Imitating Human Forms in Product Design

Transcription

Imitating Human Forms in Product Design: How
Does Anthropomorphism Work, When Does It Work,
and What Does It Affect
DISSERTATION
of the University of St. Gallen,
School of Management,
Economics, Law and Social Sciences
and International Affairs
to obtain the title of
Doctor of Philosophy in Management
submitted by
Linda Miesler
from
Germany
Approved on the application of
Prof. Dr. Andreas Herrmann
and
Prof. Dr. Oliver Gassmann
Dissertation no. 3917
ZSUZ Druckerei, Zürich, 2011
The University of St. Gallen, School of Management, Economics, Law, Social
Sciences and International Affairs hereby consents to the printing of the present
dissertation, without hereby expressing any opinion on the views herein expressed.
St. Gallen, May 13, 2011
The President:
Prof. Dr. Thomas Bieger
General Summary
It is a prevalent trend in design practice to mimic human-like shapes to foster
anthropomorphic perceptions in consumers. However, except for a few attempts in
marketing research, it is not well understood yet how consumers actually respond to
such anthropomorphic product designs. Therefore, it was the objective of my
dissertation to contribute to a broader understanding on how consumers perceive and
process anthropomorphic (face-like) product designs, and how (and when) such
human-like product forms affect consumer emotions and preferences. In five papers,
the results from a range of experimental studies are presented.
The first paper deals with the perceptual and cognitive processes underlying product
anthropomorphizing. We assumed that similar mental processes are involved when
consumers process face-like products and human faces. In particular, we studied
whether face-like product shapes trigger holistic processing in consumers, and how
such face-specific processing strategies affect consumer judgments about product
attributes such as quality.
The second paper deals with the nature of the emotional responses to face-like shapes
in product design. Applying an evolutionary psychology framework, we investigated
spontaneous affective responses to car fronts and faces, which both were manipulated
in accordance with the baby schema (“Kindchenschema”). Cues of the baby schema
(e.g., large eyes, small nose, high forehead), which are usually present in infant faces,
are known to trigger automatic positive emotions in humans, therefore, it was
compelling to examine if we find the same positive effects when baby-schema cues are
transferred to non-living objects such as cars.
The third paper addresses how context factors affect the preference for face-like
product designs. It was argued that design preferences might depend on the
consumers’ consumption goals (e.g., functional vs. emotional-hedonic), so that in
situations where a car should fulfil hedonic-emotional goals (e.g., be a friend)
anthropomorphic product designs might be preferred over less anthropomorphic ones,
whereas in situations where cars should mainly fulfil functional goals (e.g., transport)
non-anthropomorphic product designs might be preferred.
The fourth and the fifth paper have been previously published in conference
proceedings; they contain parts of the first and the second paper respectively.
Allgemeine Zusammenfassung
Es ist ein verbreiteter Trend in der Designpraxis, Produkte mit menschlichen Formen
auszustatten, um beim Konsumenten anthropomorphe Wahrnehmungen zu fördern.
Bis auf wenige Studien im Marketing wurde bisher kaum wissenschaftlich erforscht,
wie Konsumenten tatsächlich auf anthropomorphe Designs reagieren. In diesem Sinne
war es das Ziel dieser Dissertation, zu einem breiteren Verständnis darüber
beizutragen, wie Konsumenten anthropomorphe (gesichtsartige) Designs wahrnehmen
und verarbeiten, und welchen Effekt solche Designs (unter welchen Bedingungen) auf
Emotionen und Präferenzen des Konsumenten haben. In fünf Artikeln werden die
Ergebnisse aus einer Reihe von experimentellen Studien präsentiert.
Der erste Artikel beschäftigt sich mit perzeptuellen und kognitiven Prozessen, die der
Vermenschlichung von Produkten zugrunde liegen könnten. Es wurde die Annahme
überprüft, dass bei der Verarbeitung von gesichtsartigen Designs ähnliche mentale
Prozesse involviert sind wie bei der Gesichtsverarbeitung. Hierzu wurde untersucht, ob
gesichtsartige Autofronten holistische Verarbeitungsprozesse auslösen und wie sich
solche Mechanismen, die eigentlich typisch für die Verarbeitung von Gesichtern sind,
auf Urteilsprozesse von Konsumenten auswirken (z.B. Qualitätsurteile).
Der zweite Artikel beschäftigt sich mit emotionalen Reaktionen auf gesichtsartige
Designs. Da bekannt ist, dass Merkmale des Kindchenschemas (z.B. grosse Augen,
eine kleine Nase, eine hohe Stirn) aufgrund ihrer evolutionären Bedeutung beim
Betrachter automatisch positive Emotionen auslösen, wurde untersucht, ob
vergleichbare Effekte gefunden werden können, wenn das Kindchenschema auf
artifizielle Objekte wie Autofronten übertragen wird.
Der dritte Artikel beschreibt eine Studie, in der untersucht wurde, welchen Effekt
Kontexteinflüsse auf Präferenzen für gesichtsartige Designs haben. Es wurde davon
ausgegangen, dass die Ziele von Konsumenten, die sie durch den Kauf oder die
Benutzung eines Produkts versuchen zu erreichen (d.h. hedonistische vs.
utilitaristische Ziele), darüber entscheiden, welche Produktdesigns sie bevorzugen.
Der vierte und fünfte Artikel wurden bereits in Konferenzbänden veröffentlicht und
beschreiben
Ausschnitte
aus
dem
ersten
bzw.
zweiten
Artikel.
Table of Contents
I.
Paper 1 (Main Paper) ...............................................................................p. 05 – 43
Miesler, Linda, Jan R. Landwehr, Ann L. McGill, and Andreas Herrmann, “The
Face of Anthropomorphism: The Effects of Face-like Product Design Features on
Consumers’ Perceptions and Evaluations of Products.”
II.
Paper 2 (Main Paper) ............................................................................ ...p. 44 – 75
Miesler, Linda, Helmut Leder, and Andreas Herrmann, “Affective Consumer
Responses to Babies in Non-Faces and Non-Babies: An Evolutionary Perspective
of Baby-Schema Effects in Product Designs.”
III. Paper 3 (Main Paper) ...............................................................................p. 76 – 91
Miesler, Linda, “Do I Need a Car or a Friend? A Scenario-Based Investigation of
the Context Dependency of Preferences for Anthropomorphic Product Designs.”
IV. Paper 4 ................................................................................................... .p. 92 – 110
Miesler, Linda, Jan R. Landwehr, Andreas Herrmann, and Ann L. McGill (2010),
“Consumer and Product Face-to-Face: Antecedents and Consequences of
Spontaneous Face-Schema Activation,” Advances in Consumer Research, 37,
536-537.
V.
Paper 5 ................................................................................................... p.111 – 124
Miesler, Linda and Helmut Leder (2010), “The Cute Look: Baby-Schema Effects
in Product Design,” Proceedings of the 7th International Conference on Design &
Emotion, Chicago.
Paper 1: The Face of Anthropomorphism: The Effects of
Face-Like Product Design Features on Consumers’
Perceptions and Evaluations of Products
Linda Miesler(1), Jan R. Landwehr(2), Ann L. McGill(3), Andreas Herrmann(4)∗
Abstract
Regarding the product form’s impact on anthropomorphizing, we addressed two
questions: Do face-like product forms trigger mental processes which are similar to
human face-processing and, what is the effect of such anthropomorphic forms on
higher-order cognitive processes? Using pictures of cars as stimuli in three
experiments, findings reveal that car fronts that appear to have facial features
automatically trigger the human schema more than the car schema whereas car sides
do not. Findings also reveal that processing of car “faces” triggers processing similar
to human faces, specifically, greater reliance on holistic as opposed to analytical
processing. Holistic processing for cars with face-like designs also leads to less
reliance on negative information in forming product evaluations even when that
negative information is focal to the judgment. Our results provide important theoretical
implications for the concept of product anthropomorphism as well as practical ones for
the area of product design.
Introduction
Products sometimes give the appearance of having faces. Clocks and cars are common
examples of anthropomorphized objects (Aggarwal and McGill 2007; Labroo, Dhar,
and Schwarz 2008) but apart from clocks and cars many products may be seen to have
features akin to eyes, nose, or a mouth as illustrated in a recent advertising campaign
by American Express (see Figure 1) and the numerous images on websites devoted to
collecting examples of “faces in places” (e.g., http://facesinplaces.blogspot.com/).
While such examples of products with human features are common, the study of
anthropomorphized objects is not. Until now, consumer researchers have studied other
∗
(1) Linda Miesler, Doctoral Candidate, Center for Customer Insight, University of St. Gallen; (2) Jan R.
Landwehr, Assistant Professor of Marketing, Center for Customer Insight, University of St. Gallen; (3) Ann L.
McGill, Sears Roebuck Professor of General Management, Marketing and Behavioral Science, Booth School of
Business, University of Chicago; (4) Andreas Herrmann, Professor of Marketing, Center for Customer Insight,
University of St. Gallen
6
Paper 1: The Face of Anthropomorphism
aspects of anthropomorphism such as brand personality (Aaker 1997; Yoon et al.
2006) and the formation of relationships with brands (Aggarwal 2004; Fournier 1998)
but few studies have considered the effects of physical characteristics (i.e., of the mere
product form) of anthropomorphized products.
FIGURE 1
American Express Advertisement Showing a Product with Human Facial
Features
An exception is recent work by Aggarwal and McGill (2007) who explicitly
considered the influence of marketers’ efforts to present a product as possessing
human qualities. In this work, the authors take a “top-down” view of
anthropomorphism, in which the marketer proposes the metaphor between the object
and the human, for example, by having the product speak. The object then either
possesses or lacks physical features that make the metaphor easy or difficult to grasp,
respectively, for example, through facial features or body shape. Testing a schema
congruity model of evaluation (Mandler 1982; Meyers-Levy and Tybout 1989),
Aggarwal and McGill report more favorable evaluations when the product’s physical
features make it easy to see the product as human, that is, when the features are
congruent with the activated human schema.
The present research adds to the scarce literature on the nature of anthropomorphism
in consumer behavior and extends Aggarwal and McGill’s inquiry by isolating the
direct effects of a product’s physical appearance on consumer perceptions and
7
judgments. Therefore, unlike the Aggarwal and McGill studies in which the humanrelated mental concepts were activated beforehand to shape the interpretation of the
visual input, in the present research we consider spontaneous or “bottom up”
anthropomorphism. That is, we consider the effects of anthropomorphism that is
driven solely by perceptions of the physical features of the object without any further
prompting by the marketer or other outside entity, a form of anthropomorphism likely
to be relatively common. As discussed in greater detail below, we also postulate that
the bottom-up route to anthropomorphism has special characteristics and processes
that might not occur during top-down anthropomorphism, making investigation of this
type of anthropomorphism practically important and theoretically distinct.
In particular, in our research we focus on the effects of equipping an object with facial
features. The study of faces appeared a promising route to begin our inquiry because
designers explicitly attempt to imbue products with personality through facial features
(e.g., Welsh 2006). In addition, people appear to have an innate or very early acquired
preference for face stimuli (Mondloch et al. 1999) and, as noted at the outset, a
willingness, even fascination, in seeing faces in inanimate objects (Guthrie 1993). Our
study focuses on cars as a very important product category from a practical point of
view, which has also been addressed in prior theoretical work on anthropomorphism
(Aggarwal and McGill 2007, Windhager et al. 2008). Cars are an oft-discussed
example when talking about anthropomorphic forms in product design, because their
front ends share many characteristics with human faces.
In our inquiry, we address two research questions. We first explore the nature of
spontaneous anthropomorphism for objects having facial features. Although designers
and consumers may speak of a car as “having a face,” no research has clearly
examined whether people actually process those features as they would a human face
or if instead they see the product first as an object and then note that it has features like
those of people. The category “human” is well-learned and deeply familiar so makes
an easy analogy (Epley, Waytz, and Cacioppo 2007). The question remains whether
consumers actually see anthropomorphized products in human terms spontaneously,
responding to these products cognitively and emotionally as if they really were people,
or if instead they are just relying on a ready comparison to a familiar concept as in,
“Oh look, how cute, they made the car look like a tough guy.” Hence, in study 1, we
test whether processing face-like product designs might be based on the same
automatic processes that take place during human face perception. So, in study 1 we
aim at providing support for our basic idea that anthropomorphism can be
8
conceptualized as an automatic and perceptual-based process which can be triggered
by the mere product form. Our findings support the hypothesis that people perceive
objects with facial features as they would human faces, triggering the human schema
more than the relevant object schema.
Based on the findings of the first study, our second research question addresses
whether the perception of facial features in an object affects other higher-order
cognitive processes which shape subsequent product-related evaluations and
judgments. Prior work suggests that holistic perception is characteristic of the
encoding of faces (Farah et al. 1998; Hole, George, and Dunsmore 1999; Maurer, Le
Grand, and Mondloch 2002). Drawing on this research, we hypothesize that the
perception and processing of face-like product forms will lead consumers to process
information in a more holistic, configural fashion as distinct from a more linear,
analytical process. In study 2a, we test the basic effect by showing that consumers who
are exposed to facial features of products process subsequent stimuli more holistically
compared to controls. In study 2b, we show that this difference in processing for
consumers exposed to products with facial features will lead them to evaluate products
using both focal and peripheral attributes (as indication of holistic information
processing), muting effects of negative information. By investigating how consumers
process face-like products, we hope first to understand the impact of the mere product
form on anthropomorphizing, and, second, to gain deeper insights into the cognitive
processes underlying anthropomorphizing in general.
Processing of Faces – Processing of Face-Like Products
Our first research question centers on perceptions of products that have face-like
features. Even though many products can possess face-like patterns, we focus in
particular on cars because, as noted in the introduction, this particular class of products
has a highly pronounced anthropomorphic configuration and is an established research
object for studies on anthropomorphism (cf. Aggarwal and McGill 2007). Like faces,
they are a highly homogenous class of stimuli - the frontal view’s composition is
symmetrical, the car’s first-order features (the headlights, grille) resemble the facial
first-order features (the eyes, mouth), the features’ configuration is similar for faces
and cars (horizontally aligned headlights/ eyes, subjacent the grille/ mouth). In fact,
consumers are able to assign facial features to components of a car front without much
effort: they describe the headlights of cars as eye-like and the grille as the mouth (Erk
et al. 2002; Windhager et al. 2008). Similarly, as the eyes and mouth are the most
9
informative parts of the human face (Shepherd 1981, 105), the headlights and grille are
reported to be the most important features of the front end when identifying and
judging a car (Welsh 2006).
Hence, there seems to be a high morphological similarity between faces and some
product categories (i.e., cars) which is also explicitly perceived and verbalized by
consumers. But is it also valid to assume that the psychological mechanisms
underlying the processing of faces and face-like products are similar? Our assertion is
that indeed they are similar, specifically, that the anthropomorphic perception of facelike products is based on highly automatic processes that are similar to those
underlying face perception.
On its surface, this assertion of similarity of processing could appear to be at odds with
a recent investigation by Yoon and colleagues of the brand personality construct
(Yoon et al. 2006). These authors find that verbal-semantic judgments for brands (e.g.,
“Is this brand reliable?”) are processed differently than those for persons (“Is your
friend Peter reliable?”), suggesting that consumers’ understanding of brand personality
is something more akin to a metaphor than a true humanization of the brand. However,
our approach differs from Yoon et al.’s research in two important respects. First, they
investigated evaluations of abstract brands, which are mentally represented differently
than visual product designs that are the focus of our studies. Second, verbal-semantic
judgments are mainly based on the consumers’ conceptual knowledge (e.g., What does
“reliable” mean?), implying different underlying processes than the feature-based,
bottom-up route we describe here.
Additional research supports our assertion of similar psychological mechanisms for
processing anthropomorphic (face-like) product designs and human faces. For
example, one line of research attests to the high resilience of face recognition. In
general, it is known from face perception literature, that individuals are highly
sensitive to human faces. They discover faces fast and without cognitive effort
(Sergent 1989), and even very young individuals show a preference for face stimuli
over non-face stimuli (Mondloch et al. 1999; Valenza et al. 1996). Importantly, the
stored schemata of faces are remarkably resilient, that is, the face schema can
overcome other distorting influences and can even be detected in non-face but facelike configurations (e.g., man-in-the-moon illusion, face-like arrangements of fruit and
vegetables see Bruce and Young 1998). Recently, neuropsychological studies have
identified brain activation correlates which are supposed to be face-specific (N 170, M
100, activation of the fusiform face area). For example, the N 170, an event-related
10
potential that occurs 170 ms after stimulus onset, is about twice as large for faces as
for non-face stimuli (Bentin et al. 1996; Jeffreys 1996; Lui et al. 2000). What is so
impressive about the neuropsychological correlates is that they are not exclusively
elicited by real human faces but also by schematic faces comprising no more than
simple lines and dots to represent the mouth, nose, and eyes (e.g., Bentin et al. 2002;
Sagiv and Bentin 2001). Even with regard to overt behavior, presentation of a simple
three-dot sketch of a face (a “watching eyes” pattern) has been shown to serve as a
weak social cue able to elicit more altruistic behavior (e.g., a larger donation in a
dictator game) compared to a neutral dot pattern, an effect that appears to occur
outside conscious awareness (Rigdon et al. 2009). Prior research suggests therefore
that reactions to the face pattern, and apparently also the underlying psychological
processes, are very robust and flexible.
Both Guthrie (1993) and Gombrich (1973) provided an explanation for the high
resilience of face perception: They pointed to the evolutionary meaning of human
forms and stated that it is one of the most important needs of human beings to note the
presence of other human beings, hence any schema for detecting humans has priority
over other schemata (Guthrie 1993, 103). From this perspective, it is easy to
understand why consumers may automatically discover faces in inanimate objects
such as anthropomorphic products.
Further support for our similarity hypothesis comes from neuroscientific researchers
who have already provided preliminary evidence that face-like products might be
processed similarly to human faces. Especially, results from fMRI research were able
to show that cars also activate the fusiform face area (FFA), that is, the area originally
activated by human faces (Erk et al. 2002, but cf. Gauthier et al. 2000). However, the
evidence from neuropsychological studies like the one by Gauthier et al. (2000) or Erk
et al. (2002) cannot be unambiguously interpreted with regard to the effect of product
form on anthropomorphizing. First of all, the isolated impact of a product’s physical
appearance on face-like processing is not clear yet, since a range of object categories
(i.e., face-like and non face-like ones) were able to activate the FFA, a finding
Gauthier et al. explained due to the degree of familiarity with an object category.
Further, the picture stimuli employed in previous studies did not allow clear
conclusions about the direct effect of face-like forms on processing because the cars
were shown either in side or in diagonal front view, so one has to assume that in recent
studies, the face-pattern was not or was barely detectable by the participants. So, from
the existing neuropsychological studies it is still controversial whether the (face-like)
11
morphological configuration of the product per se led to the activation of the FFA or
whether other, more abstract factors accounted for the brain activation pattern found
(e.g., the level of familiarity with the category).
Consequentially, Erk et al. pointed out that from previous studies “it remains open
whether activation of the fusiform face area is due to the face-like appearance of the
cars or expertise in processing” (p. 2501). In a behavioral study, Windhager et al.
(2008) also addressed the question whether humans process car fronts like faces. They
gave indirect evidence for their assumption that car fronts are interpreted as faces so
that people draw the same inferences from car fronts as from faces (e.g., concerning
sex, maturity). However, their methodology (explicit ratings on Likert scales) did not
allow conclusions to be drawn on the underlying cognitive process.
Our research therefore builds, first, on face perception literature emphasizing the high
resilience of face detection and second, on a few preliminary empirical results
indicating that the perception of faces and of car fronts might be based on the same (or
similar) cognitive processes. Unlike previous studies, we try to isolate the effect of
mere product form on anthropomorphizing and test the basic idea that an object with
facial features may trigger a bottom-up process of recognition akin to human face
recognition in our first study:
H1:
According to a feature-triggered bottom-up process, products whose design
features imitate a human face are able to automatically activate a human faceschema.
Study 1: Do Car Fronts (But Not Car Sides) Activate a Human FaceSchema Automatically?
The first study is intended to test our basic assumptions that consumers can
anthropomorphize products spontaneously (i.e., without the pre-activation of a human
schema) and further, that face-like products are represented similarly to faces. To have
a experimental design precisely tailored to our research issue, we settled on a webbased lexical decision task (LDT) using pictures of cars, presented in front or side
view, or human faces as primes and words related to the concepts of face or car and
non-words, respectively, as targets. Based on this approach, we examined whether the
average response pattern in the face condition is similar to the response pattern in the
car front condition, but different from the response pattern in the car side condition. In
12
particular, we expected that participants who were primed with either car fronts or
human faces respond faster to face-words than to car-words, whereas participants who
were primed with car sides respond faster to car-words than to face-words. As we
compared car fronts with car sides (i.e., we used two different operationalizations of
the mental concept “car”), this experimental design also allowed us to exclude
concurring explanations for the activation of a human schema that charge the earlier
referenced neuroscientific work (cf. Gauthier et al. 2000; Erk et al. 2002). For
example, if it was the very familiar concept “car” in general which accounts for
product anthropomorphizing (e.g., as a result of marketing activities and many movies,
consumers might be highly familiar with associating cars with human beings) and not
the specific physical features, car sides should activate the human schema, too.
Design, Stimuli, and Procedure
The study was a 3 × 2 design with picture prime (car front vs. car side vs. human face)
manipulated between-participants, and word category (car related word vs. face related
word) manipulated within-participants. One-hundred sixty-five native German
speakers who were recruited by a professional market research company took part in
the online experiment (Mage = 36; SDage = 11; 56% male) and were paid an allowance
for participation. The participants’ task was to ignore the preceding picture prime and
to categorize the target stimulus, a letter string, as a word versus a non-word by
pressing a key. For all participants, words stemmed from two word categories. The
category “car-related words” consisted of nine German words which were associated
most frequently in a pretest (n = 82) with the concept car, the category “face-related
words” consisted of nine German words which were associated most frequently with
the concept human face in the same pretest.
Car-related words
wheels (Räder)
headlights (Lichter)
driver (Fahrer)
vehicle (Fahrzeug)
police (Polizei)
drive (Fahren)
steering wheel
car (Auto)
brakes (Bremsen)
Face-related
mouth (Mund)
forehead (Stirn)
head (Kopf)
face (Gesicht)
nose (Nase)
smile (Lächeln)
look (Miene)
eyes (Augen)
grin (Grinsen)
13
The two word lists did not differ with regard to mean word length and mean word
occurrence frequency. We also pretested with a separate sample of participants (n =
10) that the selected words could be unambiguously assigned to one of the two word
categories. Additionally, 18 nonwords were formed by changing at least two letters in
the car-related and face-related words, so that the non-words were still pronounceable,
but not similar in orthography to any existing word in the German language. With
regard to the preceding picture primes, participants were randomly assigned to one of
three priming conditions, so that they were primed either with black-and-white
pictures of cars shown in front view, cars shown in side view, or faces. The pictures
were displayed in the center of the monitor, fitted within a 265 by 265 pixel rectangle.
The employed car models all belonged to the compact car segment (e.g., BMW Mini,
Peugeot 207; see Figure 3) and were among the 20 most frequently sold compact cars
in Germany in 2007. The face group served as a reference group to facilitate the
subsequent interpretation of the resulting response patterns. The face pictures
comprised four male faces, and five female faces, all faces had a neutral expression
without any salient features such as glasses or a beard and were taken with kind
permission from the AR face database (Martinez and Benavente 1998). In total, every
participant responded to nine picture/car-word pairs, nine picture/face-word pairs, and
18 picture/non-word pairs with the order of presentation randomized across
participants. Before the test block, the participants completed a short practice block of
20 trials with neutral word and picture material to familiarize themselves with the
procedure. The time interval between stimulus onset and the response of the
participant (word vs. non-word by pressing the respective key) was recorded in
milliseconds for each trial. Overall, the experiment took about 12 minutes.
Reaction Time as Dependent Variable
The rationale behind the chosen methodical paradigm was as follows. We assumed
that cars shown in front view possess features like a human face, whereas cars shown
in side view apparently do not possess this face-like configuration. In accordance with
the spreading activation model (Anderson 1983; Collins and Loftus 1975; McNamara
1994) and several studies which have already proven that even pictures can serve as
effective primes in a LDT (Bajo and Canas 1989; Irvin and Lupker 1983; Kroll and
Potter 1984; Theios and Amrhein 1989; Vanderwart 1984), we expected that the
picture prime shown would activate spontaneously the appropriate concept in the
participants’ semantic network (e.g., “face”) and that the activation would spread
14
along to related concepts (e.g., “eyes”), thereby facilitating the subsequent processing
of these concepts.
In accordance with these considerations, we regard shorter latencies when responding
to face-words compared to responses to car-words as an indication of spontaneous
face-schema activation in the memory elicited by the prime picture. Conversely, we
regard shorter mean latencies for responses to car-words than to face-words as an
indication of spontaneous car schema activation. Hence, building on our first
hypothesis we expect that participants who were primed with either car fronts or
human faces should react similarly by responding faster to face-words than to carwords (or in the case of car fronts, they should at least be equally fast in responding to
face-words and to car-words), whereas participants who were primed with car sides
should respond faster to car-words than to face-words. This pattern would support the
notion that car fronts lead to spontaneous face-schema activation due to their featurebased similarity with human faces.
At first sight, it may not appear intuitive that individuals confronted with pictures of
car fronts should respond faster to face-related words than to car-related words
because one should assume that individuals still recognize the object as a car and not
as a face when asked to categorize the object. However, to make our assumptions
clearer, one has to distinguish between implicit and explicit categorization processes
(Evans 2008; Ito and Cacioppo 2000), that is, between the automatic precategorization of an object which can proceed implicitly versus the object’s explicit
categorization, for example, when one has to verbalize the category the object belongs
to. It is known from face perception literature that the pre-categorization of faces
against other objects occurs very early (e.g., Lui, Harris, and Kanwisher 2002; Seeck
and Grüsser 1992). So it is a very rapid process which uses rather coarse information
to discriminate faces from non-faces (Thorpe et al. 2001), therefore, the process should
be susceptible to false positive alarms, that is, individuals may detect a face mistakenly
where there is actually no face (such as in face-like products). With regard to the
cognitive task and task demands involved in a LDT, we assumed that the reaction
times in a LDT capture implicit and not explicit categorization processes (Neely 1991,
297) since participants are asked to ignore the prime pictures and, further, they are not
required to label the pictures (which would demand explicit categorization).
Nevertheless, we were interested in the implicit process of schema activation and,
since the face schema is a very salient and highly accessible schema with significant
biological meaning, it might also be activated by face-like objects and this activation
15
should occur with shorter latencies compared to the activation of the correct object
schema.
Results
Reaction Times for Face-related Words versus Car-related Words. Only correct
responses from word trials made within more than 250 ms but less than 1450 ms were
included in the data analysis (Ntrials = 2839). Incorrect classifications of the letter
strings were infrequent (5.4% error rate across all word trials), as were response
latency outliers (1.0%). The participants’ answers in a funnel debriefing (Bargh and
Chartrand 2000) carried out immediately after the LDT ensured that the participants
were not aware of the study’s goal, that is, they did not realize any relationship
between the prime pictures and the word targets when they performed the LDT.
We submitted the participants’ response latencies to a 3 (priming condition: car front
vs. car side vs. human face) × 2 (word category: face vs. car) between-within-subjects
ANOVA with repeated measures on the second factor. To exclude effects due to
specific aspects of the stimulus material, we used multiple operationalizations (9) of
the primes in each category (i.e., nine different car fronts; nine different car side
views; nine different human faces). To eliminate the unsystematic variance produced
by this methodological manipulation, we z-transformed response latencies within each
of the 27 individual stimuli prior to the subsequent analyses.
As expected, the prime category (car front vs. car side vs. human face) interacted
significantly with target word category (face vs. car). Participants who saw car fronts
or faces as primes in the LDT showed a different response latency pattern, compared
to participants who saw car sides as primes (F(2, 162) = 3.28, p = .04), while none of
the main effects were statistically significant (all p’s > .66) (Figure 2). More precisely,
participants who were primed with car fronts in the LDT responded significantly faster
to face-words than to car-words (t(52) = 1.69, p = .05), suggesting that the human
schema was more salient in processing these stimuli than the underlying product (car)
schema. Conversely, participants who were primed with car sides reacted significantly
faster to car-words than to face-words (t(46) = -1.87, p = .03). Comparing mean
response latency patterns in the two car conditions with the mean response latency
pattern in the face condition, latency patterns did not differ between front and face
condition (t(116) = 0.44 , p = .66) but both were significantly different from the side
condition (face vs. side: t(110) = -0.20, p = .04; front vs. side: t(98) = 2.52, p = .01)
which is in accordance with our hypothesis.
16
FIGURE 2
Results of the Lexical Decision Task with Car Words versus Face Words for
Three Priming Conditions
Note: Response latencies have been z-transformed. Smaller values indicate shorter response
latencies, that is, a stronger activation of the associated concept face or car, respectively.
Discussion
As an indication of automatic anthropomorphizing, we found that participants
responded faster to face-related words than to car-related words, but only when they
saw car fronts (and not car sides) as primes. Based on this result, we concluded that the
mere match between the physical features of a car front and a face schema resulted in
an automatic face-schema activation in accordance with a feature-triggered bottom-up
process. Therefore, we could also rule out that the concept car in general is associated
with a human schema. Participants in the car front condition also showed a similar
response latency pattern to participants in the face condition which might further
suggest similar processing of faces and car fronts since car fronts are assumed to
possess face-like feature configurations. Specifically, our findings showed that car
fronts not only brought to mind the human schema more than car sides but they also
brought to mind the human schema more than the actual product category schema.
This remarkable effect could be explained by the growing evidence from
neuropsychological studies indicating that face-selective responses have very short
latencies which makes the processing of faces special compared to the processing of
non-face objects (e.g., Farah et al. 1998; Seeck and Grüsser 1992).
17
Interestingly, our participants were not aware of the study’s goal (i.e., they did not see
any relationship between the prime pictures and the words employed), so the
participants’ responses in the LDT were not guided by expectations, which further
supported our assumption that anthropomorphizing can occur bottom-up and therefore,
automatically. This finding is in line with other authors’ theoretical assumptions (Bush
1990; Caporael 1986; Chartrand, Fitzsimons, and Fitzsimons 2008; Epley et al. 2007;
Ingram and Annable 2004), but to our knowledge we are the first to provide specific
empirical evidence to support this assumption for anthropomorphizing in product
design. In the following two studies, we searched for further evidence to support our
assumption that automatic anthropomorphism is based on cognitive processes which
are also involved in the processing of human forms (faces). Specifically, we were
interested in holistic styles of perception and information processing.
Holistic Versus Analytical Face and Product Perception
People’s outstanding performance in recognizing faces compared to other categories of
visual objects is attributed to the use of specific perceptual processing strategies that
involve the encoding of configural/holistic information that emerges rather from the
spatial relations among facial features than from the single facial features themselves
(e.g., Machhi Cassia et al. 2008). More than other visual objects, faces are supposed to
be represented and processed holistically rather than analytically. That means that
people may not represent faces in a piecemeal fashion in memory, that is, by its single
features (the eyes, mouth, nose), but they may more likely process them as a “gestalt”
(Farah et al. 1998; Hole et al. 1999; Maurer et al. 2002). Holistic processing of faces is
suggested by robust effects such as the inability of adults to recognize a single feature
(e.g., a nose) as familiar or unfamiliar when it is presented in the context of the whole
face compared to when it is presented as an isolated feature (the “part-whole
recognition effect”, Tanaka and Farah 1993; Tanaka and Sengco 1997). Another
demonstration of holistic face processing is the so-called “face composite effect”
which states that it is more difficult for individuals to recognize the top half of a face
aligned with the bottom half of a different face compared to a condition where the two
face halves are misaligned through a shift (Young, Hellawell, and Hay 1987; Konar,
Bennett, and Sekuler 2010). In the first condition, individuals perceive the two aligned
face halves as a new face by processing them holistically, in the second condition they
do not. Even if the facial features are replaced by non-face objects (e.g., eyes replaced
by two flowers) leaving the facial configuration unchanged, people still categorize
18
such “object-faces” as face (Donnelly et al. 1994). Rossion, Schiltz, and Jacquier
(2008) could validate the evidence on holistic perception found in behavioral studies
by using functional magnetic resonance imaging.
As these face-specific processing characteristics have thus far not been examined for
anthropomorphic products, we set out to extend and strengthen the evidence on facespecific processing of anthropomorphic products found in study 1 by asking if the
special characteristics of face perception (e.g., holistic object representation) can also
be found in the processing of face-like car designs. Our rationale for this investigation
is that holistic representation can be taken as supportive evidence that
anthropomorphism is an automatic bottom-up process triggered by the product’s
design features. This approach also allows us to distinguish the impact of bottom-up
anthropomorphism from anthropomorphism triggered by top-down information (e.g.,
presenting the picture of a car together with a “speech balloon” to suggest that the car
is communicating directly with the consumer). Our expectation is that top-down
anthropomorphism should lead to the deliberate construction of and search for
similarities between the object and the human form, without a similarity of the
underlying process mechanisms. As evidence for this assumption, thought protocols in
Aggarwal and McGill’s (2007) studies, which involved a more top-down process,
revealed that participants were able to verbalize the association between product forms
and human forms (introduced by phrases such as “they look like…”, p. 475). These
findings support our assumption that top-down processes, as distinct from the
automatic, bottom-up processes we examine, trigger rather the deliberate search for
similarities in surface features (comparable to analogical reasoning, e.g., the eyes of
the face are analogous to the headlights of a car) than similarities in underlying
processes (e.g., face-specific encoding). Second, holistic modes of product
representation should crucially affect product judgment formation processes and
product evaluations, so we further intend to demonstrate the effects of face-like forms
on marketing-relevant variables.
Relationship between Consumer Perception and Consumer Judgment
and Decision Making
Baumgartner (1993) emphasized that the question whether people perceive and encode
product information in a holistic or analytic fashion is an important topic to
understand, since perceptual processes form the bases for higher-order cognitive
processes like judgment and decision making. For example, a consumer who encodes
19
the product and its attributes holistically should also come up with judgments about
the product which result from rather holistic post-perceptual information processing
strategies. If our assumption is correct that face-like product designs are perceived and
represented similar to faces, that is, in a holistic manner, one would expect that due to
this processing style consumers form evaluative judgments about face-like products
differently than for other products.
Different lines of consumer research have dealt with (post-perceptual) holistic versus
analytic information processing (e.g., with regard to product categorization, attitude
formation, and judgment and decision making) (Alba and Hutchinson 1987; Holbrook
and Moore 1981; Hutchinson and Alba 1991; Monga and Roedder John 2007). These
different lines of research mostly agree with regard to the global features which
distinguish between both types of processing. In the following section, we summarize
the global features of holistic versus analytic processing which are also relevant for
our theoretical assumptions when dealing with the formation of product judgments.
In general, when processing information holistically consumers are supposed to
consider “the big picture” in a product choice situation rather than focusing on
individual product attributes. For example, individuals who process product
information holistically when they categorize a product are said to assess the product’s
overall similarity with an existing category on the basis of all the product attributes
available or attribute configurations, regardless of the attributes’ importance or
relevance for the task (Alba and Hutchinson 1987, 419). In general, the product
attributes at hand can be subdivided into focal and peripheral product attributes, with
regard to how relevant or diagnostic they are for making a judgment or decision (cf.,
Hutchinson and Alba 1991, 328). For example, one could imagine that when a
consumer has to judge the quality of a product, technical-functional product attributes
(e.g., workmanship, efficiency) are used more diagnostically to make a quality
judgment than the product’s visual design (Zeithaml 1988). Hence, according to our
differentiation, in such a situation technical product attributes might be a focal
criterion for making the judgment, whereas the product design is a rather peripheral
attribute. Individuals who process information holistically are more likely than
analytical processors to consider peripheral factors besides focal product attributes
(Monga and Roedder John 2008; Nisbett et al. 2001).
On the other hand, analytic information processing is more focused on the relevant
(i.e., diagnostic) product information accessible. Alba and Hutchinson (1987, 419)
stated that people who categorize products based on an analytical strategy only focus
20
on one or two critical attributes or attribute configurations, that is, in a judgment
situation where a product is defined by multiple attributes, the analytic information
processing style is best formalized by attributes with different importance weights. As
irrelevant attributes are assumed to have no effect, individuals processing product
information analytically do not consider peripheral information at all, or at least to a
much lesser extent than individuals who apply a holistic processing strategy (Monga
and Roedder John 2008).
Generally, it is known that individuals give more weighting to negative than positive
information in forming judgments (Eagly and Chaiken 1993). However, holistic
information processing is assumed to influence how people deal with negative product
information. Monga and Roedder John (2008) found evidence that holistic thinkers are
less susceptible to negative brand information than analytic thinkers because they also
take contextual (peripheral) information into account and do not exclusively focus on
focal product attributes. Other authors also assume that, under certain circumstances, a
holistic style of processing creates the potential for biased judgments (e.g., when
categorizing products) because diagnostic information may not be given adequate
attention, or irrelevant information may be given too much weighting (Hutchinson and
Alba 1991; Smith and Kemler Nelson 1984). However, when one sees this apparent
bias from a different angle, a positive aspect of such processing bias might be that
negative information does not get so much weighting compared to an analytic
processing approach where people analyze the product information more thoroughly.
So Burroughs (1996, 466) pointed out that applying an analytical and more reflexive
strategy in a purchase decision may let consumers recognize negative information
more easily. These ideas are formalized in the following hypotheses and tested in
studies 2a and 2b:
H2:
Similar to face perception, consumer perception of face-like products is better
described by holistic than analytic perceptual processes. (Study 2a)
Based on the assumption of differing perceptual processes depending on product shape
in the second hypotheses, we postulate further effects on subsequent cognitive
processes in our third hypothesis:
H3:
21
Holistic product perception influences judgment formation processes about
face-like products. When dealing with face-like product designs, negative
product information has a weaker impact on quality judgments compared to
non-face-like designs. (Study 2b)
As explanation for H3, we hypothesize that:
H4:
Individuals applying a holistic information-processing strategy weight focal
versus peripheral product information differently compared to individuals who
apply an analytic strategy. (Study 2b)
Study 2a: Are Face-Like Products Perceived Holistically Rather than
Analytically?
Study 2a is meant to examine whether the proportion of people who applied a holistic
strategy of perception differed between two experimentally manipulated product
design conditions. One group of participants was confronted with pictures of neutral
car designs, the other group was confronted with pictures of anthropomorphic (i.e.,
face-like) car designs.
Stimuli
The stimulus pictures used in studies 2a and 2b were modifications of the nine realistic
car models employed in study 1. In study 1, we did not systematically vary the
morphological similarity of the product design with a human face but instead varied
the angle the product was shown from (front or side). In studies 2a and 2b, by contrast,
we manipulated the product design more directly. We focused only on the cars’ front
ends and systematically varied the shape of the design features to create more or less
face-like products. More precisely, a computer graphics expert modified the car fronts’
grille and headlights in each of the nine models to increase or decrease the similarity
with a human face and for every car model we created a face-like and a neutral version
(see Figure 3). In the face-like version of each car model, the car’s headlights were
replaced by headlights which resembled human eyes and the grille had a typical
mouth-like form. In the neutral version of each car model, the local front features were
modified such that they had a square form and looked technical and without any
association with human form. To ensure that these modifications actually affected how
22
anthropomorphic people perceive the cars to be, we conducted a pretest using a student
sample of n = 256. We checked the perceived face-similarity separately for the neutral
and the face-like versions, showing the participants pictures either of the nine face-like
car models or the nine neutral car models and asking them how much each car
resembles a human face. From the nine initial car models, we chose as stimuli for our
subsequent studies the four pairs (neutral vs. face-like) which were significantly
different from each other with regard to their perceived anthropomorphic look.
FIGURE 3
Examples of the Realistic Car Models Used in Study 1 and the Four Manipulated
(Face-Like and Neutral) Car Models Used as Stimulus Material in Studies 2a and
2b
In addition, we conducted a second pretest with the four selected car models using a
student sample of n = 40. As it is known that individuals in a positive mood also tend
to process visual information holistically rather than analytically (Gasper and Clore
2002), it was important to ensure that differences in holistic versus analytic processing
were not driven by mood differences due to the exposure to face-like versus neutral
cars. Thus, participants responded to a short four-item, bipolar mood scale (badtempered – in a good temper; sad – happy; uncomfortable – comfortable; tense –
relaxed) as a control measure to rule out any mood-based alternative explanation.
According to a pre-post design, participants responded to the four items, then they
looked at the four models (either neutral or face-like condition) - each presented on a
separate page in a small booklet – and finally, they responded to the four items again
(this time, presented in inverse order to reduce memory effects). The mean of the four
mood items was computed separately for the pre- and the post-measurements
(Cronbach’s αpre = .81; Cronbach’s αpost = .84). Conducting a repeated measurement
23
ANOVA with average mood measurement as the dependent measure, point of time
(pre- vs. post measurement) as a within-participants factor and design (neutral vs. facelike) as between-participants factor, we found no differences between measurement
points (F(1, 38) = 1.20, p = .28), the design conditions (F(1, 38) = 1.02, p = .32), and,
most importantly, no interaction between measurement points and design (F(1, 38) =
0.08, p = .79). Thus, as expected, the design condition did not affect the participants’
mood which means that face-like designs do not induce a general positive mood in the
participants.
Design and Procedure
The study was a 2 × 4 design with car design (neutral vs. face-like) and car model
(model A, B, C, D) as the between-participants conditions. The objective of using
different car models was to ensure that differences between car design conditions were
not due to characteristics of a specific car brand. Sixty-six people (Mage = 39; SDage =
11; 38% male), who were recruited by a professional market research company,
participated in a short online experiment and were paid an allowance for participation.
The study consisted of an exposition and a test phase. In the exposition phase, the
participants in one design condition were exposed to the picture of a face-like car
front, whereas participants in the other design condition were exposed to the picture of
a neutral car front (in both conditions, participants saw only one out of four car models
and which one they saw was randomized). Overall, the participants were exposed to
the picture 15 times. The objective behind the exposition phase was that repeatedly
looking at the car front should activate a holistic versus analytic style of perception,
which should influence the participants’ style of perception in the test phase of the
experiment (for other instances of such cross-task processing manipulation cf. Bentin
et al. 2002; Lawson 2007; Macrae and Lewis 2002; Perfect 2003; see also MeyersLevy 1989). We expected the face-like car fronts to activate a rather holistic style of
perception, whereas the neutral car fronts to activate a more analytic style. As we
predicted quantitative differences in the proportion of people showing holistic
perception strategies versus analytic perception strategies, we expected a greater
proportion of people to apply a holistic strategy of perception in the face-like design
group compared to the proportion of people applying a holistic strategy of perception
in the neutral design group.
To assess the style of perception, in the test phase the participants had to compare
groups of geometrical figures (taken from Kimchi and Palmer 1982, 526). The
24
ensemble of figures contained a standard figure on the top and two comparison figures
side by side below the standard. The figures were made up of either small triangles or
squares with the local elements arranged such that the global configuration of the
figure was either a square or a triangle (see Figure 4). The participants had to indicate
which of the two figures was most similar to the standard figure. In one of the two
comparison figures, different elements were arranged in the same configuration as the
standard figure, in the other comparison figure, the same elements as in the standard
figure were used but were arranged in a different configuration. As one pair of figures
was similar in global form and the other pair was similar in the local elements, the
participants’ responses allowed inferences to be drawn whether the participant applied
a holistic versus analytic perception strategy when comparing the figures with the
standard figure. If participants chose the figures similar to the global form, they might
have applied a holistic strategy of perception and if they chose the figures similar in
local elements, they might have applied an analytic strategy of perception. Every
participant made three subsequent comparisons, which were presented in random
order. The whole experiment took approximately five minutes.
FIGURE 4
Groups of Geometrical Figures Used for the Test of Dominant Processing Style
(Holistic vs. Analytical) in Study 2a
Results
Analysis of the four car models showed no significant effect of the car model itself on
the perception strategies applied in the test phase, so we collapsed the data and
considered car design (neutral vs. face-like) as the only experimental condition in our
analysis.
Many participants (47%) did not appear to employ a single perception style across
trials. Therefore, prior to the data analysis we assigned the strategy to each participant
25
which the participant showed most frequently across the test trials, so that we included
one data point per participant into our analysis (binary dependent variable, 1 = holistic
perceptual strategy, 0 = analytic perceptual strategy). (As a secondary check on this
approach to the data analysis, we also analyzed the data by including only the response
style exhibited by participants’ on the first test trial after the exposition phase. This
analysis produced the same pattern of significant result as reported below.)
To test our hypothesis that face-like product designs increase the likelihood of holistic
product perception strategies in comparison to neutral product designs, we conducted a
chi-square test on the table of frequencies resulting from crossing the two levels of the
independent variable (i.e., face-like versus neutral design) with the binary coded
dependent variable (i.e., analytic versus holistic style of perception). More precisely,
we tested the one-sided hypothesis that the proportion of participants who applied a
holistic style of perception relative to the proportion of participants who applied an
analytic strategy is larger in the face-like design condition than in the neutral design
condition. Considering the proportions of participants using a holistic perceptual
strategy in the two design conditions (neutral vs. face-like), a chi square test revealed
that the strategy chosen was not independent of the product’s design in the exposition
phase (χ2(1) = 3.24, p = .04, one-sided) (see Figure 5). Participants in the face-like
design condition chose figures which were globally similar more frequently than
locally similar ones (67.7% vs. 32.3%, p = .04). On the other hand, participants in the
neutral design condition were as likely to choose figures which were similar with
regard to their local features as to choose figures which were similar with regard to
their global configurations, although they showed a slight tendency to choose locally
similar figures (45.7% vs. 54.3%, p = .37).
26
FIGURE 5
Percentage of Participants Applying a Holistic versus analytic Perception
Strategy for the Two Design Conditions in Study 2a
Discussion
The results of study 2a supported our assumption that face-like products are perceived
holistically rather than analytically, that is consumers may represent anthropomorphic
products as global “gestalts” rather than individual features. This empirical evidence
corroborates our theory that the automatic processing of face-like products is based on
the same processes as face processing, since many authors suggest that faces are
perceived and represented holistically. The result that anthropomorphic products
triggered holistic perception in our participants is not obvious, since objects such as
products generally are supposed to be perceived rather analytical, that is feature by
feature (Biederman 1987). Consequently, one could conclude that it might be possible
to change the consumers’ perception style from an analytic to a holistic mode when
presenting anthropomorphic (face-like) designed products to consumers. This
possibility to alter the consumers’ perception style is in particular relevant since we
assumed that anthropomorphic product designs do not only have an impact on how
products are perceived by consumers but also may even influence processes that are
involved in consumer judgment formation, that is, higher-order processes. We dealt
with this assumption in our last study.
27
Study 2b: Do Face-Like Product Designs Affect Consumers’ Style of
Product-Information Processing?
In study 2a, we found evidence that face-like product design may influence how a
product is perceived by consumers. The finding that face-like products might be
perceived holistically rather than analytically supports our global assumption that
anthropomorphizing face-like designs is based on similar processes involved in face
perception. As Baumgartner (1993) noted, the question whether products are perceived
by consumers holistically or analytically is highly relevant since perceptual processes
form the bases for higher-order cognitive processes such as consumer judgment and
decision making. A consumer who perceives the product as a global gestalt might
apply the product information at hand differently than a consumer perceiving the
product by its local features when it comes to making a judgment about the product.
So we investigated how face-like products might affect higher-order cognitions such
as the formation of product-related judgments. As quality judgments are, among
others, an important determinant of consumer purchase behavior, we focused on the
impact of negative product information on quality judgments. We chose the quality
construct to investigate higher-order processes because it requires the consumer to
form an overall judgment about the product’s excellence by integrating different
attributes which give information on the product’s technical/functional performance
(Zeithaml 1988). In contrast to general preference or liking ratings which are strongly
affected by non-functional factors such as product design, quality judgments should
not be affected directly by the design of a product but following our line of
argumentation might be moderated by the design of a product. In study 2b, we
expected participants who were exposed to face-like product designs to process the
accessible product information rather holistically, that is, considering both focal and
peripheral product attributes. The degree of negative product information should have
a weaker impact on the overall quality judgments of participants in the face-like design
condition because participants may pay too much attention to peripheral attributes and
less to the focal attributes. Participants in the neutral design condition may only
consider the functional attributes because they are most diagnostic to form a quality
judgment.
The study was a 2 × 3 × 4 design with car design (face-like vs. neutral), degree of
negative product information (low vs. medium vs. high), and car model (model A, B,
28
C, D) as the between-participants conditions. The pictures of the employed car models
were the same as in study 2a, thus, again multiple operationalizations were used. One
hundred thirty nine participants (Mage = 37; SDage = 11; 49% male) were recruited by a
professional market research company to take part in the web based study and were
paid an allowance for participation.
At the beginning of the experiment, the participants were told that they would be
shown a recently launched car model and that they should evaluate the new car based
on the information provided. To this end, each participant was exposed to a picture
depicting the car front which was accompanied by three product ratings. The product
ratings consisted of values on a three-point scale which indicated by the number of
stars how highly the car was rated by automobile experts on three functional attributes:
efficiency, functionality, and workmanship. To manipulate the degree of negative
product information, each attribute had either a value of three points (depicted as three
stars) or a value of one point (depicted as one star), three points meant that the car
scored high on the attribute, one point meant that the car scored only sufficiently on
the attribute. The three attributes were combined in such a way that either all of the
three attributes had three stars (factor level: low degree of negative information), one
of the three attributes had only one star (factor level: medium degree of negative
information), or two of the three attributes had only one star (factor level: high degree
of negative information). We decided to create no attribute combination where all
three attributes’ values were low because it seemed unrealistic that a car would be
launched which scored badly on all functional attributes. On the medium and high
degree of negative information levels, it was fully randomized which of the attributes
had a low (i.e., one star) or a high (i.e., three stars) value because we were not
interested in the impact of a specific functional attribute on perceived product quality.
So we had three different attribute sets on the medium and the high degree of negative
information level. The functional attributes were chosen based on the results of a
survey (n = 50) where people should report the attributes they consider most relevant
when choosing to purchase a car. These three attributes were reported as being highly
and approximately equally important and the variation in these three attributes seems
to be feasible and realistic, that is, low attribute values on the attributes chosen may
not make the car totally inacceptable for a consumer (e.g., in contrast to a knock-out
criterion like safety because a car which scores low on safety should be immediately
excluded no matter how high the car scores on other functional attributes).
29
To specify our hypotheses formulated above (H3 and H4), we expected participants
who were exposed to face-like product designs to process the provided information
rather holistically, that is, considering both focal product attributes (the functional
attributes) and peripheral attributes (the product design). The degree of negative
product information should in this condition therefore unfold a weaker impact on the
overall quality judgments because participants are expected to pay too much attention
to peripheral attributes and less attention to the focal attributes, so the negative effect
of the focal attributes should be attenuated. Participants in the neutral design condition
are however expected to only consider the functional attributes because they are most
diagnostic in order to form a reasonable quality judgment.
Dependent and Independent Variables. We asked the participants to judge the overall
quality of the newly launched car considering all the information available on a sevenpoint Likert scale with 1 = very low quality and 7 = very high quality.
In addition to the main independent variables - the design of the car and the degree of
negative information - we controlled for the participant’s ownership status (car owner
vs. non-car owner), because we assumed that participants who owned a car may
process the accessible information differently in comparison to participants who do not
possess a car, since the knowledge structures of individuals owning a car versus
individuals not owning a car should be different due to different degrees of familiarity
with the product (e.g., Cowley and Mitchell 2003; Johnson and Russo 1984).
Results
We did not find any effects of specific car model or of the random variation within the
attribute sets, so we collapsed the data and considered the design of the car and the
degree of negative information as the only factors in our analyses.
We conducted an analysis of variance with the factors - design of the car (neutral vs.
face-like), degree of negative product information (low vs. medium vs. high), and the
control variable ownership status (car owner vs. no car owner) - to check for the
expected interaction of the design of the car with the degree of negative product
information.
Quality Judgments. A significant Levene’s test, which tested the null hypothesis that
the error variance of the dependent variable quality judgment was equal across groups,
indicated that the model assumption of equal variances was not fulfilled (F(11, 127) =
3.11, p = .001). So prior to running the ANOVA, we logarithmized the quality
judgment data to meet the assumption and then we repeated the Levene’s test to
30
confirm that variance was homogenous following transformation (cf., Muller and
Fetterman 2003). Testing the three-factor model with logarithmized values of quality
judgment as dependent variable, we found the expected significant two-way
interaction of car design and degree of negative information (F(2, 127) = 3.32, p =
.04). The main effects of car design (F(1, 127) = 7.26, p = .01) and degree of negative
information (F(2, 127) = 14.71, p < .001), respectively, were also significant. Overall,
the quality judgments were not different between participants who owned a car versus
participants who did not own a car (F(1, 127) = 0.38, p = .54). The two-way
interaction of car design and ownership status was not statistically significant (F(1,
127) = 1.58, p = .21), the same held for the three-way interaction (F(2, 127) = 1.23, p
= .30). However, the two-way interaction of degree of negative information and
ownership status was statistically significant (F(2, 127) = 3.23, p = .04) which was not
surprising since we expected that individuals who owned a car differed in their
information processing from individuals who did not own a car when forming a
product judgment.
Group Comparisons. To allow a better understanding of the theoretically predicted
two-way interaction of car design and degree of negative information, we evaluated
the effect of degree of negative information on quality judgment separately for the two
design conditions. Planned contrasts revealed that face-like designs compared to
neutral designs had a compensating effect on the quality judgments: quality judgments
for cars with a neutral design decreased with an increasing level of negative product
information (Mlow = 1.43 (SD = 0.38); Mmedium = 1.22 (SD = 0.34); Mhigh = 0.89 (SD =
0.44); F(2, 66) = 10.50, p < .001; all contrasts were statistically significant with at
least p < .05). The quality judgments for face-like designed cars also decreased with an
increasing degree of negative product information, but to a much lesser extent (Mlow =
1.45 (SD = 0.28); Mmedium = 1.34 (SD = 0.30); Mhigh = 1.22 (SD = 0.30); F(2, 68) =
3.02, p = .06; only the difference between level 1 and level 3 was statistically
significant with p < .05; see Figure 6). The difference in the decrease of quality ratings
could not be explained by higher basic quality ratings for the neutral design condition
(cf., “law of initial values”; e.g., Wilder 1957), since in the low degree of negative
information group where actually all the attributes had a high value, both quality
ratings for the neutral and the face-like design conditions did not differ significantly
(Mneutral = 1.43, SD = 0.38 vs. Mface-like = 1.45, SD = 0.28; t(48) = -.23, p = .82). With
the increasing degree of negative information, the difference between the quality
31
ratings for neutral and face-like designs also increased (medium degree: ∆ = |.12|, p =
.21; high degree: ∆ = |.33|, p = .01).
FIGURE 6
Mean Quality Ratings Depending on the Three Degree of Negative Information
and Product Shape in Study 2b
Impact of Negative Product Information and Car Design on Quality Judgments. To
explain the interaction found between the factors car design and degree of negative
information, we ran a regression analysis to test whether participants in the face-like
design condition weighted focal versus peripheral information differently from
participants in the neutral design condition. Separately for both design conditions
(neutral vs. face-like), we regressed the quality judgments on the focal and peripheral
product information given. In addition to the values of the functional attributes (i.e.,
the focal product information), we included as a predictor in the regression model how
much the participants liked the design of the car (i.e., the peripheral product
information). Liking the design served as a proxy for the effect of product design on
quality judgments. Furthermore, we also included the interactions of design-liking
with all three functional attributes in the regression model (i.e., three interaction terms)
as it might be possible from a theoretical point of view that design-liking has a
different effect on quality, conditional on the value of the functional product
information (e.g., consumers might only consider the product’s visual design when the
product scores high on the functional attributes, cf. Chitturi, Raghunathan, and
32
Mahajan 2007). Based on the regression model, we intended to find further evidence
for our hypothesis that participants in the neutral design condition applied the
information more analytically, that is, integrating only the focal, highly diagnostic
product information (i.e., the functional attributes) into their overall judgments about
the car’s quality without considering peripheral information (i.e., the car’s design),
whereas we expected the participants in the face-like design condition to show a more
global information processing picture where quality should be predicted by all the
information available, that is local and peripheral information.
In order to test these assumptions, we built a linear regression model with the three
functional attributes (dummy coded: 0 for low attribute value/ one star = “negative
product information”, 1 for high attribute value/ three stars = “positive product
information”), the mean-centered values of design-liking and the three interaction
terms as predictors. With this model we found using directed t-tests that, as expected,
in the face-like design condition the coefficients of all three functional attributes and
the design-liking had a (at least marginally) significant effect on the quality ratings
(see table 1). Hence, as predicted, both focal as well as peripheral cues are used when
the design triggers a holistic perception by its anthropomorphic shape. In contrast, in
the neutral design condition only the functional attributes significantly predicted the
quality ratings (all p’s < .05), but the design-liking did not (B = -0.11, p = .36), so
design-liking was not a statistically relevant predictor of quality when the technical
attributes had low values (i.e., dummy coded as zero in the regression model).
However, two marginally significant interaction terms in the neutral design condition
indicated that design might have an amplified influence on quality judgments when the
technical values were high (dummy coded as 1 in the regression model).
33
TABLE 1
Regression Coefficients, Standard Errors, and P-Values for Face-Like versus
Neutral Design Condition
Variable
B
SE B
β
p
Face-like design (n = 71)
Attribute 1
Attribute 2
Attribute 3
Design liking
Design liking×attribute 1
0.45
0.45
0.57
0.35
-0.13
-0.02
-0.06
0.31
0.28
0.28
0.24
0.16
0.18
0.15
0.17
0.19
0.23
0.56
-0.17
-0.03
-0.07
0.08
0.06
0.02
0.07
0.21
0.46
0.35
Neutral design (n = 68)
Attribute 1
Attribute 2
Attribute 3
Design liking
0.86
1.02
0.77
-0.11
-0.19
0.41
0.32
0.33
0.36
0.31
0.32
0.21
0.28
0.23
0.27
0.29
0.25
-0.12
-0.18
0.42
0.32
0.01
0.00
0.01
0.36
0.19
0.07
0.08
Note: All three technical attributes are dummy coded with 0 = “negative product information” and
1 = “positive product information”, design liking is continuous and mean-centered. P-values are one-tailed.
Discussion
As expected, in study 2b we found that anthropomorphic product designs attenuated
the impact of negative product information on quality judgments. Quality judgments
for cars with a face-like design decreased with an increasing level of negative product
information to a much lesser extent than quality judgments for non face-like designed
cars. As an explanation for this compensating effect, we found tentative evidence that
individuals confronted with face-like product designs processed product information
more holistically, that is, they considered all the information accessible, relevant and
irrelevant, when making their judgments. So it might be possible that in a holistic
processing mode, the impact of negative product information on judgment formation
can be compensated by other less focal factors such as product design. In contrast, in
the neutral design condition we found indeed that individuals confronted with neutral
34
product designs processed the product information rather analytically. That is, they did
not take irrelevant information into account when the product scored low on the
available diagnostic attributes. However, we also found two marginally significant
interactions which could indicate that under certain circumstances (e.g., when a
product scores high on the diagnostic product attributes), individuals in a neutral
processing mode might also take peripheral information into account when forming
product judgments. One explanation for the interaction effects found in the neutral
design condition could be the hedonic dominance principle which entails that
consumers consider hedonic product attributes only when the product’s functional
attributes exceed a sufficient level (Chitturi et al. 2007). However, further research is
needed to disentangle this issue.
General Discussion and Conclusion
Marketers have long presented brands and products (intuitively or deliberately) in
anthropomorphic terms to make these products more appealing to consumers.
However, anthropomorphism is a relatively new topic in the consumer behavior
literature for which the outline of a theoretical framework is just beginning to emerge.
Hence, our research pursued the goal to contribute to a better understanding of the
cognitive processes underlying product anthropomorphism. In this research, we
suggest that there might be two routes for the processing of anthropomorphic
information, a “top down” route which has been already addressed by other authors
(Aggarwal and McGill 2007) and a “bottom up” route which was the focus of the
research presented here. By trying to understand the immediate effect of a product’s
physical appearance on anthropomorphizing we focused specifically on product forms
which resemble human faces (i.e., the front end of a car). In our first study, we
provided evidence that consumers might anthropomorphize products spontaneously,
merely due to the product’s design features. Thus, we found in a LDT that car fronts
which arguably resemble a human face activated automatically a face schema in our
participants' minds, whereas car sides which obviously do not resemble humans forms
did not. The result of our first study provided support for anthropomorphism as an
automatic process which can be triggered merely by a product’s physical form, an
issue which has not been addressed empirically before. Further, we found that faces
and car fronts elicited similar responses in our participants, therefore, our results show
that physical features might account for face-specific processing. So we were able to
clarify the evidence from other studies which also suggested that cars are processed
35
similar to faces but which could not disentangle familiarity effects from product form
effects (Erk et al. 2002; Windhager et al. 2008).
We extended this theoretical assumption and expected that the similarity in processing
between face-like forms and human faces might be also indicated by other process
characteristics which are actually typical for the encoding of human faces.
Accordingly, in our second study, we found evidence that face-like products compared
to non face-like ones were encoded and represented rather holistically than
analytically, a perceptual strategy which is reported to be unique for face perception
(Farah et al. 1998; Maurer et al. 2002). Since objects are supposed to be perceived and
encoded piecemeal (Biederman 1987), the result that we could induce holistic
perception strategies with the help of anthropomorphic car designs is remarkable. In
our third study, we tapped on the relevance of such holistic perception strategies for
consumer judgment and decision making. Hence, the results of the third study revealed
that face-like product designs which should be perceived as “gestalts” also triggered
holistic higher-order processes in the participants, indicated by the way our
participants integrated focal and peripheral product information into a global product
judgment. To our knowledge, this is the first demonstration that the mere product form
can shape higher-order cognitive processes (cf., Baumgartner 1993). This result
emphasizes that anthropomorphizing does not only affect how a product is perceived
by consumers, anthropomorphizing might also affect the information processing
strategies which account for consumer judgments and decisions.
So, to sum up, our research makes two important contributions to existing
anthropomorphism literature: First, we demonstrated the effects of physical product
features (i.e., product design) on anthropomorphism, since our results reinforce the
assumption that consumers engage in an automatic bottom-up process which is similar
to face-processing while anthropomorphizing face-like products; we were able to close
the research gap in literature concerning this issue and to clarify ambiguous results
from existing studies, respectively (e.g., Gauthier et al. 2000). Second, we identified
implications of the postulated automatic process. Anthropomorphic products might
alter the consumers’ processing style from analytic to holistic which affects both
perceptual and higher-order processes such as judgment formation. We do not expect
such effects to occur when anthropomorphizing happens top-down or concept-driven
(as opposed to bottom-up or feature-driven).
Overall, this research contributes to a better understanding of the processes which
underlie automatic product anthropomorphism, however, future research should clarify
36
some conceptual issues and questions our research did not address, but which might be
relevant for a deeper understanding of how consumers come to see products in human
terms. We adopted two psychological concepts to frame our theoretical assumptions
(i.e., bottom-up versus top-down, holistic versus analytic). In our research, we
exclusively focused on anthropomorphizing as a bottom-up process. Consequently, it
is important to point out that we did not claim that our results account for a deeper
understanding of anthropomorphism as a top-down process. Central to our studies was
the assumption that holistic encoding and processing result from face-specific bottomup processes which should occur when people anthropomorphize products
spontaneously. The prevalence of highly automatic, face-specific processes was
suggested by existing research pointing out the high resilience of face detection and
the underlying processes (e.g., Bentin et al. 2002; Guthrie 1993; Sagiv and Bentin
2001). Whether top-down processes might affect the processing strategies involved in
anthropomorphizing in the same way, was not the question we dealt with in the
research presented here. Nevertheless, we postulated that different kinds of processes
could determine bottom-up versus top-down anthropomorphizing (Wolfe et al. 2003).
Specifically, we did not expect top-down processing to trigger automatic face-specific
processes, since externally triggered anthropomorphism (as described by Aggarwal
and McGill 2007) should be based on different processes, for example more deliberate
and controlled cognitive processes such as analogical reasoning. Such rather deliberate
processes should be also involved when consumers have to assign human-like verbal
descriptions to brands – a mental process which is different from verbal judgments
about persons, as Yoon et al. (2006) have already shown. So future research could
contrast the two “routes” of anthropomorphism (bottom up versus top down) more
directly. With regard to the area of product design, our research showed that a
product’s visual features can be a very easy and powerful tool to endow a product with
a human-like personality. In future research, other product categories and other
categories of human forms beside faces could be used as stimulus materials to verify
the generalizability of the effects we found here. Further, it might be interesting to
investigate which product design features account most strongly for face-specific
processing, for example, how much do specific local features such as the car’s
headlights or grille contribute to bottom-up anthropomorphism, and what is the
contribution of the whole feature configuration? We found that face-like products are
able to trigger holistic modes of perception and processing, but there are some other
psychological effects of face stimuli on consumers’ attention and memory
performance which might also be relevant from a marketing perspective (e.g.,
37
emotional face stimuli are able to attract a lot of attention, and compared to non-face
objects faces are easier to remember) – so in the future it could be investigated
whether it is possible to find further face-specific effects apart from holistic processing
which can be triggered by anthropomorphic product designs.
Finally, recent research (Konar et al. 2010) suggests that there might exist considerable
interindividual differences in the extent of holistic face perception which might be an
interesting personality trait worth considering in future research on anthropomorphism
based on our finding that a holistic perception style can be triggered by
anthropomorphic object features.
Concluding, our research provides useful insights into the way product
anthropomorphizing may work, but it is still a long way to fully comprehend the
complex concept of anthropomorphism.
38
References
Aaker, Jennifer (1997), “Dimensions of Brand Personality,” Journal of Marketing
Research, 34 (August), 347-57.
Aggarwal, Pankaj (2004), “The Effects of Brand Relationship Norms on Consumer
Attitudes and Behavior,” Journal of Consumer Research, 31 (June), 87-101.
Aggarwal, Pankaj and Ann L. McGill (2007), “Is That Car Smiling at Me? Schema
Congruity as a Basis for Evaluating Anthropomorphized Products,” Journal of
Consumer Research, 34 (December), 468-79.
Alba, Joseph W. and J. Wesley Hutchinson (1987), “Dimensions of Consumer
Expertise,” Journal of Consumer Research, 13 (March), 411-54.
Anderson, John R. (1983), The Architecture of Cognition, Cambridge, MA: Harvard
Press.
Bajo, M. Teresa and José J. Canas (1989), “Phonetic and Semantic Activation During
Picture and Word Naming,” Acta Psychologica, 72 (2), 105-15.
Bargh, John A. and Tanya L. Chartrand (2000), “The Mind in the Middle: A Practical
Guide to Priming and Automaticity Research,” in Handbook of Research
Methods in Social and Personality Psychology, ed. Harry T. Reis and Charles
M. Judd, New York: Cambridge, 253-85.
Baumgartner, Hans (1993), “An Exploratory Investigation of Holistic and Analytic
Modes of Product Perception,” Advances in Consumer Research, 20, 673-77.
Bentin, Shlomo, Truett Allison, Aina Puce, Erik Perez, and Gregory McCarthy (1996),
“Electrophysiological Studies of Face Perception in Humans,” Journal of
Cognitive Neuroscience, 8 (6), 551-65.
Bentin, Shlomo, Noam Sagiv, Axel Mecklinger, Angela Friederici, and Yves D. von
Cramon
(2002),
“Priming
Visual
Face-Processing
Mechanisms:
Electrophysiological Evidence,” Psychological Science, 13 (2), 190-93.
Biederman, Irving (1987), “Recognition by Components: A Theory of Human Image
Understanding,” Psychological Review, 94 (2), 115-47.
Bruce, Vicki and Andrew Young (1998), In the Eye of the Beholder: The Science of
Face Perception, Oxford: Oxford University Press.
Burroughs, James E. (1996), “Product Symbolism, Self Meaning, and Holistic
Matching: The Role of Information Processing in Impulsive Buying,” Advances
in Consumer Research, 23, 463-69.
Bush, Donald J. (1990), “Body Icons and Product Semantics,” in Semantic: Visions in
Design, ed. Susann Vihma, Helsinki: UIAH, C1-C14.
39
Caporael, Linnda R. (1986), “Anthropomorphism and Mechanomorphism: Two Faces
of the Human Machine,” Computers in Human Behavior, 2 (3), 215-34.
Chartrand, Tanya L., Gráinne M. Fitzsimons, and Gavan J. Fitzsimons (2008),
“Automatic Effects of Anthropomorphized Objects on Behavior,” Social
Cognition, 26 (2), 198-209.
Chitturi, Ravindra, Rajagopal Raghunathan, and Vijay Mahajan (2007), “Form Versus
Function: How the Intensities of Specific Emotions Evoked in Functional
Versus Hedonic Trade-Offs Mediate Product Preferences,” Journal of
Marketing Research, 44 (November), 702-14.
Collins, Allan M. and Elizabeth F. Loftus (1975), “A Spreading-Activation Theory of
Semantic Processing,” Psychological Review, 82 (6), 407-28.
Cowley, Elizabeth and Andrew A. Mitchell (2003), “The Moderating Effect of Product
Knowledge on the Learning and Organization of Product Information,” Journal
of Consumer Research, 30 (December), 443-54.
Donnelly, Nick, Glyn W. Humphreys, and Jean Sawyer (1994), “Stimulus Factors
Affecting the Categorization of Faces and Scrambled Faces,” Acta
Psychologica, 85 (3), 219-34.
Eagly, Alice and Shelly Chaiken (1993), The Psychology of Attitudes, Fort Worth, TX:
Harcourt Brace Jovanovich.
Epley, Nicolas, Adam Waytz, and John T. Cacioppo (2007), “On Seeing Human: A
Three-Factor Theory of Anthropomorphism,” Psychological Review, 114 (4),
864-86.
Erk, Susanne, Manfred Spitzer, Arthur P. Wunderlich, Lars Galley, and Henrik Walter
(2002), “Cultural Objects Modulate Reward Circuitry,” NeuroReport, 13 (18),
2499–2503.
Evans, Jonathan St. B. T. (2008), “Dual-Processing Accounts of Reasoning, Judgment
and Social Cognition,” Annual Review of Psychology, 59, 255-78.
Farah, Martha J., Kevin D. Wilson, Maxwell Drain, and James N. Tanaka (1998),
“What Is ‘special’ about Face Perception?” Psychological Review, 105 (3), 48298.
Fournier, Susan (1998), “Consumers and Their Brands: Developing Relationship
Theory in Consumer Research,” Journal of Consumer Research, 24 (March),
343-73.
Gasper, Karen and Gerald L. Clore (2002), “Attending to the Big Picture: Mood and
Global versus Local Processing of Visual Information,” Psychological Science,
13 (1), 34-40.
40
Gauthier, Isabel, Pawel Skudlarski, John C. Gore, and Adam W. Anderson (2000),
“Expertise for Cars and Birds Recruits Brain Areas Involved in Face
Recognition,” Nature Neuroscience, 3 (2), 191-97.
Gombrich, Ernst H. (1973), “Illusion and Art”, in Illusion in Nature and Art, ed.
Richard L. Gregory and Ernst H. Gombrich, London: Gerald Duckworth.
Guthrie, Steward E. (1993), Faces in the Clouds: A New Theory of Religion, New
York: Oxford.
Holbrook, Morris B. and William L. Moore (1981). “Feature Interactions in Consumer
Judgments of Verbal Versus Pictorial Presentations,” Journal of Consumer
Research, 8 (June), 103-13.
Hole, Graham J., Patricia A. George, and Victoria Dunsmore, (1999), “Evidence for
Holistic Processing of Faces Viewed as Photographic Negatives,” Perception,
28 (3), 341-59.
Hutchinson, J. Wesley and Joseph W. Alba (1991), “Ignoring Irrelevant Information:
Situational Determinants of Consumer Learning,” Journal of Consumer
Research, 18 (December), 1991.
Ingram, Jack and Louise Annable (2004), “’I See You Baby, Shakin' That Ass’: User
perceptions of Unintentional Anthropomorphism and Zoomorphism in
Consumer Products,” Proceedings of the 4th Design and Emotion Conference,
Ankara, Turkey.
Irvin, Deborah J. and Stephen J. Lupker (1983), “Semantic Priming of Pictures and
Words: A Levels of Processing Approach,” Journal of Verbal Learning and
Verbal Behavior, 22 (1), 45-60.
Ito, Tiffany A. and John T. Cacioppo (2000), “Electrophysiological Evidence of
Implicit and Explicit Categorization Processes,” Journal of Experimental Social
Psychology, 36 (6), 660-76.
Jeffreys, D. Aled (1996), “Evoked Potential Studies of Face and Object Processing,”
Vision & Cognition, 3 (1), 1-38.
Johnson, Erich J. and J. Edward Russo (1984), “Product Familiarity and Learning New
Information,” Journal of Consumer Research, 11 (June), 542-50.
Kimchi, Ruth and Stephen E. Palmer (1982), “Form and Texture in Hierarchically
Constructed Patterns,” Journal of Experimental Psychology: Human Perception
and Performance, 8 (4), 521-35.
Konar, Yaroslav, Patrick J. Bennett, and Allison B. Sekuler (2010), "Holistic
Processing is not Correlated with Face-Identification Accuracy," Psychological
Science, 21 (1), 38-43.
41
Kroll, Judith F. and Mary C. Potter (1984), “Recognizing Words, Pictures, and
Concepts: A Comparison of Lexical, Object and Reality Decisions,” Journal of
Verbal Learning and Verbal Behavior, 23 (February), 39-66.
Labroo, Aparna, Ravi Dhar, and Norbert Schwarz (2008), “Of Frog Wines and
Frowning Faces: Semantic Priming, Perceptual Fluency, and Brand
Evaluation,” Journal of Consumer Research, 34 (April), 819-31.
Lawson, Rebecca (2007), “Local and Global Processing Bias Fail to Influence Face,
Object, and Word Recognition,” Visual Recognition, 15 (6), 710-40.
Lui, Jia, Alison Harris, and Nancy Kanwisher (2002), “Stages Of Processing in Face
Perception: a MEG Study,” Nature Neuroscience, 5 (9), 910-16.
Lui, Jia, Masanori Higuchi, Alec Marantz, and Nancy Kanwisher (2000), “The
Selectivity of the Occipitotemporal M170 for Faces,” Neuroreport, 11 (2),
2319-24.
Macchi Cassia, Viola, Marta Picozzi, Dana Kuefner, Emanuela Bricolo, and Chiara
Turati (2008), “Holistic Processing for Faces and Cars in Preschool-Aged
Children and Adults: Evidence from the Composite Effect,” Developmental
Science, 12 (2), 236-48.
Macrae, C. Neil and Helen L. Lewis (2002), “Do I know you? Processing Orientation
and Face Recognition,” Psychological Science, 13 (2), 194-96.
Mandler, George (1982), “The Structure of Value: Accounting for Taste,” in Affect
and Cognition: The 17th Annual Carnegie Symposium, ed. Margaret S. Clark
and Susan T. Fiske, Hillsdale, NJ: Erlbaum, 3-36.
Martinez, Aleix M. and Robert Benavente (1998), “The AR Face Database,” CVC
Technical Report, 24 (June).
Maurer, Daphne, Richard Le Grand, and Catherine J. Mondloch (2002), “The Many
Faces of Configural Processing,” Trends in Cognitive Sciences, 6 (6), 255-60.
McNamara, Timothy P. (1994), “Theories of Priming II: Types of Primes,” Journal of
Experimental Psychology: Learning, Memory, and Cognition, 20 (3), 507-20.
Meyers-Levy, Joan (1989), “Priming Effects on Product Judgments: A Hemispheric
Interpretation,” Journal of Consumer Research, 16 (1), 76-86.
Meyers-Levy, Joan and Alice M. Tybout (1989), “Schema Congruity as a Basis for
Product Evaluation,” Journal of Consumer Research, 16 (June), 39-54.
Mondloch, Catherine J., Terri L. Lewis, D. Robert Budreau, Daphne Maurer, James L.
Dannemiller, Benjamin R. Stephens, and Kathleen A. Kleiner-Gathercoal
(1999), “Face Perception during Early Infancy,” Psychological Science, 10
(September), 419-22.
42
Monga, Alokparna Basu and Deborah Roedder John (2007), “Cultural Differences in
Brand Extension Evaluation: The Influence of Analytic versus Holistic
Thinking,” Journal of Consumer Research, 33 (March), 529-36.
Monga, Alokparna Basu and Deborah Roedder John (2008), “When Does Negative
Brand Publicity Hurt? The Moderating Influence of Analytic versus Holistic
Thinking,” Journal of Consumer Psychology, 18 (4), 320-32.
Muller, Keith E. and Bethel A. Fetterman (2003), Regression and ANOVA: An
Integrated Approach using SAS Software, New York: John Wiley & Sons.
Neely, James H. (1991), “Semantic Priming Effects in Visual Word Recognition: A
Selective Review of Current Findings and Theory,” in Basis Processes in
Reading: Visual Word Recognition, ed. Derek Besner and Glyn W. Humphreys,
Hilsdale, NJ: Lawrence Erlbaum, 264-336.
Nisbett, Richard E., Kaiping Peng, Incheol Choi, and Ara Norenzayan (2001),
“Culture and Systems of Thought: Holistic versus Analytic Cognition,”
Psychological Review, 108 (2), 291-310.
Perfect, Timothy J. (2003), “Local Processing Bias Impairs Line-up Performance,”
Psychological Reports, 93 (2), 393-94.
Rigdon, Mary, Keiko Ishii, Motoki Watabe, and Shinobu Kitayama (2009), “Minimal
Social Cues in the Dictator Game,” Journal of Economic Psychology, 30 (3),
358-67.
Rossion Bruno, Christine Schiltz, and Caroline Jacquier (2008), “Individual Faces Are
Perceived Holistically in the Occipito-temporal Cortex at About 150 ms,” in
Frontiers in Human Neuroscience: 10th International Conference on Cognitive
Neuroscience, doi: 10.3389/conf.neuro.09.2009.01.369.
Sagiv, Noam and Shlomo Bentin (2001), “Structural Encoding of Human and
Schematic Faces: Holistic and Part Based Processes,” Journal of Cognitive
Neuroscience, 13 (7), 1-15.
Seeck, Margitta and Otto-Joachim Grüsser (1992), “Category-related Components in
Visual Evoked Potentials: Photographs of Faces, Persons, Flowers, and Tools as
Stimuli,” Experimental Brain Research, 92 (2), 338-49.
Sergent, Justine (1989), “Structural Processing of Faces,” in Handbook of Research on
Face Processing, ed. Andrew W. Young and Hadyn D. Ellis, North-Holland:
Elsevier Science Publishers.
Shepherd, John, Graham Davies, and Hadyn Ellis (1981), “Studies of Cue Saliency,”
in Perceiving and Remembering Faces, ed. Graham Davies, Hadyn Ellis, and
John Shepherd, London: Academic Press, 105-31.
43
Smith J. David and Deborah G. Kemler Nelson (1984), “Overall Similarity in Adults’
Classification: The Child in All of Us,” Journal of Experimental Psychology:
General, 113 (January), 137-59.
Tanaka, James N. and Martha J. Farah (1993), “Parts and Wholes in Face
Recognition,” Quarterly Journal of Experimental Psychology: Human
Experimental Psychology, 46 (2), 225-45.
Tanaka, James N. and Joseph A. Sengco (1997), “Features and Their Configuration in
Face Recognition,” Memory and Cognition, 25 (5), 583-92.
Theios, John and Paul C. Amrhein (1989), “Theoretical Analysis of the Cognitive
Processing of Lexical and Pictorial Stimuli: Reading, Naming, and Visual and
Conceptual Comparisons,” Psychological Review, 96 (1), 5-24.
Thorpe, Simon, Karl R. Gegenfurtner, Michèle Fabre-Thorpe, and Heinrich H.
Bülthoff (2001), “Detection of Animals in Natural Images Using far Peripheral
Vision,” European Journal of Neuroscience, 14 (5), 869-76.
Vanderwart, Mary (1984), “Priming by Pictures in Lexical Decision,” Journal of
Verbal Learning and Verbal Behavior, 23 (1), 67-83.
Valenza, Eloisa, Francesca Simion, Viola Macchi Cassia, and Carlo Umiltà (1996),
“Face preference at birth,” Journal of Experimental Psychology: Human
Perception and Performance, 22 (4), 892-903.
Welsh, Jonathan (2006), “Why Cars Got Angry,” Wall Street Journal, March 10, W1.
Wilder, Joseph (1957), “The Law of Initial Values in Neurology and Psychiatry,”
Journal of Nervous and Mental Disease, 125 (1), 73-86.
Windhager, Sonja, Dennis E. Slice, Katrin Schaefer, Elisabeth Oberzaucher, Truls
Thorstensen, and Karl Grammer (2008), “Face to Face: The Perception of
Automotive Designs,” Human Nature, 19 (December), 331-46.
Wolfe, Jeremy M., Serena J. Butcher, Carol Lee, and Megan Hyle (2003), “Changing
Your Mind: On the Contributions of Top-Down and Bottom-Up Guidance in
Visual Search for Feature Singletons,” Journal of Experimental Psychology:
Human Perception and Performance, 29 (2), 483-502.
Yoon, Carolyn, Angela H. Gutchess, Fred Feinberg, and Thad A. Polk (2006), “A Functional
Magnetic Resonance Imaging Study of Neural Dissociations between Brand and
Person Judgments,“ Journal of Consumer Research, 33 (June), 31-40.
Young, Andrew W., Deborah J. Hellawell, and Dennis C. Hay (1987), “Configural
Information in Face Perception,” Perception, 16 (6), 747-59.
Zeithaml, Valarie A. (1988), “Consumer Perceptions of Price, Quality, and Value: A MeansEnd Model and Synthesis of Evidence,” Journal of Marketing, 52 (3), 2-22.
Paper 2: Affective Consumer Responses to Babies in NonFaces and Non-Babies: An Evolutionary Perspective of BabySchema Effects in Product Designs
Linda Miesler(1), Helmut Leder(2), Andreas Herrmann(3) ∗
Abstract
By applying an evolutionary approach, we examined innate affective consumer
responses to features of a baby schema in product designs. Although previous studies
have suggested that consumers might perceive car fronts in a similar way as they do
human faces, it is not known how consumers respond on an affective level to
biologically significant features when these are present in artifacts. Manipulating car
fronts and human faces in accordance with the baby schema and combining facial
electromyography with cuteness ratings to assess spontaneous emotional responses, we
found that both female and male participants showed more positive affective responses
to the babyfaced than to the original stimuli, and that the effect of baby schema
features on emotions was stable over repeated exposures. These results confirm that
consumer responses to visual product designs are affected by evolutionarily-explained
determination.
Introduction
The emotional value of products is promoted as highly important by marketing
researchers and practitioners (Desmet, Overbeeke, and Tax 2001; Khalid and Helander
2006). Marketers and product designers often apply intuitive theories about how the
consumer’s mind works and what appeals to consumer emotions (Colarelli and
Dettman 2003). Human skills in reading emotions are well developed, and the variety
of emotions revealed by faces is more diverse in humans than in any other species
(Ekman 1982). Consequently, an often-found strategy to produce emotions in product
design is to create anthropomorphic features. This is reflected, for example, by a car
designer’s concern for the “face” of a car (Welsh 2006; Windhager et al. 2010). Other
examples are the creation of cute, babyish-appearing product designs, such as those of
∗
(1) Linda Miesler, Doctoral Candidate, Center for Customer Insight, University of St. Gallen; (2) Helmut
Leder, Professor of Psychology, Department of Basic Research in Psychology, University of Vienna; (3)
Andreas Herrmann, Professor of Marketing, Center for Customer Insight, University of St. Gallen
Paper 2: Affective Consumer Responses to Babies in Non-Faces and Non-Babies
45
the Volkswagen Beetle and Mini Cooper (Marcus 2002; Patton 1998). Product
designers therefore make use of a deeply embedded human trait - intuitively or
deliberately; due to the evolutionary significance of human features, consumers are
highly sensitive and attracted to them (Coss 2003; Guthrie 1993). Surprisingly,
although quite established in marketing practice (e.g., think of “sex sells” in
advertisements), evolutionary theory has been largely neglected as a useful framework
for researchers working in the fields of marketing, consumer research, and product
development (Saad 2008). This is surprising, as evolutionary adaptations are hardwired and universal; therefore, evolutionary adaptations are promising in their effects
on consumer behavior. To date, only a few studies have tested hypotheses about
consumer behavior derived from an evolutionary psychology framework (DiClemente
and Hantula 2003; Durante et al. 2010; Foxall and James 2003; Griskevicius, Shiota,
and Nowlis 2010; Saad and Gill 2003).
In the present research, we studied one type of innate perceptual process: the detection
of visual key stimuli (or visual releasers, Coss 2003) and the nature of the resulting
affective responses. Specifically, we examined responses to the baby schema
(“Kindchenschema”, Lorenz 1943) in product designs. The project had two basic
objectives. First, we studied whether adaptive affective responses to features of the
baby schema are generalized to consumer products (i.e., product designs). Second, we
studied whether such evolutionarily determined responses keep their special universal
nature even after generalization to consumer products. With ‘universal’ we refer to,
first, temporal properties and, second, gender-related properties of the elicited
affective responses, which are both relevant dimensions from a marketing point of
view. To answer these questions, we manipulated car fronts and human faces with
respect to the baby schema (e.g., by enlarging the headlights/eyes) and measured
spontaneous affective responses to all stimuli by recording facial-muscle activations
and behavioral measures.
Our research pursued three main goals. First, we extended the existing research on
anthropomorphic consumer perceptions by examining the direct impact of face-like
product designs on positive consumer emotions. Second, we explored the power of the
evolutionary psychology framework to explain consumer emotions. Third, we
demonstrated the efficiency of a non-verbal method to measure emotions reliably by
employing facial electromyography (EMG).
46
Biological Shapes in Product Designs
Product designs can be evaluated with regard to many different variables when
forming a preference judgment (e.g., how beautiful/ interesting/ innovative/ cute is a
product design?). Hitherto, most research dealing with innate design preferences has
been based on principles from Gestalt theory (cf., Arnheim 1954/1974) or the
experimental aesthetics approach (cf., Berlyne 1974). Studies examined general
stimulus dimensions and how these dimensions affect consumer preferences (e.g., Cox
and Cox 2002; Pittard, Ewing, and Jevons 2007; Veryzer 1993); dimensions
comprised, for instance, product proportions (Veryzer 1993), symmetry (Creusen,
Veryzer, and Schoormans 2010), complexity (Cox and Cox 2002), novelty (Hekkert,
Snelders, and van Wieringen 2003), innovativeness (Carbon and Leder 2005; Färber et
al. 2010), and prototypicality (Crothers, Montgomery, and Clarke 2003). A new line of
research focuses on preferences for specific shapes that mimic natural principles and
are biologically advantageous (Bar and Neta 2006). In particular, such approaches
directly test human-like shapes in product design by studying anthropomorphic
consumer perceptions (Aggarwal and McGill 2007; Chandler and Schwarz 2010;
Ingram and Annable 2004; Miesler et al. 2010). In particular, the car’s front-end is an
often-mentioned example in discussions of the perception of face-like forms in product
design. A few authors (Erk et al. 2002; Miesler et al. 2010; Windhager et al. 2010;
Windhager et al. 2008) have suggested that consumers process a car’s front-end
similarly to a human face, whereas these authors have mainly focused on the cognitive
mechanisms underlying such a face-specific processing of product designs (i.e.,
feature detection, categorization, inferences). Thus, Windhager et al. (2010) and Erk et
al. (2002) have provided evidence that consumers can detect facial features in a car
front easily. For example, Windhager et al. (2010) investigated people’s eye
movement patterns when they had to compare car fronts with human faces and found
that a car’s headlights are perceived correspondingly to the eyes, the grille
correspondingly to the nose, and the air intake or the grille correspondingly to the
mouth. In a lexical decision task, Miesler et al. (2010) provided evidence that due to
the car front’s face-like physical features, presenting a car shown in frontal view
automatically activates the mental concept of a face in the consumer’s mind, whereas
presenting a car shown from side view does not. In another study, Windhager et al.
(2008) showed that people draw the same inferences from car fronts as from human
faces (e.g., with regard to perceived sex or maturity).
47
However, the emotional effects of such anthropomorphic consumer perceptions have
not yet been examined; even though the emotional value of biological forms such as
faces is well known (Ekman 1982; Ellis and Young 1998). Even when positive affect
was measured, as in Aggarwal and McGill (2007), the increased positive affect was a
by-product of a cognitive process (i.e., successful schema congruity) rather than being
directly triggered by the product’s physical appearance.
Visual Key stimuli, Baby Schema, and Cuteness Response
It seems that it is often neglected in marketing research that “consumers are biological
and Darwinian beings” (Saad 2008, p. 426). On the other hand, according to the
framework of evolutionary psychology, all human behavior relies, to a certain degree,
on innate perceptual, cognitive, affective and/or motivational mechanisms that have
evolved through natural selection as adaptations to specific ancestral conditions (for
evolutionary psychology in marketing and consumer behavior, see, for example,
DiClemente and Hantula 2003; Durante et al. 2010; Griskevicius et al. 2010; Saad and
Gill 2000; Saad and Gill 2003). These adaptive mechanisms are functional in terms of
an increased chance of survival and reproduction. Even though modern consumers’
environments differ in some respects from the environments where these adaptive
mechanisms evolved, there are still domains of everyday life, such as mating (Buss
1994), food choice (Rozin 1976) and even aesthetic preferences (Voland and Grammer
2003), for which human preferences and behaviors are (partly) shaped by these deeply
embedded mechanisms. For instance, Griskevicius et al. (2010) recently applied
evolutionary theory to explain the influence of different positive consumer emotions
on product desirability.
With regard to perceptual adaptations, Coss (2003, p. 70) has postulated that the
detection and recognition of innate visual patterns (so-called visual key stimuli or
releasers) - including the processing of their emotional significance - occur rapidly, as
they allow the individual to immediately incorporate the stimulus’ relevance and
valence into a fast decision-making process and to quickly initiate the appropriate
adaptive behavior (in terms of approach or avoidance). For example, stimuli
signalizing a natural threat or danger (e.g., snakes) are recognized unconsciously or
pre-attentively (Brosch, Sander, and Scherer 2007; Oehman and Soares 1994).
Furthermore, Dimberg and Thunberg (1998) have provided evidence that biologically
adaptive, affective reactions are evoked spontaneously, quickly, and automatically
after a short duration of exposure to an affective stimulus. Hence, to categorize a
48
stimulus (e.g., a product) as positive or negative, evolved affective responses should
occur quickly and automatically. However, during ongoing exposure to a stimulus,
more strategic and controlled processes (e.g., in accordance with a consumer’s
learning history or social norms) might occur and modulate the initial affective
responses (e.g., Sander, Grandjean, and Scherer 2005). Therefore, innate affective
responses to visual key stimuli might be strongest immediately after the initial
appearance of a stimulus.
In the present research, we focused on one visual key stimulus, which is a prototypical
example of a biologically significant stimulus for members of the human species and
is known to have a high emotional value, the baby schema. We studied spontaneous
affective responses triggered by features of a baby schema in the first seconds of
appearance. The spontaneous reaction to infants is a positive affective orientation
towards the baby that is expressed in overt verbalizations, such as “oh, how cute!”,
often accompanied by smiling (Hildebrandt and Fitzgerald 1978; Power, Hildebrandt,
and Fitzgerald 1982; Schleidt et al. 1980). With regard to this affective “cuteness
response,” the ethologist Lorenz (1943) proposed that it is the physical appearance of
babies (e.g., a round face, large eyes, small nose and mouth, thick lips, a small chin,
and a high forehead) that serves as key stimulus to automatically trigger affection in
people, and that this, on the one hand, promotes parental nurturing and caretaking and,
on the other hand, inhibits aggression in parents and non-family members of either
gender. Consequently, offspring survival is enhanced. In accordance with Lorenz’
assumptions, behavioral studies have shown that the presence of baby-schema features
is positively correlated with perceived infant cuteness (Hildebrandt and Fitzgerald
1978; Glocker et al. 2009), adults’ motivation for caretaking (Glocker et al. 2009), and
behavioral tenderness (Sherman, Haidt, and Coan 2009). Moreover, some studies have
found that brain areas that are associated with the anticipation of reward (Glocker et al.
2009) or those that are involved in decoding a stimulus’ emotional value (Nitschke et
al. 2004) show an increased activation in the presence of cute infant faces. Brosch et
al. (2007) found an attentional bias in favor of pictures of babies when presenting
pictures of babies together with pictures of adults for less than one second, and
therefore, as responses to innate visual key stimuli in general, also baby schemaspecific responses are elicited very quickly - within a few seconds.
Importantly, human responses to features of a baby schema might be prone to
generalization, such that even non-infant objects possessing such features should
arouse the cuteness response (Lorenz 1943). In line with this assertion, Zebrowitz
49
(1997) and others found that humans are not only sensitive to baby-schema features in
infant faces, but that they also respond positively to infant animals (Sanefuji, Ohgami,
and Hashiya 2007), cute cartoon characters or dolls (Jacob, Rodenhauser, and Markert
1987), and babyfaced adults. Concerning the latter, there is clear evidence that people
are able to judge the babyfaceness of human faces of various ages (Zebrowitz and
Montepare 1992) and form trait impressions about others in accordance with these
judgments (Gorn, Jiang, and Johar 2008; Livingston and Pearce 2009; Poutvaara,
Jordahl, and Berggren 2009). Zebrowitz et al. (2003) postulated that the similarities in
responses to babies and to babyfaced adult faces are due to an overgeneralization of
the evolutionarily adaptive response of identifying babies.
Hence, in the context of product design, the question arises: do visual features of a
baby schema (e.g., large eyes and a small nose) elicit innate positive affective
responses in consumers even when they are transferred to the design of non-human,
non-living consumer products, and does it occur in the first seconds of appearance?
Because we were interested in spontaneously triggered affect, beyond explicit selfreports, we investigated consumer emotions by an implicit psycho-physiological
method, facial EMG (for a review of psycho-physiological methods in marketing, see
Wang and Minor 2008; for an empirical study, see Yoon, Gutchess, Feinberg, and
Polk 2006). As affective responses to objects also manifest in specific facial
expressions such as smiling or frowning (Ekman 1972; Ekman, Friesen, and Ancoli
1980; Ekman et al. 1982), facial EMG reliably captures changes in positive and
negative emotional states, even when overt facial expressions are absent (Cacioppo et
al. 1986; Dimberg, Thunberg, and Elmehed 2000; Topolinski et al. 2009; Winkielman
and Cacioppo 2001). As the responses triggered by product designs should be quite
mild and subtle (cf., Desmet, Hekkert, and Jacobs 2000) compared to those elicited by
emotionally stronger cues, such as angry faces, we expected the responses to
babyfaced product designs not to be accompanied by overt facial reactions. Therefore,
facial EMG was the method of choice. Two muscles are always discussed in the context of
measuring emotional responses via facial EMG. One is the Musculus corrugator supercilii,
which furrows the brow (the “frowning muscle”) and is mainly related to negative affective
states; the other is the Musculus zygomaticus major, which raises the corners of the mouth
(the “smiling muscle”) and is therefore mainly related to positive affective states. For our
purpose, the latter muscle was mainly relevant. However, some authors demonstrated
that positive affective states are not solely indicated by an increase of zygomaticus
major activity but also by a decrease of corrugator supercilii activity (Dimberg 1990).
Thus, with regard to our first research question of whether baby-schema features in
50
product designs can elicit evolved consumer affect spontaneously, we assumed that
stimuli with features of the baby schema elicit larger positive affective responses than
less-babyfaced stimuli, indicated by changes in facial muscle activations. In particular,
we postulated:
H 1: Products and faces that were manipulated in accordance with the baby schema
spontaneously elicit a larger activation of M. zygomaticus major and a lower
activation of M. corrugator supercilii compared to the original, less-babyfaced
stimulus versions.
In all studies, human faces served as a reference group to test for the internal validity
of the results found for the product designs. If the results we found for the product
designs were due to innate affective mechanisms elicited by the varying degree of
babyfaceness, we should find the same effects with human faces, which were
manipulated in the same way as the product designs. Besides the spontaneous
elicitation of positive affective responses, responses to key stimuli in general and to
the baby schema in particular are supposed to have further specific properties. As
responses to baby-schema features are relevant for survival of all members of the
human species, the nature of the affective responses is supposed to be universal
(Lorenz 1943) which refers to both the dimensions of time and gender. Because both
aspects are highly relevant from a marketing point of view as well, we formulated as
an additional research question: can further properties of innate affective responses to
features of the baby schema also be found when the responses are generalized to
consumer products?
With regard to the temporal dimension, it might be essential for companies that their
product designs attract consumers not only at the first sight, but also in the long run.
Innate responses to biologically significant stimuli feature a special temporal property:
the elicited responses are relatively stable over time and do not habituate (or habituate
very slowly, respectively). For example, Dimberg and Thunberg (1998) found that
spontaneous affective responses to emotional expressive faces did not change over
repeated exposures. The lack of habituation can be explained by the adaptive value of
the responses triggered by biologically significant stimuli (e.g., key stimuli) for
survival. Thus, for the survival of infants, it is most important that adults show the
same affectionate response to an infant at each exposure, no matter how often they are
51
exposed to it, and in consequence, adults’ responses to babies should not be suspect to
habituation. This stability of the emotional responses makes biological shapes special
compared to the more general design features (e.g., complexity) that have been
investigated before. Thus, it is known that consumer responses to product designs are
quite susceptible to repeated exposures, such that consumer preferences or affective
reactions can change over time, depending on the specific product design features
(e.g., level of visual complexity, Cox and Cox 1988, 2002; or innovativeness, Carbon,
Michael, and Leder 2008). For example, the liking for visually simple designed
products decreases quickly with repeated exposure, due to tedium, whereas the liking
for complex designs is low at the beginning but increases with repeated exposure, due
to familiarization (Cox and Cox 2002; see also Bornstein 1989), and might decrease
again after a large number of repetitions (Berlyne 1970; Tinio and Leder 2009).
Therefore, with regard to positive affective responses to features of a baby schema in
product designs, we posit the following:
H 2a: Due to their biological significance, positive affective responses to products and
faces with baby-schema features should be less susceptible to repeated exposure
(i.e., do not habituate) than responses to less-babyfaced stimulus versions.
With regard to gender, understanding differences and similarities between females and
males in consumption is compelling (cf., Saad and Gill 2000, p. 1015; Durante et al.
2010). Lorenz (1943) postulated that all humans irrespective of their gender have an
evolved perceptual bias to find baby-schema features more attractive than mature
features. Recently, Glocker et al. (2009) systematically manipulated the babyfaceness
of infant faces and found that, in line with Lorenz’ assumption, both females and
males perceived pictures of babyfaced infants as cuter than those of less-babyfaced
infants. Brosch et al. (2007) also found no gender differences in the attentional bias
toward babies, using an experimental paradigm where features of the baby schema had
to be rapidly detected. On the other hand, other authors have argued that females
should have a more pronounced preference for infantile features because their
nurturing and care-giving responsibility is greater than men’s (Harlow 1971). Thus,
some authors have reported gender differences in adults’ interest in infants and infantdirected behavior (Berman 1980; Maestripieri and Pelka 2002). While the reported
gender differences were most pronounced in studies that examined actual or
52
hypothetical overt behavior (e.g., interaction with real infants, willingness to be near
infants, and motivation to engage in care-taking activities), we studied the perceptual
sensitivity to baby-schema cues and the related spontaneous affective responses prior
to behavioral consequences. Thus, we expected in accordance with Lorenz and
existing evidence (Brosch et al. 2007; Glocker et. al 2009):
H 2b: Affective responses to products and faces that vary with regard to babyfaceness
should not be different between females and males, such that both gender
groups show larger spontaneous affective responses to products and faces with
baby-schema features than to the less-babyfaced stimulus versions.
The three main hypotheses were tested in a facial EMG study. First, a pretest was
conducted to develop and validate the stimulus materials for the main study. We
measured affective responses to the stimuli by assessing self-reported cuteness ratings
(e.g., Glocker et al. 2009). In the main study, beyond explicit verbal ratings, facial
EMG was employed to assess spontaneous and subtle affective responses to the
stimuli.
Methods
First, a pretest was conducted to select the appropriate baby-schema (~ feature size)
manipulations to be applied to the two object categories (car fronts and human faces)
and to check if the manipulations produced changes in perceived cuteness in the two
object classes that were comparable in size.
Pretest
Participants. Thirty-five participants took part in the pretest. One group of participants
rated pictures of cars (n = 19; Mage = 23 yrs, SD = 3 yrs; 74% females), and another
group rated pictures of faces (n = 16; Mage = 27 yrs, SD = 6 yrs; 75% females) for
cuteness on a seven-point Likert scale (1 = “not cute at all” to 7 = “very cute”). All
participants were students of University of Vienna.
Materials. Grayscale pictures of 16 cars shown in frontal view served as stimuli for the
product design category (picture size: 512×512 pixels). To ensure that the effects were
independent of a particular brand or segment, brand logos were eliminated, and the car
picture set comprises nine cars from the compact car segment (e.g., Fiat, Mini) and
53
seven from the middle-class segment (e.g., Mercedes, BMW). The face picture set
consisted of grayscale pictures of 16 human faces with neutral emotional expressions
(eight male and eight female faces; picture size: 370×555 pixels). The pictures were
taken from the Vienna Face Database, which contains standardized pictures of male
and female students with an age range from 18 to 25 years.
For each picture, a babyfaced version was created by a professional graphic designer
using Adobe Photoshop. Therefore, the relative sizes of three selected local features
(the headlights/eyes, the middle grille/nose, and the air intake/mouth) were
manipulated. The features and the appropriate size manipulations were selected in
accordance with literature on physical cues characterizing the baby schema (Zebrowitz
1997). Furthermore, the selected facial features clearly corresponded to the features of
a car front (e.g., the car’s headlights as human eyes), as in Windhager et al. (2010).
For each of the 16 original cars, a babyfaced version was created by enlarging the
headlights by 20% (because babies have proportionally large eyes), shrinking the
middle grille by 20% (because babies have proportionally small noses) and decreasing
the width of the air intake by 20% while simultaneously increasing its height by 20%
(because babies have small mouths but relatively thicker lips than adults). Other
authors have also employed such feature-size manipulations in a range of 10-20%
(e.g., Keating et al. 2003). Size manipulations in the face stimuli were set to 10%
because another pretest with 19 participants revealed that larger size manipulations
created unnatural face versions. The faces’ relational characteristics (e.g., distance
between the nose and upper lip, and distance between the eyes) were changed as little
as possible (for examples of stimulus materials, see Figure 1).
The effects on perceived cuteness were comparable for cars and faces, although the
size manipulations differed in quantity between the two object categories, as was
confirmed in a 2 (feature size: original versus babyfaced) × 2 (object category: car
versus face) ANOVA on the average rated cuteness of the cars and faces. We found a
significant main effect of feature size, F (1, 30) = 107.08, p < .001, ηp2 = 0.78, and the
babyfaced car fronts and faces (M = 3.85, SD = 1.13) were perceived as cuter than the
original stimuli (M = 3.33, SD = 1.09). Moreover, we found no significant main effect
of object category and no interaction of feature size with object category (both Fvalues < 1). Therefore, although the applied feature size manipulations were different
in quantity, the effects of the manipulation on perceived cuteness were comparable
between the two object categories. This warranted comparing the affective responses
to features of a baby schema in products and faces, as examined in the main study.
54
Moreover, an additional pretest (n = 25; 56% females) ensured that the feature-size
manipulation was not confounded with the perceived visual complexity of the car
designs (Mbabyfaced = 3.54, SD = 1.08; Moriginal = 3.56, SD = 1.1; F (1, 24) < 1), which
was important to test hypothesis 2a.
FIGURE 1
Examples of Manipulated (Babyfaced) and Non-Manipulated (Original) Pictures
of Cars and Faces
Facial EMG Study
Participants. Fifty-seven undergraduate students from University of Vienna
participated in the facial EMG study for partial course credit. Data of four persons had
to be excluded in both the car and face groups due to inappropriate behavior during the
experimental session (e.g., sleepiness) and/or too many movement artifacts (e.g.,
chewing and yawning). Thus, the final sample consisted of 28 car group participants
(Mage = 22 yrs, SD = 2 yrs; 61%females) and 21 face group participants (Mage = 21 yrs,
SD = 2 yrs; 67%females).
Design and Stimuli. The study design was a 2 (feature size: original versus babyfaced)
× 2 (repeated exposure: first exposure versus second exposure) × 2 (object category:
car fronts versus human faces) mixed design with repeated measurements on the first
two factors and object category as the between participants factor. Because a pretest
with seven participants had shown that the presentation of both faces and cars to the
same sample would result in carry-over-effects in responses to the two object
categories, the object categories were varied between participants. This also prevented
55
participants from reflecting on the possible relationship between car fronts and human
faces. Moreover, differences in time course could be addressed with respect to a fivesecond presentation time (see Procedure section). Overall, the stimulus set comprised a
total of 64 pictures, 32 car and 32 face pictures, according to the pretest. All stimuli
were presented in the middle of a 30-inch monitor on a medium grey background
(RGB 220, 220, 220) to reduce eyestrain from the computer monitor.
Procedure. The experiment was conducted with one participant at a time. Before the
experiment, the participants were briefed on the EMG electrode attachment procedure,
and were told that skin conductance reactions would be recorded to reduce demand
characteristics (e.g., Dimberg and Thunberg 1998; Weyers et al. 2006). Subsequently,
the skin of the muscle sites, where the electrodes were attached afterwards, was
cleaned. Participants were seated 1 m in front of the monitor. They were given brief
instructions on the overall procedure of the study, and were informed that they would
be filmed via a video camera attached to the top of the computer monitor during the
whole session. They were told that the only purpose of the camera was to enable
communication between the participant and the experimenter (e.g., in the case that an
electrode detaches) because the experimenter observed the experiment on a monitor in
a separate room.
Stimuli pictures were presented in a block design containing two consecutive
evaluation blocks. In the first block, participants rated the picture stimuli with regard
to attractiveness; in the second block, they rated the same stimuli with regard to
cuteness. Such explicit evaluations do not interfere with automatic psychophysiological responses, such as those assessed by facial EMG (cf., Glocker et al.
2009). We decided against the alternative approach to assess muscle activation in a
mere viewing task (without ratings) because in the face of complex product designs
such a method would make it difficult to explain afterwards which stimulus dimension
caused the affective response (e.g., general liking, the baby-schema cues, or
familiarity). For the same purpose of controlling for response specificity, we included
the attractiveness evaluation block in addition to the cuteness block to explore if the
affective responses differed between the two evaluation blocks. Between the two
blocks, participants took a short break of five minutes to relax the eyes. To conceal the
study’s aim (babyfaceness of product designs) participants always completed the
attractiveness block first and the cuteness block afterwards. Further, during the
attractiveness block, participants could familiarize themselves with the procedure.
Following the two evaluation blocks, the participants in the car group were exposed to
56
a third block where they rated the original version of each car with regard to
familiarity. However, because including familiarity as a control variable in our
analyses revealed that familiarity with the car brands had no effect on our main results,
we will not consider this variable later in the article. All ratings were made on 7-pointLikert scales with 1 = “not attractive/ not cute/ (not familiar)” and 7 = “very
attractive/cute/ (familiar).”
Each block began with three practice trials, which were not used in the subsequent test
trials. Within an evaluation block, each of the stimuli (cars or faces) was presented
twice to examine habituation effects (hypothesis 2a). The participants saw at first all
32 stimuli in random order, before the stimuli were newly randomized and presented a
second time. The original and the babyfaced version of the same stimulus were never
presented in a row. Each trial started with a fixation cross (3 s), followed by the
stimulus (5 s), then the rating scale appeared in the middle of the screen. The inter-trial
interval was 4 s. At the end of the experimental session, the participants’ gender and
age were assessed, and they were debriefed about what they thought the goal of the
study was and why electrodes were attached. The whole procedure took approximately
50-60 minutes.
Facial EMG Recording and Data Preprocessing. Facial EMG was recorded over the
M. zygomaticus major and the M. corrugator supercilii muscle sites of the left side of
the face (Fridlund and Cacioppo 1986). One pair of silver/silver chloride bipolar
surface electrodes (4 mm diameter/ 7 mm housings) was placed over each muscle site.
The ground electrode was located on the bone behind the right ear. Impedances of all
electrodes were reduced to less than 10 kΩ. The EMG raw signals were recorded with
a TMS International Portilab 20 channel amplifier at a sampling frequency of 2048 Hz.
Raw data were filtered offline with a 20 Hz high pass filter and a 50 Hz notch filter.
Moreover, raw data were screened offline for movement artifacts by crosschecking
salient EMG signals with the video recordings. Thus, trials containing movement
artifacts (e.g., biting, chewing, coughing, and speaking) were excluded from further
analyses.
Raw EMG signals represented changes in muscle activation in microvolts as a function
of time with 2,048 data points per second. To facilitate data processing, several further
data-processing steps were performed offline. Thus, raw data were full-wave rectified
(i.e., all data values turned into positive polarity) and integrated with a time constant of
125 ms to reduce data complexity (Topolinski et al. 2009; Weyers et al. 2006). Finally,
data were standardized to z-scores within participants and muscle sites to make
57
responses comparable across muscle sites and participants independent of the
individual participant’s reactivity (Winkielman and Cacioppo 2001). EMG activations
during stimulus presentation were averaged in five intervals of 1 second and expressed
in terms of change scores relative to a pre-stimulus baseline. The average EMG
activity during the last second of the 4-seconds fixation cross before the stimulus was
presented provided baseline values. For statistical comparisons, EMG data were
averaged over the 16 stimulus presentation trials in each of the two feature-size
conditions (original versus babyfaced), separately for the first and second stimulus
presentations within a block (first exposure versus second exposure). All offline data
processing steps were computed in Matlab 7.1 using EEGLAB toolbox (Delorme and
Makeig 2004). Statistical analyses were conducted with SPSS 18.
Results
The study’s main goal was to investigate whether features of the baby schema produce
automatic positive affective responses in faces and - even more interesting - in car
fronts (hypothesis 1), how these affective responses change over time (hypothesis 2a),
and whether the effects are stable across gender (hypothesis 2b). For each hypothesis,
we present the results separately for car fronts and human faces. The facial EMG
responses evoked during cuteness evaluations were our primary interest, but at the end
of the results section, we also briefly present results from the attractiveness evaluation
block.
Manipulation Check Based on Cuteness Ratings. To validate if our feature-size
manipulation was successful, we analyzed the self-reported cuteness ratings. We
calculated the average rated cuteness for the original and babyfaced stimuli (first
exposure only) and compared the two means for all participants. In both object
categories, the behavioral results were as expected. The babyfaced car fronts were
perceived as cuter than the original ones (Mbabyfaced = 3.72, SDbabyfaced = 0. 54 versus
Moriginal = 3.49, SDoriginal = 0.47; t (27) = 2.98, p = .006, Cohen’s d = 0.46). Therefore,
the feature size manipulation successfully influenced cuteness perception among the
facial EMG participants. The same held for the face stimuli group: participants
perceived the babyfaced faces as cuter than the original faces (Mbabyfaced = 3.90,
SDbabyfaced = 0.78 versus Moriginal = 3.50, SDoriginal = 0.66; t (20) = 5.03, p < .001,
Cohen’s d = 0.55).
58
Spontaneous Affective Responses during Cuteness Evaluation (Hypothesis 1)
To test if babyfaced car fronts and faces elicited larger positive affective responses
than the original stimuli, indicated by differences in facial EMG activity, we submitted
the participants’ facial EMG data to four 2 (feature size: original versus babyfaced) × 5
(time interval: seconds 1 to 5 after stimulus onset) repeated measurement ANOVAs,
separately for the two muscle sites of zygomaticus major and corrugator supercilii and
the two object categories (car fronts versus faces). Only responses from the first
exposure block were analyzed. The number of included trials per feature-size
condition varied between 5 and 16 (both cars and faces: M = 11 trials, SD = 3). Facial
EMG data for car fronts and human faces are plotted in Figure 2 as a function of the
time interval, with separate panels for activity over the zygomaticus major and
corrugator supercilii.
Facial EMG Responses to Car Fronts. The overall zygomaticus major activity was
larger for babyfaced car fronts (M = 0.07, SD = 0.23) than for the original cars (M = .10, SD = 0.18), during the five seconds of first stimulus exposure, F (1, 27) = 8.34, p
= .008, ηp2 = 0.24. The main effect of the time interval was not significant (F (1, 27) <
1), nor was the interaction of feature size with time interval (F (2.58, 69.75) = 2.23, p
= .101, ηp2 = 0.08; Greenhouse-Geisser corrected). Separate analyses for each of the
five time intervals revealed that babyfaced car fronts elicited a significantly larger
zygomaticus major activation than the original cars already in the first second after
stimulus onset (1st second: F (1, 27) = 4.33, p = .047, ηp2 = 0.14; 2nd second: F (1, 27)
= 7.79, p = .010, ηp2 = 0.22; 3rd second: F (1, 27) = 2.84, p = .104, ηp2 = 0.10; 4th
second: F (1, 27) = 8.94, p = .006, ηp2 = 0.25; and 5th second: F (1, 27) = 7.10, p =
.013, ηp2 = 0.21). The analyses of the activity over the corrugator supercilii data
revealed no main effect of feature size (F (1, 27) < 1), a main effect of time interval (F
(1.72, 46.45) = 4.72, p = .018; ηp2 = 0.15; Greenhouse-Geisser corrected), and no
interaction of feature size with time interval (F (1, 27) < 1). Corrugator supercilii
activity increased as a function of time interval.
Facial EMG Responses to Human Faces. The overall zygomaticus major activity was
larger for babyfaced faces (M = -0.01, SD = 0.13) than for the original faces (M = 0.08, SD = 0.24), F (1, 20) = 6.88, p = .016, ηp2 = 0.26. Neither the main effect of time
interval (F (1, 20) < 1) nor the interaction between feature size and time interval were
statistically significant (F (2.38, 47.52) = 1.01, p = .38, ηp2 = 0.05; Greenhouse-Geisser
corrected). Separate analyses for each time interval showed that the babyfaced
versions of the faces elicited a significantly larger zygomaticus major activation than
59
the original faces in three (the intermediate ones) of the five time intervals (1st second:
F (1, 20) = 1.58, p > .2, ηp2 = 0.07; 2nd second: F (1, 20) = 3.89, p = .063, ηp2 = 0.16; 3rd
second: F (1, 20) = 9.33, p = .006, ηp2 = 0.32; 4th second: F (1, 20) = 8.22, p = .010, ηp2
= 0.29; 5th second: F (1, 20) < 1). Moreover, the ANOVA revealed a main effect of
time interval on corrugator supercilii activity (F (2.44, 48.77) = 13.94, p < .001; ηp2 =
0.41, Greenhouse-Geisser corrected) but no main effect of feature size or an
interaction of feature size with time interval (for both effects F (1, 20) < 1). Corrugator
supercilii activity decreased as a function of time interval.
FIGURE 2
Average Muscle Activation Profiles of Zygomaticus Major and Corrugator
Supercilii in Response to Car Fronts and Human Faces
Cuteness block
Zygomaticus major in response to faces
0.2
mean µV change [standardized]
Zygomaticus major in response to car fronts
0.1
0.0
-0.1
-0.2
-0.3
1
2
3
4
0.2
0.1
0.0
-0.1
-0.2
-0.3
1
5
2
3
4
5
time [s]
time [s]
original car fronts/faces
babyfaced car fronts/faces
Corrugator supercilii in response to faces
Corrugator supercilii in response to car fronts
0.2
0.0
-0.2
-0.4
1
2
3
time [s]
4
5
0.2
0.0
-0.2
-0.4
1
2
3
time [s]
4
5
60
To directly contrast the zygomaticus activation differences between the babyfaced and
the original car fronts against the differences between the two feature-size conditions
found for human faces, we added object category as a between-participants factor and
conducted a 2 (feature size: original versus babyfaced) × 2 (object category: car fronts
versus human faces) ANOVA, averaged over the five seconds of stimulus exposure
(first exposure only). The analysis did not show a main effect of object category (F (1,
47) = 1.58, p > .2, ηp2 = 0.03) or an interaction effect of object category with feature
size (F (1, 47) < 1). Hence, zygomaticus activity did not differ in direction and
intensity between cars and faces. The main effect of feature size was significant (F (1,
47) = 13.54, p = .001, ηp2 = 0.22).
Habituation Effects in Facial EMG Responses (Hypothesis 2a)
To test our habituation hypothesis, we predicted that the two feature-size conditions
were different with regard to how the triggered muscle activation changed over the
two repeated exposures within the cuteness block. Thus, we analyzed whether the
facial EMG responses to the babyfaced stimuli did not habituate (decline) due to
repeated exposure, whereas responses to the original stimuli might change due to
repeated exposure. We conducted four 2 (feature size: original versus babyfaced) × 2
(repeated exposure: first exposure versus second exposure) repeated measurement
ANOVAs, separately for both muscle sites, and the two object categories. As we found
no main effect of feature size on corrugator supercilii activity when considering
responses during the first exposure block (see hypothesis 1), we tested the habituation
effects only for zygomaticus major. Per each repeated exposure condition, activity
averaged over the five seconds of stimulus presentation was included because we
found no effect of time interval on zygomaticus major activation during the first
exposure block (see also hypothesis 1). Facial EMG data for car fronts and human
faces are plotted as a function of repeated exposure in Figure 3.
Car Fronts. The ANOVA on the activity of the zygomaticus major revealed a
significant two-way interaction of feature size with repeated exposure (F (1, 27) =
4.74, p = .038, ηp2 = 0.15; see Figure 3), a main effect of feature size (F (1, 27) = 7.73,
p = .01, ηp2 = 0.22) but no main effect of repeated exposure (F (1, 27) < 1). To
interpret the interaction, we compared the facial EMG responses between the first and
second exposure, separately for the two feature-size conditions. In accordance with
hypothesis 2a, zygomaticus major responses to babyfaced car fronts did not change
significantly due to repeated exposure (MT1 = 0.07, SD = 0.23 versus MT2 = 0.01, SD =
0.14; F < 1; ηp2 = 0.03), whereas the zygomaticus major responses to the original car
61
fronts increased significantly with repeated exposure (MT1 = -0.1, SD = 0.18 versus
MT2 = 0.0, SD = 0.11; F (1, 27) = 4.36, p = .05, ηp2 = 0.14) and accounted for a slight
leveling of the activation difference between babyfaced and original car designs at the
second exposure. Thus, the effect of feature size was significant at the first (see
hypothesis 1) but not at the second exposure (F (1, 27) < 1).
Faces. The ANOVA on the activity of the zygomaticus major resulted in a significant
two-way interaction of feature size with repeated exposure (F (1, 20) = 4.73, p = .042,
ηp2 = 0.19), no main effect of feature size (F (1, 20) = 2.74, p > .1, ηp2 = 0.12), and a
significant main effect of repeated exposure (F (1, 20) = 5.17, p = .034, ηp2 = 0.21). A
comparison of the facial EMG responses between the first and second exposure,
separately for the two feature size conditions, revealed that, in accordance with
hypothesis 2a, zygomaticus major responses to babyfaced faces did not change
significantly due to repeated exposure (MT1 = 0, SD = 0.13 versus MT2 = 0.02, SD =
0.13; F (1, 20) < 1; ηp2 = 0.007), whereas the zygomaticus major responses to the
original faces increased with repeated exposure (MT1 = -0.12, SD = 0.16 versus MT2 =
0.04, SD = 0.19; F (1, 20) = 7.95, p = .01, ηp2 = 0.28) and even overran the activity
triggered by babyfaced faces at the second exposure (see Figure 3, right panel). Hence,
the simple effect of feature size was significant at the first (see hypothesis 1) but not at
the second exposure (F (1, 20) < 1).
FIGURE 3
Changes in Activation of Zygomaticus Major in Response to Car Fronts and
Human Faces over Two Stimuli Exposures
0.1
n.s.
0.0
p < .05
-0.1
-0.2
T1
Zygomaticus major in response to human faces
Zygomaticus major in response to car fronts
0.1
n.s.
0.0
p < .05
-0.1
-0.2
T2
T1
repetition over time
T2
repetition over time
original car fronts/ faces
babyfaced car fronts/ faces
Note: T1 = first exposure, T2 = second exposure; n.s. = non-significant (p > .1); values represent mean
responses to 16 stimuli averaged over 5 s presentation time.
62
Gender Differences in Facial EMG Responses (Hypothesis 2b)
To test the gender hypothesis, we analyzed whether female versus male participants
differed in their spontaneous affective responses to original versus babyfaced stimuli
in four 2 (feature size: original versus babyfaced) × 2 (time interval: seconds 1 to 5
after stimulus onset) × 2 (gender: female versus male) mixed ANOVAs, separately for
the two muscle sites and the two object categories.
Car Fronts. Of the 28 participants, 17 were female. With regard to zygomaticus major
activity, the analysis revealed that both female and male participants showed larger
affective responses to the babyfaced than to the original car fronts (main effect feature
size: F (1, 26) = 8.23, p = .008, ηp2 = 0.24; feature size × gender: F (1, 26) < 1; ηp2 =
0.007) and were not different in the overall intensity of their affective responses
triggered by the stimuli (main effect gender: F (1, 26) = 1, p > .3, ηp2 = 0.04). All the
other main and interaction effects were not significant (all F-values < 1, but feature
size × time: F (2.60, 67.48) = 2.55, p > .05, ηp2 = 0.09; Greenhouse-Geisser corrected).
With regard to corrugator supercilii, only the main effect of time interval was
significant (F (1.71, 44.47) = 3.99, p = .03, ηp2= 0.13; Greenhouse-Geisser corrected),
but none of the other main or interaction effects were (all F-values < 1, but feature size
× gender: F (1, 26) = 2.35, p > .1, ηp2 = 0.08).
Faces. Of the 21 participants, 14 were female. The ANOVA on the activity of the
zygomaticus major revealed that both female and male participants showed larger
affective responses to the babyfaced than to the original faces (main effect feature size:
F (1, 19) = 5.31, p = .03, ηp2 = 0.22; feature size × gender: F (1, 19) < 1; ηp2 = 0.006)
and were not different in the overall intensity of their affective responses (main effect
gender: F (1, 19) = 1.44, p > .2, ηp2 = 0.07). The other main and interaction effects
were not significant (all F-values < 1). With regard to corrugator supercilii activity, the
analysis revealed only a significant main effect of time interval (F (2.40, 45.59) =
12.47, p < .001, ηp2 = 0.40; Greenhouse-Geisser corrected), and all other main or
interaction effects were not significant (all F-values < 1).
Validity Check: Spontaneous Affective Responses during Attractiveness
Evaluation
To control how specific the facial EMG responses to features of a baby schema were
assessed during the cuteness evaluation block, we compared these responses with the
facial EMG responses from the attractiveness block. We explored this issue not in
terms of an additional research question but more as a validity check.
63
Manipulation Check Based on Attractiveness Ratings. We calculated the average rated
attractiveness for the original and babyfaced stimuli (first exposure only) and
compared the two means for all participants. We found that the feature-size
manipulation had neither a significant effect on the perceived attractiveness of the cars
(Moriginal = 3.70, SD = 0.46 versus Mbabyfaced = 3.61, SD = 0.49; t (23) = 1.17, p = .26,
Cohen’s d = 0.19) nor on the faces (Moriginal = 3.47, SD = 0.58 versus Mbabyfaced = 3.53,
SD = 0.71; t(22) = 0.70, p = .49, Cohen’s d = 0.08).
To check that we could not replicate the effect of feature size on positive affective
responses, we again run four 2 (feature size: original versus babyfaced) × 2 (time
interval: seconds 1 to 5 after stimulus onset) repeated measurement ANOVAs,
separately for the activation of the two muscles sites and the two object categories. In
the car group, compared to the cuteness block data, four additional participants had to
be excluded due to too many artifacts, such that the analyses were conducted on the
data of 24 participants (Mage = 22 yrs, SD = 2 yrs; 67%females). In the face group, data
of two persons had to be excluded due to too many artifacts in the attractiveness block
and inappropriate behavior during the experimental session; thus, analyses were
conducted on data of 23 participants (Mage = 21 yrs; SD = 2 yrs, 65%females). In both
object categories, the number of trials per feature-size condition varied between 5 and
16 (cars: M = 11 trials, SD = 4; faces: M = 13 trials, SD = 2).
Facial EMG Responses to Car Fronts. The ANOVA revealed no main effect of feature
size, no main effect of time interval (both F’s < 1) and no interaction of feature size
with time interval on zygomaticus major activations (F (2.41, 55.49) = 1.18, p > .3, ηp2
= 0.05; Greenhouse-Geisser corrected). Furthermore, the ANOVA revealed no main
effect of feature size (F (1, 23) = 2.13, p > .1, ηp2 = 0.09), no main effect of time
interval (F (1.83, 42.12) = 1.72, p > .1, ηp2 = 0.07; Greenhouse-Geisser corrected), and
no interaction effect between feature size and time interval (F (4, 92) = 1.61, p > .1, ηp2
= 0.07; Greenhouse-Geisser corrected) on corrugator supercilii activations.
Facial EMG Responses to Human Faces. The ANOVA revealed no main effect of
feature size (F (1, 22) < 1), a significant main effect of time interval (F (1.89, 41.58) =
7.14, p = .003, ηp2 = 0.25; Greenhouse-Geisser corrected), and no interaction between
feature size and time interval (F (1, 22) < 1) on the zygomaticus major activations.
Zygomaticus major increased as a function of time interval. Moreover, the ANOVA
revealed no main effect of feature size (F (1, 22) = 2.53, p > .1; ηp2 = 0.1), a significant
main effect of time interval (F (1.95, 42.82) = 6.92, p = .003, ηp2 = 0.24; GreenhouseGeisser corrected), and no significant interaction between feature size and time
64
interval (F (3.10, 68.28) = 2.10, p > .1, ηp2 = 0.09) on corrugator supercilii activations.
Corrugator supercilii increased as a function of time interval.
Results of the Debriefing
As facial EMG is an implicit method because it infers changes in affective responses
from changes in facial muscle activation, it is important to check whether the
participants were aware of this relationship. Hence, following our main analyses,
participants whose debriefing statements indicated that they might have had an idea of
what the purpose of the study was were eliminated, and all analyses were run again.
According to this strict criterion, seven participants in the car group and four
additional participants in the face group were excluded. For all three hypotheses, we
found the same patterns of effects as in the original analyses, although some results
failed to reach statistic significance at a 5% level due to the reduced power of this test.
Discussion and Conclusion
Discussion of Main Results
As the human mind has evolved over millions of years to enable adaptive responses to
complex environments, modern consumers’ responses to products still might be
shaped by deeply embedded perceptual, cognitive and/or affective mechanisms. In the
context of consumer emotions aroused by product design, we studied the effects of
biologically significant design features on spontaneous affective consumer responses.
We manipulated car fronts and human faces in accordance with features of the baby
schema. Assessing behavioral and facial EMG responses to these babyfaced stimuli
compared to original stimuli, for both car fronts and human faces we found effects of
our manipulation on positive affect. Babyfaced car fronts (human faces) were
perceived as cuter than the original stimuli when rated explicitly and, in line with the
behavioral data, babyfaced car fronts (human faces) elicited larger activations of the
smiling muscle, the zygomaticus major, than the original versions of the stimuli. Our
assumption that affective responses to baby-schema cues occur very quickly and
maybe automatically, which is in line with the evolutionary framework, was
supported, as differences in activation between babyfaced and original stimuli
occurred within the first second after stimulus onset for cars and within the first two
seconds for faces. No difference in corrugator activity to babyfaced cars (faces) and
the original stimuli versions was observed. Hence, the results suggest that positive
affect towards consumer products is increased due to features of the baby schema, but
65
that negative affect is not decreased; where this is in line with a two-dimensional affect
model rather than a bipolar valence continuum (see Cacioppo and Berntson, 1994, for
a discussion of the independence of positive and negative emotions; see also Cacioppo
and Gardner 1999). Moreover, there might be other, more cognitive explanations for
why we did not find any significant effects of our features-size manipulation on
corrugator supercilii because this muscle site is also responsive to cognitive load, or
fluency (Lishner, Cooter, and Zald 2008; Topolinski et al. 2009).
We further studied the universal nature of consumer affective responses to babyfaced
cars and human faces. First, we examined temporal properties and did not find changes
in affective responses to babyfaced stimuli over two exposures, which was in line with
our prediction. However, even though affective responses to the babyfaced product
designs and faces were stable, affective responses to the original stimuli increased with
repeated exposure, which was congruent with repetition effects for complex stimuli
demonstrated by other authors (e.g., Cox and Cox 2002). It was this increase of
activation triggered by the original stimuli that accounted for a slight leveling effect of
the affective responses to babyfaced and original stimuli at the second exposure.
Future studies might consider more than two exposures to study the time course of
affective responses to biological shapes in product designs more specifically (as in
Tinio and Leder 2009). Second, no effects of gender were found in facial EMG. Our
results support the gender universality of the affective responses postulated by Lorenz
(1943), such that both males and females respond positively to features of a baby
schema as in Glocker et al. (2009) and Brosch et al. (2007). However, in accordance
with other studies that reported gender differences with regard to infant-directed
behavior (Berman 1980; Maestripieri and Pelka 2002), future research might
investigate whether male and female consumers differ in their willingness to buy and
use cute products, despite the similarity of the underlying affective mechanism. In
contrast to spontaneous affect, behavior towards cute products might also be
determined by social and cultural factors (e.g., beliefs and stereotypes about
prototypical women’s and men’s products; e.g., Perkins et al. 2007). In addition to
cuteness evaluations, we also assessed attractiveness evaluations and found no effects
of our babyface manipulation on muscle activations when participants rated the car
designs (human faces) for attractiveness. Thus, the subtle babyface manipulations did
not affect the perceived overall attractiveness of either car designs or faces. Moreover,
comparing facial EMG responses from the cuteness block with results from the
attractiveness block provided a hint for the context-sensitivity of affective responses to
66
cues of a baby schema. Increased positive affect was only elicited when the detection
of a baby schema was relevant (in our case, cuteness ratings were required), which is
consistent with the assumption from evolutionary psychology that adaptive responses
are context-sensitive because they evolved to solve specific problems (Cosmides and
Tooby 1994). In our research, we focused on spontaneous affective consumer
reactions to visual key stimuli. We are aware that the automaticity of the found
affective responses can only be accepted with reservation. Although they occurred
immediately after stimulus onset, as a consequence of our study design (the cuteness
concept was “pre-activated” or framed in the participants’ mind by asking explicitly
for cuteness ratings), we do not know how facial EMG responses would have looked if
the participants had only gazed at the pictures without simultaneously completing a
rating task (e.g., Hoefel and Jacobsen 2007; Lange et al. 2003). However, we
considered it necessary to combine the facial EMG with explicit ratings because
affective responses to complex product designs can be determined by several
dimensions beside cuteness, such as aesthetic pleasure, familiarity, and novelty.
To sum up, our results not only confirmed other studies’ evidence that consumers
detect biologically significant patterns (e.g., faces) in artifacts (Erk et al. 2002;
Windhager et al. 2010; Windhager et al. 2008), but also, more importantly, that both
female and male consumers show spontaneous affective responses to product designs,
which might represent presumably innate affective responses and which were stable
over repeated exposure and were context-sensitive.
Implications for Marketing and Product Design
Product designers’ intuitions about what appeals to consumer emotions might be
correct. Our results support the idea that consumers show deeply embedded affective
responses to visual key stimuli, not only when these stimuli are present in human faces
but even when they are present in consumer product designs. Based on our findings,
the emotional value of products can be increased by making use of the knowledge that
consumer behavior is partly shaped by adaptive evolved mechanisms (e.g., by
emphasizing and maybe even exaggerating the features of visual key stimuli in product
designs, such as very large headlights). In our studies, we applied features of a baby
schema solely to car fronts, building our research on existing evidence about face-like
product designs. However, the design concept of “visual cuteness” might be applicable
to all product categories for which a deep emotional consumer-product relationship is
part of the marketing strategy. We found spontaneous affective responses to babyfaced
product designs only when cuteness ratings were required, but not when a design’s
67
attractiveness in general was evaluated. This suggests that to activate positive effects
of baby-face features, marketers either have to create a consumption situation where
the cuteness response is relevant, or cues of a baby schema have to be made even more
salient (and less subtle) than in our stimulus materials. Nonetheless, we have shown
how evolution triggers positive affective consumer responses to babies even in nonfaces and non-babies.
68
References
Aggarwal, Pankaj and Ann L. McGill (2007), “Is That Car Smiling at Me? Schema
Congruity as a Basis for Evaluating Anthropomorphized Products,” Journal of
Consumer Research, 34 (December), 468-79.
Arnheim, Rudolf (1954/1974), Art and Visual Perception: A Psychology of the
Creative Eye, Berkeley and Los Angeles: University of California Press.
Bar, Mosche and Maital Neta (2006), “Humans Prefer Curved Visual Objects,”
Psychological Science, 17 (8), 645-48.
Berlyne, Daniel E. (1970), “Novelty, Complexity, and Hedonic Value,” Perception
and Psychophysics, 8 (November), 279-86.
Berlyne, Daniel E. (1974), Studies in the New Experimental Aesthetics: Steps toward
an Objective Psychology of Aesthetic Appreciation, Washington DC:
Hemisphere.
Berman, Phyllis W. (1980), “Are Women More Responsive than Men to the Young? A
Review of Developmental and Situational Variables,” Psychological Bulletin,
88 (3), 668-95.
Bornstein, Robert F. (1989), “Exposure and Affect: Overview and Meta-Analysis of
Research, 1968-1987,” Psychological Bulletin, 106 (2), 265-89.
Brosch, Tobias, David Sander, and Klaus R. Scherer (2007), “That Baby Caught My
Eye…Attention Capture by Infant Faces,” Emotion, 7 (3), 685-89.
Buss, David (1994), The Evolution of Desire: Strategies of Human Mating, New York:
Basic Books.
Cacioppo, John T. and Gary G. Berntson (1994), “Relationship between Attitudes and
Evaluative Space: A Critical Review, with Emphasis on the Separability of
Positive and Negative Substrates,” Psychological Bulletin, 115 (3), 401-23.
Cacioppo, John T. and Wendi L. Gardner (1999), “Emotion,” Annual Review of
Psychology, 50, 191-214.
Cacioppo, John T., Richard E. Petty, Mary E. Losch, and Hai Sook Kim (1986),
“Electromyographic Activity Over Facial Muscle Regions Can Differentiate the
Valence and Intensity of Affective Reactions,” Journal of Personality and
Social Psychology, 50 (2), 260-68.
Carbon, Claus-Christian and Helmut Leder (2005), “The Repeated Evaluation
Technique (RET). A Method to Capture Dynamic Effects of Innovativeness and
Attractiveness,” Applied Cognitive Psychology, 19 (5), 587-601.
69
Carbon, Claus-Christian, Lars Michael, and Helmut Leder (2008), “Design Evaluation
by Combination of Repeated Evaluation Technique and Measurement of
Electrodermal Activity,” Research in Engineering Design, 19 (2), 143-49.
Chandler, Jesse and Norbert Schwarz (2010), “Use Does not Wear Ragged the Fabric
of Friendship: Thinking of Objects as Alive Makes People Less Willing to
Replace them,” Journal of Consumer Psychology, 20 (2), 138-45.
Colarelli, Steve M. and Joseph R. Dettman (2003), “Intuitive Evolutionary
Perspectives in Marketing Practices,” Psychology & Marketing, 20 (9), 837-65.
Cosmides, Leda and John Tooby (1994), “Origins of Domain-Specificity: The
Evolution of Functional Organization,” in ed. Lawrence A. Hirschfeld and
Susan A. Gelman, Mapping the Mind: Domain Specificity in Cognition and
Culture, New York: Cambridge University Press, 84-116.
Coss, Richard G. (2003), “The Role of Evolved Perceptual Biases in Art and Design,”
in ed. Eckard Voland and Karl Grammer, Evolutionary Aesthetics, Berlin,
Heidelberg: Springer, 69-130.
Cox, Dena S. and Anthony D. Cox (1988), “What Does Familiarity Breed?
Complexity as a Moderator of Repetition Effects in Advertisement Evaluation,”
Journal of Consumer Research, 15 (June), 111-116.
Cox, Dena S. and Anthony D. Cox (2002), “Beyond First Impressions: The Effects of
Repeated Exposure on Consumer Liking of Visually Complex and Simple
Product Designs,” Journal of the Academy of Marketing Science, 30 (2), 11930.
Creusen, Marielle E.H., Robert W. Veryzer, and Jan P.L. Schoormans (2010),
“Product Value Importance and Consumer Preferences for Visual Complexity
and Symmetry,” European Journal of Marketing, 49 (9/10), 1437-52.
Crothers, Shane, Ian Montgomery, and Robin Clarke (2003), “An Investigation into
the Role of Prototypicality in the Design of Consumer Products,” The Design
Journal, 6 (1), 52-60.
Delorme, Arnaud and Scott Makeig (2004), “EEGLAB: An Open Source Toolbox for
Analysis of Single-Trial EEG Dynamics,” Journal of Neuroscience Methods,
134 (1), 9-21.
Desmet, Pieter, Paul Hekkert, and Jan J. Jacobs (2000), “When a Car Makes you
Smile: Development and Application of an Instrument to Measure Product
Emotions,” Advances in Consumer Research, 27, 111-17.
70
Desmet, Pieter, Kees Overbeeke, and Stefan Tax (2001), “Designing Products with
Added Emotional Value: Development and Application of an Approach for
Research through Design,” The Design Journal, 4 (1), 32-47.
DiClemente, Diane F. and Donald A. Hantula (2003), “Optimal Foraging Online:
Increasing Sensitivity to Delay,” Psychology & Marketing, 20 (9), 785-809.
Dimberg, Ulf (1990), “Facial Electromyography and Emotional Reactions,”
Psychophysiology, 27 (5), 481-94.
Dimberg, Ulf and Monika Thunberg (1998), “Rapid Facial Reactions to Emotional
Facial Expressions,” Scandinavian Journal of Psychology, 39 (1), 39-45.
Dimberg, Ulf, Monika Thunberg, and Kurt Elmehed (2000), “Unconscious Facial
Reactions to Emotional Facial Expressions,” Psychological Science, 11 (1), 8689.
Durante, Kristina M., Vladas Griskevicius, Sarah E. Hill, Carin Perilloux, and Norman
P. Li (2010), “Ovulation, Female Competition, and Product Choice: Hormonal
Influences on Consumer Behavior,” Journal of Consumer Research, 37 (April),
DOI: 10.1086/656575.
Ekman, Paul (1972), “Universal and Cultural Differences in Facial Expressions of
Emotion,” in ed. James Cole, Nebraska Symposium on Motivation, Lincoln:
University of Nebraska Press, 207-83.
Ekman, Paul (1982), Emotion in the Human Face, New York: Cambridge University
Press.
Ekman, Paul, Wallace V. Friesen, and Sonia Ancoli (1980), “Facial Signs of
Emotional Experience,” Journal of Personality & Social Psychology, 39 (6),
1125-34.
Ekman, Paul, Wallace V. Friesen, and Phoebe Ellsworth (1982), “Research
Foundations,” in ed. Paul Ekman, Emotion in the Human Face, New York:
Cambridge University Press, 1-143.
Ellis, Haydn D. and Andrew W. Young (1998), “Faces in Their Social and Biological
Context”, in ed. Andrew W. Young, Face and Mind, Oxford: Oxford University
Press, 67-96.
Erk, Susanne, Manfred Spitzer, Arthur P. Wunderlich, Lars Galley, and Henrik Walter
(2002), “Cultural Objects Modulate Reward Circuitry,” NeuroReport, 13 (18),
2499-2503.
Faerber, Stella, J., Helmut Leder, Gernot Gerger, and Claus-Christian Carbon (2010),
“Priming Semantic Concepts Affects the Dynamics of Aesthetic Appreciation,”
Acta Psychologica, 135 (2), 191-200.
71
Foxall, Gordon R. and Victoria K. James (2003), “The Behavioral Ecology of Brand
Choice: How and What Do Consumers Maximize?” Psychology & Marketing,
20 (9), 811-36.
Fridlund, Alan J. and John T. Cacioppo (1986), “Guidelines for Human
Electromyographic Research,” Psychophysiology, 23 (5), 567-89.
Glocker, Melanie L., Daniel D. Langleben, Kosha Ruparel, James W. Loughead,
Ruben C. Gur, and Norbert Sachser (2009), “Baby Schema in Infant Faces
Induces Cuteness Perception and Motivation for Caretaking in Adults,”
Ethology, 115 (3), 257-63.
Glocker, Melanie L., Daniel D. Langleben, Kosha Ruparel, James W. Loughead,
Jeffrey N. Valdez, Mark D. Griffin, Norbert Sachser, and Ruben C. Gur (2009),
“Baby Schema Modulates the Brain Reward System in Nulliparous Women,”
Proceedings of the National Academy of Science, 106 (22), 9115-19.
Gorn, Gerald J., Yuwei Jiang, and Gita Venkataramani Johar (2008), “Babyfaces, Trait
Inferences, and Company Evaluations in a Public Relations Crisis,” Journal of
Consumer Research, 35 (June), 36-49.
Griskevicius, Vladas, Michelle N. Shiota, and Stephen M. Nowlis (2010), “The Many
Shades of Rose-Colored Glasses: An Evolutionary Approach to the Influence of
Different Positive Emotions,” Journal of Consumer Research, 37 (2), 238-50.
York: Oxford.
Harlow, Harry F. (1971), Learning to Love, San Francisco: Albion.
Hekkert, Paul, Dirk Snelders, and Piet C.W. Wieringen (2003), “Most Advanced, Yet
Acceptable: Typicality and Novelty as Joint Predictors of Aesthetic Preference
in Industrial Design,” British Journal of Psychology, 94 (1), 111-24.
Hildebrandt, Katherine A. and Hiram E. Fitzgerald (1978), “Facial Feature
Determinants of Perceived Infant Attractiveness,” Infant Behavior and
Development, 2 (January), 329-39.
Hoefel, Lea and Thomas Jacobsen (2007), “Electrophysiological Indices of Processing
Aesthetics: Spontaneous or Intentional Processes?” International Journal of
Psychophysiology, 65 (1), 20-31.
Perceptions of Unintentional Anthropomorphism and Zoomorphism in
Ankara, Turkey.
72
Jacob, James E., Paul Rodenhauser, and Ronald J. Markert (1987), “The Benign
Exploitation of Human Emotions: Adult Women and the Marketing of Cabbage
Patch Kids,” Journal of American Culture, 10 (Fall), 61-71.
Keating, Caroline F., David W. Randall, Timothy Kendrick, and Katherine A. Gutshall
(2003), “Do Babyfaced Adults Receive More Help? The (Cross-Cultural) Case
of the Lost Resume,” Journal of Nonverbal Behavior, 27 (2), 89-108.
Khalid, Halimahtun M. and Martin G. Helander (2006), “Customer Emotional Needs
in Product Design,” Concurrent Engineering, 14 (3), 197-206.
Lange, Kezia, Leanne M. Williams, Andrew W. Young, Edward T. Bullmore, Michael
J. Brammer, Steven C.R. Williams, Jeffrey A. Gray, and Mary L. Phillips
(2003), “Task Instructions Modulate Neural Responses to Fearful Facial
Expressions,” Biological Psychiatry, 53 (3), 226-32.
Lishner, David A., Amy B. Cooter, and David H. Zald (2008), “Rapid Emotional
Contagion and Expressive Congruence under Strong Test Conditions,” Journal
of Nonverbal Behavior, 32 (4), 225-39.
Livingston, Robert W. and Nicolas A. Pearce (2009), “The Teddy-Bear Effect: Does
Having a Baby Face Benefit Black Chief Executive Officers?” Psychological
Science, 20 (10), 1229-36.
Lorenz, Konrad (1943), “Innate Forms of Potential Experience,” Zeitschrift für
Tierpsychologie, 5, 233-519 [in German].
Maestripieri, Dario and Suzanne Pelka (2002), “Sex Differences in Interest in Infants
across the Lifespan: A Biological Adaptation for Parenting,” Human Nature, 13
(3), 327-44.
Marcus, Aaron (2002), “The Cult of Cute – The Challenge of User Experience
Design,” Interactions, 9 (November/December), 29-34.
Miesler, Linda, Jan R. Landwehr, Andreas Herrmann, and Ann L. McGill (2010),
“Consumer and Product Face-to-Face: Antecedents and Consequences of
Spontaneous Face-Schema Activation,” Advances in Consumer Research, 37,
536-37.
Nitschke, Jack B., Eric E. Nelson, Brett D. Rusch, Andrew S. Fox, Terrence R. Oakes,
and Richard J. Davidson (2004), “Orbitofrontal Cortex Tracks Positive Mood in
Mothers Viewing Pictures of Their Newborn Infants,” NeuroImage, 21 (2),
583-92.
Oehman, Arne and Joaquim J.F. Soares (1994), “Unconscious Anxiety - Phobic
Responses to Masked Stimuli,” Journal of Abnormal Psychology, 103 (2), 23140.
73
Patton, Phil (1998, October 29), “Utensils Get Cute,” The New York Times. Retrieved
September 17, 2008, from http://www.nytimes.com
Perkins, Andrew, Brad Pinter, Anthony G. Greenwald, and Mark Forehand (2007),
“Ladies and Gentlemen, Lend Me Your Attitudes…: Implicit Attitude
Formation As a Result of Group Membership and Consumption Stereotypes”,
Advances in Consumer Research, 34, 650-51.
Pittard, Narelle, Michael Ewing, and Colin Jevons (2007), “Aesthetic Theory and
Logo Design: Examining Consumer Response to Proportions across Cultures,”
International Marketing Review, 24 (4), 457-73.
Poutvaara, Panu, Henrik Jordahl, and Niclas Berggren (2009), “Faces of Politicians:
Babyfacedness Predicts Inferred Competence but not Electoral Success”,
Journal of Experimental Social Psychology, 45 (5), 1132-35.
Power, Thomas G., Katherine A. Hildebrandt, and Hiram E. Fitzgerald (1982),
“Adults’ Responses to Infants Varying in Facial Expression and Perceived
Attractiveness,” Infant Behavior and Development, 5 (1), 33-44.
Rozin, Paul (1976), “Psychological and Cultural Determinants of Food Choice,” in ed.
Trevor Silverstone, Appetite and Food Intake, Berlin: Dahlem Konferenzen,
286-312.
Saad, Gad (2008), “The Collective Amnesia of Marketing Scholars Regarding
Consumers’ Biological and Evolutionary Roots,” Marketing Theory, 8 (4), 42548.
Saad, Gad and Tripat Gill (2000), “Applications of Evolutionary Psychology in
Marketing,” Psychology & Marketing, 17 (12), 1005-34.
Saad, Gad and Tripat Gill (2003), “An Evolutionary Psychology Perspective on Gift
Giving among Young Adults,” Psychology & Marketing, 20 (9), 765-84.
Sander, David, Didier Grandjean, and Klaus R. Scherer (2005), “A Systems Approach
to Appraisal Mechanisms in Emotion”, Neural Networks, 18 (4), 317-52.
Sanefuji, Wakako, Hidehiro Ohgami, and Kazuhide Hashiya (2007), “Development of
Preference for Baby Faces Across Species in Humans (Homo Sapiens),”
Journal of Ethology, 25 (3), 249-54.
Schleidt, Margret, Wulf Schiefenhövel, Klaus Stanjek, and Renate Krell (1980),
“’Caring for a Baby’- Behavior: Reactions of Passersby to a Mother and Baby,”
Man-Environment Systems, 10 (2), 73-82.
Sherman, Gary D., Jonathan Haidt, and James A. Coan (2009), “Viewing Cute Images
Increases Behavioral Carefulness,” Emotion, 9 (2), 282-86.
74
Tinio, Pablo P.L. and Helmut Leder (2009), “Just How Stale Are Aesthetic Features?
Symmetry, Complexity and the Jaws of Massive Familiarization,” Acta
Psychologica, 130 (3), 241-50.
Topolinski, Sascha, Katja U. Likowski, Peter Weyers, and Fritz Strack (2009), “The
Face of Fluency: Semantic Coherence Automatically Elicits a Specific Pattern
of Facial Muscle Reactions,” Cognition & Emotion, 23 (2), 260-71.
Veryzer, Robert W. (1993), “Aesthetic Response and the Influence of Design
Principles on Product Preferences,” Advances in Consumer Research, 20, 22431.
Voland, Eckard and Karl Grammer (2003), Evolutionary Aesthetics, Berlin,
Heidelberg: Springer.
Wang, Yong Jian and Michael S. Minor (2008), “Validity, Reliability, and
Applicability of Psychophysiological Techniques in Marketing Research,”
Psychology & Marketing, 25 (2), 197-232.
Weyers, Peter, Andreas Mühlberger, Carolin Hefele, and Paul Pauli (2006),
“Electromyographic Responses to Static and Dynamic Avatar Emotional Facial
Expressions,” Psychophysiology, 43 (5), 450-53.
Windhager, Sonja, Florian Hutzler, Claus-Christian Carbon, Elisabeth Oberzaucher,
Kathrin Schaefer, Truls Thorstensen, Helmut Leder, and Karl Grammer (2010),
“Laying Eyes on Headlights: Eye Movements Suggest Facial Features in Cars,”
Collegium Antropologicum, 43 (3), 1075-80.
Winkielman, Piotre and John T. Cacioppo (2001), “Mind at Ease Puts a Smile on the
Face: Psychophysiological Evidence that Processing Facilitation Elicits Positive
Affect,” Journal of Personality and Social Psychology, 81 (6), 989-1000.
Yoon, Carolyn, Angela H. Gutchess, Fred Feinberg, and Thad A. Polk (2006), “A
Functional Magnetic Resonance Imaging Study of Neural Dissociations
between Brand and Person Judgments,“ Journal of Consumer Research, 33
(June), 31-40.
Zebrowitz, Leslie A. (1997), Reading Faces: Window to the Soul? Boulder, CO:
Westview Press.
Zebrowitz, Leslie A., Jean-Marc Fellous, Alain Mignault, and Carrie Andreolletti
(2003), “Trait Impressions as Overgeneralized Responses to Adaptively
75
Significant Facial Qualities: Evidence from Connectionist Modeling,”
Personality and Social Psychology Review, 7 (3), 194-215.
Zebrowitz, Leslie A. and Joann M. Montepare (1992), “Impressions of Babyfaced
Individuals across the Life Span,” Developmental Psychology, 28 (6), 1143-52.
Paper 3: Do I Need a Car or a Friend? A Scenario-Based
Investigation of the Context Dependency of Preferences for
Anthropomorphic Product Designs
Linda Miesler∗
Abstract
Past research has emphasized the need for emotions in product design experience. We
were interested in the emotional benefits provided by anthropomorphic product
designs and whether user preferences for anthropomorphic designs are universal or
vary with usage context. In line with the distinction between emotional-hedonic and
functional user needs, we proposed that users are sensitive to the emotional value a
product design communicates and therefore, users should prefer anthropomorphic
designs mainly in an emotional, but not in a functional usage context. To test the
assumption, we ran an experiment where participants were exposed to different kinds
of car choice scenarios. Based on preference rankings and ratings we found that
participants preferred anthropomorphic car designs over neutral designs in a choice
context which activated emotional user needs, and that the anthropomorphic designs
generated more positive affect than the neutral ones. Our results provide insights into
how users form product design preferences.
Introduction
Product designs do not only fulfill the immediate visual function to look aesthetically
pleasant, they may also transport non-tangible product (brand) values. For example,
Apple wants its products to be hip and modern in order to attract users who share these
values. And it is no secret that Apple’s brand values are communicated to a great
extent by its distinctive product design. In the ongoing discussion about product
success and product differentiation strategies, it is increasingly emphasized that the
more products from the same category become similar in terms of their technical
qualities, the more important it becomes to add an emotional value to a product
(Desmet, Hekkert, and Jacobs, 2000). Therefore, in particular the appearance of a
product turns out to be an important success factor (Bloch, 1995; Dumaine, 1991). But
∗
Linda Miesler, Doctoral Candidate, Center for Customer Insight, University of St. Gallen
Paper 3: Do I Need a Car or a Friend?
77
how to add such an emotional value to a product based on its design, so that users are
able to recognize the product’s emotional benefits and feel so attracted by the product
that they are willing to purchase it still remains an open question. In the present paper,
we propose anthropomorphic product forms as a means of adding an emotional value
to products. Attaching human-like attributes to products may benefit companies by
deepening user attachment to products, that is, products are not only perceived as
utility providing objects anymore - they become friends or partners the users can trust.
However, to put anthropomorphic product designs to the test and to explore the
boundaries of their application, we also asked whether is it always beneficial and
desirable for users to interact with consumer products as if they were people.
Hence, in the study presented in this paper we examined the interplay of
anthropomorphic product design features and context variables (e.g., usage context)
and how this interplay affects people’s preferences for different types of designs. To
address these issues we used a scenario-based experimental approach.
Theoretical Background
Emotional-Hedonistic User Needs
Users usually evaluate and choose products on the basis of their expectations whether
a product will meet their needs and goals (e.g., to express the social status of being hip
and modern) (Mittal, Ross, and Baldasare, 1998). One well established distinction in
consumer behavior research concerning the user needs a product has to fulfill is the
distinction between emotional-hedonic versus functional aspects of products. In their
seminal paper, Hirschman and Holbrook (1982, p. 92) defined hedonic consumption as
‘[...] those facets of consumer behavior that relate to the multi-sensory, fantasy and
emotive aspects of one’s experience with products’. In the current paper, we were
mainly interested in the emotive aspects of the user experience, that is, the users’ goal
to fulfill emotional needs (e.g., need for attachment) by interacting with a product.
Moreover, Hirschmann and Holbrook (1982) stated that according to the concept of
emotional-hedonic user experiences, the meaning of a product changes fundamentally:
a product cannot be viewed solely as an object which fulfils functional-utilitarian
goals; products are subjective symbols which convey meaning, and evoke emotions. In
the last few years, the opinion has emerged among product design researchers that it is
to a large extent the job of the product’s exterior appearance, that is of its design, to
arouse emotions in users (McDonagh-Philp and Lebbon, 2000) and, therefore, to fulfill
78
emotional user needs. Similar to Hirschmann and Holbrook, Desmet and colleagues
(Desmet, 2003; Desmet, Tax, and Overbeeke, 1999) postulated that there is no direct
relationship between the tangible features of a product design and the elicited
emotional responses; decisive for the users’ design-related emotions is the meaning the
users ascribe to a product. The meaning is supposed to result from an appraisal process
which takes the users’ goals, needs, and attitudes into account and thus, the resulting
emotions are supposed to be highly personal. In the present research, we postulated
that although product-related emotions might be individual, it might be possible to
predict product-related user emotions and, more important, to predict the resulting
choice behavior based on the elicited emotions - provided that specific context factors
are known.
Context-Dependency of Product Design Preferences
The distinction between emotional-hedonic and functional product aspects is not a
dichotomous pair of opposites, rather the two aspects can be perceived as the two end
points of a continuum. Although most products can possess both aspects (Bloch, 1995)
and users weigh both product aspects against each other to come to a trade-off (Batra
and Ahtola, 1990; Chitturi, Raghunathan, and Mahajan, 2007), mostly one of the two
aspects dominates user preferences and behaviors (Dhar and Wertenbroch, 2000).
Which aspect the users select as the dominant and more important one should depend
on the context of purchase and usage. So, consider for example a person who wants to
buy a car. Undoubtedly, in general a car can fulfill emotional-hedonic user needs (e.g.,
a pleasant look, feeling of pride) as well as utilitarian ones (e.g., be fast and efficient).
However, if the person is the owner of a grocery store and wants to buy a van to carry
goods, utilitarian goals might be weighed as much more important in a purchase
situation than how the car looks or feels like. So preferences for specific product
attributes are not stable, they are construed in the specific choice or evaluation
situation. While the point of view that people construe preferences by taking specific
characteristics of a situation into account is not new in consumer research (e.g.,
Simonson, 2008; Warlop and Ratneshwar, 1993), with regard to product design
appreciation such context dependency of design preferences is not obvious. Of course,
comparing different individuals, preferences for product design are influenced by a
multitude of factors (e.g., cultural and socio-demographic background, age, gender,
expertise; for an overview see Bloch, 1995), so that design preferences vary over
individuals (some people might adore the cute looking VW Beetle, whereas others
might dislike its toy-like design). However, one might expect that user preferences for
79
product designs are quite stable within persons (given their specific age group, gender,
subculture), that is regardless of characteristics of a given situation. To specify this
issue, we refer the assumption of stable design preferences to (1) general design tastes
or attitudes such as preference for classic styles versus modern styles, (2) preferences
for designs of products which require high user commitment (cars, electronic devices,
furniture), since preferences for low-commitment products (so called fast moving
consumer goods such as food or body care products) might depend to a large extent on
situational factors. So it was challenging to investigate whether against our assumption
of stable design preferences an individual’s preferences for product designs of highcommitment products such as cars can be also influenced by situational factors.
Emotional Product Values and Tangible Product Design Features
Given a situation where users want to predominately fulfill emotional needs by
interacting with a product, how do users find out whether they are going to select the
right product to reach their goal? The appearance of a design is an important means to
communicate information to users (c.f., product semantic approach). Thus, to infer
from the appearance of a product whether a product will fulfill their needs and wishes,
users need to be able to detect and interpret the relevant design cues which
communicate a product’s emotional and/or functional benefits. Functional aspects of a
product such as ease of use might be easy to convey via product design (e.g., by
clearly arranged buttons and very simple forms), but how can emotional values be
translated into tangible design attributes? First research attempts in consumer research
suggested that it might be possible to transfer semantic brand values such as
exclusivity into physical product design features and that users are able to read or
‘decode’ these product design features and, therefore, to experience the intended brand
value. So, Landwehr, Herrmann, Wentzel, and Labonte (upcoming) successfully
applied morphing and warping techniques to identify features in car designs (e.g.,
shape of the headlights) which users associate with values such as sportiness and
elegance. Furthermore, Desmet et al. (1999) presented a design case study where cellphones were created which aroused the intended emotional user responses by
addressing specific emotional concerns of the users. These few studies suggested that
users might be able to understand the emotional product value communicated by
tangible design features. However, if tangible designs can really transport specific
emotional values to users, it remains an open question, as to how such emotion-laden
product designs not only affect the users’ emotions, but also their actual
preference/choice behavior.
80
Anthropomorphic Product Designs
In the present study, we dealt with a specific group of tangible features of product
design. Since the study was embedded in a large scale project about people’s
perceptions of anthropomorphic product shapes (Miesler, Landwehr, Herrmann, and
McGill, 2010; Miesler and Leder, 2010), we studied preferences for face-like car
designs and whether these preferences vary with context-specific user needs.
Anthropomorphic product forms are very prevalent in product design (e.g., consider
Koziol kitchen and bath utensils, or the front end designs of cars; for an overview see
DiSalvo and Gemperle, 2003), however, just recently researchers started to investigate
these special product shapes systematically (Aggarwal and McGill, 2007; Chandler
and Schwarz, 2010). In light of the emotional-hedonic concept introduced above, as
the core assumption of the present study we supposed that product designs which can
be anthropomorphized (i.e., users can assign a human quality to the product design)
tap into emotional user needs rather than into functional ones. Thus, seeing human
qualities in a non-living object should facilitate developing an emotional relationship
and attachment to the object. So, studies have shown that anthropomorphizing
consumer products and animals generates feeling of social support (Epley, Akalis,
Waytz, and Cacioppo, 2008), and people who think about a product in human-like
terms are less willing to replace this product than people thinking about a product in
technical terms (Chandler and Schwarz, 2010). Thus, in the special case of
anthropomorphic product designs, it should be an intrinsic quality of the product
design per se to elicit positive emotions in consumers (provide an emotional benefit),
and therefore, to fulfill emotional needs.
Hypotheses
Based on the conceptual background, we assumed that users are able to assign specific
product attributes to a product contingent on the product’s tangible design features.
Further, since not only preferences for products in general, but even preferences for
specific product designs might depend on situation-specific user goals and needs
(settled on the dimension emotional-hedonic versus functional), we predicted that
those product designs should be preferred in a situation whose inferred emotional
versus functional benefits fit best with the user goals and needs relevant in this
situation. So, in a purchase or usage context which mainly addresses emotional user
needs, product designs which communicate such emotional values should be preferred.
Therefore, with regard to anthropomorphic product designs, we expected:
81
H 1: Anthropomorphic car designs are preferred over less anthropomorphic designs
in a usage context which activates emotional user needs more than functional
ones.
In addition, if our assumption was correct that people prefer anthropomorphic designs
in an emotional usage context due to the special emotional benefit provided by
anthropomorphic designs, we expected to find differences in product-related
perceptions and emotions as a function of product design.
H 2: People, who select anthropomorphically designed products as most preferred,
perceive the products as more human-like, experience more positive affect
toward the product and perceive the emotional relationship to the product as
stronger than people who select neutral designs as most preferred.
The context of usage was defined by the goals users want to fulfill in a specific
situation by using the product. To create different real life contexts which arouse more
or less emotional user needs, we applied a scenario-based approach and conducted an
experiment in which participants were exposed to different scenarios.
Methods
Methodological Approach
We applied a scenario-based experimental approach to create different real life
contexts to systematically manipulate the relevant user needs (emotional versus
functional). Using scenarios is an easy and reliable method to put someone’s
imagination into realistic everyday situations. Mostly starting with words such as
‘Imagine that you...’ the participant’s general task is to read through the narrative
description of a specific event and to imagine being the actor in the described event.
All necessary background information can be provided (such as time, place, and
personal goals). At the end of the scenario, participants are usually asked to perform a
task related to the event described in the scenario (e.g., make a hypothetical purchase
decision). The responses elicited by a scenario have high external validity, because by
imagining the event described in a scenario people remember events from their own
past experience, therefore, scenarios are able to arouse emotions and behaviors in
82
participants which are close to those experienced in real life situations, provided that
the participant is able and motivated to imagine the described event as vividly as
possible.
The study was a 2 (user needs: emotional versus functional) × 2 (product design:
anthropomorphic versus neutral) × 2 (cover story: purpose of use versus ownership)
design, whereas only the first two factors were relevant to test our hypotheses and the
third factor just represented two different operationalizations of the user needs - factor.
Each participant received one of five booklets which contained the description of one
of four different scenarios and additional rating scales (the fifth group served as
control group which was not exposed to a scenario), people were assigned randomly to
the groups. Scenarios varied with regard to the dominant user needs. So in one
scenario, emotional needs were addressed in the described situation, whereas in the
second scenario, functional needs were addressed. To guarantee that our results were
not only true in a specific situation but could be also generalized to other usage
contexts, we came up with two different cover stories. Both cover stories were built on
the same basic scenario: participants were asked to imagine that they want to choose a
car from a set of different cars, thus, they visit the car dealer where they find several
models which they take into nearer considerations. The only part which was different
between the two cover stories was the middle part, whereat for each cover story an
emotional and a functional scenario existed. The first cover story varied the experience
of using the product (purpose of use). In the scenario, the user’s expectations and
wishes which the car to choose should fulfill were either described in a very emotional
way (i.e., rather emotional needs such as car should be like a family member; be a part
of all family activities, e.g., holiday trips; like a reliable friend one can trust in every
situation) or in a cold and functional way (i.e., rather functional needs such as car has
to be fast, safe, and efficient; to transport things, e.g. after shopping; shall be worth its
money). The second cover story varied the experience of owning the product
(ownership). In the scenario either the user’s intentions to buy the car and feelings of
ownership were described (i.e., rather emotional needs such as car shall serve as status
symbol in front of neighbors; be proud of car; invest leisure time in car) or the
person’s plan to lease the car for a limited time period and advantages of leasing were
described (i.e., rather functional needs such as high flexibility, car can be turned back
after end of leasing period, be replaced by another car easily). After having read the
scenario, participants were asked to examine pictures of eight different cars which
83
were printed next to the scenario text on a separate page. Four of the cars had an
anthropomorphic design and four had a neutral design (see Figure 1). In the
anthropomorphic version of each design, the car’s headlights were replaced by
headlights which resembled human eyes and the grille had a typical mouth-like form.
In the neutral version of each design, the local front features were modified such that
they had a square form and looked technical and without any association with human
form (a pretest with a separate sample had shown that the two design conditions were
perceived as different with regard to anthropomorphic appearance). Four different
picture sequences were used to prevent order effects. First, participants were asked to
rank-order the eight different cars (‘Which car would you prefer most, second most
etc. in the described situation?’). Participants were told that all cars were equal with
regard to price and quality. Following, to study the product-related user perceptions
and emotions tackled by hypothesis 2, participants answered questions about their
most preferred car (i.e., the highest ranked car). These questions referred to the
anthropomorphic quality of the car’s appearance (three items: ‘This car appears like
being alive’, ‘This car appears like a human being’, ‘This car reminds me of a human
face’; cf., Aggarwal and McGill, 2007), general positive affect aroused by the car, that
is how pleasant the participants find the car (‘This car is pleasant’), and how much the
participants like the car (‘I like this car’), and strength of user-product relationship
(four items: ‘I would be very loyal to this car’, ‘I can trust this car’, ‘I would rely on
this car’, ‘I would feel comfortable with this car’; cf., Aaker, Fournier, and Brasel,
2004). Finally, ease of event visualization (‘I was able to visualize the described
situation’), and motivation to visualize (‘I gave my best to imagine the described
situation as well as possible’) were assessed as control variables. All answers were
given on 7-point Likert scales ranging from 1 = ‘not at all appropriate’ to 7 = ‘very
appropriate’. Last, the participants’ age, gender, and car ownership status were
retrieved.
84
FIGURE 1
Picture Stimuli
Participants
Overall, 108 participants took part in this paper-and-pencil study voluntarily
(emotional scenario condition: n = 46; functional scenario condition: n = 44; control
condition: n = 18). All participants were attendants of an executive MBA programme
at a Swiss university. Of the participants 70% were male, mean age was 34 years (SD
= 10 yrs). The majority of the participants (78.9%) were owners of a car, so most of
them were familiar with the situation of choosing a car.
Results
Data Preparation
Before analyzing the preference ranking data for testing our hypotheses, a few data
processing steps had to be performed. For each participant, the median ranks of the
four anthropomorphic and the four neutral car designs were assessed by calculating the
median value over the four ranks of a design condition (e.g., as the median is the
middle number in a sorted list of numbers, if one person assigned the ranks ‘1’, ‘2’,
‘6’, ‘8’ to the four cars with an anthropomorphic design, the median rank of this
design condition would be (2+6)/2 = 4). Please consider since the most preferred car
design had the first rank (value ‘1’), the second most preferred the second rank (value
‘2’) and so on, the smaller the value for the median rank was, the more preferred was
the design condition. Moreover, the design category (anthropomorphic versus neutral)
of the car ranked highest (the most preferred car) was recorded. All the subsequently
reported analyses were conducted with the statistical software package SPSS 18.
85
Control Variables
As the responses generated by the scenario approach depend on how well participants
were able and motivated to imagine the event described in the scenario, as a first step,
we controlled the ratings on the scales which assessed ease of visualisation, and
motivation to visualize. Averaged over both cover stories, we found that both scenario
groups were highly able and motivated to visualize the situation and that they did not
differ with regard to how easy it was for them to visualize the described situation
(Memotional scenario = 5.0 versus Mfunctional scenario = 5.0; F (1, 88) < 1, p = .88), and how
motivated they were to imagine the situation (Memotional scenario = 5.1 versus Mfunctional
scenario = 5.3; F (1, 88) < 1, p = .48).
Preference Rankings
At first, we checked the preference rankings in the control group which has not been
exposed to a scenario before completing the ranking and rating tasks. Importantly, we
did not find any preference bias toward one of the two product design conditions, that
is, in the control group the median ranks were about the same for the anthropomorphic
and the neutral designs (Mdneutral = 4.75 versus Mdanthropomorph = 4.5). Similarly, with
regard to the most preferred car (the car ranked highest), 50% of the participants in the
control group chose one of the four cars with an anthropomorphic design as favorite
car, the other 50% chose a car with a neutral design.
To test whether preferences for anthropomorphic versus neutral car designs were
influenced by the scenario a participant was exposed to (hypothesis 1), we compared
the median ranks of the anthropomorphic designs and neutral designs as a function of
the scenario conditions (i.e., were anthropomorphic car designs ranked higher, and
therefore, preferred more in the emotional scenario condition than in the functional
scenario condition and vice versa for the neutral car designs?). In Figure 2, the median
ranks of the two design categories, separately for the two scenario groups (averaged
over both cover stories) and the control group, are depicted. The median rank of the
anthropomorphic designs was Md = 3 in the emotional scenario and Md = 6.5 in the
functional scenario condition, whereas the median rank of the neutral designs was Md
= 6.25 in the emotional scenario and Md = 3.0 in the functional scenario condition. To
test whether the differences between the median ranks were statistically significant, we
applied the Mann-Whitney U test which is a non-parametric statistically procedure
usually used with ordinal data (e.g., rank orders) to test whether the distributions of
86
two independent samples are equal or different1. As expected in the first hypothesis,
we found that the anthropomorphic designs were ranked significantly higher (preferred
more) in the emotional scenario condition than in the functional scenario condition (U
= 1942.5, z = 7.63, p < .001), and accordingly, the neutral designs were ranked
significantly higher in the functional scenario condition than in the emotional scenario
condition (U = 78.50, z = -7.67, p < .001). (The results were the same when we
compared the design conditions’ median ranks between the two scenario conditions
separately for the two cover stories.)
Analyzing the design categories of the cars selected as most preferred (highest rank)
across all participants as a function of scenario revealed, that in both scenarios most of
the participants made a choice consistent with the first hypothesis, so that 89% of the
participants selected a car with anthropomorphic design features as most preferred in
the emotional scenario condition, and 80% selected a neutral car design as most
preferred in the functional scenario condition. Correspondingly, only a small subgroup
of participants made choices which were inconsistent with the first hypothesis (11%,
i.e., five participants, preferred a car from the neutral design condition in the emotional
scenario condition; 20%, i.e., nine participants, preferred a car from the
anthropomorphic design condition in the functional scenario condition).
Hence, both the preference rankings and the design of the most preferred car showed
that the participants’ design preferences were sensitive to the context described in the
scenario, so that the first hypothesis was supported.
1
see, for example, Corder and Foreman (2009) for a plain description (p. 57-78)
87
FIGURE 2
Preferences for Anthropomorphic and Neutral Car Designs as a Function of
Experimental Condition
Product-Related Perceptions and Emotions
To test the assumptions made in the second hypothesis, we conducted several analyses
of variance (ANOVAs) with product design of the most preferred car
(anthropomorphic versus neutral) as independent variable and the average ratings for
perceived anthropomorphism, liking, pleasantness, and perceived strength of userproduct relationship as dependent variables. In simple terms, an ANOVA tests whether
the means of several groups are equal or not to reveal the effect of one or more
treatments (e.g., the effect of product design).2 Again, we averaged the data over the
two cover stories. Since we intended to explore whether the preference for
anthropomorphic design features in an emotional context was due to their specific
emotional benefit (e.g., product can be seen as friend or partner due to human-like
features), we only considered participants who made choices consistent with the
assumption in the first hypothesis, i.e., who preferred anthropomorphic designs in the
emotional context, and neutral designs in the functional context, respectively. Data of
2
see, for example, Tabachnik and Fidell (2007, p. 37-53)
88
participants who made an inconsistent choice in the preference ranking task were
excluded from the analyses (n = 14; see section Preference rankings).
Perceived Degree of Anthropomorphism: At first, we tested whether anthropomorphic
car designs were perceived as more human-like than neutral car designs. For the
analysis, we calculated the scale mean over the three anthropomorphism items
(Cronbach’s α = 0.98) for each participant. In line with the second hypothesis,
participants selecting an anthropomorphic car as most preferred also perceived the car
to have stronger human-like qualities (M = 5.8) than the participants selecting a neutral
car design (M = 2), F (1, 74) = 398.32, p < .001.
Liking and Pleasantness. To test the perceived emotional value of the
anthropomorphic and the neutral car designs, we further studied as how pleasant and
likeable the most preferred car was evaluated. In line with the second hypothesis, the
anthropomorphic designs were rated on average as more pleasant (Mpleasant = 5.3) and
more likeable (Mlikeable = 4.4) than the neutral car designs (Mpleasant = 4.7, and Mlikeable =
3.9, respectively). These differences were statistically significant on a 5%- or on a
10%-level (F (1, 74) = 5.94, p = .02, and F (1, 54) = 2.89, p = .095, respectively).
Perceived User-Product Relationship. To further test the perceived emotional value of
the two car design categories, we studied the strength of the user-product relationship
for the two design categories. Calculating Cronbach’s alpha to assess the reliability of
the initial four item-scale revealed that the first item was not consistent with the other
three items, therefore, for further analysis, we omitted the first item and calculated a
scale mean only over the other three items (Cronbach’s alpha of items 2-4: α = 0.88).
Against our expectation, we found no difference between the two design categories,
that is participants perceived the strength of the (hypothetical) user-product
relationship as equally high for the anthropomorphic and the neutral car designs
(Manthropomorph = 5.8 versus Mneutral = 5.8; F (1, 74) < 1, p = .93).3 Therefore, the last
assumption of the second hypothesis was not supported.
3
As participants in both design categories rated the perceived relationship strength very highly (average value of
almost ‘6’ on a 7-point scale), it might be due to a ceiling effect that the ratings did not differentiate between the
average relationship strength ratings in the two design categories. As ceiling effects (i.e., left skewed
distribution) are an indication of non-normality, so that the normality precondition for the ANOVA is violated,
we also conducted a non-parametric test (Mann-Whitney U - test) which has less strict preconditions. However,
this test also revealed that both groups were not different with regard to the strength of the user-product
relationship (p > .1).
89
Summary and General Discussion
The first hypothesis was supported. We found that participants’ design preferences
were sensitive to the usage context, so that they preferred anthropomorphic (neutral)
car designs when (functional) emotional user needs were dominant in the described
context. Hence, we extended existing evidence (e.g., Desmet et al., 1999) and showed
that emotion-laden product designs not only influence user emotions, but they can also
affect product preferences/ choices. The design preferences examined in our study
were hypothetical, future studies could investigate real choice behavior. The second
hypothesis was only partly supported. Consistent with our expectations, participants
who preferred a car with anthropomorphic design features also perceived this car to be
more human-like than participants who preferred a neutral car design. This suggested
that it was really the car’s anthropomorphic appearance that the participants based
their preference formation on. Moreover, in line with our assumption that
anthropomorphic designs should be associated with a high emotional benefit, we found
that participants felt higher positive affect toward the anthropomorphic car designs
than toward the neutral ones. This provided evidence for the intrinsic emotional
quality of anthropomorphic product designs. Although most of the participants
selected the car designs which were supposed to match the user needs specified by the
given context (i.e., anthropomorphic designs fit the emotional needs as aroused by the
emotional scenario, and neutral designs fit the functional needs as aroused by the
functional scenario), anthropomorphic designs evoked stronger emotional responses
than neutral designs, which cannot be only explained by ‘cold’ and appraisal-based
emotions that are aroused by a fit between the users’ concerns and the product design
attributes (e.g., Desmet, 2003). Finally, the strength of user-product relationship was
not affected by the product design as expected, which was a surprising finding. One
explanation for this result could be a task effect. Both the emotional and the functional
scenarios required their participants to imagine the purchase situation and to deliberate
about the car to choose. The mere imagination of the car might have made it more
vivid and therefore attachment to the car might have increased independent of the
scenario’s content and the product design.
All in all, although users might be emotionally susceptible to anthropomorphic product
designs, they might not always prefer them. Therefore, decision makers in product
design should bear in mind that product design preferences are not universal.
Depending on the context of purchase and usage, the same design features might be
beneficial to product success in one situation, but detrimental in another.
90
References
Aaker, J., Fournier, S., & Brasel, S.A. (2004). ‘When good brands do bad.’ Journal of
Consumer Research, 31, 1-16.
Aggarwal, P. & McGill, A.L. (2007). ‘Is that car smiling at me? Schema congruity as a
basis for evaluating anthropomorphized products.’ Journal of Consumer
Research, 34, 468-479.
Batra, R. & Ahtola, O.T. (1990). ‘Measuring the hedonic and utilitarian sources of
consumer attitudes.’ Marketing Letters, 2, 159-170.
Bloch, P. H. (1995). ‘Seeking the ideal form – Product design and consumer response.’
Journal of Marketing, 59, 16-29.
Chandler, J. & Schwarz, N. (2010). ‘Use does not wear ragged the fabric of friendship:
Thinking of objects as alive makes people less willing to replace them.’ Journal
of Consumer Psychology, 20, 138-145.
Chitturi, R. Raghunathan, R. & Mahajan, V. (2007). ‘Delight by design: The role of
hedonic versus utilitarian benefits.’ Journal of Marketing, 72, 48-63.
Corder, G.W. & Foreman, D.I. (2009). Nonparametric Statistics for Non-Statisticians:
A Step-by-Step Approach. Hoboken, New Jersey: John Wiley & Sons.
Desmet, P.M.A. (2003). ‘A multilayered model of product emotions.’ The Design
Journal, 6, 4-13.
Desmet, P.M.A., Hekkert, P. & Jacobs, J.J. (2000). ‘When a car makes you smile:
Development and application of an instrument to measure product emotions.’
Advances in Consumer Research, 27, 111-117.
Desmet, P.M.A., Tax, S. & Overbeeke, K. (1999). ‘Designing products with added
emotional value: Development and application of an approach for research
through design.’ The Design Journal, 4, 32-47.
Dhar, R. & Wertenbroch, K. (2000). ‘Consumer choice between hedonic and
utilitarian goods.’ Journal of Marketing Research, 37, 60-71.
DiSalvo, C. & Gemperle, F. (2003). ‘From seduction to fulfillment: The use of
anthropomorphic form in design.’ Proceedings of the 2003 International
Conference on Designing Pleasurable Products and Interfaces (pp. 67-72).
Pittsburgh, PA: ACM Press.
Dumaine, B. (1991). ‘Design that sells and sells and…’ Fortune, 11, 86-94.
Epley, N, Akalis, S., Waytz, A. & Cacioppo, J.T. (2008). ‘Creating social connection
through inferential reproduction.’ Psychological Science, 19, 114-120.
91
Landwehr, J., Herrmann, A., Wentzel, D., & Labonte, C. (in press). ‘Verankerung von
Markenwerten im Produktdesign.’ Zeitschrift für betriebswirtschaftliche
Forschung.
McDonagh-Philp, D. & Lebbon, C. (2000). ‘The emotional domain in product design.’
The Design Journal, 3, 31-42.
Miesler, L., Landwehr, J. R., Herrmann, A. & McGill, A. L. (2010). ‘Consumer and
product face-to-face: Antecedents and consequences of spontaneous faceschema activation.’ Advances in Consumer Research, 37, 536-537.
Miesler, L., & Leder, H. (2010). ‘The cute look: Baby-schema effects in product
design.’ Design & Emotion Conference 2010, Chicago.
Mittal, V., Ross Jr., W.T. & Baldasare, P.M. (1998). ‘The asymmetric impact of
negative and positive attribute-level performance on overall satisfaction and
repurchase intentions.’ Journal of Marketing, 62, 33-47.
Simmonson, I. (2008). ‘Will I like a medium pillow? Another look at constructed and
inherent preferences.’ Journal of Consumer Psychology, 18, 155-169.
Tabachnick, B.G. & Fidell, L.S. (2007). Using multivariate statistics. Boston: Pearson
Education, Inc./ Allyn and Bacon.
Warlop, L., & Ratneshwar, S. (1993). ‘The role of usage context in consumer choice:
A problem solving perspective.’ Advances in Consumer Research, 20, 377-382.
Paper 4: Consumer and Product Face-to-Face: Antecedents
and Consequences of Spontaneous Face-Schema Activation
Linda Miesler(1), Jan R. Landwehr(2), Andreas Herrmann
(3)
, Ann L. McGill(4)∗
Abstract
In practice, designers sometimes give products a human-like appearance in the hope of
increasing liking due to anthropomorphizing. It remains an open research question,
however, whether the mere morphological shape of a product's design is sufficient to
activate a human schema. To investigate the spontaneous associations that are elicited
by a product's shape, we ran a lexical decision task contrasting human faces, car fronts
(which may resemble faces), and car sides. We examined further the effects of
anthropomorphizing on explicit product evaluations. Our results support
anthropomorphizing as an automatic process that affects explicit judgments but also
reveal a moderating factor.
Extended Abstract
The concept of anthropomorphism is gaining in popularity in marketing and product
design. Particularly in automotive design, the trend to develop cars whose fronts look
like the human face is increasing (e.g., VW Beetle, Mini). But, in striving for product
success, whether the mere morphological shape of a product's design is sufficient to
activate a human schema is, as yet, an unanswered question. In the context of
marketing-mix activities, what specific contribution can anthropomorphic product
design make to developing a product’s personality? To answer these questions,
evidence on the psychological process which underlies anthropomorphizing is needed.
Aggarwal and McGill (2007) recently proposed the schema-congruity theory to
explain how anthropomorphism works, but their experimental approach left it open if
consumers anthropomorphize products spontaneously when they see a human-like
product (i.e., according to an automatic bottom-up process) or whether it has to be
triggered externally.
∗ (1) Linda Miesler, Doctoral Candidate, Center for Customer Insight, University of St. Gallen; (2) Jan R.
Landwehr, Assistant Professor of Marketing, Center for Customer Insight, University of St. Gallen; (3) Andreas
Herrmann, Professor of Marketing, Center for Customer Insight, University of St. Gallen; (4) Ann L. McGill,
Sears Roebuck Professor of General Management, Marketing and Behavioral Science, Booth School of
Business, University of Chicago.
Paper 4: Consumer and Product Face-to-Face
93
In our project, we were particularly interested in the tendency to anthropomorphize
products due to their similarity to a human face. We chose real cars as objects of
investigation and compared car fronts with car sides, since we assumed that the design
of car fronts should resemble a human face, whereas the design of car sides should
obviously not. To gain deeper insights into the cognitive mechanisms, we investigated
in study 1 whether the activation of a face schema in memory is an automatic, featuredriven process leading to anthropomorphizing car fronts but not car sides and, in study
2, we examined the effects of anthropomorphizing on explicit product evaluations.
Study 1: To investigate whether a car might be associated spontaneously with a human
face solely due to its design features we settled on a lexical decision task (LDT) which
was performed by 165 native German speakers (Mage = 36; SDage = 11 ; 56% male).
The participants’ task was to categorize a target stimulus as a word versus a non-word
ignoring a preceding picture prime. For all participants, words stemmed from two
categories (face-words vs. car-words). With regard to the preceding picture primes,
participants were randomly assigned to one of three priming conditions, so that they
were primed either with pictures of cars presented in front view, cars presented in side
view, or faces. In total, every participant completed 36 trials. Latency for participants’
lexical decisions was recorded in milliseconds for each trial.
If the mere product design really accounts for the activation of a human schema,
different response latencies to face- and car-words between the three priming
conditions should occur. Firstly, we hypothesized that participants who were primed
with car fronts should respond faster to face-words than to car-words, whereas
participants who were primed with car sides should respond faster to car-words than to
face-words. Second, average response patterns in the face condition should be similar
to response patterns in the car front condition, but different from response patterns in
the car side condition.
As expected, the priming condition (car front, car side, face) interacted significantly
with the word target’s category (face vs. car) (F (2,162) = 3.28, p = .040). More
precisely, participants who were primed with car fronts responded significantly faster
to face-words than to car-words (t(52) = 1.69, p = .050), whereas participants who
were primed with car sides responded significantly faster to car-words than to facewords (t(46) = -1.87, p = .030). Further between-group comparisons showed that
latency patterns did not differ between the car front and face condition (t(116) = 0.44 ,
p = .660), but both were significantly different from the car side condition (face vs.
94
side: t(110) = -0.20, p = .044; front vs. side: t(98) = 2.52, p = .013) which was also
congruent with our expectations.
Study 2: The same participants who performed the LDT were asked to rate pictures of
cars on different scales (the pictures were identical to the pictures which were used in
the LDT as primes). One group of participants rated cars shown in front view, the
other group rated cars shown in side view. The scales assessed general marketing
variables (e.g., liking, willingness to pay) and specific evaluative tendencies which
should go together with anthropomorphizing (e.g., attribution of a human personality
to the car). Furthermore, we also assessed the participants’ personal disposition to
anthropomorphize cars.
We assumed as explicit evaluative consequences of automatic face-schema activation
and, therefore, anthropomorphizing that participants should rate car fronts higher than
car sides on the anthropomorphism scales and, therefore, car fronts are maybe also
evaluated better than car sides with regard to the marketing variables. As ratings on the
anthropomorphism-related scales were highly correlated (r = .74, p < .001), we created
a composite variable, named the anthropomorphism score. Overall, to our surprise,
participants did not rate car fronts significantly higher on the anthropomorphism score
than car sides (F(1, 82) = 2.45, p = .122). However, when controlling for a
participant’s personal disposition to anthropomorphize, we found the expected main
effect of the car view on the anthropomorphism score with fronts being more
anthropomorphized than sides (F(5, 71) = 8.54, p = .005). Likewise, participants were
willing to pay more for cars seen in front vs. side views (F(1, 71) = 4.55, p = .036) and
general positive affect elicited by cars was higher for front than side views (F(1, 71) =
10.65, p = .002), when controlling for personal disposition in both analyses.
General Discussion. Our results support the assumption that consumers
anthropomorphize products spontaneously solely due to anthropomorphic design
features and provide insights into the relative priority of anthropomorphic thoughts.
Specifically, findings show that car fronts not only bring to mind the human schema
more so than car sides bring forth this schema, they bring to mind the human schema
more than the actual product category schema, a remarkable effect. Furthermore,
spontaneous anthropomorphizing also seems to affect explicit product evaluations
positively under certain conditions, for example, depending on personal variables. Our
pattern of results suggests that future research is needed to examine the interplay and
relative influence of implicit and explicit anthropomorphic thoughts on consumer
behavior.
95
Consumer and Product Face-to-Face: Antecedents and Consequences
of Spontaneous Face-Schema Activation
Consumers sometimes interact with products as if they were people. They see brands
as possessing personality traits like those of people (Aaker 1997; Callcott and Phillips
1996). They form relationships with brands (Fournier 1998) and evaluate brand actions
according to norms of human relationships (Aggarwal 2004). Anthropomorphizing
brands may benefit marketers by deepening consumer attachment to brands, moving
them from utility providing vehicles to friends or partners and so engendering loyalty,
trust, and resilience in the relationship. The effect of simply seeing a product as a
person, even one who is not tied to the consumer through a deeper relationship, may in
itself affect the evaluation of a product (Aggarwal and McGill 2007). The question
then becomes how consumers come to see products in human terms. While the whole
of the marketing mix may contribute to this perception, the present research focuses on
the role of the product form on anthropomorphizing. Specifically, we examine whether
physical characteristics of a product lead to spontaneous anthropomorphizing (in
contrast to marketer-primed anthropomorphizing, cf. Aggarwal and McGill 2007) and
whether these characteristics in turn affect product evaluations.
In the following section, we first review the current status of research dealing with the
consumers’ tendency to anthropomorphize products, develop our research questions,
and then describe two empirical studies. In the literature review, we mainly focus on
evidence for one special product category, cars, because the front ends of these
products are commonly described in design practice and the business press as having
human facial features (e.g., Welsh 2006). In addition, these products have been
investigated in earlier aforementioned consumer research on marketers’ efforts to
anthropomorphize products making them especially informative in examining
consumers’ spontaneous anthropomorphism. However, while we used cars as objects
of investigation in our studies, our intent is to extend our conclusions to other product
categories.
Theoretical Background
Product Design and Anthropomorphism. Designers often create products with humanlike forms, for example, the silhouettes of many bottles are reminiscent of the human
body (DiSalvo and Gemperle 2003). Another very popular example for human-like
forms in product design is cars because the fronts of these products may look like
faces, a perception designers sometimes try to enhance (Welsh 2006). These efforts are
96
supported by research in psychology which has found that people have an innate or
very early acquired preference for face stimuli (Mondloch et al. 1999). Faces are
highly salient, being processed automatically and very quickly. This preference seems
to go so far that humans even see faces in the inanimate environment. As an
illustrative example, one could look at the numerous websites dealing with face-like
objects (e.g., http://facesinplaces.blogspot. com/) (see Fig. 1).
Perceiving human (e.g., face-like) forms and configurations in objects can be
considered the basic form of anthropomorphism since the term is derived from the
Greek expressions anthropos (“human”) and morphe (“shape” or “form”) (Epley,
Waytz, and Cacioppo 2007). More generally, the concept of anthropomorphism entails
the attribution of human-like characteristics, behaviors, feelings, or intentions to nonhuman agents which somehow look or behave in human-like fashion (Epley et al.
2007; Guthrie 1993). Thus, “anthropomorphism” describes a human behavior, the term
“anthropomorphic” relates to a product quality which should elicit the human
behavior.
FIGURE 1
Objects with Faces
source: http://facesinplaces.blogspot. com/
Cognitive Processes Underlying Anthropomorphizing. In psychology and marketing
literature, research on when and how people anthropomorphize design products like
cars is rare. Aggarwal and McGill (2007) were among the first who proposed a
cognitive theory to explain the psychological mechanism of anthropomorphism. They
claimed that a product has an anthropomorphic quality if it activates a human schema
(e.g., a face schema) in the consumer’s long-term memory. The precondition for the
activation of a human schema is a fit (congruence) between the features of the schema
stored in the memory and the object’s features (Mandler 1982). In addition, for the
special case of face-like objects, results from neuropsychological studies have
suggested that cars, whose fronts are assumed to look like a face, are processed
97
similarly to human faces. In particular, Gauthier et al. (2000) found that pictures of
cars activate the fusiform face area in the brain, that is, the area activated by human
faces. However, the evidence from neuropsychological studies like this one by
Gauthier et al. is not unambiguously interpretable because it is not clear whether the
(face-like) morphological configuration of the car per se led to the activation of the
fusiform face area or whether other, more abstract factors accounted for the brain
activation pattern found (e.g., level of familiarity with the category). Windhager et al.
(2008) also dealt with the question whether humans process car fronts like faces. They
found some indirect evidence that there might be a tendency in human information
processing to anthropomorphize cars but their methodology (explicit ratings on Likert
scales) did not allow conclusions on the underlying cognitive process to be drawn.
Process Characteristics of Anthropomorphizing. As mentioned before, Aggarwal and
McGill (2007) recently proposed the schema-congruity theory as a possible
explanation for the process of how people anthropomorphize products. However, as
they pre-activated a human schema in their study by means of experimental
manipulation, their experimental approach left it open whether we anthropomorphize
objects spontaneously when we see a human-like product (i.e., according to an
automatic bottom-up process where we match the product’s features with a human
schema) or whether we see the product as an object in the first place and its
anthropomorphization has to be triggered with the help of external means as in
Aggarwal and McGill’s study. In general, up to now, the question whether we
anthropomorphize products automatically has not yet been investigated empirically to
our knowledge. In psychological literature on anthropomorphism, there are different
positions represented with regard to the issue of how people anthropomorphize. The
majority of authors suggest that people anthropomorphize objects mostly without
conscious processing (Caporael 1986; Chartrand, Fitzsimons, and Fitzsimons 2008;
Epley et al. 2007; Ingram and Annable 2004). Bush (1990), for example, considered
anthropomorphizing products as a basic way of perceiving the environment and he
postulated that people would always see some human features in a product. However,
in contrast to the assumption of spontaneous anthropomorphizing, some authors
discuss that people can suppress the automatic process through a deliberate control
mechanism which has developed during cognitive development (Epley et al. 2007) or
that anthropomorphism might be rather modified by social context than by innate
processes (Kiesler et al. 2008).
98
Behavioral Consequences of Anthropomorphizing. Marketers and designers utilize the
consumers’ assumed tendency to anthropomorphize to make their products more
appealing (DiSalvo and Gemperle 2003; Welsh 2006). So, for marketing efforts it is
necessary that anthropomorphizing is more than an (automatic) mental process which
goes on in the consumers’ minds but that it has consequences on consumer evaluations
and behavior in real life, too. Aggarwal and McGill (2007) could show that
anthropomorphism indeed has a positive impact on product evaluation so that
anthropomorphized products are more likeable than products which do not possess this
quality (see also Ingram and Annable 2004). They explained that a product’s similarity
with a human schema has a positive impact on product evaluations because a
successful match between a schema and the product elicits a positive affective
response (Meyers-Levy and Tybout 1989). However, besides a rather unspecific
positive affective response due to schema-congruity, more specific positive effects of
anthropomorphizing could be postulated. If people possess an automatic tendency (or
even an innate "need," as some authors suggest, Caporael 1986) to anthropomorphize
the objects in their environment (e.g., to gain a feeling of mastery or to increase their
sense of wellbeing), there should be some additional benefits of anthropomorphizing
which might be observable in explicit judgments and behaviors. Therefore, some
authors suggest that the phenomenon of anthropomorphism also entails inferences
about the non-human object, for example concerning its intentions (Epley et al. 2007).
In the special case of an object’s similarity with a human face, these inferences might
relate to the fact that the object is perceived and, therefore, treated like a human on
other dimensions beside mere design qualities. So, one could assume that consumers
tend to attribute human-like personality characteristics to products whose
configurations resemble a face (as found for pets, e.g., Kwan, Gosling, and Oliver
2008), that consumers build stronger relationships with human-like products (e.g.,
loyalty, commitment), and that they even apply interaction rules in the consumerproduct relationship which are typical of interpersonal relationships (Aggarwal 2004;
Fournier 1998). Last but not least, one could also expect that affective reactions to
human-like products are more intensive than those to non human-like ones.
Overview of the Studies. To gain deeper insights into the cognitive process of
anthropomorphizing products in general and cars in particular, we were interested in
two questions. First, we investigated to what extent car fronts compared against car
sides spontaneously activate a face schema in long-term memory. Is it merely the
design of a car (i.e., the similarity of the car to a human face) which accounts for
99
anthropomorphizing according to a bottom-up feature-matching process? To do so, in
study 1, we applied a lexical decision task picking up on the schema-congruity theory
proposed by Aggarwal and McGill (2007) as theoretical background for our
experimental approach. We expected that car fronts should be associated with a human
face, whereas car sides which obviously do not possess face-like design features
should not if the mere design features of cars or products in general are sufficient for
anthropomorphizing. Second, we were interested in the effects of (spontaneous)
anthropomorphizing on explicit judgments which should go together with the tendency
to anthropomorphize (e.g., the attribution of human-like personality traits to a car) and
general product evaluations (e.g., liking). Does the tendency to anthropomorphize cars
which is assumed to be spontaneous (and maybe not conscious) affect explicit
judgments and behaviors? To assess explicit anthropomorphism-related behavior and
the consequences of anthropomorphizing, respectively, we asked people in study 2 for
explicit judgments. We expected that if car fronts activate a face schema, they also
activate specific personality characteristics associated with humans, produce feelings
of commitment and strong attachment to the car, and are better liked than car sides.
Study 1: Implicit Measurement of Anthropomorphizing
To investigate whether a face schema might be activated spontaneously seeing a car,
we settled on a web-based lexical decision task (LDT) using pictures of cars, presented
in front or side view, or faces as primes and words related to the concepts of face or
car and non-words, respectively, as targets.
Method and Procedure. One-hundred sixty-five native German speakers took part in
the online experiment in December 2008 (Mage = 36; SDage = 11; 56% male). The
participants’ task was to ignore the preceding picture prime and to categorize the target
stimulus, a letter string, as a word versus a non-word by pressing a key. (We address
issues of experimental demand and conscious awareness of the relationship between
the primes and letter strings in our debriefing for the present study and the pattern of
results for the second study.) For all participants, words stemmed from two word
categories. The category “face-related words” consisted of nine German words which
were associated most frequently in a pretest (n = 82) with the concept human face, the
category “car-related words” consisted of nine German words which were associated
most frequently with the concept car in the same pretest. The two word lists did not
differ with regard to mean word length and mean word occurrence frequency. We also
pretested with a separate sample of participants (n = 10) that the selected words could
be unambiguously assigned to one of the two word categories. Additionally, 18
100
nonwords were formed by changing at least two letters in the car-related and facerelated words, so that the non-words were still pronounceable, but not similar in
orthography to any existing word from the German language. With regard to the
preceding picture primes, participants were randomly assigned to one of three priming
conditions, so that they were primed either with pictures of cars shown in front view,
cars shown in side view, or faces. The car models employed all belonged to the
compact car segment (e.g., BMW Mini, Peugeot 207) and were among the 20 most
frequently sold compact cars in Germany in 2007. The face group served as a
reference group to facilitate the subsequent interpretation of the resulting response
patterns. In total, every participant responded to nine picture/car word pairs, nine
picture/face word pairs, and 18 picture/non-word pairs - every picture was shown four
times. Participants in the same priming condition responded to the same prime-target
pairs, but the pairs’ sequence was randomized across participants. Before the test
block, the participants completed a short practice block with neutral word and picture
material to familiarize themselves with the procedure. The latency for participants’
responses whether the target stimulus was a word or a non-word was recorded in
milliseconds for each trial.
Specific Hypothesis. The rationale behind the chosen methodical paradigm was as
follows. We assumed that cars shown in front view possess features like a human face,
whereas cars shown in side view apparently do not possess this face-like
configuration. In accordance with the spreading activation model from knowledge
representation (Anderson 1983; Collins and Loftus 1975; McNamara 1994) and
several studies which have already proven that even pictures can serve as effective
primes in a LDT (Bajo and Canas 1989; Irvin and Lupker 1983; Kroll and Potter 1984;
Theios and Amrhein 1989; Vanderwart 1984), we expected that the picture prime
shown would activate the appropriate concept in the participants’ semantic network
(e.g., “face”) and that the activation would spread along to related concepts (e.g.,
“eyes”), thereby facilitating the subsequent processing of those concepts. So for the
LDT, we postulated significantly shorter mean latencies when responding to facewords compared to responses to car-words as indication of spontaneous face-schema
activation in memory elicited by the prime picture. Conversely, we expected shorter
mean latencies for responses to car-words than to face-words when a car schema has
been spontaneously activated in the memory by the prime picture. Therefore, we
hypothesized firstly that if car fronts lead to spontaneous face-schema activation due to
their feature-based similarity with human faces and, therefore, are anthropomorphized
101
spontaneously, participants who were primed with car fronts should respond faster to
face-words than to car-words, whereas participants who were primed with car sides
should respond faster to car-words than to face-words. Second, average response
patterns in the face condition should be similar to response patterns in the car front
condition, but different from response patterns in the car side condition, if car fronts
but not car sides are processed similarly to faces.
Results. Only correct responses from word trials made within more than 250 ms but
less than 1450 ms were included in the data analysis (Ntrials = 2839). Incorrect
classifications of the letter strings were infrequent (5.4% error rate across all word
trials), as were response latency outliers (1.0%). The participants’ answers in a funnel
debriefing (Bargh and Chartrand 2000) carried out immediately after the LDT ensured
that the participants were not aware of the study’s goal, that is, they did not realize any
relationship between the prime pictures and the word targets when performing the
LDT.
We submitted the participants’ response latencies to a 3 (priming condition: car front
vs. car side vs. face) × 2 (word category: face vs. car) between-within-subjects
ANOVA with repeated measures on the second factor. To exclude effects due to
specific aspects of the stimulus material, we used multiple operationalizations (9) of
the primes in each category (i.e., nine different car fronts; nine different car side
views; nine different human faces). To eliminate the unsystematic variance produced
by this methodological manipulation, we z-transformed response latencies within each
of the 27 individual stimuli prior to the subsequent analyses.
As expected, the prime category (car front vs. car side vs. face) interacted significantly
with the word target’s category (face vs. car) with regard to mean response latencies.
Participants who saw car fronts or faces as primes in the LDT showed a different
response latency pattern compared to participants who saw car sides as primes
(F(2,162) = 3.28, p = .040) (Fig. 2). More precisely, participants who were primed
with car fronts in the LDT responded significantly faster to face-words than to carwords (t(52) = 1.69, p = .050), suggesting that the human schema was more salient in
processing these stimuli than the underlying product (car) schema. Conversely,
participants who were primed with car sides reacted significantly faster to car-words
than to face-words (t(46) = -1.87, p = .030). Comparing mean response latency
patterns in the two car conditions with the mean response latency pattern in the face
condition, latency patterns did not differ between front and face condition (t(116) =
0.44 , p = .660) but both were significantly different from the side condition (face vs.
102
side: t(110) = -0.20, p = .044; front vs. side: t(98) = 2.52, p = .013) which was also
congruent with our expectations.
To sum up, as an indication of automatic anthropomorphizing, we found that
participants responded faster to face-related words than to car-related words, but only
when they saw car fronts as primes. Participants in the car front condition also showed
a similar response latency pattern to participants in the face condition which might
suggest similar processing of faces and car fronts since car fronts are assumed to
possess face-like feature configurations. To gain further convergent evidence for the
design-driven tendency to anthropomorphize car fronts, we took a closer look at
explicit anthropomorphism-related judgments in study 2.
FIGURE 2
Mean Response Latencies to Car Words versus Face Words across the Three
Priming Conditions
Note. Response latencies have been z-transformed. Smaller values indicate shorter response
latencies, that is, a stronger activation of the associated concept face or car, respectively.
Study 2: Explicit Measurement of Anthropomorphizing
In study 1, we found that participants in a LDT responded faster to face-related words
than to car-related words, but only when they saw car fronts as primes. As car fronts
possess a face-like look, but car sides do not, we interpreted this result to indicate that
the tendency to anthropomorphize might be revealed automatically merely due to the
cars’ design features. In a second study, we asked whether the tendency to
anthropomorphize cars, which is assumed to be spontaneous (and maybe not
conscious), might be also reflected in explicit judgments and evaluations.
103
Method and Procedure. The same participants who took place in the LDT rated
pictures of nine car models on different 7-point Likert scales after the LDT. The
pictures were identical to those which have been used in the LDT as primes. One
group (n = 59) indicated how anthropomorphic they perceived the cars to be. A second
and third group of participants (n = 45 and 39, respectively) rated the car models
shown in front or side view, respectively, with regard to three general marketing
variables (liking, willingness to pay, global positive affect) and specific evaluative and
behavioral tendencies which should go together with anthropomorphizing behavior
(we call these scales anthropomorphism-related scales). One anthropomorphismrelated scale asked the participants to rate how suitable 15 different adjectives are to
describe the car shown. Adjectives were taken from Aaker’s brand personality scale
(Aaker 1997), since we needed adjectives which describe human-like characteristics
(e.g., cheerful, reliable). The second anthropomorphism-related scale asked
participants to imagine and evaluate the relationship quality with the car shown,
mainly assessing commitment/loyalty to and trust in the car (six commitment items
adapted from Aaker, Fournier, and Brasel 2004, e.g., I would be very loyal to the car, I
would stick with the car even if it let me down once or twice; four trust items adapted
from Chaudhuri and Holbrook 2001, e.g., I would trust this car, I would rely on this
car). Furthermore, with two items we also assessed the participants’ personal
disposition to anthropomorphize, asking whether the participants think that their own
car has a name and/ or a gender (adapted from Benfield, Szlemko, and Bell 2006). As
we translated all the items from English into German and had to make slight
modifications so that the items were appropriate for our context, items were translated
back into English by a native speaker who was fluent in German to guarantee
equivalence in meaning between the English and German version and, additionally, all
scales were pretested in a separate study conducted in November 2008 (n = 945,
Cronbach’s α of all scales > .90).
Specific Hypothesis. We expected as an explicit evaluative consequence of automatic
face-schema activation and, therefore, of anthropomorphizing that participants would
rate cars shown in front view (i.e., face-like cars) higher on the two assessed
anthropomorphism-related scales than car sides which do not resemble a face. That is,
we assumed that the participants would attribute human-like personality traits more to
cars which might be anthropomorphized spontaneously (i.e., cars shown in front view)
and that participants also feel a stronger relationship with anthropomorphized cars
(i.e., cars shown in front view). As a consequence, car fronts are maybe also evaluated
104
as more appealing than car sides on the marketing variables of liking, willingness to
pay, and general positive affect.
Results. Participants rated pictures of nine car models either shown in front view or
shown in side view on three marketing scales (liking, willingness to pay, and global
affect) and on two anthropomorphism-related scales (attribution of personality traits to
car, imagined relationship with car). Similar to the analysis we ran on the data from
the LDT described above, we z-transformed all ratings separately for each car model,
because we were not interested in the effects of individual car models on the
dependent measures. So, we ran all subsequent data analysis on these z-values. As
ratings on both anthropomorphism-related scales were highly correlated (r = .74, p <
.001), we created a composite variable which we named the anthropomorphism score.
Overall, to our surprise, a one-way ANOVA with front view vs. side view as between
factor and anthropomorphism score as dependent measurement showed that
participants did not rate front views significantly higher on the anthropomorphism
score than side views (0.14 vs. -0.16), (F(1, 82) = 2.45, p = .122). One explanation for
this unexpected finding concerns the indirect nature of our anthropomorphism
measures. Instead of asking directly about the extent to which our respondents saw the
cars as people (e.g., It seems like a person to me), we asked about the relationship to
the car (e.g., loyalty). A majority of participants who were presented the front angle of
the product may have seen the car in human terms but only a subset may have taken
that perception to the next level and imagined a relationship. If that is the case, we
might find an effect on our anthropomorphism measures among those participants
whose actions suggest they are willing to form a relationship with their car. To test this
explanation, we therefore controlled in a further analysis for the participants’ personal
disposition to anthropomorphize as measured by their willingness to name their car or
assign it a gender, assigning participants to one of three disposition levels based on the
two disposition items (no or weak or strong personal disposition to anthropomorphize
cars). On doing so, we found the expected main effect of car view (front vs. side) on
the anthropomorphism score with car fronts being more anthropomorphized than car
sides (0.38 vs. -0.44), (F(5, 71) = 8.54, p = .005) which was qualified by a significant
view of the car × personal disposition interaction (F(2, 71) = 3.72, p = .029) (Fig. 3).
Especially participants with a strong tendency to anthropomorphize their own car rated
car fronts higher on the anthropomorphism score than car sides (0.96 vs. -0.86), (F(1,
5) = 14.13, p = .013), whereas participants without this disposition did not evaluate car
fronts significantly different from car sides (0.02 vs. 0.07), (F(1, 41) = 0.04, p = .840).
105
With regard to the marketing variables, participants were also willing to pay more for
cars seen in front view compared to side views (0.20 vs. -0.46), (F(1, 71) = 4.55, p =
.036) and the general positive affect elicited by cars was higher for front than for side
views (0.23 vs. -0.48), (F(1, 71) = 10.65, p = .002), when controlling for the personal
disposition to anthropomorphize in both analyses. Car fronts were not liked more than
car sides (F(1, 71) = 0.82, p = .370), although this effect was directional.
FIGURE 3
Interaction between View of the Car (Front vs. Side) and Personal Disposition to
Anthropomorphize across Three Disposition Groups (None, Weak, and Strong
Disposition to Anthropomorphize)
Note: Participants with a weak or strong disposition to anthropomorphize showed a stronger
tendency to endow car fronts with human-like characteristics than car sides, whereas
participants without such a disposition did not.
General Discussion
The present research examined the effect of product features on spontaneous
anthropomorphism and product evaluations. Applying an implicit method (LDT) to
assess which concepts were activated in the participants’ minds by special picture
primes, we found that participants who were primed with cars presented in front view
showed on average similar response patterns to participants who were primed with
human faces. In contrast, participants who were primed with cars presented in side
view responded differently from the participants in the face condition. So one could
conclude from this comparison that it is indeed the particular, face-like feature
configuration only found in car fronts (but not in car sides) which accounted for the
participants’ tendency to anthropomorphize.
106
Interestingly, our participants were not aware of the study’s goal (i.e., they did not see
any relationship between the priming pictures and the words employed) which further
supported our assumption that anthropomorphizing is an automatic and maybe also
unconscious process. This is in line with other authors’ assumptions (Bush 1990;
Caporael 1986; Chartrand, Fitzsimons, and Fitzsimons 2008; Epley et al. 2007; Ingram
and Annable 2004), but to our knowledge we are among the first to have found
empirical evidence to support this assumption for anthropomorphizing in product
design. Consistent with our results from the LDT, we also found that, under certain
conditions, car fronts were rated significantly higher than car sides on explicit
anthropomorphism-related and marketing measures. So even though people might not
be aware of the underlying cognitive process which leads to anthropomorphizing (i.e.,
the activation of a human schema), the automatic tendency to anthropomorphize seems
to be reflected in explicit product inferences and evaluations under certain conditions
(Aggarwal and McGill 2007). Consumers might find an anthropomorphic product
appealing, but they cannot tell why.
Finally, we would like to point out two important implications of our results. Firstly, if
consumers have the tendency to anthropomorphize products spontaneously solely due
to anthropomorphic design features, marketers should consider a product's design as
an important addition to classical marketing-mix activities that are aimed at
developing a product's or brand's personality. Thus far, a product's design and the
corresponding communicative activities are treated quite independently in practice and
potentially supplementary benefits are not exploited (Bloch 1995). But, based on our
results, it seems likely that advertising a product as possessing human-like traits might
be much more efficient if the product's design in the first place allows spontaneous
anthropomorphizing due to its morphological shape.
Second, concerning explicit product evaluations, we found some indication that
spontaneous schema activation seems to affect explicit judgments positively, but only
under certain conditions, that is, depending on personal dispositions. These findings
suggest that design features that promote anthropomorphism will only be effective
among a segment of consumers with a chronic tendency to anthropomorphize of those
who find themselves in specific circumstances that promote seeing the human in
nonhuman forms (Epley et al. 2007). However, our findings may be limited to the
particular designs adopted in our study, which reflect current models of cars in the
European market. Future research is needed to examine the specific elements of design
that might increase anthropomorphism and explicit product evaluation among a
107
broader set of the population. In addition, future research should consider the interplay
of explicit and implicit measures of anthropomorphism and product evaluation.
108
References
Aaker, Jennifer (1997), “Dimensions of Brand Personality,” Journal of Marketing
Research, 34 (August), 347-57.
Aaker, Jennifer, Susan Fournier, and S. Adam Brasel (2004), “When Good Brands Do
Bad,” Journal of Consumer Research, 31 (June), 1-16.
Aggarwal, Pankaj (2004), “The Effects of Brand Relationship Norms on Consumer
Attitudes and Behavior,” Journal of Consumer Research, 31 (June), 87-101.
Aggarwal, Pankaj and Ann L. McGill (2007), “Is That Car Smiling at Me: Schema
Congruity as a Basis for the Evaluation for Anthropomorphized Products,”
Journal of Consumer Research, 34 (December), 468-79.
Anderson, John R. (1983), The Architecture of Cognition, Cambridge, MA: Harvard
Press.
Bajo, M. Teresa and José J. Canas (1989), “Phonetic and Semantic Activation During
Picture and Word Naming,” Acta Psychologica, 72 (2), 105-15.
Bargh, John A. and Tanya L. Chartrand (2000), “The Mind in the Middle: A Practical
Guide to Priming and Automaticity Research,” in ed. Harry T. Reis and Charles
M. Judd, Handbook of research methods in social and personality psychology,
New York: Cambridge, 253-85.
Benfield, Jacob A., William S. Szlemko, and Paul A. Bell (2006), “Driver Personality
and Anthropomorphic Attributions of Vehicle Personality Relate to Reported
Aggressive Driving Tendencies,” Personality and Individual Differences, 42
(2), 247-258.
Bloch, Peter H. (1995), “Seeking the ideal Form – Product Design and Consumer
Response,” Journal of Marketing, 59 (3), S. 16-29.
Bush, Donald J. (1990), “Body Icons and Product Semantics,” in ed. Susann Vihma,
Semantic: Visions in Design, Helsinki: UIAH, C1-C14.
Callcott, Margaret F. and Barbara J. Phillips (1996), “Observations: Elves Make Good
Cookies: Creating Likeable Spokes-Character Advertising,” Journal of
Advertising Research, 36 (5), 73-78.
Caporael, Linnda R. (1986), “Anthropomorphism and Mechanomorphism: Two Faces
of the Human Machine,” Computers in Human Behavior, 2 (3), 215-34.
Chartrand, Tanya L., Gráinne M. Fitzsimons, and Gavan J. Fitzsimons (2008),
“Automatic Effects of Anthropomorphized Objects on Behavior,” Social
Cognition, 26 (2), 198-209.
109
Chaudhuri, Arjun and Morris B. Holbrook (2001), “The Chain of Effects from Brand
Trust and Brand Affect to Brand Performance: The Role of Brand Loyalty,”
Journal of Marketing, 65 (2), 81-93.
Collins, Allan M. and Elizabeth F. Loftus (1975), “A Spreading-Activation Theory of
Semantic Processing,” Psychological Review, 82 (6), 407-28.
DiSalvo, Carl and Francine Gemperle (2003), “From Seduction to Fulfillment: The
Use of Anthropomorphic Form in Design,” in Proceedings of the 2003
international conference on Designing pleasurable products and interfaces, p.
67-72.
Epley, Nicolas, Adam Waytz, and John T. Cacioppo (2007), “On Seeing Human: A
Three-Factor Theory of Anthropomorphism,” Psychological Review, 114 (4),
864-86.
Fournier, Susan (1998), “Consumers and Their Brands: Developing Relationship
Theory in Consumer Research,” Journal of Consumer Research, 24 (March),
343-73.
Gauthier, Isabel, Pawel Skudlarski, John C. Gore, and Adam W. Anderson (2000),
“Expertise for Cars and Birds Recruits Brain Areas Involved in Face
Recognition,” Nature Neuroscience, 3 (2), 191-97.
York: Oxford.
perceptions of Unintentional Anthropomorphism and Zoomorphism in
Ankara, Turkey.
Irvin, Deborah J. and Stephen J. Lupker (1983), “Semantic Priming of Pictures and
Words: A Levels of Processing Approach,” Journal of Verbal Learning and
Verbal Behavior, 22 (1), 45-60.
Kiesler, Sara, Aaron Powers, Susan R. Fussell, and Christen Torrey (2008),
“Anthropomorphic Interactions with a Robot and a Robot-Like Agent,” Social
Cognition, 26 (2), 169-81.
Kroll, Judith F. and Mary C. Potter (1984), “Recognizing Words, Pictures, and
Concepts: A Comparison of Lexical, Object and Reality Decisions,” Journal of
Verbal Learning and Verbal Behavior, 23 (February), 39-66.
Kwan, Virginia S. Y., Samual D. Gosling, and Oliver P. John (2008),
“Anthropomorphism as a Special Case of Social Perception: A Cross-Species
110
Social Relations Model Analysis of Humans and Dogs,” Social Cognition, 26
(2), 129-42.
Mandler, George (1982), “The Structure of Value: Accounting for Taste,” in ed.
Margaret S. Clark and Susan T. Fiske, Affect and Cognition: The 17th Annual
Carnegie Symposium, Hillsdale, NJ: Erlbaum, 3-36.
McNamara, Timothy P. (1994), “Theories of Priming II: Types of Primes,” Journal of
Experimental Psychology: Learning, Memory, and Cognition, 20 (3), 507-20.
Meyers-Levy, Joan and Alice M. Tybout (1989), “Schema Congruity as a Basis for
Product Evaluation,” Journal of Consumer Research, 16 (June), 39-54.
Mondloch, Catherine, J., Terri L. Lewis, D. Robert Budreau, D. Maurer, James L.
Dannemiller, Benjamin R. Stephens, and Kathleen A. Kleiner-Gathercoal
(1999), “Face Perception during Early Infancy,” Psychological Science, 10
(September), 419-22.
Theios, John and Paul C. Amrhein (1989), “Theoretical Analysis of the Cognitive
Processing of Lexical and Pictorial Stimuli: Reading, Naming, and Visual and
Conceptual Comparisons,” Psychological Review, 96 (1), 5-24.
Vanderwart, Mary (1984), “Priming by Pictures in Lexical Decision,” Journal of
Verbal Learning and Verbal Behavior, 23 (1), 67-83.
Paper 5: The Cute Look: Baby-Schema Effects in Product
Design
∗
Linda Miesler(1), Helmut Leder(2)
Abstract
According to evolutionary theory, some visual key stimuli such as the baby schema
(“Kindchenschema”) automatically elicit strong positive responses in consumers (e.g.,
affection and approach behavior). Previous studies have suggested that consumers
perceive and process car fronts analog to human faces. However, it is not known how
consumers respond on an affective level to such biologically relevant patterns when
they are present in artifacts. In two experiments, we examined whether people would
provide similar cuteness ratings to pictures of car fronts and faces, which were
manipulated according to the baby schema. Our results showed that people were able
to detect the baby schema in car fronts. Further, people’s cuteness ratings of car fronts,
although generally lower, were comparable to their cuteness ratings of human faces.
Hence, visual key stimuli might be powerful product design features to create affectladen product designs.
Introduction
Companies profit greatly when consumers fall in love with their products on the first
sight. In marketing and product development, it is well known that as the technical
quality of consumer products becomes similar, the emotional value of products gains
in importance [1, 2, 3]. Thus, the visual appearance of a product is an important
marketing tool to appeal to consumers’ emotions, and to elicit spontaneous affective
responses such as “How beautiful!” or “How cute!”. A well-established trend in
product design is the creation of forms that somehow resemble humans, which is
reflected, for example, by a car designer’s concern for the “face” of a car [4, 5].
Product designers therefore exploit an innate human trait already known to
psychologists and anthropologists: due to the evolutionary value of human forms,
consumers are highly sensitive and attracted to them [6]. Marketing and design
professionals’ assumptions on what attracts consumers may be based intuitively on
∗
(1) Linda Miesler, Doctoral Candidate, Center for Customer Insight, University of St. Gallen; (2) Helmut
Leder, Professor of Psychology, Department of Basic Research in Psychology, University of Vienna.
112
Paper 5: The Cute Look
such evolutionary principles (e.g., consumers’ preferences are shaped by innate
motives and needs) [7]. However, the effects of such forms on product design
appreciation have not been investigated systematically.
To address the needs and preferences of consumers, marketing and design
professionals need behavioral theories. Surprisingly, evolutionary theory, which could
be useful for marketing, consumer research, and product development [8], has been
neglected. To date, only a few researchers have applied an evolutionary psychological
approach to understanding consumer behavior [9]. Evolutionary psychology explains
human behavior in terms of innate perceptual, cognitive and/or motivational
mechanisms that are assumed to have evolved through natural selection as adaptations
to ancestral conditions [10].
For product design research and application, understanding innate perceptual
processes is particularly compelling. In the present research, we investigated one type
of innate perceptual process, that is spontaneous responses to visual key stimuli (or
visual releasers [11]). We study the baby schema (“Kindchenschema”), which was
first described by Lorenz [12] as a set of physical features usually present in infant
faces. The features of the baby schema elicit spontaneous cuteness perceptions in
adults, which are important triggers of positive emotional responses to infants.
Therefore, consumers’ responses to visual key stimuli such as the baby schema are an
interesting research topic with regard to the relationship between product design and
emotions.
In summary, this paper has two intentions: to theoretically apply an evolutionary
psychology framework to product design, and to show its relevance for understanding
product appreciation; and to present two empirical studies in which we systematically
examined the relevance of specific evolutionarily-relevant physical features to product
design. The studies examined how consumers respond affectively to visual key stimuli
(i.e., the baby schema) that are presented in non-humans, that is consumer products.
Human Evolution and Visual Key Stimuli: The Baby Schema
From Biology and Psychology it is known that the presence of particular stimuli, so
called key stimuli, automatically triggers innate cognitive, affective or motivational
responses in individuals perceiving these stimuli (e.g., Zebrowitz, p. 68 [16]). Key
stimuli are usually – but not always – visual (e.g., a special animal coat pattern
formation or a specific body movement to attract potential mates). Thus, they are a
very interesting starting point when investigating the significance of human evolution
113
for product design appreciation. People’s automatic responses to key stimuli are
adaptive since they have developed during human evolution to guarantee one’s own
and the kin’s survival and reproduction. Even though modern consumers’
environments are obviously different from the environments where these adaptive
mechanisms evolved, human preferences and behaviors are still shaped by these
mechanisms. This has been shown for different domains of everyday life such as
mating [13], food choice [14] and even aesthetic preferences [15].
Undoubtedly, babies elicit spontaneous favorable responses in most perceivers [16].
The initial reaction to infants is an approach behavior such as smiling [17, 18, 19], and
people prefer pictures of infants over pictures of adults [20, 21, 22]. Lorenz [12] stated
that babies are approached positively because their physical appearance triggers
affection and nurturing in people, and that this promotes parental caretaking, and
consequently, offspring survival. There is high agreement regarding the physical
features that comprise the baby schema: round face, large eyes, small nose and mouth,
thick lips, and small chin - to name a few [23-28] (for a review of the features, see
Zebrowitz, [16], pp. 68-78). In accordance with Lorenz’ assumptions, previous studies
have shown that baby-schema features are correlated with perceived infant cuteness
[17] and adults’ motivation for caretaking [25, 29]: infants with more “babyish” facial
features (e.g., larger eyes, smaller nose) were perceived as cuter and elicited stronger
motivation in adults for caretaking.
Zebrowitz [16] (p. 78) and others suggest the baby schema to be an abstract visual
pattern which is more than the appearance quality of a specific organism, hence, it may
characterize various organisms. Equally, Lorenz [12] postulated that these features
elicit positive responses even when they are present in non-humans. In line with this
assertion, Zebrowitz [16] (p. 78) and others found that humans are not only sensitive to
baby-schema features found in infant faces, but also to those found in members of
other species. Humans respond positively to infant animals [30], cute cartoon
characters or dolls [31], and babyfaced adults. Concerning the latter, there is a lot of
evidence showing that people are able to judge the babyfaceness of human faces of
various ages [32]. Zebrowitz [33] stated that the similarities in responses to babies and
to faces that resemble babies are due to an overgeneralization of the evolutionarily
adaptive response of identifying babies (baby-face overgeneralization hypothesis).
Regarding the high affective value of the baby schema in the context of product design
research and application, the following question arises: are the features of the baby
schema so resilient that they can be seen even in non-living things such as design
114
products? If consumers respond to cute design products as they would respond to cute
living organisms, this might have important implications for product design
appreciation.
Previous Research on Face-like Product Designs
Up to now, researchers who have considered the impact of design on product
evaluations focused on general descriptive product attributes such as color and shape.
For example, Seva, Been-Lirn Duh, and Helander [34], investigated different mobile
phone designs by addressing design attributes such color, display size and shapes of
navigation buttons. Such product category-specific, exploratory approaches are
interesting when one aims to create design guidelines for one specific product
category. Results from such investigations, however, lack generalizability because
they are specific to one product category. In contrast, applying an evolutionary
psychology framework (i.e., the theory of the baby schema) allows the specification of
the values of features that could positively affect consumer perceptions across
different product categories: for example, round shapes should be perceived as cuter
than square shapes. In this regard, Khalid and Helander [3] demanded that “an explicit
requirement of emotional design is that it should be theoretically driven and
empirically grounded (…)” (p. 198). One exception to the rather exploratory design
research approaches is the upcoming research on human-like forms in product design
based on the theory of anthropomorphism (i.e., seeing non-humans as humans [6]).
Although there are many examples for human-like designed products (for an overview
see for example DiSalvo and Gemperle [35]), systematic research regarding such
anthropomorphic product forms is still scarce, but increasing [36-38]. As we address
the effects of baby-schema features on design appreciation, our research is mainly
based on previous studies that have dealt with face-like product designs. Cars are often
mentioned in discussions of face-like forms in product design. It has been suggested
that a car’s front is perceived and processed similarly to a human face. So, Windhager
et al. [5] and Erk et al. [39] have provided evidence that consumers are able to assign
facial features to components of car fronts without much effort. For example,
Windhager et al. [5] investigated people’s eye movement patterns when comparing car
fronts with human faces, and found that a car’s headlights are considered as eyes, the
grille as the nose, and the air intake or the grille as the mouth. Miesler, Landwehr,
Herrmann and McGill [38] revealed that due to the car front’s face-like physical
features, presenting a car shown in front view automatically activates a human faceschema in the consumer’s mind, whereas presenting a car shown from side view does
115
not. In another study, Windhager et al. [40] found evidence that car fronts are
interpreted as faces, and that people draw the same inferences from car fronts as from
faces (e.g., concerning perceived sex, or maturity).
Thus, biologically-related forms, such as those that are face-like, might be detected in
artifacts. However, such detection would not automatically imply that the face-like
forms would also produce product-related emotions (e.g., by eliciting affection). For
example, seeing faces in clouds does not necessarily change one’s attitude towards
clouds. Thus, from a designer’s perspective, both feature detection and affective
responses to such features are interesting. Aggarwal and McGill [36] have already
demonstrated that products with physical features that are congruent with a human
schema are liked more than products with physical features that are incongruent.
However, the affective responses found in their study (i.e., liking) were rather due to a
cognitive process (i.e., successful cognitive match between a product’s appearance and
a mental schema) than to specific affect-laden features of the product.
In the present project, we provide support for the assumption that face-like features in
car fronts are processed analog to human faces. Moreover, we investigate whether
such biologically-relevant patterns have an effect on consumers’ affective responses to
products. Specifically, we examine a special face schema that is known to be highly
affect-laden: the baby schema. Thus, we tap into the evolutionary perspective of
design appreciation by addressing the question of whether people are sensitive to
visual key stimuli in artifacts (as assumed intuitively by designers and marketers). If
this is the case, people should perceive cuteness similarly in artifacts and human faces.
The perception of cuteness has a strong emotional connotation, because as cited above
objects which are perceived as cute by consumers elicit strong positive emotions and
emotion-related behaviors such as spontaneous smiling. Therefore to examine cuteness
perceptions in consumers could inform the development of emotional product designs
which appeal to the consumers’ emotions on the first sight.
Methods
Hypotheses
We tested the effects of baby-schema features on the perception of cuteness. We used
images of car fronts that were supposed to resemble human faces. Importantly, we
expected that car fronts with local features that were manipulated according to the
baby schema (e.g., larger headlights, smaller grilles), would be perceived as cuter than
116
non-manipulated car fronts (hypothesis 1). In comparing car fronts with human faces,
we predicted that the detection of a baby schema would elicit cuteness perceptions in
both stimuli categories. However, we expected the artificial stimuli to induce weaker
cuteness responses than the faces. That is why we assumed that people would be more
sensitive to baby-schema cues in faces as compared to car fronts (hypothesis 2).
Stimuli
As described above, various features could affect the babyish appearance (the
“cuteness”) of human faces. Two restrictions determined which features we used to
manipulate the babyfaceness of both car fronts and human faces. First, as in previous
studies [5, 38-40], we selected facial features that clearly corresponded to the features
of a car front (e.g., the car’s headlights as human eyes). Second, we selected features
that had been shown to strongly affect perceived cuteness. Thus, we selected car
fronts’ headlights (~ eyes), middle grille (~ nose) and air intake (~ mouth; see Figure
1).
FIGURE 1
Car Front Features Selected for Manipulation
To create babyfaced car fronts, we manipulated the sizes of the three features. The
original stimuli were monochrome photographs of 16 different cars (picture size:
512×512 pixels) shown in front view. For each of these 16 cars, a babyfaced version
was created by enlarging the headlights by 20% (since babies have proportionally
large eyes), shrinking the middle grille by 20% (since babies have proportionally small
noses) and decreasing the width of the air intake by 20% while simultaneously
increasing its height by 20% (since babies have small mouths, but relatively thicker
lips than adults). Such feature size manipulations in a range of 10-20% have been also
used by other authors (e.g., [26]). Thus, there were two versions of each car, a nonmanipulated basic version and a manipulated babyfaced version (see Figure 2, left
117
side). As nine of the cars belonged to the compact car segment and the remaining six
cars to the medium-class segment, we included car shape as an additional factor (since
babies have rounder faces than adults). We calculated a roundness index separately for
each car using the measurement tool in Adobe Photoshop. Specifically, we measured
the height and width of each car (points with maximal distance) and divided height by
width to derive the roundness index (index = 1, perfect roundness; index < 1, ellipse).
We found that the average roundness index for compact cars was significantly higher
(and closer to a perfect circle) than the index for medium-class cars (M roundness compact =
.85, SD = 0.02; Mroundness medium = .77, SD = 0.02; t (14) = - 7.69, p < .001). This
validated our decision to include shape as an additional factor in our study, even
though we did not manipulate it systematically.
To test our second hypothesis, and to validate that the features we chose to manipulate
were relevant for cuteness perception, we applied identical manipulations to 16 human
faces (8 male and 8 female faces; picture size: 370×555 pixels). The pictures were
taken from the Vienna Face Database which contains standardized pictures of male
and female students with an age range from 18 to 25 years. A pre-study with 19
participants revealed that size manipulations of 20% would lead to unnatural face
versions (cf., the “uncanny valley”- effect [41]). Therefore, we decided to apply a
smaller size manipulation to the faces. The manipulation of the sizes of the eyes, nose
and mouth of each of the faces were identical to the manipulation of the car fronts,
although by only 10% for both enlarging and shrinking (see Figure 2, right side). The
faces’ relational characteristics (e.g., distance nose - upper lip, distance between the
eyes) were changed as little as possible.
118
FIGURE 2
Examples of the Used Stimulus Material
Participants and Procedure
One group of participants rated the 32 cars (n = 19; Mage = 23 yrs, SDage = 3 yrs; 74%
females) and another group rated the 32 faces (n = 16; Mage = 27 yrs, SDage = 6 yrs;
75% females) for cuteness on a seven-point Likert scale (1 = “not cute at all” to 7 =
“very cute”). As the participants’ task to rate the cars for cuteness was not trivial, in
both studies the participants were carefully instructed what cuteness means - especially
with respect to cars. We told participants in the car group that, apart from living
organisms, consumer products (e.g., Hello Kitty products) could also look cute. We
were careful to not point out to the physical features that can account for such a look.
The stimuli were presented in random order on a high resolution computer screen.
Each picture was presented for two seconds in the middle of the screen, then the rating
scale appeared and participants were asked to provide their rating. All participants saw
both the basic and the manipulated versions, but the pictures were randomized in such
a way that the basic and manipulated versions of the same stimulus were never
presented in sequence.
Results
Cuteness Perception of Car Fronts (Study 1)
To test our hypothesis that the manipulated, babyfaced cars will be perceived as cuter
than the basic, non-manipulated cars, we conducted a paired t-test using the 16 car
119
models as units of analysis. The analysis yielded a significant difference in the
perceived cuteness between the babyfaced (Mbabyface = 4.02, SD = 1.35) and nonmanipulated versions (Mbasic = 3.52, SD = 1.33), t (15) = - 9.59, p < .001, d = .38. The
babyfaced versions were perceived as cuter than the basic, non-manipulated versions.
At the single stimulus level (difference between basic and manipulated versions
calculated separately for the 16 car pairs), the size manipulation effect varied from an
average cuteness increase of ∆ = 0.16 (p = .30, one-sided) to ∆ = 0.84 (p = 0.008). We
also tested the effects of car shape on the cuteness ratings. A 2 (size manipulation:
basic vs. babyfaced) × 2 (car shape: compact vs. medium-class) within-betweensubjects ANOVA with repeated measures on the first factor and car shape as betweensubjects factor revealed that the cars’ shape had a significant effect on perceived
cuteness. There were higher cuteness ratings for compact cars than for medium-class
cars (Mcompact = 4.60, SD = 1.11 versus Mmedium = 2.70, SD = 0.68), F (1, 14) = 15.72, p
= .001. The effect of the feature size manipulation was also significant (F (1, 14) =
86.42, p < .001). Interestingly, there was no interaction effect between feature size
manipulation and car shape (F < 1; p = .44). The latter result suggested that it was not
only the cars’ segment and explicit preexisting knowledge structures such as “compact
cars are cute” that accounted for the perceived cuteness, but that the effect on
perceived cuteness was also due to the feature size manipulations according to the
baby schema. Although we found a statistically significant effect of feature size
manipulation on perceived cuteness, post-experiment interviews revealed that most of
the participants were unaware of these manipulations. Participants only reported some
brand association influence (such as “BMW Mini is a cute brand”), but as our analysis
showed, even basically very cute cars got cuter due to the baby-schema manipulation
(for example, absolute cuteness ratings of BMW Mini increased by 0.5 scale points
due to the manipulation).
Cuteness Perception of Human Faces (Study 2)
To test the hypothesis that people will be more sensitive to baby-schema features in
human faces than in car fronts, we applied the same feature size manipulation to
human faces and compared the cuteness ratings between the non-manipulated and
manipulated faces. As expected, a paired t-test yielded a significant difference in
perceived cuteness between the manipulated (Mbabyface = 3.67, SD = 0.87) and nonmanipulated faces (Mbasic = 3.14, SD = 0.78), t (15) = - 6.22, p < .001, d = .64. The
babyfaced versions of the faces were perceived as cuter than the non-manipulated
versions. Hence, our manipulations were suited for manipulating perceived cuteness.
120
In comparing Cohen’s effect sizes from the two studies, we found that participants’
cuteness perceptions were much more sensitive to baby-schema cues in human faces
than in car fronts (dcars = .38 versus dfaces = .64), which was in accordance with our
second hypothesis.
Discussion and Conclusions
Applying the theory of innate visual key stimuli, we predicted differences in people’s
responses to different versions of car designs. In two studies, we found that both car
fronts and human faces were perceived as cuter when - in accordance with the baby
schema - the size of the headlights (eyes) was increased, and the size of the grille and
air intake (nose and mouth) was decreased. Thus, our results indicate that people are
able to detect innate key stimuli in artifacts, which confirms the results of previous
studies on face-like product designs. Importantly, our results suggest that people do
not only detect visual key stimuli in product designs, but also show affective responses
(i.e., cuteness response), that are comparable to innate affective responses, to artifacts
possessing such features. However, our results also revealed that visual key stimuli in
human faces elicited stronger cuteness responses than cars. Thus, when innate visual
patterns such as the baby schema are generalized onto non-human stimuli, consumers’
sensitivity to these cues might be weaker.
Based on our findings, we advice product designers that to elicit strong positive
responses in consumers, they should exaggerate the characteristics of visual key
stimuli in product designs (e.g., very large headlights that would be unnaturally large
as eyes in a human face). In our studies, we focused solely on cars as product category.
However, the design concept “cuteness” might be applied to all product categories
where a strong consumer-product relationship is part of the marketing strategy, but
whose category members have rather a functional than emotional value (e.g., think of
electronic devices such as cell phones which could elicit a “protective instinct” in
consumers due to a cute appearance). On the other side, cuteness is of course less
suitable for product designs which shall represent brand values such as status or
power.
In our study, we requested spontaneous affective responses to product features
resembling innate visual key stimuli. In future studies, it might be interesting to take a
deeper look at effects of cute product designs on overt behavior. As the baby schema
also triggers behavioral responses such as caretaking, the effect of visual key stimuli
on a consumer’s willingness to purchase, use and keep a product should be further
121
investigated. Furthermore, cross-cultural effects and sex differences could also be
examined to find further evidence that consumers’ responses to human-like product
designs are shaped by innate perceptual and behavioral mechanisms.
122
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
Desmet, P. M. A., Overbeeke, C. J., & Tax, S. J. E. T. (2001). Designing products
with added emotional value: Development and application of an approach for research
through design. The Design Journal, 4, 32-47.
DiSalvo, C., & Gemperle, F. (2003). From seduction to fulfillment: The use of
anthropomorphic form in design. In Proceedings of the 2003 international conference
on designing pleasurable products and interfaces, p. 67-72.
Khalid, H.M., & Helander, M.G. (2006). Customer emotional needs in product design.
Concurrent Engineering, 14 (3), 197-206.
Welsh, J. (2006). Why cars got angry. Wall Street Journal, March 10, W1.
Windhager, S., Hutzler, F., Carbon, C.-C., Oberzaucher, E., Schaefer, K., Thorstensen,
T., Leder, H., & Grammer, K. (in press). Laying eyes on headlights: Eye movements
suggest facial features in cars. Collegium Antropologicum.
Guthrie, S.E. (1993). Faces in the clouds: A new theory of religion. New York:
Oxford.
Colarelli, S. M., & Dettman, J. R. (2003). Intuitive evolutionary perspectives in
marketing practices. Psychology & Marketing, 20 (9), 837-865.
Miller, G. (2009). Spent: Sex, evolution, and consumer behavior. New York: Penguin
Group.
Saad, G., & Gill, T. (2000). Applications of evolutionary psychology in marketing.
Psychology & Marketing, 17 (12), 1005-1033.
Lynn, M., Kampschroeder, K., & Pereira, A. (1999). Evolutionary perspectives on
consumer behavior: An introduction. In E.J. Arnould, & L.M. Scott (Eds.), Advances
in Consumer Research Volume 26 (pp. 226-230). Provo, UT : Association for
Consumer Research.
Coss, R. G. (2003). The role of evolved perceptual biases in art and design. In E.
Voland, & K. Grammer (Eds.), Evolutionary aesthetics (pp. 69-130). Berlin,
Heidelberg: Springer.
Lorenz, K. (1943). Innate forms of potential experience. Zeitschrift für
Tierpsychologie, 5, 233-519 [in German].
Buss, D. (1994). The evolution of desire: Strategies of human mating. New York:
Basic Books.
Rozin, P. (1976). Psychological and cultural determinants of food choice. In T.
Silverstone (Ed.), Appetite and food intake (pp. 286-312). Berlin: Dahlem
Konferenzen.
Voland, E., & Grammer, K. (2003). Evolutionary aesthetics. Berlin, Heidelberg:
Springer.
Zebrowitz, L. (1997). Reading faces: Window to the soul? Boulder, CO: Westview
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
123
Press.
Hildebrand, K. A., & Fitzgerald, H. E. (1978). Facial feature determinants of
perceived infant attractiveness. Infant Behavior and Development, 2 (January), 329339.
Power, T. G., Hildebrandt, K. A., & Fitzgerald, H. E. (1982). Adults’ responses to
infants varying in facial expression and perceived attractiveness. Infant Behavior and
Development, 5 (1), 33-44.
Schleidt, M., Schiefenhovel, W., Stanjek, K., & Krell, R. (1980). ‘‘Caring for a baby’’
behavior: reactions of passersby to a mother and baby. Man-Environment Systems 10
(2), 73-82.
Berry, D. S., & Zebrowitz McArthur, L. (1985). Some components and consequences
of a babyface. Journal of Personality and Social Psychology, 48 (2), 312-323.
Fullard, W., & Reiling, A. M. (1976). An investigation of Lorenz’s ‘‘babyness’’. Child
Development, 47 (4), 1191-1193.
Brosch, T., Sander, D., & Scherer, K. R. (2007). That baby caught my eye…Attention
capture by infant faces. Emotion, 7 (3), 685-689.
Berry, D. S., & Zebrowitz McArthur, L. (1985). Some components and consequences
of a babyface. Journal of Personality and Social Psychology, 48 (2), 312-323.
Cunningham, M. R. (1986). Measuring the physical in physical attractiveness: Quasiexperiments of the socio-biology of female facial beauty. Journal of Personality and
Social Psychology, 50 (5), 925-935.
Glocker, M. L., Langleben, D. D., Ruparel, K., Loughead, J. W., Gur, R. C., &
Sachser, N. (2009). Baby schema in infant faces induces cuteness perception and
motivation for caretaking in adults. Ethology 115 (3), 257-263.
Keating, C. F., Randall, D. W., Kendrick, T., & Gutshall, K. A. (2003). Do babyfaced
adults receive more help? The (cross-cultural) case of the lost resume. Journal of
Nonverbal Behavior, 27 (2), 89-108.
Masip, J., Garrido, E., & Herrero, C. (2004). Facial appearance and impressions of
'credibility': The effects of facial babyishness and age on person perception.
International Journal of Psychology, 39 (4), 276-289.
Sprengelmeyer, R., Perrett, D. I., Fagan, E.C., Cornwell, R.E., Lobmaier, J.S.,
Sprengelmayer, R., Aasheim, H.B.M., Black, I.M., Cameron, L.M., Crow, S., Milne,
N., Rhodes, E.C., & Young, A.W. (2009). The cutest little baby face: A hormonal link
to sensitivity to cuteness in infant faces. Psychological Science, 20 (2), 149-154.
Sherman, G. D., Haidt, J., & Coan, J. A. (2009). Viewing cute images increases
behavioral carefulness. Emotion, 9 (2), 282-286.
Sanefuji, W., Ohgami, H., & Hashiya, K. (2007). Development of preference for baby
faces across species in humans (Homo sapiens). Journal of Ethology, 25 (3), 249-254.
Jacob, J. E., Rodenhauser, P., & Markert, R. J. (1987). The benign exploitation of
124
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
human emotions: Adult women and the marketing of cabbage patch kids. Journal of
American Culture, 10 (Fall), 61-71.
Zebrowitz, L. A., & Montepare, J. M. (1992). Impressions of babyfaced individuals
across the life span. Developmental Psychology, 28 (6), 1143-1152.
Zebrowitz, L. A., Fellous, J.-M., Mignault, A., & Andreolletti, C. (2003). Trait
impressions as overgeneralized responses to adaptively significant facial qualities:
Evidence from connectionist modeling. Personality and Social Psychology Review, 7
(3), 194-215.
Seva, R., Been-Lirn Duh, H., & Helander, M. G. (2007). The marketing implications
of affective product design. Applied Ergonomics, 38 (6), 723-731.
DiSalvo, C., & Gemperle, F. (2003). From seduction to fulfillment: The use of
anthropomorphic form in design. In Proceedings of the 2003 international conference
on designing pleasurable products and interfaces, p. 67-72.
Aggarwal, P., & McGill, A. L. (2007). Is that car smiling at me: Schema congruity as
a basis for the evaluation for anthropomorphized products. Journal of Consumer
Research, 34 (December), 468-479.
Ingram, J., & Annable, L. (2004). ’I see you baby, shakin' that ass’: User perceptions
of unintentional anthropomorphism and zoomorphism in consumer products. In
Proceedings of the 4th Design and Emotion Conference, Ankara, Turkey.
Miesler, L., Landwehr, J. R., Herrmann, A., & McGill, A. L. (2010). Consumer and
product face-to-face: Antecedents and consequences of spontaneous face-schema
activation. Advances in Consumer Research 37.
Erk, S., Spitzer, M., Wunderlich, A.P., Galley, L., & Walter, H. (2002). Cultural
objects modulate reward circuitry. NeuroReport, 13 (18), 2499–2503.
Windhager, S., Slice, D. E., Schaefer, K., Oberzaucher, E., Thorstensen, T., &
Grammer, K. (2008). Face to face: The perception of automotive designs. Human
Nature, 19 (December), 331-346.
Seyama, J., & Nagayama, R. S. (2007). The uncanny valley: Effect of realism on the
impression of artificial human faces. Presence: Teleoperators and Virtual
Environments,16 (4), 337-351.
Curriculum Vitae
Personal Information
Name:
Linda Miesler
Date of Birth:
July, 18, 1981
Place of Birth:
Berlin, Germany
E-Mail:
[email protected]
Education
09/2008-05/2011:
University of St. Gallen, Switzerland
Doctoral Studies in Marketing
11/2009-07/2010:
University of Vienna, Austria
Visiting Scholar at the Department of Basic Research
in Psychology (financed by Swiss National Science
Foundation)
10/2001-11/2006:
Humboldt University, Berlin, Germany
Diploma Studies in Psychology (Major: Cognitive
Psychology)
07/2001:
Abitur
Informationen über die Mitarbeit der Ko-Autoren
Die vorliegenden Artikel wurden hinsichtlich der Idee, Ausführung und schriftlichen
Verfassung in wesentlichen Teilen selbständig von Linda Miesler erstellt.
Die Ko-Autoren haben den Forschungsprozess begleitet, indem sie Input und
Anregungen geliefert haben im Hinblick auf die Erstellung des experimentellen
Designs sowie die methodische Auswertung.
___________________________________
Linda Miesler
___________________________________
Prof. Dr. Andreas Herrmann

Imitating Human Forms in Product Design

Transcription

Similar documents

Face Painting for Events

Chicago Street Domination

Slane Jewelry Steve Vaubel Tess McGuire Hats William Yeoward

Skin camouflage

INTERCER AMIC DIGITAL TECHNOLOG Y

what began as a hobby / home industry in ilse appelgryn`s kitchen in

Ride 4 Smiles 2012 Program

accomplishments - City of Rock Island

SEPTEMBER 2013 FACES E

Designs Sew Fine - Tina`s Embroidery Design Collection