Does talking to a passenger affect driving performance?

By Michael G. Wilding

 

Abstract

The effect of talking to a passenger on driving performance was investigated.  It was predicted that talking to a passenger who does not modulate (change the intensity of) their conversation based on hazards in the road, would significantly decrease driving performance compared with talking to a passenger who did modulate their conversation. Participants completed a hazard perception task and answered general knowledge quiz questions either during (un-modulated condition), or in between (modulated condition) hazard perception clips, there was also a silent control condition. The general knowledge questions were used in order to simulate conversation with a passenger.  Participants observed marginally significantly fewer hazards and answered significantly fewer questions correctly when the tasks were presented at the same time (un-modulated).  There was no significant difference between reaction times across the three conditions.  Further research ideas and implications are discussed.

Introduction

In recent years, the effects of using a mobile phone whilst driving have been an area of substantial research interest.  This has resulted in a large body of research that provides convincing evidence to suggest that talking on, or operating a mobile phone whilst driving can increase the risk of being involved in an accident by significantly increasing reaction times and reducing hazard perception abilities (Mcknight & Mcknight 1993, Alm & Nilsson 1995, Strayer & Johnston 2001), as well as affecting lane position and regulation of speed (Serafin et al 1993, Pachaudi & Chapon 1994). The increase in risk can be as much as four times that of driving without using a phone (Redelmier & Tibshirani 1997), which is comparable to having a blood alcohol level higher than the legal limit.  The United Kingdom Department for Transport’s decision to ban the use of hand-held mobile telephones whilst operating a motor vehicle in December 2003 is an example of the perceived level of danger involved, and the impact that mobile phone research has had on the community.

The same deficits in driving performance are apparent even when the driver is using a hands-free kit (Boase et al 1988, Alm & Nilsson 1990, Treffner & Barrett 2004) suggesting that the problems are not caused solely by holding the phone whilst driving, but also due to the extra cognitive demand imposed by conversation.  Haigney et al (2000) found that mobile phone use during a simulated driving task significantly increases cognitive demand on the driver (as measured by heart rate changes), which resulted in a reduction of safety margins and thus increased accident risk.
 
Strayer & Johnston (2001) found that when driving and performing another task (dual-tasking), attention was focused away from driving, and instead diverted towards the most cognitively engaging of the two tasks (in this case the phone conversation) decreasing driving performance.  Dual-task theory (Kahneman 1973) suggests that there is a central bank of finite cognitive resources, which is available for carrying out tasks requiring intellectual effort.  A person’s ability to perform two or more tasks simultaneously is governed by the amount of resources each task requires from the central bank.  If the collective demand on the central bank from both tasks does not exceed its total capacity, then both tasks will be carried out as well as they would be if they were being carried out on their own.  However if the demands exceed capacity then one or both tasks will suffer as a result.  Driving and holding a conversation is an example of a dual-task situation; problems arise when the driver suffers from “mental overload” (Pachaudi & Chapon, 1994) performance is decreased due to the  fact that the total task demands of driving and holding a conversation have exceeded the available resources of the central bank.

Drivers may also have difficulty compensating for their impaired performance as they may not even be aware that their performance is affected. Boase et al (1988) interviewed drivers who regularly used hands free kits, and found that none of the participants’ thought that their driving performance was affected by doing so.  However tests using a simulated driving task found their performance to be significantly worse when holding both simple and complex phone conversations, showing that participant’s performances were adversely affected despite them thinking otherwise.

Parks (1991) noted that the reason talking on a mobile phone is seen to be more distracting whilst driving than talking to a passenger within the car is due to the timing of the conversation.  A passenger can also see the road along with the driver and will modulate the intensity of the conversation depending on their perception of the hazards present.  This helps to balance the cognitive load imposed upon the driver and will stop the central bank from becoming overloaded.  The conversation can also be partially modulated in the same way by the driver.  Somebody on the other end of the phone will not be aware of the road conditions, and indeed may not even realise that their conversation partner is driving at all.  This will mean that the conversation will not slow down or become any less cognitively demanding when hazards present themselves to the driver (i.e. no modulation takes place) requiring more attention from the driver, even when difficult driving conditions are present.

Whilst the effect of mobile phone conversations on driving abilities has been extensively investigated, the effects of other passengers in the car have received little attention.  Although it was suggested by Parks (1991) that passengers will modulate conversation according to changes in the difficulties that the driver faces, the question of what would happen if a passenger did not modulate their conversation has not been investigated in depth.

An analysis of police reports from road traffic accidents in the USA between 1995-1999 conducted by Stutts et al (2001) found that of all drivers who were reported as being distracted at the time of the crash, 10.9% attributed the cause of their distraction to be another occupant in the vehicle, whereas only 1.5% were distracted due to talking on a mobile phone.  Distracting passengers were reported as the cause of more than ten times more accidents than talking on a mobile phone.

Fairclough et al (1991) found that both talking on a mobile phone and holding a conversation with a passenger whilst completing a driving route in a car led to increased stress levels and cognitive load, as well as increased time taken to complete the circuit compared with a control group who drove on their own without talking.

Merat & Jamson (2005) found evidence to suggest that talking to an “inconsiderate” passenger whilst driving was just as distracting as talking on a mobile phone.  Participants operated an advanced driving simulator and worked their way through four driving scenarios of increasing difficulty.  These scenarios were; following a car along a straight section of road, following a car along a curved section of road, a braking event, and a coherence event, with four breaks prior to the start of each scenario.  Participants were asked to complete a sentence verification task by the experimenter (in order to simulate conversation) either in the breaks between driving (considerate passenger condition), or during the driving scenarios (inconsiderate passenger condition). In the mobile phone condition, participants were asked to complete a working memory digit recall task presented over a mobile telephone during the driving scenarios. There was also a silent control condition.

Impairment of driving ability (as measured by braking headway, maintenance of speed, and lane position) whilst talking to an inconsiderate passenger were found to be the same as talking on a mobile phone, whilst conversing with a considerate passenger was not significantly different from driving alone.  This shows that if passengers do not modulate their conversation as suggested by Parks (1991), then it becomes just as distracting as talking on a phone.  However disruptions were only apparent in the two more difficult driving events (braking and coherence), compared with the two simple ones (straight, and curved following).  This decrease in performance occurred as the task demands exceeded the capacity of the central bank.  Not only was driving performance worse in the mobile phone and inconsiderate passenger conditions, but performance was worse on the digit recall and sentence verification tasks as well, suggesting that both tasks were requiring large amounts of attention, which when combined, exceeded the capacity of the central bank, and both tasks suffered as a result.

There is evidence to suggest that talking to a passenger may even be more distracting than using a mobile phone.  Spence & Read (2003) presented participants with word triplets (e.g. “jewel, reason, under”) to shadow (repeat back out loud) whilst operating a driving simulator.  The word triplets were presented from a loudspeaker either directly in front of the participant on the dashboard, or to the side of the participant from the passenger seat.  Passengers shadowed a greater number of word triplets successfully when they were presented in front of them, rather than from the side.  This effect is caused by cross-modal links between auditory and visual processing.  Cross-modal processing theory postulates that there is a certain degree of overlap between the processing of these two different tasks.  Interference occurs when information from these two modalities (hearing, and vision) are presented from different spatial locations, due to the fact that the processing of this information uses shared cognitive resources (Driver & Spence 1994).  Participants found it harder to focus on auditory information which did not originate from the same location as the highly demanding visual information (driving).  This research shows that cognitive demand (and thus accident risk) can be reduced if both auditory and visual information are presented to the driver from the same location. 

Conversing with a passenger would present auditory information to the driver from a different location to the visual information in the road, and if the conversation is not being modulated as per Parks’ (1991) criteria,  then the conversation will pose a far greater demand on the central bank of the driver than if they were driving alone.  This, in theory should result in the conversation with a passenger being just as dangerous as talking on a mobile phone (Merat & Jamson, 2005) if not more so, as talking on mobile phone presents auditory information from a spatial location that is nearer to the driver’s visual field. (Spence & Read 2003).

Distracting passengers in the car are reported as being responsible for more than ten times as many accidents as mobile phone usage (Stutts et al 2001) yet it is an area of interest that has not been extensively investigated compared with mobile phones, despite the implications that this research could have on improving driving safety. 

The general consensus of current literature is that conversing with a passenger is not as distracting as using a mobile phone (Ishida & Matsuura 2001, Parks, 1991).  Whilst Merat & Jamson (2005) have already demonstrated that talking to an inconsiderate passenger is just as distracting as talking on a mobile phone, the task used to simulate conversation (sentence verification) was not highly representative of real life conversation.  The present study will aim to find support for findings of Merat & Jamson (2005), whilst providing a more realistic simulation of conversation.  Further contributions to the research literature will also be made by investigating the effect that talking to a passenger who does not modulate their conversation has upon the hazard perception abilities and reaction times of driver’s.  Participants will be asked general knowledge questions either during (un-modulated conversation condition), or in-between (modulated conversation condition) hazard perception events.  General knowledge questions will be used as a simulation of conversation with a passenger.  It is hypothesised that both the driving performance (measured by hazard perception scores and reaction times), and the number of questions answered will be significantly worse, when they are presented at the same time (un-modulated conversation condition), as opposed to separately (modulated conversation condition).

Method

Participants

An opportunity sample of thirty-five (17 male, 18 female) participants, aged between 17-51 (M: 24 SD: 1.11) were used.  All participants held a full UK driving licence at the time of the study.  The driving experience of the participants ranged from less than one to over thirty years.

 

Design and Procedure

Participants sat in a bucket car seat facing towards the screen.  A projector situated behind the participant projected a series of one minute long video clip onto the screen (see Figure 1).  Some of the video clips were acquired from a commercially available hazard perception test DVD, as used to practice for the driving theory test.  The remainder of the clips used were filmed using a camcorder. Each clip contained one major hazard which the participant was required to respond to by pressing a foot pedal.  Some clips contained other situations that may have been classed as hazards but to avoid ambiguity, only the most obvious hazard was used per clip.  The participants were un-aware of this fact and were instructed to press the foot pedal whenever they thought a hazard was present.  A window of time was used, starting from the earliest possible indication that a hazard was about to occur, and finishing when the driver in the video was required to take action to avoid the hazard (such as braking or altering course).  If the participant pressed the pedal within this time window this counted as correctly identifying the hazard.  If the pedal was pressed too late (after the hazard window had closed), or not at all, then this was counted as a failure to notice the hazard.  For each participant the total number of pedal presses per clip was recorded by a computer.  The amount of time elapsed at the time of the pedal press was also recorded by the computer, and was used as a measure of reaction time.

After each clip had finished participants were given a one minute break.  The screen displayed the text “please wait…” during this time.  After one minute had passed the next clip began.  This process repeated itself eight times – so in total each participant saw eight one minute clips, and had eight one minute breaks in between each clip.

A loudspeaker was also situated to the right side of participant, which could be used to play a recording of the author asking general knowledge questions when required.  The questions were used in order to simulate conversation with a passenger.  The decision was made to use pre-recorded questions in place of engaging the participant in real conversation in order to keep the experimental conditions as similar as possible between participants.  It also removes the participant’s ability to modulate the conversation themselves, as asking a question socially obliges the participant to provide a response (Mesthrie et al 2004).  Whilst in conversation the driver could choose not to respond if the driving task becomes too difficult, asking a question reduces the chance that this will happen, and thus provides a realistic simulation of un-modulated conversation with a passenger. Pre-recorded questions were also chosen over the experimenter engaging the participant in conversation, as simple “chit-chat” has previously been found to have no significant effect on driving performance (Mcknight & Mcknight 1993).

The questions used were general knowledge pub quiz questions. The questions used were classified as a “mixed-bag” selection and were not specific to any particular category of information.  32 questions were asked in total, and were presented in one minute blocks at the rate of four per minute (15 seconds per question) in order to allow participant’s time to provide an answer.  The one minute question blocks were presented in order to coincide with the hazard perception task (un-modulated conversation condition), or with the one minute breaks in between hazard perception clips (modulated conversation condition).  Participants were asked to say what they deemed to be the correct answer out loud.  Participant’s answers were recorded by the experimenter, who sat behind the participant out of their field of vision in order to remain inconspicuous.

There were three experimental conditions.  A control group completed the hazard perception task in seilence and waited throughout the breaks without answering any questions, in order to simulate driving without a passenger.  In the modulated conversation condition, participants completed the hazard perception task in silence, and responded to the quiz questions in the one minute breaks in between video clips.  This was to simulate conversation with a passenger who modulated their conversation, and hence only talked when the task demands on the driver were low.  In the un-modulated conversation condition, participants completed the hazard perception task and answered the general knowledge questions at the same time, in order to simulate un-modulated conversation with a passenger.  Participants in this condition waited through the one minute breaks in silence.

Results

Figure 2 shows the mean number of questions answered correctly by participants in the modulated and un-modulated conversation conditions. The number of questions answered correctly were analysed using an independent measures t-test.

Participants in the un-modulated conversation condition answered fewer questions correctly (M = 11.58, SE = .80),than participants in the modulated conversation condition (M = 13.58, SE = .82), resulting in a marginally significant effect of modulation of conversation on the number of questions answered correctly, t (22) = 1.74, p = .095.  This finding is consistent with that of dual-task driving studies, showing that performance on a secondary task is reduced whilst driving (Pachaudi & Chapon 1994, Strayer & Johnston 2001, Spence & Read 2003).

 

 

 

 

 

 

 

 

Figure 3 shows the mean number of hazards identified by participants in each condition.  A one-way independent measures ANOVA found that there was no significant effect of modulation of conversation on the number of hazards detected F (2, 32) = 1.80, ns.  However, planned contrasts revealed that participants in the un-modulated conversation condition detected marginally significantly fewer hazards (M = 6.42, SE = .38) than participants in both the modulated conversation (M = 7.17, SE = .27), and control groups (M = 7.18, SE = .33), t (32) = 1.91, p = .067, There was no significant difference between the number of hazards detected between the modulated conversation and control conditions t (32) =1.75, ns.  This finding is consistent with that of Merat & Jamson (2005) that driving performance can be significantly reduced by talking to an inconsiderate passenger.

 

 

 

 

 

 

 

Table 1. Mean reaction times for all eight hazards.
	Condition	Mean Reaction Time (ms)
hazard clip 1	modulated	39899 (SD: 942.56)
	un-modulated	38571 (SD: 3264.13)
	control	37962 (SD: 3058.96)
hazard clip 2	modulated	8980 (SD: 867.34)
	un-modulated	8485 (SD: 1526.23)
	control	8948 (SD: 658.02)
hazard clip 3	modulated	26629 (SD: 4268.21)
	un-modulated	25953 (SD:1780.73)
	control	24195 (SD: 782.64)
hazard clip 4	modulated	32416 (SD: 631.71)
	un-modulated	31602 (SD: 1080.96)
	control	32846 (SD: 1287.48)
hazard clip 5	modulated	18010 (SD: 661.62)
	un-modulated	18553 (SD: 1649.69)
	control	17527 (SD: 268.88)
hazard clip 6	modulated	42544 (SD: 723.56)
	un-modulated	42914 (SD: 234.43)
	control	41619 (SD: 788.28)
hazard clip 7	modulated	38570 (SD: 566.73)
	un-modulated	39295 (SD: 779.56)
	control	38654 (SD: 322.63)
hazard clip 8	modulated	28974 (SD: 1925.37)
	un-modulated	27841 (SD: 2036.87)
	control	28855 (SD: 301.33)

Table 1 shows the mean reaction times for all participants for all eight hazards. A multiple analysis of variance (MANOVA) showed no significant effect of modulation of conversation on reaction times, F (16, 14) = 1.157, ns.  Univariate analyses were all non-significant apart from clip six, F (2, 13) = 5.127, p = .023, eta2 = .441.  The large standard deviations are attributed to the ambiguous nature of what constitutes a hazard.

Discussion

The results of the present study show that talking to a passenger who is not modulating their conversation can significantly affect driving performance when compared with talking to a passenger who is modulating their conversation as per Parks (1991) criteria.  The increase in cognitive demand placed on the driver by engaging in this type of conversation is most evident when looking at the difference in the number of general knowledge questions answered correctly by participants.  Participants in the un-modulated conversation condition, correctly answered significantly fewer questions (M: 11.58) than those in the modulated conversation condition (M: 13.58). 

The deficit in general knowledge performance observed in the un-modulated conversation condition can be explained by the dual-task theory (Kahneman 1973).  The combined cognitive demand of performing the hazard perception task whilst trying to simultaneously answer general knowledge questions has exceeded the capacity of the central bank, and  performance on the general knowledge task suffered as a result of participants reaching “mental overload” (Pachaudi & Chapon, 1994).  The fact that participant’s needed to compensate for the increased cognitive demand posed by dual-tasking, as seen by decreased performance on the general knowledge questions relative to the un-modulated conversation condition, shows that attentional resources were already stretched to a dangerous level.  In a real driving situation, this could lead to an accident. This finding is also supported by other dual-task driving studies (Strayer & Johnston 2001, Spence & Read 2003).

A one-way ANOVA revealed that there was no significant difference between the total numbers of hazards identified for all participants.  However planned contrasts did reveal a marginally significant difference between participants in the un-modulated conversation condition, compared with the modulated conversation and control conditions t (32) = 1.91, p = .067, suggesting that engaging in un-modulated conversation with a passenger can lead to decreased ability to detect hazards.  There was also no significant difference between the number of hazards detected between participants in the modulated conversation condition and the control group, suggesting that if conversation is modulated correctly then both tasks can be carried out as effectively as if they were both being carried out on their own.  These findings are consistent with those of Merat & Jamson (2005).

It is important to note that Merat & Jamson (2005) only observed decreases in driving performance as a result of un-modulated conversation during the two most difficult driving situations.  It may have been the case that the hazard perception task used in the present study was not cognitively demanding enough when compared to real driving to yield a highly significant result, therefore only a tentative conclusion may be drawn from the differences in the number of hazards detected.

Wickens & Gopher (1977) demonstrated that people have voluntary control over how the resources in the central bank are allocated, thus allowing control over which task receives the larger share of cognitive resources when dual-tasking.  Participants were presented with two tasks to complete simultaneously.  The two tasks were to keep a pointer over a moving reference point, and to respond as rapidly as possible every time a number was presented on the screen.  Participants were asked either to allocate 100% priority to one task over the other or a 70/30% split.  Performance on each task was relative to the amount of priority that had been allocated to it, with performance increasing with increased priority and vice versa.  Participants in the present study may have decided to allocate more attentional resources towards the hazard perception task, and less towards the general knowledge quiz questions.

Strayer & Johnston (2001) found that accident risk is increased by talking on a phone whilst driving when the conversation becomes more cognitively engaging than the driving.  The conversation task used (general knowledge quiz questions) would have had little personal meaning compared with a real conversation that the participant was interested in.  Indeed answering questions may even have been perceived as an arduous task compared with the hazard perception, and hence may have not been as cognitively engaging as a conversation might have been.  Had the conversation task been more interesting, it may have demanded more cognitive resources from the participants.  Future research should aim to provide a conversation task which will be more cognitively engaging for the participant.  As even though the present study may have succeeded in reaching the aim of providing a more realistic simulation of conversation in terms of structure (Mesthrie et al 2004), it may not have achieved the same level of realism in terms of participant interest levels.

There was no significant difference between reaction times for all three conditions.  One major flaw with the present research which became apparent after debriefing certain participants is that there was a large amount of ambiguity as to what constituted a hazard, and this may have led to the large standard deviations reported for reaction times.  The age of the participant was particularly important in this regard.  In March 2002 the Driving Standards Agency (DSA) made the decision to introduce a hazard perception test as a compulsory part of the standard driving theory test.  Drivers who had taken their driving test after the introduction of the hazard perception test, did not generally report difficulty in identifying what constituted a hazard, as they were already familiar (and indeed had often had extensive practice) with the procedure.  However some of the participants, who had not had to partake in a hazard perception test as part of the driving theory, reported that they had more difficulty in discriminating what constituted a potential hazard during the task.  This practice effect could be investigated by further research.

The difference in amount of hazards noticed also led to less reaction times being reported in total for participants in the un-modulated conversation condition.  This may have led to only reaction times for the more obvious (and thus easier to respond to faster) reaction times being included in the analysis.  Further research could avoid this potential ceiling effect by assessing the difficulty of the hazards used prior to testing.  Comparing reaction times whilst taking into account the difficulty of the hazard may yield more conclusive results.  When looking at this result it is also important not to lose sight of the fact that participants in the un-modulated conversation condition recalled fewer hazards (M = 6.42) than participants in both the modulated conversation (M = 7.17) and control groups (M = 7.18). Although reaction times were not affected in the present research instance by conversation, it is arguably more preferable for a driver to respond later to a hazard as long as this response still occurs within a window that will allow the hazard to be avoided, rather than failing to respond to the hazard altogether which could potentially cause an accident.

In summary, there is extensive evidence to suggest that dual-tasking can lead to a decrease in performance one either both, or one of the tasks involved, when the collective task demands imposed on the performer are too great (Kahneman 1973, Wickens & Gopher 1977).  Driving and holding a conversation, is an example of dual-tasking that occurs in everyday life.  There are dual-task studies of driving to suggest that talking both on a mobile (Boase et al 1988, Pachaudi & Chapon 1994, Strayer & Johnston 2001) and with a passenger (Fairclough et al 1991, Merat & Jamson 2005) can decrease driving performance.  The current literature suggests that conversing with a passenger does not impose enough cognitive demand for it to affect driving performance (Parks 1991, Ishida & Matsuura 1991, Mcknight & Mcknight 1993). 

The present research has challenged the traditional view that that talking to a passenger does not distract the driver.  Participants’ who received a non-modulated conversation task answered marginally significantly fewer questions correctly, and failed to notice marginally significantly less hazards, when compared with participants who received a modulated conversation task, or no conversation at all.  This suggests that talking to a passenger can be significantly more distracting than driving in silence.  Although further research is needed in order to find more definite answers to the questions raised by this research, the findings have significant implications for drivers who regularly travel with passengers.  Drivers may not be aware that talking to a passenger is affecting their driving performance (Boase et al 1988), increased awareness of this effect could lead to a reduction in accidents caused by a distracting passenger, as drivers could make a concerted effort to assign more attentional resources towards driving if the passenger does not modulate their conversation, albeit at the expense of holding a conversation (Wickens & Gopher 1977).

References

 

Alm, H., & Nilsson, L. (1995). The effects of a mobile telephone task on driver behaviour in a car following situation. Accident Analysis and Prevention, 27, 707–715.

Boase, M., Hannigan, S., & Porter, J.M. (1988). ‘Sorry, can’t talk…just overtaking a lorry’. In E.D. Megaw (Ed.), Contemporary Ergonomics (pp. 527-532). U.K.: Taylor & Francis.

Driver, J.S., & Spence, C.J. (1994). Spatial synergies between auditory and visual attention. In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 311–331). Cambridge, MA: MIT Press.

Fairclough, S.H., Ashby, M.C., Ross, T. & Parkes, A. (1991). Effects of hands free telephone use on driving behaviour. In: Proceedings of the ISATA conference, Florence, Italy.

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.

Ishida, T. & Matsuura, T. (2001). The Effect of Cellular Phone Use on Driving Performance, IATTS  Research, 25(2), 6-14.

Kahneman, D. (1973). Attention and Effort, Englewood Cliffs, NJ: Prentice-Hall.

McKnight, A.J. & McKnight, A.S. (1993). The effect of cellular phone use upon driver attention. Accident Analysis and Prevention, 25(3), 259-265.

Merat, N. & Jamson, H.(2005).  Shut up I’m driving! Is talking to an inconsiderate passenger the same as talking on the mobile phone? Proceedings of the Third International Driving Symposium on Human Factor in Driver Assessment, Training and Vehicle Design, Maine, 426-432.

Mesthrie, R. Swann, J. Deumert, A. & Leap, L. L. (2004). Introducing Sociolinguistics, UK: Edinburgh University Press.

Pachaudi, G. & Chapon, A. (1994). Car phone and road safety. XIVth International Technical Conference on the Enhanced Safety of Vehicles No 94-S2-0-09.

Parks, A.M. (1991). Drivers business decision making ability whilst using carphones. In E. Lovessey (Ed.), Contemporary ergonomics: Proceedings of the Ergonomic Society annual conference (pp. 427–432). London: Taylor & Francis.

Redelmeier, D.A. & Tibshirani, R.J. (1997). Association between cellular-telephone calls and motor vehicle collisions. New England Journal of Medicine, 336, 453-458.

Serafin, C. Wen, C. & Paelke, G. (1993). Car phone usability: a human factors laboratory test. Proceedings of the Human Factors and Ergonomics Societys 37th Annual Meeting, Human Factors Society, Santa Monica, CA, 220–224.

Spence, C. & Read, L. (2003). Speech shadowing while driving: On the difficulty of splitting attention between eye and ear. Psychological Science, 14(3), 251-256.

Strayer, D.L. & Johnston, W.A. (2001). Dual task studies of simulated driving and conversing on a cellular telephone. Psychological Science, 12(6), 462-466.

Stutts, C. J. Reinfurt, W. D. Staplom, L. & Rodgman, A. E. (2001). The Role of Driver Distraction in Traffic Crashes, University of North Carolina Highway Safety Research Center, for AAA Foundation for Traffic Safety ,USA

Treffner, J. P & Barrett, R. (2004). Hands-free mobile phone speech while driving degrades co-ordination and control. Transport Research Part F, 7, 229-246.

Wickens, C. D. & Gopher, D (1977). Control theory measures of tracking as indices of attention allocation strategies. Human Factors, 19(4), 349-365.

 

First published 2008 copyright Michael G. Wilding

Hosting courtesy of Cheap Television