Driver safety and information from afar: An experimental driving simulator study of wireless vs. in-car information services

Abstract

Cars have changed from pure transportation devices to fully interactive, voice-based systems. While voice interaction in the car has previously required on-board processing, the growing speed and ubiquity of wireless technologies now enable interaction with a distant source. Will the perceived source of the information influence driver safety, responses to the information, and attitudes toward the computer system and car? A between-participants experimental design $(N=40)$ of computer proximity—in-car vs. wireless—using an advanced car simulator, found that people's driving behavior, verbal responsiveness, and attitudes are affected by computer proximity. A path analysis shows two counterbalancing effects of computer proximity on driving behavior: drivers feel more engaged with the in-car system than the wireless system, which leads to safer driving behavior; however, drivers also drive faster while using the in-car system than the wireless system, which leads to more dangerous driving behavior. Consistent with greater feelings of engagement with the in-car system, people also feel less discontentment with the in-car system and self-disclose more to the in-car system. Positive perceptions of information content also lead drivers to be more persuaded by driving recommendations. Implications for the design of wireless systems are explored.

Introduction

Computers in cars are moving from just control under the hood to actively interacting with the driver (Boehm-Davis et al., 2003; Marcus, 2004). They are providing services such as rich computer navigation assistance, providing moment-by-moment relevant information to drivers about upcoming traffic, road conditions, local weather reports, and driving directions. Such computer-based services could also provide other potentially desirable information such as travel destinations, proximate gas stations, entertainment, commerce, education, health monitoring, car maintenance instructions, etc. (Lee et al., 1999; Marcus, 2004).

With existing high demands on a driver's visual attention, many on-board computer systems utilize speech technologies. While speech-based interfaces may seem to be a safer option than graphical user interfaces in cars, there remains a concern that the cognitive demands associated with richer communication will impair driving safety. As human-computer interaction becomes more enriched and engaging, there may be problems analogous to those involving drivers talking on mobile phones on the road (McKnight and McKnight, 1993; Redelmeier and Tibshirani, 1997; Strayer et al., 2003, Strayer et al., 2005), particularly because speech-based interfaces involve two-way communication between driver and computer, unlike the existing one-way communication between the radio (and other outputs) and the driver.

Thus, as people interact with information services in cars, human perception, safety, and system use must be investigated and considered in the early stages of interface design. The practical question is how one should design such vehicle user interfaces that are helpful to drivers and, more importantly, do not hinder safe driving performance. Simultaneously, the theoretical question is what model predicts the kinds of responses observed from drivers interacting with computers while on the road.

A combination of real world mishaps and controlled experimental studies has shown that several factors significantly affect driver responses to voice interfaces in cars, including perceived voice gender, emotion, and even age. The BMW 5-series released in Germany included a voice-based navigational system, featuring a computer-generated voice with female characteristics. Although these drivers were well-aware that the voice was computer-generated, they reacted with gender stereotyped responses, ultimately rejecting the female voice and demanding a product recall (Nass and Brave, 2005). BMW switched the female voice to a male voice and re-cast the navigational system voice in the role of a co-pilot (Macneil and Cran, 2004). Along with gender, experimental studies have shown how perceived voice emotion affects not only driver attitudinal responses to the interfaces, but even safe driving behavior: drivers whose emotions are matched by the emotions of the voice interface feel more positively about the system and drive more safely (Nass et al., 2005). Perceived age also affects driver responses to in-car vehicle user interfaces. Contrary to intuitions that older drivers would prefer older voices and younger drivers would prefer younger voices, older drivers actually prefer younger voices as their Driver's Assistants rather than older voices because they attribute characteristics of younger people to those Driver's Assistants, e.g., better eyesight (Jonsson et al., 2005). While significant differences are readily predicted by evocative interface characteristics of gender, emotion, and age, there are more subtle characteristics of computer interfaces that are yet to be explored.

One of the most important and unappreciated trends in car systems is that information no longer comes solely from the car. In the traditional model of interactive car systems, the navigation system sat in the car and the driver manually controlled features such as windows, lighting, climate, music, etc. It is a historical happenstance that the computer which controls all of these services is placed in the car. Today, there is no reason why the car could not be wirelessly controlled by a computer elsewhere, e.g., remote-controlled heating for the car so that one can warm it up before going out onto the snowy roads. Taking this type of remote service further, new wireless technologies such as GPS, GPIS, and OnStar, allow for a whole new set of services to appear, including location-based services, remote emergency services, and satellite radio. For example, the OnStar system provides remote emergency services by trained emergency agents via hands-free calling for drivers with built-in microphones, connection to a cellular network, and GPS technologies.

As vehicle user interfaces are being deployed as both on-board computers and as receivers and senders of external information, communicating wirelessly from afar, there may be more at stake in these different implementations than mere engineering trade-offs. While these content-based differences could be readily studied in an experimental setting, there are more interesting questions to ask from a theoretical perspective. For example, how and why would the proximity of computer sources make a difference in user perceptions, attitudes, and behaviors? Based on existing research in computers as social actors (CASA) theory (see Section 2.1) and findings in interpersonal interaction between humans, this research focuses on the idea that computer proximity will affect people in the same way that human proximity affects them in interpersonal contexts.

This experiment was also part of a broader inquiry into how people make sense of the technological decoupling of bodies and brains from faces and voices of social actors (Nass and Brave, 2005). Though this decoupling is not typical in interpersonal interaction, it is quite common in technologically mediated interactions such as those introduced by the telegraph and telephone (Kahn and Whitehead, 1994). Similarly, older technologies such as written language have introduced a decoupling of language use from co-located and synchronous interaction with conversational partners. More recently, technologies such as live radio and the Internet also separate bodies and brains (e.g., web servers) from faces and voices (e.g., interfaces presented on each user's personal computer). The broader theoretical research question at hand is: how does the perception of communicative sources as physically distant affect interactions between communicators? In this particular context, how does the perception of computer proximity affect interactions involving drivers and vehicle computer interfaces?