Elsevier

Accident Analysis & Prevention

Volume 50, January 2013, Pages 1193-1206
Accident Analysis & Prevention

The influence of differences in the functioning of the neurocognitive attentional networks on drivers’ performance

https://doi.org/10.1016/j.aap.2012.09.032Get rights and content

Abstract

Considerable research efforts are currently being devoted to analysing the role that the attentional system plays in determining driving behaviour, with the ultimate objective of reducing the number of attention-related accidents. The present study aims to assess the influence of differences in the functioning of the three attentional networks (executive control, attentional orienting and alerting) when drivers have to deal with some common hazardous situations, for example, when an oncoming car or a pedestrian unexpectedly crosses their trajectory. Multiple measures of participants’ attentional functioning were obtained from a computer-based neurocognitive test: the Attention Networks Test for Interactions and Vigilance (ANTI-V). These measures were compared to performance in a driving simulator where different types of hazardous situations were presented. Correlation and linear regression analyses revealed significant associations between individual attentional measures and driving performance in specific traffic situations. In particular, a higher attentional orienting score on the ANTI-V was associated with safer driving in situations where a single precursor anticipated the hazard source, whereas in complex situations with multiple potential hazard precursors, higher attentional orienting scores were associated with delayed braking. Additionally, partial evidence of a relationship between crash occurrence and the functioning of the executive control and the alerting networks was found. Overall, the current research would be helpful to better understand the role that each attentional network (executive control, attentional orienting and alerting) play in safe driving, and thus to develop efficient countermeasures to reduce attention-related crashes.

Highlights

► The association of attentional neurocognitive networks and driver performance is analysed. ► Drivers’ attentional functioning is measured using a neurocognitive test, the ANTI-V. ► Driving performance is measured using a simulator presenting hazardous situations. ► Attentional orienting is helpful in situations where a precursor anticipated a hazard. ► Attentional orienting can be negative in situations with multiple potential hazards.

Introduction

Driving a vehicle is a complex multi-tasking activity, in which all cognitive resources should be applied in a coordinated way to safely complete a journey. Of the different cognitive resources, research efforts have increasingly been devoted to analysing the role played by the attentional system in driving behaviour, with the aim of reducing the number of road traffic accidents (e.g., Regan et al., 2011). In fact, driver distraction and inattention are considered among the major contributing causes of road traffic casualties and their negative impact on road safety is expected to further increase in the immediate future, mainly due to the proliferation of some potentially distracting in-vehicle technologies (e.g. Kircher, 2007, Klauer et al., 2006, Ranney, 2008, Regan et al., 2011, Stutts et al., 2001).

The present study will provide additional evidence to explain the influence of different attentional functions (such as executive control, attentional orienting, phasic alertness and vigilance) when drivers have to deal with common hazardous situations, such as when an oncoming car or a pedestrian unexpectedly crosses their trajectory. To achieve this objective, multiple measures of attentional function were obtained individually from participants using a single computer-based neurocognitive test termed the Attention Networks Test for Interactions and Vigilance (ANTI-V; Roca et al., 2011), and these measures were compared with performance in a driving simulator where a number of hazardous situations had to be safely negotiated. Although some previous attempts have been made with other neurocognitive tests to link attentional network functions to driving behaviour (e.g., Weaver et al., 2009), the relationship between the attentional components assessed by the ANTI-V and performance data from a driving simulator presenting hazardous situations is still unclear.

As the result of a decade of neurocognitive research on human attention, a quick and easy computer-based task has been designed with the aim of measuring participants’ performance in some basic components of attention. The original task is known as the Attention Networks Test or ANT (Fan et al., 2002) and is a combination of the cued reaction time (Posner, 1980) and the flanker task (Eriksen & Eriksen, 1974). Participants are required to determine as fast as possible the direction of a central arrow (left or right), and the efficiency of the three attentional networks is assessed by measuring the influence on performance of alerting signals, spatial cues and distracting flankers.

The ANT is based on a widely accepted neurocognitive model of human attention, i.e. the three attentional networks model (Posner, 1994, Posner, 2008, Posner and Petersen, 1990). According to this model, three relatively independent neural networks are responsible for controlling the different attentional functions: the executive control, attentional orienting and alerting networks. First, the executive control network involves mechanisms for ignoring distracters and resolving cognitive conflict, and is usually assessed by using Stroop, Simon or flanker tasks (e.g., Callejas et al., 2004, Fan et al., 2002). While driving, executive control could be a crucial factor in complex traffic situations, where multiple potentially hazardous elements should be concurrently monitored and drivers have to focus their attention on the most relevant ones whereas ignoring the others (such as a busy crossroad junction). Second, the attentional orienting network is aimed at selecting information from the sensory input by allocating the attentional focus to a potentially relevant area or object in the visual field, and is usually assessed by presenting valid, invalid and neutral spatial cues in reaction time tasks (e.g., Callejas et al., 2004, Fan et al., 2002). Regarding the driving task, for example, it has been suggested (Crundall et al., 2012, Garay-Vega et al., 2007) that the existence in traffic situations of elements foreshadowing a hazard may act as a cue helping drivers drawing their attention to its potential location (for instance, a bus stopped may indicate the potential location of pedestrians about to cross the road). Finally, the alerting network is necessary to achieve and maintain a state of high sensitivity to incoming stimuli (Posner, 2008). This network is related to performance in tasks that involve both phasic alertness (i.e., the increased readiness to respond after a warning signal) and tonic alertness or vigilance (i.e. the ability to maintain attention over a prolonged period of time) (see, for example, Posner, 2008, Sturm and Willmes, 2001). Both components of the alerting network are considered to play a relevant role while driving. For example, phasic alertness is involved in situations where a warning signal is available to the driver to avoid a hazard (as provided, for instance, by a lane departure or a collision avoidance warning system; e.g., May & Baldwin, 2009). In addition, low vigilance is considered to be one of the main causes of road accidents, especially after sleep loss or during prolonged driving (Åkerstedt et al., 2011, Campagne et al., 2004, Lal and Craig, 2001, Larue et al., 2011).

The validity of the ANT measures is solidly supported by evidence from different disciplines, such as neuroscience, neuropsychology and experimental psychology (e.g., Fan et al., 2002, Fan et al., 2005, Ishigami and Klein, 2010, Posner, 2008). As a consequence, this attentional test and its variations are currently being used in a wide range of basic and applied studies, and also in the driver behaviour and road traffic safety areas (see for example, López-Ramón et al., 2011, Roca et al., in press, Weaver et al., 2009). Some alternative versions of the test are currently available, such as the ANTI-V, which includes an extra measure of vigilance in addition to the executive control, attentional orienting and phasic alertness indices (Roca et al., 2011, Roca et al., 2012). This test may be especially useful in driver behaviour studies, since low vigilance is considered a major cause of road traffic accidents (Åkerstedt et al., 2011, Campagne et al., 2004, Lal and Craig, 2001, Larue et al., 2011). In addition, by employing this test, the influence of each specific attentional component (executive control, attentional orienting, phasic alertness and vigilance) on driving behaviour can be analysed separately.

Previous research has tried to predict driving performance by using the attentional scores provided by the Attentional Networks Test. Weaver et al. (2009) used the original version of the ANT (Fan et al., 2002) and the Manitoba Road Test (Weaver et al., 2009) in both a simulated driving evaluation and an on-road test. The Manitoba Road Test is a demerit-based scoring system, aimed at assessing driving performance. Demerit points are given for the commission of certain infractions, such as speed, turning or signal violations. In this study, two overall performance measures from the ANT (global reaction time and global accuracy) were good predictors of overall performance in the driving simulator. However, no association was found between the three separate functions of attention (executive control, attentional orienting and phasic alerting) and driving performance in the simulator or the on-road test.

Weaver et al. (2009) found these results surprising, since the attentional functions are considered to play an important role while driving. Thus, looking for potential associations in different driving situations and using alternative driving performance measures was recommended, since the Manitoba Road Test may not have been appropriate to tap all the different aspects of attention. The lack of significant results may have occurred for a number of reasons, such as the gap between the driving task and the attentional measures considered. The driving task that participants were asked to complete in Weaver et al. (2009)’s study (i.e. a hazard-free route where the commission of certain infractions was assessed) may not have represented those situations where the correct functioning of the attentional networks is most in demand (whereas, for example, hazardous situations are likely to place the greatest demand on attentional functions). Unfortunately, pragmatic and ethical issues prevent the study of truly hazardous real-world situations (unless one can undertake a large scale naturalistic study; see Klauer et al., 2006). Still, simulated driving environments provide a compromise between the realism of the task and the ability to place participants in controlled yet hazardous situations.

As a consequence, the main objective of the current study is to analyse specific relationships between individual differences in the functioning of the three attentional networks (executive control, attentional orienting and alerting) and the participants’ performance in a driving simulator presenting particular hazardous situations in which demands are made on attention to avoid crashing. These driving performance measures should be considered more appropriate for evaluating the role played by the attentional system than previous studies using, for example, indices based on traffic infractions while engaged in everyday driving (e.g., Weaver et al., 2009).

General driving performance measures will be obtained in this study, together with separate measures in different types of traffic situations, such as the three categories of hazard proposed by Crundall et al., 2010, Crundall et al., 2012: Behavioural Prediction, Environmental Prediction, and Dividing and Focusing Attention situations (see description below). It can be suggested that the relationship of each attentional network to driving performance might not be unidirectional in any given traffic situation. On the contrary, a specific attentional function could be helpful in some specific situations, whereas in some other cases it might be associated with a worse performance. For example, focusing our attention on a preceding vehicle (and thus partially ignoring the traffic environment) will be helpful when this vehicle unexpectedly brakes, but this attentional behaviour can be dangerous when a pedestrian suddenly steps out in front of one's car. As a consequence, some qualitative differences might be found in the relationship of a specific attentional network score with performance in different traffic situations.

Crundall et al., 2010, Crundall et al., 2012 proposed that three categories of driving hazards could be differentiated according to the relationship between potential precursors and the actual hazard source: Behavioural Prediction, Environmental Prediction, and Dividing and Focusing Attention hazardous situations (see Table 1). First, Behaviour Prediction hazards can be avoided if the drivers anticipate the behaviour of a visible precursor (i.e. a pedestrian or another vehicle) before it becomes a hazard source (for example, an oncoming motorcycle that suddenly invades the participant's trajectory). Second, in Environmental Prediction situations the hazard source is not visible before the hazard is triggered (for example, a child stepping out from behind an ice-cream van). It should be noted that the precursors to Environmental Prediction hazards are part of the environment and conceal the hazard source (such as the ice-cream van hiding the child), while the precursors to Behavioural Prediction hazards are the same stimuli as the hazard sources (for example, the motorcycle is both a precursor and a hazard source). Third, Dividing and Focusing Attention situations require the drivers to monitor multiple sources of potential risk before selecting one as the actual hazard. This is a more complex category containing potential hazards from both Behavioural Prediction and Environmental Prediction categories, but specifically in this category, more than one hazard is visible at the point at which the hazard triggered (for example, when driving over a crossroads junction, there is traffic from the right that fails to give way, while a hazard from the left was equally plausible).

By using this taxonomy of hazards, different results may be hypothesized depending on the attentional function considered and the characteristics of the simulated driving situation.

First, previous research (Crundall et al., 2012, Garay-Vega et al., 2007) suggested that the existence in traffic situations of elements foreshadowing a hazard may act as a cue helping drivers drawing their attention to its potential location. As a consequence, an association between a higher attentional orienting score in the ANTI-V (which measures the influence of peripheral spatial cues on participants’ performance) and better resolution of Behavioural Prediction and Environmental Prediction hazards (in which a precursor could be used to foreshadow the hazard source) might be expected.

Second, Dividing and Focusing Attention hazards involve complex situations where multiple potential hazards are concurrently present, and thus executive control might be necessary to ignore non-hazardous traffic elements. According to Crundall et al. (2012), dividing attention across the multiple potential hazards in Dividing and Focusing Attention situations must give way to focusing attention upon the actual hazard when it occurs. Therefore, we may propose that a reduced executive control score in the ANTI-V (which reflects a participant's higher ability to ignore distracters) might be specifically associated to better performance in Dividing and Focusing Attention situations.

Third, regarding the alerting network in the ANTI-V, none of the driving situations include the presentation of warning signals and thus no particular association is expected with the phasic alertness score. Similarly, the current study was not specifically interested in vigilance during driving and, therefore, did not contain any task demands to manipulate vigilance (e.g., using a prolonged task or sleep deprivation). Yet, it is possible to find significant results with the phasic alertness or the vigilance measures, since both alerting network components are considered to play a relevant role in safely driving, as previously claimed.

A good understanding of how the attentional networks influence common driving situations could be useful in various ways. For example, as raised by Weaver et al. (2009), each neural network is mainly influenced by a specific neurotransmitter (executive control by dopamine, attentional orienting by acetylcholine, and alerting by noradrenaline; Posner, 2008). Therefore, it is possible to anticipate the possible impact of some medications or some clinical conditions on drivers’ performance in particular situations. For example, drugs or clinical conditions modulating the cholinergic system would have an impact on driving situations associated with the attentional orienting network, whereas performance in driving situations associated with other attentional networks might be preserved. Also, it could be possible to use the ANTI-V to obtain individual patterns of attentional dysfunction (e.g., from clinical populations or from subclinical elderly drivers), and then create driving rehabilitation programmes customised to their needs.

In addition, data obtained in the current study might provide further evidence of the validity of hazard detection-based driver simulators, such as the FAROS GB3 used in this study. If relevant associations between the attentional indices from the ANTI-V and performance in the driving simulator are observed, this would suggest that simulator is able to tackle important cognitive components of real driving (i.e. the attentional functions of executive control, attentional orienting and alertness).

Section snippets

Participants

A sample of 42 students from the University of Nottingham volunteered for this study. Twenty were females (48%) and their mean age was 22 (St. Dev. = 4). Each of them had a valid UK driving licence and a minimum experience of 12 months since passing the driving test (mean = 4.5 years; St. Dev. = 3.7). Also, normal or corrected-to-normal vision was required. None of them had previous experience with the driver simulator.

The attentional test (ANTI-V)

The Attention Networks Test for Interactions and Vigilance (ANTI-V) was used to

The attentional test (ANTI-V)

The usual main effects and interactions expected with the ANTI-V task were obtained in the 2 (Congruency) × 3 (Validity) × 2 (Warning) repeated-measures ANOVA (see Fig. 3). RT results revealed that: (a) participants were faster when the central and the flanker cars were congruent as compared to when they were incongruent (618 ms and 691 ms, respectively; F(1,38) = 122.89; p < .001; eta2 = .76); (b) participants were faster when a valid cue was presented in comparison to the no cue and invalid conditions

Discussion

The main objective of the present study was to analyse the relationship between differences in the functioning of the three attentional networks (executive control, attentional orienting and alerting) and drivers’ behaviour. With this aim, individual measures of attentional functioning were obtained by using a single computer-based neurocognitive test (i.e. the Attention Networks Test for Interactions and Vigilance) and these were then compared with participants’ performance in a driving

Conclusions and limitations of the study

In conclusion, the current study has been successful in obtaining evidence to show that individual measures in attentional functioning (executive control, attentional orienting, phasic alertness and vigilance) could be differently associated with simulated driving performance in specific hazardous situations. In particular, a higher score in attentional orienting has been associated with a safer driving performance in situations where there is a single precursor anticipating the hazard source

Acknowledgments

Financial support was provided by the Ministerio de Ciencia e Innovación (PSI2010-15883, PSI2011-22416 and PSI2011-29504), the Junta de Andalucía (PO7-SEJ-02613) and the Engineering and Physical Sciences Research Council (EP/D035740). None of the funding sources had a direct involvement in the study design, in data collection, analysis or interpretation, in the writing of the report or in the decision to submit the paper for publication. Also, we would like to thank the Dirección General de

References (40)

  • G.S. Larue et al.

    Driving performance impairments due to hypovigilance on monotonous roads

    Accident Analysis and Prevention

    (2011)
  • J.F. May et al.

    Driver fatigue: The importance of identifying causal factors of fatigue when considering detection and countermeasure technologies

    Transportation Research Part F

    (2009)
  • A. Miyake et al.

    The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis

    Cognitive Psychology

    (2000)
  • M.A. Regan et al.

    Driver distraction and driver inattention: definition, relationship and taxonomy

    Accident Analysis and Prevention

    (2011)
  • J. Roca et al.

    Measuring vigilance while assessing the functioning of the three attentional networks: The ANTI-Vigilance task

    Journal of Neuroscience Methods

    (2011)
  • J. Roca et al.

    The effects of sleep deprivation on the attentional functions and vigilance

    Acta Psychologica

    (2012)
  • W. Sturm et al.

    On the functional neuroanatomy of intrinsic and phasic alertness

    Neuroimage

    (2001)
  • G. Underwood et al.

    Driving simulator validation with hazard perception

    Transportation Research Part F

    (2011)
  • B. Weaver et al.

    Using the Attention Network Test to predict driving test scores

    Accident Analysis and Prevention

    (2009)
  • K. Ball et al.

    Visual attention problems as a predictor of vehicle crashes in older drivers

    Investigative Ophthalmology & Visual Science

    (1993)
  • Cited by (23)

    • Dual-task interference in simulated car driving: The psychological refractory period effect when not only the second, but also the first task is ecologically relevant

      2022, Applied Ergonomics
      Citation Excerpt :

      A further limitation of our study is that we did not formally assess our participants’ cognitive functioning. Driving skills are known to be related to cognitive abilities such as attention (Marciano and Yeshurun, 2015; Roca et al., 2013; Wang et al., 2015), working memory capacity (Guo et al., 2020; Louie and Mouloua, 2019; Wood et al., 2016) and executive functions (León-Domínguez et al., 2017; Mäntylä et al., 2009). A comprehensive battery of cognitive tests therefore might have reduced unobserved heterogeneity and revealed associations between cognitive abilities and performance on our driving task.

    • Validation of the attention-related driving errors scale in novice adolescent drivers

      2021, Accident Analysis and Prevention
      Citation Excerpt :

      The two subscales of the DBQ (errors and violations) tested specific relationships between types of aberrant driving. The performance-based measures included in the present study were chosen in consideration of previous research evaluating relationships between the ARDES and performance-based attention (López-Ramón et al., 2011; Montes et al., 2016; Roca et al., 2013a). A computerized CPT task containing two subtests was administered: Phase “X” (sustained attention) and Phase “AX” (selective attention; Rosvold et al., 1956).

    • On the origins and evolution of the Attention Network Tests

      2021, Neuroscience and Biobehavioral Reviews
    • Predicting driver reaction time and deceleration: Comparison of perception-reaction thresholds and evidence accumulation framework

      2021, Accident Analysis and Prevention
      Citation Excerpt :

      Based on these results, it appears that drivers determine deceleration rate in braking maneuvers based on the strength of evidence (i.e., the value of tau-inverse) at the start of braking. This finding can also be related to the association of attentional orienting with driving performance (Posner, 1980; Roca et al., 2013). Similar to the finding in Roca et al. (2013), when the lead vehicle brake lights act as a single potential hazard precursor as in this study, the drivers with higher attentional orienting will show safer driving behavior (e.g., earlier braking).

    View all citing articles on Scopus
    View full text