The premise underlying certain common traffic laws is that multi-tasking behind the wheel–tuning a radio or placing phone calls while driving, for example–is safe so long as it is “hands-free.”
A new study of drivers interacting with automotive voice-controlled information systems shoots that premise in the heart.
Not only is there a significant cognitive load–and thus, heightened distraction–associated even with speech-based functions, researchers found, but vocal interactions impose a detrimental burden on the brain that lingers on even after a given task has ended.
This effect persists long enough that even a vehicle traveling at only 25 mph can cover the length of more than three football fields before a motorist can focus once again solely on driving.
The study, conducted for the AAA Foundation for Traffic Safety, was headed by Dr. David Strayer, a professor of cognition and neural science at the University of Utah. It can be accessed in full, free of charge by clicking here.
Strayer and his team recruited 257 drivers, evenly balanced between males and females, ranging in age from 21 to 70, with five to 55 years of driving experience and no at-fault accidents in the last five years. After passing a defensive driving course, they were assigned among 10 different models of 2015 vehicles, each equipped with its manufacturer’s auditory/vocal in-vehicle information system (IVIS) that enabled voice-controlled phone calling, number dialing, and radio tuning.
After being familiarized with his or her IVIS system, each driver maneuvered a 2.7-mile street loop through a light-traffic suburban residential area, obeying stop signs and stoplights and the 25 mph speed limit.
As they drove, the subjects completed a series of six intermittent vocal tasks, including tuning their radio to various AM and FM stations and making phone calls by voice-dialing or using their contacts lists. They were allowed a minute and a half to complete each task, and their responses were recorded by monitoring equipment that measured reaction time, accuracy, and gaze tracking.
Afterward, the subjects rated how mentally and physically demanding the tasks were, how much time pressure they felt, and how “insecure, discouraged, irritated, stressed, and annoyed” they were by the various activities.
After five days at home when they were to practice with their IVIS system, the subjects returned for another round on the test course.
In analyzing their findings, the study team reached these conclusions:
- Using a voice-based system to complete common tasks produced “a significant increase” in a motorist’s cognitive workload, compared to simply concentrating on driving. Overall, the workload ratings were “surprisingly high,” by the researchers’ calculations, averaging 3.34 on a five-point scale and ranging from 2.37 to 4.57.
The average and upper ranges were “significantly more demanding than typical cell phone conversations, which have cognitive workload levels around 2.3,” Strayer writes.
This “should serve as a caution that these voice-based interactions can be very mentally demanding and ought not to be used indiscriminately while operating a motor vehicle.”
Not surprisingly, the systems that were “more intuitive, less complex,” and faster in completing tasks had lower workload scores “than more rigid, error-prone, time-consuming ones.” Unfortunately, the researchers point out, “[m]any of the systems that are currently available tend to be complex and error prone, with inconsistent behavior….”
- Older drivers experienced a “significantly greater” cognitive impact than those who were younger and more likely to be tech savvy. Even when just driving, the reaction time of older drivers tended to be about 18% slower than that of younger subjects. “In essence, the age-related differences…doubled when participants interacted with the IVIS.” This gap is amplified even further “as the complexity of the task increases,” Strayer writes.
This is important, he points out, because older drivers (55-64 years old) are the “most frequent purchasers of new vehicles” and thus are likely to encounter voice-based systems that “induce high levels of cognitive workload.”
- Practice doesn’t seem to matter much. “Even after five days of practice, there were still large [cognitive] costs associated with IVIS interactions,” compared with non-distracted driving, Strayer writes. “It appears that the impairments from using the IVIS cannot be practiced away…. IVIS interactions that were easy on the first day were also easy after five days of practice, and those IVIS interactions that were difficult on the first day were [still] relatively difficult to perform after five days of practice.”
He cites what he calls “the Power Law of Learning,” which posits that “the biggest improvements occur early in training…. This implies that any additional practice [beyond five days] will have diminishing returns….”
- Voice-based tasks affect gaze behavior. Instead of encouraging drivers to keep their eyes on the road, “we found that many participants routinely glanced at the [visual] displays” on their dashboards during interactions, even though the tasks had no visual requirement, Strayer notes. The frequency of important glances “to the forward roadway and side- and rear-view mirrors” correspondingly decreased.
“[I]t is increasingly evident that natural visual scanning behavior is fundamentally coupled to cognitive processing demands,” he states. “Quite simply, it is incorrect to assume that talking to your car is an ‘eyes-free’ activity.”
- Perhaps the biggest stunner of the study is “evidence of persistent [cognitive] interference following the IVIS interactions.” The impact of a momentary cognitive diversion lingered for up to 27 seconds after an IVIS interaction had been completed before returning to the baseline performance of sole-focused driving.
“To put this in context,” Strayer explains, “at 25 mph a vehicle would have traveled 988 feet before the residual [cognitive] cost had completely dissipated”–the length of more than three football fields.
“Just because a driver terminates a call,” Strayer warns, “does not mean that they are no longer impaired. This “lingering act of disengaging,” he says, has “implications for self-regulatory strategies, such as choosing to dial or send a text message at a stoplight, because the costs of these interactions are likely to persist when the light turns green,” as the driver reestablishes “situation awareness of the driving environment that was lost during the IVIS interaction.”