Vous êtes sur la page 1sur 20

Product personality in physical

interaction
Pieter M. A. Desmet, Juan Carlos Ortz Nicolas and Jan P. Schoormans,
Department of Industrial Design, Delft University of Technology,
Landbergstraat 15, 2628 CE Delft, The Netherlands
The possibility of designing physical humaneproduct interactions with
predened personalities was explored in two studies. In the rst study, 60
participants reported the personality of two devices that were developed to be
identical in terms of appearance and dierent in terms of interaction style, i.e.
dominant versus elegant. In the second study, 75 respondents reported the
personality of the same devices but this time with a dominant instead of a neutral
appearance. The results indicated that it was possible to design interaction
devices with dierent personalities, and that the eect of appearance is stronger
than that of interaction style.
2008 Elsevier Ltd. All rights reserved.
Keywords: design research, product design, psychology of design, product
personality
P
ersonality can be dened as a persons dynamic and organized set of
non-physical characteristics that uniquely inuences his or her
cognition, aect, and behaviour in various situations (Ryckman,
2004). In everyday life, we assess the personalities of individuals in order
to predict what we can expect of them (Carver and Scheier, 1996). An
awareness of personality thus helps us to operate successfully in our social
context. For instance, we feel less reluctant to ask a favour from a person
whom we believe to be friendly and kind-hearted than from someone with
a reserved and unkind personality. Personality serves a function in know-
ing other people and, in a sense, ourselves, because it helps us to identify
both dierences and similarities between people. The concept of personal-
ity is conceived in a similar way in the science of psychology. Although
there are an abundance of theoretical traditions in personality psychology,
two main aims have consistently been reported in its rich and varied
history. The rst is to develop tools and methods that enable us to
construct a coherent picture of a person and his or her major psycholog-
ical processes, and the second is to develop tools and methods that enable
us to study and understand individual dierences; in other words, how
people dier from each other (Carver and Scheier, 1996).
Corresponding author:
Pieter M. A. Desmet
p.m.a.desmet@
tudelft.nl
www.elsevier.com/locate/destud
0142-694X $ - see front matter Design Studies 29 (2008) 458e477
doi:10.1016/j.destud.2008.06.003 458
2008 Elsevier Ltd. All rights reserved. Printed in Great Britain
What does this have to do with product design? If personality is a concept that
represents those qualities of a person that inuence his or her cognition, aect,
and behaviour, one would assume that the use of this concept is limited to
describing humans. It would not seem to make sense to attribute personality
to things that cannot have or display thoughts, feelings, and intentional behav-
iour. And yet that is what people do, all the time. One attributes personality to
non-animate objects in the same way as one attributes personality to other
people. We systematically infer a persons personality from his or her
appearance. This is useful to us because making a quick judgement call about
someones personality may improve our chances of successful social interac-
tion, whatever our intention may be in this interaction. You do not ask for
directions from just anyone in the street; instead you pick the person who
seems friendly and harmless enough to respond positively to your inquiry.
Your assessment will be largely based on facial features, postures, clothes,
hairstyle, etc. Although ones appearance is not a part of ones personality,
the impact of appearance on how we infer a persons personality is supported
by many experimental results (Jones, 1990). Ones visual appearance includes
visual-static and visual-dynamic characteristics (Borkenau and Liebler, 1995).
Examples of the rst are the way of dress, hair colour and style, make-up,
facial features like the shape of eyes and nose, and examples of the latter are
gait, mimicking, facial expressions, and gestures. Studies have shown that
both types of visual characteristics inuence inferences about personality traits
(Montepare and Zebrowitz-McArthur, 1988; Zebrowitz, 1990; Borkenau and
Liebler, 1992).
Our ability, or rather, tendency, to infer personality from appearance does not
stop with people. We do this just as easily with animals (one might be
astonished by the extent to which sophisticated human traits are sometimes
attributed to pets) and non-animate objects. Researchers in various domains
have shown that people also use personality characteristics to describe brands
(Biel, 1993; Aaker, 1997), stores (Sirgy et al., 2000; dAstous and Le vesque,
2003), product appearance (Janlert and Stolterman, 1997; Jordan, 1997;
Govers, 2004), and computers (Nass et al., 1995; Nass and Lee, 2001). People
often think and talk about products as having a personality and relate to them
accordingly (Janlert and Stolterman, 1997). Govers et al. (2004) have called
this product personality, as people use human personality characteristics to
describe their impression of a product. Govers (2004) argued that product
personality cannot be reduced to distinct product attributes but rather refers
to a holistic description of the total product. Hence, the term non-physical
in the denition of human personality can be replaced with intangible
when adapting this denition to apply to product personality. In addition,
Govers (2004) demonstrated that, analogously to human personality, appear-
ance is a major determinant in the perception of product personality, and
assembled and validated a set of personality characteristics that are useful
for describing product appearance.
Product personality in physical interaction 459
1 Designing product personality
The phenomenon of product personality has not yet been widely studied. Jan-
lert and Stolterman (1997) wrote a conceptual paper about the character of
things, and Jordan (1997, 2002) conducted two applied studies in which the
participants were asked to think about products as if they were people and
to assign human personality traits to them. Govers et al. (2004) instructed de-
sign students to design irons that are either happy, cute, or tough. The re-
searchers selected three designs for each characteristic. Respondents were
instructed to rate the (drawings of the) irons on the three target traits. The re-
sults supported the hypothesis that the happy design was rated as happier, the
tough as tougher, and the cute as cuter than the other designs. Nass et al.
(1995) focused their study on dynamic interaction style rather than on static
appearance. They manipulated a software program in order to make two ver-
sions, one with a submissive and one with a dominant interaction style.
1
Re-
spondents were instructed to perform a standardized problem-solving task
on the computer. After nishing the task, they rated the personality of the
computers. The results conrmed the hypothesis that the dominant software
would be perceived as more dominant and the submissive software as more
submissive.
These last two studies are particularly interesting because they indicate that, to
a certain extent, it is possible to design products with a predened personality.
It remains unclear, however, to what degree the physical interaction style of
a product can be designed to have or contribute to a predened personality.
Govers et al. (2004) focused on (pictures of) product appearance, and Nass
et al. (1995) focused on virtual agents.
2
Given the fact that one judges
a persons personality partly on the basis of this persons dynamic appearance,
we assume that this would also apply to consumer products. For that reason,
the present paper addresses the following question: is it possible to convert
predened personality traits into dynamic humaneproduct interaction?
With dynamic humaneproduct interaction, we refer to the act of physically
manipulating a product. To investigate this question a research through
design approach (see Frayling, 1993; Zimmerman et al., 2007) was adopted,
in which two interaction devices were designed with the aim of expressing
dierent personalities in their behaviour. It was decided to develop devices
(instead of using existing consumer products) because this facilitated a high
degree of freedom in manipulating the stimulis behaviour. This design free-
dom enabled us to explore the possibilities of actually designing for product
personality. Two studies were performed with these devices. Study 1 was
designed to test the extent to which the participants perceived the intended
personalities. Study 2 was designed to investigate the inuence of appearance
on the perceived personality of these interaction devices. The present paper
briey describes the development of the devices, and subsequently reports
the two studies.
460 Design Studies Vol 29 No. 5 September 2008
2 Study 1: personality in interaction
The aim of Study 1 was to explore the possibility of designing interaction
devices that have dierent and predened personalities. Two devices were
developed: one with an intended dominant and one with an intended elegant
interaction personality. In Study 1, it was tested whether the device with an
intended dominant interaction was perceived as having a more dominant
and less elegant personality than the device with an intended elegant
interaction.
2.1 Method
2.1.1 Participants
Sixty Dutch students of the Department of Industrial Design from Delft
University of Technology (31 males and 29 females) between 21 and 40 years
of age (M26.13; SD2.96) participated. They were recruited from the
university cafeteria of Delft University of Technology and received a small
gift for their participation. This test was a between-participant procedure.
Half of the group of participants responded to an interaction device that
was intended to be dominant, and the other half responded to an interaction
device that was intended to be elegant.
2.1.2 Stimulus materials
Study 1 focused exclusively on dynamic physical interaction. This implies that
we manipulated product attributes that can only be perceived when users
physically interact with the product. In order to prevent visual appearance
from inuencing the results, two devices were designed that were identical in
terms of appearance, but dierent in terms of dynamic behaviour. The devel-
opment of these devices involved a series of iterative steps. Two workshops
were performed in order to select suitable personality traits and to explore
what kind of interaction characteristics contribute to these personality traits.
Below, a short overview of the main steps is provided. For a detailed account
of the workshops and design process, see Ort z Nicola s (2006).
2.1.3 Workshop 1: selecting personality traits
The aim of the rst workshop was to identify personality traits that can be
expressed in physical humaneproduct interaction, and to select two of these
traits to be used as the basis for designing two interaction devices. The set
of 20 distinct traits included in Govers (2004) product personality scale served
as a reference: cheerful, open, relaxed, easygoing, pretty, cute, idiosyncratic,
provocative, interesting, lively, dominant, obtrusive, untidy, childish, silly,
boring, aloof, serious, modest, and honest. Nine masters students in industrial
design participated. They rst performed an individual task to sensitize them
for the topic of personality in interaction. Various products (for example,
coee machines, computer mice, and music players) were provided. Partici-
pants were asked to describe the personality of these products in as much detail
Product personality in physical interaction 461
as possible while interacting with them. Eighteen of the 20 reference set traits
were mentioned at least once in these descriptions; pretty and cute were the
only two traits that were not mentioned at all. In addition, some traits that
are not included in the reference set were also mentioned. The two additional
traits mentioned most frequently were elegant and gentle. These particular two
traits are represented by a more extensive set that Govers (2004) used as the
basis for the personality scale with 20 trait items. They were omitted from
that initial set because Govers found in a questionnaire study that they were
not often used to describe product personality. We decided to add them to
the reference set for the current study because they were often used to describe
the personality of physical interaction behaviour in this part of the workshop.
In the second part of the workshop the participants explored in groups to what
extent they would be able to design each of the 22 personality traits in physical
interaction behaviour. These explorations indicated that it is harder to design
interaction to have certain personality traits (like, for example, untidy and
serious) than others (like, for example, easygoing and lively). During the
design explorations, seven traits were identied as traits that the participating
designers would be able to design in physical interaction behaviour: boring,
dominant, easygoing, elegant, gentle, honest, and lively. From these seven
elegant and dominant were selected for developing the interaction devices.
These traits were selected because (a) a comparison study requires a minimum
of two stimuli that are manipulated on one attribute, and (b) these two traits
are dissimilar without representing poles of a single dimension (see Carver and
Scheier, 1996).
2.1.4 Workshop 2: exploring design attributes
The aim of the second workshop was to generate insights into how an elegant
and a dominant personality can be designed in physical product interactions.
For this reason, we explored the relationships between product attributes and
perceived personalities. Seven masters students in industrial design partici-
pated. Five aspects of physical product interaction, proposed by Janlert and
Stolterman (1997), Klooster and Overbeeke (2005), and Saer (2006), were
used as a framework for the workshop: force, sound, motion, texture, and
performance. These aspects were used to explore the subjective product
attributes that contribute to a dominant versus an elegant interaction person-
ality. Our approach was to describe subjective instead of objective product
attributes because personality refers to a holistic description of the total prod-
uct (Govers, 2004). Moreover, this approach has been reported to be useful in
design projects that are characterized by experiential design goals or intentions
(see Hekkert et al., 2003; Desmet et al., 2007).
One of the activities was a series of role-playing situations. An example is
a typical restaurant scenario in which one participant acted as a dominant cus-
tomer, and another as an elegant waiter, or vice versa. The customer and
waiter demonstrated their personality with bodily and facial expressions and
462 Design Studies Vol 29 No. 5 September 2008
behaviour. They were allowed to make sounds, but not to talk. Another task
involved playing a dominant or elegant product. One participant portrayed
the product, which was either dominant or elegant, and the other portrayed
the user. After each task, participants discussed what product attributes
contributed to the intended quality of interaction. Two workshop sessions
resulted in a set of attributes, which is summarized in Table 1. These were
used in the design stage to develop interaction devices that are intended to
be experienced as either dominant or elegant.
2.1.5 Interaction devices
Several device designs were developed and tested in an iterative design process.
Figure 1 shows three initial designs. Design A is a sliding block puzzle in which
interaction personality is inuenced by manipulating the materials, weights,
and textures of the blocks and the base. Design B is a pantograph in which
personality was inuenced by manipulating the weights of the beams and
the textures and materials of the joints. Design C is a rope that is attached
to a mechanism with weights and elastic strings to inuence interaction
personality.
Device design C was selected because it provided more possibilities for
inuencing the interaction personality than the other two design ideas. The
selected devices are shown in Figures 2 and 3: two variants of a mechanism
with a handle on a rope. The handle can be pulled from the container and re-
turns to the neutral position when no force is applied. The rope is attached to
a mechanism that is designed to give the interaction e the action of pulling e
a particular personality by using design elements such as weights, slopes,
shapes, elastic strings, and textures.
Table 1 Subjective product attributes involved in interaction behaviour that is perceived as dominant or elegant
Interaction dimension Dominant interaction Elegant interaction
Force Heavy Light
Energetic Energetic
Rigid (requires strength) Easy
Sound Loud Soft
Fragmented Harmonious
Discordant Clear
Motion Slow Extended
Bumpy Smooth
Firm Supple
Texture Rough Fluid
Irregular Rened
Solid Flexible
Performance Restrictive Balanced
Complicated Graceful
Irregular Accessible
Product personality in physical interaction 463
2.1.6 Dominant interaction device (D-Int)
The rope is attached to a steel wheel that is placed on a slope. The wheel is
connected to an elastic string that reacts against pulling: the required force
increases with the distance between the handle and the container. The shape
of the slope was designed to disrupt the uency of interaction, forcing the
user to make short movements that require additional threshold force. Glass
marbles were inserted in various points of the slope to introduce additional
Figure 1 Three interaction device concepts
Figure 2 Design intended for dominant interaction
464 Design Studies Vol 29 No. 5 September 2008
discontinuities in the interaction. The metal wheel is partly covered with
rubber to generate an irregular sound.
2.1.7 Elegant interaction device (E-Int)
The rope is attached to a horizontally supported plate. Another rope connects
the other side of the plate to a counterweight. A small wheel with a tension
spring is attached to the base of the plate. The wheel rolls along an undulating
Figure 3 Design intended for elegant interaction
Figure 4 Design intended for dominant appearance
Product personality in physical interaction 465
path. The materials and surfaces of the path and wheel were selected to
generate a uid tactile quality and a harmonious sound.
2.1.8 Interaction task
Since the stimulus was not the device as such but the physical interaction with
the device, respondents were required to perform a prescribed interaction task.
Seven aluminium tubes (10 mm diameter) were hung on ropes down from the
ceiling in front of the interaction device. The task was to take the handle of the
device (in any hand) and to reach out to touch each of the aluminium tubes
with the handle (see Figure 4). The participant was free to choose in which
order to touch the tubes, but after touching each tube the handle was to be
returned to the starting position. The distance between the tubes and the oor
varied between 10 and 140 cm and the distance between the tubes and the
devices container varied between 40 and 150 cm.
2.1.9 Questionnaire
A two-part questionnaire was developed. In the rst part, participants
answered general questions about their age and gender. In the second part,
participants answered questions about the perceived personality and attributes
of the interaction device. Seven personality traits were included: boring,
dominant, honest, easygoing, gentle, lively, and elegant. These were the
personality traits that were often used to describe product personality in phys-
ical interaction in workshop 1. The following question was posed: If you
would consider this object to have a personality, what would it be? In
addition, ten attributes were included: I found this object very balanced, acces-
sible, exible, heavy, loud, complicated, rened, restrictive, smooth, and
required a lot of strength to use. These ten attributes were selected from those
that had been used in the design phase of stimuli development. This set of ten
was selected because it represents the ve aspects of product interactions and
comprises the attributes that the designer of the devices identied as particu-
larly helpful in the design process. All personality and attribute items were
measured with 7-point Likert scales. Five versions of the questionnaire were
made, with the personality and attribute order randomized in each.
2.1.10 Procedure
The test was completed individually. In the experiment, the interaction device
was hidden behind a big screen that was designed to look like a room divider
that was part of the room structure. The screen blocked the respondents view
of the device. In Study 2 we will refer to this as the neutral appearance.
Written instructions explaining the test procedure and the interaction task
were handed to the respondent when he or she entered the room. One of the
ve randomly selected questionnaires was also handed to the respondent.
After reading the instructions, the respondent performed the task and lled
out the questionnaire in as much time as he or she needed. All respondents
nished the task within 30 min.
466 Design Studies Vol 29 No. 5 September 2008
2.2 Results
2.2.1 Product personality
To investigate the dierences in perceived personality between the two
interaction devices, a repeated measures MANOVA was performed with
Stimulus (two levels) as the between-participant factor, and the seven person-
ality traits as the dependents. The multivariate test showed a signicant eect
of Stimulus [F(7,52) 6.74, p <0.001], indicating that the two devices were
perceived as having dierent personalities. Univariate tests showed a main
Stimulus eect (p <0.05; see Table 2) for all personality traits except for
honest and elegant, which indicates that according to the participants the
devices (D-Int and E-Int) diered in terms of how boring, dominant,
easygoing, gentle, and lively they are.
The last column in Table 2 shows the mean dierences for personality traits
between the D-Int and the E-Int devices. These dierences indicate that the
personality of D-Int is perceived as more dominant and lively than the person-
ality of E-Int. The personality of E-Int is perceived as more boring, easygoing,
and gentle than the personality of D-Int. The D-Int device rates highest on the
traits dominant and lively, whereas the E-Int device rates highest on the traits
easygoing and honest.
2.2.2 Product attributes
To explore the relationships between perceived personality and perceived
product attributes, a repeated measures MANOVA was performed with
Stimulus (two levels) as the between-participant factor, and the ten measured
attributes as the dependents. The multivariate test showed a signicant eect
of Stimulus [F(10,49) 8.27, p <0.001], indicating that the two devices dier
in terms of perceived attributes. Univariate tests showed a main Stimulus eect
Table 2 Mean ratings and pairwise comparison for seven personality traits
Perceived personality Interaction device Dierence
(D-Int E-Int)
D-Int E-Int
M SD M SD
Boring 3.03 1.43 4.20 1.49 1.17*
Dominant 4.87 1.43 3.30 1.73 1.57*
Honest 3.93 1.55 4.63 1.40 0.70
Easygoing 2.47 1.20 4.70 1.70 2.23*
Gentle 2.57 1.50 4.03 1.43 1.47*
Lively 4.67 1.69 3.30 1.47 1.37*
Elegant 3.03 1.88 3.20 1.40 0.17
Based on estimated marginal means. *The mean dierence is signicant at the 0.05 level. Adjustment for multiple compar-
isons: Bonferroni.
Product personality in physical interaction 467
(p <0.05; see Table 2) for all attributes, except loud, rened, and restrictive,
indicating an inuence of these attributes on the perceived personality.
The last column in Table 3 shows the mean dierences for perceived attributes
between the D-Int and the E-Int devices. D-Int is perceived as more heavy and
complicated than E-Int, and E-Int is perceived as more balanced, accessible,
exible, and smooth than D-Int. The D-Int device rates highest on the attri-
butes restrictive, heavy, and loud, and the E-Int device rates highest on the
attributes accessible, restrictive, and exible.
In order to investigate the association between the attributes and the dominant
and elegant product personality, bivariate (Pearson) correlations were
computed between each of the product attributes and the personality traits
dominant, elegant, and easygoing (Table 4). The personality trait easygoing
was included in the analyses because the personality of E-Int is perceived as
more easygoing than elegant.
Table 4 indicates that perceived dominance is signicantly correlated with the
attributes heavy, complicated, smooth, and strength required: a perceived
dominance personality is associated with interactions that are perceived as
complicated, heavy, not smooth, and requiring a lot of strength. Perceived
elegance correlates signicantly with balanced, exible, loud, rened, and
smooth. This indicates that an elegant personality is associated with interac-
tions that are rened, balanced, smooth, exible, and not loud. Easygoing
correlates signicantly with balanced, exible, heavy, complicated, rened,
smooth, and strength required: an easygoing personality is associated with
interactions that are not heavy, do not require strength, are smooth, exible,
and not complicated. The results of the rst study indicate that we succeeded,
Table 3 Mean ratings and pairwise comparison for ten perceived product attributes
Perceived product attribute Interaction device Dierence
(D-Int E-Int)
D-Int E-Int
M SD M SD
Balanced 3.00 1.60 4.27 1.57 1.27*
Accessible 4.03 1.67 5.00 1.34 0.97*
Flexible 3.13 1.61 4.67 1.49 1.53*
Heavy 4.23 1.28 2.00 1.05 2.23*
Loud 4.20 1.97 4.13 2.06 0.07
Complicated 3.57 1.45 2.47 1.48 1.10*
Rened 2.73 1.46 3.27 1.57 0.53
Restrictive 4.70 1.42 4.67 1.60 0.53
Smooth 2.33 1.30 3.70 1.73 1.37*
Requires strength 3.97 1.77 1.57 0.77 2.40*
Based on estimated marginal means. *The mean dierence is signicant at the 0.05 level. Adjustment for multiple
comparisons: Bonferroni.
468 Design Studies Vol 29 No. 5 September 2008
at least for a large part, in designing an intended product interaction
personality by varying these ten product attributes.
3 Study 2: inuence of appearance on personality
Study 2 was designed to explore the interaction between product interaction
personality and product appearance personality. For this study additional
data was collected and compared with the data of Study 1.
3.1 Method
3.1.1 Participants
Seventy-ve Dutch students of the Department of Industrial Design from
Delft University of Technology (39 males and 36 females) between 21 and
55 years of age (M27.16; SD7.55) participated. None participated in
the workshops or in Study 1. They were recruited from the university cafeteria
of Delft University of Technology and received a small gift for their participa-
tion. This test was a between-participant procedure, in which the number of
participants was equally divided over the stimuli.
3.1.2 Stimulus materials
In this study, three stimuli were used. One was a visual stimulus, and the other
two were combined visual and interaction stimuli. Table 5 gives an overview of
the stimuli used in Study 1 and 2.
A container was designed and built to have a dominant visual appearance.
Several designs were made, and with the help of a pilot study (not reported
here) the most dominant was selected (Figure 4). The basic dimensions of
the container were 200 by 150 by 60 cm. This container (D-App) was the rst
stimulus of Study 2. The second stimulus was the dominant interaction device
placed in the dominant appearance container (D-Int-D-App), and the third
Table 4 Correlations between ten perceived product attributes and three perceived personality traits
Perceived product attribute Perceived personality trait
Dominant Elegant Easygoing
Balanced 0.10 0.48** 0.31*
Accessible 0.17 0.15 0.19
Flexible 0.23 0.34** 0.51**
Heavy 0.45** 0.13 0.61**
Loud 0.09 0.33* 0.05
Complicated 0.51** 0.03 0.42**
Rened 0.10 0.49** 0.30*
Restrictive 0.22 0.19 0.25
Smooth 0.44** 0.40** 0.53**
Requires strength 0.47** 0.03 0.58**
**Correlation is signicant at the 0.01 level (two-tailed); *correlation is signicant at the 0.05 level (two-tailed).
Product personality in physical interaction 469
stimulus was the elegant interaction device placed in the dominant appearance
container (E-Int-D-App). Respondents that responded to stimulus D-App did
not interact with, but only observed the object. Respondents that responded to
stimulus D-Int-D-App and stimulus E-Int-D-App performed a task that was
identical to the task in Study 1.
3.1.3 Questionnaire and procedure
A questionnaire was developed to measure the same personality trait items as
the questionnaire that was used in Study 1. Five versions of the questionnaire
were made, with the personality and character order randomized in each. The
procedure was identical to the rst part of the study.
3.2 Results
The dataset of Study 2 was combined with the dataset of Study 1. To explore
the relationship between interaction, appearance, and personality, a repeated
measures MANOVA was performed with Appearance (two levels; Intended
Neutral Appearance versus Intended Dominant Appearance) and Interaction
(two levels; Intended Dominant Interaction versus Intended Elegant Interac-
tion) as the between-participant factors, and the seven personality traits as
the dependents. The multivariate test showed a signicant eect of Appear-
ance [F(7,100) 6.54, p <0.001], Interaction [F(7,100) 4.14, p <0.001],
and the two-way interaction [F(7,100) 3.97, p <0.01]. Univariate tests
showed main eects (p <0.05) for all personality traits, except for honest
and boring (last column Table 6).
Table 6 shows that the personality of the interaction devices with a dominant
appearance is perceived as more dominant, lively, and less easygoing and
gentle than the personality of the interaction devices with a neutral appear-
ance. As we also found a signicant Interaction Appearance interaction
eect, we decided to proceed with analyses of appearance eects per interac-
tion device. For each interaction device, a repeated measures ANOVA was
performed with Appearance (two levels) as the between-participant factor,
and the personality traits as the dependents. Table 7 shows the results for
Table 5 Stimuli used in Study 1 and in Study 2
Study Stimulus code Intended interaction
personality
Intended appearance
personality
Study 1 D-Int Dominant Neutral
E-Int Elegant Neutral
Study 2 D-App e Dominant
D-Int-D-App Dominant Dominant
E-Int-D-App Elegant Dominant
470 Design Studies Vol 29 No. 5 September 2008
the dominant interaction device, and Table 8 shows the results for the elegant
interaction device.
Table 7 indicates that of all personality traits, both stimulus D-Int-D-App and
D-Int rate highest on the trait dominant. The last column in Table 7 shows the
mean dierences for personality traits between the two stimuli. An interesting
nding is that the personality of the dominant interaction device with
a dominant appearance is perceived as less lively and more boring than the
personality of the device with a neutral appearance.
Table 8 indicates that of all personality traits, stimulus E-Int rates highest on
the trait easygoing, and stimulus E-Int-D-App rates highest on the trait
dominant. So, even though the interaction is perceived as easygoing, the
personality is still predominantly rated as dominant. The last column in
Table 8 shows the mean dierences for personality traits between the two
stimuli. It indicates that the personality of the elegant interaction device
Table 7 Appearance effect on perceived personality of dominant interaction device
Perceived
personality
Intended dominant interaction
Intended neutral
appearance (D-Int)
Intended dominant
appearance
(D-Int-D-App)
Dierence (intended
neutral intended
dominant appearance)
Boring 3.03 4.28 1.25*
Dominant 4.87 5.52 0.65
Honest 3.93 4.28 0.35
Easygoing 2.47 2.64 0.17
Gentle 2.57 2.56 0.01
Lively 4.67 2.84 1.83*
Elegant 3.03 2.32 0.71
Based on estimated marginal means. *The mean dierence is signicant at the 0.05 level.
Adjustment for multiple comparisons: Bonferroni.
Table 6 Appearance effect on perceived personality
Perceived personality Intended neutral
appearance
Intended dominant
appearance
Dierence
(neutral dominant)
M SD M SD
Boring 3.62 1.56 4.12 1.64 0.50
Dominant 4.08 1.76 5.60 1.28 1.52*
Honest 4.28 1.51 4.10 1.53 0.18
Easygoing 3.58 1.84 2.72 1.62 0.86*
Gentle 3.30 1.63 2.72 1.50 0.58*
Lively 3.98 1.71 2.84 1.46 1.14*
Elegant 3.12 1.65 2.28 1.26 0.84*
Based on estimated marginal means. *The mean dierence is signicant at the 0.05 level.
Adjustment for multiple comparisons: Bonferroni.
Product personality in physical interaction 471
with a neutral appearance is perceived as more easygoing, gentle and elegant,
and less dominant than the same interaction device with a dominant
appearance.
4 Discussion
The purpose of this research was to determine if designers can convert
predened personality traits into dynamic humaneproduct interactions. The
results indicate that we (partly) succeeded in designing these predened
personalities. Signicantly dierent personalities were attributed to the two
interaction devices. The device that was intended to be dominant was indeed
perceived as more dominant than the device that was intended to be elegant.
The dominant interaction device was found to be perceived as heavier, more
complicated, less smooth, and requiring more strength than the elegant inter-
action device; each of these attributes were found to be related to perceived
dominance. However, no signicant device eect on perceived elegance was
found. This means that the device that was intended to be elegant was not per-
ceived as more elegant than the other device. Even so, it was also found that
although the elegant interaction device was not perceived as more elegant, it
was perceived as more easygoing than the dominant interaction device. This
may indicate that we had designed a device with an easygoing rather than
an elegant interaction. The results show that an elegant personality is to
some degree associated with the same attributes as an easygoing interaction:
both are associated with interactions that are balanced, exible, rened, and
smooth. The results also indicate some dierences. An easygoing personality
is associated with some interaction attributes that were not correlated with
an elegant personality: not-heavy interactions that do not require strength
and are not complicated. The fact that the elegant interaction device rated
signicantly higher than the dominant interaction device on these three
attributes supports the idea that the designer had created an easygoing rather
than an elegant interaction.
Table 8 Appearance effect on perceived personality of elegant interaction device
Perceived
personality
Intended elegant interaction
Intended neutral
Appearance
(E-Int)
Intended dominant
appearance
(E-Int-D-App)
Dierence (intended
neutral intended
dominant appearance)
Boring 4.03 3.96 0.07
Dominant 2.33 5.68 3.35*
Honest 4.30 3.92 0.38
Easygoing 5.80 2.80 3.00*
Gentle 5.00 2.88 2.12*
Lively 3.40 2.84 0.56
Elegant 4.33 2.24 2.09*
Based on estimated marginal means. *The mean dierence is signicant at the 0.05 level.
Adjustment for multiple comparisons: Bonferroni.
472 Design Studies Vol 29 No. 5 September 2008
Appearance was found to dominate product personality. The elegant interac-
tion device with a dominant appearance rated higher on dominance than on
any other personality trait (including elegant and easygoing). So although
interaction style does inuence personality, in this study the inuence of
appearance outweighed it. Note, however, that the personality of the elegant
interaction device with a neutral appearance was perceived as more easygoing,
gentle and elegant, and less dominant than the same interaction device with
a dominant appearance.
An additional interesting nding was that the inuence of appearance diered
between the two interaction personalities. In the case of the dominant interac-
tion, adding a dominant appearance made it less lively and more boring.
Maybe the personality of the interaction is more predictable when it is congru-
ent with the personality of the appearance, and therefore perceived as more
boring. This would match the ndings of Ludden et al. (2004) who showed
that incongruence between the senses can elicit surprise.
5 Conclusions
In conclusion, the results of the current studies suggest that (a) to some degree
designers can create predened product personalities with dynamic interac-
tion, (b) the eect of appearance on personality is more powerful than the
eect of dynamic interaction, and (c) perceived product personality is not
a straightforward blend of the eect of appearance and the eect of dynamic
interaction.
In this study, the eect of appearance on product personality was stronger
than the eect of interaction. This seems comparable with how we assess the
personality of people: the rst, and thus in most cases visual, impression has
the most immediate impact. If we extend this comparison, we may wonder
what the inuence of time on this eect might be. Once you get to know
a person, his or her behaviour may become more important than appearance
in your assessment of his or personality. Perhaps the same applies to products.
An interesting additional research step would be to explore the relative inu-
ence on perceived personality of interaction and appearance at several stages
of product purchase and ownership. One may also wonder whether the order
in which appearance and interaction are experienced would have an eect on
their relative importance to personality. This may be tested by blindfolding
respondents and having them experience the interaction rst before removing
the blindfold and having them experience the appearance. A complicating
factor would be that blindfolding will also inuence experience and therefore
possibly perceived personality.
Note that we did not control for the personality of the participants. Interaction
is determined not only by the product, but also by the participant, and there-
fore the personality of interaction may also be inuenced by both the
Product personality in physical interaction 473
personality of the product and the personality of the user. The concept of the
interaction device and task (moving a handle to tubes hanging in the air) itself
may also have had a personality, that is, a baseline personality, which was not
tested for. Another complexity in studying product personality is that one
should be aware that sometimes a word can have more than one meaning.
The word tough, for example, can be used to refer to a non-physical trait,
but also to a physical property. When we say a person is tough (or weak)
we probably do not refer to the physical properties of his or her body, but
rather to his or her personality. We have to be careful not to confuse the literal
and metaphorical senses of words in our product personality research; surely
a metal chair will be judged as tougher than a plastic one, but that does not
necessarily mean that people judge this chair as having a tougher personality.
In other words, the fact that we use the same word to describe the personality
of a person (this person is soft) and to describe a product (this product is soft)
does not necessarily imply that the same meaning is intended; we may merely
use the same words to refer to dierent things.
What is the use of designing for interaction personality? Personalities are
mental tools used to handle a complex reality and therefore to reduce the
mental eort involved in dealing with people and products. In ascribing
a certain personality to a product we make a very simple, but powerful descrip-
tion that frequently will be accurate enough to help us to manage the task of
successfully using the product (see Janlert and Stolterman, 1997). The fact that
dynamic behaviour also inuences the perceived personality oers the possibil-
ity of designing dynamic personalities that can adapt to the usage context,
thereby helping the user to successfully use the product without having to
deal with complex information about its inner structure. In addition, Fiske
and Taylor (1991) proposed that in social situations people prefer to interact
with individuals who behave consistently as compared with individuals who
behave inconsistently. Consistency makes it easier to predict how people will
behave and respond during social interaction, which may also apply to
humaneproduct interaction. Note that reducing the mental load is not the
only possible design strategy. In some cases it is more favourable to deliber-
ately use inconsistencies to design products that are perceived as innovative
and rich in their interaction (see Overbeeke et al., 2002; Desmet, 2003).
Product personality does inuence purchase decisions (Govers, 2004), and our
study demonstrates that appearance is not the only means of inuencing prod-
uct personality. A recent trend is to have consumers try out the products in the
store, which strengthens the necessity to understand the qualities of interaction.
It is especially interesting to explore the eects of appearance personality e
interaction personality (in)congruency. A device with a neutral appearance
and dominant interaction was found to be much more lively and less boring
than a device with a dominant appearance and dominant interaction. This
indicates that it may be worthwhile to invest resources in exploring interaction
474 Design Studies Vol 29 No. 5 September 2008
personality in design projects, although this approach may be more complex
than restricting the focus on appearance personality because interaction
personality is dicult to simulate (i.e. it cannot be drawn, it has to be felt).
Most of us are familiar with the experience of misjudging someones personal-
ity. It can be a disappointing experience when a person who appears to be kind
turns out to display behaviour that is not so kind. But it can also prove to be
a pleasant surprise when, for example, a person that looks boring turns out
to be witty and funny. When designing the appearance of a product to have
a particular personality, the designer should be aware that this personality
sets up expectations about the personality of interaction. Designers can seek
to understand these expectations in order to generate consistent experiences,
or in some cases perhaps in order to play with them e designing products
that facilitate intense and multifaceted interactions which are not experienced
as straightforward and eortless but rather as unique and fascinating.
1. The dominant computer used strong language expressed in the form of assertions and
commands, displayed a high condence level, and always went rst in the interaction.
Conversely, the submissive computer used weaker language expressed in the form of
questions and suggestions, displayed a low condence level, and always went second
in the interaction. Note that in this study the functional information conveyed by the
computer was not manipulated; only the style of communication was manipulated. All
responses were pre-programmed, and there was no natural language processing or arti-
cial intelligence employed.
2. In the eld of human computer interaction, many research initiatives focus on the per-
sonality of computers as generated with agents (Min Lee and Nass, 2003; Masuch
et al., 2006) or characters (Hayes-Roth and Doyle, 1998; Neubauer, 2004).
References
Aaker, J L (1997) Dimensions of brand personality, Journal of Marketing Re-
search Vol 34 pp 347e356
dAstous, A and Le vesque, M (2003) A scale for measuring store personality,
Psychology and Marketing Vol 41 pp 455e469
Biel, A L (1993) Converting image into equity in D A Aaker and A L Biel (eds)
Brand equity and advertising Lawrence Erlbaum, Hillsdale, USA pp 67e82
Borkenau, P and Liebler, A (1995) Observable attributes as manifestations and
cues of personality and intelligence, Journal of Personality Vol 63 pp 1e25
Borkenau, P and Liebler, A (1992) The cross-modal consistency of personality:
inferring strangers traits from visual or acoustic information, Journal of Research
in Personality Vol 26 pp 128e204
Carver, C S and Scheier, M F (1996) Perspectives on personality (3
rd
edn) Allyn &
Bacon, Needham Heights, USA
Desmet, P M A (2003) Measuring emotion: development and application of an
instrument to measure emotional responses to products in M A Blythe,
A F Monk, K Overbeeke and P C Wright (eds) Funology: from usability to
enjoyment Kluwer Academic Publishers, Dordrecht, NL pp 111e123
Desmet, P M A, Porcelijn, R and Van Dijk, M B (2007) Emotional design:
application of a research-based design approach, Knowledge, Technology and
Policy Vol 20 pp 141e155
Fiske, S T and Taylor, S E (1991) Social cognition McGraw-Hill, New York, USA
Frayling, C (1993) Research in art and design, Royal College of Art Research
Papers Vol 1 pp 1e5
Product personality in physical interaction 475
Govers, P C M (2004) Product personality, unpublished doctoral dissertation,
Delft University of Technology, Delft, NL
Govers, P C M, Hekkert, P and Schoormans, J P L (2004) Happy, cute and
tough: can designers create a product personality that consumers understand?
in D MacDonagh, P Hekkert, J Van Erp and D Gyi (eds) Design and emotion,
the experience of everyday things Taylor & Francis, London, UK pp 345e349
Hayes-Roth, B and Doyle, P (1998) Animate characters, Autonomous Agents and
Multi-Agent Systems Vol 1 No 2 pp 195e230
Hekkert, P, Mosterd, M and Stomp, G (2003) Dancing with a machine: a case of
experience-driven design in Proceedings of the 2003 International Conference on
Designing Pleasurable Products and Interfaces, Pittsburgh, PA, USA, ACM Press,
New York, USA pp 114e119
Janlert, L E and Stolterman, E (1997) The character of things, Design Studies
Vol 18 pp 297e314
Jones, E E (1990) Interpersonal perception W.H. Freeman and Company,
New York, USA
Jordan, P W (1997) Products as personalities in S A Robertson (ed) Contemporary
ergonomics Taylor & Francis, London, UK pp 73e78
Jordan, PW(2002) The personalities of products inWSGreen and PWJordan (eds)
Pleasure with products: beyond usability Taylor & Francis, London, UK pp 49e60
Klooster, S, and Overbeeke, C J (2005) Designing products as an integral part of
choreography of interaction, in Proceedings of the 1st European Workshop on
Design and Semantics of Form and Movement, New Castle, UK pp 23e35
Ludden, G D S, Schierstein, H N J, and Hekkert, P (July 2004) Surprises elicited
by products incorporating visualetactual incongruities, in Paper Presented at the
Fourth International Conference on Design and Emotion, Ankara, Turkey
Masuch, M, Hartman, K and Schuster, G (2006) Emotional agents for interactive
environments in Proceedings of the Fourth International Conference on Creating,
Connecting and Collaborating through Computing, IEEE Computer Society,
Washington, USA Vol 00 pp 96e102
Min Lee, K and Nass, C (2003) Designing social presence of social actors in
human computer interaction in Proceedings of the SIGCHI conference on Human
Factors in Computing Systems, Ft. Lauderdale, Florida, USA, ACM Press, New
York, USA Vol 00 pp 289e296
Montepare, J M and Zebrowitz-McArthur, L (1988) Impressions of people created
by age-related qualities of their gaits, Journal of Personality and Social Psychology
Vol 55 pp 547e556
Nass, C and Lee, K M (2001) Does computer-synthesized speech manifest
personality? Experimental tests of recognition, similarity-attraction, and consis-
tency-attraction, Journal of Applied Experimental Psychology Vol 7 No 3 pp
171e181
Nass, C, Moon, Y, Fogg, B J, Reeves, B and Dryer, D C (1995) Can computer
personalities be human personalities, International Journal of HumaneComputer
Studies Vol 43 pp 223e239
Neubauer, B J (2004) Designing articial personalities using Jungian theory,
Journal of Computing Science Vol 20 No 1 pp 297e305
Ortz Nicola s, J C (2006) Product personality in interaction, unpublished masters
thesis, Delft University of Technology, Delft, NL
Overbeeke, C J, Djajadiningrat, J P, Hummels, C C M and Wenseen, S A G
(2002) Beauty in usability: forget about ease of use! in W S Green and
P W Jordan (eds) Pleasure with products: beyond usability Taylor & Francis,
London, UK pp 9e18
Ryckman, R (2004) Theories of personality Thomson/Wadsworth, Belmont, USA
476 Design Studies Vol 29 No. 5 September 2008
Saer, D (2006) Designing for interaction: creating smart applications and clever
devices Peachpit Press, Berkeley, CA
Sirgy, M J, Grewal, D and Mangleburg, T (2000) Retain environment,
self-congruity, and retail patronage: an integrative model and a research agenda,
Journal of Business Research Vol 49 pp 127e138
Zebrowitz, L A (1990) Social perception Brooks Cole, Pacic Grove, USA
Zimmerman, J, Forlizzi, J and Evenson, S (2007) Research through design as
a method for interaction design research in Proceedings of the CHI conference
on Human Factors in Computing Systems, San Jose, CA, USA, ACM Press,
New York, USA pp 493e502
Product personality in physical interaction 477

Vous aimerez peut-être aussi