using ambient lighting in persuasive communication

Using ambient lighting in persuasive communication

Citation for published version (APA):Lu, S. (2017). Using ambient lighting in persuasive communication. Eindhoven: Technische UniversiteitEindhoven.

Document status and date:Published: 05/04/2017

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 13. Jun. 2020

https://research.tue.nl/en/publications/using-ambient-lighting-in-persuasive-communication(e2572ba0-8c82-4e49-948d-2c1e1c57a04e).html

Using Ambient Lighting in PersuasiveCommunication

Using A

mbient Lighting in Persuasive C

omm

unication

Eindhoven University of TechnologyDepartment of Industrial Engineering & Innovation Sciences

Shengnan Lu

Shengnan Lu

Invitation

You are cordially invitedto the public defense of my

PhD dissertation:

Using Ambient Lighting in Persuasive Communication

Location:Auditorium, Collegezaal 4,

Eindhoven University of Technology

Time:Wednesday at 16.00

April 5th, 2017

Shengnan [email protected]

Using Ambient Lighting in PersuasiveCommunication

✐✐

✐✐

✐✐

✐✐

The work described in this dissertation has been carried out at the Eindhoven Univer-sity of Technology, the Netherlands. An electronic copy of this dissertation in PDFformat is available from the TU/e library website (http://www.tue.nl/lib)

All rights reserved. Reproduction of this publication in whole or in part is prohibitedwithout the prior permission from the author.

Cover visual design: Veronica Magica, Athens, Greece.Cover inspiration: Shengnan Lu & Sofia Fountoukidou, Eindhoven, the Netherlands.

Typeset with LATEX.Printed on Biotop paper.Printed by Ipskamp Drukkers B.V., Nijmegen, the Netherlands.

A catalogue record is available from the Eindhoven University of Technology Library.ISBN: 978-90-386-4248-2NUR: 741Copyright ©2016 - 2017, Shengnan Lu, the Netherlands.

✐✐

✐✐

✐✐

✐✐

Using Ambient Lighting in Persuasive Communication

PROEFSCHRIFT

ter verkrijging van de graad van doctoraan de Technische Universiteit Eindhoven,

op gezag van de rector magnificus prof.dr.ir. F.P.T. Baaijens,voor een commissie aangewezen door het College voor Promoties,

in het openbaar te verdedigen op woensdag 5 april 2017 om 16:00 uur

door

Shengnan Lu

geboren te Chongqing, China.

✐✐

✐✐

✐✐

✐✐

Dit proefschrift is goedgekeurd door de promotoren en de samenstelling van de pro-motiecommissie is als volgt:

voorzitter: prof.dr. I.E.J. Heynderickx1e promotor: prof.dr. C.J.H. Middencopromotor, dr. J.R.C. Hamleden: prof.dr. J.H. Eggen

prof.dr. D.V. Keyson (Technische Universiteit Delft)prof.dr. B.E.R. de Ruyter (Radboud Universiteit Nijmegen)prof.dr. M. Tscheligi (University of Salzburg)

Het onderzoek dat in dit proefschrift wordt beschreven is uitgevoerd in overeenstem-ming met de TU/e Gedragscode Wetenschapsbeoefening.

✐✐

✐✐

✐✐

✐✐

对大自然的敬畏，让我能感恩并且享受它所赐予我们的一切体验！每天一睁开眼，宛如游泳的少年在爱琴海探出水面，不快，不慢。

.

.——谨以我的博士论文，献给我

最爱的爸爸和妈妈遇见你们是我这一生最幸运的事

.

.卢圣南

二零一七年二月于荷兰..

✐✐

✐✐

✐✐

✐✐

vi

✐✐

✐✐

✐✐

✐✐

Table of contents

1 General introduction 11.1 Persuasive Technology . . . . . . . . . . . . . 21.2 Ambient Persuasive Technology . . . . . . . . 101.3 Using Lighting in Ambient PT . . . . . . . . . 121.4 Psychological Perspective . . . . . . . . . . . . 171.5 Research Goals . . . . . . . . . . . . . . . . . 201.6 Overview of the Chapters . . . . . . . . . . . 21

2 The role of color associations 252.1 Introduction . . . . . . . . . . . . . . . . . . . 252.2 Study 2.1 . . . . . . . . . . . . . . . . . . . . 30

2.2.1 Method . . . . . . . . . . . . . . . . . 322.2.2 Results . . . . . . . . . . . . . . . . . . 352.2.3 Discussion . . . . . . . . . . . . . . . . 38

2.3 Study 2.2 . . . . . . . . . . . . . . . . . . . . 382.3.1 Method . . . . . . . . . . . . . . . . . 412.3.2 Results . . . . . . . . . . . . . . . . . . 42

vii

✐✐

✐✐

✐✐

✐✐

Table of contents

2.3.3 Discussion . . . . . . . . . . . . . . . . 452.4 General discussion . . . . . . . . . . . . . . . 46

3 The role of color contexts 533.1 Introduction . . . . . . . . . . . . . . . . . . . 543.2 Study 3.1 . . . . . . . . . . . . . . . . . . . . 57


3.3 Study 3.2 . . . . . . . . . . . . . . . . . . . . 643.3.1 Method . . . . . . . . . . . . . . . . . 643.3.2 Results . . . . . . . . . . . . . . . . . . 663.3.3 Discussion . . . . . . . . . . . . . . . . 69

3.4 General discussion . . . . . . . . . . . . . . . 69

4 The role of color temperature experiences 734.1 Introduction . . . . . . . . . . . . . . . . . . . 744.2 Study 4.1 . . . . . . . . . . . . . . . . . . . . 77


4.3 Study 4.2 . . . . . . . . . . . . . . . . . . . . 824.3.1 Method . . . . . . . . . . . . . . . . . 834.3.2 Results . . . . . . . . . . . . . . . . . . 884.3.3 Discussion . . . . . . . . . . . . . . . . 91

viii

✐✐

✐✐

✐✐

✐✐

Table of contents

4.4 General discussion . . . . . . . . . . . . . . . 92

5 General discussion 975.1 Understanding the associative processes . . . . 985.2 Understanding the experiential processes . . . 1025.3 Contributions to design and society . . . . . . 1055.4 Limitations and directions for future research 1075.5 Ethical considerations . . . . . . . . . . . . . 1085.6 Conclusion . . . . . . . . . . . . . . . . . . . . 110

References 113

Summary 131

List of Figures 135

Appendices 141

A Thermostat Task 143A.1 Study 2.1 & Study 2.2 . . . . . . . . . . . . . 143A.2 Study3.2 . . . . . . . . . . . . . . . . . . . . . 146

B Questionnaire 147B.1 Study 3.1 & Study 3.2 . . . . . . . . . . . . . 147B.2 Study 4.1 & Study 4.2 . . . . . . . . . . . . . 149

Publication List 155

ix

✐✐

✐✐

✐✐

✐✐

Table of contents

Acknowledgements 157

CurricuLum Vitae 161

x

✐✐

✐✐

✐✐

✐✐

1 | General introduction

Since life evolved under the continuous influence of light, it is notsurprising that practically all forms of life, including the human

being, has developed a variety of responses to light and its characteristics.

Light supports human beings living functionally: it helps them to observethe world, and more interestingly, may also help them to communicate.More than 2000 years ago, the Greek already started using light as acommunication tool by using sunlight reflected by shields to give ordersto distal soldiers (see Figure 1.1a). More recently, in the late 19th century,the heliograph was invented for instantaneous optical communicationover long distances by flashes of sunlight reflected by a mirror (see Figure1.1b). Nowadays, we may no longer need shields or heliographs to sendmessages, but we still use light in many other forms of communication.

In brief, the current dissertation aims to study lighting as a communica-tion medium embedded in the ambient environment with the intentionto facilitate behavior change, especially, energy-conservation behavior.The focus of the research presented in this dissertation is on understan-ding the psychological mechanisms underlying the effectiveness of am-bient lighting persuasive technology. These effects will be tested in thedomain of sustainable behavior within the context of persuasive techno-logy. In the next section, we will first describe this persuasive technology

1

✐✐

✐✐

✐✐

✐✐

Chapter 1. General introduction

(a) (b)

Figure 1.1: Examples of human beings using light to communicate in history: (a) signalwith a shield; and (b) heliograph.

context.

1.1 Persuasive Technology

Studying persuasion goes back a long way. It was already acknowledgedby the ancient Greeks, who worshiped and studied Peitho (Πειθω: thegoddess of persuasion, see Frazer, 1898) and has been the subject ofhuman investigation for more than two thousand years (e.g., Aristotle’sRhetoric and Topics, see Cope, 1867). It was not until the twentiethcentury that ideas about persuasion and their link to empirical observa-tions were studied in science. Early work ranged from case studies of theebb and flow of public opinion in which the authors speculated aboutthe causes of observed shifts in attitudes (e.g., Lazarsfeld, Bernard, &Hazel, 1948), to content analyses of political propaganda in which theauthors speculated about the attitudinal and behavioral consequences ofpropaganda messages (e.g., Lasswell, Casey, & Smith, 1935).

Next to much of the early work that was investigated in the domain ofsociology and politics, psychology and communication scholars extended

2

✐✐

✐✐

✐✐

✐✐

1.1. Persuasive Technology

the research area of persuasion, and defined persuasion as “human com-munication that is designed to influence others by modifying their beliefs,values, attitudes, or behaviors.” (Simons, 1976). In reviewing the historyof research on communication and persuasion, social psychologists usedmostly the experimental approach to explore a wide variety of research onattitude and behavior change (for an overview see Delia, 1987). Remark-ably, in the 1950s, the Yale group was the first to examine systematicallythe variables that continue to be of interest today such as attitude struc-ture (Hovland & Rosenberg, 1960), individual differences (Hovland &Janis, 1959), message order effects (Hovland, 1957) and many others.

More recently, psychology and communication scholars defined persua-sion as “...a symbolic process in which communicators try to convinceother people to change their attitudes or behaviors regarding an issuethrough the transmission of a message in an atmosphere of free choice.”(Perloff, 2010, p.12). This definition implies that persuasion concerns apersuader who, by the means of communication, changes the attitudesor behaviors of a target.

Given this definition, persuasion is everywhere, playing an essential rolein politics, religion, marketing, psychotherapy, education, and day-to-daysocial interactions (see e.g., Cialdini, 2007; Moyer-Gusé, 2008; Murphy &Shleifer, 2004). For instance, politicians may use all kinds of strategies topersuade people to vote for them; governments team up with supermar-kets to persuade people to eat more healthily; schools try to persuadestudents to believe that intelligence is malleable, and your partner maypersuade you to cook pasta while you want to eat pizza tonight.

With the development of persuasion research, psychologists explored thebenefits of persuasive communication that are not limited to a human-to-human context: information sources could also be other sources thanjust human beings. For example, artificial agents can convey persua-sive messages, just as human sources can. These forms of persuasivecommunication differ from human-to-human persuasive communication,

3

✐✐

✐✐

✐✐

✐✐


in which they use a different source, can deliver kinds of different mes-sages, can use different channels, and can have different goals. Thatis, non-human sources can deliver more diverse messages that containthe information or meaning the source is hoping to convey. Also, non-human sources can use various sensory channels than the traditional ver-bal channel only (e.g., speaking, seeing, touching, smelling and tastingor combinations of these, see Berlo, 1960). Finally, non-human sourcescan be programmed to be almost anywhere, anytime and thereby canstrive for more goals, no matter as a group or an individual. Therefore,during the last decade, the study of persuasion in a human-to-technologycontext has emerged and grown into a global area of research and de-sign within the human-technology interaction domain. These interactivetechnologies and systems have the potential to engage in the same per-suasion processes that humans do, and have distinct advantages overhuman persuaders. For example, these interactive systems are persis-tent, allow anonymity, can store, access and manipulate huge volumesof data. Also, they can for example use many modalities (e.g., graph-ics, rich audio and video, animation or simulation) to influence people’spreferences for visual, audio, or textual experiences while enhancing thepersuasive power of these systems (e.g., Fogg, 2002; King, Dent, & Miles,1991).

Fogg (1999) was one of the first to articulate the importance of persua-sion in human-computer interaction for the design of interactive systemswith the intent to change human attitudes or behaviors. With his bookon the topic, Fogg (2002) initiated a research field that is now called per-suasive technology. Many efforts of persuasive technology researchershave been made to study interactive computing systems designed tochange people’s attitudes or behaviors (see e.g., IJsselsteijn, de Kort,Midden, Eggen, & van den Hoven, 2006; Midden, Kaiser, & Mccalley,2007; Oinas-Kukkonen & Harjumaa, 2008). These systems do not usecoercive or deceptive means to change attitudes or behaviors. Instead,

4

✐✐

✐✐

✐✐

✐✐


persuasive technology aims at inducing voluntary changes of behaviorswithout taking away control from its users (Fogg, 1999, 2002).

These voluntary changes of behaviors can be accomplished in many areassuch as in the domain of health, in which persuasive technology could helpusers to maintain an active lifestyle by tracking physical activity (Con-solvo, Everitt, Smith, & Landay, 2006; Consolvo, Klasnja, McDonald,& Landay, 2009), to moderate nutrition (Whiteley, Bailey, & McInnis,2008), to quit smoking (Khaled et al., 2008), or to increase their aware-ness about maternal health (Parmar & Keyson, 2008); in the domain ofsafety, in which persuasive technology could educate users of the advan-tages of driving safely and reducing their overall driving speed (Millar& Millar, 2000); in the domain of marketing, in which persuasive tech-nology could adapt to individual preferences by tracking consumers andtheir preferences over multiple stores and across media (Amit & Zott,2001; Zanker & Jessenitschnig, 2009); and in the domain of environmen-tal sustainability, in which persuasive technology could create “energyawareness” and promoting energy-related behavioral change (McCalley& Midden, 2002; Sauer, Schmeink, & Wastell, 2007; Svane, 2007).

Persuasive technology can have various functions: it can work as a socialactor capable of establishing a social relationship that forms the basisof social influence such as social approval (e.g., Ham & Midden, 2014)or social comparison (e.g., Midden, Kimura, Ham, Nakajima, & Kleppe,2011); it can be a medium that allows for novel experiences such assounds or touch that create and it can provide tools that guide andsupport behavior (for an overview see Fogg, 2002). In the next section,we will take a closer look at using persuasive technology as a tool forproviding feedback.

Feedback has often been used as a persuasive tool to influence people’sbehavior. It has long been acknowledged in the field of psychology thatfeedback or knowledge of results can have a positive effect on performance(e.g., Becker & Seligman, 1978). That is, by providing feedback about the

5

✐✐

✐✐

✐✐

✐✐


effects of people’s behavior, people can easily monitor the consequencesof their behavior and adapt their behavior in response to that feedback.

In the domain of environmental sustainability, a basic feature of per-suasive technologies (and a cornerstone of behavioral change in general)is appropriate feedback about resource consumption (Darby, 2006) tocreate energy awareness (Midden et al., 2007; Redström, 2006) and toencourage sustainable behavior (Midden, McCalley, Ham, & Zaalberg,2008). Prior to the 1990’s, experiments indicated that electronic devicesmight contribute to the efficiency and effectiveness of behavioral inter-ventions, but technology just was not yet smart enough in most casesto make these devices very successful. By then, psychologists have yetmerely touched upon the opportunities offered by intelligent systems topromote (energy conservation) behavior. Earlier studies often used sim-ple technologies like written messages based on daily or weekly meterreading (e.g., Winett, Kagel, Battalio, & Winkler, 1978), while some re-searchers used electronic displays. More recently, in conjunction withthe growth of technologies, the number of persuasive feedback systemsgrew rapidly (e.g., providing smart electricity meter feedback). Interest-ingly, feedback can take many different forms. Therefore, in the followingsections, we will discuss different types of feedback and their potentialadvantages and drawbacks.

First, persuasive technology can use factual feedback. Several effortsexplored this type of feedback, which allows users to judge for examplelevels of household energy consumption (see e.g., Abrahamse, Steg, Vlek,& Rothengatter, 2005; Lockton, Harrison, & Stanton, 2008). For exam-ple, McCalley and Midden (2002) demonstrated in several studies thatimmediate feedback by adding an energy meter to the user interface ofa washing machine could persuade users to use less energy. The role ofthe energy meter is to provide a clearly explicit point of reference for im-proved billing (see Figure 1.2 left). Likewise, the Wattcher (see Wattcher@Online, 2016) is a display that uses numbers to give feedback about

6

✐✐

✐✐

✐✐

✐✐


the home’s energy consumption (see Figure 1.2 right). These numberspresent exactly how many kWh a person is using at that point in time,which means that the information that feedback provides is very specificfor this particular user and his or her particular behavior and situation.

Figure 1.2: Factual feedback: (left) energy meter and (right) Wattcher.

Second, persuasive technology can use social feedback (see e.g., Ham &Midden, 2010). Earlier research suggested that social reinforcement hasbeen applied widely in many domains such as child education, therapeu-tic programs, health behavior and social interaction as a mechanism forbehavioral change (Bandura & McDonald, 1963). Utilizing this socialmechanism in the persuasive system, Ham and Midden (2010) proposedsocial feedback by making use of a social robot. This social feedback asprovided by a robotic agent can show signs of disapproval or negative so-cial incentives (e.g., a sad face or a smiling face) as feedback (see Figure1.3). Remarkably, their research indicates that the behavioral effects ofsocial feedback are greater compared to the effects of factual feedback.In addition to the social robot, social feedback can also be providedby other non-human sources, like a smart, on-screen computer agent.For example, research showed that a human-like artificial social agent(a virtual human-like torso and head) can effectively influence users byproviding social feedback (e.g., disapproval of their energy consumption,see Ruijten, Midden, & Ham, 2015).

Third, persuasive technology can use a subtle, unobtrusive, and cogni-tively efficient type of feedback: ambient feedback (e.g., Miller, Rich, &

7

✐✐

✐✐

✐✐

✐✐


Figure 1.3: Social feedback as provided by a robotic agent.

Davis, 2009; Maan, Merkus, Ham, & Midden, 2011). Ishii and Ullmer(1997) suggested the use of ambient media such as sound, light, airflow,and water movement to act as background interfaces with cyberspace andwork at the periphery of human perception. Wisnesky and colleagues(1998) supported this definition of ambient technology, by describingambient technology as being at the “periphery of our attention”, need-ing “subconscious understanding” and possibly involving “all our senses”.Given this definition, ambient feedback has been widely investigated andapplied in practice (e.g., Arroyo, Bonanni, & Selker, 2005; Wilson, Lil-ley, & Selker, 2013). For example, an ambient device called Energy Orb(Energy Orb @Online, 2016, see Figure 1.4 left) was used to provideenergy consumption feedback that changed color dependent on the time-of-use tariff in operation (e.g., by glowing red when energy usage reachesa certain level and green otherwise). Another example of an ambient de-vice is the Power Aware Cord (Gustafsson & Gyllenswärd, 2005), whichcan visualize the current uptake of electricity of a connected appliancethrough glowing pulses, flow, and intensity of light (see Figure 1.4 right).Similar to the ambient feedback like Energy Orb, Show-me can displaythe amount of water that is used during a shower by a series of LEDs ina row assembled on a stick in the shower cabin, in which an increasingwater level is visualized (Kappel & Grechenig, 2009).

Also, ambient feedback can change people’s behaviors by making a parti-cular behavior fun to do. Piano Stairs (see Piano Stairs @Online, 2016),

8

✐✐

✐✐

✐✐

✐✐

1.2. Ambient Persuasive Technology

The interface of the Power-Aware Cord invites users to plug in different appliances and experiment with how these relate to each other in terms of energy. Our idea is that, by integrating the information about electric flows where they actually are, will result in a more intuitive, playful and metaphorical display than what would normally be produced. Such an approach might inspire users of the Power-Aware Cord to explore and reflect upon the energy consumption of other electrical devices in their home, using the cord.

Figure 1. The Power-Aware Cord prototype.

Our aim is also that users might perceive the light patterns as the actual electricity in the cord, if not on a direct level then at least on an intuitive and metaphoric level. By this, we mean that people might talk about and refer to the light just as if it would be the electricity itself, even if they on a logical level would realize that it is just a representation.

With the Power-Aware Cord, users’ actions, such as plugging or unplugging electrical devices into sockets, immediately result in a response from the cord, giving the user direct feedback and the feeling of both seeing and interacting with electricity. The rich affordances stemming from the Power-Aware Cord, as a tangible, real-world object adds to its intuitiveness [3].

Representing the amout of electricity with light, and not a numerical display, supports the notion of ambient information displays that does not force the user to walk up to the device but rather to receive the information at a glance from a distance. This enables the interface to be accessed from any place in the room in which it is situated, and thus constantly providing subtle information on energy usage. This could be useful in order to change behaviours

and awareness over time or to detect unnecessary stand-by consumption.

Related Work Studies of energy information displays in homes have been made in “The Effects of Information on Residential Demand for Electricity” by Isamu Matsukawa [2]. In this paper, he argues that consumers need information in order to act rationally. He establishes that the display used in this example does affect the energy consumption with its graphs and numerical information, even if the effect is small.

There are several attempts to visualize abstract information. Similarities to our concept can be found in Chatzitsakyris et al. proposed Ambient Media-implementations [1]. Here the geographical location of subway-lines was visualized by light in the side-walks of a city. In this ambient-media example, invisible occurrences in time and space are made visible, in reference to the time and space in which they occur.

“The videocard game” by Buur and Soendergaard, is an example of augmented reality, augmenting the object [5]. By embedding computer monitors into a table and providing physical playing cards with electronic ID tags, a tool for video analysis was created. The artifacts augmented in this example will very clearly appear augmented to the user.

The work by Raby & Dunne in their Design Noir [6] can be mentioned both as inspiration and related work in the way they design everyday objects that visualize hidden properties such as electromagnetic fields stemming from devices in the surroundings.

THE PROTOTYPE The light in the Power-Aware Cord is obtained by the use of electroluminescent wire. This wire contains a phosphor layer that glows with an intensive blue-green light when an alternating current is introduced. Due to the color of the phosphor, the wire appears to be white when unpowered, a feature that greatly reduces its appearance in the device when unpowered. This hides the underlying construction from the user and adds to the unexpected experience when the device is powered and starts to glow. Since the cable will shift in colour when going from unlit to lit mode the user can easy detect low intensity levels.

Three luminescent wires are bound together with ordinary copper wires for electric conduction. The wires are twisted together to improve the flexibility of the resulting cable. Twisting the wires also gives the possibility to create an effect of motion through the cable by powering each of the three luminescent wires in turn. The whole structure is coated with a layer of transparent silicone.

The resulting functionality in the Power-Aware Cord is an extension cable capable of delivering up to 10 amperes load. The resulting power range that can be measured is 0 – 2200 W. The three microprocessor controlled light

CHI 2005 | Late Breaking Results: Posters April 2-7 | Portland, Oregon, USA

1424

Figure 1.4: Ambient feedback: (left) Energy Orb and (right) Power Aware Cord.

a well-known artwork, plays notes of a melody or piano tones when peo-ple walk on them, with the goal of stimulating people to take the stairsinstead of the elevator or escalator. A final example of effective ambientfeedback is the living plant display trash bin developed by Holstius andcolleagues (2004). Holstius and colleagues were inspired by the plant pho-totropism from nature and designed a living plant display in the ambientenvironment, in which recyclable waste or trash can trigger directionalbursts of light that gradually induced the plant displays to lean towardthe more active container. This ambient designed to provide cumulativeambient feedback about recycling behavior was shown to encourage re-cycling behavior effectively. Therefore, as described in these examples,feedback in persuasive technologies as provided by ambient stimuli mayeffectively influence people in a subtle and unobtrusive way.

Persuasive technology that uses especially this kind of psychological pro-cesses (like ambient feedback) is labeled Ambient Persuasive Technology(Davis, 2008; Ham & Midden, 2010), which was shown to be effectivewhile being at the “periphery of our attention” (see Wisneski et al.,1998, p.22), “without conscious attention” (see Ham, Midden, & Beute,2009, p.5).

9

✐✐

✐✐

✐✐

✐✐


1.2 Ambient Persuasive Technology

Ambient persuasive technology aims to exploit the human peripheralperception capabilities, while most types of conventional persuasive tech-nologies are only effective if the user spends ample attention to them. Inmany daily situations, people might not be motivated or lack the cogni-tive capacity to consciously process relatively complex information suchas factual feedback (see Bargh, 1997, for a discussion of the automaticnature of human thinking). Inspired by Weiser’s vision of ubiquitouscomputing (Weiser, 1993), ambient persuasive technology employs theapproach that the “information is moved off the screen into the physicalenvironment, manifesting itself as subtle changes in form, movements,sound, color, smell, temperature, or light” (Wisneski et al., 1998, p.23),instead of employing the attention-demanding approach.

Indeed, more and more, persuasive technologies are integrated into ourenvironment to deliver persuasive messages, in which humans are notexpected to have high cognitive capacity and spend conscious attention(e.g., Ham & Midden, 2010; Pousman & Stasko, 2006). Earlier researchby Ishii and Ullmer (1997) suggested the use of ambient media to actas background interfaces with cyberspace and work at the periphery ofhuman perception. This allows new forms of influencing through subtlecues in the ambient environment reflecting changes in light, odor, sound,airflow and so on (e.g., Baron, 1997; De Ruyter & Aarts, 2004; Ishii &Ullmer, 1997). For example, studies indicated that ambient stimuli, suchas pleasant ambient odors (e.g., baking cookies, roasting coffee) in the air,can significantly influence the behavior of people (e.g., Baron & Bronfen,1994; Baron & Thomley, 1994; Warm, Dember, & Parasuraman, 1991).That is, passersby were more likely to help the accomplice when pleasantfragrances were present in the air than when they were absent (Baron& Thomley, 1994). Also, ambient stimuli, such as colored light, wereused with the intention to reduce water consumption by tracking and

10

✐✐

✐✐

✐✐

✐✐

1.2. Ambient Persuasive Technology

displaying information about water consumption at the sink itself (seeFigure 1.5, Arroyo et al., 2005).

(a) (b)

Figure 1.5: Waterbot: (a) indicating warm water; while (b) indicating cold water.

In this sense, ambient stimuli providing feedback through the visual-ization of sensed data, in which users process the persuasive messagesoccasionally without needing high cognitive, seem well suited to provideusers with feedback in a more subtle manner than direct informationpresentation (see Kim, Hong, & Magerko, 2010).Would persuasive interventions that employ this kind of ambient persua-sive technology be able to influence a user without receiving consciousattention from that user? Ham and colleagues (2009) devised a form ofpersuasive technology to which the user could not pay conscious attentionbecause it employed subliminal feedback. That is, participants had tomake 90 different choices (relating to energy consumption) and receivedfeedback about the correctness of their choice through the presentationof a smiling or sad face. For some participants, these faces were presentedonly for 22 ms, which prohibited conscious perception of these stimuli.Results suggested that this subliminal feedback led to the same levelsof influencing the user as presenting the faces supraliminally (150 ms).Interestingly, earlier studies confirmed these cognitively-efficient effectsby showing the non-conscious influence of ambient (e.g., olfactory) cueson behavior, while participants were unaware of this influence (Holland,Hendriks, & Aarts, 2005).

11

✐✐

✐✐

✐✐

✐✐


These findings suggested that persuasive technology can influence users’behaviors or attitudes without receiving conscious attention from thatuser. So, ambient persuasive technology might prove to be effective instimulating behavior, especially for people who are not motivated tospend too much attention to certain messages (because they lack moti-vation or processing capacity for a specific topic). Therefore, ambientpersuasive technology is able to influence attitudes or behaviors with-out conscious attention to that persuasive technology – not only becauseit necessitates a combination of knowledge about very new insights into(unconscious) psychological mechanisms and very new forms and designsof persuasive technology, but also because a lot of human day-to-day be-haviors are driven by unconscious motives (Bargh & Chartrand, 1999)and thereby the importance of ambient persuasive technologies that in-fluence these motives might be major.

1.3 Using Lighting in Ambient PT

Designing ambient persuasive technology with light is a promising newdomain that has gained attention in recent years. In particular, the in-tegration of luminaires into ambient interactive systems enables a novelway of using light in ambient persuasive technology (see e.g., Ham &Midden, 2010; Magielse & Ross, 2011). These systems use different cha-racteristics of lighting such as color, brightness, color temperature, LEDposition, and their combination.

The characteristics of lighting make it possible to induce behavior changethat only requires minimal attention and cognitive effort by the userwhile being effective. Lighting has specific qualities that make it partic-ularly suitable for helping users to attain their goals and serve as ambientpersuasive technology. That is, lighting can provide information directlyrelated to the situation and the task at hand (e.g., by showing a flow ofred waves projected on the ceiling towards an open window, indicating

12

✐✐

✐✐

✐✐

✐✐

1.3. Using Lighting in Ambient PT

heat loss), which allows the user to be easy to understand the informationwith low taxing. Also, lighting can be very cheap, energy-friendly, andeasy to install, while maintaining comfort and aesthetically pleasing. In-terestingly, lighting can be seen by other people present in a room as well,which induces social pressures as a persuasive mechanism and allowingfor social coordination to enhance group performance (see Shteynberg& Galinsky, 2011). Finally, lighting can trigger other mediating sys-tems, like walls or devices with light-sensitive coatings, which create avast variety of persuasive applications. In particular, modern-day light-ing systems (e.g., LED) provides us with new opportunities to mediateinformation and function as persuasive lighting. These overwhelmingadvantages and potential opportunities that are caused by an increase inthe variety and freedom of light-source form factor and characteristics oflight (e.g., built-in associative information, low conspicuity of light, andflexibility of lighting dimensions such as intensity, hue, beam shape), setlighting apart from other (ambient) persuasive mediums.

In parallel with technological developments, the body of knowledge onthe effects of lighting is growing, and we are increasingly aware of thepossible effects that light may have on people. Light affects people visu-ally (seeing), biologically and psychologically, and therefore has a majorimpact on the perception and experience of our environment and every-thing in it. Lighting also influences our social relations (Magielse & Ross,2011) and in many cases carries information. As a result of increasinginsights into the possible effects of light, many applications can be con-ceived of that go well beyond simply illuminating an environment for usto see.

Earlier research started by focusing on illumination and visual percep-tion, and showed a multitude of possible effects through the visual sys-tem. On the whole, researchers addressed issues such as visual taskperception, object recognition, contrast perception and visual comfortin different light settings (for an overview see Boyce, 2014). This has

13

✐✐

✐✐

✐✐

✐✐


resulted in guidelines for the amount of illumination and tolerated glarefor various situations and tasks (e.g., Vos, 1984).

However, researchers and designers realized that light should not only beanalyzed from a perceptual point of view, but should also be regarded asa driver of bodily, cognitive, emotional and behavioral responses by theperceiver in different contexts of everyday life. In other words, researchhas shifted towards the physiological and psychological effects of light.

A number of studies have shown the physiological effects of light. In par-ticular, light plays an important role in regulating important biochemicalprocesses and neuroendocrine control (for instance, melatonin and cor-tisol) via the eye (Hughes & Neer, 1981). Light exposure is the mostimportant stimulus for synchronizing the biological clock (Czeisler etal., 1989), suppressing pineal melatonin production (Brainard, Rollag,& Hanifin, 1997) and elevating core body temperature (Badia, Myers,Boecker, Culpepper, & Harsh, 1991). For example, shifting light fromlow (i.e., warm) to high (i.e., cool) color temperature can promote vas-cular tension regulated by the sympathetic nervous system, resulting inreducing the decrease of deep body-part temperatures (Higashihara etal., 2002).

A number of studies have also shown the psychological effects of light.In our daily life, we may notice that people are usually more cheerfulon a sunny day than people on a dark and foggy day. In the literature,empirical research has generated a variety of evidence for the psycholo-gical effects of light. Many studies have indicated the effects of light onfor example attention (Giusa & Perney, 1974), perceived guilt (Aspinall& Dewar, 1980), time-estimation task performance (Delay & Richard-son, 1981), interpersonal communication (Giusa & Perney, 1974), per-formance of various cognitive tasks and interpersonal behaviors (Baron,Rea, & Daniels, 1992), mood and decision making (Belcher & Kluczny,1987; McClaughan, Aspinall, & Webb, 1996), self-reported quality of life(Sörensen & Brunnström, 1995) and environmental atmosphere (Vogels,

14

✐✐

✐✐

✐✐

✐✐

1.3. Using Lighting in Ambient PT

2008).

Among different characteristics of light, color(ed lighting) has been ex-tensively investigated with respect to its psychological effects. For exam-ple, a series of experiments indicated that red lighting was more arousingthan colored lighting (e.g., green and blue), as measured by an increasedskin conductance (e.g., Gerard, 1958; Jacobs & Hustmyer, 1974; Wilson,1966), or as measured by an decreased electroencephalography (EEG) al-pha activity (e.g., Ali, 1972; Küller, Mikellides, & Janssens, 2009). Thispsychological effect of red may be mediated by associations with pastexperiences. That is, red is often associated with danger and is oftenused as a warning sign, while the color such as blue is associated withthe sea or clear sky (Kaiser, 1984). Also, short wavelength colors suchas blue and green can evoke more pleasant feelings than long wavelengthcolors such as orange and red (Bellizzi & Hite, 1992; Kwallek, Lewis,Lin-Hsiao, & Woodson, 1996; Valdez & Mehrabian, 1994).

Earlier research indicates that lighting can be effectively used as persua-sive technology. That is, a number of studies revealed that persuasivefeedback that consisted of lighting changes could influence sustainablebehavior (see e.g., Arroyo et al., 2005; Froehlich, 2009; Ham & Midden,2010; Maan et al., 2011; Van den Bos et al., 2008; Wisneski et al., 1998).For example, in the 1970’s, Becker and Seligman (1978) investigated theeffectiveness of light that went on “in a highly visible part of the home”whenever the air conditioner was on, while the outside temperature was20 ◦C or lower. In this way, homeowners were told that the signalingdevice indicated when they could cool their house effectively by open-ing the windows and turning off their air conditioner. In homes thatcontained such signaling device, an average of 15.7% saving in energyconsumption was found. More recently, ambient lighting devices, suchas the Energy Orb, could change color dependent on the time-of-usetariff in operation (e.g., by glowing red when energy usage reaches a cer-tain level and green otherwise). Using the same “traffic light” metaphor,

15

✐✐

✐✐

✐✐

✐✐


Kuznetsov and Paulos (2010) designed an ambient lighting persuasivetechnology for promoting water conservation. More specially, a greenlight is displayed near the faucet as long as the consumption is belowthe average of everything previously measured. When the user is aboutto exceed the average, the light changes to yellow. It becomes red if thewater has been running for longer than one standard deviation aboveaverage. Finally, it even flashes. More examples of using lighting as anambient persuasive medium in the domain of environmental sustainabil-ity have been proposed, such as Waterbot (Arroyo et al., 2005), PowerAware Cord (Gustafsson & Gyllenswärd, 2005), Show-Me (Kappel &Grechenig, 2009), and Energy Curtain (Redström, 2006). These am-bient lighting displays have lead people to reflect on their behavior andreconsider sustainability and environmental issues beyond heating energyor water usage.

Ambient lighting can not only be used effectively as persuasive techno-logy for promoting environmentally friendly behavior, but also for influ-encing other kinds of behaviors. In the domain of healthy lifestyle, anambient lighting device called MoveLamp can indicate a person’s recentphysical activity (by changing colors from green to red based on his orher step count via a pedometer) in the person’s workplace and encouragehim or her to integrate a little physical activity into the day (Fortmann,Stratmann, Boll, Poppinga, & Heuten, 2013). Likewise, a series of LED’sranging from green to yellow to red in color, was used to display (on abracelet) if a person has moved enough (see NikeFuel @Online, 2016).Also, twinkly white lights that would be triggered whenever someone ap-proached them, in order to lure and encourage people to take the stairs(Rogers, Hazlewood, Marshall, Dalton, & Hertrich, 2010). Lighting asan ambient persuasive medium has also been employed in other domainssuch as justice judgments (Van den Bos et al., 2008), time managementduring meetings (Occhialini, Van Essen, & Eggen, 2011).

Ambient lighting used as persuasive technology might be so effective be-

16

✐✐

✐✐

✐✐

✐✐

1.4. Psychological Perspective

cause it needs only little amounts of cognitive attention. Interestingly,Ham and Midden (2010) empirically tested the effect of ambient lightingfeedback through a lamp that could gradually change color (see Figure1.6) dependent on the amount of energy consumption of the partici-pant in a certain task, and compared these effects to a more widely-usedfactual feedback (i.e., numerical feedback in kWh). Ham and Midden(2010) found that participants who processed ambient lighting feedbackabout their energy consumption in a certain task could easily performa second task at the same time, whereas participants who processedfactual energy consumption feedback could not. This research by Hamand Midden (2010) confirmed that ambient lighting persuasive techno-logy can still be effective in situations in which more focal persuasivetechnology (e.g., factual-evaluative feedback) loses its effectiveness.

Figure 1.6: Ambient lighting feedback as used in the research by Ham and Midden (2010).

1.4 Psychological Perspective

To understand the psychological mechanisms underlying the (ambient)persuasive technology, we will discuss different process models that in-fluence people’s attitudes and behaviors in the following sections.

Approaches to persuasion can be ordered according to the amount ofcognitive effort involved in the processes that people focus on (Petty,Cacioppo, & Goldman, 1981). Psychologists often explain and predicthow implementations of influence strategies affect consumer attitudes

17

✐✐

✐✐

✐✐

✐✐


and behaviors using dual-process theories (for an overview see Chaiken& Trope, 1999; Smith & DeCoster, 2000).

In the domain of persuasion, the Elaboration Likelihood Model (ELM)(Petty & Cacioppo, 1986) is a dual-process theory that proposed thatpersuasive messages can affect attitudes through two routes: central andperipheral routes, based on the amounts of thoughtful information pro-cessing or the individual’s subjective elaboration demand. More spe-cifically, the central route is characterized by an individual’s criticallycognitive thinking about task-related arguments and relative merits toforming judgments about the target behavior. On the other hand, theperipheral route is characterized by the individual using simple cues orinferences heuristically in evaluating the target behavior, without cogni-tive thinking (Petty & Cacioppo, 1986; Bhattacherjee & Sanford, 2006).The latter occurs through a more heuristic application.

More recently, the Reflective-Impulsive Model (Strack & Deutsch, 2004)integrated concepts from motivational science into a dual-process sys-tem, and proposed that consumer behavior is controlled by two interact-ing systems that follow different operating principles: reflective systemand impulsive system. In the reflective process, conscious thoughts leadto deliberate decisions. This process requires a high amount of cog-nitive capacity, and therefore, distractions will easily interfere with itsoperation. The impulsive process, in contrast, occurs automatically andrequires little cognitive capacity. The impulsive system is conceived ofas an associative network (see Smith, 1998) and the basis of the im-pulsive process is the spreading of activation of associations. In thisimpulsive system, the associative links are relatively stable and changeonly gradually through learning (e.g., Johnson & Hirst, 1993; McClel-land, McNaughton, & O’Reilly, 1995; Smith & DeCoster, 2000). Asa consequence, the impulsive system has low flexibility but is fast andneeds no attentional resources. For example, if we see a red traffic light,associations, such as, dangerous, warning and negativity, may be acti-

18

✐✐

✐✐

✐✐

✐✐

1.4. Psychological Perspective

vated in the impulsive system. Such associations have previously beenwell learned and will be more readily accessible and guide the subse-quent process. These associative links in the impulsive system will pro-vide a cognitively fast shortcut to influence people’s behavior (e.g., roadusers will stop before the red traffic light). Generally speaking, in thereflective-impulsive model, the reflective system generates behavioral de-cisions that are based on knowledge about facts and values, whereas theimpulsive system elicits behavior through associative links.

From the perspective of dual-process models (e.g., Strack & Deutsch,2004), many persuasive technologies, such as factual feedback (see Sec-tion 1.1), do require users to consciously and critically think aboutenergy-related arguments (e.g., the amount of energy consumption, kWh,see McCalley & Midden, 2002) to form judgments about energy conser-vation behavior. As suggested by earlier research (Bargh, 1997), most ofthe psychological processes that people use for processing external stimuliin day-to-day situations, often are associative in nature, unconscious, fastand automatically activated upon the perception of those stimuli. There-fore, persuasive technologies such as colored ambient lighting feedback(see Ham & Midden, 2010) mainly involve cognitively-efficient processesthrough more associative links in the impulsive system. That is, users canprocess these energy-related arguments (e.g., more red for high energyconsumption, see also Maan et al., 2011) through pre-existing color asso-ciations without needing too much cognitive efforts. More interestingly,processes in the impulsive system may also be accompanied by experien-tial processes (Strack & Deutsch, 2004). These experiential processes arealso associative in nature but involve less cognition (e.g., Dijksterhuis,Bargh, & Miedema, 2000; Zajonc, 1980) compared to associative proc-esses in the impulsive domain of the cognitive system. For example, aperson can change their behavior when he/she feels uncomfortable (i.e.,outside of his/her comfort zone) without necessarily spending cognitiveefforts on processing external stimuli.

19

✐✐

✐✐

✐✐

✐✐


In the current thesis, we will focus on understanding the psychologi-cal mechanisms, especially cognitively-efficient processes, which couldexplain the effectiveness of using ambient lighting in persuasive techno-logy.

1.5 Research Goals

As described at the beginning of this chapter, most types of conven-tional persuasive technologies are only effective if the user spends ampleattention to them. However, in many situations, people might not bemotivated or lack the cognitive capacity to consciously process relativelycomplex information such as factual-evaluated feedback (Bargh, 1997).Cognitive tasks may require information processing that taxes workingmemory. Such tasks may hinder feedback information to be processeddue to limited processing capacity (for a discussion on cognitive capacitysee Smyth, 1994, p.142). Ambient lighting could deliver more sensorypersuasive messages, in which the processes of this type of persuasivetechnology often are associative in nature. These associative processesin the impulsive system have low flexibility but are fast and need lit-tle (or no) attentional resources (Strack & Deutsch, 2004). Furthermore,processes in the impulsive system may also be accompanied by an experi-ential state of awareness. That is, without necessarily knowing its origin,people may experience a feeling with its distinct phenomenal quality. Forexample, a person may have a visual perception of lightness or darkness,a pleasant or unpleasant feeling, or the experience of pain or familiaritywithout knowing the concepts or categories of light, pleasantness, painor familiarity. Thus, the experiential processes in the impulsive systemare associative in nature, but involve less cognition (e.g., Dijksterhuis etal., 2000; Zajonc, 1980).

Therefore, based on previous research and theories, we set two main goalsfor this thesis:

20

✐✐

✐✐

✐✐

✐✐

1.6. Overview of the Chapters

(1) Understanding the associative processes of ambient lightingpersuasive technology. More specifically, we aimed to explore the roleof color associations of ambient lighting for the effectiveness of ambientpersuasive technology (Chapter 2), and of color associations in theircontext (Chapter 3).

(2) Understanding the experiential processes of ambient light-ing persuasive technology. More specifically, we aimed to explore theinfluence of ambient lighting on people’s (thermal) experiences (Chapter4).

1.6 Overview of the Chapters

In Chapter 2, we present two studies that investigate the effects ofcolor associations of ambient lighting feedback on the ease of processingthe feedback messages, and furthermore on energy conservation behavior.More specifically, we argue that colors of ambient lighting feedback couldcarry meanings that have strong and consistent associations with energyconsumption, which could help users easily process feedback messages,and thereby enhance the effectiveness of ambient persuasive technology.An online pre-test was conducted to investigate the strength of diffe-rent color associations (of lighting feedback) with energy consumption.Two studies were conducted to investigate the effects of color associationstrength (strongly-associated vs. weakly-associated) and color associa-tion consistency (consistent vs. inconsistent) on the ease of processingof feedback messages, and furthermore on users’ energy consumptionbehavior.

Thereby, Chapter 2 will help us uncover important underlying mecha-nisms behind the effectiveness of ambient lighting persuasive technology,by studying the effects of color associations of ambient lighting feedbackon ease of processing. Still, we argue that for understanding the roleof color associations, the context in which these colors are presented is

21

✐✐

✐✐

✐✐

✐✐


crucial. That is, associations of color are highly context-dependent asindicated by Color-In-Context theory (Elliot & Maier, 2012).

Therefore, in Chapter 3, we aim to investigate the influence of a color’scontext (the context in which colors were presented as energy feedback)on user’s interpretation of feedback messages, and furthermore, on theeffectiveness of this feedback (for influencing energy conservation be-havior). We argue that the (at least with respect to energy-related)information that color-based ambient persuasive technology carries ishighly context-dependent. More specifically, a fitting context in whichcolor-based ambient persuasive technology is presented could shape theassociations of colors, which makes the feedback messages directly under-standable. Moreover, this context can increase the persuasive power ofambient persuasive technology to promote energy conservation behavior.

For understanding the psychological mechanisms underlying effectivenessof ambient lighting feedback, we study in Chapter 2 and Chapter 3 theassociative processes of feedback messages. These associative processesare one part of the impulsive cognitive processes (Strack & Deutsch,2004). Importantly, processes in the impulsive cognitive system, to somedegree, may also consist of experiential processes (Strack & Deutsch,2004). These experiential processes are also associative in nature butinvolve less cognition (e.g., Dijksterhuis et al., 2000; Zajonc, 1980).

Therefore, in Chapter 4, we extend our understanding of the mecha-nisms behind the effectiveness of ambient lighting persuasive technologyby exploring the experiential process. More specifically, we argue thatthermal experience is one of the main reasons people have for conservingenergy or not (see also Becker et al., 1981; Heijs & Stringer, 1988). There-fore, we proposed that ambient lighting might be used to influence user’sthermal experiences (e.g., by emitting a “warm” or “cold” light) andthereby influence user’s room temperature perception and motivation tochange room temperature. We conducted two studies to investigate theeffects of color temperature of room lighting on participants’ perception

22

✐✐

✐✐

✐✐

✐✐

1.6. Overview of the Chapters

of room temperature, and their motivation to change room temperaturesettings.

In Chapter 5, we first discuss the main findings and contributions ofthe studies presented in the thesis. Then, we identify general limitationsand directions for future research. In addition, we briefly address theethical consideration of using ambient lighting in persuasive technology.Finally, we present the general conclusion of the thesis.

23

✐✐

✐✐

✐✐

✐✐


24

✐✐

✐✐

✐✐

✐✐

2 | The role of color associa-tions

Although earlier research (e.g., Arroyo et al., 2005; Maan et al.,2011; Wilson et al., 2013) showed that ambient lighting serving

as persuasive feedback can be effective for influencing energy conserva-tion behavior, the underlying mechanisms behind this effectiveness re-main unclear. In this chapter, we argue that colors of lighting feedbackcould carry meaning that has pre-existing associations with energy con-sumption. These associations could help users easily process the feedbackmessages, and thereby enhance the effectiveness of ambient persuasivetechnology.

2.1 Introduction

According to the International Energy Outlook 2015, the growth ofglobal primary energy consumption continued to accelerate in 2014 de-

This chapter is largely based on:Lu, S., Ham, J. & Midden, C. (2016). The influence of color association strength and consistencyon ease of processing of ambient lighting feedback. Journal of Environmental Psychology. (Vol. 47,pp. 204-212).Lu, S., Ham, J. & Midden, C. (2014). Using ambient lighting in persuasive communication: Therole of pre-existing color associations. In L.Gamberini, A. Spagnolli, & L. Chittaro (Eds.): PER-SUASIVE 2014, LNCS 8462 (PP. 167-178). Springer International Publishing Switzerland 2014.

25

✐✐

✐✐

✐✐

✐✐

Chapter 2. The role of color associations

spite stagnant global economic growth (Energy Information Administra-tion, 2015). The threats of growing greenhouse gasses and exhaustion ofnatural resources have urged nations worldwide to seek for substantialreduction in energy consumption. Next to the importance of technolo-gical solutions such as efficient systems and renewable energy sources,consumer behavior plays a crucial role in reducing the level of energyconsumption. For instance, an actual 1 ◦C decrease, from 20 ◦C to 19 ◦C,results in heating energy savings of 7% (Briand & Pras, 2010). There-fore, the question of how to promote pro-environmental behavior (e.g.,energy conservation behavior) has become highly relevant in the domainof energy sustainability (Midden & Ham, 2012).

Various interventions and incentives have been used to promote energyconservation behavior (see Abrahamse et al., 2005; Midden et al., 2007).One strategy is embedding feedback in user-system interaction, and suchfeedback interventions have been employed to promote energy conserva-tion behavior. For example, McCalley and Midden (2002) demonstratedin several studies that immediate feedback by adding an energy meter tothe user interface of a washing machine could persuade users to use 21%less energy. Another, more recent, novel smart home system (Jahn etal., 2010) used intuitive user interfaces that could show energy consump-tion data, (e.g. energy price, energy source, standby consumption). Thiskind of feedback allowed users to judge levels of household energy con-sumption and consume energy efficiently. Relatedly, Ham and Midden(2010) proposed another form of interactive feedback by making use ofa social robot. This social robot showed signs of disapproval or negativesocial incentives (e.g., a sad face) as feedback about energy consumption.This form of social feedback (i.e., a persuasive robotic agent) was foundto be able to more effectively induce behavior change among humanusers, compared to the effects of interactive factual-evaluative feedback(directly indicating the amount of kWh, see Ham & Midden, 2010).

However, some forms of persuasive technology that were designed to

26

✐✐

✐✐

✐✐

✐✐

2.1. Introduction

change people’s attitudes or behaviors in a predetermined way might notbe very practical for many day-to-day situations. For example, factualfeedback messages may be relatively complex and loose their persuasivepower in situations that require high cognitive capacity. Likewise, socialfeedback requires user’s focal attention because users have to focus onthe feedback messages consciously. So, in daily situations, people mightlack the focal attention or cognitive capacity to consciously process thesefeedback information (Bargh & Williams, 2006).

Concerning a form of feedback that could be less sensitive to a lack offocal attention and cognitive capacity, earlier research (e.g., Maan et al.,2011) proposed to use another form of feedback that was easier to pro-cess: ambient feedback. Ishii and Ullmer (1997) suggested the use ofambient media such as light, airflow, and sound to act as backgroundinfluences and work at the periphery of human perception. The conceptof ambient persuasive technology was investigated by research (see e.g.,E. Aarts, Markopoulos, & De Ruyter, 2007; Davis, 2008; Ham, Midden,& Beute, 2009; Martinez & Geltz, 2005; Schmidt, 2005; Wisneski et al.,1998). According to earlier research, ambient persuasive technology wasshown to be effective while being provided at the “periphery of our atten-tion”(Wisneski et al., 1998, p.22), “without conscious attention”(Ham etal., 2009, p.5).

In line with the proposed concept of ambient persuasive technology, Ar-royo and colleagues (2005) presented ambient feedback through provid-ing colored lighting reflecting the temperature of the water without al-tering the function of the sink. For example, one of their designs wascalled HeatSink and it consisted of colored LEDs mounted around thefaucet aerator that could illuminate the stream of water with a red lightwhen water was hot, and blue when water was cold. This HeatSinkcould inform subtle behavior change regarding water use while increas-ing the functionality of the sink. Also, a case study by Wilson and col-leagues (2013) suggested that effective ambient feedback could be pro-

27

✐✐

✐✐

✐✐

✐✐


vided through light and sound on the status of the heating system intandem with the status of their windows, so to convey the energy conse-quences of their behavior directly. For example, if a window is opened intandem with a detected increase in radiator surface temperature, the col-ored light corresponding to temperature immediately displays a warninglight (e.g., red) to indicate waste.

Feedback given through this kind of ambient persuasive technology canhelp users in those day-to-day situations, in which users lack cognitivecapacity or focal attention to process feedback information. In line withearlier research (Petty, Cacioppo, & Kasmer, 2015; Strack & Deutsch,2004), the processing of this more simple information, as provided byambient feedback, requires less cognitive capacity and attentional re-sources compared to the processing of complex information, as providedby factual feedback.

Recent research indeed confirmed that ambient persuasive technologycan still be effective in situations in which more focal persuasive techno-logy (e.g., factual-evaluative feedback as described above) loses its effec-tiveness. That is, research by Maan and colleagues (2011) explored thefundamental characteristics of ambient feedback. This research testedthe effect of feedback through a lamp that could gradually change colordependent on the amount of energy consumption of the participant in acertain task, and compared these effects to a more widely-used factualfeedback (i.e., numerical feedback in kWh). Results indicated that feed-back through ambient lighting was more effective for influencing energyconservation behavior than numerical feedback. In addition, processingambient lighting feedback seemed easier in the sense that performing anadditional cognitive load task did not interfere with the feedback. How-ever, the underlying mechanisms which could explain the effectiveness ofprocessing ambient feedback messages remain largely unclear.

In the current research, we explore the role of associations. We arguethat colors of lighting feedback could carry meaning that has pre-existing

28

✐✐

✐✐

✐✐

✐✐

2.1. Introduction

associations with the target behavior (i.e., energy conservation behaviorin the current research). These associations could help users easily un-derstand the feedback messages, and thereby, increase the persuasiveeffectiveness of ambient lighting feedback. For instance, the color pair,red vs. green (as used in Maan et al., 2011), indicated higher energyconsumption levels for shades of red and shades of green indicated lowerconsumption levels. It seems that these pre-existing associations of col-ored lighting feedback lead to less energy consumption compared to thenumerical feedback that lacks these associations. However, this researchdid not study the content of the associations and the nature of the asso-ciative processes.

Concerning the importance of associations, earlier research has shownthat associations, especially color associations, can be very important inobject recognition tasks, like detecting a target object (e.g., animal) ineither color or gray-scale natural scenes (Otsuka & Kawaguchi, 2009).These color associations can help people to easily detect the target objectfor the reason that color information in the image is similar to the dailyvision, and linking to pre-existing associations.

Therefore, the main aim of the current research is to test whether lightingfeedback with colors that have pre-existing associations with energy sav-ing as target behavior, is easier to process compared to lighting feedbackwith colors that do not have these associations. We argue that lightingfeedback with colors that have pre-existing associations with energy con-sumption could help users easily process the feedback messages. Suchassociative processes are less dependent of user’s cognitive capacity. Incontrast, when lighting feedback uses colors lacking these pre-existing as-sociations, it would be a lot more difficult to learn whether the messageindicates high or low energy consumption.

In addition to the effect of associations on the ease of processing feedbackmessages, we are also interested in exploring the effect of associations onenergy consumption behavior. As an exploratory test, we added energy

29

✐✐

✐✐

✐✐

✐✐


consumption scores as our second dependent variable. We argue thatthese pre-existing associations of colored lighting feedback could promoteusers to consume less energy than the lighting feedback with colors thatlack these associations.

2.2 Study 2.1In Study 2.1, preceding the lab experiment, an online survey was con-ducted to explore the strength of different color associations with energyconsumption. Fifty-two participants (average age 23.0 years old, SD =.40) were invited to fill in a questionnaire about the association strengthof 21 color pairs (e.g., red vs. purple; yellow vs. blue). For each colorpair, we posed the question “Please indicate whether you think a fol-lowing specific color pair is strongly associated with high vs. low energyconsumption?” Results of this survey showed that participants reportedthe color pair, red vs. green, to be the most strongly associated withenergy consumption, and the color pair, yellow vs. purple, to be themost weakly associated with energy consumption.Therefore, following this online survey, we defined two types of light-ing feedback: strongly-associated vs. weakly-associated lighting feed-back (see Table 2.1). The strongly-associated lighting feedback couldgradually change colors between red and green dependent on heatingenergy consumption through an ambient LED wall washer. Likewise,the weakly-associated lighting feedback could gradually change colorsbetween yellow and purple. Using these two types of colored ambientlighting as energy feedback, a lab experiment was set up in which par-ticipants had the opportunity to conserve energy in a series of tasks (acomparable thermostat task that was successfully used in earlier research,see Maan et al., 2011) while receiving the lighting feedback.First of all, to test whether strongly-associated lighting feedback waseasier to process than weakly-associated lighting feedback, we manipu-

30

✐✐

✐✐

✐✐

✐✐

2.2. Study 2.1

Table 2.1: The colors used as strongly-associated feedback and weakly-associated feedbackin Study 2.1.

strongly-associated feedback weakly-associated feedbackred = high*green = low

yellow = highpurple = low

*high = high energy consumption level; and low = low energy consumption level

lated the cognitive load of participants. To manipulate cognitive load,we distracted half of the participants while they were performing theenergy conservation task. For this, we used an additional cognitive loadtask consisting of a random number recognition task. The cognitive loadtask was comparable to the successful cognitive load manipulation usedin the research by Marsh and Hicks (1998). We argued that those par-ticipants who were requested to perform this additional cognitive loadtask, a lack of strong associations of lighting feedback could interferewith the processing of feedback messages, leading to a longer completiontime of the energy conservation task. Therefore, we expected an interac-tion effect (H2.1.1) of color association strength × cognitive load on taskcompletion time. That is, for participants who received lighting feedbackwith colors that had weak associations, performing the additional cog-nitive load task would cause slower processing of the feedback messages(i.e., resulting in longer programming time of the thermostat task). Incontrast, for participants who received lighting feedback with colors thathad strong associations, performing this additional cognitive load taskwould not increase the processing time of those feedback messages.

Second, as an exploratory test, we argued that lighting feedback withcolors that had strong energy associations could help users correctly in-terpret feedback messages, and thereby promote users to consume lessenergy. Compared to the main manipulation of feedback type (light-ing feedback vs. factual feedback) in research by Maan and Colleagues(2011), we had a comparable manipulation of feedback type by usingstrongly-associated vs. weakly-associated lighting feedback. Therefore,

31

✐✐

✐✐

✐✐

✐✐


in line with their findings, we expected the main effect (H2.1.2) of colorassociation strength on energy consumption behavior. That is, strongly-associated lighting feedback would have stronger persuasive effects (lead-ing to lower energy consumption) than weakly-associated lighting feed-back. In other words, participants who received strongly-associated light-ing feedback would consume less energy in the thermostat task thanparticipants who received weakly-associated lighting feedback. In linewith the findings by Maan and colleagues (2011), we did not expect theinteraction effect on energy consumption behavior in this experimentalsetup.

2.2.1 Method

Participants and Design

Ninety-nine native Dutch participants (54 male and 45 female) took partin this experiment, with an average age of 23.4 years old (SD = 4.97). Ofthose, most of them were students from Eindhoven University of Tech-nology. We invited only participants who were not color blind. All ofthem were randomly assigned to one of four experimental conditions:2 (color association strength: strongly-associated vs. weakly-associatedwith energy consumption) × 2 (cognitive load: low vs. high). The experi-ment lasted 30 minutes. Participants were paid e 5 for their participationat the end of the session.

Materials and Procedures

All the tasks we used in this experiment were fully computerized, andthe lighting feedback was provided via the wall washers (LEDs that illu-minate the side walls). Participants could perceive the lighting feedbackat the periphery of their perception while paying focal attention to ascreen-based task (see Figure 2.1).

32

✐✐

✐✐

✐✐

✐✐

2.2. Study 2.1

(a) Red (b) Green (c) Yellow (d) Purple

Figure 2.1: Programmable LED wall washers on the side walls indicating (a) high and (b)low energy consumption levels in strongly-associated lighting feedback condition; (c) highand (d) low energy consumption level in weakly-associated lighting feedback condition.

Upon arrival, each participant was asked to read and sign an informedconsent form stating the general purpose of the research and their will-ingness to participate in this research. The procedure of the experimentwas visualized in Figure 2.2. First, participants were requested to sitin front of the computer, on which an LCD thermostat was presented.This simulated programmable thermostat panel (see Figure 2.3), a modelof an available central heating interface in Dutch households, was usedto collect data about participants’ energy consumption scores while per-forming virtual temperature-setting tasks for several scenarios (e.g., “It isafternoon and you are not at home. The outside temperature is 20 ◦C”).For each scenario, participants were asked to strive for two goals: (1)to consume as little energy as possible for heating the room, and (2) tomaintain a temperature level that they deemed as comfortable as possi-ble.

Introductiontothethermostatpanel

0min 10

2practicetrials

10experimentaltrials

15 25 30min

Demographicquestion

Introductiontolightingfeedbackfeedbacktask

5

Figure 2.2: The timeline of the experiment in Study 2.1.

After the introduction to the simulated programmable thermostat paneland the lighting feedback (the only feedback in the experiment), parti-

33

✐✐

✐✐

✐✐

✐✐


cipants had to complete two practice trials. The ten experimental trialsthat followed were similar to the practice trials, with scenarios varyingoutside temperature from −5 ◦C to 19 ◦C, varying daily periods, and pres-ence or absence of the user. When participants hit the “finish” button foreach scenario, we registered the energy consumption scores correspond-ing to the final setting, and the total amount of time a participant usedfor that scenario.

Figure 2.3: A simulated programmable heating system interface used to gather data aboutthe participant’s energy consumption behavior in Study 2.1.

For the cognitive load manipulation, only participants in the high loadcondition performed an additional cognitive load task while setting thethermostat. This cognitive task was comparable to the successful cog-nitive load manipulation in earlier research (Marsh & Hicks, 1998), andwas presented during all ten experimental scenarios. Participants hearda random series of recorded spoken numbers from 1 to 30 (in Dutch)with a 3 seconds interval. We registered the number of correct responses(pressing the space bar whenever they heard an odd number).

The temperatures that participants set during each scenario were used todetermine the lighting colors. In the strongly-associated lighting feedbackcondition, the LED light was given a red color in the peripheral environ-

34

✐✐

✐✐

✐✐

✐✐

2.2. Study 2.1

ment when energy consumption (based on the temperatures participantsset on the programmable thermostat) was at a high level, and a greencolor when at a low level. When a participant’s thermostat settings leadto energy consumption between the high and low level, the peripherallighting was set to a color between green and red in different saturationwith the same brightness level. White was set as the neutral color cor-responding to medium energy consumption level in this experiment (seeFigure 2.4). In the weakly-associated lighting feedback condition, yellowlighting represented high energy consumption and purple represented lowenergy consumption.

… … … … … …

VeryhighEnergyconsumptionlevel:

Medium Verylow

Figure 2.4: The link between energy consumption level and the color of strongly-associatedlighting feedback (color can continuously change according to the thermostat setting).

Finally, participants answered several demographic questions, were de-briefed and thanked for their participation.

We constructed two main dependent variables: (1) task completion time,by averaging the times a participant spent on each of the 10 scenarios,and (2) energy consumption scores per scenario, by using the energyconsumption scores corresponding to the final setting of each scenario.

2.2.2 Results

To test whether lighting feedback with colors that had strong associationswith energy consumption was easier to process than lighting feedbackthat used colors with weak associations with energy consumption, weanalyzed the average task completion time per scenario and submittedthese to a 2 (color association strength: strongly-associated vs. weakly-associated) × 2 (cognitive load: low vs. high) between-subject ANOVA.

35

✐✐

✐✐

✐✐

✐✐


As advised by earlier research (Ratcliff, 1993), the response time shouldexclude outliers of more than two SDs from the average response time.Therefore, we reported our analysis of the variable task completion timewith and without these (six out of 99) participants who scored outliers.

Supporting our first hypothesis (H2.1.1), results showed the expected in-teraction effect of color association strength × cognitive load on task com-pletion time, F (1, 89) = 4.33, p = .04, η2 = .046 (see Figure 2.5). Thatis, participants receiving weakly-associated lighting feedback spent moretime to program the thermostat task while performing an additionalcognitive load task (M = 39.5s, SD = 13.4) compared to participantsreceiving weakly-associated lighting feedback while not performing thisadditional cognitive load task (M = 30.2s, SD = 5.4), F (1, 89) = 8.72,p = .004, η2 = .089. In contrast and as expected, this difference wasnot found between participants who received strongly-associated lightingfeedback in the high cognitive load condition (M = 35.3s, SD = 8.65)and participants receiving strongly-associated lighting feedback in thelow cognitive load condition (M = 35.0s, SD = 12.5), F (1, 89) = 0.01,p = .92, η2 < .0011.

To explore the effects of color associations of lighting feedback on energyconsumption behavior, the 10 energy consumption scores (correspond-ing to the final setting of each of the 10 scenarios) of one participantwere submitted to the a 2 (color association strength: strongly-associatedvs. weakly-associated) × 2 (cognitive load: low vs. high) × 10 (scenarios)repeated-measures ANOVA, with scenarios as the repeated factor. In

1When we included the six outliers on the measure for task completion time, the interactioneffect was comparable, F (1, 95) = 3.76, p = .055, η2 = .038. More specifically, participants receivingweakly-associated lighting feedback needed more time to program the thermostat in the high cogni-tive load condition (M = 44.3s, SD = 16.4) compared to participants receiving weakly-associatedlighting feedback in the low cognitive load condition (M = 30.2s, SD = 5.4), F (1, 95) = 13.89,p < .001, η2 = .128. While, this difference was not found between participants receiving strongly-associated lighting feedback in the high cognitive load condition (M = 38.9s, SD = 15.2) and partic-ipant receiving strongly-associated lighting feedback in the low cognitive load condition (M = 35.0s,SD = 12.5), F (1, 95) = 1.11, p = .29, η2 = .012.

36

✐✐

✐✐

✐✐

✐✐

2.2. Study 2.1

0

10

20

30

40

50

taskco

mpletiontim

e(s)

strongly-asscociated weakly-associatedColorassociationstrength

lowcognitiveload highcognitiveload

Figure 2.5: Averaged task completion time as a result of color association strength (strongly-associated vs. weakly-associated) and cognitive load (low vs. high) in Study 2.1. Error barsattached to each column in the figure indicate 95% confidence intervals.

line with our second hypothesis (H2.1.2), results showed that partici-pants who received lighting feedback with colors that had strong as-sociations with energy consumption used less energy (EMM = 646.8,SE = 27.8) than participants who received lighting feedback with colorsthat have weak associations with energy consumption (EMM = 727.0,SD = 28.7), F (1, 95) = 4.03, p = .048, η2 = .041 (see Figure 2.6).

0

100

200

300

400

500

600

700

800

averagee

nergyc

onsumptions

core

(kWh)


Figure 2.6: Averaged energy consumption score as a result of color association strength(strongly-associated vs. weakly-associated) in Study 2.1. Error bars attached to each columnin the figure indicate 95% confidence intervals.

37

✐✐

✐✐

✐✐

✐✐


2.2.3 Discussion

As expected, results of Study 2.1 showed that only participants whoreceived lighting feedback with colors that had strong energy associa-tions, processed the feedback messages independent of being distractedor not (i.e., participants were requested to perform an additional cogni-tive load task while programming the thermostat task). In other words,for participants who received lighting feedback with colors that lackedthese strong associations, it was more difficult to process the energy con-sumption feedback messages, and thereby this additional cognitive loadtask hindered their processing time of feedback messages. Also, resultsrevealed that these strong associations lead to a stronger persuasive ef-fectiveness of this lighting feedback than the weak associations of lightingfeedback, resulting in less energy consumption in the thermostat task.

2.3 Study 2.2

To extend our insight in the processing of color associations, we de-signed a second study to explore the other factor of color associations,in which we added lighting feedback with colors that were inconsistentwith pre-existing energy associations (e.g., more red for lower energy con-sumption, and more green for higher energy consumption). In Study 2.2,we created a second color association factor, consistency, which allowedfor independently manipulating color association consistency and colorassociation strength (as used in Study 2.1). More specifically, we cre-ated inconsistent versions of strongly-associated and weakly-associatedlighting feedback. In these inconsistent versions of lighting feedback,we reversed the meaning of colors to indicate the energy consumptionlevel. That is, for the inconsistent strongly-associated lighting feedback,users received red light when their energy consumption was low, andgreen light when their energy consumption was high. For the incon-

38

✐✐

✐✐

✐✐

✐✐

2.3. Study 2.2

sistent weakly-associated lighting feedback, users received yellow lightwhen their energy consumption was low and purple light when theirenergy consumption was high.

Different from the first study, in Study 2.2, all participants were re-quested to perform an additional cognitive load task while the colorassociation strength and the color association consistency were manip-ulated between participants. Also, to improve the behavioral measure,in Study 2.2, we added a more sensitive measure of energy consumptionbehavior by collecting the energy consumption scores based on partici-pants’ actions during the whole thermostat task (as suggested in earlierresearch by Ruijten, Midden, & Ham, 2015). However, in Study 2.1, weonly collected the energy consumption scores corresponding to the finalsetting of each scenario.

Results of Study 2.1 showed that lighting feedback that had colors withstrong energy associations could help users to easily process feedbackmessages (e.g., colors indicated high or low energy consumption), andthereby, resulting in lower energy consumption in the thermostat task.Besides the color association strength (i.e., strong vs. weak associations),we argued that lighting feedback would be more effective when colorsare consistent with pre-existing associations compared to colors whichwere inconsistent with pre-existing associations. For instance, for thesame color pair that was used to have inconsistent associations withpre-existing associations (see Table 2.2), we explored whether strongly-associated lighting feedback could still help users easily process thesefeedback messages.

In Study 2.2, our main aim is to investigate the effects of color associ-ation consistency of lighting feedback on the ease of processing of feed-back messages. We argued that the consistency of color’s strong energyassociations could help users to easily process the feedback messages(e.g., whether that color indicated high or low energy consumption), andthereby, increase the persuasive effectiveness of this lighting feedback

39

✐✐

✐✐

✐✐

✐✐


Table 2.2: The colors used for two factors of color associations (color association strengthand color association consistency) in Study 2.2.

color association consistencyconsistent inconsistent

color association strength strongly-associated red = high*green = low

green = highred = low

weakly-associated yellow = highpurple = low

purple = highyellow = low

*high = high energy consumption level; and low = low energy consumption level

(for influencing energy conservation behaviors). In other words, lightingfeedback with colors that lacked consistent and strong energy associa-tions may slow down users’ processing of feedback messages. As arguedabove, we expected the interaction effect (H2.2.1) of color associationstrength and color association consistency on the task completion time.That is, lighting feedback with colors that were inconsistent with strongenergy associations would interfere with users’ processing of feedbackmessages, resulting in an increase of time for the thermostat program-ming task. However, for colors with weak energy associations, reversingthe meaning of colors (i.e., more purple for either lower or higher energyconsumption) would moderate users processing of these feedback mes-sages to a smaller extent.

To explore the effects of color association consistency on energy consump-tion behavior, we included the energy consumption scores correspondingto the final setting of each scenario (the same measure as used in Study2.1) and the dynamic energy consumption scores after each action (themore sensitive measure added in Study 2.2) as two dependent variables.

Therefore, we expected the main effect (H2.2.2) of color association con-sistency on energy consumption scores per scenario. That is, inconsi-stent versions of lighting feedback would decrease the persuasive powerof lighting feedback, resulting in consuming more energy per each taskscenario. Additionally, as discussed above, in order to detect the rel-atively small and indirect behavior effects, we added a more sensitive

40

✐✐

✐✐

✐✐

✐✐

2.3. Study 2.2

measure of energy consumption behavior by collecting the energy con-sumption scores based on participants’ actions during the whole thermo-stat task. Therefore, we also expected (H2.2.3) that participants wouldreduce more energy consumption per action during the thermostat taskwhen receiving consistent lighting feedback compared to participants re-ceiving inconsistent lighting feedback.

2.3.1 Method


One-hundred-and-one native Dutch participants (69 male and 32 female)were recruited by using a local database of participants. We invited onlyparticipants who had not taken part in our Study 2.1 and who were notcolor blind. All of them were randomly assigned to one of four experi-mental conditions: 2 (color association strength: strongly-associated vs.weakly-associated with energy consumption) × 2 (color association con-sistency: consistent vs. inconsistent). This experiment also lasted 30minutes. Participants were paid e 5 for their participation at the end ofthe session.

Materials and Procedures

The experimental procedure of the current study was completely identi-cal to the experimental procedure of Study 2.1, with the exception thatall participants in the current experiment have to perform an additionalcognitive load task (whereas only half of the participants have to performthis task in Study 2.1). Moreover, we also collected energy consumptionscores based on the difference of energy consumption between every set-ting and the previous setting (caused by an action performed by theparticipant, e.g., increasing the thermostat temperature). This valueshowed the immediate effect of the lighting feedback on participants’

41

✐✐

✐✐

✐✐

✐✐


energy use. A positive energy use indicated that a participant used moreenergy than in the previous setting, and a negative energy use indicatedthat the participant used less energy compared to the previous setting.

We constructed three main dependent variables: (1) task completiontime, by averaging the times a participant needed on each of the 10 sce-narios, (2) energy consumption scores per scenario, by using the energyconsumption scores corresponding to the final setting of each scenario(the identical measure of energy scores as used in Study 2.1), and (3) thechange of energy consumption scores per action, by using the differenceof energy consumption scores between every setting and the previoussetting.

2.3.2 Results

To investigate the interaction effects of the color association strengthand consistency of lighting feedback on the processing time of feedbackmessages, we submitted the average task completion time per scenario toa 2 (color association strength: strongly vs. weakly) × 2 (color associationconsistency: consistent vs. inconsistent) between-subject ANOVA. As inStudy 2.1, we reported our analysis of the variable task completion timewith and without six (out of 101) outliers (more than two SDs from theaverage task completion time, the same as in Study 2.1, Ratcliff, 1993).In line with our first hypothesis (H2.2.1), results showed the interactioneffect on task completion time, F (1, 91) = 9.90, p = .002, η2 = .098 (seeFigure 2.7). That is, only participants receiving lighting feedback withcolors that are inconsistent with strong energy associations, spent moretime on programming the thermostat task (M = 46.7s, SD = 12.0) thanparticipants receiving lighting feedback with colors that are consistentwith strong energy consumption (M = 37.9s, SD = 9.6), F (1, 91) =8.06, p = .005, η2 = .083. In contrast and as expected, this differencewas not found for the participants who received consistent or inconsistent

42

✐✐

✐✐

✐✐

✐✐

2.3. Study 2.2

feedback with colors that have weak energy associations, F (1, 91) = 2.49,p = .118, η2 = .027. In other words, for colors that lacked strong energyassociations, reversing the meaning of the two colors did not moderateusers processing time of these feedback messages2.

0

10

20

30

40

50

taskco

mpletiont

ime(

s)


consistent inconsistent

Figure 2.7: Averaged task completion time as a result of color association strength (strongly-associated vs. weakly-associated) and color association consistency (consistent vs. inconsi-stent) in Study 2.2. Error bars attached to each column in the figure indicate 95% confidenceintervals.

To explore the effects of the color association consistency of lightingfeedback on energy consumption behavior, we analyzed two dependentvariables, respectively. First, the 10 energy consumption scores of oneparticipant (over the 10 scenarios) were submitted to a 2 (color asso-ciation strength: strongly-associated vs. weakly-associated) × 2 (colorassociation consistency: consistent vs. inconsistent) × 10 (scenarios)repeated-measures ANOVA, with scenarios as a repeated factor. Notsupporting our second hypothesis (H2.2.2), results indicated that parti-cipants receiving consistent lighting feedback did not consume less (ormore) energy in the thermostat programming task (i.e., average energy

2When we included the six outliers on the measure for task completion time, the interaction effectwas comparable, F (1, 97) = 3.76, p = .056, η2 = .037. More specifically, only participants receivinglighting feedback with colors that are inconsistent with strong energy associations, spent more timeon programming the thermostat task (M = 47.9s, SD = 14.9) than participants receiving lightingfeedback with colors that are consistent with strong energy consumption (M = 39.4s, SD = 12.0),F (1, 97) = 5.38, p = .022, η2 = .053. However, this difference was not found for the participantswho received consistent or inconsistent feedback with colors that have weak energy associations,F (1, 97) = .18, p = .67, η2 = .002.

43

✐✐

✐✐

✐✐

✐✐


consumption scores, EMM = 666.3, SE = 29.2) than participants re-ceiving inconsistent lighting feedback (EMM = 705.4, SE = 28.6),F (1, 94) = .91, p = .34, η2 = .010 (see Figure 2.8).

0

100

200

300

400

500

600

700

800

averagee

nergyc

onsumption

score(

kWh)

consistent inconsistentColorassociationconsistency

Figure 2.8: Averaged energy consumption score as a result of color association consistency(consistent vs. inconsistent) in Study 2.2. Error bars attached to each column in the figureindicate 95% confidence intervals.

Second, to explore the behavior effects, in the current study, we addeda more sensitive measure of energy consumption behavior. Therefore,the change of energy consumption scores per action3 were submittedto a Linear Mixed-Effects Model with a 2 (color association strength:strongly-associated vs. weakly-associated with energy consumption) ×2 (color association consistency: consistent vs. inconsistent) between-subjects design with the specific thermostat task scenario as the randomfactor. Supporting our third hypothesis (H2.2.3), participants receiv-ing consistent lighting feedback decreased more energy per each action(EMM = −14.2, SE = 2.11) than participants receiving inconsistentlighting feedback (EMM = −10.1, SE = 1.96), F (1, 4623.83) = 8.48,p = .004 (see Figure 2.9).

3change of energy consumption scores per each action = the current-action energy consumptionscore − the previous-action energy consumption score.

44

✐✐

✐✐

✐✐

✐✐

2.3. Study 2.2

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

thec

hang

eofene

rgyc

onsumption

scores(kWh)

consistentinconsistentColorassociationconsistency

Figure 2.9: The change of energy consumption scores per action as a result of color asso-ciation consistency (consistent vs. inconsistent) in Study 2.2. Error bars attached to eachcolumn in the figure indicate 95% confidence intervals.

2.3.3 Discussion

The main aim of Study 2.2 was to investigate the effects of the colorassociation consistency of lighting feedback on the ease of processing offeedback messages. In line with our hypotheses, results showed that forparticipants who received the inconsistent version of strongly-associatedlighting feedback, more time was needed to process the feedback mes-sages compared to participants who received the consistent version ofstrongly-associated lighting feedback. However, when participants pro-cessed feedback messages without strong pre-existing color associationsto energy consumption, reversing the meaning of the two colors (e.g.,either yellow indicates high or low energy consumption level) did notmoderate users processing time of these feedback messages.

Although the results of Study 2.2 did not show the main effects on energyconsumption scores at a trial level, results of this study did show theexpected main effect on energy behavioral change, that is, the consistentversion of lighting feedback caused more energy behavioral changes thanthe inconsistent version of lighting feedback.

The evidence of the main effects on energy behavioral change, insteadof on energy consumption scores per trial, might be due to various rea-

45

✐✐

✐✐

✐✐

✐✐


sons. Most importantly, compared to the energy consumption scores pertrial (corresponding to the final setting of each scenario), the sensitivebehavioral measure by collecting energy consumption scores per actionincluded large parts of causes for the variation in energy consumptionscores during a given period, such as between two actions. Therefore, itwas not surprising the change of energy consumption scores was moresensitive in detecting these main effects, compared to the energy con-sumption scores at the trial level.

2.4 General discussion

The aim of this research was to empirically test the psychological mecha-nisms of color associations behind the effective ambient lighting feedback(as used in Maan et al., 2011) on energy conservation behavior. In thisresearch, we argued that colored lighting feedback could carry mean-ing that had pre-existing associations with energy consumption. Thesestrong and consistent associations could help users easily process thefeedback messages, and thereby enhance effectiveness of ambient persua-sive technology. In contrast, for colors that had weak energy associationsor colors that were inconsistent with these strong energy associations,such lighting feedback could hinder users in easily processing the feed-back messages (as indicated by slower processing of feedback messages)and result in a decrease of the effectiveness of this lighting feedback.Therefore, two studies were conducted to test these effects. In the firststudy, to investigate the effects of color associations of lighting feedbackon the ease of processing of the feedback messages, we manipulated colorassociation strength (strongly-associated vs. weakly-associated) and cog-nitive load (high vs. low) between participants. In line with our expec-tations, results of Study 2.1 showed that only for participants receivinglighting feedback with colors that have strong associations with energyconsumption, additional cognitive load tasks, i.e., in the high cognitive

46

✐✐

✐✐

✐✐

✐✐

2.4. General discussion

load, did not lead to a slower processing of that feedback messages. Inother words, these strong associations of lighting feedback indeed helpedusers to easily process the feedback messages even with limited cognitivecapacity. Furthermore, results of Study 2.1 found that participants whoreceived this strongly-associated lighting feedback indeed consumed lessenergy in the thermostat task than participants who received weakly-associated lighting feedback. These findings indicated that such strongassociations could help users easily process the feedback messages, andalso increase the persuasive power of this ambient lighting persuasivetechnology.

To extend our insight in the role of color associations in the ease of pro-cessing of lighting feedback, in Study 2.2, we added a second color associ-ation factor, consistency, and investigated the effects of lighting feedbackwith colors that were inconsistent with strong energy associations (i.e.,the inconsistent version of strongly-associated lighting feedback could forexample be: more red for lower energy consumption, and more green forhigher energy consumption). We argued that such inconsistent lightingfeedback would hinder users in easily processing the feedback messages,and thereby the effectiveness of lighting feedback with colors that hadinconsistent associations would be low. In line with our expectations, re-sults of Study 2.2 showed that participants who received the inconsistentversion of strongly-associated lighting feedback processed the feedbackmessages more slowly (i.e., longer programming time of the thermostattask) compared to participants who received the consistent version ofstrongly-associated lighting feedback. However, when using feedbackwithout these strong color associations to energy consumption, reversingthe meaning of the two colors did not moderate users processing timeof these feedback messages. That is to say, without these consistentand strong energy associations of lighting feedback, the processing timeof feedback messages of such lighting feedback would be slowed down.Moreover, Study 2.2 found that participants who received consistent ver-

47

✐✐

✐✐

✐✐

✐✐


sions of lighting feedback consumed less energy in the thermostat taskthan participants who received inconsistent versions of lighting feedback.

So, both studies present evidence that ambient persuasive technologythat uses colors with strong and consistent energy associations can ef-fectively influence user energy consumption behavior. Study 2.1 showedthis main effect on energy consumption behavior in the settings of par-ticipants at the end of a scenario, in the thermostat task. In addition,Study 2.2 showed this energy effect in each separate temperature settingthat participants chose, right after participants received lighting feed-back. Still, Study 2.2 presented no evidence for an influence of colorassociations (of lighting feedback) on the settings of participants at thetrial level that was found in Study 2.1). This might be explained, firstof all, by the difference between the manipulations of color associationsin Study 2.1 and 1.2. That is, these two studies used different mainindependent variables, which were color association strength in Study2.1 and color association consistency in Study 2.2. These two differentmanipulations of color associations had different effect sizes. So, it isstraightforward that these two main effects are not comparable acrossstudies. Second, we argue that in Study 2.2, all participants were cogni-tively busy (i.e., they were requested to perform an extra cognitive loadtask) and thereby the additional cognitive load of the (in)consistencymanipulation could only have a small effect. More specifically, all par-ticipants in Study 2.2 were under cognitive load while only half of theparticipants in Study 2.1 were under cognitive load. Therefore, we arguethat the additional cognitive load (caused by the inconsistency of light-ing feedback in Study 2.2) could evoke a smaller effect than the addi-tional cognitive load (caused by the color association strength of lightingfeedback in Study 2.1). Both our arguments could be explaining whyonly the more sensitive measure (assessing energy consumption behaviorright after a participant received lighting feedback) was able to showan effect of the color association manipulation. Future research could

48

✐✐

✐✐

✐✐

✐✐


compare these behavioral effects of color association consistency whenparticipants have enough cognitive capacity (e.g., without performingthe additional cognitive load task) to process these feedback messages.

As we expected, the consistent version of red vs. green lighting feed-back has a strong pre-existing association with energy consumption andfurthermore leads to lower energy consumption in the task. However,besides the ease of processing the feedback messages, avoiding red am-bient lighting could be an additional motivating factor for lower energyconsumption. As suggested by Elliot and colleagues (2007), red is alsoassociated with the danger of failure in achievement contexts and evokesavoidance motivation. So for participants who even do not understandthe pre-existing color associations could still reach to a good energy con-servation level by simply avoiding red ambient lighting. Although thisadditional motivator seems an advantage of red vs. green lighting feed-back, the importance of the role of color associations with energy con-sumption in ambient lighting feedback that we concluded in this researchmay decline.

The present research shows that lighting feedback with colors that haveconsistent and strong associations with energy consumption can enhancethe persuasive effectiveness of the feedback. However, the role of colorassociations is only one of the various features of successful feedback. Asillustrated by earlier research (Fischer, 2008; Roberts & Baker, 2003), rel-evant features of feedback that may determine its effectiveness included:frequency, duration, content, breakdown, medium and way of presenta-tion, comparisons, and combination with other instruments, etc. Furtherresearch could improve ambient lighting feedback by for example project-ing colored text-shaped lighting instead of colored lighting in the ambientenvironment, and combining more features in a more appealing way.

This current research shows the persuasive effectiveness of using ambi-ent lighting as feedback in the domain of energy conservation. However,we did not test the energy consumption of this ambient lighting system

49

✐✐

✐✐

✐✐

✐✐


itself. This could be a task for future research. Given the availability ofhighly efficient LED technology, ample opportunities to provide lightingfeedback on available technical systems, and possibilities to scaling downthe size of the ambient environment, applying ambient lighting in prac-tice plausibly could be effective as well as energy efficient. For example,instead of illuminating complete side walls to give lighting feedback, fu-ture research could also explore more energy-efficient lighting feedbackby using small LED lamps mounted surrounding the thermostat panelor a (computer) screen to provide lighting feedback (e.g., like AmbilightTV).

There are several limitations to the studies reported in this manuscript.An important potential limitation is that, in both studies, participantsperformed a virtual thermostat task without the physical room temper-ature being adapted. Earlier research (McCalley & Midden, 2002) in-dicated comparable energy saving results (washing machine tasks) afterreceiving interactive feedback in a field study as in a lab study. How-ever, thermostat tasks involve physical temperature comfort experiences,and this measurement of energy conservation behavior is more sensi-tive in a field setting. For instance, people who feel uncomfortably coldwould very likely not decrease the room temperature to save more heat-ing energy at home. In line with earlier research, comfort experienceoften has a dominant influence on people’s (energy consumption) beha-vior (e.g., Becker, Seligman, Fazio, & Darley, 1981; Heijs & Stringer,1988). Moreover, discomfort might be one of the reasons for not savingnatural resources (see Merkus, 2012; Ohnaka, Tochihara, & Watanabe,1994). Although participants in the current research were requested toset room temperature as comfortable as possible in a given scenario,their comfort experience would not be increased by the desired comfort-able room temperature because no physical room temperature changed.This gap between the desired high-comfort experience and the currentlow-comfort experience would trigger less energy conservation behavior.

50

✐✐

✐✐

✐✐

✐✐


Therefore, future research could collect participants’ energy consump-tion scores of the thermostat task with the physical room temperaturebeing adapted, in which stronger effects of lighting feedback on energyconsumption behavior could be detected. Also, new research could focuson the effectiveness of experience-based ambient persuasive technologyby exploring the effects of ambient lighting on people’s perception ofroom temperature, and their tendency to change room temperature set-tings. With respect to the issue about thermal experiences in ambientpersuasive technology, we will further discuss in Chapter 4.

In sum, the current research contributes to our understanding of the un-derlying psychological mechanisms behind effective ambient persuasivetechnology. Our research shows the importance of the role of strong andconsistent associations of ambient colored lighting that increases the easeof processing feedback messages. The value of these associative processesis not limited to energy consumption behavior, but can easily be gener-alized to other behavioral domains, such as health, safety. In general,colored lighting can be employed as effective persuasive interventions,and the current research contributes to the scientific literature the firstpieces of knowledge on exploring the psychological mechanisms of howto help users to easily process the feedback messages by using color as-sociations.

51

✐✐

✐✐

✐✐

✐✐


52

✐✐

✐✐

✐✐

✐✐

3 | The role of color contexts

Chapter 2 helped us understand the important role of color asso-ciations in associative processes by studying the effects of color

associations of ambient lighting feedback on ease of processing of feed-back messages. Although colors are widely used as feedback in ambientpersuasive technology (e.g., Jahn et al., 2010; Merkus, 2012; Wilson etal., 2013), associations that colors have seem highly context dependent(see Elliot & Maier, 2012). However, research on the effects (e.g., on acolor’s meaning) of color context is still quite limited. Therefore, to bet-ter understand the role of color associations in associative processes, wealso argue that the context in which color-based persuasive technologyis presented is crucial.

In Chapter 3, we extend our insights in the associative processing of col-ored ambient lighting feedback and investigate the influence of the con-text in which colors are presented on user’s interpretation of the feedbackmessages, and furthermore, on the effectiveness of this ambient feedbackfor influencing energy conservation behavior.

This chapter is largely based on:Lu, S., Ham, J., & Midden, C. (2016). Red Radiators Versus Red Tulips: The Influence of Con-text on the Interpretation and Effectiveness of Color-Based Ambient Persuasive Technology. In:Meschtscherjakov A., De Ruyter B., Fuchsberger V., Murer M., Tscheligi M. (eds) Persuasive Tech-nology. PERSUASIVE 2016, LNCS 9638. (pp. 303-314). Springer International Publishing.

53

✐✐

✐✐

✐✐

✐✐

Chapter 3. The role of color contexts

3.1 Introduction

Various studies of the influence of color presented evidence for a linkbetween colors and temperature-related experiences. For example, the“Hue-Heat” Hypothesis (see Bennett & Rey, 1972; Mogensen & English,1926) indicated that colors with “cold” hues (e.g., blue, green and violet),lead to the perception of an object’s temperature being lower, while colorswith “warm” hues (e.g., red, yellow and orange), lead to the perceptionof an object’s temperature being higher. Likewise, as shown by Frangerand colleagues (1977), a room with red lighting led to the perception of ahigher room temperature, whereas a room with blue lighting created theperception of a lower temperature (even though room temperature wasobjectively the same). Also, earlier research showed that yellow lightingin a room could induce a higher perceived room temperature than bluelighting (Winzen et al., 2013).

However, in some other studies that also used colors to test the “Hue-Heat” Hypothesis, different results have been reported refuting or notconfirming this hypothesis (see Candas & Dufour, 2005). For instance,Ho and colleagues (2014) found effects apparently in contrast with thered-hot/blue-cold associations, which indicated that a blue object wasmore likely to be judged as warm than a red object of the same physicaltemperature (Ho, Iwai, Yoshikawa, Watanabe, & Nishida, 2014). Also,several studies were unable to demonstrate that colors had any effectson the perception of temperature (Berry, 1961; Candas & Dufour, 2005).Separate from the “Hue-Heat” hypothesis literature, other research alsoshowed results that conflict. For example, some studies confirmed thehypothesis that warm colors could evoke more arousal than cool colors(Jacobs & Suess, 1975). In contrast, other research presented evidenceopposing this hypothesis. That is, brain research showed that blue lightled to more arousal compared to red light (Yoto, Katsuura, Iwanaga,& Shimomura, 2007). In addition, several researchers pointed out that

54

✐✐

✐✐

✐✐

✐✐

3.1. Introduction

red enhances cognitive performance (Kwallek, Lewis, & Robbins, 1988;Mehta & Zhu, 2008) while, in contrast, others found that red reducescognitive performance (Mehta & Zhu, 2008; Soldat, Sinclair, & Mark,1997), or found no relation between color and performance (Ainsworth,Simpson, & Cassell, 1993; Elliot, Maier, Moller, Friedman, & Meinhardt,2007; Etnier & Hardy, 1997). Finally, some earlier research showed thatcolors could influence human emotional state (Kaya & Epps, 2004), whileothers studies did not find such evidence (O’Connor, 2011).

These conflicting results might be due to various reasons. An influentialtheory that provides an explanation for these diverging effects is Color-in-Context theory by Elliot & Maier (2012). This theory proposes thatthe effects of colors are highly context dependent. That is, it argues thatcolor is an integrated part of the environment and cannot be separatedfrom its form, texture, and other surroundings. Therefore, influences ofcolors can only be observed in context. For instance, the color red cancarry a negative meaning (e.g., failure, danger, threat) for individualsin achievement contexts (Elliot & Maier, 2012). Likewise, research byMehta & Zhu (2008) showed that in contexts in which people have toavoid making mistakes, priming them with the color red made partici-pants more afraid of making mistakes. However, in other contexts, thecolor red can carry a positive meaning and have affiliation-related im-plications, such as in contexts involving sexual interaction (see Elliot &Maier, 2012). More specifically, earlier research showed that red can fa-cilitate males’ attraction to females (Elliot & Niesta, 2008), and likewise,females’ attraction to males (Elliot et al., 2010).

Current Research

In the research area of persuasive technology, colors have been widelyused as media to stimulate behavioral change (see e.g., Arroyo et al.,2005; Fogg & Hreha, 2010; Lapides, Sharlin, & Greenberg, 2009; Midden

55

✐✐

✐✐

✐✐

✐✐


et al., 2007; Occhialini et al., 2011). Colors, such as red and green, havebeen employed to present energy feedback to promote energy conserva-tion behavior (Jentsch et al., 2011; Maan et al., 2011; Merkus, 2012).For example, in an earlier study, participants were given energy con-sumption feedback by changing the shirt color of onscreen avatars thatrepresented participants. The avatar with a red shirt represented theperson who used the largest amount of energy, while the avatar withthe green shirt represented the person who used the smallest amount ofenergy (Midden et al., 2011). Results of this earlier research suggestedthat using these colors as feedback in this context was very effective formotivating participants to save energy. More recent research showedcomparable results, now using colored lighting to provide feedback. Forexample, in Chapter 2, we used red and green lighting to provide am-bient lighting feedback about energy consumption. Results of Chapter2 indicated that colors that have pre-existing associations with energyconsumption (e.g., red/green that is associated with high/low energyconsumption) can enhance the persuasive power of ambient feedback ina context that is related to energy consumption.

In the current chapter we argue that the message that a color carriesmay vary dependent on the context in which a color is perceived. So, incertain contexts, colors such as red are interpreted to indicate high energyconsumption and thereby help users to understand the negative feedbackmessages. In some other contexts, the color red can be interpreted asindicating good and alive, for example, in a nature context (e.g., whendetecting ripe fruit) (Elliot & Maier, 2012).

Based on a pretest, we investigated the importance of color context intwo specific contexts: one was an energy-related context in which colorscould be projected on the form of a heating radiator, and the otherwas an energy-unrelated context in which colors could be projected onthe form of a tulip. In the pretest, participants had indicated that aheating radiator was an object that was strongly associated with energy

56

✐✐

✐✐

✐✐

✐✐

3.2. Study 3.1

consumption, whereas a tulip was an object that was weakly associatedwith energy consumption.

Specifically, in current research, we argue that the color context couldfacilitate activation of the color associations. That is, presenting colors(e.g., red) in an energy-related context (e.g., projecting red on an object -the form of a radiator - to indicate high energy consumption) would makeenergy consumption feedback messages (e.g., high energy consumptionlevel) more clear compared to presenting colors in an energy-unrelatedcontext (e.g., projecting red on an object - the form of a tulip - to in-dicate high energy consumption). Also, presenting colors (e.g., red) inan energy-related context would moderate the assessment of an object’swarmth (e.g., projecting red on a radiator is associated with warmth)compared to presenting colors in an energy-unrelated context. More-over, we argue that presenting colors in energy-related contexts can in-crease the effectiveness of the color-based ambient persuasive technology,leading to stronger energy conservation behavioral change.

3.2 Study 3.1

In Study 3.1, we focused on the influence of context (in which colors wereperceived) on user’s interpretation of the color messages when using itas a feedback mechanism in colored-based ambient persuasive techno-logy. Two specific colors (i.e., red and blue) were presented in either anenergy-related context or an energy-unrelated context. We argued thatthe color associations could be shaped by the color context. Therefore,we expected (H3.1.1) an interaction effect of color and context on thetwo assessments of the stimuli: user’s interpretation of the feedback mes-sages (i.e., energy consumption level) and user’s assessment of an object’swarmth (i.e., warmth association). More specifically, for user’s interpre-tation of the feedback messages, we expected that participants wouldinterpret the color red in an energy-related context as indicating higher

57

✐✐

✐✐

✐✐

✐✐


energy consumption level than the color red in an energy-unrelated con-text. However, we expected that participants would not interpret thecolor blue in an energy-related or in an energy-unrelated context diffe-rently. For user’s assessment of an object’s warmth, we expected thatparticipants would perceive the color red in an energy-related context tobe warmer than the color red in an energy-unrelated context. In con-trast, this different would not be expected for the color blue between twocontexts.

In the literature on the “Hue-Heat” Hypothesis that we mentioned above,the color red and blue were often chosen to test the relationship betweencolors and temperature-related experiences. Specifically, in our currentresearch, presenting the color red on a radiator seems to be more naturalthan presenting the color blue on a radiator. That is, people may easilyassociate the red radiator with the hot radiator that glows, whereasthe blue radiator may not be easily associated with a cold radiator.Therefore, the color red may have stronger effects than the color blue inthese interaction effects.

Moreover, we argued that presenting “right” colors in a “right” contextwould help users the most to correctly interpret the feedback messagesand enhance their assessments of an object’s warmth. In our case, pre-senting “right” colors in a “right” context means that presenting the colorred in an energy-related context to indicate a high energy consump-tion level and to be associated with warmth. Therefore, we expected(H3.1.2) that participants would interpret the color red in an energy-related context as indicating the highest energy consumption level andbeing the warmest, compared to the other three conditions: the colorred in an energy-unrelated context, and the color blue in an energy-related or energy-unrelated context. More specifically, concerning user’sinterpretation of the feedback message, we expected that participantswould interpret the color red in an energy-related context as indicat-ing the highest energy consumption level, compared to the color red in

58

✐✐

✐✐

✐✐

✐✐

3.2. Study 3.1

an energy-unrelated context and the color blue in an energy-related orenergy-unrelated context. Concerning user’s assessment of an object’swarmth, we expected that participants would perceive the color red inan energy-related context to be the warmest, compared to the partici-pants in the other conditions.

3.2.1 Method


Fifty-one students (average age 21.6 years old, SD = 2.15) of EindhovenUniversity of Technology (23 male and 28 female) were recruited by usinga local participant database. All participants were native Dutch speakersand were randomly assigned to one of the four experimental conditions:2 (context: energy-related vs. energy-unrelated) × 2 (color: red vs. blue).The experiment lasted approximately 15 minutes. Participants receivede 3 for their participation.

Experimental produce

Before participants entered the lab, the room was adapted to fit one ofthe four experimental conditions. That is, we changed only the picturepresented on the TV screen present in the room to show the relatedlighting stimulus (i.e., a red radiator, a red tulip, a blue radiator or ablue tulip; see Figure 3.1).

After entering the lab room, participants were requested to look aroundthe room and then remember (for later recall) as many objects in thisroom as possible (with the purpose of making them spend attention tothe ambient environment). Then, participants were asked to fill out aquestionnaire. This questionnaire contained a variety of questions aboutthe room and the object on the TV screen. Among these questions werethe two questions of focus. That is, the questionnaire asked participants

59

✐✐

✐✐

✐✐

✐✐


(a) red radiator (b) red tulip (c) blue radiator (d) blue tulip

Figure 3.1: Four between-subject experimental conditions of Study 3.1: (a) the color redin an energy-related context, (b) the color red in an energy-unrelated context, (c) the colorblue in an energy-related context and (d) the color blue in an energy-related context.

for their interpretations of the color (i.e., either red or blue) on either theradiator or tulip (using the question “How do you judge the meaning thatthis object conveys? I think this object means...”, to which participantscould answer by choosing an option on a 7-point rating scale ranging from1 = very low energy consumption to 7 = very high energy consumption).Also, the questionnaire asked participants for the warmth perception ofthe color on either radiator or tulip (using the question “How warmdo you feel the object projected on the screen is? I feel the object issomething...”, to which ranging from 1 = very cold to 7 = very warm).Finally participants were debriefed and thanked for their participation.

3.2.2 Results

To analyze the interaction effect of color and context on participant’s in-terpretation of the feedback messages and participant’s assessment ofan object’s warmth, we submitted our two dependent variables to a2 (context: energy-related vs. energy-unrelated) × 2 (color: red vs. blue)multivariate analysis of variance (MANOVA).

Not in line with our hypothesis (H3.1.1), results of this MANOVA didnot show the expected full interaction effect of color and context onparticipant’s interpretation of the feedback messages and assessment ofan object’s warmth, F (2, 46) = 1.98, p = .15, η2 = .079. Within the sameanalysis, univariate ANOVAs on the outcome variables did not reveal

60

✐✐

✐✐

✐✐

✐✐

3.2. Study 3.1

the significant full interaction effects on participant’s interpretation ofthe feedback messages, F (1, 47) = 2, 19, p = .15, η2 = .045, and onparticipant’s assessment of an object’s warmth, F (1, 47) = 3, 50, p =.068, η2 = .069.

However, as we discussed above in Section 3.2, we argued that parti-cipants may naturally associate for example the red radiator with thehot radiator that glows, whereas the blue radiator may not be naturallyassociated with cold or icy radiator. Therefore, the effects of contextfor the color red was expected to be more visible than the effects ofcontext for the color blue. So, we provided additional support for ourhypothesis (H3.1.1) by reporting the simple effects of context withinthe same MANOVA. That is, results revealed the expected simple ef-fect of context for the color red, F (2, 46) = 5.59, p = .007, η2 = .195.In contrast and as expected, no simple effect of context for the colorblue was found, F (2, 46) = .17, p = .85, η2 = .007. More specifically,within the same analysis, results of univariate ANOVAs on the two out-come variables showed the expected simple effect that the color red inan energy-related context was interpreted to indicate higher energy con-sumption level, F (1, 47) = 7.29, p = .010, η2 = .134, and to be warmer,F (1, 47) = 9.01, p = .004, η2 = .161, compared to the color red in anenergy-unrelated context. However, as expected, univariate ANOVAsdid not reveal the simple effects of context for the color blue on bothoutcome variables.

Moreover, supporting our second hypothesis (H3.1.2), planned contrastanalysis on our outcome variables by comparing the condition (presentingthe color red in an energy-related context) to the other three conditions(presenting the color red in an energy-unrelated context, the color bluein an energy-related context and the color blue in an energy-unrelatedcontext). Results of this analysis revealed that participants interpretedthe color red in an energy-related context as indicating the highest energyconsumption level and being the warmest, F (2, 46) = 25.4, p < .001,

61

✐✐

✐✐

✐✐

✐✐


η2 = .525. Also, within the same analysis, separate planned contrastanalyses on two outcome variables revealed the expected effects thatpresenting the color red in an energy-related context was interpretedas indicating the highest energy consumption level, F (1, 47) = 17.50,p < .001, η2 = .271 (see Figure 3.2), and being the warmest F (1, 47) =50.47, p < .001, η2 = .518 (see Figure 3.2), compared to the other threeconditions.

0

1

2

3

4

5

6

7

theinterpretationo

fene

rgy

consum

ptionlevel

radiator tulipContext

red blue

Figure 3.2: Averaged participant’s interpretation of energy consumption level for all con-ditions in Study 3.1: the color red in an energy-related context (i.e., radiator) and in anenergy-unrelated context (i.e., tulip); and the color blue in an energy-related context (i.e.,radiator) and in an energy-unrelated context (i.e., tulip). Error bars attached to each columnin the figure indicate 95% confidence intervals.

0

1

2

3

4

5

6

7

thea

ssessm

ento

fanob

ject's

warmth

radiator tulipContext

red blue

Figure 3.3: Averaged participant’s assessment of an object’s warmth for all conditions inStudy 3.1: the color red in an energy-related context (i.e., radiator) and in an energy-unrelated context (i.e., tulip); and the color blue in an energy-related context (i.e., radiator)and in an energy-unrelated context (i.e., tulip). Error bars attached to each column in thefigure indicate 95% confidence intervals.

62

✐✐

✐✐

✐✐

✐✐

3.2. Study 3.1

In addition, we explored the relationship between the two dependentvariables. Results indicated that participant’s interpretations of feedbackmessages correlated to participant’s assessment of an object’s warmth(r = .625, p < .001).

3.2.3 Discussion

Although no full interaction effect was found in Study 3.1, results showedthe expected simple effects of context for the color red on user’s inter-pretations of feedback messages and on user’s assessment of an object’swarmth. That is, participants interpreted the color red as indicatinghigher energy consumption level and being warmer in an energy-relatedcontext compared to the color red in an energy-unrelated context. Thesefindings supported our expectations that for the color red, people per-ceived the colors in contexts differently with respect to the informationthat the color carried.

Moreover and in line with our expectations, results of planned contrastanalyses in Study 3.1 showed that presenting colors (e.g., red) in anenergy-related context indeed made the feedback message (e.g., highestenergy consumption level) the most clear, and moreover, enhanced thecorrect assessment of an object’s warmth (e.g., warmest) the most.

Still, although Study 3.1 showed that participants could easily under-stand the information that color-based persuasive technology conveyedin a context that consistent with the color, the important question re-mains of whether this color-consistent context (in which the color-basedfeedback was presented) can influence users’ energy conservation beha-vior. Therefore, the second study was designed with the following twopurposes: (1) to replicate the effects of context on users’ interpretationof feedback messages and user’s assessment of an object’s warmth thatcolor-based persuasive technology carries, and (2) to further investigate

63

✐✐

✐✐

✐✐

✐✐


the influence of context on the effectiveness of color-based ambient feed-back for promoting users’ energy conservation behavior.

3.3 Study 3.2

The first aim of Study 3.2 was to replicate the effects of context on user’sinterpretation of feedback messages and user’s assessment of an object’swarmth that we found in Study 3.1. We argued that an energy-relatedcontext would strengthen the associations of color with high energy con-sumption and warmth, which could make the feedback messages directlyunderstandable. Therefore, we expected (H3.2.1) that participants wouldinterpret the color red in an energy-related context as indicating a higherenergy consumption level and as being warmer than the color red in anenergy-unrelated context.

Secondly, to further investigate the context effect on energy conserva-tion behavior, we argued that the energy-related context could increasethe effectiveness of this color-based energy feedback to facilitate energyconservation behavior. Therefore, we developed an interactive energyfeedback system, in which the object (i.e., either a radiator or tulip)could show colors varying from very weak red to very intense red de-pending on people’s energy use. We expected (H3.2.2) that participantswould save the most energy when they receive the color-based ambientfeedback in an energy-related context.

3.3.1 Method

Participant and Design

Fifty students (average age 23.1 years old, SD = 3.14) of Eindhoven Uni-versity of Technology (32 male and 18 female) were recruited by using

64

✐✐

✐✐

✐✐

✐✐

3.3. Study 3.2

a local participant database. They participated in the current experi-ment in which color-based ambient feedback was presented in either anenergy-related context (i.e., projecting red as energy feedback on theradiator) or an energy-unrelated related context (i.e., projecting red onthe tulip). All participants were native Dutch speakers. The experimentlasted approximately 30 minutes, and participants received e 5 for theirparticipation.

Experimental Procedure

Participants were welcomed at the entrance of the lab. Each participantwas asked to read and sign a consent form stating the general purpose ofthe research to indicate their willingness to participate in this research.After entering the lab, the participant was seated in front of a laptopon a desk. All the instructions and tasks used in this experiment werefully computerized and were all in Dutch. First of all, participants wereasked to read a magazine for a period of five minutes, while either a redradiator (in the energy-related context condition) or a red tulip (in theenergy-unrelated context condition) was presented on the TV (see Figure3.4). After five minutes, the computer indicated that the reading periodwas over. Then participants were asked to fill out a questionnaire thatwas identical to the questionnaire used in Study 3.1 about the object onthe TV screen and the room.After answering these questions, we assessed energy consumption beha-vior by asking participants to perform a virtual thermostat task on thelaptop (which is identical to the task that was successfully used in Chap-ter 2 and earlier research, see Maan et al., 2011). That is, participantsreceived instructions on how to program a virtual room thermostat inter-face, and were asked to program this thermostat for two practice scenar-ios. After two practice scenarios, participants completed programmingtasks on the thermostat for ten experimental scenarios. For each sce-nario, the scenario itself was first described to the participants (e.g., It is

65

✐✐

✐✐

✐✐

✐✐


(a) feedback system via radiator (b) feedback system via tulip

Figure 3.4: The interactive ambient energy feedback system by presenting colors in (a) anenergy-related context and in (b) an energy-unrelated context.

night and you are going to bed. It is −10 ◦C outside), and then partici-pants were asked to program the thermostat by setting the temperaturefor the each room, while receiving interactive color-based energy feed-back (feedback in the energy-related context condition, see Figure 3.5;feedback in the energy-unrelated context condition, see Figure 3.6) aftereach change of the temperature settings, until they pressed the ‘Ready’button. Finally, participants were asked to answer several demographicquestions, thanked for participation, and debriefed.

(a) (b) (c) (d)

Figure 3.5: In the energy-related context, the interactive color-based energy feedback systemshowed colors varying between very weak and intense red depending on people’s energy usefrom (a) indicating very low energy consumption to (d) very high energy consumption.

3.3.2 Results

Replicating the effects of context on user’s interpretation of the feedbackmessages and user’s assessment of an object’s warmth that we found in

66

✐✐

✐✐

✐✐

✐✐

3.3. Study 3.2

(a) (b) (c) (d)

Figure 3.6: In the energy-unrelated context, the interactive color-based energy feedback sys-tem showed colors varying between very weak and intense red depending on people’s energyuse from (a) indicating very low energy consumption to (d) very high energy consumption.

Study 3.1, we submitted the two dependent variables to a 2 (context:energy-related vs. energy-unrelated) MANOVA.

In line with our hypothesis (H3.2.1), results of this MANOVA showedthat participants interpreted the color red in an energy-related context asindicating a higher energy consumption level and as being warmer thanthe color red in an energy-unrelated context, F (2, 47) = 12.98, p < .001,η2 = .36. Also, the same as in previous study, univariate ANOVAs onthe outcome variables revealed the significant main effects of context forthe color red on participant’s interpretation of the feedback messages,F (1, 48) = 18.28, p < .001, η2 = .28, and on participant’s assessment ofan object’s warmth, F (1, 48) = 10.69, p = .002, η2 = .18 (see Figure 3.7and Figure 3.8).

0

1

2

3

4

5

6

7

theassessmen

tofanob

ject's

warmth

radiatortulipContext

Figure 3.7: Averaged participant’s assessment of an object’s warmth as a result of context(energy-related, i.e., radiator vs. energy-unrelated, i.e., tulip) in Study 3.2. Error barsattached to each column in the figure indicate 95% confidence intervals.

Moreover, to analyze the influence of context on the effectiveness of this

67

✐✐

✐✐

✐✐

✐✐


1

2

3

4

5

6

7

theinterpretatio

nofene

rgy

consum

ptionlevel

RadiatorTulipContext

Figure 3.8: Averaged participant’s interpretation of energy consumption level as a result ofcontext (energy-related, i.e., radiator vs. energy-unrelated, i.e., tulip) in Study 3.2. Errorbars attached to each column in the figure indicate 95% confidence intervals.

color-based ambient persuasive technology for influencing user’s energyconservation behavior, we collected the energy consumption scores basedon participant’s actions during the whole thermostat task as a depen-dent variable. This dependent variable consisted of the differences ofenergy consumption scores between every setting and the previous setting(caused by an action performed by the participant, e.g., increasing thethermostat temperature). Therefore, the change of energy consumptionscores per each action4 was submitted to a Linear Mixed-Effects Modelto assess the relationship between the color context (radiator vs. tulip)and the energy consumption score with the specific thermostat task asthe random factor. Supporting our second hypothesis (H3.2.2), resultsshowed that participants who received color-based energy feedback inthe energy-related context (i.e., radiator) reduced their energy consump-tion more (EMM = −13.6, SE = 2.25) than participants who receivedcolor-based energy feedback in the energy-unrelated context (i.e., tulip)(EMM = −10.1, SE = 2.3), F (1, 1715.9) = 4.84, p = .028 (see Figure3.9).

4The change of energy consumption scores per each action = the current-action energy con-sumption score − the previous-action energy consumption score.

68

✐✐

✐✐

✐✐

✐✐


-19

-17

-15

-13

-11

-9

-7

-5

-3

-1

thec

hang

eofene

rgy

consum

ptionscores(kWh)

RadiatorTulipContext

Figure 3.9: The change of energy consumption scores per action as a result of context(energy-related, i.e., radiator vs. energy-unrelated, i.e., tulip) in Study 3.2. Error barsattached to each column in the figure indicate 95% confidence intervals.

3.3.3 Discussion

So, results of Study 3.2 showed that participants could correctly interpretthe color red as indicating a higher energy consumption level and be-ing warmer in an energy-related context. Most interestingly, presentingcolor-based energy feedback messages in an energy-related context led tothe most energy conservation behavior. Thereby, these findings revealedthat the color context can help users to correctly interpret the informa-tion that color-based feedback carries, and moreover, can increase theeffectiveness of this ambient persuasive technology to facilitate energyconsumption behavior.


Colors have been widely used as media to promote behavioral changes(e.g., Ham & Midden, 2010; Jahn et al., 2010; Wilson et al., 2013). In thecurrent research, we argued that the information that color-based am-bient persuasive technology carries was highly context-dependent. Twostudies were designed to investigate the effects of a color’s context (i.e.,the context in which colors were presented as energy feedback) on users’

69

✐✐

✐✐

✐✐

✐✐


interpretation of the information that colors carried. Also, we investi-gated the effect of color context on the effectiveness of this color-basedambient persuasive technology to influence users’ energy consumptionbehavior. Results of both studies showed that participants to whomwe presented colors in an energy-related context (e.g., projecting red onan object in the form of a radiator indicated high energy consumption)could most optimally interpret the information that a color carried (e.g.,red indicated high energy consumption and was more associated withwarmth). Moreover, results showed that participants to whom we pre-sented colors in an energy-related context saved the most energy whenthey were receiving color-based energy feedback as compared to partici-pants in an energy-unrelated context.

In line with earlier research (Busch et al., 2012) and the Color-In-Contexttheory (Elliot & Maier, 2012), our findings confirmed that the contextcould shape the associations of color. Also, using colors as ambient feed-back to promote energy conservation behavior, we found that a color-energy-consistent context can facilitate activation of the associations ofthe color red with high energy consumption or concept of warm. Further-more, this context can enhance the energy-saving performance promotedby color-based energy feedback. Remarkably, the current research im-plies that by taking into consideration the context of ambient persuasivetechnology, the effectiveness of color-based ambient persuasive techno-logy could be increased. This mechanism is not only limited to energyfeedback (as used in Chapter 2) but can be generalized to other do-mains, like health lifestyle (e.g., presenting the color red as feedback inhealth-related context).

The current research opens up a new field of research. That is, basedon the first findings of the current studies, future research might inves-tigate the influence of context for colors on user comfort perceptionsand thereby on user’s energy consumption behavior. Indeed, comfortexperiences were proposed to be one of the dominant influences on peo-

70

✐✐

✐✐

✐✐

✐✐


ple’s energy conservation behavior (Becker et al., 1981; Heijs & Stringer,1988). Besides using colors in different contexts as feedback to indicateenergy information, future research can also investigate whether user’scomfort experiences will be influenced by using color-based persuasivetechnology in different contexts. For instance, future research might in-vestigate whether the existing example of projecting a glowing red fire ona wall in the restaurant could actually increase the costumer’s thermalcomfort experience, compared to projecting the color red as a context-independent ambient lighting (e.g., projecting red lighting on a plainwall).

Another important consideration related to studying comfort experiencesis that in our second study, participants performed the temperature-setting task on a simulated thermostat without the physical room tem-perature being adapted. However, in a day-to-day environment, behav-ioral changes in the settings of a heating thermostat directly lead to thechanges of the room temperature, which might stimulate more people’sbehavioral change. Future research could investigate the relationshipbetween ambient persuasive technology, comfort experiences and energyconsumption behavior, in an environment in which these variables areconnected to one another. For example, room temperature changes afterpeople change settings of a thermostat. In such research, a simulatedthermostat could be connected to the heating system in the room.

Additionally, future research may extend the current findings by notonly presenting color-based ambient energy feedback on a TV screen.That is, future research might replicate and also extend these findings byinvestigating the effects of colors and the role of their context when colorsare presented in a more ecological context, for example when presentingan actual red glow on an actual physical radiator (e.g., Gyllensward,Gustafsson, & Bang, 2006). By doing this, this psychological mechanismof color context on facilitating activation of color associations might beapplied to different household energy-consumption devices in real and

71

✐✐

✐✐

✐✐

✐✐


make the perception of the energy consumption less explicit while stilleffective (see e.g., Jensen, 2003; Matsukawa, 2004).

In sum, our research showed that a color-energy-consistent context inwhich ambient persuasive technology is presented can facilitate activa-tion of the associations of color with high energy consumption and theconcept of warm, which makes the feedback messages correctly under-standable. Moreover, this context can increase the persuasive powerof ambient persuasive technology to promote energy conservation beha-vior. These findings deepen our insight into the associative processes ofcolored ambient lighting feedback by making clear that context can fa-cilitate activation of the color associations and furthermore help peopleeasily process the feedback messages. This could also have implicationsfor the design of future ambient persuasive technology as the current fin-dings show that such design should take into consideration the contextin which ambient persuasive technology is used.

72

✐✐

✐✐

✐✐

✐✐

4 | The role of color temper-ature experiences

As described in the general introduction, we studied in Chapter2 and Chapter 3 the associative processes underlying the ef-

fectiveness of ambient persuasive lighting feedback. These associativeprocesses are one part of the impulsive cognitive processes (Strack &Deutsch, 2004). Importantly, processes in the impulsive cognitive sys-tem, to a certain extent, may also consist of experiential processes (Strack& Deutsch, 2004). These experiential processes are also associative innature, but involve less cognition (see Dijksterhuis et al., 2000; Zajonc,1980). In the current chapter, we investigate whether these experientialprocesses could also contribute to explaining the effects of ambient light-ing persuasive technology. As argued in the previous chapters, lighting,as a medium of ambient persuasive technology (Davis, 2008; Ham et al.,2009; Maan et al., 2011), can be used as a successful feedback mediumof energy consumption. The basic persuasive strategies employed by

This chapter is largely based on:Lu, S., Ham, J., & Midden, C. (2015). Persuasive technology based on bodily comfort experiences:The effect of color temperature of room lighting on user motivation to change room temperature.In: MacTavish T., Basapur S. (Eds) Persuasive Technology. PERSUASIVE2015, LNCS 9072 (pp.83-94). Springer Cham. (Best Paper Award).

73

✐✐

✐✐

✐✐

✐✐

Chapter 4. The role of color temperature experiences

these forms of ambient persuasive technology that use colored lightingare persuasive interventions that necessitate associative processing.

In the current chapter, we extend our understanding of the psychologicalmechanisms behind the impulsive processes of ambient persuasive tech-nology by exploring the experiential processes. For this, we propose anew perspective on ambient lighting persuasive technology, which mightinfluence people through thermal experiences.

4.1 Introduction

Numerous social psychological studies have established the fact that so-cial and psychological experiences can influence people’s assessment oftheir physical environments (Asch, 1951; Helson, Blake, Mouton, & Olm-stead, 1956; Sherif, 1948). For example, earlier research showed that so-cial exclusion experiences could make people literally feel colder (Zhong& Leonardelli, 2008). Likewise, other research showed that people’s be-liefs regarding the current temperature settings and their own abilitiesto adjust to warm or cold conditions can influence their thermal ex-periences (Howell & Kennedy, 1979; Howell & Stramler, 1981). Also,research showed that a reported (relatively) high temperature can in-crease perceived comfort even when the actual room temperature re-mained the same (Stramler, Kleiss, & Howell, 1983). More recently,research demonstrated that a sustainable mindset (i.e., being aware ofacting environmentally-friendly) can induce people to perceive the roomtemperature to be physically higher (Taufik, Bolderdijk, & Steg, 2015).

In this chapter, we propose a new perspective on ambient persuasive tech-nology, which may influence people through thermal experience. Earlierresearch indicated that one important guiding factor (at least with re-spect to saving energy by decreasing room temperature) in people’s en-vironment is their thermal comfort experience (Becker et al., 1981; Heijs

74

✐✐

✐✐

✐✐

✐✐

4.1. Introduction

& Stringer, 1988). That is, thermal comfort is one of the main reasonspeople have for conserving energy or not, and thermal experiences aremore important than an awareness of the reality of the energy crisis orone’s disposable budget (see Becker et al., 1981). Likewise, research (Rit-sema, Midden, & Heijden, 1982; Van Raaij & Verhallen, 1983; Verhallen& Van Raaij, 1981) indicated the experience of thermal comfort as aprominent determinant of an energy-conscious attitude. In other words,discomfort experiences were considered as a cause (or as the main cause)to not save natural resources (e.g., Merkus, 2012; Ohnaka et al., 1994).Therefore, we argue that more attention is needed for thermal experi-ences in persuasive strategies, and we propose to investigate the wayin which persuasive technology could employ experiential processes andmore specifically thermal experiences.

Persuasive technology that is based on thermal experience can be dis-tinguished from persuasive technology that uses other kinds of persua-sive strategies by (as least) the following characteristics: 1. Insteadof information-based strategies (e.g., directly indicating the amount ofkWh, or the color associations with energy consumption), Experience-Based (EB) persuasive technology uses persuasive strategies that areexperience-based (e.g., feelings of comfort in a specific environment);2. Information-based persuasive strategies involve mostly cognitive pro-cesses, while EB persuasive strategies involve mainly bodily and per-ceptual processes; 3. Information-based persuasive strategies influencepeople’s behavior consciously (or at least largely consciously), and EBpersuasive strategies mainly have an unconscious influence on people’sbehavior; and 4. Information-based persuasive strategies have an impacton people’s thinking, whereas EB persuasive strategies have a relativelydirect impact on people’s experiences.

In the current chapter, we assess the effectiveness of the experience-basedforms of persuasive technology we proposed. Therefore, two studies wereconducted to explore whether ambient lighting could influence people’s

75

✐✐

✐✐

✐✐

✐✐


thermal experiences, and their willingness to save energy (by decreasingthe room temperature).

Basic physics states that incandescence is heat made visible. That is,all objects with a temperature greater than absolute zero emit thermalradiation or electromagnetic waves as temperature rises, making veryhot objects appear as if they are emitting light. Because people oftenexperience light along with heat, this connection between light and heatmay be deeply rooted psychologically.

Among various characteristics of light, color temperature of light hasbeen investigated with respect to its non-visual effects in physiologicalmeasurements, such as melatonin secretion, body temperature, corticalarousal level, autonomic nerve tone, hormonal secretion and motor func-tion (e.g., M. Aarts et al., 2007; Sato & Yasukouchi, 1999; Yasukouchi& Ishibashi, 2005). For example, body heat loss might be inhibited ifpeople were exposed to light with lower color temperature (Iseki & Ya-sukouchi, 2000; Tottori, Kozaki, Yasukouchi, Iwahashi, & Ikeda, 2000).Besides the physiological effects of color temperature of light, earlier re-search also demonstrated the psychological effects on people’s preferencesof their environments and their atmosphere perception (Vogels, 2008).For example, people prefer low color temperature (i.e., “warm” lighting)in winter, and high color temperature (i.e., ‘cool’ lighting) in summer(Nakamura & Oki, 2000). Also, based on several studies, Ishikawa (1993)estimated that household energy consumption for heating systems couldbe reduced by 5% to 8% by using “warm” lighting in winter and “cold”lighting in summer, while maintaining the same thermal comfort level.

To investigate the effectiveness of EB persuasive technology, in the cur-rent chapter we conducted two experiments in which we tested the effec-tiveness of persuasive technology that manipulated thermal perceptionswith the goal of influencing user’s tendency for changing energy consump-tion behavior. That is, we performed two studies in which we changedlighting settings in a room by manipulating room lighting color tempera-

76

✐✐

✐✐

✐✐

✐✐

4.2. Study 4.1

ture (“warm” vs. “cold” lighting). Based on the aforementioned findingsof earlier research, we expected in Study 4.1 to find evidence that roomlighting color temperature influences participants’ perceptions of roomtemperature, and their tendency to change room temperature. More in-terestingly, in Study 4.2 of this chapter, instead of measuring people’stendency to change temperature, we added a thermostat task to investi-gate people’s temperature-setting behavior.

4.2 Study 4.1

Therefore, in the first study of this chapter, we expected (H4.1.1) themain effect of color temperature of room lighting on the perception ofthe room temperature and participant’s tendency to change the roomtemperature. More specifically, we expected that participants in the“warm” lighting condition would perceive the room temperature to behigher and express a stronger tendency to decrease the room temperaturethan participants in the “cold” lighting condition.

4.2.1 Method

Participant and design

Sixty-six visitors of Eindhoven University of Technology participated inthis study. Of these, thirty-eight participants were high school students(average age 17.2 years old, SD = 1.87) and twenty-eight were theirparents (average age 50.1 years old, SD = 3.86). All participants werenative Dutch speakers and participated during a visit of the lighting labat the school of Innovation Sciences, Eindhoven University of Technologyduring the University Open Day. Six groups (approximately ten visitorsfor each group) were randomly assigned to either the “warm” lightingcondition or the “cold” lighting condition.

77

✐✐

✐✐

✐✐

✐✐


Experimental procedure

Before the groups entered the lab room, the lighting of the room wasset to a “warm” lighting condition (2700 Kelvin), or to a “cold” lightingcondition (5500 Kelvin). The lighting in the room was programmed tothe specific settings (“warm” or “cold”) before the group entered theroom, and stayed in those settings until after the group had left theroom.

After entering the lab room, the participants were given a short (ver-bal) general introduction to this lighting lab, in which a confederateexplained about the technicalities of the various lighting systems in theroom (e.g., ceiling and wall luminaries, wall washers) and the way thewall washers (LEDs that illuminate the wall) can be programmed. Inthese explanations, neither the ceiling luminaries and their lighting set-tings nor potential influences of lighting on people were discussed, so nopsychological effects related to the experiment were explained. Duringthese explanations, lighting conditions in the room remained unaltered,and no demonstrations of changes in the settings of any of the lightingsystems were given. After this introduction (of approximately 3 min-utes), each participant was asked whether he or she wanted to partici-pate in a short study, by filling out a one-page questionnaire (in Dutch,as described in the Materials section). Answering the questions of thequestionnaire took approximately 2 minutes. After this, the lighting labdemonstrations continued, and the confederate (lab tour guide) providedmore extensive explanations including details on the ceiling lighting, howlighting might be used to influence users. Also, the confederate debriefedall members of the group on the purpose of the study they had partici-pated in and thanked them for their participation. The whole visit tothe lighting lab lasted approximately 20 minutes.

78

✐✐

✐✐

✐✐

✐✐

4.2. Study 4.1

Materials

To assess perceived room temperature, perceived lighting temperature,and participant’s tendency to change room temperature, we asked parti-cipants to fill out a questionnaire. This questionnaire contained a varietyof questions about the lab room, among them the three questions of fo-cus. That is, this questionnaire asked participants first of all for theirperceptions of the room temperature (using the question “What do youthink about the temperature in this room?”, to which they could answeron a 7-point rating scale ranging from 1 very cold to 7 very warm).

Also, the effectiveness of the color temperature manipulation was checkedby asking participants for their perceptions of the lighting temperaturein the room (using the question “What do you think about the lighting inthis room?” to which they could answer by indicating whether “Warm”was applicable on a 7-point rating scale ranging from 1 absolutely notapplicable to 7 very much applicable).

Additionally, the questionnaire asked participants for their tendency tochange the temperature in the room (using the question “If you couldadjust the temperature in this room, by how many degrees would youlike to change it?” to which participants could answer on a 7-point ratingscale, ranging from −3 ◦C to 3 ◦C).

Finally, participants were asked to indicate their age, gender and thenumber of layers of clothing they were wearing.

4.2.2 Results

First of all, to check whether the color temperature manipulation ofroom lighting was successful, results of non-parametric Mann-Whitneytest showed that participants in the “warm” lighting condition perceivedthe room lighting to be warmer (M = 4.00, SD = 1.83) than participants

79

✐✐

✐✐

✐✐

✐✐


in the “cold” lighting condition (M = 5.48, SD = 1.21), U = 290.00,z = −3.30, p = .001, r = −.41 (see Figure 4.1).

0

1

2

3

4

5

6

7

thew

armthperceptionof

room

lighting

lowhighColortemperatureoflight

Figure 4.1: Averaged warmth perception of room lighting as a result of color temperature(“warm” vs. “cold”) of room lighting in Study 4.1. Error bars attached to each column inthe figure indicate 95% confidence intervals.

A total of three univariate (with a z-score > 2.9) and multivariate (witha Mahalanobis distance > 13.82) outliers were found on the measurefor participant’s tendency to change room temperature. Therefore, weexcluded these three outliers from our analyses.

To analyze the main effect of color temperature of room lighting onparticipant’s perception of room temperature and participant’s tendencyto change room temperature, we submitted the dependent variables to a2 (color temperature: warm vs. cold) MANOVA.

Supporting our hypothesis (H4.1.1), results of this MANOVA revealedthe expected main effect of color temperature on participant’s percep-tion of the room temperature and his or her tendency to change roomtemperature, F (2, 58) = 3.84, p = .027, η2 = .117.

Within this MANOVA, univariate ANOVA on participant’s tendency tochange room temperature showed the expected effect that participants inthe “warm” lighting condition expressed a stronger tendency to decreasethe room temperature by an average of 1.91 ◦C (SD = .96) than par-ticipants in the “cold” lighting condition who wanted to decrease room

80

✐✐

✐✐

✐✐

✐✐

4.2. Study 4.1

temperature by an average of 1.13 ◦C (SD = 1.45), F (1, 59) = 7.55,p = .008, η2 = .113 (see Figure 4.2).

-3

-2.5

-2

-1.5

-1

-0.5

0

thet

ende

ncyt

ochangethe

room

tempe

rature(oC)


Figure 4.2: Averaged participant’s tendency to change the room temperature as a result ofcolor temperature (“warm” vs. “cold”) of room lighting in Study 4.1. Error bars attachedto each column in the figure indicate 95% confidence intervals.

0

1

2

3

4

5

6

7

thewarmthperceptionof

room

tempe

rature


Figure 4.3: Averaged warmth perception of room temperature as a result of color tempera-ture (“warm” vs. “cold”) of room lighting in Study 4.1. Error bars attached to each columnin the figure indicate 95% confidence intervals.

However, univariate ANOVA (within the same MANOVA) on partici-pant’s perception of the room temperature did not reveal the expectedeffect. That is, no evidence was found that participants in the “warm”lighting condition perceived the room temperature to be higher (M =5.11, SD = .96) than participants in the “cold” lighting condition (M =4.83, SD = 1.04), F (1, 59) = 1.72, p = .195, η2 = .028 (see Figure 4.3).

81

✐✐

✐✐

✐✐

✐✐


4.2.3 Discussion

In sum, results of the MANOVA in Study 4.1 showed that color tempera-ture of room lighting can influence participant’s perception of room tem-perature (i.e., perceive the room temperature to be warmer) and partici-pant’s tendency to change room temperature (i.e., express a stronger ten-dency to decrease the room temperature). Thereby, Study 4.1 presentsevidence that color temperature of room lighting can have a significanteffect on people’s perception of room temperature and their tendency tochange the room temperature.

However, the psychological mechanisms for a stronger tendency to de-crease the room temperature are still unclear. Crucially, earlier research(e.g., Becker et al., 1981; Heijs & Stringer, 1988) showed that comfortof a dwelling is one of the main reasons for conserving energy or not.We argued that user comfort might be one of the most important de-terminants of user motivation to change room temperature settings anda crucial psychological variable when users process ambient persuasivetechnology.

Therefore, we conducted a follow-up study that incorporated a user com-fort experience measure. Additionally, the psychological mechanismscausing a stronger tendency to decrease the room temperature might alsobe effective for actually reducing energy consumption. Thus, in Study4.2, we investigated whether a person in a room with “warm” lightingwould diminish his or her energy consumption in a task in which parti-cipants were requested to program the heating thermostat.

4.3 Study 4.2

For Study 4.2, we argued that color temperature of room lighting mightinfluence a person’s room temperature perception, the comfort experi-

82

✐✐

✐✐

✐✐

✐✐

4.3. Study 4.2

ences of the room lighting, the comfort experiences of the room temper-ature, and also the person’s energy consumption behavior in the room(i.e., actually taking measures to increase room temperature).Therefore, in Study 4.2, we first wanted to replicate Study 4.1. Also, weincluded measures for perceived comfort of room lighting, perceived com-fort of the room temperature and a behavioral energy consumption mea-sure as additional dependent variables. We argued that “warm” lightingalso had positive effects on these three additional variables. More spe-cifically, we expected (H4.2.1) that participants in the room with the“warm” lighting would indicate the room lighting to be more comfort-able than participants in the room with the “cold” lighting. We alsoexpected that participants in the “warm” lighting condition would per-ceive the room temperature to be warmer (H4.2.2) and more comfortable(H4.2.3), compared to participants in the “cold” lighting condition. Fi-nally, concerning participants’ energy consumption behavior (room tem-perature setting), we expected (H4.2.4) that participants in the roomwith the “warm” lighting would consume less energy in the thermostatprogramming tasks (i.e., setting lower temperature) than participants inthe room with “cold” lighting.

4.3.1 Method

Participants and design

Forty-eight students (average age 22.1 years old, SD = 2.76 ) of Eind-hoven University of Technology (29 male and 19 female) were recruitedby using a local participant database. They participated in the currentexperiment in either the “warm” lighting condition first and then the“cold” lighting condition, or in the “cold” lighting condition first andthen the “warm” lighting condition. Whether a participant received the“warm” or the “cold” lighting condition first, was strictly counterbal-anced. All participants were native Dutch speakers. The experiments

83

✐✐

✐✐

✐✐

✐✐


lasted approximately 45 minutes, and participants received 7 euros fortheir participation.

Experimental procedure

Participants were welcomed in the central hall of the lab building. Eachparticipant was asked to read and sign a consent form stating the gen-eral purpose of the research and their willingness to participate in thisresearch. In the lab room hidden from view for the participant, a ther-mometer was put on a table in the corner of the room to record theaverage room temperature during the experiment.

Before participants entered the lab room, for half of the participantsthe lighting of the room was set to “warm” lighting in the first session(and to “cold” lighting in the second session). For the other half of theparticipants, the lighting in the room was set to “cold” lighting in thefirst session (and to “warm” lighting in the second session). We usedfluorescent ceiling lamps with a color temperature of 2500 Kelvin as“warm” and 6100 Kelvin as “cold” lighting conditions (see Figure 4.4).

(a) (b)

Figure 4.4: Ambient lighting environment used in Study 4.2: (a) “warm” room and (b)“cold” room.

84

✐✐

✐✐

✐✐

✐✐

4.3. Study 4.2

For the first session (see Figure 4.5 for a graphical representation of theexperimental procedure of Study 4.2), participants were asked to read amagazine for a period of ten minutes. For this, we supplied an issue ofa Dutch popular science magazine (named “Quest”) on the table rightnext to the laptop. After ten minutes, the computer indicated that thereading period was over and participants were asked to answer a setof fourteen questions that asked them to evaluate the lab room. Thisquestionnaire consisted of fourteen questions (see the Material sectionbelow).

Lightingcondition1

0min 10

ReadingMagazine

Questionnaire

Thermostattask

15 20 25 35 40 45

Lightingcondition2

Thermostattask

Questionnaire

WhenparticipantoutsidetheLab WhenparticipantinsidetheLab

ReadingMagazine

Figure 4.5: The timeline of the experiment in Study 4.2.

After answering these questions, we assessed energy consumption beha-vior by asking participants to perform a thermostat task (a task identi-cal to the one used in Chapter 2) on the laptop. That is, participantsreceived instructions on how to program a virtual room thermostat in-terface and were asked to program this thermostat for several scenar-ios. After two practice scenarios, participants completed programmingthe thermostat for six experimental scenarios. For each scenario, thescenario itself was first described to the participants (e.g., “You and afriend are going to watch a television show for the next half hour.”) andthen the participant was asked to program the thermostat by setting thetemperature for the room. For each scenario, participants were asked tostrive for two goals: to consume as little energy as possible (for heat-ing the room), and to maintain a temperature level that they deemedas comfortable as possible. After programming the room thermostat forthe six scenarios, participants were told to leave the lab room and to

85

✐✐

✐✐

✐✐

✐✐


take a seat in the hall way outside of the lab room.

Figure 4.6: A simulated programmable heating system interface used to gather data aboutthe participant’s energy consumption behavior in Study 4.2.

After a five-minute break, the second session of the experiment started.Participants were asked to retake their seats inside the lab room. Theexperimental procedure of the second session was completely identical tothe first session, with the exception that the lighting in the laboratorywas manipulated to the other condition before the participant enteredthe room (i.e., out of view of the participants). After completion of thesecond session, participants were asked to answer demographic questions.

Materials

In each of the two sessions, participants were asked to fill out a setof fourteen questions (in Dutch). First of all, to assess participants’perception of the lighting temperature in the room, participants wereasked to answer four questions about the room lighting temperature(“What do you think about the lighting in this room?”, “Do you think thelighting in this room is warm or cold?”, “What feeling does the lightingin this room give you?”, and “What do you think the light in this roomlooks like?”) to which participants could answer by choosing an optionon a 7-point rating scale ranging from 1 very cold to 7 very warm. Weconstructed a reliable (Cronbach’s α = .86) measure of perceived lightingtemperature by averaging participant’s answers to these four questions.

86

✐✐

✐✐

✐✐

✐✐

4.3. Study 4.2

These four questions were used to check whether the room lighting colortemperature manipulation was successful.

Then, to assess participants’ perceived comfort of the room and its light-ing temperature, participants were asked to answer four questions abouttheir evaluation of the room and its lighting temperature (“How com-fortable do you think this room is?”, “How comfortable do you thinkthe lighting in this room is?”, “Do you think the room lighting is com-fortable?”, “Do you think the room lighting is pleasant?”). Participantscould answer these questions by choosing an option on a 7-point rat-ing scale ranging from 1 very uncomfortable to 7 very comfortable. Weconstructed a reliable (Cronbach’s α = .78) measure of room lightingcomfort evaluation by averaging a participant’s answers to these fourquestions.

Next, to assess participants’ perception of the room temperature, parti-cipants were asked to answer two questions about the room temperature(“How does the temperature of this room feel?”, “Does the air in thisroom feel warm or cold?”). Participants could answer these questions bychoosing an option on a 7-point rating scale ranging from 1 very cold to7 very warm. We constructed a reliable (Cronbach’s α = .83) measureof perceived lighting temperature by averaging participant’s answers tothese four questions.

Finally, to assess participants’ perceived comfort of the room tempera-ture, participant were asked to answer four questions about their evalu-ation of the room temperature (“How comfortable do you experience theroom temperature?”, “How comfortable do you think the room temper-ature is?”, “Do you think the room temperature is pleasant?”, “Is theroom temperature agreeable?”). Participants could answer these ques-tions by choosing an option on a 7-point rating scale ranging from 1very uncomfortable/very unpleasant/very disagreeable to 7 very com-fortable/very pleasant/very agreeable. We constructed a reliable (Cron-bach’s α = .95) measure of evaluated room temperature by averaging a

87

✐✐

✐✐

✐✐

✐✐


participant’s answers to these four questions.

4.3.2 Results

Different from Study 4.1 (and also Study 3.1 & 3.2), Study 4.2 containedtwo repeated measures (once measured in the warm lighting condition,and once measured in the cool lighting condition) of each dependentvariable (insert names of the three measures), and these two repeatedmeasures were not statistically independent. Therefore, we used a re-peated measures ANOVA to analyze these results, and not a MANOVAas in Study 4.1. The reason for this is that in a MANOVA, observationsshould be statistically independent (see Field, 2009, p.603). So, to an-alyze the effects of the room lighting color temperature on our variousdependent variables, we submitted our three dependent variables to three2 (color temperature: “warm” vs. “cold”) repeated-measures ANOVAs.

As a manipulation check on whether the room lighting color temperaturemanipulation was successful, results showed that when the room lightingwas set to “warm”, participants indicated that they perceived the roomlighting to be warmer (M = 5.11, SD = .72) than when the room lightingwas set to “cold” (M = 3.15, SD = .84), F (1, 47) = 175.5, p < .001,η2=.79 (see Figure 4.7).

Supporting our first hypothesis (H4.2.1), results showed that when thelighting was set to “warm”, participants indicated the lighting temper-ature in the room to be more comfortable (M = 5.05, SD = .86)than when the lighting was set to “cold” (M = 4.15, SD = 1.00),F (1, 47) = 34.8, p < .001, η2 = .46 (see Figure 4.8).

In addition to the perception of lighting temperature and perceived com-fort of lighting temperature, we were especially interested in the effectson participant’s perception of room temperature and perceived comfortof room temperature.

88

✐✐

✐✐

✐✐

✐✐

4.3. Study 4.2

0

1

2

3

4

5

6

7

thew

armthperceptionof

room

lighting


Figure 4.7: Averaged warmth perception of the room lighting as a result of color temperature(“warm” vs. “cold”) of room lighting in Study 4.2. Error bars attached to each column inthe figure indicate 95% confidence intervals.

0

1

2

3

4

5

6

7

perceivedcomfortofro

om

lighting


Figure 4.8: Averaged perceived comfort of room lighting as a result of color temperature(“warm” vs. “cold”) of room lighting in Study 4.2. Error bars attached to each column inthe figure indicate 95% confidence intervals.

In line with our second hypothesis (H4.2.2), results indicated that whenthe room lighting was set to “warm”, participants perceived the roomtemperature to be warmer (M = 5.05, SD = .86) than when the lightingwas set to “cold” (M = 4.72, SD = .81), F (1, 47) = 9.9, p = .003,η2 = .17 (see Figure 4.9).

However, results provided no evidence in support of our third hypothesis(H4.2.3), that is, they presented no evidence that when room lightingwas set to “warm”, participants perceived the room temperature to bemore (or less) comfortable (M = 5.01, SD = 1.03) than when the roomlighting was set to “cold” (M = 4.83, SD = .90), F (1, 47) = 2.10,

89

✐✐

✐✐

✐✐

✐✐


0

1

2

3

4

5

6

7

warmthofroo

mte

mpe

rature


Figure 4.9: Averaged warmth perception of room temperature as a result of color tempera-ture (“warm” vs. “cold”) of room lighting in Study 4.2. Error bars attached to each columnin the figure indicate 95% confidence intervals.

p = .16, η2 = .04 (see Figure 4.10).

0

1

2

3

4

5

6

7

perceivedcomfortofro

om

tempe

rature


Figure 4.10: Averaged perceived comfort of room temperature as a result of color tempera-ture (“warm” vs. “cold”) of room lighting in Study 4.2. Error bars attached to each columnin the figure indicate 95% confidence intervals.

Finally, results did not support our final hypothesis (H4.2.4). That is,participants in the room with the “warm” lighting did not consume lessenergy in the thermostat programming tasks (i.e., setting a temperature,M = 19.68 ◦C, SD = 1.26) than participants in the room with “cold”lighting (M = 19.75 ◦C, SD = 1.34), F (1, 47) = .87, p = .36, η2 = .02(see Figure 4.11).

90

✐✐

✐✐

✐✐

✐✐

4.3. Study 4.2

0

5

10

15

20

room

tempe

raturese

tting(

o C)


Figure 4.11: Averaged room temperature setting as a result of color temperature (“warm”vs. “cold”) of room lighting in Study 4.2. Error bars attached to each column in the figureindicate 95% confidence intervals.

4.3.3 Discussion

In sum, supporting the findings in Study 4.1, results of Study 4.2 re-vealed the significant effect of color temperature of room lighting on par-ticipant’s perception of room temperature. Moreover, results of Study4.2 also revealed the significant effect of color temperature on people’scomfort experience of the room lighting. However, in contrast to ourexpectations, there were no significant effects on people’s comfort experi-ence of the room temperature and people’s energy conservation behavior(i.e., setting a lower temperature).

Although no main effect was found on the comfort experience of the roomtemperature, interestingly, the comfort experience of the room temper-ature was strongly correlated to the comfort experience of the roomlighting (r = .585, p < .001) and to the temperature perception of theroom temperature (r = −.312, p = .031). That is to say, participantswho rated the room temperature to be comfortable perceived the roomlighting to be also comfortable, but perceived the room temperature tobe cold. Therefore, in this experimental setup, it seemed that comfortexperience of the room temperature could be associated with how peopleperceived the comfort of the room lighting and how people perceived the

91

✐✐

✐✐

✐✐

✐✐


physical room temperature.

Study 4.2 did not provide evidence that room lighting color temperatureinfluenced actual energy saving behavior. That is, results of Study 4.2did not show that when participants were in a room with “warm” lightingsettings, they programmed the room thermostat differently from whenthey were in a room with “cold” lighting settings. The lack of behavioralchanges in our thermostat programming tasks may be due to a varietyof reasons. Crucially, in our lighting lab, the actual room temperaturecould not be controlled and was relatively high (approximately 22.9 ◦Cin Study 4.1, which was conducted in March; approximately 24.6 ◦C inStudy 4.2, which was conducted in June). Indeed, all participants ofStudy 4.2 indicated the room temperature to be quite comfortable, inthe both “warm” and “cold” lighting conditions. Thereby, participantsmay have lacked the motivation to change their behavior to attain acomfortable status (change the settings of the thermostat to a higher orlower temperature). In contrast, when people are in an uncomfortablestatus (e.g., feel slightly cold or hot), people might consume more energy(by increasing room temperature) in the “cold” lighting condition thanpeople in the “warm” lighting condition (or vice-versa when they feelhot). Relatedly, earlier research (Nakamura & Oki, 2000) suggested thatpreference in the color temperature of general room lighting may varydepending on the previous experiences of air temperature before enteringthe room. So, future research might find the effect of room lighting colortemperature on behavior change by diminishing but also by increasingthe physical temperature in the experimental room, as long as it becomessomewhat uncomfortable.


In this chapter, we extended our understanding of the psychologicalmechanisms of effective ambient persuasive system lighting by exploring

92

✐✐

✐✐

✐✐

✐✐


the experiential processes. More specifically, the current research inves-tigated the effect of color temperature of room lighting (as a Experience-Based form of persuasive technology) on participants’ perceptions ofroom lighting temperature, their experiences of the comfort related tothe room lighting temperature and related to the room temperature,and also their tendency to change room temperature and participant’stemperature-setting behavior.

Results of both studies indicated that our manipulation of room lightingwas effective in showing that participants in the “warm” lighting condi-tion perceived the lighting in the room to be warmer than participantsin the “cold” lighting condition. More importantly, results of our secondstudy supported our expectation that our manipulation of room light-ing influenced participants’ judgment of the temperature in the room.That is, in Study 4.2, participants in the room with the “warm” lightingperceived the room temperature to be warmer than participants in theroom with the “cold” lighting.

However, this effect of lighting color temperature on participants’ judg-ment of the temperature in the room was only found when the lightingconditions were manipulated within participants (as in Study 4.2), andwas not found when these conditions were manipulated between partici-pants (as in Study 4.1). Although in these two studies, the effects showedthe same direction, the effect of color temperature on participant’s per-ception of room temperature in Study 4.1 did not reach a significantlevel5. Probably, differences between participants (and their judgmentsof room temperature) caused more noise in Study 4.1 on our dependentvariable of room temperature perception making it more difficult to findeffects of our manipulation of lighting color temperature on that variablein Study 4.1.

5We conducted a power analysis before executing Study 4.1. This analysis showed that to attain80% power and when expecting a medium sized effect, the current number of participants is sufficient.

93

✐✐

✐✐

✐✐

✐✐


Interestingly, results of Study 4.1 also suggested that our manipulationof room lighting influenced participant’s tendency to adjust the roomtemperature. That is, in Study 4.1, participants in the “warm” lightingcondition expressed a stronger tendency to decrease the room temper-ature than participants in the “cold” lighting condition. This evidenceindicated that user tendency to save energy and related energy-savingbehavior (e.g., setting a lower room temperature) could be enhancedwhen people were in a room with “warm” lighting settings.

Furthermore, results of Study 4.2 showed that when the room lightingwas set to “warm”, participants indicated the lighting temperature in theroom to be more comfortable than when the lighting was set to “cold”.However, results of Study 4.2 provided no evidence that our manipula-tion of room lighting influenced participant’s comfort experience of theroom itself. One reason might be that comfortable room temperaturemay vary considerably among individuals. Such differences between par-ticipants (caused by e.g., just having had a meal, taken a shower orwearing a sweater, see in Heijs & Stringer, 1988) may have caused noisethat made it more difficult to find effects of our manipulation of roomlighting temperature. Future research investigating these effects couldbe conducted in real homes as a between-subjects design, additionallyincreasing the external validity of the experiment by collecting the dataover an extended period of time during normal usage of lights.

In sum, our results from the current chapter are promising by suggestingthat incorporating thermal experiences in persuasive technology seems tobe a potential strategy to influence people’s temperature-related behav-iors. Particularly, people who are exposed to a “warm” lighting wouldperceive the room temperature to be warmer and show stronger tenden-cies to decrease room temperature. However, as we discussed above, ourresearch in Chapter 4 did not reveal the significant effects of color tem-perature on people’s comfort experience of the room temperature andpeople’s energy conservation behavior. On the basis of these findings,

94

✐✐

✐✐

✐✐

✐✐


additional research questions can be addressed. Of special interest couldbe the mediation factors that could help explain the effect of color tem-perature on people’s energy conservation behaviors.

Taken as a whole, these findings contribute to the scientific literaturecrucial evidence that the effectiveness of persuasive technology couldalso be enhanced by targeting thermal experiences via mainly bodily andperceptual processes. This kind of persuasive technology can help usersdecrease their energy consumption, by using a category of persuasivestrategies that targets user thermal experiences.

95

✐✐

✐✐

✐✐

✐✐


96

✐✐

✐✐

✐✐

✐✐

5 | General discussion

The work presented in this dissertation aimed to study lighting asa communication medium embedded in the ambient environment

with the intention to facilitate behavior change. As described in thegeneral introduction of this dissertation, ambient lighting has specificqualities that makes it particularly suitable to serve as effective ambi-ent persuasive technology and it helps users to attain their goals whileneeding only a limited amount of cognitive attention (see e.g., Maan etal., 2011; Rogers et al., 2010; Wilson et al., 2013). Still, the psychologi-cal mechanisms behind the effectiveness of ambient lighting persuasivetechnology remain largely unclear. The current thesis focused on empiri-cally exploring the cognitively-efficient processes underlying this effectiveambient lighting persuasive technology. In particular, we studied the as-sociative processes (Chapter 2 & 3), and experiential processes (Chapter4) of the persuasive technology that uses ambient lighting.

In the current chapter, we first discuss the findings and contributions ofthe studies that were performed in section 5.1 and section 5.2. Then,in section 5.3, we specify some contributions to persuasive design andsociety. Next, in section 5.4, we identify general limitations across allempirical studies and discuss the directions for future research. In section5.4, we will briefly address the ethical considerations of using ambient

97

✐✐

✐✐

✐✐

✐✐

Chapter 5. General discussion

lighting persuasive technology to change people’s behavior. Finally, insection 5.5, we present the general conclusion of this thesis.

5.1 Understanding the associative processes

As described in the general introduction of this thesis, earlier researchsuggested that the characteristics of light (e.g., its color) serving as am-bient persuasive technology, make it possible to create behavior change(e.g., energy conservation), which require only minimal attention andcognitive effort by the user while still being effective (see also e.g., Kimet al., 2010; Rogers et al., 2010; Maan et al., 2011). However, the psy-chological mechanisms underlying the effectiveness of processing ambientlighting persuasive technology remain largely unclear. We argued thatthe color associations (see e.g., Otsuka & Kawaguchi, 2009) and color’scontexts (see e.g., Elliot & Maier, 2012) of ambient lighting persuasivetechnology could help users easily process the feedback messages. Theseprocesses are distinguished in dual-process models of cognitive processes(e.g., the reflective-impulsive model as proposed by Strack and Deutsch,2004) as impulsive processes, consisting mainly of associative processes.These associations slowly develop when people cognitively create rep-resentations of the environment, and change only gradually throughlearning (see e.g., Johnson & Hirst, 1993; McClelland, McNaughton, &O’Reilly, 1995; Smith, 1998; Smith & DeCoster, 2000). Therefore, thefirst research goal in this thesis was to understand the associative proc-esses of persuasive ambient lighting system. We studied the efficiency ofthese associative processes by investigating two characteristics of ambi-ent lighting feedback: the role of color associations of ambient lightingfeedback (Chapter 2), and the role of color context (the context in whichcolors were presented as ambient feedback, in Chapter 3).

98

✐✐

✐✐

✐✐

✐✐

5.1. Understanding the associative processes

The role of color associations

In Chapter 2 we studied the effects of color associations of ambient light-ing feedback on the ease of processing of feedback messages, and on theeffectiveness of this ambient lighting persuasive technology for influenc-ing energy conservation behavior. We argued that colors of ambientlighting feedback could carry meanings that have strong and consistentassociations with energy consumption, which could help users easily pro-cess feedback messages, and thereby enhance the effectiveness of ambientpersuasive technology for influencing energy conservation behavior. Thetwo studies described in Chapter 2 tested the color association strength(strong vs. weak) and the color association consistency (consistent vs.inconsistent) of ambient lighting feedback. In line with our expecta-tions, the results of Chapter 2 indicate that ambient lighting feedbackwith colors that have strong and consistent associations with energy con-sumption would be easily processed (i.e., the processing time of feedbackmessages would not be slowed down by additional cognitive load task),and would increase the persuasive power for stimulating energy conserva-tion behavior. So, both studies presented evidence that ambient lightingpersuasive technology that uses colors with strong and consistent energyassociations can effectively influence user energy consumption behaviorwhile requiring little cognitive capacities.

These findings thereby contributed to uncovering the psychological mech-anisms behind ambient lighting feedback effects by showing the associa-tive processes through color associations.

As argued in the general introduction, there is a general cognitive limiton the abilities to do mental work (for a cognitive capacity overview seeSmyth, 1994). Our findings showed that, even though people are underhigh cognitive load, color associations can help people to easily detectthe target message for the reason that color information is linking to pre-existing associations. Beyond this finding, we showed the importance of

99

✐✐

✐✐

✐✐

✐✐


color associations for behavior change interventions through persuasivetechnology. In addition to earlier research on ambient lighting feedback(e.g., Maan et al., 2011; Wilson et al., 2013), our studies extend theresearch on ambient persuasive technology by investigating the contentof the color’s associations and the nature of the associative process (seeStrack & Deutsch, 2004).

The role of color contexts

To understand the role of color associations in associative processes, wealso argued that the context in which these colors are presented is crucial.That is, associations of color seem highly context dependent as indicatedby Color-In-Context theory (Elliot & Maier, 2012, see Chapter 3 of thisthesis).

Therefore, in Chapter 3, to extend our insight in the associative process-ing of colored ambient lighting feedback, we investigated the influenceof a color’s context on user’s interpretation of color’s feedback messages,and furthermore, on the effectiveness of this feedback (for influencingenergy conservation behavior). We argued that color-based feedback ishighly context dependent (at least with respect to energy-related infor-mation). More specifically, a consistent context in which color-basedambient persuasive technology is presented could shape the associationsof colors and make the feedback messages more easily understandable.Furthermore, this consistent context could increase the persuasive powerof ambient persuasive technology to promote energy conservation beha-vior. We expected that presenting colors (e.g., red) in an energy-relatedcontext makes feedback messages (e.g., high energy consumption level)more clear and can enhance the correct interpretation of the color (e.g.,warm), compared to presenting colors in an energy-unrelated context.Additionally, presenting colors in energy-related contexts could increasethe effectiveness of the color-based ambient persuasive technology, lead-

100

✐✐

✐✐

✐✐

✐✐

5.1. Understanding the associative processes

ing to stronger energy conservation behavioral changes. These hypothe-ses were tested in two studies. Findings of both studies supported ourmain hypotheses. That is, the color red in an energy-related contextwas interpreted to be warmer and as related to a higher energy con-sumption level than red in an energy-unrelated context. Also, present-ing color-based ambient feedback in an energy-related context promotedmore energy conservation behavior than presenting this feedback in anenergy-unrelated context.

Thereby, Chapter 3 extended our insight into the associative processesof ambient lighting persuasive technology by making clear that a fit-ting context in which ambient persuasive technology is presented canshape the associations of colors (see also Elliot & Maier, 2012), whichmakes the feedback messages directly understandable, and furthermore,influences user’s interpretation and the effectiveness of this ambient per-suasive technology.

In addition, although colors are widely used as feedback in ambient per-suasive technology (e.g., Jahn et al., 2010; Merkus, 2012; Wilson et al.,2013) and also highly context dependent (Elliot & Maier, 2012), theempirical evidence for the influence of color’s context on color’s interpre-tations and the effectiveness of persuasive technology is still limited. Thecurrent studies make innovative contributions to the body of persuasivetechnology literature on how contexts in which ambient stimuli (e.g.,colors) are presented, shape the associative processes and furthermoreinfluence the effectiveness of this persuasive technology.

In general, Chapter 2 and Chapter 3 reveal the importance of associativeprocesses consisting of associative links (e.g., presenting red lighting asenergy feedback on a radiator) that can help users easily process feedbackmessages, and thereby enhance the effectiveness of ambient persuasivetechnology. In particular, our research addresses the important rolesof color associations and color’s contexts of ambient lighting persuasivetechnology, and points out that the ease of processes of persuasive tech-

101

✐✐

✐✐

✐✐

✐✐


nology can be enhanced by utilizing these associative links.

These findings advance the research domain of ambient persuasive tech-nology by examining the cognitively-efficient associative processes (seee.g., Petty & Cacioppo, 1986; Strack & Deutsch, 2004) underlying the ef-fectiveness of ambient lighting persuasive technology. In addition, thesefindings support earlier theoretical models for predicting human beha-vior, such as Motivation-Opportunity-Abilities Model by (Olander &ThØgersen, 1995) or Behavioral Model by Fogg (2009), by indicatingthat ambient lighting can boost user’s motivation to change behavior(e.g., ambient lighting feedback motivates people by providing the con-sequences of their current behavior).

5.2 Understanding the experiential proc-esses

As described in the general introduction, for understanding the underly-ing mechanisms behind the effectiveness of ambient lighting persuasivetechnology, associative processes are one part of the impulsive domainof cognitive processes. Importantly, processes in the impulsive systemmay also be accompanied by experiential processes (Strack & Deutsch,2004). These experiential processes are also associative in nature, butinvolve less cognition (e.g., Dijksterhuis et al., 2000; Zajonc, 1980), com-pared to associative processes in the impulsive domain of the cognitivesystem. For example, a person may have a visual perception of lightnessor darkness, a pleasant or unpleasant feeling, or the experience of pain orfamiliarity without knowing the concepts or categories of light, pleasant-ness, pain or familiarity. Therefore, we extended our understanding ofthe psychological mechanisms behind the effectiveness of ambient light-ing persuasive technology by exploring the experiential processes. So,the second research goal of this thesis was to understand the experien-

102

✐✐

✐✐

✐✐

✐✐

5.2. Understanding the experiential processes

tial processes of ambient lighting persuasive technology.

The role of color temperature experiences

In Chapter 4, we extended our understanding of the underlying mecha-nisms of ambient lighting persuasive technology by exploring the experi-ential processes, instead of associative process. Therefore, we proposed anew perspective on ambient lighting persuasive technology, which mightinfluence people through thermal experience. In this chapter, we in-vestigated the effects of color temperature of ambient lighting (e.g., byemitting “warm” or “cold” light) on user’s room temperature percep-tion, their willingness to change the room temperature, and their energyconservation behavior. We argued that color temperature of ambientlighting (e.g., “warm” light) would influence people’s thermal experience(e.g., perceiving the room temperature to be high), their willingnessto change the room temperature and their energy conservation beha-vior (by setting a lower room temperature). We expected that roomtemperature in the lighting condition with a lower color temperature(i.e., “warm” light) would be perceived higher and more comfortable,compared to room temperature in the lighting condition with a highercolor temperature (“cold” light). Additionally, using “warm” lighting inambient lighting persuasive technology would enhance the effectivenessof Experience-Based persuasive technology (i.e., people would expressa stronger motivation to decrease the room temperature, and influencetheir energy conservation behavior by setting a lower room temperature),compared to using “cold” lighting in ambient lighting persuasive tech-nology. In line with our expectations, our findings showed that colortemperature of room lighting can influence user’s perception of roomtemperature, and can influence user’s willingness to change the roomtemperature. However, our second study in this chapter provided no ev-idence that these effects of room lighting color temperature also extendedto actual energy conservation behavior.

103

✐✐

✐✐

✐✐

✐✐


As discussed in Chapter 4, the lack of behavioral changes in our ther-mostat programming tasks (in Study 4.2) may have been due to a va-riety of reasons, such as the physical room temperature in our lightinglab could not be controlled and was relatively high (participants indeedindicated the room temperature to be quite comfortable, in the both“warm” and “cold” lighting conditions). Thereby, participants may havelacked the motivation to change their behavior to attain a comfortablestatus. Whereas when people are in an uncomfortable status (e.g., feelslightly cold or hot), they might consume more energy (by increasingroom temperature) in the “cold” lighting condition than people in the“warm” lighting condition (or vice-versa when they feel hot). In addi-tion to the physical room temperature in our lab, previous experiencesof air temperature before entering the lab may also influence people’spreference in the color temperature of general room lighting (Nakamura& Oki, 2000).

In line with earlier research indicating that thermal experience may havean influence on people’s (energy consumption) attitude and behavior(Becker et al., 1981; Heijs & Stringer, 1988), the current research suggeststhat ambient persuasive technology (e.g., manipulating color tempera-ture of ambient lighting) could influence people’s thermal experience, andmight enhance its persuasive effectiveness. Thereby, findings of Chapter4 contribute to the field of ambient persuasive technology by proposinga new perspective on ambient persuasive technology: experience-basedpersuasive technology. In the research area of persuasive technology, va-rious models have been proposed that capture the working mechanismsof persuasive technology (e.g., the three roles of persuasive technology,Fogg, 2002; see also Oinas-Kukkonen & Harjumaa, 2009). Of these mod-els, only the medium role for persuasive technology (as proposed by Fogg,2002) includes the concept of experience. That is, persuasive technologiescan use both interactivity and narrative to create persuasive experiencesthat support rehearsing a behavior, empathizing, or exploring causal re-

104

✐✐

✐✐

✐✐

✐✐

5.3. Contributions to design and society

lationships. However, we argue that the current research is the first tostress and provide evidence for the importance of bodily thermal experi-ences for influencing human behavior and thinking. So, using a categoryof persuasive strategies that targets people’s thermal experiences (seeBecker et al., 1981; Heijs & Stringer, 1988) can enhance the experientialprocesses of ambient lighting persuasive technology. The investigationsof these experiential processes in our research can be regarded as the firstpiece of knowledge on exploring the ambient persuasive technology thatis designed to influence people subconsciously while remaining effective.

5.3 Contributions to design and society

Besides the scientific contributions discussed above, the work presentedin this thesis also makes contributions relevant for designers to createambient persuasive technology by utilizing lighting. First, our researchextends the forms and medium of ambient persuasive technology by usingambient lighting, instead of using traditionally physical or screen-basedambient display (Pousman & Stasko, 2006; Streppel, 2015). Lighting asan ambient medium provides a great opportunity for designers withinthe context of persuasion, in which different shapes and sizes of ambi-ent persuasive display are possible from small sculptures, to wall-sizedinstallations, or even to the whole physical environment. In addition,our research provides a new approach for stimulating behavioral changethrough (bodily thermal) experiences when designers create ambient per-suasive technology. As pointed out by Aarts and colleagues (2007), theemphasis on perceived user value in the design of ambient intelligentsystems has changed from usability towards creating user experiencessuch as presence, connectedness, and immersion. Our research can beregarded as a first step to further ambient persuasive design, in whichambient persuasive systems can influence people through bodily experi-ence.

105

✐✐

✐✐

✐✐

✐✐


Finally, the work presented in this thesis makes particular contributionsrelevant for environmental sustainability. Regarding the major environ-mental risks of CO2-emissions and climate change that policy-makersand psychologists face (e.g., Gardner & Stern, 1996), a purely technolo-gical approach to reduce energy consumption often leads to disappoint-ing results due to changes in user behavior (such as rebound effects, seeMidden et al., 2007). Utilizing persuasive technology to promote sustai-nable behavior of people, whether as individuals, as households, or associeties, has often been investigated (see Midden et al., 2008; Steg &Vlek, 2009). Interestingly, different from many cognitively-demandingpersuasive technologies designed to encourage energy conservation be-havior (e.g., numerical feedback about energy consumption, kWh), thework in this dissertation proposes to employ ambient lighting in persua-sive technology, which is an innovative and easy approach for people tofollow through color associations of ambient lighting. For policy makerswho help constructing people’s environments, our research may extendtheir tool kits (e.g., Schneider & Sidney, 2009; Weaver, 2009) by incorpo-rating ambient lighting persuasive technology. This additional tool mighthelp effectively change citizen behavior. Also, policy makers could pro-mote to create an ambient environment with persuasive lighting that canbe integrated unobtrusively in our daily routine life, and may interveneeven on a level of automatic behavior through for example influencingthermal experiences by manipulating the color temperature of ambientlight. Therefore, our findings contribute to the domain of environmentalsustainability by offering an easy, unobtrusive, while effective solution tomake the behaviors, whether as individuals or as societies, sustainable.

106

✐✐

✐✐

✐✐

✐✐

5.4. Limitations and directions for future research

5.4 Limitations and directions for futureresearch

Throughout all empirical studies in this dissertation, we already identi-fied specific limitations and potential areas for future research. In ad-dition, we may identify a number of more general limitations across allstudies in the current dissertation.

All participants presented in our studies were native Dutch. On the onehand, these culture-consistent samples strongly supported our argumen-tation that for example using lighting feedback with colors that havestrong associations with energy consumption in an energy-related con-text could help users easily process the feedback messages. On the otherhand, these color or context associations with energy consumption mightvary among different cultures. These cultural differences may further in-fluence people’s perceptual judgments (Roberson, Davies, & Davidoff,2000). Therefore, although the essence of the associative mechanismmay remain similar, exploring associative processes of (ambient) persua-sive technology in different cultures could be a valuable topic for futureresearch.

Also, in most of our studies, we investigated the effects of ambient light-ing persuasive technology on temperature-setting behavior (e.g., in Study2.1, 2.2, 3.2 and 4.2). However, in our experiments, participants per-formed the temperature-setting behavioral tasks in the lab in which wedid not manipulate the physical room temperature according to the tem-perature settings. In Chapter 2, we specifically discussed the gap betweenthe desired high-comfort experience and the current low-comfort experi-ence (see also Becker et al., 1981; Heijs & Stringer, 1988). More gene-rally, people might have a strong willingness to change their behaviorswhile experiencing discomfort (i.e., when people are outside their comfortzone, see Gauthier, 2011; Ohnaka et al., 1994). Future research could

107

✐✐

✐✐

✐✐

✐✐


investigate the behavioral change of participants by assigning them toan environment with a temperature outside of the participant’s thermalcomfort zone (as set by ANSI/ASHRAE standard 55, 2013) while theyconduct the temperature-setting tasks. Also, future research could in-vestigate the factors that may also affect the thermal comfort zone, suchas clothing insulation level, activity level, and demographics (De Dear,Brager, Reardon, & Nicol, 1998).

Furthermore, as argued in the general introduction of this dissertation,associative processes are one part of the impulsive domain of the cog-nitive processes. These processes in the impulsive system may also beaccompanied by experiential processes (see e.g., Strack & Deutsch, 2004;Maas & Van den Bos, 2009). However, earlier research (e.g., Gawronski& Bodenhausen, 2006; Reeder & Pryor, 2008) mostly focused on associa-tive processes (and cognitively-demanding processes such as reflectivelypropositional processes). The current research studied not only asso-ciative processes, but also the role of thermal experiences, and therebyextended our knowledge on the importance of experiential processes (as acognitive process in the impulsive domain). Our current research can beconsidered a starting point for further research to study the relationshipbetween associative and experiential processes.

5.5 Ethical considerations

In the general introduction of this dissertation, we already addressed thatpersuasive technology aims at inducing the voluntary change of behaviorswithout taking away control from its users (see also Fogg, 1999, 2002).Despite the obvious good intentions behind persuasive technologies likeserving social values (environmental sustainability or senior care, see e.g.,Ham & Spahn, 2015) or goals of the user (healthy lifestyle or comfortexperiences, see e.g., Chatterjee and colleagues, 2009), many, if not all,persuasive technologies could have unintended effects too (Berdichevsky

108

✐✐

✐✐

✐✐

✐✐

5.5. Ethical considerations

& Neuenschwander, 1999). From an ethical perspective, it is imperativethat humans are not deceived or coerced by persuasive technology todo something they would not want to do, and humans should be awareof final goals of the persuasive technology and consequences of theirbehavioral changes (see Berdichevsky & Neuenschwander, 1999; Smids,2012; Spahn, 2012; Verbeek, 2006).

Ambient lighting persuasive technology, as proposed in this thesis, canbe categorized as part of the more encompassing phenomenon of tech-nological mediation. This concept (as developed by Ihde 1990; Latour1994; see also Verbeek 2010) indicates the ways in which technologiesinevitably and often implicitly help to shape human actions and per-ceptions by establishing relations between users and their environment –sometimes even without users being aware of this persuasive system. Ourresearch shows that ambient lighting persuasive technology has potentialto influence user behavior through cognitively-efficient processes or evensolely based on their thermal experience, like the thermal experiences inour case. This influence while requiring minimal attention could raise avariety of ethical issues, for instance, whether could we design ambientlighting persuasive technology that persuades users to decrease the roomtemperature beyond user’s awareness (e.g., they were exposed to a roomwith ’warm’ lighting) or while they did not notice the final goals of per-suasion, such as, saving energy (see Berdichevsky & Neuenschwander,1999)?

In addition, our research (e.g., Study 2.1) employed colored ambientlighting (by using LED wall washers) to provide energy consumptionfeedback. Participants were exposed to an intense and saturated redlighting in the entire room when they set a relatively high room temper-ature, which may cause eye-strain or headache (e.g., Main, Vlachoniko-lis, & Dowson, 2000). Persuasive designers themselves would not bewilling to be exposed to an unpleasant condition while being persuadedto change behavior (see “Golden Rule” in Flew, 1979). Therefore, in-

109

✐✐

✐✐

✐✐

✐✐


ducing such unintended effects, for example, physical discomfort, whileinfluencing a person’s behavior cannot be regarded as being very ethical.These unintended effects of applying ambient lighting in persuasive tech-nology should be ethically discussed when designers create more appliedpersuasive systems.

5.6 Conclusion

The primary aim of the work in this dissertation was to improve ourunderstanding of the psychological mechanisms underlying effective am-bient persuasive lighting for facilitating energy-conservation behavior.By evaluating the effects of color associations of ambient lighting, the ef-fects of context in which ambient lighting was presented, and the effectsof color temperature on energy consumption behavior, we deepened ourinsights of the psychological processes underlying the effectiveness of pro-cessing ambient lighting persuasive technology. Also, we have providedempirical evidence that effective persuasive technology can use ambi-ent lighting as a communication medium with the intention to stimulatebehavior change while remaining cognitively efficient. In addition, weextended the role of ambient stimuli within the context of persuasivetechnology by embedding ambient lighting feedback in our environment,and meanwhile, proposed an experience-based perspective on persuasivetechnology through ambient lighting.

In sum, the findings from the work presented in the current thesis re-veal associative and experiential processes of effective ambient lightingpersuasive technology. The value of these psychological processes is notlimited to using ambient lighting as persuasive technology in the domainof environmental sustainability, but can easily be generalized to per-suasive communication to facilitate all kinds of behavior change (e.g.,healthy lifestyle) and help to solve societal issues (e.g., sustainability).The Greek started using lighting as a communication tool over 2000 years

110

✐✐

✐✐

✐✐

✐✐

5.6. Conclusion

ago. We can now observe how new lighting technology will help creatingnew tools for persuasive communication.

111

✐✐

✐✐

✐✐

✐✐

General discussion

112

✐✐

✐✐

✐✐

✐✐

References

Aarts, E., Markopoulos, P., & De Ruyter, B. (2007). The persuasiveness of ambientintelligence. In Security, privacy, and trust in modern data management (pp. 367–381). Springer Berlin Heidelberg.

Aarts, M., van Hoof, J., & Schoutens, A. (2007). High colour temperature lightimproves restless behaviour and circadian rhythm of institutionalised older adultswith dementia. In Proceedings of the symposium light, performance and quality oflife (pp. 56–60).

Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A review of inter-vention studies aimed at household energy conservation. Journal of EnvironmentalPsychology, 25 (3), 273–291.

Ainsworth, R. A., Simpson, L., & Cassell, D. (1993). Effects of three colors in an officeinterior on mood and performance. Perceptual and Motor Skills, 76 (1), 235–241.

Ali, M. (1972). Pattern of eeg recovery under photic stimulation by light of differentcolors. Electroencephalography and Clinical Neurophysiology, 33 (3), 332–335.

Amit, R., & Zott, C. (2001). Value creation in e-business. Strategic managementjournal, 22 (6-7), 493–520.

Arroyo, E., Bonanni, L., & Selker, T. (2005). Waterbot: exploring feedback andpersuasive techniques at the sink. In Proceedings of the SIGCHI conference onHuman factors in computing systems (pp. 631–639). ACM.

Asch, S. E. (1951). Effects of group pressure upon the modification and distortion ofjudgments. Groups, leadership, and men, 222–236.

Aspinall, P., & Dewar, J. (1980). Lighting and perceived guilt. Lighting Researchand Technology, 12 (3), 140–143.

113

✐✐

✐✐

✐✐

✐✐

References

Badia, P., Myers, B., Boecker, M., Culpepper, J., & Harsh, J. (1991). Bright lighteffects on body temperature, alertness, eeg and behavior. Physiology & behavior ,50 (3), 583–588.

Bandura, A., & McDonald, F. J. (1963). Influence of social reinforcement and thebehavior of models in shaping children’s moral judgment. The Journal of Abnormaland Social Psychology, 67 (3), 274.

Bargh, J. A. (1997). The automaticity of everyday life. In Advances in socialcognition, vol. 10 (pp. 1–61).

Bargh, J. A., & Chartrand, T. L. (1999). The unbearable automaticity of being.American psychologist, 54 (7), 462–479.

Bargh, J. A., & Williams, E. L. (2006). The automaticity of everyday life. CurrentDirections in Psychological Science, 15 (1), 1–4.

Baron, R. A. (1997). The sweet smell of... helping: Effects of pleasant ambient fra-grance on prosocial behavior in shopping malls. Personality and Social PsychologyBulletin, 23 , 498–503.

Baron, R. A., & Bronfen, M. I. (1994). A whiff of reality: Empirical evidenceconcerning the effects of pleasant fragrances on work-related behavior1. Journal ofApplied Social Psychology, 24 (13), 1179–1203.

Baron, R. a., Rea, M. S., & Daniels, S. G. (1992). Effects of indoor lighting (il-luminance and spectral distribution) on the performance of cognitive tasks andinterpersonal behaviors: The potential mediating role of positive affect. Motiva-tion and Emotion, 16 (1), 1–33.

Baron, R. A., & Thomley, J. (1994). A whiff of reality positive affect as a potentialmediator of the effects of pleasant fragrances on task performance and helping.Environment and Behavior , 26 (6), 766–784.

Becker, L. J., & Seligman, C. (1978). Reducing air conditioning waste by signalingit is cool outside. Personality and Social Psychology Bulletin, 4 (3), 412–415.

Becker, L. J., Seligman, C., Fazio, R. H., & Darley, J. M. (1981). Relating attitudesto residential energy use. Environment and Behavior , 13 (5), 590–609.

Belcher, M., & Kluczny, R. (1987). The effects of light on decision making: Someexperimental results. In Proceeding CIE 21st Session, Venice (Vol. 1, pp. 354–357).

114

✐✐

✐✐

✐✐

✐✐

References

Bellizzi, J. A., & Hite, R. E. (1992). Environmental color, consumer feelings, andpurchase likelihood. Psychology & marketing, 9 (5), 347–363.

Bennett, C. A., & Rey, P. (1972). What’s so hot about red? Human Factors: TheJournal of the Human Factors and Ergonomics Society, 14 (2), 149–154.

Berdichevsky, D., & Neuenschwander, E. (1999). Toward an ethics of persuasivetechnology. Communications of the ACM , 42 (5), 51–58.

Berlo, D. K. (1960). Communication: An introduction to theory and practice. NewYork: Holt, Rinehart, and Winston.

Berry, P. C. (1961). Effect of colored illumination upon perceived temperature.Journal of Applied Psychology, 45 (4), 248.

Bhattacherjee, A., & Sanford, C. (2006). Influence processes for information techno-logy acceptance: An elaboration likelihood model. MIS quarterly, 805–825.

Boyce, P. R. (2014). Human factors in lighting. CRC Press.

Brainard, G. C., Rollag, M. D., & Hanifin, J. P. (1997). Photic regulation of melatoninin humans: ocular and neural signal transduction. Journal of Biological Rhythms,12 (6), 537–546.

Briand, G., & Pras, B. (2010). Lighting and perceived temperature: energy-savinglevers to improve store evaluations? In M. Campbell, J. Inman, & R. Pieters(Eds.), NA-Advances in Consumer Research Volume 37 (pp. 312–318). Duluth,MN : Association for Consumer Research.

Busch, M., Schrammel, J., Flu, M. A., Kruijff, E., & Tscheligi, M. (2012). Persuasivestrategies report. CURE-Center for Usability Research & Engineering.

Candas, V., & Dufour, A. (2005). Thermal comfort: multisensory interactions?Journal of physiological anthropology and applied human science, 24 (1), 33–36.

Chaiken, S., & Trope, Y. (Eds.). (1999). Dual-process theories in social psychology.Guilford Press.

Chatterjee, S., & Price, A. (2009). Healthy living with persuasive technologies:framework, issues, and challenges. Journal of the American Medical InformaticsAssociation, 16 (2), 171–178.

115

✐✐

✐✐

✐✐

✐✐

References

Cialdini, R. B. (2007). Influence: The Psychology of Persuasion. New York: HarperCollins.

Consolvo, S., Everitt, K., Smith, I., & Landay, J. A. (2006). Design requirementsfor technologies that encourage physical activity. In Proceedings of the SIGCHIConference on Human Factors in Computing Systems (pp. 457–466). ACM.

Consolvo, S., Klasnja, P., McDonald, D. W., & Landay, J. A. (2009). Goal-settingconsiderations for persuasive technologies that encourage physical activity. In Pro-ceedings of the 4th international conference on persuasive technology (p. 8). ACM.

Cope, E. M. (1867). An introduction to aristotle’s rhetoric. Wm. Brown ReprintLibrary.

Czeisler, C. A., Kronauer, R. E., Allan, J. S., Duffy, J. F., Jeweit, M. E., Brown, E. N.,& Ronda, J. M. (1989). Bright Light Induction of Strong (Type 0) Resetting ofthe Human Circadian Pacemaker. Science, 244 , 1328.

Darby, S. (2006). The Effectiveness of Feedback on Energy Consumption: A Reviewfor DEFRA of the Literature on Metering, Billing and Direct Displays. Technicalreport, Environmental Change Institute, University of Oxford.

Davis, J. (2008). Towards Participatory Design of Ambient Persuasive Technology.Presented at Persuasive Pervasive Technology and Environmental SustainabilityWorkshop at Pervasive 2008: Sydney, Australia.

De Dear, R. J., Brager, G. S., Reardon, J., & Nicol, F. (1998). Developing an adaptivemodel of thermal comfort and preference/discussion. ASHRAE transactions, 104 ,145.

Delay, E. R., & Richardson, M. A. (1981). Time estimation in humans: Effects ofambient illumination and sex. Perceptual and Motor Skills, 53 (3), 747–750.

Delia, J. G. (1987). Communication research: A history. In C. Berger & S. Chaffee(Eds.), Handbook of communication science (pp. 20–98). Beverly Hill, CA: SagePublication.

De Ruyter, B., & Aarts, E. (2004). Ambient intelligence: visualizing the future. InProceedings of the working conference on Advanced visual interfaces (pp. 203–208).ACM.

116

✐✐

✐✐

✐✐

✐✐

References

Dijksterhuis, A., Bargh, J. A., & Miedema, J. (2000). Of men and mackerels: Atten-tion, subjective experience, and automatic social behavior. In H. Bless & J. Forgas(Eds.), The message within: The role of subjective experience in social cognitionand behavior (pp. 37–51). Philadephia: Psychology Press.

Elliot, A. J., & Maier, M. A. (2012). Color-in-context theory. Advances in experi-mental social psychology, 45 , 61–125.

Elliot, A. J., Maier, M. A., Moller, A. C., Friedman, R., & Meinhardt, J. (2007).Color and psychological functioning: The effect of red on performance attainment.Journal of Experimental Psychology: General, 136 (1), 154–168.

Elliot, A. J., & Niesta, D. (2008). Romantic red: red enhances men’s attraction towomen. Journal of personality and social psychology, 95 (5), 1150–1164.

Elliot, A. J., Niesta Kayser, D., Greitemeyer, T., Lichtenfeld, S., Gramzow, R. H.,Maier, M. A., & Liu, H. (2010). Red, rank, and romance in women viewing men.Journal of Experimental Psychology: General, 139 (3), 399–417.

Energy orb @Online. (2016). Retrieved from http://www.ambientdevices.com

Etnier, J. L., & Hardy, C. J. (1997). The effects of environmental color. Journal ofSport Behavior , 20 (3), 299–312.

Fanger, P., Breum, N., & Jerking, E. (1977). Can colour and noise influence man’sthermal comfort? Ergonomics, 20 (1), 11–18.

Field, A. (2009). Discovering statistics using spss (third edition). Sage publications.

Fischer, C. (2008). Feedback on household electricity consumption: A tool for savingenergy? Energy Efficiency, 1 (1), 79–104.

Flew, A. (1979). A dictionary of philosophy. London: Pan Books in association withThe MacMillan Press.

Fogg, B. (2002). Persuasive technology: Using computers to change what we thinkand do. CA: Morgan Kaufmann.

Fogg, B., & Hreha, J. (2010). Behavior wizard: a method for matching targetbehaviors with solutions. In Proceedings of the 5th International Conference onPersuasive technology (pp. 117–131). Springer Berlin Heidelberg.

117

✐✐

✐✐

✐✐

✐✐

References

Fogg, B. J. (1999). Persuasive technologies-introduction. Communication of theACM , 42 (5), 26–29.

Fogg, B. J. (2009). A behavior model for persuasive design. In Proceedings of the 4thInternational Conference on Persuasive Technology (pp. 40:1–40:7). ACM.

Fortmann, J., Stratmann, T. C., Boll, S., Poppinga, B., & Heuten, W. (2013).Make me move at work! an ambient light display to increase physical activity. InProceedings of pervasivehealth 2013 (pp. 274–277). IEEE.

Frazer, J. G. (1898). Pausanias’s description of greece (Vol. 5). Macmillan andCompany, limited.

Froehlich, J. (2009). Promoting energy efficient behaviors in the home throughfeedback: The role of human-computer interaction. In Proceedings of HCIC WinterWorkshop 2009 (Vol. 9, pp. 1–11).

Gardner, G. T., & Stern, P. C. (1996). Environmental problems and human behavior.Allyn & Bacon.

Gauthier, S. (2011). What are people’s responses to thermal discomfort? sensingclothing and activity levels using sensecam. In Proceedings of the Conference onBehavior, Energy and Climate Change.

Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional proc-esses in evaluation: an integrative review of implicit and explicit attitude change.Psychological Bulletin, 132 (5), 692.

Gerard, R. M. (1958). Differential effects of colored lights on psychophysiologicalfunctions (Doctoral dissertation). University of California.

Giusa, F. F. L., & Perney, L. R. (1974). Further studies on the effects of brightnessvariations on attention span in a learning environment. Journal of the IlluminatingEngineering Society, 3 (3), 249–252.

Gustafsson, A., & Gyllenswärd, M. (2005). The power-aware cord: energy awarenessthrough ambient information display. In CHI’05 extended abstracts on Humanfactors in computing systems (pp. 1423–1426). ACM.

Gyllensward, M., Gustafsson, A., & Bang, M. (2006). Visualizing energy consumptionof radiators. In International conference on persuasive technology (pp. 167–170).

118

✐✐

✐✐

✐✐

✐✐

References

Ham, J., & Midden, C. (2010). A persuasive robotic agent to save energy: theinfluence of social feedback, feedback valence and task similarity on energy con-servation behavior. In S. Ge, H. Li, J. Cabibihan, & Y. Tan (Eds.), InternationalConference on Social Robotics 2010 (pp. 335–344). Springer Berlin Heidelberg.

Ham, J., & Midden, C. (2014). A Persuasive Robot to Stimulate Energy Conservation:The Influence of Positive and Negative Social Feedback and Task Similarity onEnergy-Consumption Behavior. International Journal of Social Robotics, 6 (2),163–171.

Ham, J., Midden, C., & Beute, F. (2009). Can ambient persuasive technology per-suade unconsciously?: using subliminal feedback to influence energy consumptionratings of household appliances. In Proceedings of the 4th International Conferenceon Persuasive Technology. ACM.

Ham, J., & Spahn, A. (2015). Shall i show you some other shirts too? the psychologyand ethics of persuasive robots. In R. Trappl (Ed.), A Construction Manual forRobots’ Ethical Systems (pp. 63–81). Springer International Publishing.

Heijs, W., & Stringer, P. (1988). Research on residential thermal comfort: Some con-tributions from environmental psychology. Journal of Environmental Psychology,8 , 235–247.

Helson, H., Blake, R. R., Mouton, J. S., & Olmstead, J. A. (1956). Attitudesas adjustments to stimulus, background, and residual factors. The Journal ofAbnormal and Social Psychology, 52 (3), 314-322.

Higashihara, Y., Tsutsumi, Y., Kozaki, T., Kitamura, S., Ueda, S., Horinouchi, K.,. . . Yasukouchi, A. (2002). Effects of color temperature of lighting sources on bodytemperature in the living room and bedroom at night. Journal of PhysiologicalAnthropology and Applied Human Science, 21 (6), 317–318.

Ho, H.-N., Iwai, D., Yoshikawa, Y., Watanabe, J., & Nishida, S. (2014). Combiningcolor and temperature: A blue object is more likely to be judged as warm than ared object. Scientific reports, 4 (5527), 1–5.

Holland, R. W., Hendriks, M., & Aarts, H. (2005). Smells like clean spirit non-conscious effects of scent on cognition and behavior. Psychological Science, 16 (9),689–693.

119

✐✐

✐✐

✐✐

✐✐

References

Holstius, D., Kembel, J., Hurst, A., Wan, P.-H., & Forlizzi, J. (2004). Infotropism:living and robotic plants as interactive displays. In Proceedings of the 5th conferenceon designing interactive systems: processes, practices, methods, and techniques (pp.215–221).

Hovland, C. I. (1957). The order of presentation in persuasion. New Haven, CT:Yale University Press.

Hovland, C. I., & Janis, I. L. (Eds.). (1959). Personality and persuasibility. NewHaven, CT: Yale University Press.

Hovland, C. I., & Rosenberg, M. J. (1960). Attitude organization and change. NewHaven, CT: Yale University Press.

Howell, W. C., & Kennedy, P. A. (1979). Field validation of the fanger thermal com-fort model. Human Factors: The Journal of the Human Factors and ErgonomicsSociety, 21 (2), 229–239.

Howell, W. C., & Stramler, C. S. (1981). Contribution of psychological variablesto the prediction of thermal comfort judgments in real world settings. ASHRAETrans.;(United States), 87 (5), 609–621.

Hughes, P. C., & Neer, R. M. (1981). Lighting for the elderly: A psychobiologicalapproach to lighting. Human Factors: The Journal of the Human Factors andErgonomics Society, 23 (1), 65–85.

Ihde, D. (1990). Technology and the lifeworld: From garden to earth. Bloomingtion,IN: Indiana University Press.

IJsselsteijn, W., de Kort, Y., Midden, C., Eggen, B., & van den Hoven, E. (2006).Persuasive technology for human well-being: setting the scene. In Proceedingsof International Conference on Persuasive Technology (pp. 1–5). Springer BerlinHeidelberg.

Iseki, T., & Yasukouchi, A. (2000). Effect of color temperature of fluorescent lamps onrectal temperature during and after heat exposure. In Proceedings of 44th Meetingon Physiol Anthropol (pp. 72–73).

Ishii, H., & Ullmer, B. (1997). Tangible bits: towards seamless interfaces betweenpeople, bits and atoms. In Proceedings of the ACM SIGCHI conference on Humanfactors in computing systems (pp. 234–241). ACM.

120

✐✐

✐✐

✐✐

✐✐

References

Ishikawa, Y. (1993). Energy-saving effect of heating/cooling system by illuminantcolor. Journal of the Illuminating Engineering Institute of Japan, 77 (11), 690–692.

Jacobs, K. W., & Hustmyer, F. E. (1974). Effects of four psychological primarycolors on GSR, heart rate and respiration rate. Perceptual and motor skills, 38 (3),763–766.

Jacobs, K. W., & Suess, J. F. (1975). Effects of four psychological primary colors onanxiety state. Perceptual and motor skills, 41 (1), 207–210.

Jahn, M., Jentsch, M., Prause, C. R., Pramudianto, F., Al-Akkad, A., & Reiners,R. (2010). The energy aware smart home. In Proceedings of 5th InternationalConference on Future Information Technology (pp. 1–8). IEEE.

Jensen, O. M. (2003). Visualisation turns down energy demand. In Proceedings of the2003 summer study of the european council for an energy efficient economy (pp.451–454).

Jentsch, M., Jahn, M., Pramudianto, F., Simon, J., & Al-Akkad, A. (2011). AnEnergy-Saving Support System for Office Environments. In A. Salah & B. Lepri(Eds.), Human Behavior Understanding 2011. LNCS (Vol. 7065, pp. 83–92).Springer Berlin Heidelberg.

Johnson, M. K., & Hirst, W. (1993). MEM: Memory subsystems as processes. InA. Collins, S. Gathercole, M. Conway, & P. Morris (Eds.), Theories of memory(Vol. 1, pp. 241–286). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Kaiser, P. K. (1984). Physiological response to color: A critical review. ColorResearch & Application, 9 (1), 29–36.

Kappel, K., & Grechenig, T. (2009). Show-me: water consumption at a glance topromote water conservation in the shower. In Proceedings of the 4th InternationalConference on Persuasive Technology (pp. 26:1–26:6). ACM.

Kaya, N., & Epps, H. (2004). Relationship between color and emotion: A study ofcollege students. College Student Journal, 38 (3), 396–405.

Khaled, R., Fischer, R., Noble, J., & Biddle, R. (2008). A qualitative study ofculture and persuasion in a smoking cessation game. In Proceedings of InternationalConference on Persuasive Technology (pp. 224–236). Springer Berlin Heidelberg.

121

✐✐

✐✐

✐✐

✐✐

References

Kim, T., Hong, H., & Magerko, B. (2010). Design requirements for ambient displaythat supports sustainable lifestyle. In Proceedings of the 8th ACM Conference onDesigning Interactive Systems (pp. 103–112). ACM.

King, W. C., Dent, M. M., & Miles, E. W. (1991). The persuasive effect of graphicsin computer-mediated communication. Computers in Human Behavior , 7 (4), 269–279.

Küller, R., Mikellides, B., & Janssens, J. (2009). Color, arousal, and performance – Acomparison of three experiments. Color Research & Application, 34 (2), 141–152.

Kuznetsov, S., & Paulos, E. (2010). Upstream: motivating water conservation withlow-cost water flow sensing and persuasive displays. In Proceedings of the SIGCHIConference on Human Factors in Computing Systems (pp. 1851–1860). ACM.

Kwallek, N., Lewis, C., Lin-Hsiao, J., & Woodson, H. (1996). Effects of ninemonochromatic office interior colors on clerical tasks and worker mood. ColorResearch and Application, 21 (6), 448–458.

Kwallek, N., Lewis, C. M., & Robbins, A. S. (1988). Effects of office interior coloron workers’ mood and productivity. Perceptual and Motor Skills, 66 (1), 123–128.

Lapides, P., Sharlin, E., & Greenberg, S. (2009). HomeWindow: an augmented realitydomestic monitor. In Proceedings of the 4th ACM/IEEE International Conferenceon Human Robot Interaction (pp. 323–324). ACM.

Lasswell, H., Casey, R., & Smith, B. (1935). Propaganda and promotional activities:An annotated bibliography. Minneapolis: University of Minnesota Press.

Latour, B. (1994). On Technical Mediation – Philosophy, Sociology, Genealogy.Common Knowledge, 3 (2), 29–64.

Lazarsfeld, P. F., Bernard, B., & Hazel, G. (1948). The people’s choice: How thevoter makes up his mind in a presidential campaign (2nd ed.). New York: ColumbiaUniversity Press.

Lockton, D., Harrison, D., & Stanton, N. (2008). Making the user more efficient:Design for sustainable behaviour. International Journal of Sustainable Engineering,1 (1), 3–8.

Maan, S., Merkus, B., Ham, J., & Midden, C. (2011). Making it not too obvious:the effect of ambient light feedback on space heating energy consumption. EnergyEfficiency, 4 (2), 175–183.

122

✐✐

✐✐

✐✐

✐✐

References

Maas, M., & Van den Bos, K. (2009). An affective-experiential perspective on reac-tions to fair and unfair events: Individual differences in affect intensity moderatedby experiential mindsets. Journal of Experimental Social Psychology, 45 (4), 667–675.

Magielse, R., & Ross, P. R. (2011). A design approach to socially adaptive lighting en-vironments. In Proceedings of the 9th ACM SIGCHI Italian Chapter InternationalConference on Computer-Human Interaction: Facing Complexity (pp. 171–176).ACM.

Main, A., Vlachonikolis, I., & Dowson, A. (2000). The wavelength of light causingphotophobia in migraine and tension-type headache between attacks. Headache:The Journal of Head and Face Pain, 40 (3), 194–199.

Marsh, R. L., & Hicks, J. L. (1998). Event-based prospective memory and execu-tive control of working memory. Journal of Experimental Psychology: Learning,Memory and Cognition, 24 , 336–349.

Martinez, M., & Geltz, C. (2005). Utilizing a pre-attentive technology for modifyingcustomer energy usage. In Proceedings of the european council for an energy-efficient economy (Vol. 30).

Matsukawa, I. (2004). The effects of information on residential demand for electricity.The Energy Journal, 1–17.

McCalley, L. T., & Midden, C. J. H. (2002). Energy conservation through product-integrated feedback: The roles of goal-setting and social orientation. Journal ofEconomic Psychology, 23 (5), 589–603.

McClaughan, C., Aspinall, P., & Webb, R. (1996). The effects of lighting upon moodand decision making. In Proceedings of CIBSE National Lighting Conference, Uni-versity of Bath (pp. 237–245). London: Chartered Institution of Building ServicesEngineers.

McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Why there arecomplementary learning systems in the hippocampus and neocortex: insights fromthe successes and failures of connectionist models of learning and memory. Psy-chological review, 102 (3), 419–457.

Mehta, R., & Zhu, R. (2008). Blue or Red? Exploring the Effect of Color on CognitiveTask Performances. Science, 323 (5918), 1226–1229.

123

✐✐

✐✐

✐✐

✐✐

References

Merkus, B. (2012). A shower meter to save water and energy: The influence oftwo feedback sources, goals, and the role of comfort on conservation behavior in alab and field experiment (Unpublished master’s thesis). Eindhoven University ofTechnology.

Midden, C., & Ham, J. (2012). Persuasive technology to promote pro-environmentalbehavior. In L. Steg, A. E. van den Berg, & J. I. De Groot (Eds.), EnvironmentalPsychology: An Introduction (pp. 243–225). New York: Wiley.

Midden, C., Kaiser, F. G., & Mccalley, L. T. (2007). Technology’s four roles inunderstanding individuals’ conservation of natural resources. Journal of SocialIssues, 63 (1), 155–174.

Midden, C., Kimura, H., Ham, J., Nakajima, T., & Kleppe, M. (2011). Persuasivepower in groups: The influence of group feedback and individual comparison feed-back on energy consumption behavior. In Proceedings of International Conferenceon Persuasive Technology (pp. 1:1–1:8). ACM.

Midden, C., McCalley, T., Ham, J., & Zaalberg, R. (2008). Using persuasive tech-nology to encourage sustainable behavior. Paper presented at the Pervasive Per-suasive Technology and Environmental Sustainability Workshop, 6th InternationalConference on Pervasive Computing, Sydney.

Millar, M. G., & Millar, K. U. (2000). Promoting safe driving behaviors: The influenceof message framing and issue involvement. Journal of Applied Social Psychology,30 (4), 853–866.

Miller, T. M., Rich, P., & Davis, J. (2009). ADAPT : Audience Design of AmbientPersuasive Technology. In Proceedings of CHI’09 Extended Abstracts on HumanFactors in Computing Systems (pp. 4165–4170). ACM.

Mogensen, M. F., & English, H. B. (1926). The apparent warmth of colors. TheAmerican Journal of Psychology.

Moyer-Gusé, E. (2008). Toward a theory of entertainment persuasion: Explaining thepersuasive effects of entertainment-education messages. Communication Theory,18 (3), 407–425.

Murphy, K., & Shleifer, A. (2004). Persuasion in politics. American EconomicReview, 94 (2), 435–439.

124

✐✐

✐✐

✐✐

✐✐

References

Nakamura, H., & Oki, M. (2000). Influence of air temperature on preference for colortemperature of general lighting in the room. Journal of the Human-EnvironmentalSystem, 4 (1), 41–47.

Nikefuel @Online. (2016). Retrieved from http://secure-nikeplus.nike.com/plus/what-is-fuel/

Occhialini, V., Van Essen, H., & Eggen, B. (2011). Design and evaluation of anambient display to support time management during meetings. In IFIP Conferenceon Human-Computer Interaction (pp. 263–280). Springer Berlin Heidelberg.

O’Connor, Z. (2011). Color psychology and color therapy: Caveat emptor. ColorResearch & Application, 36 (3), 229–234.

Ohnaka, T., Tochihara, Y., & Watanabe, Y. (1994). The effects of variation in bodytemperature on the preferred water temperature and flow rate during showering.Ergonomics, 37 (3), 541–546.

Oinas-Kukkonen, H., & Harjumaa, M. (2008). A systematic framework for designingand evaluating persuasive systems. In Proceedings of International Conference onPersuasive Technology (pp. 164–176). Springer Berlin Heidelberg.

Oinas-Kukkonen, H., & Harjumaa, M. (2009). Persuasive systems design: Keyissues, process model, and system features. Communications of the Association forInformation Systems, 24 (1), 486–500.

Olander, F., & ThØgersen, J. (1995). Understanding of consumer behavior as aprerequisite for environmental protection. Journal of Consumer Policy, 18 (4),345–385.

Otsuka, S., & Kawaguchi, J. (2009). Direct versus indirect processing changes theinfluence of color in natural scene categorization. Attention, perception & psy-chophysics, 71 (3), 1588–1597.

Parmar, V., & Keyson, D. (2008). Persuasive technology for shaping social beliefs ofrural women in India: An approach based on the theory of planned behaviour. InProceedings of International Conference on Persuasive Technology (pp. 104–115).Springer Berlin Heidelberg.

Perloff, R. M. (2010). The dynamics of persuasion: Communication and attitudes inthe 21st century. Routledge.

125

✐✐

✐✐

✐✐

✐✐

References

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of per-suasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (pp.123–205). New York: Academic.

Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement asa determinant of argument-based persuasion. Journal of Personality and SocialPsychology, 41 (5), 847–855.

Petty, R. E., Cacioppo, J. T., & Kasmer, J. A. (2015). The role of affect in theelaboration likelihood model of persuasion. In L. Donohew, H. Sypher, & E. Higgins(Eds.), Communication, Social Cognition, and Affect (pp. 117–147). PsychologyPress.

Piano stairs @Online. (2016). Retrieved from http://www.thepianostairs.com

Pousman, Z., & Stasko, J. (2006). A taxonomy of ambient information systems: Fourpatterns of design. In Proceedings of the Working Conference on Advanced VisualInterfaces, AVI’06 (pp. 67–74). ACM.

Ratcliff, R. (1993). Methods for dealing with reaction time outliers. PsychologicalBulletin, 114 (3), 510-532.

Redström, J. (2006). Persuasive design: Fringes and foundations. In Internationalconference on persuasive technology (pp. 112–122).

Reeder, G. D., & Pryor, J. B. (2008). Dual psychological processes underlying publicstigma and the implications for reducing stigma. Mens Sana Monographs, 6 (1),175–186.

Ritsema, B. S., Midden, C. J. H., & Heijden, P. G. M. (1982). Energiebesparing ingezinshuishoudingen: attitudes, normen en gedragingen, een landelijk onderzoek.Energieonderzoek Centrum Nederland.

Roberson, D., Davies, I., & Davidoff, J. (2000). Color categories are not universal:Replications and new evidence from a stone-age culture. Journal of ExperimentalPsychology: General, 129 (3), 369–398.

Roberts, S., & Baker, W. (2003). Towards effective energy information: Improvingconsumer feedback on energy consumption. A report to Ofgem Centre for Sustai-nable Energy, Bristol, UK.

126

✐✐

✐✐

✐✐

✐✐

References

Rogers, Y., Hazlewood, W. R., Marshall, P., Dalton, N., & Hertrich, S. (2010).Ambient influence: Can twinkly lights lure and abstract representations triggerbehavioral change? In Proceedings of the 12th ACM International Conference onUbiquitous Computing (pp. 261–270). ACM.

Ruijten, P. A., Midden, C. J., & Ham, J. (2015). Lonely and susceptible: Theinfluence of social exclusion and gender on persuasion by an artificial agent. Inter-national Journal of Human-Computer Interaction, 31 (11), 832–842.

Sato, M., & Yasukouchi, A. (1999). Non-visual effects of room illumination. InM. Sato, H. Tokura, & S. Watanuki (Eds.), Recent Advances in Physiological An-thropology (pp. 3–10). Fukuoka: Kyushu University Press.

Sauer, J., Schmeink, C., & Wastell, D. (2007). Feedback quality and environmentallyfriendly use of domestic central heating systems. Ergonomics, 50 (6), 795–813.

Schmidt, A. (2005). Interactive Context-Aware Systems Interacting with AmbientIntelligence. In G. Riva, F. Vatalaro, F. Davide, & M. Alcaniz (Eds.), AmbientIntelligence (pp. 159–178). Amsterdam: IOS Press.

Schneider, A., & Sidney, M. (2009). What is next for policy design and socialconstruction theory? 1. Policy Studies Journal, 37 (1), 103–119.

Sherif, M. (1948). An outline of psychology. New York: Harper.

Shteynberg, G., & Galinsky, A. D. (2011). Implicit coordination: Sharing goals withsimilar others intensifies goal pursuit. Journal of Experimental Social Psychology,47 (6), 1291–1294.

Simons, H. W. (1976). Persuasion: Understanding, practice, and analysis. Reading,MA: Addison Wesley.

Smids, J. (2012). The voluntariness of persuasive technology. In Proceedings ofInternational Conference on Persuasive Technology (pp. 123–132). Springer BerlinHeidelberg.

Smith, E. R. (1998). Mental representation and memory. In D. Gilbert, S. Fiske,& G. Lindzey (Eds.), Handbook of Social Psychology (Vol. 1, pp. 391–445). NewYork: Oxford University Press.

Smith, E. R., & DeCoster, J. (2000). Dual-process models in social and cognitivepsychology: Conceptual integration and links to underlying memory systems. Per-sonality and Social Psychology Review, 4 (2), 108–131.

127

✐✐

✐✐

✐✐

✐✐

References

Smyth, M. M. (1994). Cognition in action (second edition). Psychology Press.

Soldat, A. S., Sinclair, R. C., & Mark, M. M. (1997). Color as an environmental pro-cessing cue: External affective cues can directly affect processing strategy withoutaffecting mood. Social Cognition, 15 (1), 55–71.

Sörensen, S., & Brunnström, G. (1995). Quality of light and quality of life: Anintervention study among older people. Lighting Research and Technology, 27 (2),113–118.

Spahn, A. (2012). And lead us (not) into persuasion…? persuasive technology andthe ethics of communication. Science and engineering ethics, 18 (4), 633–650.

Standard, A. (2013). Standard 55-2013 – thermal environmental conditions for hu-man occupancy. American Society of Heating, Refrigerating and Air-ConditioningEngineering, Atlanta, GA.

Steg, L., & Vlek, C. (2009). Encouraging pro-environmental behaviour: An inte-grative review and research agenda. Journal of environmental psychology, 29 (3),309–317.

Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of socialbehavior. Personality and Social Psychology Review, 8 (3), 220–247.

Stramler, C. S., Kleiss, J. A., & Howell, W. C. (1983). Thermal sensation shiftsinduced by physical and psychological means. Journal of Applied Psychology, 68 (1),187–193.

Streppel, L. (2015). Evoking Behavioral Change with Ambient Devices. In E. vonZezschwitz et al. (Eds.), Human Computer Interaction in the Internet of ThingsEra (pp. 85–91). CiteSeer.

Svane, Ö. (2007). Helping, Informing or Coaxing the Consumer? Exploring Persua-sive Technology as applied to Household’s Energy Use. Policy, 1 (11), 1-11.

Taufik, D., Bolderdijk, J. W., & Steg, L. (2015). Acting green elicits a literal warmglow. Nature Climate Change, 5 (1), 37–40.

Tottori, D., Kozaki, T., Yasukouchi, A., Iwahashi, M., & Ikeda, K. (2000). Theeffect of bathroom lighting on body temperature change. Journal of PhysiologicalAnthropology and Applied Human Science, 19 (2), 116-117.

128

✐✐

✐✐

✐✐

✐✐

References

Valdez, P., & Mehrabian, A. (1994). Effects of color on emotions. Journal of Expe-rimental Psychology: General, 123 (4), 394–409.

Van den Bos, K., Ham, J., Lind, E. A., Simonis, M., Van Essen, W. J., & Rijpkema,M. (2008). Justice and the human alarm system: The impact of exclamation pointsand flashing lights on the justice judgment process. Journal of Experimental SocialPsychology, 44 (2), 201–219.

Van Raaij, W. F., & Verhallen, T. M. (1983). A behavioral model of residentialenergy use. Journal of Economic Psychology, 3 (1), 39–63.

Verbeek, P. (2010). What things do: Philosophical eflections on technology, agency,and design. Pennsylvania State University Press.

Verbeek, P.-P. (2006). Persuasive technology and moral responsibility toward anethical framework for persuasive technologies. Persuasive, 6 , 1–15.

Verhallen, T. M., & Van Raaij, W. F. (1981). Household behavior and the use ofnatural gas for home heating. Journal of Consumer Research, 8 (3), 253–257.

Vogels, I. (2008). Atmosphere metrics. In J. Westerink, M. Ouwerkerk, T. Overbeek,W. Pasveer, & B. de Ruyter (Eds.), Probing Experience: From Assessment ofUser Emotions and Behavior to Development of Products (pp. 25–41). SpringerNetherlands.

Vos, J. J. (1984). Disability glare–a state of the art report. CIE Journal, 3 (2),39–53.

Warm, J. S., Dember, W. N., & Parasuraman, R. (1991). Effects of olfactory stimula-tion on performance and stress. Journal of the Society Cosmetic Chemists, 42 (3),199–210.

Wattcher @Online. (2016). Retrieved from http://www.wattcher.nl/en/

Weaver, R. K. (2009). Target compliance: The final frontier of policy implementation.Governance Studies at Brookings.

Weiser, M. (1993). Some computer science issues in ubiquitous computing. Commu-nications of the ACM , 36 (7), 75–84.

Whiteley, J. A., Bailey, B. W., & McInnis, K. J. (2008). Using the internet to pro-mote physical activity and healthy eating in youth. American Journal of LifestyleMedicine, 2 (2), 115–127.

129

✐✐

✐✐

✐✐

✐✐

References

Wilson, G. D. (1966). Arousal Properties of Red Versus Green. Perceptual and MotorSkills, 23 , 947–949.

Wilson, G. T., Lilley, D., & Bhamra, T. (2013). Design Feedback Interventions ForHousehold Energy Consumption Reduction. In Proceedings of 16th Conference ofthe European Roundtable on Sustainable Consumption and Production (ERSCP)& 7th Conference of the Environmental Management for Sustainable Universities(EMSU) (ERSCP-EMSU 2013) (Vol. 6, pp. 4–7). Istanbul, Turkey: SustainableDevelopment and Cleaner Production Center, Bogaziçi University.

Winett, R. A., Kagel, J. H., Battalio, R. C., & Winkler, R. C. (1978). Effects ofmonetary rebates, feedback, and information on residential electricity conservation.Journal of Applied Psychology, 63 (1), 73–80.

Winzen, J., Albers, F., & Marggraf-Micheel, C. (2013). The influence of colouredlight in the aircraft cabin on passenger thermal comfort. Lighting Research andTechnology, 0 , 1–11.

Wisneski, C., Ishii, H., Dahley, A., Gorbet, M., Brave, S., Ullmer, B., & Yarin, P.(1998). Ambient Displays: Turning Architectural Space into an Interface betweenPeople and Digital Information. In N. Streitz, S. Konomi, & H. Burkhardt (Eds.),Cooperative buildings: Integrating information, organization, and architecture (Vol.1370, pp. 22–32). Springer Berlin Heidelberg.

Yasukouchi, A., & Ishibashi, K. (2005). Non-visual effects of the color temperatureof fluorescent lamps on physiological aspects in humans. Journal of PhysiologicalAnthropology and Applied Human Science, 24 (1), 41–43.

Yoto, A., Katsuura, T., Iwanaga, K., & Shimomura, Y. (2007). Effects of objectcolor stimuli on human brain activities in perception and attention referred to eegalpha band response. Journal of Physiological Anthropology, 26 (3), 373–379.

Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. AmericanPsychologist, 35 (2), 151.

Zanker, M., & Jessenitschnig, M. (2009). Case-studies on exploiting explicit cus-tomer requirements in recommender systems. User Modeling and User-AdaptedInteraction, 19 (1-2), 133–166.

Zhong, C.-B., & Leonardelli, G. J. (2008). Cold and lonely does social exclusionliterally feel cold? Psychological Science, 19 (9), 838–842.

130

✐✐

✐✐

✐✐

✐✐

Summary

The work in this dissertation focused on understanding the psycholo-gical mechanisms underlying the effectiveness of ambient lighting

persuasive technology in the energy conservation domain. Earlier re-search has shown that ambient lighting can be successfully employedas a medium to promote energy conservation behavior (e.g., by pro-viding immediate energy feedback, such as red light = high energy con-sumption), and such ambient persuasive technology had stronger persua-sive effects on energy conservation behavior, compared to (widely used)factual-evaluative feedback (e.g., directly indicating the amount of kWh).Feedback given through this kind of ambient persuasive technology canhelp users in those day-to-day situations, in which users lack cognitivecapacity or motivation to consciously process relatively complex typesof feedback (e.g., numerical feedback). However, the underlying mecha-nisms behind the effectiveness of ambient persuasive lighting still remainlargely unclear.

The main research question in this dissertation was: What are the un-derlying psychological mechanisms of the effectiveness of using light inambient persuasive technology on energy conservation behavior? We ar-gue that the color associations and color’s contexts of ambient lightingcould help users to easily process feedback messages. These processes

131

✐✐

✐✐

✐✐

✐✐

Summary

are identified in dual-process models of cognitive processing (e.g., thereflective-impulsive model) as impulsive processes. These associationsslowly develop when people are cognitively creating representations ofthe environment, and change only gradually through learning. Therefore,the first research goal in this dissertation was to understand the asso-ciative processes of ambient lighting persuasive technology. We studiedthe efficiency of these associative processes by investigating two charac-teristics of ambient lighting feedback: the role of color associations ofambient lighting feedback, and the role of color context. Processes inthe impulsive cognitive system, to some degree, may also consist of ex-periential processes. Experiential processes are also associative in naturebut involve less cognition.

Therefore, the second research goal in this dissertation was to understandthe experiential processes of ambient lighting persuasive technology. Weproposed a new perspective on exploring the influence of ambient light-ing on people’s thermal experiences. More specifically, in Chapter 2, weargued that colors of ambient lighting feedback could carry meaningsthat have strong and consistent associations with energy consumption,which could help users to easily process feedback messages, and therebyenhance the effectiveness of ambient persuasive technology. An onlinesurvey was conducted to investigate the strength of different color associ-ations (of lighting feedback) with energy consumption. Two experimen-tal studies were conducted to investigate the effects of color associationstrength (strongly-associated vs. weakly-associated) and color associa-tion consistency (consistent vs. inconsistent) on the ease of processingof feedback messages, and furthermore on users’ energy consumptionbehavior. In line with our expectations, results showed that lightingfeedback that used colors with strong (Study 2.1) and consistent (Study2.2) energy associations could help users to easily process the feedbackmessages, and lead to more energy conservation behavior.

In Chapter 3, we argued that for understanding the role of color as-

132

✐✐

✐✐

✐✐

✐✐

Summary

sociations, the context in which these colors are presented is crucial.That is, the information that color-based ambient persuasive techno-logy carries is highly context-dependent. More specifically, a consistentcontext in which color-based ambient persuasive technology is presentedcould shape the associations of colors, which makes the feedback mes-sages directly understandable. Moreover, this context can increase thepersuasive power of ambient persuasive technology to promote energyconservation behavior. Two studies were conducted to investigate ef-fects of context (in which color-based ambient lighting was presented) onuser’s interpretation of the information that lighting carries, and moreimportantly, on the effectiveness of this ambient persuasive lighting (forinfluencing energy conservation behavior). Confirming our expectations,results showed that participants perceived the color red in an energy-related context to be warmer and as related to a higher energy con-sumption level, compared to the color red in an energy-unrelated con-text. Also, participants who received ambient lighting feedback in anenergy-related context consumed the lowest amount of energy (Study3.1 & 3.2).

In Chapter 4 we extended our understanding of the underlying mecha-nisms of ambient lighting persuasive technology by exploring the experi-ential processes, instead of associative processes. We argued that thermalexperiences are a main factor when making decisions about energy con-servation. Therefore, we proposed that ambient lighting might be usedto influence user’s thermal experiences (e.g., by emitting a ‘warm’ or‘cold’ light) and thereby influence user’s room temperature perceptionand their willingness to change room temperature. We conducted twostudies to investigate the effects of brightness level of room lighting andthe effects of color temperature of room lighting on participants’ percep-tions of room temperature, and their willingness to change room tem-perature settings. Supporting our expectations, results indicated thatcolor temperature (Study 4.1 & 4.2) can influence a user’s perception of

133

✐✐

✐✐

✐✐

✐✐

Summary

the temperature in the room, and influence user’s willingness to changeroom temperature. However, our second study in this chapter providedno evidence that these effects of room lighting color temperature alsoextended to actual energy conservation behavior.

Finally, in the General Discussion, we discuss the main findings andcontributions of the studies presented in this dissertation. Also, we iden-tify general limitations and directions for future research. In addition,we briefly address the ethical consideration of using ambient lighting inpersuasive technology.

In sum, the findings from the work presented in this dissertation re-veal associative and experiential processes of effective ambient lightingpersuasive technology. By evaluating the effects of color associations ofambient lighting, the effects of context of which ambient lighting, andthe effects of color temperature on thermal experiences, we deepened ourinsights into the psychological processes underlying the effectiveness ofambient lighting persuasive technology.

134

✐✐

✐✐

✐✐

✐✐

List of Figures

1.1 Examples of human beings using light to communicate inhistory: (a) signal with a shield; and (b) heliograph. . . . 2

1.2 Factual feedback: (left) energy meter and (right) Wattcher. 71.3 Social feedback as provided by a robotic agent. . . . . . . 81.4 Ambient feedback: (left) Energy Orb and (right) Power

Aware Cord. . . . . . . . . . . . . . . . . . . . . . . . . . 91.5 Waterbot: (a) indicating warm water; while (b) indicating

cold water. . . . . . . . . . . . . . . . . . . . . . . . . . . 111.6 Ambient lighting feedback as used in the research by Ham

and Midden (2010). . . . . . . . . . . . . . . . . . . . . . 17

2.1 Programmable LED wall washers on the side walls indi-cating (a) high and (b) low energy consumption levels instrongly-associated lighting feedback condition; (c) highand (d) low energy consumption level in weakly-associatedlighting feedback condition. . . . . . . . . . . . . . . . . 33

2.2 The timeline of the experiment in Study 2.1. . . . . . . . 332.3 A simulated programmable heating system interface used

to gather data about the participant’s energy consumptionbehavior in Study 2.1. . . . . . . . . . . . . . . . . . . . 34

2.4 The link between energy consumption level and the colorof strongly-associated lighting feedback (color can contin-uously change according to the thermostat setting). . . . 35

135

✐✐

✐✐

✐✐

✐✐

List of Figures

2.5 Averaged task completion time as a result of color associa-tion strength (strongly-associated vs. weakly-associated)and cognitive load (low vs. high) in Study 2.1. Errorbars attached to each column in the figure indicate 95%confidence intervals. . . . . . . . . . . . . . . . . . . . . . 37

2.6 Averaged energy consumption score as a result of color as-sociation strength (strongly-associated vs. weakly-associated)in Study 2.1. Error bars attached to each column in thefigure indicate 95% confidence intervals. . . . . . . . . . 37

2.7 Averaged task completion time as a result of color associa-tion strength (strongly-associated vs. weakly-associated)and color association consistency (consistent vs. inconsi-stent) in Study 2.2. Error bars attached to each columnin the figure indicate 95% confidence intervals. . . . . . . 43

2.8 Averaged energy consumption score as a result of colorassociation consistency (consistent vs. inconsistent) inStudy 2.2. Error bars attached to each column in thefigure indicate 95% confidence intervals. . . . . . . . . . 44

2.9 The change of energy consumption scores per action asa result of color association consistency (consistent vs.inconsistent) in Study 2.2. Error bars attached to eachcolumn in the figure indicate 95% confidence intervals. . 45

3.1 Four between-subject experimental conditions of Study3.1: (a) the color red in an energy-related context, (b)the color red in an energy-unrelated context, (c) the colorblue in an energy-related context and (d) the color bluein an energy-related context. . . . . . . . . . . . . . . . . 60

136

✐✐

✐✐

✐✐

✐✐

List of Figures

3.2 Averaged participant’s interpretation of energy consump-tion level for all conditions in Study 3.1: the color red inan energy-related context (i.e., radiator) and in an energy-unrelated context (i.e., tulip); and the color blue in anenergy-related context (i.e., radiator) and in an energy-unrelated context (i.e., tulip). Error bars attached to eachcolumn in the figure indicate 95% confidence intervals. . 62

3.3 Averaged participant’s assessment of an object’s warmthfor all conditions in Study 3.1: the color red in an energy-related context (i.e., radiator) and in an energy-unrelatedcontext (i.e., tulip); and the color blue in an energy-relatedcontext (i.e., radiator) and in an energy-unrelated context(i.e., tulip). Error bars attached to each column in thefigure indicate 95% confidence intervals. . . . . . . . . . 62

3.4 The interactive ambient energy feedback system by pre-senting colors in (a) an energy-related context and in (b)an energy-unrelated context. . . . . . . . . . . . . . . . . 66

3.5 In the energy-related context, the interactive color-basedenergy feedback system showed colors varying betweenvery weak and intense red depending on people’s energyuse from (a) indicating very low energy consumption to(d) very high energy consumption. . . . . . . . . . . . . . 66

3.6 In the energy-unrelated context, the interactive color-basedenergy feedback system showed colors varying betweenvery weak and intense red depending on people’s energyuse from (a) indicating very low energy consumption to(d) very high energy consumption. . . . . . . . . . . . . . 67

137

✐✐

✐✐

✐✐

✐✐

List of Figures

3.7 Averaged participant’s assessment of an object’s warmthas a result of context (energy-related, i.e., radiator vs.energy-unrelated, i.e., tulip) in Study 3.2. Error bars at-tached to each column in the figure indicate 95% confi-dence intervals. . . . . . . . . . . . . . . . . . . . . . . . 67

3.8 Averaged participant’s interpretation of energy consump-tion level as a result of context (energy-related, i.e., radi-ator vs. energy-unrelated, i.e., tulip) in Study 3.2. Errorbars attached to each column in the figure indicate 95%confidence intervals. . . . . . . . . . . . . . . . . . . . . . 68

3.9 The change of energy consumption scores per action as aresult of context (energy-related, i.e., radiator vs. energy-unrelated, i.e., tulip) in Study 3.2. Error bars attached toeach column in the figure indicate 95% confidence intervals. 69

4.1 Averaged warmth perception of room lighting as a resultof color temperature (“warm” vs. “cold”) of room lightingin Study 4.1. Error bars attached to each column in thefigure indicate 95% confidence intervals. . . . . . . . . . 80

4.2 Averaged participant’s tendency to change the room tem-perature as a result of color temperature (“warm” vs.“cold”) of room lighting in Study 4.1. Error bars attachedto each column in the figure indicate 95% confidence in-tervals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

4.3 Averaged warmth perception of room temperature as aresult of color temperature (“warm” vs. “cold”) of roomlighting in Study 4.1. Error bars attached to each columnin the figure indicate 95% confidence intervals. . . . . . . 81

4.4 Ambient lighting environment used in Study 4.2: (a) “warm”room and (b) “cold” room. . . . . . . . . . . . . . . . . . 84

4.5 The timeline of the experiment in Study 4.2. . . . . . . . 85

138

✐✐

✐✐

✐✐

✐✐

List of Figures

4.6 A simulated programmable heating system interface usedto gather data about the participant’s energy consumptionbehavior in Study 4.2. . . . . . . . . . . . . . . . . . . . 86

4.7 Averaged warmth perception of the room lighting as aresult of color temperature (“warm” vs. “cold”) of roomlighting in Study 4.2. Error bars attached to each columnin the figure indicate 95% confidence intervals. . . . . . . 89

4.8 Averaged perceived comfort of room lighting as a result ofcolor temperature (“warm” vs. “cold”) of room lightingin Study 4.2. Error bars attached to each column in thefigure indicate 95% confidence intervals. . . . . . . . . . 89

4.9 Averaged warmth perception of room temperature as aresult of color temperature (“warm” vs. “cold”) of roomlighting in Study 4.2. Error bars attached to each columnin the figure indicate 95% confidence intervals. . . . . . . 90

4.10 Averaged perceived comfort of room temperature as a re-sult of color temperature (“warm” vs. “cold”) of roomlighting in Study 4.2. Error bars attached to each columnin the figure indicate 95% confidence intervals. . . . . . . 90

4.11 Averaged room temperature setting as a result of colortemperature (“warm” vs. “cold”) of room lighting in Study4.2. Error bars attached to each column in the figure in-dicate 95% confidence intervals. . . . . . . . . . . . . . . 91

139

✐✐

✐✐

✐✐

✐✐

List of Figures

140

✐✐

✐✐

✐✐

✐✐

Appendices

141

✐✐

✐✐

✐✐

✐✐

✐✐

✐✐

✐✐

✐✐

A | Thermostat Task

On the following sections, the set of trails as used in the thermostat taskthat used in Chapter 2 are presented in Dutch and the translated versionin English.

A.1 Study 2.1 & Study 2.2

Dutch (original) English (translated)

(1) Het is avond en je bent thuis. (1) It is evening and you are atHet is buiten 3 ◦C. Stel voor home. The outside tempera-de verschillende vertrekken de ture is 3 ◦C. Set the temper-centrale verwarming in. ature for the different rooms

in the house.

(2) Het is middag en je bent niet (2) It is afternoon and you arethuis. Het is buiten 20 ◦C. not at home. The outside tem-Stel voor de verschillende ver- perature is 20 ◦C. Set the tem-trekken de centrale verwarm- perature for the different roomsing in. in the house.

143

✐✐

✐✐

✐✐

✐✐

Appendices. Thermostat Task

(3) Het is ochtend en je bent thuis. (3) It is morning and you are atHet is buiten −5 ◦C. Stel voor home. The outside temper-de verschillende vertrekken de ature is −5 ◦C. Set the tem-centrale verwarming in. perature for the different rooms

in the house.

(4) Het is ochtend en je bent thuis. (4) It is morning and you are atHet is buiten 15 ◦C. Stel voor home. The outside tempera-de verschillende vertrekken de ture is 15 ◦C. Set the temper-centrale verwarming in. ature for the different rooms

in the house.

(5) Het is middag en je bent niet (5) It is afternoon and you arethuis. Het is buiten 8 ◦C. Stel not at home. The outside tem-voor de verschillende vertrek- perature is 8 ◦C. Set the tem-ken de centrale verwarming perature for the different roomsin. in the house.

(6) Het is zondagmiddag en je (6) It is a Sunday afternoon andbent thuis. Het is buiten 19 ◦C. you are at home. The outsideStel voor de verschillende ver- temperature is 19 ◦C. Set thetrekken de centrale verwarm- temperature for the differenting in. rooms in the house.

(7) Het is middag en je bent niet (7) It is afternoon and you arethuis. Het is buiten 18 ◦C. not at home. The outside tem-Stel voor de verschillende ver- perature is 18 ◦C. Set the tem-trekken de centrale verwarm- perature for the different roomsing in. in the house.

144

✐✐

✐✐

✐✐

✐✐



in the house.


in the house.

(10) Het is avond en je geeft vanavond (10) It is evening and tonight youthuis een feestje. Het is buiten give a party at home. The16 ◦C. Stel voor de verschil- outside temperature is 16 ◦C.lende vertrekken de centrale Set the temperature for theverwarming in. different rooms in the house.

(11) Het is nacht en je ligt in bed. (11) It is night and you are inHet is buiten −1 ◦C. Stel voor bed. The outside tempera-de verschillende vertrekken de ture is −1 ◦C. Set the tem-centrale verwarming in. perature for the different rooms

in the house.

(12) Het is nacht en je ligt in bed. (12) It is night and you are inHet is buiten 14 ◦C. Stel voor bed. The outside tempera-de verschillende vertrekken de ture is 14 ◦C. Set the temper-centrale verwarming in. ature for the different rooms

in the house.

145

✐✐

✐✐

✐✐

✐✐


A.2 Study3.2

(1) U moet zometeen in deze ruimte (1) You will have to do exercisegaan fitnessen. in this room.

(2) U moet zometeen in deze ruimte (2) You will have to go to sleepgaan slapen. in this room.

(3) U (alleen) gaat hier het komende (3) You (alone) are going to readhalf uur een boek lezen. a book in the next half hour.

(4) U (alleen) gaat hier de komende (4) You (alone) are going to watchtwee uur een film kijken. a movie in the next 2 hours.

(5) U en een vriend gaan hier het (5) You and a friend are going tokomende half uur een tv-programma watch a TV show in the nextkijken. half hour.

(6) U en drie vrienden gaan hier (6) You and three friends are go-de komende 2 uur een bord- ing to play a board game inspel spelen. the next 2 hours.

(7) U en de experimentleider gaan (7) You and the experiment leaderheir het komende half uur een are going to have a couple ofkop thee drinken. tea in the next half hour.

(8) U en vijf andere studenten (8) You and five friends are goinggaan de komende 2 uur in deze to study in this room in theruimte studeren. next 2 hours.

146

✐✐

✐✐

✐✐

✐✐

B | Questionnaire

On the following sections, these questionnaires are presented in Dutchand the translated version in English.

B.1 Study 3.1 & Study 3.2

Dutch (original) English (translated)

(1) Hoe warm vindt u het object (1) How warm do you feel the ob-geprojecteerd op het scherm? ject projected on the screenIk vind het object... is? I feel the object is some-! heel koud thing...! koud ! very cold! enigzins koud ! cold! noch koud, noch warm ! slightly cold! enigzins warm ! neither cold nor warm! warm ! slightly warm! heel warm ! warm

! very warm

147

✐✐

✐✐

✐✐

✐✐

Appendices. Questionnaire

(2) Wat vind u van het object (2) How do you think of the ob-dat geprojecteerd is op het ject on the screen? I thinkscherm? Ik vind het object... the object is...! heel negatief ! very negative! negatief ! negative! enigzins negatief ! slightly negative! noch negatief, noch posi- ! neither negative nor po-

tief sitive! enigzins positief ! slightly positive! positief ! positive! heel positief ! very positive

(3) Wat denkt u dat de betekenis (3) How do you judge the mean-is van dit geprojecteerd ob- ing that this object conveys?ject? Ik denk dat dit object I think this object means...te maken heeft met... ! very low energy consump-! heel erg laag energiever- tion

bruik ! low energy consumption! laag energieverbruik ! slightly low energy con-! enigzins laag energiever- sumption

bruik ! neither low nor high energy! noch laag, noch hoog en- consumption

ergieverbruik ! slightly high energy con-! enigzins hoog energiever- sumption

bruik ! high energy consumption! hoog energieverbruik ! very high energy consump-! heel erg hoog energiever- tion

bruik

(4) In hoeverre denk je dat dit ge- (4) To what extend do you thinkprojecteerd object te maken this object is related to energy

148

✐✐

✐✐

✐✐

✐✐


heeft met het energieverbruik? consumption? I think this ob-Ik denk dat dit object... ject is...! heel erg zwak gerelateerd ! very weakly-related with

is aan het energieverbruik energy consumption! zwak gerelateerd is aan ! weakly-related with energy

het energieverbruik consumption! enigzins zwak gerelateerd ! slightly weakly-related with

is aan het energieverbruik energy consumption! noch zwak, noch sterk gere- ! neither weakly- nor strongly-

lateerd is aan het energie- related with energy con-verbruik sumption

! enigzins sterk gerelateerd ! slightly strongly-related withis aan het energieverbruik energy consumption

! sterk gerelateerd is aan ! strongly-related with energyhet energieverbruik consumption

! heel erg sterk gerelateerd ! very strongly-related withis aan het energieverbruik energy consumption

B.2 Study 4.1 & Study 4.2

(1) Wat vindt u van het licht (1) What do you think aboutin deze ruimte? the lighting in this room?! heel koud ! very cold! koud ! cold! enigzins koud ! slightly cold! noch koud, noch warm ! neither cold nor warm! enigzins warm ! slightly warm! warm ! warm! heel warm ! very warm

149

✐✐

✐✐

✐✐

✐✐


(2) In hoeverre is de temper- (2) Do you think the lightingatuur van het licht koud/ in this room is warm orwarm? cold?

...as above...

(3) Wat voor gevoel geeft het (3) What feeling does the light-licht u? ing in this room give you?

...as above...

(4) Hoe ziet het licht er vol- (4) What do you think thegens u uit? light in this room looks

like?...as above...

(5) Hoe comfortabel vindt u (5) How comfortable do youdeze kamer? think this room is?! heel oncomfortabel ! very uncomfortable! oncomfortabel ! uncomfortable! enige oncomfortabel ! slightly uncomfortable! noch oncomfortabel, noch ! neither uncomfortable nor

comfortabel comfortable! enige comfortabel ! slightly comfortable! comfortabel ! comfortable! heel comfortabel ! very comfortable

(6) Hoe comfortabel voelt u (6) How comfortable do youzich in deze ruimte, als u think the lighting in thisdenkt aan hoe deze ruimte room is?nu verlicht is? ...as above...

(7) Is de temperatuur van de (7) Do you think the temper-

150

✐✐

✐✐

✐✐

✐✐


huidige verlichting aange- ature of the room light-naam? ing is pleasant? (not cor-! heel onaangenaam rect translation?)! onaangenaam ! very unpleasant! enige onaangenaam ! unpleasant! noch onaangenaam, noch ! slightly unpleasant

aangenaam ! neither unpleasant nor pleas-! enige aangenaam ant! aangenaam ! slightly pleasant! heel aangenaam ! pleasant

! very pleasant

(8) Is de temperatuur van de (8) Do you think the temper-huidige verlichting aantrekke- ature of the room light-lijk? ing is attractive? (not cor-! heel onaantrekkelijk rect translation?)! onaantrekkelijk ! very disagreeable! enige onaangenaam ! disagreeable! noch onaantrekkelijk, noch ! slightly disagreeable

aantrekkelijk ! neither disagreeable nor! enige aantrekkelijk agreeable! aantrekkelijk ! slightly agreeable! heel aantrekkelijk ! agreeable

! very agreeable

(9) Hoe voelt voor u de al- (9) How does the tempera-gemene temperatuur van ture of this room feel?deze ruimte aan? ! very cold! heel koud ! cold! koud ! slightly cold! enigzins koud ! neither cold nor warm

151

✐✐

✐✐

✐✐

✐✐


! noch koud, noch warm ! slightly warm! enigzins warm ! warm! warm ! very warm! heel warm

(10) Voelt de lucht in deze (10) Does the air in this roomruimte warm of koud aan? feel warm or cold

...as above...

(11) Hoe comfortabel voelt (11) How comfortable do youu zich in deze ruimte, als experience the room tem-u denkt aan de temper- perature?atuur in deze ruimte? ! very uncomfortable! heel oncomfortabel ! uncomfortable! oncomfortabel ! slightly uncomfortable! enige oncomfortabel ! neither uncomfortable nor! noch oncomfortabel, noch comfortable

comfortabel ! slightly comfortable! enige comfortabel ! comfortable! comfortabel ! very comfortable! heel comfortabel

(12) Hoe comfortabel vindt (12) How comfortable do youu de huidige temperatuur think the room tempera-in deze ruimte? ture is?

...as above...

(13) Is de huidige algemene (13) Do you think the roomtemperatuur in deze ruimte temperature is pleasant?aangenaam? ! very unpleasant! heel onaangenaam ! unpleasant

152

✐✐

✐✐

✐✐

✐✐


! onaangenaam ! slightly unpleasant! enige onaangenaam ! neither unpleasant nor pleas-! noch onaangenaam, noch ant

aangenaam ! slightly pleasant! enige aangenaam ! pleasant! aangenaam ! very pleasant! heel aangenaam

(14) Is de huidige algemene (14) Is the room temperaturetemperatuur in deze ruimte agreeable?aantrekkelijk? ! very disagreeable! heel onaantrekkelijk ! disagreeable! onaantrekkelijk ! slightly disagreeable! enige onaangenaam ! neither disagreeable nor! noch onaantrekkelijk, noch agreeable

aantrekkelijk ! slightly agreeable! enige aantrekkelijk ! agreeable! aantrekkelijk ! very agreeable! heel aantrekkelijk

153

✐✐

✐✐

✐✐

✐✐


154

✐✐

✐✐

✐✐

✐✐

Publication List

Journal publicationsLu, S., Ham, J. & Midden, C. (2016). The influence of color association

strength and consistency on ease of processing of ambient lightingfeedback. Journal of Environmental Psychology (Vol. 47, pp. 204-212).

Lu, S., Ham, J. & Midden, C. (2016). The Effect of Brightness Levelof Room Lighting on Thermal Experiences and User Motivation toChange Room Temperature. Manuscript in preparation.

Conference publicationsLu, S., Ham, J., & Midden, C. (2016). Red Radiators Versus Red Tulips:

The Influence of Context on the Interpretation and Effectiveness ofColor-Based Ambient Persuasive Technology. In: MeschtscherjakovA., De Ruyter B., Fuchsberger V., Murer M., Tscheligi M. (eds) Per-suasive Technology. PERSUASIVE 2016, LNCS 9638. (pp. 303-314).Springer International Publishing.

Lu, S., Ham, J., Akker, M. & Midden, C. (2016). The persuasive poten-tial of lighting: Exploring user lighting setting preferences for a warm

155

✐✐

✐✐

✐✐

✐✐

Publication List

room atmosphere and energy consumption feedback. In Proceedingsof International Conference on Ambient Computing, Applications, Ser-vices and Technologies (pp. 1:1-1:2).

Lu, S., Ham, J., & Midden, C. (2015). Persuasive technology basedon bodily comfort experiences: The effect of color temperature ofroom lighting on user motivation to change room temperature. In:MacTavish T., Basapur S. (Eds) Persuasive Technology. PERSUA-SIVE2015, LNCS 9072 (pp. 83-94). Springer Cham. (Best PaperAward).

Lu, S., Ham, J. & Midden, C. (2014). Using ambient lighting in per-suasive communication: The role of pre-existing color associations.In: L.Gamberini, A. Spagnolli, & L. Chittaro (Eds.): PERSUASIVE2014, LNCS 8462 (PP. 167-178). Springer International PublishingSwitzerland 2014.

156

✐✐

✐✐

✐✐

✐✐

Acknowledgments

I still remember reading Chinese every morning in a crowed classroomwhen I was a boy, as if it was yesterday. In the meantime, life has

been treating me generously providing all kinds of possibilities over thepast 27 years. Now, at the end of my official student life, I realize howlucky I am to choose this wonderful journey and to be accompanied byall the people I met. With this book in my hand, I would like to expressmy sincere gratitude to all people that contributed in many differentways to the completion of this doctoral dissertation.The deepest gratitude of mine goes, first and foremost, to my promoter:prof. Cees Midden, and daily supervisor: dr. Jaap Ham. When I lookback throughout my PhD, one of the things that I really enjoyed andappreciated is our weekly conversations about work and life. Dear Cees,thank you for always ‘persuasively’ allowing me to explore the interest-ing research directions, and helping me to always keep a sharp eye ontheoretical framework from the beginning to the end of my PhD project.Whenever I felt confused or lost in research, I could always find the an-swers and solutions from you. Besides work, I very much honored theway that you respected different opinions. I have learned and benefitedfrom you a lot in both my work and personal life. Dear Jaap, thank youfor offering me a great start of my PhD life even before I came to theNetherlands. I feel so lucky to spend most of my research time together

157

✐✐

✐✐

✐✐

✐✐

Acknowledgements

with you in the last several years. You had the magic power to alwayskeep me motivated and inspired. From the theoretical models, the pro-gramming to the HUGE documents needed for my special PhD case, youtried to help me out in one way or another during different ph(f)as(c)esof my PhD. Without either of you, I cannot imagine the completion ofmy PhD project and this dissertation.

Besides my supervisors, I would like to extend my gratitude to the restof my thesis committee: prof. Berry J.H. Eggen, Prof. David V. Keyson,prof. Boris E.R. de Ruyter and prof. Manfred Tscheligi, for their insight-ful comments and encouragement, but also for the hard question whichincented me to widen my research from various perspectives.

During my PhD I also followed the main master program at Human-Technology Interaction, TU/e. Here, I would like to thank the followingteachers (who later became my colleagues) for allowing me to join thecourses and spending time on my assignments and projects: dr. Ray-mond Cuijpers, dr. Antal Haans, prof. Wijnand IJsselsteijn, prof. JamesJuola, prof. Armin Kohlrausch, prof. Yvonne de Kort, dr. Daniël Lakens,prof. Chris Snijders, dr. Martijn Willemsen and so on. The knowledgethat I have gained from your courses was very valuable for my own PhDresearch.

At Human-Technology Interaction group, I would like to thank all won-derful fellow PhD students and colleagues who made my time in theNetherlands as a PhD student unforgettable. In particular, to my for-mer officemates, Leon and Frank, you are the very first officemates whointroduced me to Dutch culture. Our talks about everything in office orduring lunch made me love this new place more and more. To Peter,you are always helpful in different occasions and thank you for the greattime in Italy and our weekly squash. To all the officemates at IPO 1.39,although we did not meet every day at office in the last two years, I verymuch enjoyed our talks about work and parties whenever we met eachother. To Ellen and Anita, you may not be able to imagine how impor-

158

✐✐

✐✐

✐✐

✐✐

Acknowledgements

tant you are to me, as a foreign student. Thank you for all the help, fromthe beginning when I picked up my laptop to the end when you helpedme setting up the defense date. Also, I very appreciated your creativitiesfor the HTI events. I would like to thank Martin, Aart and Jan-Roeloffor your lab and technical support. I would also like to thank Wout vanBommel for your lab support when I conducted my experiment in theDepartment of the Built Environment.

A Special thank goes to my two paranymphs on the day of my defense:Dai Yi & Sofia Fountoukidou. Dear Sofia: you are more than a colleagueto me, and you are like my sister that I have wished for. Your intelligenceand consideration can always provide me great supports and suggestionsin my most difficult and unexpected period of my PhD. Your humor andcreativity made our conference visit in Salzburg, and trips in Viennaor within the Netherlands so colorful and beautiful (as you are). Bestwishes to your own PhD and hope I can be your paranymph at yourPhD ceremony too. Dear Yi: friending with you for over 15 years is sucha luxury to me. From Chongqing (2001), Chengdu (2011) to Holland(2014), I feel so lucky to have you in the different stages of my life. Youare the one whom I can share everything with, about work and privatelife. You are the one who never complains about my hours-talks and whosupports my choices during the last years. You are also the one who hasthe boldness to face/take all the challenges in life, which makes you as arole model for me to look up to throughout my student life. Best wishesto your new career and hope I can be your paranymph at your weddingceremony.

Friendship, is a word that I always cherish, within or outside academicenvironment. To my best friends in Chongqing, our friendship (withsuch a big group) is the source that I can always rely on no matter whenand where I was. Especially, many thanks to Xuan Liu, we telephonedeach other almost everyday in the last several months, we just talkedtoo much;) !!! To all my friends in Holland, my PhD life without you

159

✐✐

✐✐

✐✐

✐✐

Acknowledgements

would be so gray. To whom we had lots of cozy dinners and drinks,to whom we went to gyms/sports together, to whom we had traveledtogether, and to whom we spend so many day and night events together.Of all these friends, special thank goes to the friends I met at Dynamo,you are awesome! Thank you all for helping me building my dancinglife in Holland (somehow I dreamed to be a dancer when I was little).Moreover, special thank goes to my best roommates in Eindhoven forthe last three years: dr. Xiulei Cao and dr. Jiquan Wang. Thank youfor all the interesting talks, great food, and also for taking care of mewhen I was sick, allowing my RenXin and making me literally feel athome. Best wishes for both of you in your future.

Although I cannot list all the names of people who I want to thank, Iwould like to keep your names in my heart. To some of whom I mentionedin this book, to some of whom I kept in touch with, to some of whom Ikept memories of you, and some of whom have changed (and are goingto change) the trajectory of my life.

Finally, to my Mom and Dad. I am so proud to be born in a “teacher-family” (both my parents and grandpa are teachers), who raised me witha love of science, and provided me with lots of freedom in all my pursuits.You are my best friends, who give me confidence/courage and who arealways willing to join me for crazy adventures. Without you, I wouldnot be here now, writing these words. Grandma, you taught me how tobe a kind person, and you succeeded. You will forever be missed.

Shengnan Lu

Eindhoven, February 2017

160

✐✐

✐✐

✐✐

✐✐

CurricuLum Vitae

Shengnan Lu was born on 10 March 1989, inChongqing, the People’s Republic of China. Hegrew up and completed the secondary and highschool in Chongqing. From 2007, he startedstudying in Yingcai Honors College of Universityof Electronic Science and Technology of China.This Yingcai Honors College only enrolls the top

3% of the whole students in the university, which is based on the scoresof College Entrance Examination. Here, he obtained his Bachelor’s de-gree in Communication Engineering (Excellent Undergraduate StudentAward) in 2011. During his Bachelor program, he won the First Prizeat National Mathematic Contest in Modeling of China in 2009, and alsoSecond Prize at Interdisciplinary Contest in Modeling (ICM) in 2010.Since the beginning of 2012, he started a PhD project at Eindhoven Uni-versity of Technology (TU/e) in the Netherlands, in which he studiedthe psychological mechanisms underlying the effectiveness of using Am-bient Persuasive Technology in behavioral change. As while being a PhDstudent, he has also followed the main master program in Human Tech-nology Interaction at TU/e. During his PhD, he presented his work at anumber of national and international conferences. For his work, in June2015, he won the Best Paper Award at the 10th International Conferenceon Persuasive Technology, in Chicago, Illinois, USA.

161

using ambient lighting in persuasive communication

Documents