Innovative in-vehicle technologies are increasing the scope of activities drivers can perform while driving. They increase drivers’ communication capabilities, and provide various safety benefits. One ubiquitous manifestation of the new technologies is the in-vehicle cellular phone and the use of portable cellular phones while driving. As the number of people with cell phones increases, so does the number of drivers who use in-vehicle cell phones. In a nationally representative telephone survey conducted by the National Highway Traffic Safety Administration (NHTSA) in April 2002, more than 60% of the respondents said they have a cellular phone, and about 30% used the phone to make or receive calls while driving (Royal, 2003). This is a 10% increase in the number of people using phones while driving, based on an NHTSA-sponsored observational survey conducted a year and a half earlier (Utter, 2001).1

Talking on the phone while driving is a distracting task, but it is not the only potentially distracting task observed on the road. The same NHTSA-sponsored survey revealed that talking or listening to someone on the phone is less frequent than talking to other passengers (81% of respondents), changing radio stations, cassettes, or CDs (66%), and eating or drinking (49%) (Royal, 2003). Furthermore, the actual frequency of use, at least of hand-held phones, may not be that high. In one nationally representative observational study conducted in 2000 (Utter, 2001) at 690 controlled intersections, across all times, all locations, and all drivers, only 3.0% of the drivers were seen using a hand-held cell phone (however, this is most likely an underestimate of the true use of cell phones, because the frequency of drivers using hands-free phones could not be assessed).

From a cognitive, human-information-processing perspective, the use of the phone while driving constitutes a time-sharing situation in which driving is the primary task (hopefully) and talking on the phone is the secondary task. As such, the phone task can be distracting, disruptive to driving, and potentially dangerous. This is the general conclusion of NHTSA reviews of the literature in this area (Goodman, 1997; Goodman, Tijerina, Bents, and Wierwille, 1999). This general conclusion is supported by laboratory and driving simulation studies (McKnight and McKnight, 1993; Alm and Nilsson, 1993; Strayer and Johnston, 2001), closed-track road studies (Ishida and Matsuura, 2001; Hancock, Lesch, and Simmons, 2003), and open-road studies (Brown, Tickner, and Simmonds, 1969; Harbluk, Noy, and Eizenman, 2002). When the driving or driving-like tasks are sufficiently difficult, all these studies show some deterioration in these tasks when secondary cognitive perceptual-motor tasks must be performed in parallel to the driving task. In addition, two epidemiological studies have demonstrated an association between in-vehicle phone use and crashes (e.g., Redelmeier and Tibshirani, 1997; Laberge-Nadeau, Maag, at el., 2001).

A paradigm common to all the experimental studies that have been reviewed is that both the driving task and the distracting cell phone task are experimenter-paced (unless driving speed is one of the measures of driving performance), and the study compares the performance of one or more control conditions (no distraction or listening to music) with one or more distracting conditions (math operations, short-term memory tasks, conversations, message comprehension and recall, etc.) in a situation where each condition is presented only once, or for one block of trials.

However, these two common features – fixed paced tasks and one trial or block of trials in each condition – are not typical of driving in general nor of driving while performing other tasks in particular. Driving and conversing on the phone are both partially self-paced tasks. Drivers can often adjust their speed to control the rate of information input, and they can often pause in the conversation or postpone a conversation when the driving task is too demanding. In addition, as in other tasks in life, performance improves with practice and feedback, so that the more practiced a task is, the more it (or parts of it) can be automated; consequently, more attentional capacity is left for other tasks. Thus, practice at a task can enable a person to time-share it effectively with other tasks. This has been repeatedly demonstrated in studies on perceptual-motor task performance and time-sharing of various tasks (e.g., Wickens and Hollands, 1999).

In driving, much of the experience-based improvement is based on subtle and gradual changes in information acquisition and processing. This has been demonstrated in changes in visual search (Mourant and Rockwell, 1972), and in the automation of shifting gears (Shinar, Meir, and Ben-Shoham, 1998). There is also evidence for the self-paced nature of driving, as a means of compensating for information overload or underload, and it comes from several studies of driving performance and behavior. Thus it has been demonstrated that drivers who are fatigued tend to speed up (to increase the rate of information input and thus reduce the boredom of the driving task) (Fuller, 1984). To counter overload from using the cell phone while driving, drivers often increase their safety margins by increasing headways to vehicles ahead (Brookhuis de Vries and Waard, 1991; Ishida and Matsuura, 2001), or by reducing their speed (Ishida and Matsuura, 2001; Waugh, Glumm, et al., 2000). Finally, when fatigued, driving performance can actually be aided by a “distracting” task such as a conversation (Drory, 1985), or a cognitively challenging game (Oron, Ronen, Shinar, and Cassuto, 2002).

In addition to pacing or modifying the driving, drivers can also pace the phone conversation. This was actually demonstrated in one study where driver response times to questions asked over the phone were delayed relative to responses made to the same question when the conversant was a passenger in the car, or when the questions were asked while the car was stationary (Waugh, Glumm, et al., 2000).

In terms of crash involvement, the actual contribution of engaging in a phone conversation is hard to assess, and has been estimated from as low as 10 fatalities per year to as high as a 1,000 (Hahn and Dudley, 2002). In a NHTSA-sponsored telephone survey of a nationally representative sample of drivers, Royal (2003) reports that 3.5% have had a crash within the past 5 years which they attributed to distraction (of any kind), and only 0.1% of the drivers (or 0.5% of those using cell phones while driving) attributed a crash to distraction from a cell phone. Royal concludes that based on driver perceptions, cell phones – at their current use levels – account for only 3% of all distracted-driving crashes. In terms of cost-benefit ratio, researchers from the Harvard Center for Risk Analysis concluded that a ban on the use of cell phones while driving would cost $700,000 per quality-adjusted life-year saved (Lissy, Cohen, Park, and Graham, 2000), which is less cost-effective than many other approaches to saving lives. However, it is very possible that as the use of cell phones spreads to more drivers and to more and longer conversations, the actual crash risks due to phone distractions will increase as well. An interesting historical perspective is provided by Goodman (1997) who noted that in 1938 concerns similar to those that are raised now with respect to cell phones were raised with respect to allowing drivers to listen to the radio while driving. This perspective raises the issue of whether or not the introduction of new cognitive tasks to the driving situation necessarily impairs driving significantly, beyond the level that drivers are willing to tolerate, and if so under what conditions and for how long. Like other psychomotor skills, it is possible that the complex task of sharing the driving task with the listening/talking phone task can be improved with practice.

The objective of this research was to determine if there are practice-related improvements in the use of cell phones during a simulated driving task. This is the second study we have conducted on this issue. In the first study (Shinar, Tractinsky, and Compton, 2002), we examined the effects of practice on drivers’ ability to combine a phone task and driving in a simulator. The drivers were all students with 5+ years of driving experience. The participants drove an STI simulator either without a phone, with a hands-free phone, or with a hand-held phone. The distraction task was either a comprehension task in which drivers heard a short essay and were then questioned on it, a short-term memory-span task where the drivers heard a sequence of six 1-2 digit numbers and then were asked to state whether a seventh number was one of the six or not, and a math operations task where the drivers had to perform a string of operations on single-digit numbers. To assess the effects of practice, the drivers performed the driving task with the different distractions on five sessions over a period of two weeks. In general, performance on some of the driving measures (but not all) was poorer in the presence of the phone distraction task. However, no discernable learning effects were obtained on either the driving task measures, the phone task measures, or their combination. We concluded that the reason for the absence of any practice effect was that the driving task was too easy for these participants, and that they reached their asymptotic level already on the first session.

To further examine learning/practice effects, in the present study the driving task difficulty was increased by (1) requiring the drivers to drive at higher speeds than before, and (2) by testing two more groups of drivers: novice drivers with minimal driving experience and older drivers who were expected to have more difficulty in time-sharing the driving and phone tasks (Hancock et al, 2003; McKnight and McKnight, 1993; Strayer and Johnston, 2001). Thus, the purpose of the study was to see if, when the driving task is sufficiently difficult or the driver is less capable, there is a learning effect whereby drivers learn to combine the driving task with the phone task. With respect to the distraction, we used two types of tasks: a math operations task (as in the previous study), and an emotionally-involving conversation. The math operations task has been shown to be sufficiently taxing to interfere with driving performance (McKnight and McKnight, 1993; Shinar, et al., 2002), but its validity as a surrogate for a phone conversation is questionable. The conversation task has the needed ecological validity but suffers from the fact that it is much less structured and – to be emotionally-involving - has to be individually-tailored to each participant.

1As these studies involved different methodologies and sampling plans (self-reports of use over time versus direct observations in specific locations), their results are not directly comparable.