1 3

Exp Brain Res (2014) 232:1085–1093DOI 10.1007/s00221-013-3792-x

REsEaRch aRtIclE

Assessing the benefits of multisensory audiotactile stimulation for overweight individuals

Xiaoang Wan · Charles Spence · Bingbing Mu · Xi Zhou · Cristy Ho

Received: 5 October 2013 / accepted: 19 November 2013 / Published online: 6 December 2013 © springer-Verlag Berlin heidelberg 2014

Introduction

Driving safety is undoubtedly an issue of global impor-tance. It has been estimated that 1.2 million people are killed while 50 million are injured in road accidents every year (World health Organization 2004, 2013a). One of the most common types of road accidents is front-to-rear-end collision (e.g., Evans 1991; horowitz and Dingus 1992), for which driver distraction or inattention has been high-lighted as a common cause (Wang et al. 1996). In order to promote safe driving, comprehensive warning systems have been developed, with both unimodal and multisensory warning signals being intensively investigated in order to try and identify the types of signals that may be most effec-tive in terms of mitigating car accidents (see spence and ho 2008a, b, for reviews).

Other than visual warning signals, auditory or vibro-tactile warning signals have been shown to effectively improve drivers’ responses in potential front-to-rear-end collision situations (ho et al. 2005, 2006a, b; ho and spence 2005; lee et al. 2004, 2006; see spence 2010 for a general review). compared to unisensory (auditory or tactile) signals, audiotactile multisensory warning signals can be even more effective in terms of capturing a person’s spatial attention (Fitch et al. 2007; ho et al. 2007, 2009; santangelo and spence 2007). For instance, in a driving simulator study, when the participants were asked to follow another car, which might suddenly decelerate, their braking responses were faster when they were alerted by audiotac-tile signals than when they were warned by unimodal audi-tory or vibrotactile signals (ho et al. 2007).

When these findings from the psychophysical literature on tactile and/or audiotactile stimulation are applied to the field of driving safety, researchers have often tried to deliver vibrotactile warnings to the driver’s seat belt via the

Abstract We report an experiment designed to examine whether individuals who are overweight would perform differently when trying to detect and/or discriminate audi-tory, vibrotactile, and audiotactile targets. the vibrotactile stimuli were delivered either to the participant’s abdomen or to his hand. thirty-six young male participants were classified into normal, underweight, or overweight groups based on their body mass index. all three groups exhibited a significant benefit of multisensory (over the best of the unisensory) stimulation, but the magnitude of this benefit was modulated by the weight of the participant, the task, and the location from which the vibrotactile stimuli hap-pened to be presented. For the detection task, the over-weight group exhibited a significantly smaller benefit than the underweight group. In the discrimination task, the over-weight group showed significantly more benefits than the other two groups when the vibrotactile stimuli were deliv-ered to their hands, but not when the stimuli were delivered to their abdomens. these results might raise some interest-ing questions regarding the mechanisms underlying audio-tactile information processing and have applied relevance for the design of the most effective warning signal (e.g., for drivers).

Keywords Multisensory · Overweight · Vibrotactile · Detection · Discrimination

X. Wan · B. Mu · X. Zhou Department of Psychology, school of social sciences, tsinghua University, Beijing 100084, china

X. Wan · c. spence · c. ho (*) crossmodal Research laboratory, Department of Experimental Psychology, University of Oxford, Oxford OX1 3UD, UKe-mail: cristy.ho@psy.oxon.org

1086 Exp Brain Res (2014) 232:1085–1093

1 3

part on top of a driver’s abdomen (ho et al. 2006a, 2007; lee et al. 2004), by means of foot pedals (Birrell et al. 2013), or via the steering wheel itself (sato et al. 1998; suzuki and Jansson 2003). It should, however, be noted that the location on the body surface from which the tactile stimuli are presented can exert a dramatic influence on a person’s responses to them. For instance, people are slower to detect tactile stimuli delivered to their foot than those delivered to their upper arm (Bergenheim et al. 1996).

audiotactile interactions are different when stimuli are presented from behind a person’s head (e.g., as compared to when they are presented from in front of their head). that is, people were more accurate to judge the temporal order of auditory and tactile stimuli when they were presented at different locations behind their heads than when they were presented at the same position from behind their heads (Kitagawa et al. 2005), whereas no such effect was shown when these stimuli were presented in front of their heads (Zampini et al. 2005). however, very few studies have been conducted in order to determine which is the better solution in terms of presenting tactile and/or audiotactile stimuli to drivers and which part(s) of the human body are more effi-cient to respond to tactile and/or audiotactile warning sig-nals (ho and spence 2008; spence and ho 2008b).

In particular, if vibrotactile and audiotactile stimuli are delivered to drivers via the part of the seat belt sitting on top of their abdomens, say, it remains unknown whether people of different body sizes and/or having levels of body fat (especially abdominal fat) would benefit similarly from audiotactile stimulation. Previous research has demon-strated that tactile responses to vibrations can be influenced by the age, sex, and hand preference of the participants (see Verrillo and Bolanowski 2009, for a review). critically, overweight/obesity has become a global health problem: It is estimated that approximately 5 million people are obese and that around 1.5 billion adults worldwide were over-weight in 2008 (Finucane et al. 2011). Overweight and obesity are defined by abnormal or excessive body fat and are indexed by body mass index (BMI). BMI is calculated by dividing one’s body mass (in kilograms) by the square of their height (in meters). currently, adults having a BMI between 18.5 and 25 are considered to have normal weight, adults whose BMI is lower than 18.5 are considered to be underweight, while those with a BMI in excess of 25 but lower than 30 are considered overweight (World health Organization 2013b). It has been estimated that around 60 % of the world’s population could become overweight or obese by the year 2030 (Kelly et al. 2008).

Importantly, overweight/obesity not only raises health-related concerns, but also raises issues around the potential impairment of cognitive abilities (see sellbom and Gunstad 2012, for a review). Overweight people with metabolic abnormalities out of a large sample of more than 6,000

adults showed more rapid cognitive decline than their nor-mal weight counterparts over a 10-year period (singh-Man-oux et al. 2012). higher BMI was associated with slower processing speed and poorer performance on a selective attention test (cournot et al. 2006), as well as poorer execu-tive functioning (Gunstad et al. 2007). some research has demonstrated that greater body weight also increases the risk of severe injury and death in vehicle crashes (e.g., Mock et al. 2002). It is therefore going to be important to determine whether overweight drivers would also ben-efit from vibrotactile and/or audiotactile warning signals as much as their normal weight counterparts.

the purpose of the present study was, then, relatively straightforward. Using detection and discrimination tasks in a laboratory setting, we wanted to examine whether body size and body fat measured by BMI would influence the processing of (and latency of responses to) auditory, vibro-tactile, and audiotactile targets. the vibrotactile stimuli were delivered to two different regions of the participant’s body, namely their abdomen and their non-dominant hand. these two bodily regions were chosen to simulate the situ-ation in which vibrotactile and/or audiotactile warning sig-nals were delivered to drivers via the seat belt and steering wheel, respectively.

We anticipated that the overweight individuals would benefit from audiotactile multisensory stimulation as com-pared to unimodal stimulation, considering the robustness of this phenomenon. however, we also thought it possible that the overweight participants would benefit less from audiotactile multisensory stimulation than normal or under-weight individuals. Due to the increased abdominal fat in those who are overweight, we thought it possible that they would be less sensitive to vibrotactile stimuli delivered to their abdomens than normal or underweight individuals and therefore they will show less benefit from audiotactile stimulation as compared to their responses to the auditory stimulation. On the other hand, it also seemed reasonable to expect that overweight individuals would show more benefit from audiotactile stimulation as compared to vibro-tactile stimulation, which was due in part to any impair-ment they showed in response to the unisensory vibrotactile stimulation.

Methods

Participants

thirty-six male participants (mean age of 22.5 years, rang-ing from 19 to 33 years) from mainland china took part in this study in return for cNY 50. they were assigned to three groups based on their BMI, with 12 participants in each group (see table 1). all of the participants were

1087Exp Brain Res (2014) 232:1085–1093

1 3

right-handed by self-report. they all had normal hearing and a normal sense of touch by self-report.

apparatus and materials

a Pc-based computer with a 17-inch monitor was used to run the experiments. Matlab 2009a with psychophys-ics toolbox (PtB) 3 was used to present the stimuli and to record the participants’ responses. Responses were col-lected via a keyboard. a tactor (Bc461-1 100, Oticon ltd., somerset, New Jersey) attached to a Velcro belt was used to deliver the vibrotactile stimuli. During the abdominal blocks, this belt was fastened around the participant’s waist on top of his clothing with the tactor placed in the mid-dle of the participant’s stomach approximately 2 cm above the navel. During the hand blocks, this belt was fastened around the participant’s left hand with the tactor placed in the center of the participant’s palm. the tactor was driven by a 250-hz signal presented via the output of a soundcard on the computer through a power amplifier. headphones were used to present the auditory stimuli (56.0 dB measured near the earpads). the audiotactile stimuli consisted of the simultaneous presentation of both the vibrotactile and the auditory stimuli. In order to mask any noise produced by the operation of the tactor, white noise was presented (57.4 dB measured near the earpads) throughout the experimental session, including white noise played via the headphones and white noise resulting from the operation of an air puri-fier in the laboratory. the air purifier was turned on during the experiment to mask any noise produced by the operation of the tactor and to avoid the necessity of having to present loud noise directly to participants’ ears via the headphones. In addition, the participant wore a gown (provided by the experimenters) to mask the noise produced by the operation of the tactor during the blocks of trials that involved abdom-inal stimulation. During the blocks that involved hand stim-ulation, the experimenters placed another gown to cover the participant’s left hand, and the participant was instructed to keep his left hand still under the cover of this gown to mask any noise produced by the operation of the tactor.

Design and procedure

Before scheduling the participants, the experimenters con-tacted the participants by email or phone to make sure that they were in a generally good state of health and to ask about their weight and height in order to estimate their BMI. after the participants had arrived in the laboratory and signed the consent forms, the experimenters measured their weight, height, waistline, the length of their upper arms (from shoulder to elbow), the length of their lower arms (from elbow to wrist), the length of their hands (from wrist to the top of their middle finger), hand width, and hand circumference. the experimenters calculated each participant’s BMI and assigned them to one of the three groups based on their score.

Body Weight (normal, underweight, or overweight), tar-get type (auditory, vibrotactile, or audiotactile), and tar-get Position (abdomen vs. hand) were all manipulated as independent variables for both the detection and discrimi-nation tasks, with Body Weight as a between-groups fac-tor and the other two being within-group factors. Both the auditory and vibrotactile targets were unisensory targets, whereas the audiotactile target was a multisensory target, which was a combination of the auditory and vibrotactile targets. an aBBa design was used to counterbalance the order of tasks for each participant, with two blocks of the same task being grouped together. that is, each participant first completed 2 blocks of the detection task with targets in one position, followed by two blocks of the discrimination task with targets in the same position. Next the participant completed 2 blocks of the discrimination task with targets in the other position, followed by 2 blocks of the detection task with targets in the second position. half of each group of participants completed the abdominal blocks first, and the other half completed the hand blocks first. Within each block of trials, equal numbers of three different types of targets (auditory, vibrotactile, and audiotactile) were mixed and presented in a random order.

In the detection task, the participants were instructed to use their right index finger to press the M key on the

Table 1 Information regarding the three groups of participants tested in the present study (with standard deviations in parentheses)

Underweight group Normal weight group Overweight group

Number of participants 12 12 12

BMI range 15.8–18.2 19.9–25.0 25.6–29.8

Mean BMI 17.3 (0.60) 22.1 (1.78) 27.3 (1.48)

Mean weight (kg) 53.26 (3.20) 65.95 (6.39) 84.24 (10.68)

Mean height (m) 1.75 (0.04) 1.73 (0.07) 1.75 (0.08)

Mean waistline (cm) 70.00 (3.19) 80.33 (5.25) 92.75 (5.93)

Mean arm length (cm) 57.42 (3.00) 57.83 (2.44) 58.75 (3.79)

Mean hand length (cm) 18.42 (0.90) 18.33 (0.91) 18.96 (1.51)

Mean hand circumference (cm) 19.50 (0.80) 20.42 (0.90) 21.63 (1.40)

1088 Exp Brain Res (2014) 232:1085–1093

1 3

keyboard whenever they heard a beeping sound from the headphones and/or felt a vibration from the belt. Each sound or vibration was presented for 200 ms. a detection trial was terminated if the participants failed to make any response within 3,000 ms of target onset, and the next trial was then started. as for the discrimination task, the partici-pants were instructed to use their right index finger to press the B key on the keyboard whenever they heard a short beep and/or feel a short vibration (both 150 ms in dura-tion) and to use their right middle finger to press the N key whenever they heard a long beep and/or feel a long vibra-tion (both 250 ms in duration). a discrimination trial was terminated if the participants failed to make any response within 5,000 ms, and the next trial was then started. Each participant completed a total of 8 blocks with 90 trials in each block. the participants were given 20 practice trials before each detection task and 24 practice trials before each discrimination task. No feedback was provided to the par-ticipants during the course of the experiment.

Results

the detection task

the three groups of participants made very few errors in the detection task (99.5 % correct in the normal weight group, 99.6 % in the underweight group, and 99.3 % in the overweight group). Rts lower than 100 ms or in excess of 1,500 ms were excluded (e.g., chang and Ro 2005, for sim-ilar exclusion criteria), resulting in 1 % of the data being discarded from the data analyses. Mean Rts in each condi-tion for each group were calculated based on those trials in which the participants made correct responses and are shown in Fig. 1.

We performed a 3 (Body Weight: normal, underweight, or overweight) × 3 (target type: auditory, vibrotactile, or audiotactile) × 2 (target Position: abdomen vs. hand) aNOVa on the Rt data, with Body Weight as a between-groups factor while target type and target Position being within-group factors. this analysis revealed a significant main effect of target type, F(2, 66) = 90.15, p < .001. Planned pairwise comparison with Bonferroni correction for multiple testing revealed that responses to the audio-tactile targets were significantly faster (296 ms) than to the auditory (339 ms) or vibrotactile targets (336 ms), both t(35) > 11.27, p < .001; whereas responses to the auditory and vibrotactile targets were comparable, t(35) < 1, n.s. the main effect of target Position was also significant, F(1, 33) = 19.82, p < .001, with participants responding more rapidly when the vibrotactile stimuli were delivered to their hand (309 ms) than to their abdomen (338 ms). this analysis also revealed a significant interaction between

target type and Body Weight, F(4, 66) = 3.92, p < .01, and a significant interaction between target type and target Position, F(2, 66) = 13.65, p < .001. None of other effects was significant, all F < 1.58, p > .22. In order to interpret the two significant interaction terms, we performed the fol-lowing analyses.

In order to examine the extent to which the three dif-ferent groups of participants benefitted from multisensory audiotactile stimulation as compared to unimodal vibrotac-tile stimulation, we performed a 3 (Body Weight: normal, underweight, or overweight) × 2 (target type: vibrotac-tile vs. audiotactile) × 2 (target Position: abdomen vs. hand) aNOVa on the Rt data. this analysis revealed a significant main effect of target type, F(1, 33) = 154.97, p < .001, but this effect was qualified by a significant inter-action with Body Weight, F(2, 33) = 3.60, p = .038. that is, all three groups of participants responded more rapidly to the multisensory targets than to the unisensory targets, all t(11) > 5.28, p < .01, but the overweight group showed less of a benefit of multisensory stimulation (28 ms) than the underweight group (51 ms), t(22) = 2.60, p < .05. the normal weight group showed a benefit of 39 ms, which appeared to be greater than that of the overweight group but smaller than that of the underweight group, but nei-ther of these trends reached statistical significance, both ts < 1.57, p > .13. the main effect of target Position was also significant, F(1, 33) = 20.84, p < .001, suggesting that responses were faster when the vibrotactile stimuli were delivered to the participant’s hand (301 ms) than to his abdomen (332 ms); but this effect was qualified by an interaction between target type and target Position, F(1, 33) = 24.73, p < .001. that is, participants showed a benefit from multisensory stimulation regardless of whether the vibrotactile targets were delivered to their abdomen or hand, both t(11) > 7.41, p < .01, but they showed a greater benefit when the vibrotactile targets were delivered to their abdomen (52 ms) than to their hand (27 ms), t(11) = 5.10, p < .001. None of other effects was significant, all Fs < 1.

similarly, in order to examine how the three differ-ent groups of participants benefited from multisensory audiotactile stimulation as compared to unimodal audi-tory stimulation, we performed a 3 (Body Weight: normal, underweight, or overweight) × 2 (target type: auditory vs. audiotactile) × 2 (target Position: abdomen vs. hand) aNOVa on the Rt data. this analysis revealed a sig-nificant main effect of target type, F(1, 33) = 153.81, p < .001, and a significant main effect of target Position, F(1, 33) = 13.11, p < .01. these results therefore suggest that participants’ responses were faster to the audiotactile stimuli (296 ms) than to the auditory stimuli (339 ms), and they were faster when the vibrotactile stimuli were deliv-ered to their hand (307 ms) than to their abdomen (328 ms). None of other effects was significant, all Fs < 2.38, p > .10.

1089Exp Brain Res (2014) 232:1085–1093

1 3

the discrimination task

as for the discrimination task, the normal, underweight, and overweight groups performed with a high level of accuracy (94.6 %, 92.7 %, and 91.8 %, respectively). Rts shorter than 250 ms or longer than 2,500 ms were excluded from data analyses (e.g., hazeltine et al. 2007, for simi-lar data exclusion criteria). this resulted in the removal of <1 % of the data. Mean Rts in each condition for each group were calculated based on trials in which the partici-pants made correct responses and are shown in Fig. 1.

We first performed a 3 (Body Weight: normal, under-weight, or overweight) × 3 (target type: auditory, vibro-tactile, and audiotactile) × 2 (target Position: abdomen vs. hand) aNOVa on the Rts and accuracy data. this analy-sis revealed a significant main effect of target type on the

Rt data, F(2, 66) = 22.36, p < .001, and on the accuracy data, F(2, 66) = 15.21, p < .001. Planned pairwise com-parison with Bonferroni correction for multiple testing sug-gested that responses to the audiotactile targets (722 ms, 95.2 %) were faster than to the auditory targets (740 ms, 94.2 %), t(35) = 3.47, p < .01, but with a comparable level of accuracy, t(35) < 1, n.s. Furthermore, responses to the audiotactile targets were also significantly faster and more accurate than to the vibrotactile targets (757 ms, 89.6 %), both t(35) > 5.91, p < .001; while responses to the audi-tory targets were significantly faster and more accurate than to the vibrotactile targets, both t(35) > 3.04, p < .05. the main effect of target Position on participants’ Rts was also significant, F(1, 33) = 11.50, p < .01, as was the effect on accuracy, F(1, 33) = 8.20, p < .01. this suggests that responses were faster and more accurate when the

Fig. 1 the upper panel shows mean Rts for the auditory, vibrotactile, and audiotactile targets in the detection task, while the middle and lower panels show mean Rts and accuracy data for the auditory, vibrotactile, and audiotactile tar-gets in the discrimination task, respectively. the three groups of columns on the left show the results when the vibrotactile stimuli were delivered to the participant’s abdomen, whereas those on the right highlight the results when the vibrotactile stimuli were delivered to the participant’s hand instead. Error bars show the standard errors of the means

200

250

300

350

400

450

Det

ecti

on R

Ts (

ms)

Audio Vibrotactile Audiotactile

500

600

700

800

900 D

iscr

imin

atio

n R

Ts (

ms)

70%

75%

80%

85%

90%

95%

100%

Underweight Normal Overweight Underweight Normal Overweight

Dis

crim

inat

ion

Acc

urac

y

Abdomen Hand

1090 Exp Brain Res (2014) 232:1085–1093

1 3

vibrotactile stimuli were delivered to the participant’s hand (728 ms, 94.0 %) than to his abdomen (752 ms, 92.0 %). the analysis also revealed a significant interaction between target Position and Body Weight on the Rt data, F(2, 33) = 4.26, p = .023, but it just failed to reach statistical significance in the accuracy data, F(2, 33) = 2.73, p = .08. there was also a significant interaction between target type and target Position in the Rt data, F(2, 66) = 5.95, p < .01, and in the accuracy data, F(2, 66) = 8.60, p < .001. to interpret the interactions between target Position and Body Weight and between target type and target Position, we performed the following analyses.

to investigate how the three different groups of partici-pants benefited from multisensory audiotactile as compared to unisensory vibrotactile stimulation in the discrimina-tion task, we performed 3 (Body Weight: normal, under-weight, or overweight) × 2 (target type: vibrotactile vs. audiotactile) × 2 (target Position: abdominal vs. hand) aNOVas on the Rt and accuracy data. these analyses revealed a significant main effect of target type on the Rt data, F(1, 33) = 59.37, p < .001, and on the accuracy data, F(1, 33) = 34.06, p < .001; but these effects were quali-fied by a significant interaction between target type and target Position on the Rt data, F(1, 33) = 5.90, p = .021, and on the accuracy data, F(1, 33) = 10.68, p = .011. that is, participants showed a benefit from multisensory stimulation on both their Rt and accuracy when the vibro-tactile cues were delivered to their abdomen or hand, all t(11) > 2.84, p < .01, with a greater benefit being observed for both measures when the vibrotactile cues were deliv-ered to their abdomen (49 ms, 8.5 %) than to their hand (22 ms, 2.7 %), both t(11) > 2.46, p < .05. the main effect of target Position on the Rts was also significant, F(1, 33) = 12.78, p < .001, as was the effect on accuracy, F(1, 33) = 8.31, p < .01; but these effects were also qualified by a significant interaction between target type and Body Weight on the Rt data, F(2, 33) = 4.56, p = .018. that is, the overweight group were around 60 ms faster when the vibrotactile stimuli were delivered to their hand than to their abdomen, t(11) = 4.08, p < .01; this effect just failed to reach statistical significance for the normal weight group (27 ms), t(11) = 2.15, p = .055; whereas the underweight group did not show this effect, t(11) < 1, n.s.

We also performed similar analyses to examine how the three different groups of participants benefited from mul-tisensory stimulation as compared to auditory stimulation in the discrimination task. these analyses revealed a sig-nificant main effect of target type on the Rt data, F(1, 33) = 11.40, p < .001, suggesting that responses to the audiotactile targets were faster (722 ms) than to the auditory targets (740 ms). the main effect of target Position was also significant on the Rt data, F(1, 33) = 5.59, p = .024, suggesting that participants responded more rapidly

when the vibrotactile stimuli were delivered to their hand (725 ms) than to their abdomen (738 ms); but this effect was qualified by a target type by Body Weight interaction in the Rt data, F(2, 33) = 3.76, p = .034. that is, both the overweight and normal weight groups responded more rapidly when the vibrotactile stimuli were delivered to their hand than to their abdomen, both t(11) > 2.32, p < .05, and this effect was comparable for these two groups, t(22) < 1, n.s.; whereas the underweight group did not show any such effect, t(11) < 1, n.s. In addition, the three-way interaction between target type, target Position, and Body Weight just failed to reach statistical significance in terms of the Rt data, F(2, 33) = 2.75, p = .076, ηp

2 = .143, but was significant in the accuracy data, F(2, 33) = 5.08, p = .012, ηp

2 = 235. that is, the overweight group showed a greater benefit from audiotactile stimulation when the vibrotactile stimuli were delivered to their hand than to their abdomen, on both the Rts (32 ms vs. 10 ms) and the accuracy (2.6 % vs. 0.6 %), both t(11) > 2.87, p < .05; whereas the normal and underweight groups did not show these effects, all t(11) < 1.67, p > .12.

Body weight and other body parameters

We also analyzed whether the three groups differed in any of the body parameters measured in this study. the results showed that they differed in waistline and hand circum-ference. compared to the normal group, the underweight group had a significantly smaller waistline, t(22) = 5.83, p < .001, and significantly smaller hand circumference, t(22) = 2.64, p = .015; whereas the overweight group had a significantly larger waistline, t(22) = 5.44, p < .001, and significantly larger hand circumference, t(22) = 2.52, p = .020. however, the three groups had comparable arm length, hand length, and height, all ts < 1.23, p > .23, sug-gesting that nerve conduction latencies for vibrotactile stimuli may be comparable across the three groups despite their varying weights.

Finally, we calculated the correlation between partici-pants’ body weight and their Rts or Rt benefits from mul-tisensory stimulation. as for the detection task, there was a significant positive correlation between their body weight and Rts for the auditory targets, regardless of whether the vibrotactile targets were delivered to their abdomen, r = 0.353, p = .035, or to their hand, r = 0.376, p = .024. there was also a significant positive correlation between their body weight and their Rts to the audiotactile targets, regardless of whether the vibrotactile stimuli were delivered to their abdomen, r = 0.396, p = .017, or hand, r = 0.374, p = .025; but there was no such effect for the vibrotactile targets, both p > .28. however, when the vibrotactile stim-uli were delivered to their abdomen, there was a signifi-cant negative correlation between body weight and the Rt

1091Exp Brain Res (2014) 232:1085–1093

1 3

benefit observed from multisensory as compared to unisen-sory vibrotactile stimulation, r = −0.035, p = .036, but no such effect was observed when comparing the Rts in the audiotactile condition to those in the auditory condition, or either type of Rt benefits when the vibrotactile stimuli were delivered to their hands, all p > .09. In contrast, there were no significant correlations between body weight and Rts or Rt benefits in the discrimination task, all p > .29.

Discussion

the influence of target position on audiotactile stimulation

For the detection and discrimination tasks, participants’ responses were overall faster in the detection task, and faster and more accurate in the discrimination task, when vibrotactile stimuli of the same objective intensity were delivered to the participant’s hand than to his abdomen. these results are consistent with the literature showing that the location of tactile stimulation affects response laten-cies (see also Bergenheim et al. 1996; Kitagawa et al. 2005; Zampini et al. 2005). that is, when the intensity of the vibrotactile signals was kept constant, the hands may repre-sent a preferable location from which to present unimodal vibrotactile warning signals as compared to the abdomen. In other words, in order to deliver comparable benefits, the intensity of the vibrotactile signals delivered to an individu-al’s abdomen may need to be stronger than those delivered to their hands.

One possible explanation for this could be that the par-ticipants simply paid more attention to their hands, as they used them to press keys in order to perform the detection and discrimination tasks. the extant literature demon-strates that vibrotactile stimuli do not always automatically capture people’s attention (santangelo and spence 2007; spence 2010). therefore, when the vibrotactile stimuli were delivered to their abdomen, they had to deploy their attention to two different regions of their body, that is, their abdomen and hand, leading to a cost in the efficiency of their responding (see also lakatos and shepard 1997). to test this explanation, foot responses may be used in future research to compare whether the responses to vibrotactile stimuli delivered to the hands and abdomens are compara-ble or different from each other, given that in both cases, the participants will be required to deploy attention to two different regions of their bodies.

a second possible explanation could be the difference in nerve conduction velocities, which may be affected by temperature, nerve type, or the location on the nerve (e.g., campbell et al. 1981; Gallace and spence 2014), as the hands and abdomen are at somewhat different distances

from the brain and may be transduced by somewhat dif-ferent nerve fibers. a third possibility could be the differ-ence in the skin types receiving the vibrotactile stimuli. In this study, vibrotactile stimuli delivered to the hands were directly delivered to the palms, that is, to a hairless gla-brous skin site (for reviews, see Gallace and spence 2014; Verrillo and Bolanowski 2009), whereas vibrotactile stim-uli delivered to the abdomen were registered by a hairy skin site covered by clothing. the difference in the skin types as well as whether the skin directly receives the vibrotactile stimuli might have caused the difference in performance.

On the other hand, compared to their responses to vibro-tactile stimulation, our participants showed a greater ben-efit from audiotactile stimulation when the vibrotactile targets were delivered to their abdomens than when they were delivered to their hands. Yet, as can be seen in Fig. 1, responses to the audiotactile stimuli for which vibrotactile stimuli were delivered to the participant’s abdomen were facilitated as compared to responses to the unimodal vibro-tactile stimuli delivered to his abdomen, but were compa-rable to the responses to the unimodal vibrotactile stimuli delivered to his hand. these results can be taken to suggest that if vibrotactile stimuli are delivered to an individual’s abdomen, then audiotactile stimulation can significantly improve his/her responses.

the influence of body mass on audiotactile stimulation

the results of the present study demonstrate that all of the participants, and that includes the overweight, nor-mal, and underweight participants, benefitted from multi-sensory stimulation, but that the magnitude of this benefit depended on the nature of the task, the location the vibro-tactile stimuli were delivered to, and the nature of the base-line measure (relative to their responses to the vibrotactile or auditory stimuli). Interestingly, the effects of audiotactile stimulation on the responses of our overweight individu-als revealed a somewhat different pattern of results for the detection and discrimination tasks. For the detection task, relative to the responses to unimodal vibrotactile stimuli, the three groups of participants all demonstrated a benefit from multisensory stimulation, while the overweight group showed less of a benefit than the underweight group. this effect was also confirmed by the significant negative cor-relation between all participants’ body weight and Rt benefits from audiotactile stimulation compared to vibro-tactile stimulation. as can be seen in Fig. 1, this effect is likely due to the fact that the overweight group exhibited much slower responses in the unimodal vibrotactile con-dition in which the vibrotactile target stimuli happened to be delivered to their abdomen. this might be attributed to the increased abdominal fat accumulation making the over-weight group less sensitive to the vibrotactile stimuli.

1092 Exp Brain Res (2014) 232:1085–1093

1 3

By contrast, for the discrimination task, compared to the responses to unimodal auditory stimuli, the overweight group exhibited a greater benefit from audiotactile stimu-lation when the vibrotactile stimuli were delivered to their hands than to their abdomens, whereas the normal and underweight groups did not show these effects. Note that the overweight participants had significantly larger waist-lines and hand circumferences than the other two groups of participants, but the three groups all received vibrotactile signals of the same intensity. It will be an interesting ques-tion for future research to test whether overweight, normal, and underweight individuals need vibrotactile warnings of different intensities in order to achieve comparable baseline performance and whether they will have the same benefit if auditory signals are added in this case.

In summary, the results of the present study have impor-tant implications for the design of in-car warning signals. critically, overweight drivers would benefit from multisen-sory stimulation, and they are also more sensitive to such stimuli when the stimuli are delivered to their hands than those delivered to their abdomen. In conclusion, then, it may not be advisable to use unimodal vibrotactile warnings for drivers who happen to be overweight. Given the large overweight population worldwide (Finucane et al. 2011), tactile or multisensory warning system designers should be fully aware of the possible effect of a driver’s body mass on the functioning of their warning system. One should be cautious when designing a standard warning system for drivers with different body sizes.

Acknowledgments this research was supported in part by tsing-hua University cultural Inheritance and Innovation Program by the asia Research center of tsinghua University, and by the tsinghua-santander Young Faculty program.

References

Bergenheim M, Johansson h, Granlund B, Pedersen J (1996) Experimental evidence for a sensory synchronization of sen-sory information to conscious experience. In: hameroff sR, Kaszniak aW, scott ac (eds) toward a science of conscious-ness: the first tucson discussions and debates. MIt Press, cam-bridge, pp 303–310

Birrell sa, Young Ms, Weldon aM (2013) Vibrotactile pedals: provi-sion of haptic feedback to support economical driving. Ergonom-ics 56(2):282–292

campbell WW Jr, Ward lc, swift tR (1981) Nerve conduction velocity varies inversely with height. Muscle Nerve 4:520–523

chang E, Ro t (2005) Inhibition of return in perception and action. Vis cogn 12:443–472

cournot M, Marquié Jc, ansiau D et al (2006) Relation between body mass index and cognitive function in healthy middle-aged men and women. Neurology 67:1208–1214

Evans l (1991) traffic safety and the driver. Van Nostrand Reinhold, New York

Finucane MM, stevens Ga, cowan MJ et al (2011) National, regional, and global trends in body-mass index since 1980:

systematic analysis of health examination surveys and epidemi-ological studies with 960 country-years and 9.1 million partici-pants. lancet 377:557–567

Fitch GM, Kiefer RJ, hankey JM, Kleiner BM (2007) toward devel-oping an approach for alerting drivers to the direction of a crash threat. hum Factors 49:710–720

Gallace a, spence c (2014) In touch with the future: the sense of touch from cognitive neuroscience to virtual reality. Oxford Uni-versity Press, Oxford

Gunstad J, Paul Rh, cohen Ra, tate DF, spitznagel MB, Gordon E (2007) Elevated body mass index is associated with execu-tive dysfunction in otherwise healthy adults. compr Psychiatry 48:57–61

hazeltine E, aparicio P, Weinstein a, Ivry RB (2007) configural response learning: the acquisition of a nonpredictive motor skill. J Exp Psychol hum 33:1451–1467

ho c, spence c (2005) assessing the effectiveness of various audi-tory cues in capturing a driver’s visual attention. J Exp Psychol-appl 11:157–174

ho c, spence c (2008) the multisensory driver: implications for ergonomic car interface design. ashgate Publishing, aldershot

ho c, tan hZ, spence c (2005) Using spatial vibrotactile cues to direct visual attention in driving scenes. transp Res F traffic 8:397–412

ho c, Reed NJ, spence c (2006a) assessing the effectiveness of ‘‘intu-itive’’ vibrotactile warning signals in preventing front-to-rear-end collisions in a driving simulator. accid anal Prev 38:989–996

ho c, tan hZ, spence c (2006b) the differential effect of vibro-tactile and auditory cues on visual spatial attention. Ergonomics 49:724–738

ho c, Reed NJ, spence c (2007) Multisensory in-car warning signals for collision avoidance. hum Factors 49:1107–1114

ho c, santangelo V, spence c (2009) Multisensory warning sig-nals: when spatial correspondence matters. Exp Brain Res 195:261–272

horowitz aD, Dingus ta (1992) Warning signal design: a key human factors issue in an in-vehicle front-to-rear-end collision warning system. In: Proceedings of the human Factors society, pp 1011–1013

Kelly t, Yang W, chen cs, Reynolds K, he J (2008) Global burden of obesity in 2005 and projections to 2030. Int J Obes 32:1431–1437

Kitagawa N, Zampini M, spence c (2005) audiotactile interactions in near and far space. Exp Brain Res 166:528–537

lakatos s, shepard RN (1997) time-distance relations in shifting attention between locations on one’s body. Percept Psychophys 59:557–566

lee JD, hoffman JD, hayes E (2004) collision warning design to mitigate driver distraction. In: Proceedings of chI, vol 6. acM, New York, pp 65–72

lee JD, McGehee DV, Brown tl, Marshall D (2006) Effects of adap-tive cruise control and alert modality on driver performance. transp Res Rec 1980:49–56

Mock cN, Grossman DG, Kaufman RP, Mack cD, Rivara FP (2002) the relationship between body weight and risk of death and seri-ous injury in motor vehicle crashes. accid anal Prev 34:221–228

santangelo V, spence c (2007) Multisensory cues capture spatial attention regardless of perceptual load. J Exp Psychol human 33:1311–1321

sato K, Goto t, Kubota Y, amano Y, Fukui K (1998) a study on a lane departure warning system using a steering torque as a warn-ing signal. In: Proceedings of the international symposium on advanced vehicle control (aVEc’98), pp 479–484

sellbom Ks, Gunstad J (2012) cognitive function and decline in obe-sity. J alzheimers Dis 30:89–95

singh-Manoux a, czernichow s, Elbaz a et al (2012) Obesity pheno-types in midlife and cognition in early old age: the Whitehall II cohort study. Neurology 79:755–762

1093Exp Brain Res (2014) 232:1085–1093

1 3

spence c (2010) crossmodal spatial attention. In: Kingstone a, Miller MB (eds) annals of the New York academy of sciences (the year in cognitive neuroscience), vol 1191. Wiley-Blackwell, New York, pp 182–200

spence c, ho c (2008a) Multisensory warning signals for event per-ception and safe driving. theor Issues Ergon sci 9:523–554

spence c, ho c (2008b) tactile and multisensory spatial warning sig-nals for drivers. IEEE trans haptics 1:121–129

suzuki K, Jansson J (2003) an analysis of driver’s steering behav-iour during auditory or haptic warnings for the designing of lane departure warning system. JsaE Rev 24:65–70

Verrillo Rt, Bolanowski sJ (2009) tactile responses to vibration. In: havelock D, Kuwano s, Vorlander M (eds) handbook of signal processing in acoustics. springer, New York, pp 1185–1213

Wang Js, Knipling RR, Goodman MJ (1996) the role of driver inat-tention in crashes: new statistics from the 1995 crashworthiness data system. In: 40th annual proceedings of the association for

the advancement of automotive medicine, Vancouver, 7–9 Oct 1996

World health Organization (2004) World report on road traffic injury prevention. World health Organization Web. http://www.who.int/violence_injury_prevention/publications/road_traffic/world_report/en. accessed 2 Oct 2013

World health Organization (2013a) Global status report on road safety. World health Organization Web. http://www.who.int/vio-lence_injury_prevention/road_safety_status/2013/en. accessed 4 Oct 2013

World health Organization (2013b) Obesity and overweight. World health Organization Web. http://www.who.int/mediacentre/factsheets/fs311/en. accessed 5 sept 2013

Zampini M, Brown t, shore DI et al (2005) audiotactile temporal order judgments. acta Psychol 118:277–291