Cursing the driver who swerved in front of you is one thing, but cursing at your own car is quite another. It may become increasingly common as cars take on an ever-expanding role of keeping those in the driver’s seat driving safely.

For many people, the experience of being gently reprimanded by a GPS system is no longer novel. (Mine seems to be forever sighing and saying “Recalculating”). But as researchers look for ways to make cars safer, they are not only investigating how voice systems can lighten the load by performing related tasks (like making phone calls as discussed in this issue’s article “Listen Up!”), they are looking at what kinds of voices drivers will respond to best if they are being corrected for bad behavior and whether certain voices or types of “conversation” encourage better driving.

Dr. Clifford Nass, director of Stanford University’s Communication between Humans and Interactive Media (CHIMe) program, has been investigating how differences in tone of voice, words used, and the gender and age of the computer-generated voice talking to a driver affects the driver’s response.

His research has shown, among other things, that gender still plays a role: male voices typically get more attention and respect than female voices, from both male and female drivers. In fact, a few years ago BMW in Germany had to change its voice system from a female voice to a male because male drivers got angry being told what to do by a woman.

Another interesting finding was that older drivers respond better to younger voices. Dr. Nass speculates that this may be because older drivers

think of younger people as having better vision, and thus being more likely to see things the driver may have missed.

On the other hand, regardless of driver age, one has to be careful that the voice highlighting things the driver may have missed uses psychological principles effectively. In its research, Dr. Nass’s lab tested the following series of statements in a simulated driving situation: “You’re not driving very well and you need to pay more attention.” Most people respond better to praise than criticism, so it was perhaps not

surprising that that the drivers resented this critique and their driving got worse. It deteriorated even more with a second warning telling them,

“You really need to be more careful.” Finally, the voice told the driver to pull over, resulting in such anger that the at least one driver crashed the car.

Nevertheless, Dr. Nass is optimistic that as we learn how to find the right balance of gender, tone, and words, car commands will ultimately

improve the way people drive. For instance, a perky voice can be irritating to a driver who is upset, but inspire better driving among happy drivers. Conversely, a calm, even slightly depressed-sounding voice actually improved the driving of upset drivers by 40 percent. Of course, for this information to be useful, the voice automation system has to be able to detect whether a driver is driving erratically because he’s upset or because he’s happy but distracted. Maybe it will ultimately need to be combined with physiological sensors and breathalyzers so it can turn on the appropriate voice when needed.

Ultimately, it may come down not to an issue of what we are technologically capable of, but how much we are willing to have our cars know about us. If your car knows you were happy driving home from a secret rendezvous, do you really want to risk it letting your spouse know when she gets into the car.