The Scottish Government

« Previous | Contents | Next »

Listen

EXTERNAL-TO-VEHICLE DRIVER DISTRACTION

CHAPTER THREE THEORY

3.1 Psychological theory relating to driver distraction is derived from psychological theories of attention.

AROUSAL THEORY

3.2 The first major theory of attention in modern psychology was stated by Donald Hebb in 1955. Hebb stated that attention was a function of arousal. 'Arousal' has a technical definition in psychology. However, it means roughly the same as the words 'excited' or 'interested' in ordinary language. It is a physiological state, which can be measured quantitatively (via EEG measurements, moisture levels on skin and other methods). Hebb's theory was an adaption of the Yerkes-Dodson (Yerkes and Dodson, 1908) law, and stated that human beings seek out an optimal level of arousal. Too low a level of arousal ('anxiety') would lead to a situation of boredom. Too high a level of arousal ('anxiety') would lead to a state of stress. This can be shown in graphic form in figure 3.1, below.

Figure 3.1: The 'Yerkes-Dodson' Law

3.3 Other psychologists have demonstrated that the situation is rather more complex than this basic model would allow. For example, it has become clear that the level of arousal we seek varies over time. In the morning, when we wake up, we only want low levels of arousal (quiet music, a coffee). In the evening we will tend to want more arousal (a walk, perhaps a movie, or a restaurant). When going to sleep at night, we will wish for lowered arousal again. Moreover there are probably individual differences in the amount of arousal we seek as well. Some people (termed 'sensation seekers') seem to seek high levels of arousal and will tend to make lifestyle choices accordingly (for example, in becoming professional gamblers, or pursuing 'extreme sports') (Zuckerman, 1979).

3.4 Earlier in the twentieth century the Russian psychologist Pavlov had noted what he termed the Orientation Reaction (OR) (Pavlov, 1927). This is the reaction of an animal to unanticipated stimulus. The Canadian psychologist D.E. Berlyne describes a few of the features of this reaction as follows: 'The pupil of the eye dilates….the eyes open wide and turn toward a source of visual stimulation…the head turns towards a source of sound' (Berlyne, 1960: 81). It was Berlyne who first linked the OR to arousal theory on the one hand and information theory on the other. Information theory (as defined by Shannon, 1948), stated that information could be quantified (in terms of how many 'bits') it contained. Berlyne's innovation was to see that information could modify arousal. So for example, if we are underaroused, we can seek information (read a newspaper, pick up a book, watch television) to raise our arousal levels. On the other hand, if there is 'too much information' (for example, if we are asked to perform a task for which we are inadequately trained) we will attempt to lower our levels of arousal (Shinar et al. 1978: 18. cf. also Matthews et al, 1996).

3.5 Berlyne's work was mainly concerned with discovering what kind of information had what kind of effect on arousal. Here he made a number of discoveries, three of which are particularly relevant to the present study.

Novelty

3.6 In experiments, Berlyne discovered that information that was 'novel' (in a way that could be defined in terms of Shannon's theory) was both more likely to raise arousal and more likely to provoke the Orientation Reaction. For example, in one experiment, pictures of animals were projected on a screen for ten seconds. One animal slide was kept the same for the duration for the experiment, whereas the other was constantly altering and showing new animals. It was found (by tracking eye movements) that the 'novel' animal was looked at for more and more time (relative to the 'non-novel' animal) as the experiment progressed (Berlyne, 1960). See also Friedman (1979).

Surprisingness (or incongruity)

3.7 This refers to an informational element in relationship to other informational elements. As patterns occur, we begin to construct hypotheses as to 'what will happen next', and if these expectations are violated, we are 'surprised'. Thus in an experiment showing geometrical displays of 'dots' on a screen, the dots with a fundamentally different pattern from the 'norm' were looked at for longer, and provoked more 'arousal' than the 'normal' patterns (Berlyne, 1957). This was backed up by an experiment by Denny on rats, in which hungry rats were given the 'choice' of two mazes to seek food. One was the 'normal' maze they usually found food in, and one was a new maze, with a different order. The rats overwhelmingly chose the 'new' maze, indicating they were 'bored' with the old one, and sought new stimuli to raise their arousal (Denny, 1957).

Complexity

3.8 The information content of shapes and figures (for example) can be quantified. So for example, one can say a line is made up of one 'bit' of information, two lines are made up of two 'bits' of information, an equilateral triangle, three bits, and so on. Again, it was discovered that the more complex the shapes the more they were looked at in an experimental situation.

3.9 Therefore, novel, surprising and complex figures and sounds will be more arousing than boring repetitive and simple shapes and sounds. It should be noted that in terms of what subjects liked, shapes that were over complex, over-'novel' and over surprising were disliked as much as shapes that were boring and simple. People preferred an intermediate point: not too arousing, not too boring.

3.10 It is clear how this relates to a theory of attention. Attention is posited in this theory as a means to an end: a means of modulating arousal via the pursuit of information. Information will be sought out if the subject is too bored, and avoided if the subject is too stressed. In the 'underarousal' state the subject will be more likely to produce the 'Orientation Reaction' to complex, incongruous and novel phenomena. This OR will be what we term 'distraction'. It should be noted that these findings have been broadly confirmed by studies of drivers and signs. Drivers were more likely to notice signs/billboards that were bright, contained colour contrasts (black and white, for example), and were large: in other words, that contained arousing, stimulating information. It seems highly likely that 'neon' 'flashing' signs would be even more 'arousing' in this sense, and, hence, distracting (Forbes et al, 1968).

3.11 Now there is an obvious question here that will be dealt with later. If the subject is performing a task which s/he finds boring, and information presents itself, is the Orientation Reaction conscious or unconscious? That is, can the subject choose to 'override' this reaction, or, physiologically, is the impulse towards arousal too strong? If the latter were the case, this would obviously have strong implications for driver distraction.

BROADBENT'S 'FILTER' THEORY

3.12 The other main 'stream' of attention theory derives from Broadbent. It should be noted that Broadbent saw his theory as an adjunct to Hebb's theory. He never denied the important role that arousal played in attention processes, nor that the subject was actively seeking information (as opposed to passively receiving it) (see, for example, Broadbent 1958: 126-127). Nevertheless the emphasis of his theory was somewhat different.

3.13 Broadbent also used information theory. However, he concentrated on the fact that a given communication medium (for example a 'phone line) has a limited capacity: only a certain amount of information can pass through it. He then went on to hypothesise that human information processing had similar properties. Therefore, the 'medium' of information transmission into the brain (that is, the ears or eyes) would have a limited capacity to transfer information to the brain. Broadbent's theory is, therefore, a 'bottleneck' theory. He hypothesised that because of this limitation, 'filters' have evolved to limit the amount of information that can be processed.

3.14 To test his hypothesis, Broadbent carried out a number of experiments, mainly concerned with attending to two information sources at the one time. For example, in one experiment subjects had to attend to speech while (occasionally) a buzzer would sound. Broadbent found that task performance was reduced, and concluded that the information transmission aspect of the senses had been 'overloaded', and that information transmission was therefore a 'single channel': more than one information source would lead to overload.

3.15 It should be noted that the essence of Broadbent's theory, that the senses of human beings are single channel information transducers prone to 'bottlenecks', is false. This was demonstrated soon afterwards by Treisman, who demonstrated that there had to be more than one information-transmission channel. For example, subjects were asked to carry out a listening task played into one ear whilst 'distracting' information was played in the other: there was no performance decrement (although there was when more than one other information source was provided) (Treisman, 1964b).

3.16 More importantly, Moray (1959), showed that if the 'second' information source contained information that was personally important to the subject, then the subject would attend to it: in other words, it was not filtered out completely.

3.17 Treisman used this finding to create a new theory, that instead of being wholly filtered out, extraneous information was attenuated. That is, instead of being 'filtered out' (and, therefore, not processed), it would be accorded less importance, and processed, as it were, in a preliminary fashion. If its information features were 'recognised' then it would be further processed and, if necessary, brought to consciousness (Treisman, 1964b). In other words, the main features of Broadbent's theory (that information-channels have a limited capacity) seem to be true, but his specific predictions were false. Information is not filtered out, and there is not just one information channel.

The Theories of Neisser

3.18 At this point the findings of Neisser (1976) should be discussed. In the 'seventies, Neisser and others carried out various versions of the 'dual task' experiments above, and bore out their main findings. However, he then carried on to investigate what effect practice had on these activities. For example Hirst et al (1980) discovered that with practice, subjects could learn to take dictation and read a story at the same time. This suggested that, whereas at any given time the attentional ability of a subject was limited this limit could vary over time, and could in fact be increased with practice. Therefore, even though in absolute terms the capacity of the information transmission channel was fixed, the subject's ability to process the information provided could be increased (by analogy this is the same as the way data is transmitted on the Internet: when files are compressed more information can be transmitted. Nothing has changed in the hardware, but with the addition of particular software, the computer 'learns' to accept and process more information).

3.19 This fits in well with Hebb's original theories as proposed above, given that the 'arousal point' varies over time. At some times the same information will lead to overarousal (stress), whilst at others it will be 'just right'. To use the example of time of day, the amount of information available at a night-club or crowded bar will be acceptable, whereas it would lead to overload at seven in the morning. In the current context, it might be predicted that whereas to begin with, new drivers would have difficulty in talking to a passenger and driving at the same time, with practice this would become easier. This was demonstrated experimentally by Brown and Poulton (1961).

3.20 Neisser also pointed out that whereas the limited-capacity channel argument was undoubtedly correct in an absolute sense, it could be argued that it was a slightly strange way of conceptualising the situation. A television can only show one channel at a time, but this is not normally regarded as a problem. We 'channel flick' until we find the channel we wish to watch, and then we watch it. In a sense we 'filter out' extraneous information, but usually because we are not interested in it: not because we run the risk of being 'overwhelmed' or 'overloaded'. This argument becomes stronger when we remember that (unlike a television) human beings can indeed process information from more than one channel at the same time (with practice).

3.21 Finally Allport (1993) points out that current view of brain physiology identifies many attentional 'sub-systems' in the brain (cf. also Castiello, 1997), and to state that there is one 'attention' system is a gross oversimplification. In other words, attention is an intrinsically multi-channel phenomenon. (Note: physiologically, arousal also consists of many subsystems, but positing one arousal system functions adequately as a theoretical construct (Green, 1987)).

3.23 The questions that will be asked in the next section are:

• Is the Orientation Reaction (OR) automatic?
• What attracts attention? (i.e. raises arousal?)
• What happens to attention when another task is being performed?

PERCEPTION AND PSYCHOPHYSICS

3.24 Berlyne's theories predicted that arousing information would be of a specific kind. He demonstrated that bright, stimulating, (etc.) lights triggered the Orientating Reaction (OR), an automatic response strategy that Berlyne saw as being part of the constant search for arousal. This view is supported by current research into the psychological study of search patterns and visual stimulus.

3.25 Such experiments tend to be laboratory experiments in which subjects are seated in front of a computer and are asked to perform tasks. Occasionally points of light ('singletons') which are not connected with the task are shown. Eye movements can be tracked, to see whether they are registered are not.

3.26 Here some jargon must be explained before the relevance of these experiments to driver distraction can be shown. It has been shown in (in Yantis and Johnston, 1990) that there are two main modes of visual search. Visual mode A is the 'default state'. It is used when the subject is not engaged in a task and is not searching for something. This visual search is 'broad'. That is, the subject 'takes in the whole picture' and pays attention to peripheral lights and actions.

3.27 In 'search' or 'attentional' mode (visual mode B) on the other hand, visual search is narrow. The subject concentrates on the immediate visual field. S/he is concentrating, and pays less attention to objects at the periphery of the visual field (for a specific example taken from the driving experience, see Crundall et al 1999, and Mourant and Rockwell, 1974).

3.28 The strength of this effect should not be underestimated. For example, in one experiment (Simons and Chabris, 1999) subjects were asked to view a basketball-like game and follow the number of times one of the team possessed the ball. Meanwhile, a woman in a gorilla suit walked briefly into the game. When asked afterwards, subjects did not recall seeing anything unusual. Because it was not task relevant, it was not processed. This has also been shown in more specific visual experiments in which observers were not distracted by an 'abrupt onset' (i.e. a singleton suddenly appearing on the computer screen) when performing a task which required attention (for example, Yantis and Johnston, 1990, and Yantis and Jonides, 1990). However, this finding has many exceptions, as will become obvious. For example, Theeuwes also discovered that when subjects were searching for an unknown shape they could easily become distracted by something with an 'arousing' (bright) colour even though they were not looking for something of that colour (Theeuwes, 1991). However Pashler discovered that when looking for a known object (that is, they were told to look for a square or triangle), it was far harder to slow down their search rate by distraction (Pashler 1988).

3.29 There are further debates in this field (specifically concerning whether there is more than one kind of 'search mode'), but the key findings are obvious. That is, in 'passive' or 'normal' visual search then subjects are open to what might be termed 'normal' distraction. That is, some bright light or colourful shape, from the corner of vision, might grab their attention. However in 'search mode' this is much less likely. But here there is another twist. If, in search mode, the subject is searching for something specific (i.e. a 'stop' sign s/he knows to be there), s/he is unlikely to be distracted. But if they are searching for something and they do not know if it is there or not (for example, the approach to a junction where the subject does not know whether there is a stop sign or not), then s/he is liable to have the search rate slowed by extraneous distracters, as s/he has to search through the whole visual field (for a demonstration of this in a driving context cf Brown and Cole, 1969). Bahcall and Kowler (1997) found that ability to identify characters was worsened if the characters were closer together. This again demonstrates that visual 'clutter' slows response times and has a negative impact on performance. This was demonstrated in a driving context by Hughes and Cole (1984), who discovered that many safety signs were not being noticed by drivers in situations of high visual 'clutter'. This was corroborated by Brown, and Monk (1975). Elements which seem to add to confusion are variability in the size of background elements and mean luminescence (Cole and Jenkins, 1984: Jenkins and Cole, 1982) (Note: it is also possible that because 'cluttered' scenes have less 'structure' they might also confuse search strategies, which seem to improve when the target is located in a clear, coherent, visual scene (Biederman, 1972)).

3.30 So there is a difference here between distraction per se (that is, having one's attention 'distracted' from the specific task), and complexity of visual field. The 'complexity of visual field' issue is specifically that of time. The driver is not 'distracted' in the same sense that a driver driving on a motorway may have his/her attention distracted by a bright flashing billboard. However, in terms of slowing up the search process (considering the vehicle will probably be moving, perhaps towards an interchange or junction), this might still have an impact on safety, because the driver will simply run out of time to search the visual field, and will have to make a quick, ad hoc judgement as to whether to proceed or not (in other words, to decide whether the sign is simply not there, or that not enough time was available to look for it). This decision, of course, could be the wrong one.

3.32 This still leaves the issue of the 'first' (low arousal) kind of distraction described above. This would be a situation when the driver was distracted from his/her task: that is, it would be 'bottom up', (or exogenous (Theeuwes, 1991)) or being distracted by task irrelevant data. We have seen that the driver (conceptually) can be distracted when in 'broad' 'non-search' mode. But is distraction possible in search mode as well or is all vision 'controlled' or 'top down' (endogenous)? For example, if one was driving and concentrating intently on the road (perhaps driving in such a manner because it was foggy, or there was snow on the ground) and a flashing light occured in the extreme right of the visual field. Would this be processed? If it were, this would be 'bottom up': unconsciously having our gaze drawn to an object when we don't 'want it' to be.

3.33 Here the debate has raged, but some consensus is now emerging. Mack and Rock (1998) decided that such 'bottom up processing' in search mode was impossible, after an ingenious series of experiments. For example subjects were asked to observe a target variable, whilst 'distracters' were shown at various distances from the target. The interesting point is that subjects did not even recall having seen the distracter. However, Theeuwes argues for the opposite conclusion (Theeuwes and Godjin, in press). Instead of eliciting a verbal response (i.e. asking subjects whether they 'remembered' the distracter), Theeuwes measured the length of time it took to perform the task. He discovered that the task took longer when there were distracters, regardless of whether subjects 'remembered' seeing the distracter at all. The problem is how to reconcile the views of Mack/Rock on the one hand, and Theeuwes on the other.

3.34 The solution, it seems, is to make a distinction between conscious and unconscious eye movements. Theeuwes showed that when presented with a visual distracter in 'attentional (or search) mode', when the distracter is not the object being searched for, the eye is automatically drawn to it, but unconsciously and for a short period of time (a few milliseconds). The brain does not register this, and so the subject has no memory of having seen the distracter. In non-attentional mode, on the other hand, the Orientation Reaction (OR) is directed to the distracter, but it is also consciously processed. Harbluck and Noy (2002) demonstrated this in a driving/experimental context. Using a sophisticated eye tracking device, they found that as arousal/information/cognitive demands increased, eye movements tended to concentrate more and more on the centre of the visual field and less and less on peripheries (although it should be noted that this is just a tendency: the eye usually tends to concentrate on activity on the central of the visual field, or, in the case of drivers, in the direction the car is moving (Cole and Hughes, 1990: Wolfe, O'Neill, and Bennett, 1998. For more details of eye movements while driving, see Land, 2001: Land and Horwood, 1995: and Land and Lee, 1994)). Moreover, saccadic eye movements (the eye 'darting around') decreased, with the gaze becoming more fixed (note: 'Saccadic' eye movements are not to be compared negatively to some other kind of eye movement. The normal behaviour of the eye is saccadic 'sampling' of the environment. The eye rarely rests on any one object in the visual scene for longer than a third of a second (Potter, 1975)). This kind of visual concentration would mitigate against visual distraction from the road signs/other vehicles etc. on the far left or far right of the visual field. (Note: when asked in the Harbluck and Noy experiment, subjects noted that 'distraction' increased as task difficulty increased, but they clearly meant distraction by the task, not external-to-vehicle visual distraction (Harbluck and Noy, 2002)). So therefore, distraction is less likely in these situations: but not impossible. As Wolfe writes: 'It seems fair to say that there are paradigms where singletons must capture attention even if the subject does not want this to occur' (Wolfe, 1998: 19) (italics added). It seems, therefore, that 'bottom up' (involuntary) distraction (albeit 'unconscious') is possible even in 'search mode' (although the kind of object likely to distract may still be affected by what is being searched for (Casimir et al, 2002)).

3.35 What, therefore distracts attention? As one might expect from Hebb's arousal theory, we would expect high information, novel and surprising events to attract attention, and this is in fact what we find. Irwin et al, (2000) demonstrated that 'abrupt onsets' (i.e. objects that appear 'instantaneously') and luminescence attract attention involuntarily. This bears out Berlyne's point about surprisingness and novelty, and again confirms the arousal (OR) hypothesis. It also bears out the finding that the OR is, in some circumstances, involuntary. In terms of billboards and signs, Coetze (2003) stresses that high information signs (quantified in terms of bits) will require more time to be processed, and, of course, Berlyne stressed that high information ('interesting') signs are more likely to be looked at, and for longer (Friedman, 1979), than 'boring' signs.

3.36 The relevance of these experiments for driver distraction, and their explanation in terms of arousal theory are as follows. As the pre-eminent visual theorist of our time, J.J. Gibson (1979) argued, we are active, information seeking organisms, existing in a certain environment, with which we are 'structurally coupled' (Varela, 1991). We seek information to regulate our arousal levels (Wallace et al, 2002). When we are aroused by a task (that is because it is stressful or at least interesting) we concentrate on it: and our field of attention narrows. The more aroused we are, therefore, the less likely we are to be distracted consciously. However we are 'distracted' automatically and unconsciously even in this situation, by novel, 'surprising' stimuli, but the visual system quickly decides what is and is not relevant. If it is not seen as being relevant then we don't consciously 'perceive' it (Ruz and Lupiáñez 2002). Despite this, even this automatic distraction may slow down response times.

AROUSAL AND UNDERAROUSAL

3.37 The key difference between the Broadbent and the Hebb view of perception is that Broadbent's theory only explains cognitive (or channel) overload. The Hebb theory does this too, but it also predicts difficulties due to cognitive underload.

3.38 Therefore, this theory predicts not one but two problems with driver attention: overarousal and underarousal (Young and Stanton, 2002). In arousal (search mode) situations, the driver is less likely to be distracted, although, as we have seen it is still possible. However, s/he may not yet have learned to cope with the amount of information that needs to be processed. In low arousal (broad 'scanning' mode) the driver is likely to be distracted because s/he is bored, and needs stimulation in the form of information.

3.39 The Catch 22, as regards underarousal, is, however, that drivers need information in order to keep themselves interested in what they are doing. So for example, de Fockert, Rees, Frith and Lavie write: 'Several studies have shown that distracters that could not be ignored in situations of low perceptual load (….) were successfully ignored in situations of high perceptual load' (de Fockert, 2001). However the information that is needed to raise arousal/attention (perceptual load) might in itself function as a distracter.

3.40 As a general rule, therefore, drivers 'self-regulate' their attention adequately. In low arousal conditions (boredom), s/he daydreams, or looks at the external environment. In high arousal conditions (risky situations) s/he concentrates on the road, and the task. Problems arise when risk has, as it were, objectively risen, but the driver does not realise it has, so arousal stays low. Holohan, sums up this idea best when he writes: 'The most dangerous traffic situation is that in which the driver is unable to perceive the danger. If the danger is not perceived, the reticular system is not brought into play. Consequently the driver's level of arousal is low, and his/her reactions to external stimuli will not be as rapid as they would have been had the arousal level been higher' (Holohan, 1979, 36) (see also Haga, 1984).

SPECIFIC VEHICLE AND VISION BASED EXPERIMENTS

3.42 As well as 'pure' psychophysics perception experiments, there have also been experiments that deal specifically with information processing effects in vehicles.

Arousal

3.43 Boadle (1976) showed that at night (i.e. in low information/arousal situations) vigilance and task performance decrease quickly. In low arousal situations subjects were more likely to respond, but less likely to respond accurately, to task demands. Giambra (1995) showed that Task Unrelated Imagery and Thought (TUIT) or 'daydreaming' was more associated with the performance of uninteresting or uninvolving tasks. This suggests that in low arousal situations (as drivers become bored), they are more likely to become distracted (i.e. to seek non-driving related information), and less likely to make accurate decisions as a result of the information gained.

3.44 It is possible that this state is related to the phenomenon known as 'Highway Hypnosis' (Brown, 1991). This is a state in which drivers in a state of low arousal (usually at night, in low traffic situations), start to 'dream' or 'sleep while awake'. It is often associated with the phenomenon of 'land train' accidents in Australia. Land trains are extremely long trucks which travel the Australian outback. Accidents involving these trucks are not associated with city driving, as one might expect, but with the long, featureless roads in the outback, with little or no surrounding traffic. The implication is that the lorry driver went into a state of 'highway hypnosis' and simply left the road. It has been suggested that the introduction of 'visual or audio stimulation' would be an effective remedy to this problem. In other words, 'highway hypnosis' seems to be associated with low arousal, and the solution is to raise arousal (Moses, 1995: 62).

3.45 An important corollary to this, and important in the context of distraction, is 'phototaxis', sometimes called the 'moth effect' or 'the fascination phenomenon' (Charles et al 1990). This seems to be associated with 'highway hypnosis', except that in this case, the driver 'fixates' on a light (either by the side of the road, or the headlights of the car in front). Sometimes the driver simply fixates on a car parked on the side of the road, and drives into it (frequently a police car with its lights on), or else fails to brake adequately when the car in front brakes. Again, it is associated with long featureless roads and low arousal. It should be noted that the moth effect (unlike what is generally understood as 'external' distraction) can occur in the centre or the peripheries of the visual field (for example, a driver may fixate (or become absorbed (Tellegen, A., & Atkinson, G. (1974)) on the lights of the car in front of him/her, or on a police car or road sign on the side of the road).

3.46 Chapman discovered (Chapman and Underwood, 1998) that in high danger situations (i.e. high arousal situations) drivers narrowed their visual search, and thus 'focused attention'. More specifically, novice drivers had longer fixation times, suggesting that they had less experience in 'managing' the high arousal of dangerous driving. Interestingly (and counter-intuitively) information-poor rural roads provoked longer visual fixation times (and therefore, presumable, higher arousal levels) than information-rich urban streets. This suggests that perhaps drivers sometimes compensate in 'low arousal situations' (in the 'information-poor' areas) by raising their arousal levels to stay alert, or perhaps it simply means drivers are more used to driving in urban areas and find rural areas more 'interesting'. In any case, it suggests that the high information load of urban driving can sometimes be counteracted by the driver becoming 'blasé' and reverting to 'broad attention mode': this leading to a greater risk of distraction.

Individual Differences and Impact on Attention

3.47 These experiments are not directly relevant to driver distraction per se, but detail general driving characteristics related to attention. Avolio, Kroeck and Panek (1985) carried out 'dual task' experiments similar to those above. Subjects were asked to fill in 'Perceptual Style' questionnaires (which measure ability to 'separate' objects from their background) and tests of selective attention. Poor performance on these tests was correlated with poor task performance. This suggests that drivers who have not yet learned to adequately manage attention are more likely to have accidents (for support of this hypothesis, see Gopher, 1982 and Hakinnen, 1979).

3.48 In an experimental study, Rosenbloom and Wolf (2002) assessed the relation between risky driving behaviour and high scores on Zuckerman's Sensation Seeking Scale. As predicted, 'sensation seekers' were more likely to engage in a number of 'risky' behaviours. Unfortunately 'distractibility' was not one of the features tested for. Nevertheless, this indicates an area for future research.

3.49 Gopher (1982) carried out a number of tests on pilots concerning the ability to 'divide attention'. He discovered that the ability to divide attention was correlated with ability to fly safely measured by a number of parameters. Hakinnen et al (1979) discovered similar results. One of the parameters tested was the ability to maintain attention. Again this suggests a link between ability to sustain attention and safe driving.

Alcohol

3.50 In a Spanish study (Rossello et al, 1999), it was discovered that the consumption of alcohol, even below the (Spanish) legal limit, reduced the ability to divide attention. This would certainly indicate that (for example) ability to use a mobile phone and drive would be lessened by the consumption of alcohol. Implications for external-to-vehicle distraction are less clear. For example in another study, Marinkovic et al (2001) discovered that the consumption of moderate amounts of alcohol decreased the function of a neurophysical response 'evoked by … novel and rarely occurring stimuli, regardless of whether they are task relevant' (Marinkovic, op cit.: 536). This suggests of course that drivers under the influence of alcohol would be less likely to be distracted by task irrelevant external phenomena (such as billboards) although, of course, also less likely to notice task relevant stimuli (such as oncoming traffic).

Factors Affecting Distraction

3.51 There follows a discussion of whether there are significant individual differences in propensity towards distraction, either biological or caused by specific behaviours.

Age

3.52 Given that, as stated, external-to-vehicle distraction tends to be visual, the effect of age on driver's visual acuity is particularly important. It has been generally noted in a number of studies that visual acuity tends to decrease with age (Staplin et al, 1997). This process begins at about 45 y/o, and increases rapidly at 65+ (Klein, 1991). However, it has proved surprisingly difficult to demonstrate a causal relationship between lack of visual acuity and accident rates. It seems likely that drivers with poorer visual acuity compensate by driving more carefully and more slowly (and not driving in hazardous conditions), thus counteracting the effect, at least to a certain extent. Moreover, the necessity for 20/20 vision in normal driving conditions is not immediately apparent . Some studies have found links between certain aspects of visual perception and 'riskiness' in driving, but the link between this and distraction is not clear (Burg 1971, Evans and Ginsburg, 1985). More to the point it is not clear why decline in visual acuity should have a strong effect (or any effect) on driver distraction: especially given that these distractions, as mentioned above, tend to be visual.

3.53 Therefore there are other studies in this field which are possibly more relevant. Staplin et al (1997) have demonstrated that selective attention and attention switching abilities deteriorate with age. This fits in with the attention model proposed by Neisser, in which attention is a skill. Like most skills, it is immature in the late teens and early twenties, improves up until middle age, and then deteriorates. A number of laboratory experiments have taken place which test the hypothesis that multitasking is harder for the elderly. These usually use the methodology discussed above, and it seems that there is a link between inability to perform selective attention tasks of the sort discussed above (that is, where there is an auditory task which must be carried out concurrently with a driving task), and age. (See for example Brouwer et al, 1991)).

3.54 Following on from these experiments, various other studies have demonstrated that older adults respond more slowly to 'surprising' phenomena than younger drivers, thus suggesting that perhaps old people's propensity to have accidents where 'distraction' is counted as a variable is due to other features of the ageing process other than the strictly attentional: that is, slower reaction times, and other factors.

Young Drivers

3.55 Young drivers are disproportionately represented in accident statistics. In Great Britain car drivers aged 17 to 21 account for 4.4% of license holders, but represent 13% of all car drivers involved in accidents (Department of Transport, 1998). The AAA North Carolina (Stutts et al, 2001) project found that young drivers (under 20) were the most likely to be caught in a distraction related accident, although this was mostly related to internal-to-vehicle distraction.

3.56 This might seem to be contradicted by the data on old drivers above. But it must be remembered that most of the pure driver distraction data relates to multitasking (and not, therefore distraction proper). It may well be that young drivers are more susceptible to distraction given their high desire for arousal and risk (see below).

3.57 However, as always, we should remember that these are merely hypotheses. A study by Ramney and Pulling (1989) discovered no link between a battery of cognitive tests and driving ability. It is possible that the small numbers (n=50) were responsible for this, but it should serve as a reminder that these issues are complex.

Fatigue/Age

3.58 Ingwerson, (1995) discovered that the 17-25 age group were heavily over-represented in an Australian database for fatigue-related accidents. Given that fatigue seems to be strongly related to 'relaxing attention' (Kenny, 1995) which can in some cases lead to the driver losing control of the car (especially in information poor monotonous driving conditions), this seems to suggest a link between youth and inability to sustain attention. This fits in well with arousal theory. Zuckerman (1979) argues that young people are more likely to be 'sensation seekers', and that, therefore, in 'boring' driving conditions they are more likely to 'drift off' or otherwise shift attention away from the driving task. Again, this would make them more likely to fall prey to 'phototaxis' or the 'moth effect' given that attentional narrowing (and concentration on the task) would give way to a broader attentional sweep, during which one is more likely to be distracted.

CONCLUSION

3.59 It seems highly likely that young (17-25) and old (65+) drivers have particular attentional difficulties and that the consumption of alcohol and lack of sleep generally reduce the driver's ability to multitask, process information, and carry out adequate visual searches. This would therefore seem to indicate that drivers in these two age groups, because they cannot 'control' their attentional skills, run a higher risk of distraction and that this risk would be increased by the influence of alcohol, lack of sleep etc.

« Previous | Contents | Next »