The relationship between cognitive function and hazard perception in younger drivers

Highlights

• Examines cognitive correlates of different aspects of hazard perception ability for younger drivers.

• Visuo-spatial skills, executive function and global cognitive function predicts hazard perception accuracy.

• Inhibitory control predicts anticipation of hazards.

• No cognitive skills predicted action responses times to hazards.

• Cognitive measures could enhance graduated driver licensing systems.

Abstract

Driving is a complex task; with research suggesting cognitive function plays a significant role in driver behaviour. Recent studies have investigated the role of cognitive function in younger drivers who are experiencing brain maturation and are over-represented in crash statistics. Emerging evidence suggests poor cognitive functioning is one explanation for this high crash risk. For younger drivers, the relationship between cognitive function and driving ability has been consistently shown for speeding and lane deviations. However, the driving skill most consistently linked to crash involvement is hazard perception, which is the ability to anticipate and respond to potentially dangerous traffic situations. The aim of this study was to investigate the cognitive correlates of hazard perception in younger drivers. Seventy-nine undergraduate students completed a hazard perception test and a battery of cognitive tests that have previously demonstrated a relationship with safe driving. The newly created hazard perception test measured accuracy of hazard identification, response times when anticipating the hazard, and response times when taking action to avoid the hazard. Hazard perception accuracy was significantly related to visuo-spatial skills, executive functioning and global cognitive status. Anticipation response times were significantly related to inhibitory control, with no significant relationship found between cognitive function and action response times. These findings are discussed in line with limitations in the study. Future research into the role of specific cognitive domains could lead to the enhancement of hazard perception testing for licensing with cognitive training and assessment, with the potential to reduce the crash risk of vulnerable younger drivers.

Introduction

Driving is a complex and cognitively demanding task that is widely considered to be one of the most dangerous activities an individual can engage in. Mistakes or errors can lead to crashes; and this is reflected in global road trauma statistics where worldwide 1.35 million people are killed annually and is the leading cause of death for those aged under 29 (World Health Organisation, 2018). Numerous explicit skills have been defined in an effort to understand how better and safer driving can be facilited. One of the driving skills consistently linked with crash risk is hazard perception, which is the ability to anticipate potentially dangerous road and traffic situations (Horswill & McKenna, 2004). The importance of hazard perception skills to younger drivers has been recognised by governments such that assessment of hazard perception skills has become part of Graduated Driver Licensing systems in countries such as Australia, the United Kingdom, and the Netherlands (Australian Government, 2018, Bates et al., 2014, Drive Right Netherlands, 2019).

While it is a well-established phenomenon that younger drivers (<25 years) have a significantly higher crash risk compared to older, more experienced drivers, their crash risk is similar to that of drivers over 65 years of age (Bureau of Infrastructure, Transport and Regional Economics, 2013). This has prompted a significant body of research aimed at investigating factors that contribute to younger and older driver crash risk. Given younger drivers are experiencing physical and cognitive brain maturation and older drivers are experiencing normal age-related cognitive decline, recent studies have suggested that cognitive function may play a role in driver errors (Foy et al., 2016, Mathias and Lucas, 2009). The majority of studies have focused on the role of cognitive function across a range of driving behaviours such as speeding and lane deviations. However, there has been limited research into the cognitive factors associated with the driving skill of hazard perception.

The role of cognitive function in safe driver behaviour has largely been studied in older drivers, with a focus on assessing fitness to drive (Aksan, Anderson, Dawson, Uc, & Rizzo, 2015). Research has shown that the cognitive domains of executive function, attention, visuo-spatial skills, memory, and overall mental status are associated with the crash risk of older drivers who are experiencing age-related cognitive changes (Anstey, Wood, Lord, & Walker, 2005). Given younger drivers also have a heightened crash risk, a recent study by Ledger, Bennett, Chekaluk, & Batchelor (2019a) found that cognitive function was related to speeding and lane deviations in a driving simulator for both older and younger drivers. There was no difference in the strength of this relationship between the two age groups. The cognitive domains that were shown to be associated with driving performance were executive function, attention, visuo-spatial skills, memory, inhibition, psychomotor skills, and global cognitive status (Bennett et al., 2016, Ledger et al., 2019a).

Executive function is the ability to direct thoughts and actions towards goal-directed behaviour (Miyake et al., 2000). In older drivers, poor executive function was related to increased crash rates (Daigneault, Joly, & Frigon, 2002). Likewise in a simulator study examining executive function of younger drivers, results suggest the cognitive processes that were important for safe lane changing included planning, memory, decision-making, and mental flexibility (Mäntylä, Karlsson, & Marklund, 2009). Similarly, younger drivers with poor executive functioning were more likely to engage in dangerous driving behaviours such as speeding (Hayashi, Foreman, Friedel, & Wirth, 2018).

Attention is broadly defined as the capacity of an individual to focus on relevant information pertaining to the task being performed while ignoring irrelevant information (Luna, Garver, Urban, Lazar, & Sweeney, 2004). A study of older drivers suggested failures in selected, divided or sustained attention were related to on-road driving errors including speeding, lane deviations and signalling errors (McKnight & McKnight, 1999). Likewise, attentional capacity of younger drivers has also been found to mediate younger driver crash risk (Regan, Lee, & Victor, 2013).

Safe driving also relies on visuo-spatial skills, or the ability to accurately perceive and process rapidly changing visual stimuli in ones surrounding environment (Anstey et al., 2005). Studies in both older and younger drivers found poor visuo-spatial skills significantly predicted on-road driver errors and discriminated between safe and unsafe drivers (Aksan et al., 2015, Lincoln et al., 2006, McKnight and McKnight, 2003).

Appropriate use of memory systems to encode, store and retrieve information is another critical aspect of cognition that is vital for driving tasks such as navigation, recall of plans and performing learned ‘automatic’ skills such as changing gears (Anstey et al., 2005). A study in older drivers found those with poorer memory had poorer performance on a range of driving tasks (Hu, Trumble, Foley, Eberhard, & Wallace, 1998). Likewise, a study in younger drivers found those with higher working memory capacity performed better on a simulated lane change task (Ross et al., 2014).

Response inhibition, a component of executive function, is the ability to withhold a dominant or automatic response (Miyake et al., 2000). A study of drivers across the lifespan found poor inhibitory control was related to driver errors including speeding, unsafe following distances and lane deviations (Adrian, Moessinger, Charles, & Postal, 2019). Findings suggest the ability to suppress an initiated response is particularly important when sudden changes in the road environment require an alternative action, such as choosing to wait at an intersection after identifying the presence of a previously unseen oncoming car. Response inhibition has been widely studied in younger drivers either as an isolated domain or as a component of executive function (Walshe, McIntosh, Romer, & Winston, 2017). Studies of younger drivers have found that errors on an inhibition tasks (e.g. Stop-signal task, Stroop colour-naming task) were positively associated with lane deviations (Jongen, Brijs, Komlos, Brijs, & Wets, 2011); collisions (Ross et al., 2015) and speeding in a driving simulator (Hatfield, Williamson, Kehoe, & Prabhakharan, 2017).

Psychomotor skills are another important cognitive domain, with safe driving requiring visuo-motor control of the vehicle and manual dexterity to coordinate a rapid response to changing traffic conditions (Emerson et al., 2012, Shanmugaratnam et al., 2010). Poor psychomotor skills were related to increased collision rates and time spent speeding in a simulator study of older drivers, suggesting psychomotor skills are important in braking and accelerator control (Alosco, Spitznagel, Cleveland, & Gunstad, 2013). Little research has been done into the relationship between psychomotor performance and driving for young drivers. A simulator study by Zicat, Bennett, Chekaluk, and Batchelor (2018) finding no relationship between psychomotor performance and speeding or lane deviation, however this has not been examined over a range of driving behaviours in this cohort.

Finally, mental status or global cognitive status is frequently assessed in older drivers as part of a fitness to drive assessment (Reger et al., 2004). A commonly used test is the Mini-Mental State Exam (MMSE), which measures a range of cognitive domains, including orientation, language, concentration, working memory, recall, and visuo-spatial skills (Folstein, Folstein, & McHugh, 1975). A driving simulator study of older drivers found the MMSE was highly predictive of driver safety (Mathias & Lucas, 2009).Emerging research suggests that the MMSE might also be a useful predictor of driving performance in young drivers with a simulator study by Ledger, Bennett, Chekaluk, Batchelor, & Di Meco (2019b) predicted overall driving performance over and above age. Further research on the relationship with cognitive status and driving in younger drivers is needed.

The role of individual cognitive domains in driving behaviour is well supported in the literature and underpins brief screening instruments used to assess fitness to drive (Anstey, Horswill, Wood, & Hatherly, 2012). However, a composite battery of tests assessing multiple cognitive domains was found to be a better predictor of driving performance (Bennett et al., 2016). While the majority of studies have focused on driver behaviours such as speeding and lane deviations, there is a growing body of research into the role of cognitive function in hazard perception.

Hazard perception involves a number of cognitive processes in the detection, interpretation, and reaction to potential traffic conflicts that appear in a complex environment (Crundall, 2016, Horswill et al., 2009, Wetton et al., 2011). Hazard perception skills require a proficient coordination of multiple cognitive systems - efficient visual search strategies (Mackenzie & Harris, 2015); focusing and prioritising attention and memory of past situations (Wood, Hatley, Furley, & Wilson, 2016); mental flexibility to predict the appearance of potential hazards, and planning and executing a course of action (McInerney & Suhr, 2015). Despite the cognitive complexity of hazard perception, the development of tests that can assess hazard perception skills has largely been without a theoretical framework (Moran, Bennett, & Prabhakharan, 2019).

Hazard perception skills are typically measured by watching a series of short video clips of potential traffic conflicts on a computer (McKenna & Crick, 1994). Response times (temporal response) and accuracy (spatial response) of hazard perception is captured through a range of methods including button press, touch screen, eye tracking and questionnaire (Moran et al., 2019). Temporal responses are typically based on a priori defined hazard windows for each scenario. Response windows are the time from when the hazard can be first anticipated or appears until it is no longer deemed a hazard, or action has to be taken to avoid a crash such as braking or swerving (Grayson & Sexton, 2002). The use of response windows aids in linking the temporal response to the appearance of the hazard, rather than random or inappropriate responding. Likewise, spatial responses are linked to the a priori defined hazard location using methods such as touch screen or questionnaires (e.g. ‘What was the hazard you responded to?’)

Recent developments in hazard perception testing have used a theoretical framework based on Endsley (1995) situation awareness (SA; Moran et al., 2019). SA involves broad cognitive processes in the perception of stimuli in a dynamic environment, comprehension of what has been perceived and projection of future status (Endsley, 1995). SA has been measured using the Situation Global Awareness Technique (SAGAT) in which clips of hazardous scenarios are occluded prior to the hazard developing (Endsley et al., 1998, Crundall, 2016). This is followed by questions, ‘What was the hazard? Where was the hazard? What happens next? (WWW; Jackson et al., 2009, Castro et al., 2016, Ventsislavova et al., 2016). The final question ‘What happens next?’ has formed the basis for developing hazard prediction tests (rather than perception tests) that assess decision-making following identification of the hazard (Gugliotta et al., 2017, Ventsislavova and Crundall, 2018).

The validity of hazard perception tests is based on a well-established paradigm that older, more experienced drivers will have quicker response times, more efficient and broader visual search patterns, and more accurate hazard perception and prediction, than younger, novice drivers (Crundall, 2016, Horswill et al., 2009, McKenna and Crick, 1994, Scialfa et al., 2011, Wetton et al., 2011). Hazard perception ability increases over time and with experience on the roads; but begins to decline in older drivers from 65 years of age at a time when normal age-related cognitive decline occurs (Horswill et al., 2009). In younger drivers, hazard perception ability has been shown to improve over time until a few years following licensure (Borowsky, Shinar, & Oron-Gilad, 2010). This coincides with normal brain maturation, which is occurring at the same time as younger drivers are gaining experience on the roads. It can be difficult to disentangle the role of cognitive changes and driving experience on hazard perception ability in younger drivers due to their confounding effect. However for older drivers, they cannot become less experienced and so the majority of research into cognitive function and hazard perception has been in older driver populations.

A study by McInerney and Suhr (2015) used a battery of cognitive tests and found executive function, attention, visuo-spatial skills, memory, inhibition and overall cognitive function contributed individually to the variance in hazard perception errors (defined as a failure to identify the hazard). Collectively, these cognitive domains accounted for 36.7% of the variance. Horswill et al. (2008) found that selected and divided attention, as well as processing speed, were related to hazard response times in older drivers. Additional cognitive variables such as working memory and psychomotor skills have also been shown to be important for hazard perception in older drivers, with the effects of cognitive function independent of age (Anstey et al., 2012).

There have been limited studies investigating the relationship between specific cognitive domains and hazard perception in younger drivers. A study by Ross et al. (2015) found poorer inhibitory control, as denoted by longer reaction times on a stop-signal task, was positively related to hazard perception response times in a driving simulator study of younger drivers (Verbruggen & Logan, 2008). Also in a younger driver sample, Wood et al. (2016) found poor working memory was associated with longer hazard response times when cognitive load was increased by asking participants to complete a computer-based hazard perception test and a secondary auditory tone task simultaneously. There was no association found between working memory and hazard perception response times when cognitive demands were lower in the single task condition. By also analysing eye gaze behaviour such as duration of fixation on a hazard as a measure of attention, results suggest the ability to inhibit distractions, shift between tasks, update working memory and focus attention, were found to be important in hazard perception (Wood et al., 2016).

The limited body of research into hazard perception in younger drivers has focused on specific cognitive domains. There has been no research that replicates the use of a battery of cognitive tests, which have been shown to be more predictive of driver safety in older populations than assessing individual domains (Bennett et al., 2016). Furthermore, hazard perception research and test development has been largely without a theoretical framework (Moran et al., 2019). There has been a narrow application of frameworks such as Endsley (1995) situation awareness, to explain the cognitive processes involved in hazard perception. Given the complex interplay between multiple cognitive domains and safe driving in older populations, there is an opportunity to apply a broader theoretical framework to understanding cognitive function and hazard perception in younger drivers, such as Uc and Rizzo (2008) information processing model for driver error.

Uc and Rizzo (2008) proposed the information processing model following studies investigating the crash risk of individuals with neurodegenerative disorders such as Alzheimer’s disease. Their model, which is consistent with the well-established human information processing model (Wickens, 1992), proposes that from the introduction of a stimulus, such as a car turning across the driver’s path, there are four stages at which cognitive impairment can cause errors that lead to unsafe driving behaviours such as failing to brake.

The first stage involves perception and attention to stimuli through sensory input and interpretation of the road and traffic environment; followed by the second stage in which memories guide the planning of a response. The third stage involves executing the response (e.g. braking), and feedback occurs in the fourth stage so that experiences can guide subsequent actions. These four stages work holistically in guiding safe driving behaviour. While the Uc and Rizzo (2008) information processing model has largely been studied in older drivers, Ledger et al. (2019a) and Zicat et al. (2018) found it equally applicable to younger drivers. Based on Uc and Rizzo (2008) model, cognitive errors in perceiving and interpreting the complex and rapidly changing roadway; failure of attention and short term memory to assimilate information; and correct planning and executing appropriate acceleration, braking and steering responses can lead to driver errors. While the information processing model has been the framework used to assess cognitive function in a range of driving behaviours, there is limited research that applies this model specifically to hazard perception.

While studies suggest that individual cognitive domains are related to hazard perception, to the authors’ knowledge there have been no studies in younger drivers that replicate those in older driver populations, in which a battery of cognitive tests are used to assess the key domains associated with hazard perception ability. It seems prudent to gain a more holistic understanding of the role of cognitive function in younger drivers given driver licensing occurs during a phase of brain maturation and with limited on-road driving experience. Therefore, the aim of this study is to investigate the individual and overall cognitive correlates of hazard perception in younger drivers. Specifically, it is hypothesised that poorer performances on a battery of cognitive function tests that assesses a range of cognitive domains will be positively related to lower accuracy scores and longer response times on a hazard perception test.