INTELLIGENT MUSIC SELECTION TO INFLUENCE

DRIVER BEHAVIOUR

An Empirical Study

Perry MacNeille, Kacie Theisen

and

Oleg Gusikhin

Ford Motor Company Research and Advanced Engineering, 2101 Village Road, Dearborn, Michigan, U.S.A.

Keywords: Music perception, Music cognition, Utility music, Functional music, Driver behaviour.

Abstract: There is a belief that driving competency and style is influenced by music choices, yet there is little

scientific study into the effects music choices have on the way people drive. This paper describes a

preliminary explorative study conducted to find evidence of music influencing driving behaviour to justify

further research into the area. Three main effects were considered in this study: 1) that music either

enhances or impedes the driving activity; 2) that driving behaviour is influenced by whether the subject

likes the music being played; 3) that driving behaviour is changed by the tempo of the music being played.

The speed holding ability of 39 experienced drivers employed in a large company was tested using a vehicle

simulator to observe evidence that speed holding control is influenced by one or more of the main effects.

1 INTRODUCTION

Audio entertainment is a prevalent activity in

vehicles that accompanies driving and is almost as

old as the automotive industry itself, dating back to

1930. Nowadays, it is hard to imagine vehicles

without factory installed entertainment options, such

as radios and CD players. Furthermore, automakers

are constantly seeking ways to improve vehicle

entertainment offerings to keep up with

advancements in consumer electronics and consider

this area an important aspect of product

differentiation and competitive advantage.

There are many audio entertainment options

including news, talk shows, books on tape, foreign

language tutorials, and of course music. From this

set of options, music is probably the most common

audio activity in vehicles, especially among younger

drivers. Therefore, the focus of this paper and

associated project will be on music even though the

concepts developed could apply and be extended to

other audio options, like the news.

It is believed that music may significantly

influence driving performance and safety. In fact,

music as well as other forms of entertainment may

serve as a remedy against boredom and driver

fatigue during long and/or repetitive drives. On the

other hand, music is rarely cited as a source of

distraction (Bayly 2008).

Experimental psychology offers a compelling

argument that there is a strong relationship between

music, the listening situation, and arousal (a state of

heightened physiological activity), and the

performance of the accompanying task. In this case,

the listening situation refers to the driving

environment. It is also known that drivers often

instinctively regulate music based on the driving

situation; playing high tempo loud music on an

empty highway to fight boredom, and turning the

volume down while performing a difficult

manoeuvre in congested urban traffic (Dibben,

2007). If the car music system can sense the

environment and automatically select the most

appropriate music for the given condition it might

improve driving performance and safety while

enhancing the music listening experience.

In recent years there have been significant

advancements in three relevant areas that make the

development of such a system possible.

Digital Music - The rise of digital music enabled

instantaneous access to a vast variety of the musical

choices. First of all, MP3 players, such as the iPod,

allow storage and direct access to thousands of user

owned songs and provide a variety of ways to

arrange, on the fly, the music being played. Music

played can be selected by artist, genre, playlist,

random shuffle, or user-defined tags, such as mood,

without the need of changing CDs or tapes. More

MacNeille P., Theisen K. and Gusikhin O..

INTELLIGENT MUSIC SELECTION TO INFLUENCE DRIVER BEHAVIOUR - An Empirical Study.

DOI: 10.5220/0003687800830090

In Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2011), pages 83-90

ISBN: 978-989-8425-75-1

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

recently, access to music has been further expanded

by the introduction of mobile internet music players,

such as the Slacker mobile player and iPhone

applications of Pandora and LastFM. These players

can tap into virtually unlimited internet musical

collections using 3G networks or Wi-Fi, allowing

for high control over music selection. As a result

music could be used to achieve very specific ends in

very specific circumstances.

Vehicle Sensory, Navigation and Traffic

Information - The rapid growth of the number and

quality of vehicle sensors allows reliable

identification of the driving situation, and bio-

physical state and emotions of the driver.

Furthermore, with the proliferation of navigation and

traffic information services we can identify not only

the current driving conditions but also the upcoming

driving environment. This would allow for upfront

planning of the next music choice.

Advances in the Music Knowledge Methods and

Tools - MP3 players allow for music selection

guidance using MP3 ID3 tags. Selection based on

these tags is either limited to editorial data, such as

artist and genre, or requires the daunting task of

organizing the entire music collection into special

purpose playlists or tagging each individual song.

Recognizing the need for a more elaborative music

selection and fuelled by the digital music technology

and web 2.0 phenomenons there has been significant

growth in internet-based music services in the last

decade. These services are a class of recommender

systems that allow organization and selection of

music by different qualitative criteria far beyond

simply genre and artist. These services range from

playlist generation and automatic tagging of

privately owned digital music to personalized

internet radio stations. The core of these services is a

knowledge base comprised of an extensive musical

catalogue. Different services exploit information in

the knowledge base in different ways to customize

music selections for individual listeners. Methods

used range from collaborative filtering or tagging by

the listener community to elaborate annotations by

professional musicologists regarding the qualitative

features of a song from its digital properties. As a

result, the availability of such music knowledge

enables music classification based on the anticipated

cognitive response and appropriateness for different

driving environments.

These advances create promise for a system that can

automatically select the next music piece that is

most appropriate for the given driving environment

and influence desirable driver behaviour. In order to

accomplish this task one of the main questions is,

how does the music influence driving behaviour? In

this paper we report the results of the preliminary

study to explore such an influence based on the

different types of music.

This paper is organized as follows. The next

section presents a review of the current state-of-the-

art in music knowledge methods and tools, and

reviews the products that explore this knowledge to

provide personalized music recommendations or

even a music station. Section 3 describes the

experiment and section 4 reviews the results.

Finally, the paper will conclude with summary and

future research recommendations.

2 MUSIC RECOMMENDATION

TECHNOLOGIES

Up until now people have had two main options for

car entertainment: broadcast based (radio) and

private selection based (MP3, CD, etc.). With the

broadcast based option listeners get to listen to new

and unexpected songs. To contrast this, private

owned collections provide the ability to listen to

music that the listeners own and thus presumably

like. In addition, they can listen to certain songs

whenever they like, however this option lacks the

excitement of new and unexpected music. iPod

Shuffle was an attempt to somewhat remedy this

issue by randomly picking songs from a private

collection. This technique of random music selection

can be considered one of the first music

recommendation techniques.

In the last few years there has been tremendous

internet-based growth of personalized musical

services dubbed Music 2.0 similar to Web 2.0. These

services attempt to capture personal musical

preferences of the given listener through interactive

voting and try to offer musical choices that

correspond to their musical tastes. Some of these

services organize the personal music library in

playlists according to given criteria, some select

music from the personal collection on the fly, and

some work as internet radio. In the case of personal

internet radio the stations provide an interactive

capability to tailor the station to the user's personal

preferences by skipping a song, banning certain

songs, or expressing that they really like a song.

Personalized Internet music services, such as

Last.fm, Musicovery, or Pandora, take the listening

experience to a whole new level by allowing users to

create their own personalized radio stations and

ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics

music experience instead of simply listening to what

the station chooses to webcast.

Last.fm (http://www.last.fm/) uses collaborative

filtering to recommend music you might like.

Collaborative filtering works by keeping track of

songs the user likes and listens to and is based on the

assumption that the musical taste of the specific

listener can be matched against the community.

Last.fm uses what they refer to as Scrobbling to

keep track of the music you listen to for use with

collaborative filtering techniques and to chart

popular music. One of the drawbacks of

collaborative filtering is that it only works well with

popular songs because songs lesser known might not

have substantial data. Another problem with this

approach is that it does not take into account the

structure of the song.

Musicovery (http://musicovery.com/) allows

users to intuitively find music based on their mood

and preferences for different genres and time

periods. This is accomplished through a graphical

interface based on Liveplasma technology and the

"mood pad". Forty musical parameters are used to

describe each song and place it on the mood pad

according to the likeliness that the song will please

the user according to their selected mood. Mood is

selected on a two dimensional scale with the y axis

ranging from calm to energetic and the x axis

ranging from dark to positive.

Pandora (http://www.pandora.com/) bases its

music recommendation on the Music Genome

Project. The Music Genome is a database of songs

where each song is described by a set of

characteristics. Each characteristic is referred to as a

gene and different songs have different numbers of

genes. For example, a rock song may have 150

genes while a classical song may have 500 genes.

These genes are collected into logical groups called

chromosomes and a set of chromosomes make up a

genome. Musicologists carefully classify each song

in the database by this method so that songs can be

compared on the basis of their genes taking into

account that some genes are more important than

others.

Listeners can tune into a currently playing station

or create their own personalized station. Stations are

started by entering the name of a song, artist, or

album which is referred to as the seed of the station.

This seed is then examined to determine which

distinguishing aspects of a song are present in its

genome and populate the station with songs with

similar features. While listening to music users can

provide feedback regarding whether or not they like

a selected song. Future music selections use this

feedback to refine the importance of each gene to an

individual listener.

Users can listen to Pandora online or on the go

with Pandora mobile. In the last two years Pandora

has been integrated into many telematics systems

and constitutes a popular vehicle infotainment

option.

In general Internet music stations provide

listeners with music they like based on feedback

from the listener. They usually allow users to create

multiple stations based on specific music features,

such as an artist or genre. Users can switch between

multiple stations or allow the system to switch for

them through a random shuffle mode, such as

Pandora mix mode. Since this shuffle is random,

these systems do not take into account the listeners

environment.

More recently products are being developed

which include music selection based on the user's

current environment and personal context.

Nike+ is one example, which is currently widely

available to consumers. This system has users place

a sensor in the bottom of their running shoe that

keeps track of each step taken. A user can then

choose between using the Nike+ Sportband or their

iPod. Either option provides runners with

information about their pace, distance, calories

burned, and more. After a run, users can log on to

the Nike+ webpage and upload information from

their run to track their progress, set goals, and

connect with other runners.

If paired with an iPod nano, users can create their

own personalized playlists to accompany their

workouts or download popular workout playlists

(referred to as Sport iMixes) from iTunes. US patent

20060107822 A1 filed by Apple claims the

capability of affecting a user's mood and behaviour

during an activity, such as exercise, by controlling

the speed of the music being listened to. The speed

of the music can either be selected to match the pace

of the activity or to drive the pace of the activity.

In a research context, several universities have

projects underway which aim to use Bio feedback to

select the music being played. The hypothesis

behind these applications is that a user's context

affects what they want to listen to. The University of

Maryland Baltimore County has developed a

concept called XPod which is intended to reside on a

PDA (Dornbush, 2007). Once XPod is initiated, a

BodyMedia SenseWear device is used to wirelessly

transmit data to a server to determine the user's level

of activity and emotion. Data collected by the

BodyMedia device includes transversal and

longitudinal acceleration, galvanic skin response,

INTELLIGENT MUSIC SELECTION TO INFLUENCE DRIVER BEHAVIOUR - An Empirical Study

skin temperature, heat flow, and nearbody

temperature. This information is used to determine if

the user is in an active, passive, or resting state and

is passed to a neural network engine which

compares the user's current activity level, state and

time to past song preferences to make the next music

selection. This selection is then sent back to the

client device. Lifetrack and AndroMedia are two

additional developments that utilize sensor data to

extract contextual information and recommend

suitable songs for the current situation.

Based on this discussion it is reasonable to

assume that context-aware music selection would be

not only possible, but also very beneficial within the

driving environment. Modern vehicles already know

a lot about drivers and driving conditions (current

and upcoming through digital maps) and this

information can be used to select a specific song (or

station). Following existing research in context-

based music selection the vehicle system can learn

the preferred music in the given driving conditions.

However, it is possible that not only driving

environment and conditions influence the music we

prefer, but also music influences the driver's

behaviour. Thus a more ambitious goal would be to

not only select the music preferred for the

conditions, but also select the music that can provide

a subtle influence on the driver's behaviour.

The next section describes our first attempt to

answer the question does music influence driver

behaviour and is there some kind of general pattern

how it influences behaviour.

3 THE EXPERIMENTAL

PROCEDURE

For this study there were 39 subjects, 16 female and

23 males, who were all Ford Motor Company

employees with at least 2 years driving experience.

The largest age population was 41-50 years old as

shown in Figure 1, which differs from many studies

based on college aged students. A list of songs was

selected from several genres and drivers were asked

to select the genre they liked the best and the genre

they liked the least. Each driver then drove a

simulator without music, with their favourite genre

of music, and with their least favourite genre during

which their speed and throttle position was

monitored. The data was later studied to determine if

there was any change in driving behaviour between

those three conditions.

Three hypotheses were tested in the experiment.

The first that music does influence driving, the

second that hearing music you prefer or dislike

affects driving and the third that music tempo affects

driving.

Under 20 20-30 31-40 41-50 51 +

Figure 1: Distribution of ages by categories of study

participants.

To test these hypotheses, each of the subjects

was brought into a room with a vehicle simulator,

and asked to complete a questionnaire about their

demographics and musical preferences.

Next the subject was given a driving task. In this

task the subject is asked to drive the simulator at an

appropriate speed without help from a speedometer.

In this way the subject is forced to judge speed on

the basis of optical flow rather than watching a

speedometer. The course is on a 2-lane one-way

road without stop signs or traffic lights with a striped

centreline and solid shoulder lines that curve and

change altitude. The landscapes surrounding the

road are generally rolling hills covered with grass

and occasional trees. There are very few other

vehicles on the road and the environment is visually

sparse. If the subject drives too fast he/she will lose

control of the car and skid off the road with the

appropriate tire squeal sounds.

The task is fairly difficult for many people, in

that many subjects' speed drifted considerably.

Because the simulator is not motion-based there are

no proprioceptive sensations of movement such as

the feeling of lateral acceleration from rounding a

curve or the feeling of acceleration or deceleration

from applying the throttle or brake respectively.

Each driver was asked to perform this driving

task two times: once while listening to their

favourite genre of music and once listening to their

least favourite. Genre preferences were identified

from the questionnaire they answered before

realizing they would be driving to their least

favourite music. To ensure results were not biased

by the order the genre was played, alternate subjects

heard their favourite and least favourite genre first.

Each run was divided into 6 parts:

ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics

1. The Music is off and the Speedometer is

Displayed. This period from 0-120 seconds was

used to get the subject up to normal driving speed

and to get adjusted to judging speed on the

simulator.

2. The Music is off and the Speedometer is off.

This period between 120-240 seconds was used to

establish a baseline for the subject's speed holding

without a speedometer, which was continued

throughout the remained of the experiment.

3. The Music is Turned on at 240 Seconds with

the first song of the chosen genre. Alternate subjects

had the songs ordered with increasing tempo or

decreasing tempo.

4. Song 2 is played

5. Song 3 is played

6. Song 4 is played

The purpose of this procedure was to look for

evidence of 3 main effects: that music influences

driving, that the subjects' like or dislike of the music

influences driving, or that tempo influences driving.

3.1 The Questionnaire before the Drive

Several questions were asked of drivers before the

drive to frame their general disposition about music.

The subject's age group was asked because there is

considerable evidence that musical preferences are

set at particular times in a person's life, and the

popular music of that time may influence their

lifelong preferences. In our study 24 of the 39

subjects answered "yes" when asked if their current

favourite genre is the same now as when they were

growing up. This is consistent with the findings of

(Mulder et al., 2009) and others who found that

music taste is well developed in adolescence and

crystallizes in middle age. It was also found that

preferred artists may change a lot during a lifetime,

but genre preference is much less likely to change

and style is by far the most stable.

The selection of genres used for this study was

somewhat arbitrary. There are several systems of

genre, and the understanding of genre types may

differ between individuals to some extent. When

asked, 36 of the 39 subjects reported that they felt

the songs in the playlist for their favourite genre

were representative of that genre.

Pop and Rock were substantially more popular

than all other genres (see Figure 2) and Alternative

and Electronica were ranked lowest. Pop and Rock

are likely the best known and most familiar genres

while Alternative and Electronica are less known

which may account for their apparent lack of

popularity in our study. No attempt was made in this

study to find out what were the true preferences of

the subjects other than asking them what they like to

listen to. Selected questions from the questionnaire

include:

1. List the nine genres in order of preference.

2. Is your favourite genre in the list

3. What was your favourite genre growing up?

4. Is your favourite growing up still your favourite

today?

5. How much time do you spend listening to music

in a day?

6. What kind of music do you listen to in a car?

7. Gender

8. Age group

9. Most prominently listened to type of music

10.Average number of hours spent listening to music

in a day

11.Subject's overall impression of vehicle audio

entertainment.

2423

ternati

Coun

Elec

ronica

Jaz

Pop

Hip-Ho

Rock

Figure 2: Number of subjects that selected each genre as

their favourite.

3.2 The Playlists

Nine commonly defined genres were selected and

populated with four songs each in order of changing

tempo. With some subjects the tempo increased

while with other subjects the tempo decreased. The

songs averaged 3 minutes and 44 seconds long, with

the longest being 5 minutes and 37 seconds and the

shortest being 2 minutes and 46 seconds. Each

playlist averaged 14 minutes and 33 seconds with

the longest (Rock) 20 minutes and 8 seconds and the

shortest (Reggae) 10 minutes and 33 seconds.

3.3 The Vehicle Simulator

The vehicle simulator (Figure 3) consists of two

seats, a steering wheel and a pedal cluster from a

Ford Motor Company vehicle mounted on a

stationary frame. Computers create engine sounds

through an audio system and display a synthetic

scene with the appearance that the subject is driving

INTELLIGENT MUSIC SELECTION TO INFLUENCE DRIVER BEHAVIOUR - An Empirical Study

a car on the road with no stops. The scene is a loop

that repeats several times during a single course.

The simulator also has LCD screens in the

location of rear-view mirrors, in the location of a

centre-stack display and the instrument cluster.

The computer system is capable of playing

playlists from the Rhapsody™ music player, timed

to start with a delay of 4 minutes from when the

simulator begins a course.

Figure 3: The stationary vehicle simulator at Ford

Research and Innovation Center that was used for this

experiment.

3.4 The Final Questionnaire

Throughout the experiment the subject was not

informed specifically what the experiment was

trying to determine. However, at the end of each run

the subject was asked to fill out a questionnaire

consisting of questions about what they thought of

the music. The questionnaire questions follow:

1. Do you think the music you just heard was

representative of the _________ genre?

2. Did you like the music?

3. Did you know any of the songs?

4. If yes, which song did you know?

5. Were you bored while driving?

6. Do you think the music you listened to influenced

your driving?

7. If yes, how was your driving influenced?

8. Do you have any other comments regarding your

drive?

4 RESULTS

With respect to the first main effect, that music

influences driving behaviour, we found 48% of the

cases where we felt we could observe a change

immediately after the music began to play. Figure 4

shows a case where the subject's speed increases and

speed fluctuations become more pronounced when

the music starts. In Figure 5 the least favourite music

is played and speed decreases instead of increasing.

This supports the second main effect that like or

dislike of the music also influences driving

behaviour.

One observation from comparing Figure 4 and

Figure 5 is that while in both figures speed holding

variability increased after the music started, in

Figure 5 where the music is the least favourite the

variability is lower, which indicates that different

music can have different effects on the driving

behaviour.

120

0 240 480 720 960 1200

Time (Seconds)

Speed (MPH)

Speedometer, No Music No Speedometer, No Music First Song

Second Song Fourth Song Series6

Figure 4: This graph shows the speed holding behaviour of

one subject exemplifying the effect of music on some

drivers' speed holding behaviour. Immediately after music

from the favoured genre begins to play at 180 second the

driver's speed increases and fluctuates with greater

amplitude.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Speedometer, No Music No Speedometer, No Music

First Song Second Song

Fourth Song Series6

Figure 5: This graph shows the speed holding behaviour of

the same subject as in Figure 4 when listening to the least

favoured genre. As in Figure 4, speed holding becomes

more erratic when the music is played, but the speed

decreases instead of increasing.

Figure 6 and Figure 7 demonstrate the same

pattern for a different subject. This subject had much

better speed control than the subject of Figure 4 and

Figure 5, but still exhibited finer speed control

fluctuations when listening to the least favourite

genre in Figure 7 when compared to the favourite

genre in Figure 6.

It should be noted at this point that speed holding

profiles such as the one in Figure 4 are indicative of

speed being controlled by vehicle dynamics. In these

ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics

cases, the subject increases speed to over 100 MPH

where the simulator is very difficult to control in the

curves, so the subject slows in the curves and then

picks up speed on the straightaway. Other subjects

are apparently able to maintain a more steady speed

as with the subject in Figure 6 and Figure 7. Most

likely the two types of subjects are controlling their

speed in different ways.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Speedometer, No Music No Speedometer, No Mus ic First Song

Second Song Third Song Fourth Song

Figure 6: This is one example of a subject with good speed

control listening to their favourite genre.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Figure 7: This is the speed holding model for the subject

of Figure 6 while listening to the least favourite genre.

We also looked at the data for evidence that a

driver listening to the favourite genre will travel at a

different speed than a driver listening to the least

favourite genre. There were 6 subjects out of the 39

(15%) that exhibited faster driving when listening to

their least favourite genre than when listening to

their favourite genre. The opposite was true for 7 out

of the 39 (18%) that drove faster when listening to

their favourite genre. Examples of the shift in

driving speed are found in Figure 8 where the

subject drives faster when listening to the favourite

genre and in Figure 9 where the subject drives

slower while listening to the favourite genre.

The third main effect that this study was intended

to test is whether or not musical tempo influences

how fast a subject drives. This general idea is

supported by considerable research (Turley, 2000),

(Miller, 2003) that temp may influence how quickly

shoppers move through a grocery store or eat their

dinner in a restaurant. One might also conjecture that

musical tempo could alter how fast a subject will

drive in the simulator.

We found that 5 out of our 39 subjects (13%)

speed up as the tempo of the music is increased and

that 4 out of 39 (10%) drove at a slower speed. Two

examples of this are found in Figure 10 and Figure

11.

120

0 240 480 720 960 1200 1440

Time (seconds)

Speed (MPH)

Favorite Least Favorite

Figure 8: Subject drives at a higher speed while listening

to the favourite genre. Just before 960 seconds while

listening to their favourite genre, the drive loses control of

the car and spins out due to excessive speeds.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Least Favorite Favorite

Figure 9: Subject drives at a lower speed while listening to

the favourite genre.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Speedometer, No Music No Speedometer, No Music

First Song Second Song

Third Son

Fourth Son

Figure 10: This figure shows speed holding for a subject

listening to the least favourite genre while the tempo

increases. It demonstrates that some subjects' speed

increased as the tempo increased.

120

0 240 480 720 960 1200

Time (seconds)

Speed (MPH)

Speedometer, No Music No Speedometer, No Music Firs t Song

Second Song Third Song Fourth Song

Figure 11: This graph shows the speed holding for a

subject listening to the favourite genre while the tempo

increases. The speed gradually increases as the tempo

increases.

INTELLIGENT MUSIC SELECTION TO INFLUENCE DRIVER BEHAVIOUR - An Empirical Study

5 CONCLUSIONS

The advancements in personalized digital music

technologies and their integration into vehicle

systems create promise for the next generation of

automotive music entertainment that will match

specific music with the driving environment and

influence driver behaviour in the desired direction.

Although there is a substantial amount of anecdotal

evidence that music influences the way we drive, the

research of this subject to support the development

of an automatic recommender system is rather

limited. This paper describes our preliminary study

into the influence of music on driving behaviour. In

our study 89.7% of the subjects exhibited some

change in driving behaviour during the experiment,

of which a significant fraction of the subjects

exhibited this change immediately when the music

was turned on. However, the specific change in

behaviour in response to the music preference or

tempo appears to be very personal. More work in

this area is warranted to explore these results further.

Larger data sets are needed to sufficiently

characterize driver behaviour as a result of the music

they listen to. Naturalistic studies would be very

beneficial to eliminate the effects of a reduced visual

scene and lack of proprioceptive inputs on speed

control. Relevant methods of classifying both music

and driver behaviour need to be discovered and

implemented in future experiments.

ACKNOWLEDGEMENTS

The authors wish to thank Michael Gunawan, a 2009

Ford Motor Company High School intern who was

instrumental in conducting the experiment in this

paper and analyzing the data.

REFERENCES

Bayly, M., Young, K. L., and Regan, M. A., Sources of

Distraction inside the Vehicle and Their Effects on

Driving Performance in Young, K. L., Lee, J. D. and

Regan, M. A. (Eds.), Driver distraction: Theory,

Effects and Mitigation, Boca Raton, FL: CRC Press,

2008, pp.191-213.

Dibben N. and Williamson, V. J.; An exploratory survey

of in-vehicle music listening, Psychology of Music

October 2007 35: 571-589.

Dornbush, S., Joshi, A., Segall, Z., Oates, T.; A Human

Activity Aware Learning Mobile Music Player, Juan

C. A., J. C., and Shapiro D. (Eds.), Advances in

Ambient Intelligence: Frontiers in Artificial

Intelligence and Applications, Volume 164, 2007:

107-122.

Miller, J. M.; Peden, B. F.; Complexity and Degree of

Tempo Modulation as Factors in Productivity; Psi Chi

Journal of Undergraduate Research; The National

Honor Society in Psychology Vol. 8, No. 1, 2003: 21-

27.

Mulder, J., Bogt, T., Tom F. M., Raaijmakers, Q.,

Gabhainn, S. N., Sikkema, P., From death metal to

R&B? Consistency of music preferences among Dutch

adolescents and young adults; Psychology of Music

January 2010 vol. 38 no. 1: 67-83

Turley, L. W.; Milliman, Ronald E.; Atmospheric Effects

on Shopping Behaviour: A Review of the

Experimental Evidence; J BUSN RES 2000. 49. 193–

211; 2000 Elsevier Science Inc.

ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics