G., & Harrison, J. E. (2009). Serious injury due to land transport crashes, Australia

2006–07. Injury research and statistics. Canberra, Australia: Australian Institute of

Health and Welfare.

Hockey, G. R. J., Wastell, D. G., Sauer, J. (1998). Effects of sleep deprivation and user

interface on complex performance: a multilevel analysis of compensatory control.

Human Factors , 40, 233-253.

Hodgdon, J., Bragg, B. W. E., & Finn, P. (1981). Young driver risk-taking research: The

state of the art. National Highway Traffic Safety Administration. Springfield, VA:

National Technical Information Service.

Holland, C. A. (1993). Self-bias in older drivers‟ judgements of accident likelihood.

Accident Analysis & Prevention, 25, 431-441.

Holubowycz, O. T., & McLean, A. J. (1980). Evaluation of a road safety program for

automotive apprentices. Adelaide, Australia: Road Accident Research Unit,

University of Adelaide.

226

Hooper, C. J., Luciana, M., Conklin, H. M., & Yarger, R. S. (2004). Adolescents‟

performance on the development of decision making and ventromedial prefrontal

cortex. Developmental Psychology, 40, 1148-1158.

Horberry, T., Anderson, J., Regan, M. A., Triggs, T. J., & Brown, J. (2006). Driver

distraction: The effect of concurrent in-vehicle task, road environment complexity

and age on driving performance. Accident Analysis & Prevention, 38, 185-191.

Horneman, C. (1993). Driver education and training: A review of literature. Report No.

6/93. Sydney, Australia: Roads and Traffic Authority.

Horrey, W. J., & Wickens, C. D. (2003). Multiple resource modeling of task interference in

vehicle control, hazard awareness and in-vehicle task performance. Proceedings of

the Second International Driving Symposium on Human Factors in Driver

Assessment, Training, and Vehicle Design. Iowa City, IA: Human Factors Research

Program, University of Iowa.

Horrey, W. J., Wickens, C. D., & Consalus, K. P. (2006). Modeling drivers‟ visual

attention allocation while interacting with in-vehicle technologies. Journal of

Experimental Psychology: Applied, 12, 67-86.

Horswill, M. S., & McKenna, F. P. (2004). Drivers‟ hazard perception ability: Situation

awareness on the road. In S. Banbury & S. Tremblay (Eds.), A cognitive approach

to situation awareness: Theory and application. Aldershot, UK: Ashgate.

Hunter, D. R. (2002). Risk Perception and Risk Tolerance in Aircraft Pilots. Washington,

DC: Federal Aviation Administration.

227

Insurance Institute for Highway Safety (2009). Teen Fact Sheet 2008. Arlington, VA:

Insurance Institute for Highway Safety.

International Traffic Safety Data Analysis Group (2012). Road Safety Annual Report 2010.

Paris, France: International Transport Forum.

Isler, R. B., Starkey, N. J., & Sheppard, P. (2011). Effects of higher-order driving skill

training on young, inexperienced drivers‟ on-road driving performance. Accident

Analysis & Prevention, 43, 1818-1827.

Isler, R. B., Starkey, N. J., & Williamson, A. R. (2009). Video-based road commentary

training improves hazard perception of young drivers in a dual task. Accident

Analysis & Prevention, 41, 445-452.

Isreal, J., Chesney, G., Wickens, C. D., & Donchin, E. (1980). P300 and tracking difficulty:

Evidence for a multiple capacity view of attention. Psychophysiology, 17, 259-273.

Ivers, R., Senserrick, T., Boufous, S., Stevenson, M., Chen, H.-Y., Woodward, M., &

Norton, R. (2009). Novice drivers‟ risky driving behavior, risk perception, and crash

risk: Findings from the DRIVE study. American Journal of Public Health, 99,

1638-1644.

Jaeggi, S. M., Buschkuehl, M., Perrig, W. J., & Meier, B. (2010). The concurrent validity

of the N-back task as a working memory measure. Memory, 18, 394-412.

James, W. (1890). Principles of psychology. New York, NY: Holt.

Jenness, J. W., Lattanzio, R. J., O'Toole, M., Taylor, N., Pax, C. (2002). Effects of manual

versus voice-activated dialing during simulated driving. Perceptual & Motor Skills,

94, 363-379.

228

Job, R. F. S. (1990). The application of learning theory to driving confidence: The effect of

age and the impact of random breath testing. Accident Analysis & Prevention, 22,

97-107.

Johannsdottir, K. R., & Herdman, C. M. (2010). The Role of Working Memory in

Supporting Drivers' Situation Awareness for Surrounding Traffic. Human Factors,

52, 663-673

Jonah, B. A. (1986). Accident Risk and Risk-Taking Behaviour amoung Young Drivers.

Accident Analysis & Prevention, 18, 255-271.

Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice Hall

Kahneman, D., Ben-Ishai, R., & Lotan, M. (1973). Relation of a test of attention to road

accidents. Journal of Applied Psychology, 58, 113-115.

Kamp, R. J. A., Dolmas, D. H. J. M., Van Berkel, H. J. M., & Schmidt, H. G. (2013). The

effect of midterm peer feedback on student functioning in problem-based tutorials.

Advances in Health Sciences Education, 18, 199-213.

Kantowitz, B. H., & Knight, J. (1976). Testing tapping time sharing: Auditory secondary

task. Acta Psychologica, 40, 343-362.

Kaptein, N. A., Theeuves, J., & Van der Horst, R. (1996). Driving simulator validity: Some

considerations. Journal of the Transportation Research Board, 1550, 30-36.

Katila, A., Keskinen, E., & Hatakka, M. (1996). Conflicting goals of skid training. Accident

Analysis & Prevention, 28, 785-789.

229

Katila, A., Keskinen, E., Hatakka, M., & Laapotti, S. (2004). Does increased confidence

amoung novice drivers imply a decrease in safety? The effects of skid training on

slippery road accidents. Accident Analysis & Prevention, 36, 543-550.

Keskinen, E. (1996). Why Do Young Drivers Have More Accidents? Bergisch Gladbach,

Germany: Bundesanstalt für Straßenwesen.

Keskinen, E., Hatakka, M., Katila, A., Laapotti, & Peraaho. (1999). Driver training in

Finland. IATSS Research, 23, 78-84.

Keskinen, E., Hatakka, M., Katila, A., & Laapotti, S. (1992). Was the renewal of Driver

Training successful? Turku, Finland: University of Turku.

Komaki, J., Heinzmann, A. T., & Lawson, L. (1980). Effects of Training and Feedback:

Component Analysis of a Behavioural Safety Program. Journal of Applied

Psychology, 65, 261-270.

Koneci, C., Ebbesen, E. B., & Koneci, D. K. (1976). Decision processes and risk taking in

traffic: Driver response to the onset of yellow light. Journal of Applied Psychology,

6, 359-367.

Kramer, F. A., Larish, F. J., & Strayer, D. L. (1995). Training of attentional control in dual

task settings: a comparison of young and old adults. Journal of Experimental

Psychology: Applied, 1, 50-76.

Kuiken, M. J., & Twisk, D. (2001). Safe driving and the training of calibration: A literature

review. Leidschendam, the Netherlands: SWOV Institute for Road Safety Research.

230

Langford, J. (1999). Does Pre-Driver Training Work? Yes, No, Maybe! Paper presented at

the 1999 Road Safety Research, Policing and Education Conference, Canberra,

Australia.

Lajunen, T., & Summala, H. (2003). Can we trust self-reports of driving? Effects of

impression management on driver behaviour questionnaire responses.

Transportation Research, Part F, 6, 97-107.

Langley, J. D., Wagenaar, A. C., & Begg, D. J. (1995). An Evaluation of the New Zealand

Graduated Driver Licensing System. Accident Analysis & Prevention, 28, 139-146.

Larkin, J., McDermott, J., Simon, D. P., & Simon, H. A. (1980). Expert and novice

performance in solving physics problems. Science, 208, 1335-1342.

Lawton, R., Parker, D., Manstead, A. S. R., & Stradling, S. G. (1997). The Role of Affect

in Predicting Social Behaviours: The Case of Road Traffic Violations. Journal of

Applied Social Psychology, 27, 1258-1276.

Leon-Carrion, J., Garcia-Orza, J., & Perez-Santamaria, F. J. (2004). The development of

the inhibitory component of the executive functions in children and adolescents.

International Journal of Neuroscience, 114, 1291-1311.

Lima, M. L., Barnett, J., & Vala, J. (2005). Risk perception and technological development

at a societal level. Risk Analysis, 25, 1229-1239.

Lin, M.-L., & Fearn, K. T. (2003). The provisional license: night-time and passenger

restrictions: A literature review. Journal of Safety Research, 34, 51-61.

Logie, R. H. (1995). Visuo-spatial working memory. Hove, UK: Psychology Press.

231

Logie, R. H., Baddeley, A. D., Mane, A., Donchin, E., & Sheptak, R. (1989). Working

memory in the acquisition of complex cognitive skills. Acta Psychologica, 71, 53-

87.

Lonero, L. P. (1999). Reinventing driver education: strategic directions for Driver Ed in the

21st century. Paper presented at the 1st International Novice Drivers Conference,

Toronto, Canada.

Lowrance, W. W. (1980). The nature of risk. In R. C. Schwing & W. A. Albers Jr (Eds.),

Societal risk assessment. New York, NY: Plenum Press.

Luciana, M., Conklin, H. M., Cooper, C. J., & Yarger, R. S. (2005). The development of

nonverbal working memory and executive control processes in adolescents. Child

Development, 76, 697-712.

Luna, B. (2004). Algebra and the adolescent brain. Trends in Cognitive Science, 8, 437-

439.

Luna, B., Garver, K. E., Urban, T. A., Lazar, N. A., & Sweeney, J. A. (2004). Maturation of

cognitive processes from late childhood to adulthood. Child Development, 75, 1357-

1372.

Lund, A. K., Williams, A. F., & Zodar, P. (1986). High school driver education: Further

evaluation of the DeKalb County study. Accident Analysis & Prevention, 18, 349-

357.

Lynam, D. (1996). Prospects for improving driver training in Europe. In H. Simpson (Ed.),

New to the Road: Reducing the risks for young motorists – Proceedings of the First

232

Annual Symposium of the Youth Enhancement Service (pp. 103-114). Los Angeles,

CA: University of California, Los Angeles.

Lynam, D., & Twisk, D. A. (1995). Car Driver Training and Licensing Systems in Europe

– Report prepared by members of Forum of European Road Research Institutes

(FERSI) and supported by European Commission Transport Directorate (DG VII).

Crowthorne, UK: Transport Research Laboratory (TRL).

Macdonald, W. A. (1994). Young Driver Research Program – A Review of Information on

Young Driver Performance Characteristics and Capacities. Melbourne, Australia:

Monash University Accident Research Centre.

Malaterre, G., Ferrandez, F., Fleury, D., & Lechner, D. (1988). Decision Making in

Emergency Situations. Ergonomics, 31, 643-655.

Martens, M. H. (2000). Assessing road sign perception: A methodological review.

Transportation Human Factors, 2, 347-357.

Masten, S. V., & Hagge, R. A. (2004). Evaluation of California‟s graduated driver licensing

program. Journal of Safety Research, 35, 523-535.

Matthews, G., Desmond, P. A. (2002). Task-induced fatigue effects on simulated driving

performance. The Quarterly Journal of Experimental Psychology, 55A, 659–686.

Matthews, M. L., & Moran, A. R. (1986). Age differences in male drivers' perception of

accident risk: The role of perceived driving ability. Accident Analysis & Prevention,

18, 229-313.

Mayhew, D. R. (2003). The learner's permit. Journal of Safety Research, 34, 35-43.

233

Mayhew, D. R., Fields, M., & Simpson, H. M. (2000). Why 16? Arlington, VA: Insurance

Institute for Highway Safety.

Mayhew, D. R., & Simpson, H. M. (1995). The role of driving experience: Implications for

training and licensing of new drivers. Toronto, Canada: Insurance Bureau of

Canada.

Mayhew, D. R., & Simpson, H. M. (1996). Effectiveness and role of driver education in a

graduated licensing system: Summary. Ottawa, Canada: Traffic Injury Research

Foundation.

Mayhew, D. R., & Simpson, H. M. (2002). The safety value of driver education and

training. Injury Prevention, 8, 3-8.

Mayhew, D. R., Simpson, H. M., & Pak, A. (2003). Changes in collision rates among

novice drivers during the first months of driving. Accident Analysis & Prevention,

35, 683-691.

Mayhew, D. R., Simpson, H. M., Williams, A. F., & Desmond, K. (2003). Specific and

long term effects of Nova Scotia‟s graduated licensing program. Traffic Injury

Preventation, 4, 91-97.

Mayhew, D. R., Simpson, H. M., Williams, A. F., & Ferguson, S. A. (1996). Effectiveness

and Role of Driver Education & Training in a Graduated Licensing System. Journal

of Public Health Policy, 19, 51-67.

Mayhew, D. R., Simpson, H. M., Williams, A. F., & Ferguson, S. A. (1998). Effectiveness

and Role of Driver Education and Training in a Graduated Licensing System.

Journal of Public Health Policy, 19, 51-67.

234

McCarley, J. S., Vais, M., Pringle, H., Kramer, A. F., Irwin, D. E. & Strayer, D. L. (2004).

Conversation disrupts change detection in complex driving scenes. Human Factors,

46, 424-436.

McCartt, A. T., Shabanova, V. I., & Leaf, W. A. (2003) Driving experience, crashes and

traffic citations of teenage beginning drivers. Accident Analysis and Prevention, 35,

311–320.

McCartt, A. T., Mayhew, D. R., Braitman, K. A., Ferguson, S. A., & Simpson, H. M.

(2009). Effects of age and experience on young driver crashes: Review of recent

literature. Traffic Injury Prevention, 10, 209-219.

McCartt, A. T., Teoh, E. R., Fields, M., Braitman, K. A., & Hellinga, L. A. (2010).

Graduated Licensing Laws and Fatal Crashes of Teenage Drivers: A National

Study. Traffic Injury Prevention, 11, 240-248.

McCormick, I. A., Walkey, F. H., & Green, D. E. (1986). Comparative Perception of

Driver Ability – A Confirmation and Expansion. Accident Analysis & Prevention,

18(3), 205-208.

McEvoy, S. P., Stevenson, M. R., McCartt, A. T., Woodward, M., Howarth, R. W.,

Palamara, P., & Cercarelli, L. R. (2005). Role of mobile phones in motor vehicle

crashes resulting in hospital attendance: a case-crossover study. British Medical

Journal. 331, 428-30.

McKenna, F. P., & Crick, J. (1994). Hazard perception in drivers: A methodology for

testing and training. Crowthorne, UK: Transport Research Laboratory.

235

McKenna, F. P., & Horswill, M. S. (1999). Hazard perception and its relevance for driver

licensing. Journal of the International Association of Traffic and Safety Sciences,

23(1), 36-41.

McKenna, F. P., Stanier, R. A., & Lewis, C. (1991). Factors underlying illusory self-

assessment of driving skill in males and females. Accident Analysis & Prevention,

23, 45-52.

McKnight, A. J. (1992). Driver licensing in Victoria. Melbourne, Australia: Monash

University Accident Research Centre.

McKnight, A. J., & Resnick, J. (1993). Youthful Driver at Risk. In K. Young (Ed.),

Workshop to Identify Training Requirements Designed to Reduce Young Driver

Risk Taking and improve Decision Making Skills. Washington, DC: US Dept of

Transportation, National Highway Traffic safety Administration.

McKnight, A. J., & McKnight, A. S. (1993). The effect of cellular phone use upon driver

attention. Accident Analysis and Prevention, 25, 259-265.

McLeod, S. A. (2008). Working Memory, Retrieved Decemeber, 2011 from

http://www.simplypsychology.org/working%20memory.html

Milech, D., Glencross, D., & Hartley, L. (1989). Skill acquisition by young drivers:

Perceiving, interpreting and responding to the driving environment (Report No.

MR4). Canberra, Australia: Federal Office of Road Safety.

Miller, G. A. (1956). The magic number seven, plus or minus two: Some limits on our

capacity for processing information. Psychological Review, 63, 81-93.

236

Milosevic, S., & Gajic, R. (1986). Presentation factors and drivers' characteristics affecting

road-sign registration. Ergonomics, 29, 807-815.

Mesken, J., Lajunen, T., & Summala, H. (2002). Interpersonal violations, speeding

violations and their relation to accident involvement in Finland. Ergonomics, 45,

469-483.

Molesworth, B. R. C., Bennett, L., & Kehoe, E. J. (2011). Promoting learning, memory,

and transfer in a time-constrained, high hazard environment. Accident Analysis &

Prevention, 43, 932-938.

Molesworth, B. R. C., Tsang, M. H., & Kehoe, E. J. (2011). Rehearsal and verbal

reminders in facilitating compliance with safety. Accident Analysis & Prevention,

43, 991-997.

Molesworth, B. R. C., & Wiggins, M. W. (2007). Improving the Retention of Skills during

Training. Proceedings of the 45th Annual SAFE Symposium. Reno, NV: SAFE

Association.

Molesworth, B. R. C., Wiggins, M. W., & O'Hare, D. P. (2003). Personalising risk and

training to improve the risk management behaviour of pilots. Paper presented at the

6th International Australian Aviation Psychology Symposium., Sydney, Australia.

Molesworth, B. R. C., Wiggins, M. W., & O'Hare, D. P. (2006). Improving pilots' risk

assessment skills in low-flying operations: The role of feedback and experience.

Accident Analysis & Prevention, 38, 954-960.

237

Montag, I., & Comrey, A. L. (1987). Internality and Externality as Correlates of

Involvement in Fatal Driving Accidents. Journal of Applied Psychology, 72(3), 339-

343.

Naatanen, R., & Summala, H. (1976). Road User Behaviour and Traffic Accidents. Oxford,

UK: North Holland Publishing.

Nairne, J. S. (2002). Remembering over the short-term: The case against the standard

model. Annual Review of Psychology, 53, 53-81.

Naylor, G., & Briggs, J. (1963). Effects of task complexity and task organization on the

relative efficiency of part and whole training methods. Journal of Experimental

Psychology, 65, 217-224.

Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood, NJ: Prentice

Hall.

Neyens, D. M., & Boyle, L. N. (2008). The influence of driver distraction on the severity of

injuries sustained by teenager drivers and their passengers. Accident Analysis &

Prevention, 40, 254-259.

NHTSA. (2008). 2007 Traffic Safety Annual Assessment – Highlights. Washington, DC:

NHTSA‟s National Center for Statistics and Analysis.

Nyberg, A., & Engstrom, I. (1999). The driver training concept “The Insight” – an

evaluation. Linkoping, Sweden: Swedish National Road & Transport Research

Institute (VTI).

Organization for Economic Co-operation and Development. (2006). Young Drivers: The

road to safety Organization of Economic Cooperation and Development and the

238

European Conference of Ministries of Transport Report ITRD. Paris, France:

OECD Publishing.

Oude Egberink, H., Lourens, P. F., & van der Molen, H. H. (1986). Driving strategies

among younger and older drivers when encountering children. Accident Analysis &

Prevention, 18(4), 315-324.

Paas, F., van Gog, T., & Sweller, J. (2010). Cognitive Load Theory: New

Conceptualizations, Specifications, and Integrated Research Perspectives.

Educational Psychology Review, 22, 115-121.

Parker, D., Manstead, A. R., Stradling, S. G., & Reason, J. T. (1992). Determinants and

intentions to commit driving violations. Accident Analysis and Prevention, 24, 117-

131.

Parker, D., Reason, J. T., Manstead, A. R., & Stradling, S. G. (1995). Driving errors,

driving violations, and accident involvement. Ergonomics, 38, 1036-1048.

Parker, D., West, R.J., Stradling, S.G., & Manstead, A. R. (1995). Behavioural

characteristics and involvement in different types of traffic accident. Accident

Analysis and Prevention, 27, 571-581.

Payne, S., Brownlea, A., & Hall, A. (1984). Evaluation of Quensland defensive driving

course. Canberra, Australia: Federal Office of Road Safety.

Pelz, D. C., & Krupat, E. (1974). Caution profile and driving record of undergraduate

males. Accident Analysis & Prevention, 6, 45-58.

239

Peräaho, M., Keskinen, E., & Hatakka, M. (2003). Driver competence in a hierarchical

perspective: Implications for driver education. Turku, Finland: University of Turku,

Traffic Research..

Petzoldt, T., Weiß, T., Krems, J. F., & Bannert, M. (2011). The development of a cognitive

skills training to support driver education – Experimental validation of theoretical

underpinnings. Paper presented at the Proceedings of the 6th International Driving

Symposium on Human Factors in Driver Assessment, Training, and Vehicle

Design, Lake Tahoe, CA.

Phye, G. D. (1991) Advice and Feedback during Cognitive Training: Effects at Acquisition

and Delayed Transfer. Contemporary Educational Psychology, 16, 87-94.

Phye, G. D. (1989). Schemata training and transfer of an intellectual skill. Journal of

Educational Psychology, 81, 347-352.

Polderack, R. A., Selco, S. L., Field, J. E., & Cohen, N. J. (1999). The relationship between

skill learning and repetition priming: Experimental and computational analyses.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 208-

235.

Porte, M. C., Xeroulis, G., Reznick, R. K., Dubrowski, A. (2007). Verbal feedback from an

expert is more effective than self-accessed feedback about motion efficiency in

learning new surgical skills. The American Journal of Surgery, 193, 105-110.

Prabhakharan, P., & Molesworth, B. R. C. (2011). Repairing Faulty Scripts to Reduce

Speeding Behaviour in Young Drivers. Accident Analysis & Prevention, 43, 1696-

1702.

240

Prabhakharan, P., Molesworth, B. R. C., & Hatfield, J. (2012). Impairment of a Speed

Management Strategy in Young Drivers under High Cognitive Workload. Accident

Analysis & Prevention, 47, 24-29.

Preusser, D. F., Ferguson, S. A., & Williams, A. F. (1998a). The effect of teenage

passengers on fatal crash risk of teenage drivers. Accident Analysis & Prevention,

30, 217-222.

Preusser, D. F., Ferguson, S. A., & Williams, A. F. (1998b). The effect of teenage

passengers on the fatal crash risk of teenage drivers. Accident Analysis &

Prevention, 30, 217-222.

Quimby, A. R. (1988). In-car observation of unsafe driving actions (Report ARR 153).

Berkshire, England: Transport and Road Research Laboratory.

Raz, A., Kirsch, I., Pollard, J., & Nitkin-Kaner, Y. (2006). Suggestion Reduces the Stroop

Effect. Psychological Science, 17, 91-95.

Reason, J. (2000). A cognitive engineering perspective on maintenance errors. In R.

Amalberti & N. Sarter (Eds.), Cognitive Engineering in the Aviation Domain.

Mahwah, NJ: Lawrence Erlbaum Associates.

Reason, J., Manstead, A. S. R., Stradling, S. G., Baxter, J., & Campbell, K. (1990). Errors

and violations on the roads: A real distinction. Ergonomics, 33, 1315-1332.

Reason, J., Parker, D., & Free, R. (1994). Bending the Rules: The Varieties, Origins and

Management of Safety Violations. Leiden, The Netherlands: Rijks Uniiversiteit

Leiden.

241

Redelmeier, D. A., & Tibshirani, R. J. (1997). Association between cellular-telephone calls

and motor vehicle crashes. The New England Journal of Medicine, 336, 453-458.

Reed, M. P., & Green, P. A. (1999). Comparison of driving performance on-road and in a

low-cost simulator using a concurrent telephone dialling task. Ergonomics, 42,

1015-1037.

Regan, M. A., Deery, H. A., & Triggs, T. J. (1998). Training of attentional control in

novice car drivers: a simulator study. Proceedings of the Human Factors and

Ergonomics Society, 42(20), 1452-1456.

Regan, M. A., Deery, H. A., Triggs, T. J., Kowadlo, N., McCormick, L., Tierney, P.,

Tomasevic, N., & Wilson, N. (1998). Investigation into the effectiveness of variable

priority training in enhancing attentional control and risk perception in learner and

probationary drivers. Melbourne, Australia: Monash University Accident Research

Centre.

Regan, M. A., Lee, J. D., & Young, K. L. (2009). Driver Distraction: Theory, Effects and

Mitigation. Boca Raton, FL: CRC Press.

Regan, M. A., Triggs, T. J., & Godley, S. T. (2000). DriveSmart: A CD-ROM Skills

training product for novice car drivers. Paper presented at the Road Safety

Research, Policing and Education Conference, Brisbane, Australia.

Rice, T. M., Peek-Asa, C., & Kraus, J. F. (2004). Effects of the California graduated driver

licensing program. Journal of Safety Research, 35, 375-381.

242

Richard, C. M., Wright, R. D., Ee, C., Prime, S. T., Shimizu, Y., & Vavrik, J. (2002).

Effect of a concurrent auditory task on visual search performance in a driving-

related image-flicker task. Human Factors, 44, 108-119.

Richardson, B., Sorenson, J., & Soderstrom, J. (1987). Explaining the Social and

Psychological Impacts of a Nuclear Power Plant Accident. Journal of Applied

Social Psychology, 17, 16-37.

Road and Traffic Authority. (2010). Road traffic crashes in New South Wales: Statistical

statement for the year ended 31 Decemember 2009. Sydney, Australia: NSW

Centre for Road Safety.

Road and Traffic Authority. (2011). NSW speeding zoning guidelines. Sydney, Australia

NSW Centre for Road Safety.

Roberti, J. W. (2007). A review of behavioral and biological correlates of sensation

seeking. Journal of Research in Personality, 38, 256-279.

Rohrmann, B. (2004). Technological versus socio-psychological risk management. Paper

presented at the 2003 Australian Disaster Conference, published as CD-ROM,

Canberra, Australia.

Roy, M., & Chi, M. T. H. (2005). Self-explanation in a multi-media context. In R. Mayer

(Ed.), Cambridge Handbook of Multimedia Learning. Cambridge, UK: Cambridge

Press.

Rumelhart, D. E., McClelland, J. L., & Group., P. R. (1986). Parallel distributed

processing, Vol. 1: Foundations. Cambridge, MA: MIT Press.

243

Sagberg, F. (2000). Evaluation of the 16-year age limit of driver training: crash risk after

the practical test. Oslo, Norway: Transportøkonomisk institutt.

Salvucci, D. D. (2001). An integrated model of eye movements and visual encoding.

Cognitive Systems Research, 1, 201-220.

Salvucci, D. D. (2002). Modeling driver distraction from cognitive tasks. Proceedings of

the 24th Annual Conference of the Cognitive Science Society. Hillsdale, NJ:

Lawrence Erlbaum Associates.

Salvucci, D. D. (2005). A multitasking general executive for compound continuous tasks.

Cognitive Science, 29, 457-492

Salvucci, D. D., & Taatgen, N. A. (2008). Threaded Cognition: An Integrated Theory of

Concurrent Multitasking. Psychological Review, 115, 101-130.

Schacter, D. L., & Tulving, E. (1994). What are the memory systems of 1994? Cambridge,

MA: MIT Press.

Schacter, D. L., Wagner, A. D., & Buckner, R. L. (2000). Memory systems of 1999. In E.

C. Tulving, F. I. M. (Ed.), The Oxford Handbook of Memory (pp. 627-643). New

York, NY: Oxford University Press.

Schank, R. C., & Abelson, R. P. (1977). Scripts, Plans, Goals and Understanding.

Chichester, UK: Wiley (Halsted Press).

Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information

processing: I. Detection, search, and attention. Psychological Review, 84, 1-66.

Schmidt, R. A., & Wulf, G. (1997). Continuous Concurrent Feedback Degrades Skill

Learning: Implications for Training and Simulation. Human Factors, 39, 509-525.

244

Senserrick, T., & Haworth, N. (2005). Review of literiture regarding national and

international young driver training, licensing and regulatory systems (Report No.

239). Melbourne, Australia: Monash University Accident Research Centre.

Senserrick, T., Kallan, M. J., & Winston, F. K. (2007). Child passenger injury risk in

sibling versus non-sibling teen driver crashes: a US study. Injury Prevention, 13,

207-210.

Senserrick, T., & Swinburne, G. C. (2001). Evaluation of an Insight driver training program

for young drivers. Melbourne, Australia: Monash University Accident Research

Centre.

Senserrick, T., & Whelan, D. (2003). Graduated Driver Licensing: Effectiveness of

Systems & Individual Components. Melbourne, Australia: Monash University

Accident Research Centre.

Shebilske, W. L., Regan, J. W., Arthur, W., & Jordon, J. A. (1992). A dyadic protocol for

training complex skills. Human Factors, 34, 369-374.

Shiffrin, R. M. (1976). Capacity limitations in information processing, attention, and

memory. In W. K. Estes (Ed.), Handbook of learning and cognitive processes (Vol.

4). Hillsdale, NJ: Erlbaum.

Shiffrin, R. M. (1999). 30 years of memory. In C. Izawa (Ed.), On human memory:

Evolution progress and reflection of the 30th anniversary of the Atkinson-Shiffrin

model. Mahwah, NJ: Lawrence Erlbaum Associates.

Shinar, D. (2007). Traffic Safety and Human Behaviour. Oxford, UK: Elsevier.

245

Shinar, D., & Drory, A. (1983). Sign registration in daytime and night-time driving. Human

Factors, 25, 117-122.

Shope, J. T., & Molnar, L. J. (2003). Graduated driver licensing in the United States:

evaluation results from the early programs. Journal of Safety Research, 34, 63-69.

Shope, J. T., & Molnar, L. J. (2004). Michigan‟s graduated driver licensing program:

Evaluation of the first four years. Journal of Safety Research, 35, 337-344.

Simon, F., & Corbett, C. (1996). Road traffic offending, stress, age, and accident history

among male and female drivers. Ergonomics, 39, 757-780.

Simpson, H. M. (2003). The evolution and effectiveness of graduated licensing. Journal of

Safety Research, 34, 25-34.

Slovic, P., Fischhoff, B., & Lichtenstein, S. (1979). Rating the risks. Environment, 21, 14-

39.

Smith, E. E., & Jonides, J. (1999). Storage and executive processes in the frontal lobes.

Science, 283, 1657-1661.

Sokolowska, J., & Pohorille, A. (2000). Models of risk and choice: Challenge or danger.

Acta Psychologica, 104, 339-369.

Soliday, S. M. (1975). Development and preliminary testing of a driving hazard

questionnaire. Perceptual & Motor Skills, 41, 763-770.

Soliday, S. M., & Allen, J. A. (1972). Hazard perception in automobile drivers: Age

differences. Chapel Hill, NC: University of North Carolina Highway Safety

Research Centre.

246

Sowerbutts, T. D. (1975). An analysis, in terms of convictions, of a volunteer, behind the

wheel, advanced driver training course. Sydney, Australia: Traffic Accident

Research Unit.

Sperling, G. (1960). The information that is available in brief visual presentations.

Psychological Monographs, 74, 1-29.

Spicer, R. A. (1964). Human factors in traffic accidents. Final Report (Research Grant No

AC-55). Washington, DC: United States Public Health Service.

Spolander, K. (1982). Inexperienced drivers' behaviour, abilities and attitudes. Linkoping,

Sweden: Swedish National Road and Transport Research Institute.

Srinivasan, R., & Jovanis, P. P. (1997). Effect of in-vehicle route guidance systems on

driver workload and choice of vehicle speed: Findings from a driving simulator

experiment. In Y. I. Noy (Ed.), Ergonomics and safety of intelligent driver

interfaces. Hillsdale, NJ, Lawrence Erlbaum Associates.

Stock, J. R., Weaver, J. K., Ray, H. W., Brink, J. R., & Sadoff, M. G. (1983). Evaluation of

safe performance secondary school driver education curriculum project: Final

report. Washington, DC: Department of Transportation.

Struckman-Johnson, D. L., Lund, A. K., Williams, A. F., & Osborne, D. W. (1989).

Comparative effects of driver improvement programs on crashes and violations.

Accident Analysis & Prevention, 21, 203-215.

Sulin, R. A., & Dooling, D. J. (1974). Intrusion of a thematic idea in retention of prose.

Journal of Experimental Psychology, 103, 255-262.

247

Svenson, O. (1981). Are We All Less Risky and More Skillful than Our Fellow Drivers?

Acta Psychologica, 47, 143-148.

Sweller, J. (1988). Cognitive Load during Problem Solving: Effects on Learning. Cognitive

Science, 12, 257-285.

Sweller, J. (1994). Cognitive Load Theory, Learning Difficulty and Instructional Design.

Learning and Instruction, 4, 295-312.

Tabachnick, & Fidell. (2007). Using Multivariate Statistics (5th ed.). Boston, MA: Allyn

and Bacon.

Teague, R. C., Gittelman, S. S., & Park, O. (1994). A Review of the Literature on Part-Task

and Whole-Task Training and Context Dependency. Alexandria, VA: U.S. Army

Research Institute for the Behavioral and Social Sciences.

Transport for NSW (2010). Road Traffic Crashes in New South Wales – Statistical

Statement for the year ended 31 December 2010. Sydney, Australia: Centre for

Road Safety, Transport for NSW.

Treisman, A. M. (1964). Verbal cues, language, and meaning in selective attention.

American Journal of Psychology, 77, 206-219.

Treisman, A. M., & Gelade, G., (1980). A Feature-Integration Theory of Attention.

Cognitive Psychology, 12, 97-136.

Trick, L. M., Enns, J. T., Mills, J., & Vavrik, J. (2004). Paying attention behind the wheel:

a framework for studying the role of attention in driving. Theoretical Issues in

Ergonomic Science, 5, 385-424.

248

Triggs, T. J. (1981). Signalised Intersections and Driver Performance. Melbourne,

Australia: Road Safety and Traffic Authority.

Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.),

Organisation of memory. London, UK: Academic Press.

Twigger-Ross, C. L., & Breakwell, G. M. (1999). Relating risk experience,

venturesomeness and risk perception. Journal of Risk Research, 2, 73-83.

Twisk, D. A. (1995). Liability of novice drivers and the role of the driver training. In D. R.

Lynam & D. A. Twisk (Eds.), Car driver training and licensing systems in Europe.

Berkshire, UK: Transport Research Laboratory.

Ulmer, R. G., Ferguson, S. A., Williams, A. F., & Preusser, D. F. (2001). Teenage crash

reduction associated with delayed licensure in Connecticut. Journal of Safety

Research, 32, 31-34.

Ulmer, R. G., Preusser, D. F., Williams, A. F., Ferguson, S. A., & Farmer, C. M. (2000).

Effect of Florida‟s graduated licensing program on the crash rate of teenage drivers.

Accident Analysis & Prevention, 32, 527-532.

Ulmer, R. G., Williams, A. F., & Preusser, D. F. (1997). Crash involvements of 16-year old

drivers. Journal of Safety Research, 28, 97-103.

Underwood, G., Crundall, D., & Chapman, P. (2002). Selective searching while driving:

The role of experience in hazard detection and general surveillance. Ergonomics,

45(1), 1-12.

249

Utter, D. (2001). Passenger Vehicle Cell Phone Use Results from the Fall 2000, National

Occupant Protection Use Survey. Washington, D.C.: National Highway Traffic

Safety Administration Research Note.

Vernick, J. S., Li, G., Ogaitis, S., MacKenzie, E. J., Baker, S. P., & Gielen, A. C. (1999).

Effects of high school driver education on motor vehicle crashes, violations, and

licensure. American Journal of Preventative Medicine, 16(1), 40-46.

Wallace, P., Haworth, N., & Regan, M. A. (1989). Best Training Methods For Teaching

Hazard Perception and Responding By Motorcyclists. Report No. 236. Melbourne,

Australia: Monash University Accident Research Centre.

Wasielewski, P. (1984). Speed as A Measure of Driver Risk: Observed Speeds versus

Driver and Vehicle Characteristics. Accident Analysis & Prevention, 16, 89-103.

Watts, G. R., & Quimby, A. R. (1980). Aspects of road layout that affect drivers‟

perception and risk taking. Crowthorne, UK: Transport and Road Research

Laboratory.

WHO (2007). Youth and Road Safety. In T. Toroyan & M. Peden (Eds.). Geneva,

Switzerland: World Health Organization.

WHO (2009). Global Status Report on Road Safety: Time for Action. In T. Toroyan & M.

Peden (Eds.). Geneva, Switzerland: World Health Organization.

WHO (2011). Road traffic injuries (Fact sheet No. 358) Retrieved Decemeber, 2011, from

http://www.who.int/mediacentre/factsheets/fs358/en/

250

Wickens, C. D. (1976). The effects of divided attention on information processing in

tracking. Journal of Experimental Psychology: Human Perception and

Performance, 2, 1-13.

Wickens, C. D. (1980). The structure of attentional resources. In R. Nickerson (Ed.),

Attention and performance VIII. Hillsdale, NJ: Erlbaum.

Wickens, C. D. (1992). Engineering psychology and human performance. (2nd ed.). New

York, NY: Harper Collins.

Wickens, C. D. (2002). Multiple resources and performance prediction. Theoretical Issues

in Ergonomics Science, 3, 159-177.

Wickens, C. D. (2005). Multiple resource time sharing models. In N. Stanton, A. Hedge, K.

Brookhuis, E. Salas, & H. Hendrick (Eds.), Handbook of human factors and

ergonomics methods. Boca Raton, FL: CRC Press.

Wickens, C. D. (2008). Multiple Resources and Mental Workload. Human Factors, 50,

449-455.

Wickens, C. D., & Colcombe, A. (2007). Dual-task performance consequences of imperfect

alerting associated with a cockpit display of traffic information. Human Factors, 49,

839-850.

Wickens, C. D., Dixon, S. R., & Ambinder, M. S. (2006). Workload and automation

reliability in unmanned air vehicles. In N. J. Cooke, H. Pringle, H. Pedersen, & O.

Connor (Eds.), Advances in human performance and cognitive engineering

research: Vol. 7. Human factors of remotely operated vehicles. Amsterdam:

Elsevier.

251

Wickens, C. D., Goh, J., Helleberg, J., Horrey, W., & Talleur, D. A. (2003). Attentional

models of multi-task pilot performance using advanced display technology. Human

Factors, 45, 360-380.

Wickens, C.D., Helleburg, J., Goh, J., Xu, X., & Horrey, W.J. (2001). Pilot Task

Management: Testing an Attentional Expected Value Model of Visual Scanning

(Technical Report ARL-01-14/NASA-01-7). Savoy, IL: University of Illinois,

Aviation Research Lab.

Wickens, C. D., & Horrey W. J. (2009) Models of Attention, Distraction, and Highway

Hazard Avoidance. In M. A. Regan, J. D. Lee & K. L. Young (Eds.), Driver

Distraction: Theory, Effects and Mitigation. Boca Raton, FL: CRC Press.

Wickens, C. D., Lee, J. D., Liu, Y., & Becker, S. E. G. (2004). An Introduction to Human

Factors Engineering (2nd ed.). Upper Saddle River, NJ: Pearson/Prentice Hall.

Wildavsky, A. (1993). The comparative study of risk perception: A beginning. In B. Ruck

(Ed.), Risk is a construct: Perceptions of risk perception. Munich, Germany:

Knesebeck.

Williams, A. F. (1999). Graduated licensing comes to the United States. Injury Prevention,

5, 133-135.

Williams, A. F. (2003). Teenage drivers: Patterns of risk. Journal of Safety Research, 34, 5-

15.

Williams, A. F. (2006). Young driver risk factors: Successful and unsuccessful approaches

for dealing with them and an agenda for the future. Injury Prevention, 12, i4-i8.

252

Williams, A. F. (2009). Licensing age and teenage driver crashes: A review of the

evidence. Traffic Injury Prevention, 10, 9-15.

Williams, A. F., & Ferguson, S. A. (2004). Driver education renaissance? Injury

Prevention, 10, 4-7.

Williams, A. F., Ferguson, S. A., & McCartt, A. T. (2007). Passenger effects on teenage

driving and opportunities for reducing the risks of such travel. Journal of Safety

Research, 38, 381-390.

Williams, A. F., Ferguson, S. A., & Wells, J. K. (2005). Sixteen-years-old drivers in fatal

crashes, United States, 2003. Traffic Injury Prevention, 6, 202-206.

Williams, A. F., Preusser, D. F., & Ferguson, S. A. (1997). Analysis of the fatal crash

involvements of 15-year-old drivers. Journal of Safety Research, 28, 49-54.

Williams, A. F., & Shults, R. A. (2010). Graduated driver licensing research, 2007–present:

A review and commentary. Journal of Safety Research, 41, 77-84.

Williams, A. F., & Wells, J. K. (1995). Deaths of teenagers as motor vehicle passengers.

Journal of Safety Research, 26, 161-167.

Williams, J. C. (1988). A data-based method for assessing and reducing human error to

improve operational performance. Paper presented at the Institute for Electrical and

Electronic Engineers Fourth Conference on Human Factors and Power Plants,

Monterey, CA.

Williamson, A., Lombardi, D. A., Folkard, S., Stutts, J., Courtney, T. K., & Connor, J. L.

(2012). The link between fatigue and safety. Accident Analysis & Prevention, 43,

498-515.

253

Wogalter, M. S., Young, S. I., Brelsford, J. W., & Barlow, T. (1999). The relative

contributions of injury severity and likelihood information on hazard-risk

judgements and warning compliance. Journal of Safety Research, 30(3), 151-162.

Woolley, J. (2000). In-car driver training at high schools: A literature review. Adelaide,

Australia: Transport SA.

Young, K. L., & Regan, M. A. (2007). Driver distraction: A review of the literature. In I. J.

Faulks, M. A. Regan, M. Stevenson, J. Brown, A. Porter & J.D. Irwin (Eds.).

Distracted driving. Sydney, NSW: Australasian College of Road Safety.

Young, K., Regan, M. A., & Hammer, M. (2003). Driver Distraction: A Review of

Literature. Melbourne, Australia: Monash University Accident Research Centre.

Zaidel, D. M., Paalberg, W. T., & Shinar, D. (1979). Indicators of driver inattention. Paper

presented at the 23rd Conference of the American Association for Automotive

Medicine, Morton Grove, IL.

Zoller, U., Pushkin, D. (2007) Matching Higher-Order Cognitive Skills (HOCS) promotion

goals with problem-based laboratory practice in a freshman organic chemistry

course. Chemistry Education Research and Practice, 8(2), 153-171.

Zuckerman, M. (1979). Sensation Seeking: Beyond the Optimal Level of Arousal. Hillsdale,

NJ: Lawrence Erlbaum.

Zuckerman, M., Eysenck, S., & Eysenck, H. J. (1978). Sensation Seeking in England and

America: Cross Cultural, Age, and Sex Comparisons. Journal of Consulting and

Clinical Psychology, 46(1), 139-149.

254

Appendices

Appendix A – Experiment 1: 5 km practice track (STISIM script)

0, ROAD,4,2,1,1,0,3,3,.1,.1,0,-1,-1,-1,2.5,-1,2.5,-1,3,-1,3,0 0, BSAV, 1, .03, Prasannah Experiment 1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 23, 24, 26, 27, 28, 39, 42, 43, 44, 47, 50 0, SIGN, 100, 30, C:\STISIM\Data\Signs\Sp60mph.3ds 0, SIGN, 100, 351, C:\STISIM\Data\Signs\Sp40mph.3ds 0, TREE, 200, 50, 0, 10.67, 30.48, 0 0, BLDG, 182.88, -24.38, H*2;4;5;13;7;9 0, BLDG, 198.12, 24.38, H*2;4;5;13;7;9 0, BLDG, 213.36, -24.38, H*2;4;5;13;7;9 0, BLDG, 228.6, 24.38, H*2;4;5;13;7;9 0, BLDG, 243.84, -24.38, H*2;4;5;13;7;9 0, BLDG, 259.08, 24.38, H*2;4;5;13;7;9 0, BLDG, 274.32, -24.38, H*2;4;5;13;7;9 0, BLDG, 289.56, 24.38, H*2;4;5;13;7;9 0, BLDG, 304.8, -24.38, H*2;4;5;13;7;9 0, BLDG, 320.04, 24.38, H*2;4;5;13;7;9 0, BLDG, 335.28, -24.38, H*2;4;5;13;7;9 0, BLDG, 350.52, 24.38, H*2;4;5;13;7;9 0, BLDG, 365.76, -24.38, H*2;4;5;13;7;9 0, BLDG, 381, 24.38, H*2;4;5;13;7;9 0, BLDG, 396.24, -24.38, H*2;4;5;13;7;9 0, BLDG, 411.48, 24.38, H*2;4;5;13;7;9 0, BLDG, 426.72, -24.38, H*2;4;5;13;7;9 0, BLDG, 441.96, 24.38, H*2;4;5;13;7;9 0, BLDG, 457.2, -24.38, H*2;4;5;13;7;9 0, BLDG, 472.44, 24.38, H*2;4;5;13;7;9 0, BLDG, 487.68, -24.38, H*2;4;5;13;7;9 0, BLDG, 502.92, 24.38, H*2;4;5;13;7;9 0, BLDG, 518.16, -24.38, H*2;4;5;13;7;9 0, BLDG, 533.4, 24.38, H*2;4;5;13;7;9 0, BLDG, 548.64, -24.38, H*2;4;5;13;7;9 0, BLDG, 563.88, 24.38, H*2;4;5;13;7;9 0, BLDG, 594.36, -24.38, H*2;4;5;13;7;9 0, V, 0, 281.94, -9.45, 1, *18~35 0, V, 0, 312.42, -9.45, 1, *18~35 0, V, 0, 327.66, -9.45, 1, *18~35 0, V, 0, 403.86, -9.45, 1, *18~35 0, V, 0, 469.39, -9.45, 1, *1~4;18~35 255

0, V, 0, 487.68, -9.45, 1, *10~12;18~35 0, V, 0, 518.16, -9.45, 1, 19 0, A, 0, 304.8, 9.45, *4~6;18~35 0, A, 0, 350.52, 9.45, *2~4;18~35 0, A, 0, 417.58, 9.45, *1;18~35 0, A, 0, 545.59, 9.45, *2~4;18~35, 12, -3.96, 15.24, 2, 3, -2.74, 15.24, 2 0, A, 0, 466.34, 9.45, *10~12 0, A, 0, 478.54, 9.45, *1~12;-9;-8;-5;18~35 0, A, 0, 505.97, 9.45, *1~12;-9;-8;-5 0, A, 0, 530.35, 9.45, *1~12;-9;-8;-5;18~35 0, A, 0, 542.54, 9.45, *1~12;-9;-8;-5 0, A, 0, 553.21, 9.45, *1~12;-9;-8;-5;18~35 0, V, 0, 652.27, -8.53, 1, *10~12;18~35 0, V, 0, 670.56, -8.53, 1, *10~12;18~35 0, V, 0, 682.75, -8.53, 1, *10~12;18~35 0, V, 0, 694.94, -8.53, 1, *10~12;18~35 0, V, 0, 707.14, -8.53, 1, *10~12;18~35 0, V, 0, 719.33, -8.53, 1, *10~12;18~35 0, V, 0, 774.19, -8.53, 1, *10~12;18~35 0, V, 0, 786.38, -8.53, 1, *10~12;18~35 0, V, 0, 804.67, -8.53, 1, *10~12;18~35 0, V, 0, 822.96, -8.53, 1, *10~12;18~35 0, V, 0, 835.15, -8.53, 1, *10~12;18~35 0, V, 0, 896.11, -8.53, 1, *10~12;18~35 0, V, 0, 914.4, -8.53, 1, *10~12;18~35 0, V, 0, 926.59, -8.53, 1, *10~12;18~35 0, V, 0, 969.26, -8.53, 1, *10~12;18~35 0, A, 0, 691.9, 8.53, *1~12;-9;-8;-5;18~35 0, A, 0, 707.14, 8.53, *1~12;-9;-8;-5;18~35 0, A, 0, 719.33, 8.53, *1~12;-9;-8;-5 0, A, 0, 731.52, 8.53, *1~5;18~35 0, A, 0, 457.2, 8.53, *1~5;18~35 0, A, 0, 487.68, 8.53, *1~5;18~35 0, A, 0, 752.86, 8.53, *1~12;-9;-8;-5;18~35 0, A, 0, 768.1, 8.53, *1~12;-9;-8;-5;18~35 0, A, 0, 810.77, 8.53, *1~12;-9;-8;-5 0, A, 0, 822.96, 8.53, *1~5;18~35 0, A, 0, 835.15, 8.53, *1~5;18~35 0, A, 0, 899.16, 8.53, *1~5;18~35 30.48, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 45.72, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 60.96, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 76.2, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 91.44, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 256

106.68, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 121.92, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 137.16, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 152.4, A, 18.29, 182.88, 1.83, *10~14 152.4, A, 18.29, 213.36, 1.83, *18~27 152.4, A, 18.29, 243.84, 5.49, *18~27 152.4, A, 18.29, 274.32, 1.83, *37~45 152.4, A, 18.29, 304.8, 1.83, *18~27 152.4, A, 18.29, 335.28, 5.49, *37~45 152.4, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 167.64, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 182.88, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 198.12, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 200, TREE, 0, 0, 0, 0, 0, 0 213.36, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 228.6, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 243.84, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 259.08, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 261, LS, 40,0 274.32, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 289.56, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 300, V, 0, 652.27, -8.53, 1, *10~12;18~35 300, V, 0, 670.56, -8.53, 1, *10~12;18~35 300, V, 0, 682.75, -8.53, 1, *10~12;18~35 300, V, 0, 694.94, -8.53, 1, *10~12;18~35 300, V, 0, 707.14, -8.53, 1, *10~12;18~35 300, V, 0, 719.33, -8.53, 1, *10~12;18~35 300, V, 0, 774.19, -8.53, 1, *10~12;18~35 300, V, 0, 786.38, -8.53, 1, *10~12;18~35 300, V, 0, 804.67, -8.53, 1, *10~12;18~35 300, V, 0, 822.96, -8.53, 1, *10~12;18~35 300, V, 0, 835.15, -8.53, 1, *10~12;18~35 300, V, 0, 896.11, -8.53, 1, *10~12;18~35 300, V, 0, 914.4, -8.53, 1, *10~12;18~35 300, V, 0, 926.59, -8.53, 1, *10~12;18~35 300, V, 0, 969.26, -8.53, 1, *10~12;18~35 300, A, 0, 691.9, 8.53, *1~12;-9;-8;-5;18~35 300, A, 0, 707.14, 8.53, *1~12;-9;-8;-5;18~35 300, A, 0, 719.33, 8.53, *1~12;-9;-8;-5 300, A, 0, 731.52, 8.53, *1~5;18~35 300, A, 0, 457.2, 8.53, *1~5;18~35 300, A, 0, 487.68, 8.53, *1~5;18~35 300, A, 0, 752.86, 8.53, *1~12;-9;-8;-5;18~35 300, A, 0, 768.1, 8.53, *1~12;-9;-8;-5;18~35 257

300, A, 0, 810.77, 8.53, *1~12;-9;-8;-5 300, A, 0, 822.96, 8.53, *1~5;18~35 300, A, 0, 835.15, 8.53, *1~5;18~35 300, A, 0, 899.16, 8.53, *1~5;18~35 304.8, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 320.04, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 335.28, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 350.52, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 381, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 402.34, V, 0, 609.6, -8.23, 1, *18~35 409.96, V, 0, 609.6, -8.23, 1, *18~35 411.48, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 411.48, A, 0, 609.6, 8.23, *18~35 417.58, A, 0, 609.6, 8.23, *18~35 419.1, V, 0, 609.6, -8.23, 1, *18~35 426.72, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 432.82, V, 0, 609.6, -8.23, 1, *18~35 441.96, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 445.01, V, 0, 609.6, -8.23, 1, *18~35 457.2, A, 18.29, 182.88, 1.83, *10~14 457.2, A, 18.29, 213.36, 1.83, *18~27 457.2, A, 18.29, 243.84, 5.49, *18~27 457.2, A, 18.29, 274.32, 1.83, *37~45 457.2, A, 18.29, 304.8, 5.49, *18~27 457.2, A, 18.29, 335.28, 5.49, *37~45 457.2, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 472.44, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 484.63, BLDG, 609.6, -14.33, B*6;3;1 487.68, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 493.78, V, 0, 609.6, -8.23, 1, *18~35 502.92, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 505.97, V, 0, 609.6, -8.23, 1, *18~35 518.16, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 518.16, V, 0, 609.6, -8.23, 1, *18~35 533.4, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 536.45, V, 0, 609.6, -8.23, 1, *18~35 548.64, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 548.64, V, 0, 609.6, -8.23, 1, *18~35 563.88, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 563.88, V, 0, 609.6, -8.23, 1, *18~35 579.12, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 700, BLDG, 459.08, 24.38, H*2;4;5;13;7;9 700, BLDG, 474.32, -24.38, H*2;4;5;13;7;9 700, BLDG, 489.56, 24.38, H*2;4;5;13;7;9 258

700, BLDG, 504.8, -24.38, H*2;4;5;13;7;9 700, BLDG, 520.04, 24.38, H*2;4;5;13;7;9 700, BLDG, 535.28, -24.38, H*2;4;5;13;7;9 700, BLDG, 550.52, 24.38, H*2;4;5;13;7;9 700, BLDG, 565.76, -24.38, H*2;4;5;13;7;9 700, BLDG, 581, 24.38, H*2;4;5;13;7;9 700, BLDG, 596.24, -24.38, H*2;4;5;13;7;9 700, BLDG, 611.48, 24.38, H*2;4;5;13;7;9 700, BLDG, 626.72, -24.38, H*2;4;5;13;7;9 700, BLDG, 641.96, 24.38, H*2;4;5;13;7;9 762, A, 18.29, 213.36, 1.83, *18~27 762, A, 18.29, 231.65, 5.49, *10~14 762, A, 18.29, 243.84, 1.83, *18~27 762, A, 18.29, 274.32, 5.49, *10~14 762, A, 18.29, 289.56, 5.49, *10~14 762, A, 18.29, 304.8, 1.83, *1~12 914.4, A, 0, 152.4, 8.23, *18~35 1000.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 1030.48, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1045.72, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1060.96, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1076.2, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1091.44, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1106.68, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1121.92, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1137.16, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1150, TREE, 200, 50, 0, 10.67, 30.48, 0 1152.4, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1152.8, C,0, 0, 200, 0, -3.46063E-03 1167.64, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1182.88, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1198.12, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1213.36, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1228.6, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1243.84, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1259.08, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1274.32, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1289.56, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1304.8, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1320.04, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1335.28, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 1350.4, SIGN, 5, 457.2 1350.52, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 1381, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 259

1458, LS, 60,0 1600, TREE, 0, 0, 0, 0, 0, 0 1892.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 1902.4, C,0, 0, 400, 0, 7.46063E-03 1975, TREE, 200, 50, 0, 10.67, 30.48, 0 2300, V, 0, 652.27, -8.53, 1, *10~12;18~35 2300, V, 0, 670.56, -8.53, 1, *10~12;18~35 2300, V, 0, 682.75, -8.53, 1, *10~12;18~35 2300, V, 0, 694.94, -8.53, 1, *10~12;18~35 2300, V, 0, 707.14, -8.53, 1, *10~12;18~35 2300, V, 0, 719.33, -8.53, 1, *10~12;18~35 2300, V, 0, 774.19, -8.53, 1, *10~12;18~35 2300, V, 0, 786.38, -8.53, 1, *10~12;18~35 2300, V, 0, 804.67, -8.53, 1, *10~12;18~35 2300, V, 0, 822.96, -8.53, 1, *10~12;18~35 2300, V, 0, 835.15, -8.53, 1, *10~12;18~35 2300, V, 0, 896.11, -8.53, 1, *10~12;18~35 2300, V, 0, 914.4, -8.53, 1, *10~12;18~35 2300, V, 0, 926.59, -8.53, 1, *10~12;18~35 2300, V, 0, 969.26, -8.53, 1, *10~12;18~35 2300, A, 0, 691.9, 8.53, *1~12;-9;-8;-5;18~35 2300, A, 0, 707.14, 8.53, *1~12;-9;-8;-5;18~35 2300, A, 0, 719.33, 8.53, *1~12;-9;-8;-5 2300, A, 0, 731.52, 8.53, *1~5;18~35 2300, A, 0, 457.2, 8.53, *1~5;18~35 2300, A, 0, 487.68, 8.53, *1~5;18~35 2300, A, 0, 752.86, 8.53, *1~12;-9;-8;-5;18~35 2300, A, 0, 768.1, 8.53, *1~12;-9;-8;-5;18~35 2300, A, 0, 810.77, 8.53, *1~12;-9;-8;-5 2300, A, 0, 822.96, 8.53, *1~5;18~35 2300, A, 0, 835.15, 8.53, *1~5;18~35 2300, A, 0, 899.16, 8.53, *1~5;18~35 2350, LS, 80,0 2411.48, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2426.72, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2441.96, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2457.2, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2472.44, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2484.63, BLDG, 609.6, -14.33, B*6;3;1 2487.68, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2500, BLDG, 459.08, 24.38, H*2;4;5;13;7;9 2500, BLDG, 474.32, -24.38, H*2;4;5;13;7;9 2500, BLDG, 489.56, 24.38, H*2;4;5;13;7;9 2500, BLDG, 504.8, -24.38, H*2;4;5;13;7;9 260

2500, BLDG, 520.04, 24.38, H*2;4;5;13;7;9 2500, BLDG, 535.28, -24.38, H*2;4;5;13;7;9 2500, BLDG, 550.52, 24.38, H*2;4;5;13;7;9 2500, BLDG, 565.76, -24.38, H*2;4;5;13;7;9 2500, BLDG, 581, 24.38, H*2;4;5;13;7;9 2500, BLDG, 596.24, -24.38, H*2;4;5;13;7;9 2500, BLDG, 611.48, 24.38, H*2;4;5;13;7;9 2500, BLDG, 626.72, -24.38, H*2;4;5;13;7;9 2500, BLDG, 641.96, 24.38, H*2;4;5;13;7;9 2500, BLDG, 657.2, -24.38, H*2;4;5;13;7;9 2500, BLDG, 672.44, 24.38, H*2;4;5;13;7;9 2500, BLDG, 687.68, -24.38, H*2;4;5;13;7;9 2500, BLDG, 702.92, 24.38, H*2;4;5;13;7;9 2500, BLDG, 718.16, -24.38, H*2;4;5;13;7;9 2500, BLDG, 733.4, 24.38, H*2;4;5;13;7;9 2500, BLDG, 748.64, -24.38, H*2;4;5;13;7;9 2500, BLDG, 763.88, 24.38, H*2;4;5;13;7;9 2500, BLDG, 794.36, -24.38, H*2;4;5;13;7;9 2502.92, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2518.16, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2533.4, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2548.64, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2563.88, BLDG, 609.6, 24.38, H*2;4;5;13;7;9 2579.12, BLDG, 609.6, -24.38, H*2;4;5;13;7;9 2892.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 2950, TREE, 0, 0, 0, 0, 0, 0 3152.4, C,0, 0, 300, 0, 3.46063E-03 3275, TREE, 200, 50, 0, 10.67, 30.48, 0 3300, V, 0, 652.27, -8.53, 1, *10~12;18~35 3300, V, 0, 670.56, -8.53, 1, *10~12;18~35 3300, V, 0, 682.75, -8.53, 1, *10~12;18~35 3300, V, 0, 694.94, -8.53, 1, *10~12;18~35 3300, V, 0, 707.14, -8.53, 1, *10~12;18~35 3300, V, 0, 719.33, -8.53, 1, *10~12;18~35 3300, V, 0, 774.19, -8.53, 1, *10~12;18~35 3300, V, 0, 786.38, -8.53, 1, *10~12;18~35 3300, V, 0, 804.67, -8.53, 1, *10~12;18~35 3300, V, 0, 822.96, -8.53, 1, *10~12;18~35 3300, V, 0, 835.15, -8.53, 1, *10~12;18~35 3300, V, 0, 896.11, -8.53, 1, *10~12;18~35 3300, V, 0, 914.4, -8.53, 1, *10~12;18~35 3300, V, 0, 926.59, -8.53, 1, *10~12;18~35 3300, V, 0, 969.26, -8.53, 1, *10~12;18~35 3300, A, 0, 691.9, 8.53, *1~12;-9;-8;-5;18~35 261

3300, A, 0, 707.14, 8.53, *1~12;-9;-8;-5;18~35 3300, A, 0, 719.33, 8.53, *1~12;-9;-8;-5 3300, A, 0, 731.52, 8.53, *1~5;18~35 3300, A, 0, 457.2, 8.53, *1~5;18~35 3300, A, 0, 487.68, 8.53, *1~5;18~35 3300, A, 0, 752.86, 8.53, *1~12;-9;-8;-5;18~35 3300, A, 0, 768.1, 8.53, *1~12;-9;-8;-5;18~35 3300, A, 0, 810.77, 8.53, *1~12;-9;-8;-5 3300, A, 0, 822.96, 8.53, *1~5;18~35 3300, A, 0, 835.15, 8.53, *1~5;18~35 3300, A, 0, 899.16, 8.53, *1~5;18~35 3350, LS, 60,0 3500, BLDG, 459.08, 24.38, H*2;4;5;13;7;9 3500, BLDG, 474.32, -24.38, H*2;4;5;13;7;9 3500, BLDG, 489.56, 24.38, H*2;4;5;13;7;9 3500, BLDG, 504.8, -24.38, H*2;4;5;13;7;9 3500, BLDG, 520.04, 24.38, H*2;4;5;13;7;9 3500, BLDG, 535.28, -24.38, H*2;4;5;13;7;9 3500, BLDG, 550.52, 24.38, H*2;4;5;13;7;9 3500, BLDG, 565.76, -24.38, H*2;4;5;13;7;9 3500, BLDG, 581, 24.38, H*2;4;5;13;7;9 3500, BLDG, 596.24, -24.38, H*2;4;5;13;7;9 3500, BLDG, 611.48, 24.38, H*2;4;5;13;7;9 3500, BLDG, 626.72, -24.38, H*2;4;5;13;7;9 3500, BLDG, 641.96, 24.38, H*2;4;5;13;7;9 3500, BLDG, 657.2, -24.38, H*2;4;5;13;7;9 3500, BLDG, 672.44, 24.38, H*2;4;5;13;7;9 3500, BLDG, 687.68, -24.38, H*2;4;5;13;7;9 3500, BLDG, 702.92, 24.38, H*2;4;5;13;7;9 3500, BLDG, 718.16, -24.38, H*2;4;5;13;7;9 3500, BLDG, 733.4, 24.38, H*2;4;5;13;7;9 3500, BLDG, 748.64, -24.38, H*2;4;5;13;7;9 3500, BLDG, 763.88, 24.38, H*2;4;5;13;7;9 3500, BLDG, 794.36, -24.38, H*2;4;5;13;7;9 3600.4, SIGN, 4, 457.2 3959.08, TREE, 0, 0, 0, 0, 0, 0 4000.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 4002.4, C,0, 0, 400, 0, -7.46063E-03 4300, TREE, 200, 50, 0, 10.67, 30.48, 0 4458, LS, 80,0 5000, ES, 0

262

Appendix B – Experiment 1: 10.5 km training track (STISIM script)

0, BSAV, 1, .03, Prasannah Experiment 1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 23, 24, 26, 27, 28, 39, 42, 43, 44, 47, 50 0, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 0, TREE, 200, 50, 0, 10.67, 30.48, 0 0, SIGN, 100, 60.96, C:\STISIM\Data\Signs\Sp60mph.3ds 0, A, 28.96, 457.2, 1.83, *1~59 60.96, A, 28.96, 457.2, 1.83, *1~59 213.36, A, 28.96, 457.2, 5.49, *1~59 375.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 396.24, A, 28.96, 457.2, 1.83, *1~59 457.2, BLDG, 563.88, -3.05, B21 457.2, BLDG, 579.12, 3.05, B20 472.44, A, 28.96, 457.2, 5.49, *1~59 573.02, A, 28.96, 457.2, 1.83, *1~59 676.66, A, 28.96, 457.2, 1.83, *1~59 756.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 843, LS, 40,0 883.92, A, 28.96, 457.2, 5.49, *1~59 975.36, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 975.36, TREE, 0, 0, 0, 0, 0, 0 1000.4, SIGN, 5, 457.2 1100.24, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 1100.24, TREE, 200, 50, 0, 10.67, 30.48, 0 1219.2, A, 28.96, 457.2, 5.49, *1~59 1224, LS, 60,0 1341.12, A, 28.96, 457.2, 5.49, *1~59 1615.44, A, 28.96, 457.2, 1.83, *1~59 1652.8, C,0, 0, 500, 0, 5.46063E-03 1828.8, A, 28.96, 457.2, 5.49, *1~59 1920.24, A, 28.96, 457.2, 1.83, *1~59 2164.08, A, 28.96, 457.2, 1.83, *1~59 2350.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 263

2407.92, BLDG, 548.64, -3.05, B26 2407.92, BLDG, 548.64, 3.05, B17 2407.92, BLDG, 598.64, -3.05, B5 2407.92, BLDG, 598.64, 3.05, B12 2590.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 2743.2, A, 28.96, 457.2, 5.49, *1~59 2818, LS, 40,0 2825.4, SIGN, 5, 457.2 2895.6, A, 28.96, 457.2, 1.83, *1~59 2926.08, TREE, 0, 0, 0, 0, 0, 0 3050, TREE, 200, 50, 0, 10.67, 30.48, 0 3058, LS, 80,0 3248, A, 28.96, 457.2, 5.49, *1~59 3278.48, A, 28.96, 457.2, 1.83, *1~59 3369.92, A, 28.96, 457.2, 1.83, *1~59 3452.8, C,0, 0, 300, 0, 2.46063E-03 3461.36, A, 28.96, 457.2, 1.83, *1~59 3575.4, SIGN, 4, 457.2 3705.2, A, 28.96, 457.2, 5.49, *1~59 3796.64, A, 28.96, 457.2, 1.83, *1~59 4010, A, 28.96, 457.2, 5.49, *1~59 4040.48, A, 28.96, 457.2, 5.49, *1~59 4162.4, A, 28.96, 457.2, 1.83, *1~59 4175.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 4214.8, C,0, 0, 400, 0, -7.46063E-03 4284.32, A, 28.96, 457.2, 5.49, *1~59 4375.76, A, 28.96, 457.2, 1.83, *1~59 4472, BLDG, 940.08, -3.05, G25 4472, BLDG, 940.08, 3.05, G29 4472, BLDG, 990, -3.05, G23 4472, BLDG, 990, 3.05, G21 4497.68, A, 28.96, 457.2, 5.49, *1~59 4619.6, A, 28.96, 457.2, 1.83, *1~59 4643, LS, 80,0 4711.04, A, 28.96, 457.2, 5.49, *1~59 4825.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 4924.4, A, 28.96, 457.2, 1.83, *1~59 5075.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 5292.4, LS, 40,0 5370, TREE, 0, 0, 0, 0, 0, 0 5375.4, SIGN, 4, 457.2 5500, TREE, 200, 50, 0, 10.67, 30.48, 0 5542.4, LS, 60,0

264

5772, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 152.4, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0 5914.8, C,0, 0, 200, 0, -2.46063E-03 6076.8, A, 28.96, 457.2, 1.83, *1~59 6525, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 6534, A, 28.96, 457.2, 5.49, *1~59 6629.68, BLDG, 457.2, -3.05, B3 6629.68, BLDG, 457.2, 3.05, B10 6629.68, BLDG, 487.68, -3.05, B15 6629.68, BLDG, 487.68, 3.05, B8 6629.68, BLDG, 527.68, -3.05, B7 6629.68, BLDG, 527.68, 3.05, B5 6655.92, A, 28.96, 457.2, 5.49, *1~59 6719.93, A, 28.96, 457.2, 5.49, *1~59 6800.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 6960.72, A, 28.96, 457.2, 1.83, *1~59 6992.2, LS, 40,0 7025.6, TREE, 0, 0, 0, 0, 0, 0 7072.64, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 7200, TREE, 200, 50, 0, 10.67, 30.48, 0 7204.56, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 7204.56, A, 28.96, 457.2, 1.83, *1~59 7210.4, SIGN, 5, 457.2 7267.2, LS, 80,0 7409.36, A, 28.96, 457.2, 5.49, *1~59 7752.8, C,0, 0, 400, 0, 7.46063E-03 7771.49, A, 28.96, 457.2, 1.83, *1~59 7910.4, SIGN, 4, 457.2 8475.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 8514.8, C,0, 0, 200, 0, -5.46063E-03 8942.4, LS, 80,0 9450.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 9458, A, 28.96, 457.2, 5.49, *1~59 9600.64, BLDG, 487.2, -3.05, B15 9600.64, BLDG, 487.2, 3.05, B11 9600.64, BLDG, 517.68, -3.05, B12 9600.64, BLDG, 517.68, 3.05, B14 9610.4, A, 28.96, 457.2, 1.83, *1~59 265

9742, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 9747.56, A, 28.96, 457.2, 5.49, *1~59 9917.4, LS, 40,0 10050.6, TREE, 0, 0, 0, 0, 0, 0 10067.6, A, 28.96, 457.2, 1.83, *1~59 10130, TREE, 200, 50, 0, 10.67, 30.48, 0 10209.2, LS, 60,0 10500, ES

266

Appendix C – Experiment 1: 21 km test track (STISIM script)

0, BSAV, 1, .03, Prasannah Experiment 1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 23, 24, 26, 27, 28, 39, 42, 43, 44, 47, 50 0, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 0, TREE, 200, 50, 0, 10.67, 30.48 0, SIGN, 100, 60.96, C:\STISIM\Data\Signs\Sp60mph.3ds 0, A, 28.96, 457.2, 1.83, *1~59; -5 60.96, A, 28.96, 457.2, 1.83, *1~59; -5 213.36, A, 28.96, 457.2, 5.49, *1~59; -5 375.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 385.8. LS, 40, 457.2 396.24, A, 28.96, 457.2, 1.83, *1~59; -5 457.2, BLDG, 563.88, -3.05, B21 457.2, BLDG, 579.12, 3.05, B20 457.2, BLDG, 661.42, -3.05, B18 457.2, BLDG, 658.37, 3.05, B22 472.44, A, 28.96, 457.2, 5.49, *1~59; -5 573.02, A, 28.96, 457.2, 1.83, *1~59; -5 676.66, A, 28.96, 457.2, 1.83, *1~59; -5 766.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 883.92, A, 28.96, 457.2, 5.49, *1~59; -5 975.36, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 975.36, TREE, 0, 0, 0, 0, 0 1000.4, SIGN, 5, 457.2 1158.24, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 1158.24, TREE, 200, 50, 0, 10.67, 30.48 1219.2, A, 28.96, 457.2, 5.49, *1~59; -5 1234, LS, 60,0 1341.12, A, 28.96, 457.2, 5.49, *1~59; -5 1615.44, A, 28.96, 457.2, 1.83, *1~59; -5 1652.8, C,0, 0, 400, 0, 4.46063E-03 1828.8, A, 28.96, 457.2, 5.49, *1~59; -5 1920.24, A, 28.96, 457.2, 1.83, *1~59; -5 267

2164.08, A, 28.96, 457.2, 1.83, *1~59; -5 2350.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 2407.92, BLDG, 548.64, -3.05, B19 2407.92, BLDG, 548.64, 3.05, B17 2570.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 2743.2, A, 28.96, 457.2, 5.49, *1~59; -5 2818, LS, 40,0 2825.4, SIGN, 4, 457.2 2895.6, A, 28.96, 457.2, 1.83, *1~59; -5 2970, TREE, 0, 0, 0, 0, 0 3020, TREE, 200, 50, 0, 10.67, 30.48 3038, LS, 60,0 3248, A, 28.96, 457.2, 5.49, *1~59; -5 3278.48, A, 28.96, 457.2, 1.83, *1~59; -5 3369.92, A, 28.96, 457.2, 1.83, *1~59; -5 3452.8, C,0, 0, 304.8, 0, -8.46063E-03 3461.36, A, 28.96, 457.2, 1.83, *1~59; -5 3575.4, SIGN, 5, 457.2 3705.2, A, 28.96, 457.2, 5.49, *1~59; -5 3796.64, A, 28.96, 457.2, 1.83, *1~59; -5 4010, A, 28.96, 457.2, 5.49, *1~59; -5 4040.48, A, 28.96, 457.2, 5.49, *1~59; -5 4162.4, A, 28.96, 457.2, 1.83, *1~59; -5 4175.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 4214.8, C,0, 0, 304.8, 0, 8.46063E-03 4284.32, A, 28.96, 457.2, 5.49, *1~59; -5 4375.76, A, 28.96, 457.2, 1.83, *1~59; -5 4497.68, A, 28.96, 457.2, 5.49, *1~59; -5 4572, BLDG, 840.08, -4.57, G21 4619.6, A, 28.96, 457.2, 1.83, *1~59; -5 4643, LS, 60,0 4711.04, A, 28.96, 457.2, 5.49, *1~59; -5 4825.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 4924.4, A, 28.96, 457.2, 1.83, *1~59; -5 5050.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 5292.4, LS, 40,0 5375.4, SIGN, 4, 457.2 5375, TREE, 0, 0, 0, 0, 0 5425, TREE, 200, 50, 0, 10.67, 30.48 5517.4, LS, 80,0 5772, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 152.4, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0 5914.8, C,0, 0, 400, 0, -4.46063E-03 6076.8, A, 28.96, 457.2, 1.83, *1~59; -5 268

6525, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 6534, A, 28.96, 457.2, 5.49, *1~59; -5 6640.68, BLDG, 457.2, -3.05, B15 6640.68, BLDG, 457.2, 3.05, B13 6640.68, BLDG, 487.68, -3.05, B14 6640.68, BLDG, 487.68, 3.05, B8 6640.68, BLDG, 518.16, -3.05, B7 6640.68, BLDG, 518.16, 3.05, B2 6640.68, BLDG, 548.64, -6.1, B7 6640.68, BLDG, 548.64, 6.1, B5 6655.92, A, 28.96, 457.2, 5.49, *1~59; -5 6719.93, A, 28.96, 457.2, 5.49, *1~59; -5 6800.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 6960.72, A, 28.96, 457.2, 1.83, *1~59; -5 6992.2, LS, 40,0 7075, TREE, 0, 0, 0, 0, 0 7082.64, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 7200, TREE, 200, 50, 0, 10.67, 30.48 7204.56, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 7204.56, A, 28.96, 457.2, 1.83, *1~59; -5 7210.4, SIGN, 5, 457.2 7267.2, LS, 80,0 7409.36, A, 28.96, 457.2, 5.49, *1~59; -5 7752.8, C,0, 0, 304.8, 0, 8.46063E-03 7771.49, A, 28.96, 457.2, 1.83, *1~59; -5 7910.4, SIGN, 4, 457.2 8475.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 8514.8, C,0, 0, 304.8, 0, -8.46063E-03 8942.4, LS, 80,0 9400.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 9400.64, SOBJ, 609.6, 0, 0,0,0,0, C:\STISIM\Data\Buildings\TownBlock_1.Lmm 9458, A, 28.96, 457.2, 5.49, *1~59; -5 9610.4, A, 28.96, 457.2, 1.83, *1~59; -5 9650.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 9747.56, A, 28.96, 457.2, 5.49, *1~59; -5 9867.4, LS, 40,0 9905, TREE, 0, 0, 0, 0, 0 10067.6, A, 28.96, 457.2, 1.83, *1~59; -5 269

10075, TREE, 200, 50, 0, 10.67, 30.48 10117.4, LS, 60,0 10500, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 10500, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 10500, A, 28.96, 457.2, 1.83, *1~59; -5 10560.96, A, 28.96, 457.2, 1.83, *1~59; -5 10713.36, A, 28.96, 457.2, 5.49, *1~59; -5 10854.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 10864.8. LS, 40, 457.2 10896.24, A, 28.96, 457.2, 1.83, *1~59; -5 10957.2, BLDG, 563.88, -3.05, B21 10957.2, BLDG, 579.12, 3.05, B20 10957.2, BLDG, 661.42, -3.05, B18 10957.2, BLDG, 658.37, 3.05, B22 10972.44, A, 28.96, 457.2, 5.49, *1~59; -5 11073.02, A, 28.96, 457.2, 1.83, *1~59; -5 11176.66, A, 28.96, 457.2, 1.83, *1~59; -5 11266.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 11383.92, A, 28.96, 457.2, 5.49, *1~59; -5 11475.36, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 11480, TREE, 0, 0, 0, 0, 0 11500.4, SIGN, 5, 457.25 11558.24, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 11657, TREE, 200, 50, 0, 10.67, 30.48 11719.2, A, 28.96, 457.2, 5.49, *1~59; -5 11734, LS, 60,0 11841.12, A, 28.96, 457.2, 5.49, *1~59; -5 12115.44, A, 28.96, 457.2, 1.83, *1~59; -5 12152.8, C,0, 0, 400, 0, 4.46063E-03 12328.8, A, 28.96, 457.2, 5.49, *1~59; -5 12420.24, A, 28.96, 457.2, 1.83, *1~59; -5 12664.08, A, 28.96, 457.2, 1.83, *1~59; -5 12850.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 12907.92, BLDG, 548.64, -3.05, B19 12907.92, BLDG, 548.64, 3.05, B17 270

13070.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 13243.2, A, 28.96, 457.2, 5.49, *1~59; -5 13318, LS, 40,0 13325.4, SIGN, 4, 457.2 13395.6, A, 28.96, 457.2, 1.83, *1~59; -5 13426.08, TREE, 0, 0, 0, 0, 0 13480, TREE, 200, 50, 0, 10.67, 30.48 13538, LS, 60,0 13748, A, 28.96, 457.2, 5.49, *1~59; -5 13778.48, A, 28.96, 457.2, 1.83, *1~59; -5 13869.92, A, 28.96, 457.2, 1.83, *1~59; -5 13952.8, C,0, 0, 304.8, 0, -8.46063E-03 13961.36, A, 28.96, 457.2, 1.83, *1~59; -5 14075.4, SIGN, 5, 457.2 14205.2, A, 28.96, 457.2, 5.49, *1~59; -5 14296.64, A, 28.96, 457.2, 1.83, *1~59; -5 14510, A, 28.96, 457.2, 5.49, *1~59; -5 14540.48, A, 28.96, 457.2, 5.49, *1~59; -5 14662.4, A, 28.96, 457.2, 1.83, *1~59; -5 14675.8, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 14714.8, C,0, 0, 304.8, 0, 8.46063E-03 14784.32, A, 28.96, 457.2, 5.49, *1~59; -5 14875.76, A, 28.96, 457.2, 1.83, *1~59; -5 14997.68, A, 28.96, 457.2, 5.49, *1~59; -5 15119.6, A, 28.96, 457.2, 1.83, *1~59; -5 15143, LS, 60,0 15211.04, A, 28.96, 457.2, 5.49, *1~59; -5 15272, BLDG, 640.08, -4.57, G31 15272, BLDG, 640.08, 4.57, G32 15325.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 15424.4, A, 28.96, 457.2, 1.83, *1~59; -5 15550.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 15792.4, LS, 40,0 15868, TREE, 0, 0, 0, 0, 0 15875.4, SIGN, 4, 457.2 15920, TREE, 200, 50, 0, 10.67, 30.4 16017.4, LS, 80,0 16272, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 152.4, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0 16576.8, A, 28.96, 457.2, 1.83, *1~59; -5 17025, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 17034, A, 28.96, 457.2, 5.49, *1~59; -5 17140.68, BLDG, 457.2, -3.05, B13 17140.68, BLDG, 457.2, 3.05, B14 271

17140.68, BLDG, 487.68, -3.05, B15 17140.68, BLDG, 487.68, 3.05, B8 17140.68, BLDG, 518.16, -3.05, B4 17140.68, BLDG, 518.16, 3.05, B13 17140.68, BLDG, 548.64, -6.1, B7 17140.68, BLDG, 548.64, 6.1, B5 17155.92, A, 28.96, 457.2, 5.49, *1~59; -5 17219.93, A, 28.96, 457.2, 5.49, *1~59; -5 17300.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 17460.72, A, 28.96, 457.2, 1.83, *1~59; -5 17492.2, LS, 40,0 17573, TREE, 0, 0, 0, 0, 0 17582.64, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 17702, TREE, 200, 50, 0, 10.67, 30.48 17704.56, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, C:\STISIM\Data\Textures\Grass05.Jpg, 6, 17704.56, A, 28.96, 457.2, 1.83, *1~59; -5 17710.4, SIGN, 5, 457.2 17767.2, LS, 80,0 17909.36, A, 28.96, 457.2, 5.49, *1~59; -5 18252.8, C,0, 0, 304.8, 0, 8.46063E-03 18271.49, A, 28.96, 457.2, 1.83, *1~59; -5 18410.4, SIGN, 4, 457.2 18975.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp80mph.3ds 19014.8, C,0, 0, 304.8, 0, -8.46063E-03 19442.4, LS, 80,0 19900.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp40mph.3ds 19900.64, SOBJ, 609.6, 0, 0,0,0,0, C:\STISIM\Data\Buildings\TownBlock_1.Lmm 19958, A, 28.96, 457.2, 5.49, *1~59; -5 20110.4, A, 28.96, 457.2, 1.83, *1~59; -5 20150.2, SIGN, 100, 457.2, C:\STISIM\Data\Signs\Sp60mph.3ds 20247.56, A, 28.96, 457.2, 5.49, *1~59; -5 20367.4, LS, 40,0 20410, TREE, 0, 0, 0, 0, 0 20567.6, A, 28.96, 457.2, 1.83, *1~59; -5 20575, TREE, 200, 50, 0, 10.67, 30.48 20617.4, LS, 60,0 21000, ES

272

Appendix D – Experiment 1: Cases Examples and Consequences

Turbocharged Lancer Case

Police were called to the scene of a motor vehicle accident on the M5 motorway in the early hours of a Wednesday morning. David, the driver of the vehicle, was rushed to hospital with severe head trauma, internal bleeding and multiple broken bones and fractures to his ribs and collarbone. Witnesses driving in a nearby vehicle reported that they noticed the turbocharged Mitsubishi Lancer he was driving from about 200m behind them and within a few seconds had speed past to at least 100m ahead, where they lost sight of him around a bend. Minutes later the witnesses discovered the Mitsubishi on its roof after having ricocheted off several cement barriers on both sides of the road. Crash scene investigators validated the witness‟s reports that the driver had been excessively speeding, placing his speed on impact at approximately 160 km/h.

Ramifications Legislation in NSW makes it an offence to travel at a speed greater than that specified by the speed limit. The driver of the vehicle was exceeding the specified speed limit of 100 km/h by more than 45 km/h. As a result, they received a $1,744 fine, 6 demerit points and had their licence disqualified for a minimum of 6 months. Following a court proceeding however, the driver‟s maximum penalty can be increased to $2,200, with an unlimited license disqualification (i.e. they can never drive on NSW roads).

Street Racing Case

Emergency services were called to the scene of a multi-vehicle car accident on Liverpool Road, Bankstown, late Saturday evening. Police arrived to find two cars smash into one another, the first vehicle with its engine completely collapsed into the left hand side of the second vehicle. Whilst all 3 occupants of the first vehicle and the driver of the second escaped with relatively minor cuts and wounds, a 19 year old female in the front passenger side of the second vehicle had become trapped, pinned down by her legs. Emergency service personnel were required to use the jaws-of-life to cut through the car‟s body to rescue the 19 year old. She was taken to hospital with severe wounds and multiple shattered bones from her knees through to her feet; she was not expected to walk for months. Upon questioning, it was revealed that the drivers were street racing with their parent‟s cars. Both drivers of the vehicles claimed that they were travelling at a speed of about 140 km/h, which was 60 km/h over the speed limit, although their passengers reported even higher speeds. The accident occurred when the driver of the first vehicle lost control of his vehicle after a bump on the road and veered sharply into the side of second vehicle.

Ramifications

273

Legislation in NSW makes it an offence to travel at a speed greater than that specified by the speed limit. The driver of the vehicle was exceeding the specified speed limit of 80 km/h by more than 60 km/h. As a result, they received a $1,744 fine, 6 demerit points and had their licence disqualified for a minimum of 6 months. However as a result of negligent driving where grievous bodily harm has occurred, in the absence of a specific court order, their license disqualification was increased to 3 years. If a court proceeding was to follow, the maximum disqualification can be unlimited (i.e. they can never drive on NSW roads), there can be a maximum fine of $2,200 and a maximum gaol sentence of 9 months.

Traffic Lights Case Susanna, a mother of 3, was taken to hospital to be treated for severe shock and after hitting a pedestrian with her motor vehicle at a traffic light intersection on George St, Sydney, late Thursday afternoon. The 29-year-old male pedestrian received fatal head injuries as a direct result of the accident. Upon investigation, Susanna reported to police that she was rushing home after work to prepare a birthday dinner for one of her sons. As she approached the aforementioned traffic light intersection, the light had turned from green to amber. Susanna, who had been driving at the specified 50 km/h speed limit, admitted to accelerating up to a speed of about 70 km/h even though she had plenty of space to slow down and stop at the intersection. About 10m before the intersection, she noticed the male pedestrian crossing the road. Susanna admitted that she didn‟t even have a chance to brake; it had happened too quickly for her to react.

Ramifications Legislation in NSW makes it an offence to travel at a speed greater than that specified by the speed limit. The driver of the vehicle was exceeding the specified speed limit (50 km/h) by more than 10 km/h but not more than 20 km/h. As a result, they received a $197 fine and 3 demerit points. However, as a result of negligent driving where death has occurred, in the absence of a specific court order, their license was disqualified for 3 years. If a court proceeding was to follow, the maximum disqualification can be unlimited (i.e. they can never drive on NSW roads), there can be a maximum fine of $3,300 and a maximum gaol sentence of 18 months.

274

Appendix E – Experiment 1: Participant Information Statement

Improving Drivers‟ Risk Management Behaviour

You have been invited to participate in a study titled „Improving Drivers‟ Risk Management Behaviour‟. I, Prasannah Prabhakharan, hope to investigate the effectiveness of different training methods on driving performance. You have been selected because you are a licensed student between the ages of 16-25

The experiment will be conducted over 2 one hour sessions spaced apart by one week. In the first session, you will answer a series of questionnaires, participate in a test known as the Implicit Association Test (IAT) and may also participate in a driving simulation task. In second session, you will complete another IAT and will be involved in a driving simulation task and a post experiment questionnaire.

While the likelihood is extremely low, there is the possibility that you will experience some psychological distress operating the driving simulator. It is also possible, though uncommon, that you may experience physical symptoms such as motion sickness, dizziness, fatigue, and/or nausea. In the rare event that any of these occur, please notify the researcher immediately and the experiment will be ceased without delay.

The personal benefits of participating in this research are limited. However, from the wider communities perspective, the expected benefit of the research include a more thorough understanding of young drivers‟ risk management behaviour. The natural extension of this research also will aim to implement more effective training methods in driver education programs.

Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will only be disclosed with your permission, except as required by law. If you give us your permission by signing this document, we plan to publish the results in journals and conference proceedings. Please note with all publications, information will be provided in such a way that you cannot be identified.

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]).

You will be able to obtain a summary of the research results on UNSW Aviation‟s home web page under „News‟.

Your decision as to whether or not you would like to participate will not prejudice your future relations with The University of New South Wales and/or the Department of Aviation. If you decide to participate, you are free to withdraw your consent and to discontinue participation at any time without prejudice.

If you have any questions, please feel free to ask the experimenter before signing this consent form. If you have any additional questions later, please contact either Prasannah Prabhakharan

275

([email protected]) or Dr Brett Molesworth ([email protected]). You will be given a copy of this form to keep. PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM (continued)

Improving Drivers‟ Risk Management Behaviour

You are making a decision whether or not to participate. Your signature indicates that, having read the information provided above, you have decided to participate.

…………………………………………………… .……………………………………………………. Signature of Research Participant Signature of Witness

…………………………………………………… .…………………………………………………….PRASANNAH PRABHAKHARAN (Please PRINT name) (Please PRINT name)

PRIMARY RESEARCHER …………………………………………………… .……………………………………………………. Date Nature of Witness

------

REVOCATION OF CONSENT

Improving Drivers‟ Risk Management Behaviour

I hereby wish to WITHDRAW my consent to participate in the research proposal described above and understand that such withdrawal WILL NOT jeopardise any treatment or my relationship with The University of New South Wales, and/or the Department of Aviation.

…………………………………………………… .……………………………………………………. Signature Date

…………………………………………………… Please PRINT Name

The section for Revocation of Consent should be forwarded to Dr. Brett Molesworth, Room 205c Old Main Building UNSW Sydney 2052.

276

Appendix F – Experiment 1: Participant Script

WEEK 1 BACKGROUND QUESTIONNAIRES In the first section of this session, you will complete a series of questionnaires and scales. It is critical to that you answer all the questions honestly and from your point of view.

If at any stage during the test, there are instructions that you don‟t understand, please ask the researcher to clarify.

Between each subsection, there will be a concluding screen. Take a moment, should you need, before pressing finish to continue to the next subsection.

I would like to reiterate that your answers are completely confidential and your data is deidentified so answers will remain completely anonymous.

Also could I get you switch your phone off so it will not distract you during the experiment.

INTRO TO CASES/RULE/CONTROL Now that you have completed the background questions, we can now move into the main task. Further instructions will follow.

INTRO TO 10.5 KM DRIVE The next section of this session is the primary task for this week.

This section involves a 10.5 km drive on the simulator. In the unlikely event that you feel uncomfortable or nauseous during the drive, please let the experimenter know immediately and the research will be discontinued.

For this task, you need to assume the following role. You are a new employee of News Limited, a newspaper company. Your job puts you in charge of driving a delivery vehicle to deliver newspapers to subscribed businesses and residents. Your department within News Limited has a reputation as the most reliable and efficient departments within the organisation. As a new employee you need to ensure that you do not let the department down.

There are a number of delivery points, which are along a predefined route. Any buildings along the way will signify a delivery point. You do not have to stop the vehicle to deliver the newspapers; another employee will throw them out the window for you.

Your task is to deliver all of the newspapers to each subscriber as quickly as possible whilst driving in accordance with NSW road rules. News Limited will hold you personally liable for any infringements you incur during your drive. 277

FEEDBACK During your delivery drive, you exceeded the speed limit on _____ separate occasions. At one point you exceeded the speed limit by _____km/h. For this violation, considering you are a ______license holder, you would receive _____ demerit point/s and receive a fine of $_____. (For P1, As of July 2009, this offence is an immediate disqualification of your licence.)

More importantly you jeopardised the safety of both yourself and your passengers. In addition, the lives of any bystanders you came into contact with you were also jeopardised as a result of this driving behaviour

ALL PARTICIPANTS Take a moment to reflect on speeding behaviour and the consequences associated.

DEBREIF The debrief about what has happened during this week will be discussed at the end of next week.

If I could get you not to mention any aspect of this experiment to friends, just so they don‟t know what the experiment is about beforehand.

------

WEEK2

INTRO TO SECOND WEEK IAT Similar to the first week, you will be asked to complete an Implicit Association Task. This will only take about 10 minutes.

INTRO FOR TEST DRIVE In this section of the session, you will get to drive the simulator as a test run. The main purpose of this test drive is make sure you are familiar with the controls and to make sure the simulator does not make you feel sick or nauseous. In the unlikely event that you do feel unwell, please let the researcher know immediately and we the research will be discontinued .

INTRO TO 20 KM DRIVE This section involves a 20 km drive on the simulator. In the unlikely event that you feel uncomfortable or nauseous during the drive, please let the experimenter know immediately and the research will be discontinued.

For this task, you need to assume the following role. You are a new employee of News Limited, a newspaper company. Your job puts you in charge of driving a delivery vehicle 278

to deliver newspapers to subscribed businesses and residents. Your department within News Limited has a reputation as the most reliable and efficient departments within the organisation. As a new employee you need to ensure that you do not let the department down.

Thank you for your time and participating in the research.

279

Appendix G – Experiment 1: Percentage of distance speeding (PSY output files)

Table 16

Means and Standard Deviations (40 km/h Zone Percentage of Speeding)

n Mean Standard Deviation

Group 1 (Case) 15 7.389 4.795

Group 2 (Rule) 14 10.459 4.791

Group 3 (Episodic) 15 3.421 3.272

Group 4 (Ctrl) 14 7.76 4.812

Table 17

Between contrast coefficients (40 km/h Zone Percentage of Speeding)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

280

Table 18

Analysis of Variance Summary Table (40 km/h Zone Percentage of Speeding)

Source SS df MS F

Case v Ctrl B1 0.997 1 0.997 0.050

Rule v Ctrl B2 50.988 1 50.988 2.570

Episodic v Ctrl 136.311 1 136.311 6.871

Error 1071.243 54 19.838

Table 19

Raw CIs (scaled in Dependent Variable units; 40 km/h Zone Percentage of Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.371 1.655 -4.461 3.719

Rule v Ctrl B2 2.699 1.683 -1.461 6.858

Episodic v Ctrl -4.339 1.655 -8.428 -0.249

Note: Bonferroni 95% Simultaneous Confidence Intervals

281

Table 20

Approximate Standardized CIs (scaled in Sample SD units; 40 km/h Zone Percentage of

Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.083 0.372 -1.001 0.835

Rule v Ctrl B2 0.606 0.378 -0.328 1.540

Episodic v Ctrl -0.974 0.372 -1.892 -0.056 B3

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 21

Means and Standard Deviations (60 km/h Zone Percentage of Speeding)

n Mean Standard Deviation

Group 1 (Case) 15 15.478 14.905

Group 2 (Rule) 14 19.822 12.183

Group 3 (Episodic) 15 5.553 5.998

Group 4 (Ctrl) 14 16.224 10.952

282

Table 22

Between contrast coefficients (60 km/h Zone Percentage of Speeding)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

Table 23

Analysis of Variance Summary Table (60 km/h Zone Percentage of Speeding)

Source SS df MS F

Case v Ctrl B1 4.040 1 4.040 0.031

Rule v Ctrl B2 90.583 1 90.583 0.689

Episodic v Ctrl 824.612 1 824.612 6.270

Error 7102.401 54 131.526

283

Table 24

Raw CIs (scaled in Dependent Variable units; 60 km/h Zone Percentage of Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.747 4.262 -11.277 9.783

Rule v Ctrl B2 3.597 4.335 -7.113 14.308

Episodic v Ctrl -10.671 4.262 -21.202 -0.141

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 25

Approximate Standardized CIs (scaled in Sample SD units; 60 km/h Zone Percentage of

Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.065 0.372 -0.983 0.853

Rule v Ctrl B2 0.314 0.378 -0.620 1.248

Episodic v Ctrl -0.930 0.372 -1.849 -0.012

Note: Bonferroni 95% Simultaneous Confidence Intervals

284

Table 26

Means and Standard Deviations (80 km/h Zone Percentage of Speeding)

n Mean Standard Deviation

Group 1 (Case) 15 7.573 8.92

Group 2 (Rule) 14 5.641 6.772

Group 3 (Episodic) 15 2.003 2.831

Group 4 (Ctrl) 14 6.441 5.688

Table 27

Between contrast coefficients (80 km/h Zone Percentage of Speeding)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

285

Table 28

Analysis of Variance Summary Table (80 km/h Zone Percentage of Speeding)

Source SS df MS F

Case v Ctrl B1 9.277 1 9.277 0.223

Rule v Ctrl B2 4.481 1 4.481 0.108

Episodic v Ctrl 142.644 1 142.644 3.434

Error 2242.837 54 41.534

Table 29

Raw CIs (scaled in Dependent Variable units; 80 km/h Zone Percentage of Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 1.132 2.395 -4.786 7.049

Rule v Ctrl B2 -0.800 2.436 -6.819 5.219

Episodic v Ctrl -4.438 2.395 -10.356 1.479

Note: Bonferroni 95% Simultaneous Confidence Intervals

286

Table 30

Approximate Standardized CIs (scaled in Sample SD units; 80 km/h Zone Percentage of

Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 0.176 0.372 -0.743 1.094

Rule v Ctrl B2 -0.124 0.378 -1.058 0.810

Episodic v Ctrl -0.689 0.372 -1.607 0.230

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 31

Means and Standard Deviations (Overall Percentage of Speeding)

n Mean Standard Deviation

Group 1 (Case) 15 29.416 25.525

Group 2 (Rule) 14 33.601 17.416

Group 3 (Episodic) 15 9.646 8.311

Group 4 (Ctrl) 14 30.425 20.433

287

Table 32

Between contrast coefficients (Overall Percentage of Speeding)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

Table 33

Analysis of Variance Summary Table (Overall Percentage of Speeding)

Source SS df MS F

Case v Ctrl B1 7.368 1 7.368 0.020

Rule v Ctrl B2 70.618 1 70.618 0.196

Episodic v Ctrl 3126.699 1 3126.699 8.677

Error 19459.324 54 360.358

288

Table 34

Raw CIs (scaled in Dependent Variable units; Overall Percentage of Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -1.009 7.054 -18.439 16.422

Rule v Ctrl B2 3.176 7.175 -14.552 20.904

Episodic v Ctrl -20.779 7.054 -38.210 -3.349

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 35

Approximate Standardized CIs (scaled in Sample SD units; Overall Percentage of

Speeding)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.053 0.372 -0.971 0.865

Rule v Ctrl B2 0.167 0.378 -0.767 1.101

Episodic v Ctrl -1.095 0.372 -2.013 -0.176

Note: Bonferroni 95% Simultaneous Confidence Intervals

289

Appendix H – Experiment 1: Number of zone violations (PSY output files)

Table 36

Means and Standard Deviations (40 km/h Zone Frequency of Violation)

n Mean Standard Deviation

Group 1 (Case) 15 9.6 0.828

Group 2 (Rule) 14 10 0

Group 3 (Episodic) 15 6.933 3.751

Group 4 (Ctrl) 14 9.714 0.611

Table 37

Between contrast coefficients (40 km/h Zone Frequency of Violation)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

290

Table 38

Analysis of Variance Summary Table (40 km/h Zone Frequency of Violation)

Source SS df MS F

Case v Ctrl B1 0.095 1 0.095 0.024

Rule v Ctrl B2 0.571 1 0.571 0.146

Episodic v Ctrl 56.003 1 56.003 14.306

Error 211.390 54 3.915

Table 39

Raw CIs (scaled in Dependent Variable units; 40 km/h Zone Frequency of Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.114 0.735 -1.931 1.702

Rule v Ctrl B2 0.286 0.748 -1.562 2.133

Episodic v Ctrl -2.781 0.735 -4.598 -0.964

Note: Bonferroni 95% Simultaneous Confidence Intervals

291

Table 40

Approximate Standardized CIs (scaled in Sample SD units; 40 km/h Zone Frequency of

Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.058 0.372 -0.976 0.860

Rule v Ctrl B2 0.144 0.378 -0.789 1.078

Episodic v Ctrl -1.406 0.372 -2.324 -0.487

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 41

Means and Standard Deviations (60 km/h Zone Frequency of Violation)

n Mean Standard Deviation

Group 1 (Case) 15 8.133 2.532

Group 2 (Rule) 14 8.714 1.637

Group 3 (Episodic) 15 6.267 3.674

Group 4 (Ctrl) 14 8.786 2.19

292

Table 42

Between contrast coefficients (60 km/h Zone Frequency of Violation)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

Table 43

Analysis of Variance Summary Table (60 km/h Zone Frequency of Violation)

Source SS df MS F

Case v Ctrl B1 3.082 1 3.082 0.443

Rule v Ctrl B2 0.036 1 0.036 0.005

Episodic v Ctrl 45.951 1 45.951 6.601

Error 375.881 54 6.961

293

Table 44

Raw CIs (scaled in Dependent Variable units; 60 km/h Zone Frequency of Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.652 0.980 -3.075 1.770

Rule v Ctrl B2 -0.071 0.997 -2.535 2.392

Episodic v Ctrl -2.519 0.980 -4.942 -0.097

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 45

Approximate Standardized CIs (scaled in Sample SD units; 60 km/h Zone Frequency of

Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.247 0.372 -1.165 0.671

Rule v Ctrl B2 -0.027 0.378 -0.961 0.907

Episodic v Ctrl -0.955 0.372 -1.873 -0.037

Note: Bonferroni 95% Simultaneous Confidence Intervals

294

Table 46

Means and Standard Deviations (80 km/h Zone Frequency of Violation)

n Mean Standard Deviation

Group 1 (Case) 15 4.6 2.028

Group 2 (Rule) 14 4.429 1.604

Group 3 (Episodic) 15 3.333 2.41

Group 4 (Ctrl) 14 5.214 1.718

Table 47

Between contrast coefficients (80 km/h Zone Frequency of Violation)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

295

Table 48

Analysis of Variance Summary Table (80 km/h Zone Frequency of Violation)

Source SS df MS F

Case v Ctrl B1 2.733 1 2.733 0.700

Rule v Ctrl B2 4.321 1 4.321 1.107

Episodic v Ctrl 25.620 1 25.620 6.565

Error 210.719 54 3.902

Table 49

Raw CIs (scaled in Dependent Variable units; 80 km/h Zone Frequency of Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.614 0.734 -2.428 1.200

Rule v Ctrl B2 -0.786 0.747 -2.631 1.059

Episodic v Ctrl -1.881 0.734 -3.695 -0.067

Note: Bonferroni 95% Simultaneous Confidence Intervals

296

Table 50

Approximate Standardized CIs (scaled in Sample SD units; 80 km/h Zone Frequency of

Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.311 0.372 -1.229 0.607

Rule v Ctrl B2 -0.398 0.378 -1.332 0.536

Episodic v Ctrl -0.952 0.372 -1.870 -0.034

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 51

Means and Standard Deviations (Overall Frequency of Violation)

n Mean Standard Deviation

Group 1 (Case) 15 22.333 5.038

Group 2 (Rule) 14 23.143 2.958

Group 3 (Episodic) 15 16.533 9.265

Group 4 (Ctrl) 14 23.714 3.931

297

Table 52

Between contrast coefficients (Overall Frequency of Violation)

Contrast Group 1 Group 2 Group 3 Group 4

Case v Ctrl B1 1 0 0 -1

Rule v Ctrl B2 0 1 0 -1

Episodic v Ctrl 0 0 1 -1

Table 53

Analysis of Variance Summary Table (Overall Frequency of Violation)

Source SS df MS F

Case v Ctrl B1 13.810 1 13.810 0.398

Rule v Ctrl B2 2.286 1 2.286 0.066

Episodic v Ctrl 373.410 1 373.410 10.774

Error 1871.638 54 34.660

298

Table 54

Raw CIs (scaled in Dependent Variable units; Overall Frequency of Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -1.381 2.188 -6.787 4.025

Rule v Ctrl B2 -0.571 2.225 -6.070 4.927

Episodic v Ctrl -7.181 2.188 -12.587 -1.775

Note: Bonferroni 95% Simultaneous Confidence Intervals

Table 55

Approximate Standardized CIs (scaled in Sample SD units; Overall Frequency of

Violation)

Contrast Value SE CI limits

Lower Upper

Case v Ctrl B1 -0.235 0.372 -1.153 0.684

Rule v Ctrl B2 -0.097 0.378 -1.031 0.837

Episodic v Ctrl -1.220 0.372 -2.138 -0.302

Note: Bonferroni 95% Simultaneous Confidence Intervals

299

Appendix I – Experiment 2: 10 km track (STISIM script)

0, BSAV, 0, 1, Prasannah Experiment 1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 23, 24, 26, 27, 28, 39, 42, 43, 44, 47, 50 0, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 0, TREE, 200, 50, 0, 10.67, 30.48, 0 0, SIGN, 100, 60.96, C:\STISIM\Data\Signs\Sp60mph.3ds 0, A, 16.67, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 50, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 100.96, A, 16.67, 457.2, 1.83, *1~59; -5 ;-9 ; -15; -36; -46; -49 213.36, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 396.24, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 457.2, BLDG, 563.88, -3.05, B21 457.2, BLDG, 579.12, 3.05, B20 472.44, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 550, LS, 60,0 550, LS, 80,0 573.02, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 676.66, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 728, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 883.92, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 975.36, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 975.36, TREE, 0, 0, 0, 0, 0, 0 1000.4, SIGN, 5, 500 1100.24, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 1100.24, TREE, 200, 50, 0, 10.67, 30.48, 0 1219.2, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 1220, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp60mph.3ds 1228, LS, 40,0 1341.12, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 1615.44, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 1652.8, C,0, 0, 500, 0, 5.46063E-03 1720, LS, 60,0 1828.8, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 300

1833, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 1920.24, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 2164.08, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 2333, LS, 40,0 2362, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp60mph.3ds 2407.92, BLDG, 548.64, -3.05, B26 2407.92, BLDG, 548.64, 3.05, B17 2407.92, BLDG, 598.64, -3.05, B5 2407.92, BLDG, 598.64, 3.05, B12 2743.2, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 2771, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 2825.4, SIGN, 5, 500 2862, LS, 60,0 2895.6, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 2926.08, TREE, 0, 0, 0, 0, 0, 0 3050, TREE, 200, 50, 0, 10.67, 30.48, 0 3248, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 3271, LS, 80,0 3278.48, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 3369.92, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 3452.8, C,0, 0, 300, 0, 2.46063E-03 3458, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp60mph.3ds 3461.36, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 3575.4, SIGN, 4, 500 3705.2, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 3796.64, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 3958, LS, 60,0 4010, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 4021, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 4040.48, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 4162.4, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 4214.8, C,0, 0, 400, 0, -7.46063E-03 4284.32, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 4375.76, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 4472, BLDG, 940.08, -3.05, G25 4472, BLDG, 940.08, 3.05, G29 4472, BLDG, 990, -3.05, G23 4472, BLDG, 990, 3.05, G21 4497.68, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 4521, LS, 40,0 4581, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 4619.6, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 4711.04, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 4924.4, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 301

5081, LS, 80,0 5099, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp60mph.3ds 5370, TREE, 0, 0, 0, 0, 0, 0 5375.4, SIGN, 4, 500 5500, TREE, 200, 50, 0, 10.67, 30.48, 0 5599, LS, 60,0 5676, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 5772, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 152.4, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0 5914.8, C,0, 0, 200, 0, -2.46063E-03 6076.8, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 6154, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 6176, LS, 80,0 6534, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 6629.68, BLDG, 457.2, -3.05, B3 6629.68, BLDG, 457.2, 3.05, B10 6629.68, BLDG, 487.68, -3.05, B15 6629.68, BLDG, 487.68, 3.05, B8 6629.68, BLDG, 527.68, -3.05, B7 6629.68, BLDG, 527.68, 3.05, B5 6654, LS, 40,0 6655.92, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 6665, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp60mph.3ds 6719.93, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 6960.72, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 7025.6, TREE, 0, 0, 0, 0, 0, 0 7072.64, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, , , C:\STISIM\Data\Textures\Concrete.Tga, 24, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, C:\STISIM\Data\Textures\Concrete.Tga, 12, 7165, LS, 60,0 7200, TREE, 200, 50, 0, 10.67, 30.48, 0 7204.56, ROAD, 3.66, 4, 2, 2, 0.15, 3.05, 3.05, 0.12, 0.12, 0, -1, -1, 0, 3.05, 0, 3.05, 0, 3.05, 0, 3.05, 0, C:\STISIM\Data\Textures\Road15.Jpg, 12, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, C:\STISIM\Data\Textures\Grass11.Jpg, 6, 7204.56, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 7210.4, SIGN, 5, 500 7233, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 7409.36, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 7733, LS, 80,0 7745, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 7752.8, C,0, 0, 400, 0, 7.46063E-03 7771.49, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 302

7910.4, SIGN, 4, 500 8245, LS, 40,0 8370, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp80mph.3ds 8514.8, C,0, 0, 200, 0, -5.46063E-03 8870, LS, 80,0 8913, SIGN, 100, 500, C:\STISIM\Data\Signs\Sp40mph.3ds 9413, LS, 40,0 9458, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 9600.64, BLDG, 487.2, -3.05, B15 9600.64, BLDG, 487.2, 3.05, B11 9600.64, BLDG, 517.68, -3.05, B12 9600.64, BLDG, 517.68, 3.05, B14 9610.4, A, *0~5, 457.2, 1.83, *1~59;-5 ;-9 ; -15; -36; -46; -49 9747.56, A, *0~5, 457.2, 5.49, *1~59;-5 ;-9 ; -15; -36; -46; -49 10000, ES

303

Appendix J – Experiment 2: Visual aid for Mental Arithmetic Task

Example of Number Addition Task 3 6 4 8 +1 4 7 5 9

304

Appendix K – Experiment 2: Participant Information Statement

Improving Drivers‟ Risk Management Behaviour

You have been invited to participate in a study titled „Improving Drivers‟ Risk Management Behaviour‟. I, Prasannah Prabhakharan, hope to investigate the effectiveness of different training methods on young driver performance. You have been selected because you currently hold a provisional licence (or higher) and are between the ages of 16-24.

The experiment will be conducted over 2 one hour sessions spaced apart by one week. In the first session, you will answer a series of questionnaires and scales, participate in a test known as the Implicit Association Test (IAT), a number addition task and will also participate in a driving simulation task. In second session, you will be involved in a driving simulation task and a post experiment questionnaire.

For your participation, you will receive a $40AUD (forty dollar) book voucher from the Australian Booksellers Association, which you will receive at the end of the second week. Benefits of participating from the perspective of the wider communities include a more thorough understanding of young drivers‟ risk management behaviour. The natural extension of this research also will aim to implement more effective training methods in driver education programs to reduce young driver accidents and fatalities on our roads.

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (p: 9385 4234, f: 9385 6648, e: [email protected]).

You will be able to obtain a summary of the research results on UNSW Aviation‟s home web page under „News‟.

305

If you have any questions, please feel free to ask the experimenter before signing this consent form. If you have any additional questions later, please contact either Prasannah Prabhakharan ([email protected]) or Dr Brett Molesworth ([email protected]). You will be given a copy of this form to keep.

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM (continued)

Improving Drivers‟ Risk Management Behaviour

You are making a decision whether or not to participate. Your signature indicates that, having read the information provided above, you have decided to participate.

…………………………………………………… .……………………………………………………. Signature of Research Participant Signature of Witness

…………………………………………………… .…………………………………………………….PRASANNAH PRABHAKHARAN (Please PRINT name) (Please PRINT name)

PRIMARY RESEARCHER …………………………………………………… .……………………………………………………. Date Nature of Witness

------

REVOCATION OF CONSENT

Improving Drivers‟ Risk Management Behaviour

…………………………………………………… .……………………………………………………. Signature Date

…………………………………………………… Please PRINT Name 306

The section for Revocation of Consent should be forwarded to Dr. Brett Molesworth, Room 205c Old Main Building UNSW Sydney 2052.

Appendix L – Experiment 2: Participant Script

If at any stage during the test, there are instructions that you don‟t understand, please ask the researcher to clarify.

Between each subsection, there will be a concluding screen. Take a moment, should you need, before pressing finish to continue to the next subsection.

I would like to reiterate that your answers are completely confidential and your data is deidentified so answers will remain completely anonymous.

Also could I get you switch your phone off so it will not distract you during the experiment.

INTRO TO 10KM DRIVE The next section of this session is the primary task for this week.

This section involves a 10km drive on the simulator. In the unlikely event that you feel uncomfortable or nauseous during the drive, please let the experimenter know immediately and the research will be discontinued.

SECOND WEEK DRIVE

307

In addition to this driving task, you will be asked to complete a basic number addition task. The researcher will read out a 4 digit number periodically throughout the drive. Your task is add ONE to each 4 digits the researcher reads out and answer with the new 4 digit number.

FEEDBACK During your delivery drive, you exceeded the speed limit on _____ separate occasions. At one point you exceeded the speed limit by _____km/h. For this violation, considering your are a ______license holder, you would receive _____ demerit point/s and receive a fine of $_____. (For P1, As of July 2009, this offence is an immediate disqualification of your licence.)

ALL PARTICIPANTS Take a moment to reflect on speeding behaviour and the consequences associated.

DEBREIF The debrief about what has happened during this week will be discussed at the end of next week.

If I could get you not to mention any aspect of this experiment to friends, just so they don‟t know what the experiment is about beforehand.

WEEK2 INTRO FOR TEST DRIVE In this section of the session, you will get to drive the simulator as a test run. The main purpose of this test drive is make sure you are familiar with the controls and to make sure the simulator does not make you feel sick or nauseous. In the unlikely event that you do feel unwell, please let the researcher know immediately and we the research will be discontinued .

INTRO TO 20KM DRIVE This section involves a 20km drive on the simulator. In the unlikely event that you feel uncomfortable or nauseous during the drive, please let the experimenter know immediately and the research will be discontinued.

308

Thank you for your time and participating in the research.

309

Appendix M – Experiment 3: Participant Information Statement

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM

Resource Allocation Training to Improve Cognitive Performance.

You have been invited to participate in a study titled „Resource Allocation Training to Improve Cognitive Performance‟. I, Prasannah Prabhakharan, hope to investigate the effectiveness of a training method to teach cognitive skills. You have been selected as a possible participant because you are a between the ages of 16-25 and hold a current driver‟s licence.

If you decide to participate, the experiment will be conducted over 2 sessions spaced apart by one week. In the first session, complete a demographic questionnaire as a well as a computer task. The second session will be almost identical to the first. Both sessions are expected to take no longer than 1 hour.

While the likelihood is extremely low, there is the possibility that you will experience some psychological distress during these tasks. Due to the nature of the tasks, it is possible, you may experience symptoms such as cognitive fatigue (that is, you feel „mentally drained‟). It is also possible, though uncommon, that you may experience physical symptoms such as motion sickness, dizziness, fatigue, and/or nausea. In the rare event that any of these occur and you feel you cannot continue, please notify the researcher immediately and the experiment will be ceased without delay.

For your participation, you will received a $40AUD (forty dollar) book voucher from the Australian Booksellers Association, which you will receive at the end of the second session. Benefits for participating in this research from the perspective of the wider community include helping us gain a more thorough understanding of how the brain processes information and the malleability of these functions. The extension of this research also will aim to incorporate the results into training procedures in many industries.

Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission, except as required by law. If you give us your permission by signing this document, we plan to publish the results in a journal and present the findings as a conference proceeding. In any publication, information will be provided in such a way that you cannot be identified.

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be investigated promptly and you will be informed of the outcome.

You will be able to obtain a summary of the research results on UNSW Aviation‟s home web page under „News‟.

Your decision as to whether or not you would like to participate will not prejudice your future relations with The University of New South Wales and/or the School of Aviation. If you decide to

310

participate, you are free to withdraw your consent and to discontinue participation at any time without prejudice.

If you have any questions, please feel free to ask the researcher before signing this consent form. If you have any additional questions later, please contact either Prasannah Prabhakharan ([email protected]) or Dr Brett Molesworth ([email protected]).

You will be given a copy of this form to keep.

311

THE UNIVERSITY OF NEW SOUTH WALES

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM (continued)

Resource Allocation Training to Improve Cognitive Performance.

You are making a decision whether or not to participate. Your signature indicates that, having read the information provided above, you have decided to participate.

…………………………………………………… .……………………………………………………. Signature of Research Participant Signature of Witness

…………………………………………………… Prasannah Prabhakharan (Please PRINT name) (Please PRINT name)

…………………………………………………… Primary Researcher Date Nature of Witness

………………………………………………………………………………………………………………………………………………

REVOCATION OF CONSENT

Resource Allocation Training to Improve Cognitive Performance.

…………………………………………………… .……………………………………………………. Signature Date

……………………………………………………

The section for Revocation of Consent should be forwarded to Dr. Brett Molesworth, Room 205c Old Main Building UNSW Sydney 2052.

312

Appendix N – Experiment 3: Visual aid for the dual-task

X X X X

X V X X

X X X X

Press “A” in this situation

LEFT EAR RIGHT EAR

“ 9 ” “ 5 ” Press “L” in this situation because both are ODD

“ 4 ” “ 2 ” Press “L” in this situation because both are EVEN

“ 8 ” “ 3 ” DO NOT Press “L” in this situation because one is EVEN, the other is ODD 313

Appendix O – Experiment 3: Participant Script

Week 1 Introduction

In the first section of this session, you will complete a series of questionnaires and scales. It is critical to that you answer all the questions honestly and from your point of view.

If at any stage during the test, there are instructions that you don‟t understand, please ask the researcher to clarify.

I would like to reiterate that your answers are completely confidential and your data is de-identified so answers will remain completely anonymous.

Also could I get you switch your phone off so it will not distract you during the experiment.

Dual Task

The session is primarily focused on a computer task. Listen carefully to these instructions to completely understand what is required.

You will part take in a number of 1-minute blocks. In each block, you will receive a series of visual and auditory stimuli simultaneously. Your task will be to respond correctly to both of the stimuli.

The visual task is a visual search task. Here, you will see a matrix of letters. In some cases, all the letter will be the same. In some cases, you may notice that a few are different. Your task is simply press “A” if there is one or more letters that are different to the rest of the matrix.

If the matrix has all the same letters, you don‟t need to respond.

SHOW EXAMPLE

The auditory task is a number discrimination task and will be played to you through the headphones. For this task, you will hear 2 numbers between 1 and 9, one in each ear. Your task is to press “L” if both numbers are ODD OR if both numbers are EVEN.

If the one number is odd and other is one even, you don‟t need to respond.

SHOW EXAMPLE

You will have 1.5 seconds to come up with both answers before the next trial commences.

These blocks can get quiet repetitive but the poorer you perform, the longer the task takes. It is in your best interest to try and stay as motivated as possible in order to complete the first session as quickly as possible.

314

Once the block is finished you will be asked to turn off the monitor and pass the keyboard to the researcher.

FOR IMPLICIT GROUP ONLY

In each block there are a number of targets to respond to. If you feel you are not responding enough to one or both task/s, chances are you are missing the targets.

FOR EXPLICIT GROUP ONLY

For these tasks, we‟re after a specific level of performance. Following each trial, you will be given feedback about your performance in order to amend it correctly. Please take this feedback into account after each block and aim to perform accordingly.

FEEDBACK

If correct

In the block you just completed, the results indicate that you performed at the desired level. Please perform at a comparable level in the next block as well.

If close to correct level of performance

In the block you just completed, the results indicate that you performed slightly/significantly below the desired level on the visual task (and you slightly/significantly below the desired level on the auditory task). It is important that you focus on improving your performance on this/these task to achieve the desired level of performance.

Week 2

Welcome back to this week‟s session. This week will be identical to last week‟s session, where you will receive the dual visual and auditory task.

Remember, you press „A‟ if one or more letter in the matrix of letters is different to the rest and you press „L‟ if the two the numbers you hear in both ears are either both ODD or both EVEN.

SHOW EXAMPLE

The session will take approximately the same amount of time as last week.

315