Reorienting Driver Attention with Dynamic Tactile Cues Cristy Ho, Rob Gray, and Charles Spence Abstract—A series of three experiments was designed to investigate whether the presentation of moving tactile warning signals that are presented in a particular spatiotemporal conﬁguration may be particularly effective in terms of facilitating a driver’s response to a target event. In the experiments reported here, participants’ visual attention was manipulated such that they were either attending to the frontal object that might occasionally approach them on a collision course, or else they were distracted by a color discrimination task presented from behind. We measured how rapidly participants were able to initiate a braking response to a looming visual target following the onset of vibrotactile warning signals presented from around their waist. The vibrotactile warning signals consisted of single, double, and triple upward moving cues (Experiment 1), triple upward and downward moving cues (Experiment 2), and triple random cues (Experiment 3). The results demonstrated a signiﬁcant performance advantage following the presentation of dynamic triple cues over the static single tactile cues, regardless of the speciﬁc conﬁguration of the triple cues. These ﬁndings point to the potential beneﬁts of embedding dynamic information in warning signals for dynamic target events. These ﬁndings have important implications for the design of future vibrotactile warning signals. Index Terms—Haptic I/O, human factors, human information processing, automotive Ç 1 INTRODUCTION I N recent years, there has been a great deal of interest in the development of multisensory driver assistance sys- tems (e.g., see [1], [2]). While the trend for the future would appear to be the development and installation of completely autonomous driver systems that perform the majority of ‘accident-prone’ driving tasks (e.g., parking, lane changing, etc.) at the push of a button or increasingly via voice, or some other form of gestural, control (e.g., [3]), the design of collision warning systems remains an important area of applied research (see [4]). These systems are designed to keep drivers informed with regard to critical driving situa- tions that they may want to intervene in, should the autono- mous driver systems (for whatever reason) go astray. The development of effective collision warning systems repre- sents a potentially low-cost solution to reducing the number of accidents on the road, which are often associated with the loss of life and substantial ﬁnancial costs (see [5]). Research in this area has demonstrated that the presen- tation of tactile warning signals can offer an effective means of alerting a driver to a potential danger and rapidly orienting their spatial attention to the location, or direction, of interest in the environment (see [6] for a review). It is often said that tactile warnings may be more effective than equivalent warning signals presented in other sensory modalities. The reason for this being that the sense of touch is relatively uninvolved in the task of driving and any other secondary tasks that a driver may engage in, such as talk- ing on a phone or using the in-car navigation system [7]. There is certainly already some evidence to support such a proposal (e.g., [8], [9]). However, one limitation with the use of tactile warnings is the difﬁculty associated with con- veying meaning (or some information about the collision event other than merely a spatial direction) to the driver. In the case of visual and auditory warning signals, the cues themselves are typically semantically meaningful in that they are naturally mapped onto certain well-learnt external events. A red exclamation mark, for instance, is generally understood as some kind of high importance error signal; whereas a car horn sound is interpreted as the driver of a vehicle trying to attract someone else’s attention [10]. Meanwhile, in the case of tactile warning signals, it is unclear whether there exist only a limited number of uni- versally recognised tactile patterns or vibrations that intui- tively signify one thing or another (e.g., rumble strips on the road indicating a driver’s approach to a roundabout). It is also unclear whether with people’s increasingly wide- spread exposure to tactile interfaces, drivers will more eas- ily be able to master various different types of symbolic tactile signals. This is similar to the case of mastering the auditory cues associated with a rear parking sensor system before their meaning become instantly recognisable. In this regard, previous research on the presentation of structured abstract tactile information (known as tactile icons, or ‘tactons’) has demonstrated that people can reli- ably identify no more than two to three different vibrotac- tile parameters (e.g., rhythm, roughness, and spatial location; e.g., [11], [12]). This appears to suggest that the number of distinct types of information that can be commu- nicated via vibrotactile cues by a given system may be quite limited. For instance, Jones and her colleagues have  C. Ho and C. Spence are with the Department of Experimental Psychology, University of Oxford, Oxford OX1 3UD, United Kingdom. E-mail: cristy.ho@psy.oxon.org, charles.spence@psy.ox.ac.uk.  R. Gray is with the School of Sport and Exercise Sciences, University of Birmingham, Birmingham B15 2TT, United Kingdom. E-mail: r.gray2@bham.ac.uk. Manuscript received 17 July 2013; revised 28 Oct. 2013; accepted 12 Nov. 2013. Date of publication 2 Dec. 2013; date of current version 14 May 2014. Recommended for acceptance by R. Klatzky. For information on obtaining reprints of this article, please send e-mail to: reprints@ieee.org, and reference the Digital Object Identiﬁer below. Digital Object Identiﬁer no. 10.1109/TOH.2013.62 86 IEEE TRANSACTIONS ON HAPTICS, VOL. 7, NO. 1, JANUARY-MARCH 2014 1939-1412 ß 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.