ArticlePDF Available

A Tiger and a President: Imperceptible Celebrity Facial Cues Influence Trust and Preference

December 2012
Journal of Consumer Research 39(4):769-783

DOI:10.1086/665412

Authors:

Robin J. Tanner

University of Wisconsin–Madison

Ahreum Maeng

University of Kansas

Neuroscientiﬁc research suggests that the brain has evolved speciﬁc capabilities enabling automatic social judgments of others to be made based on facial properties alone. However, little research in marketing has considered the consequences of how facial imagery is automatically processed. We explore automatic perceptions of familiarity by using morphing software to digitally combine unfamiliar faces with those of Tiger Woods and George Bush. Despite a complete lack of conscious recognition, trustworthiness ratings of the composite faces are clearly inﬂuenced by the celebrities in question. This appears to be due to implicit recognition being sufﬁcient for individuals to automatically access their own summary valence judgments of either Woods or Bush.

Content uploaded by Ahreum Maeng

Content may be subject to copyright.

Content uploaded by Robin J. Tanner

Content may be subject to copyright.

Journal of Consumer Research, Inc.

A Tiger and a President: Imperceptible Celebrity Facial Cues Influence Trust and Preference

Author(s): Robin J. Tanner and Ahreum Maeng

Source:

Journal of Consumer Research,

Vol. 39, No. 4 (December 2012), pp. 769-783

Published by: The University of Chicago Press

Stable URL: http://www.jstor.org/stable/10.1086/665412 .

Accessed: 13/08/2013 13:22

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to

digitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

769

䉷 2012 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 39 ● December 2012

A Tiger and a President: Imperceptible

Celebrity Facial Cues Inﬂuence Trust

and Preference

ROBIN J. TANNER

AHREUM MAENG

Neuroscientiﬁc research suggests that the brain has evolved speciﬁc capabilities

enabling automatic social judgments of others to be made based on facial prop-

erties alone. However, little research in marketing has considered the conse-

quences of how facial imagery is automatically processed. We explore automatic

perceptions of familiarity by using morphing software to digitally combine unfamiliar

faces with those of Tiger Woods and George Bush. Despite a complete lack of

conscious recognition, trustworthiness ratings of the composite faces are clearly

inﬂuenced by the celebrities in question. This appears to be due to implicit rec-

ognition being sufﬁcient for individuals to automatically access their own summary

valence judgments of either Woods or Bush. Alternative explanations based on a

perceptual-ﬂuency account, or implicit recognition sufﬁcient to perceive speciﬁc

trait ratings, are ruled out. These ﬁndings suggest that the marketing practice of

digitally manipulating the attractiveness of facial imagery risks overlooking the

important inﬂuence of familiarity.

uman faces are arguably the most important visual stimuli

we perceive. They are quite staggeringly complex and

represent an exceptionally rich source of social information,

which we have evolved to be extremely adept at decoding

(Engell, Haxby, and Todorov 2007). For example, faces not

only reveal a tremendous amount of information about our

underlying emotions, but the mere adoption of certain facial

expressions is sufﬁcient to moderate the underlying physiology

and subjective experience of emotion (Ekman 1992). Further-

more, we begin to mimic facial expressions soon after birth

(Hatﬁeld, Cacioppo, and Rapson 1993), infer movement tra-

Robin J. Tanner is assistant professor of marketing, Wisconsin School of

Business, University of Wisconsin–Madison, 975 University Avenue, Madison,

WI 53706 (rtanner@bus.wisc.edu). Ahreum Maeng is a PhD candidate in mar-

keting, Wisconsin School of Business, University of Wisconsin–Madison, 975

University Avenue, Madison, WI 53706 (amaeng@bus.wisc.edu). Correspon-

dence should be sent to the ﬁrst author. The authors thank the Fuqua brown

bag reading group, Tanya Chartrand, the editor, the associate editor, and the

three reviewers for suggestions that signiﬁcantly improved the article. This work

was supported by a grant from the UW Madison Graduate School to the ﬁrst

author.

Ann McGill served as editor and Ravi Dhar served as associate editor for this

article.

Electronically published March 21, 2012

jectories and intentions from peoples’ gazes (Nummenmaa,

Hyona, and Hietanen 2009), and are astonishingly adept at

rapidly identifying angry faces in crowds (Hansen and Hansen

1988). In fact, evidence from cognitive, developmental, and

evolutionary psychology, as well as from cognitive neu-

roscience, supports the face-speciﬁcity hypothesis, which

suggests that the face represents a “special” (and possibly

unique) class of stimuli whose processing is handled by

discrete and specialized neural circuits (Kanwisher and

Yovel 2006).

Indeed, converging evidence suggests that the human brain

has evolved speciﬁc capabilities that enable automatic social

judgments of others (such as how trustworthy they are) to be

made based on facial properties alone (Critchley et al. 2000;

Engell et al. 2007; Winston et al. 2002). That humans have

evolved to be highly sophisticated (and automatic) interpreters

of facial information has potentially profound implications

for marketers. After all, from decorative models in cosmetics

ads to celebrity endorsers of sodas, facial imagery is nearly

ever-present across all forms of marketing communication. It

is thus rather surprising how little research in marketing has

directly explored the implications of how people automati-

cally code (and respond to) facial stimuli.

While there has certainly been a growing awareness

among marketing researchers that environmental cues can

automatically inﬂuence consumer behavior, this work has

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

770 JOURNAL OF CONSUMER RESEARCH

to date largely focused on examining the effects of lexical

and brand primes on speciﬁc goals and traits (Chartrand

et al. 2008; Fitzsimons, Chartrand, and Fitzsimons 2008).

Although automatic responses to some more overtly social

cues (such as the presence of interpersonal mimicry; Tanner

et al. 2008) have been studied, relatively little research has

speciﬁcally examined automatic evaluation of facial cues. The

research that does exist has tended to focus on certain effects

of speciﬁc expressions, such as in research on emotional con-

tagion (Howard and Gengler 2001; Small and Verochi 2009)

rather than chronic (emotionally neutral) features of the face.

However, given the emerging neuropsychological evidence

that our brains are essentially hardwired to automatically pro-

cess facial information, we believe that research on automatic

processing of facial imagery merits considerably broader at-

tention from marketing scholars.

The current research was conceived to explore one aspect

of automatic facial processing that has the potential to be

highly relevant to marketers. Speciﬁcally, we are interested

in whether perceptions of the trustworthiness of unfamiliar

faces (i.e., of apparently unknown individuals) can be in-

ﬂuenced by incorporating subtle facial cues of familiar peo-

ple into the faces in question. This question was motivated,

in part, by the ﬁnding from the facial morphing literature

that subtly (and imperceptibly) incorporating elements of an

individual’s own face into an otherwise unfamiliar face in-

creases reported trust in, and/or preference for, the unfamiliar

face (Bailenson et al. 2009; DeBruine 2002). Importantly,

since the self is simultaneously both similar and familiar, it

is unclear from this stream of research whether a similarity-

or familiarity-based process underlies the observed positive

effects. A primary goal of our research was to better un-

derstand these competing explanations. We do this by re-

lying on faces that are familiar but not similar to those of

our participants. Speciﬁcally, we investigate the effect of

subtly incorporating elements of celebrity images into oth-

erwise unfamiliar faces.

The remainder of this article is structured as follows. First,

we provide a brief overview of research on automatic facial

processing, which suggests that perceptions of trust can be

automatically determined based on facial cues alone. Sec-

ond, we review the existing ﬁnding from the digital morph-

ing literature that subtle incorporation of self-images into

unfamiliar faces increases perceived trust in the individuals

pictured, and we discuss both similarity- and familiarity-

based explanations that could underlie this effect. Finally,

we present four studies that explore the effects of morphing

unknown faces with highly familiar celebrity images. These

studies enable us to make the following claims. First, morph-

ing unknown faces to contain imperceptible cues of familiar

individuals can lead to similar positive effects on perceived

trustworthiness as does self-morphing. Second, this effect

appears to be driven by implicit recognition of the speciﬁc

familiar individual concerned rather than by either a more

general ﬂuency effect or by some aspect of their facial struc-

ture that they may share with others. Third, this implicit

recognition appears sufﬁcient for a broad approach/avoid

valence pertaining to the familiar individual to be perceived

but is insufﬁcient for speciﬁc trait assessments of them to

be accessed from memory.

AUTOMATIC FACIAL PROCESSING

Accurate social judgments have in all likelihood been a

signiﬁcant determinant of human survival and indeed of

the survival of our primate ancestors (Winston et al. 2002).

For example, from an evolutionary biology perspective, the

ability to quickly discern whether a stranger might be hostile

and/or untrustworthy could signiﬁcantly enhance an indi-

vidual’s ultimate chance of reproductive success (Cacioppo

2004). In fact, research suggests that almost all the stimuli

we encounter are evaluated on a positive/negative valence

dimension (Bargh 1997; Osgood, Suci, and Tannenbaum

1957), with these evaluations evoking approach and avoid

behavioral tendencies (Corwin 1921; Lewin 1935). More-

over, in social contexts, speciﬁcally, these valence evalua-

tions serve as a central input into person-impression for-

mation (Fiske, Cuddy, and Glick 2007; Kim and Rosenberg

1980).

Considerable evidence now suggests that many of these

social judgments are actually made entirely automatically,

that is, without any conscious deliberation or intent (Bargh

and Chartrand 1999; Devine 1989; Hassin, Uleman, and

Bargh 2005). This automatic “social intelligence” appears

to be supported by speciﬁc cognitive and neural mechanisms

(Winston et al. 2002), especially in the area of facial per-

ception (Kanwisher and Yovel 2006). The amygdala region,

in particular, appears to play a central role in making au-

tomatic socially relevant judgments on the basis of facial

appearances (Adolphs, Tranel, and Damasio 1998; Winston

et al. 2002). Indeed, Todorov and Engel (2008) argue that

a primary function of the amygdala may be to provide con-

tinuous vigilance by evaluating objects and agents prior to

interacting with them and that one way this is achieved is

by automatically evaluating the valence of faces (see also

Amaral 2002; Whalen 1998). Once rendered, these valence

evaluations are then used to guide approach/avoidance be-

haviors (Chen and Bargh 1999).

Of particular relevance to the continuing debate as to the

extent to which judgments can truly be determined entirely

automatically, a notable feature of these apparently auto-

matic face-based judgments is that they originate in a part

of the brain that is neuroanatomically dissociable from the

location where explicit judgments based on facial infor-

mation are made (Critchley et al. 2000). Put differently, in

layman’s terms, automatic (and unconscious) processing of

facial information is potentially unique in that it can actually

be attributed to a speciﬁc location in the brain. That evo-

lution has chosen to endow us with such speciﬁc capabilities

for automatic facial processing only serves to reinforce our

contention that marketers would be wise to pay greater at-

tention to how the implicit processing of facial information

might inﬂuence consumer behavior.

From a marketing perspective, a particularly important

insight from this body of research is that individuals appear

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 771

to automatically categorize faces in terms of their perceived

trustworthiness (Engell et al. 2007). Indeed, when assessing

unfamiliar faces, trust ratings have been demonstrated to

provide the best approximation of the core underlying va-

lence evaluation (Oosterhof and Todorov 2008). This ca-

pability enables individuals to form judgments about a novel

individual’s trustworthiness after as little as a 50-millisecond

exposure to their face (Todorov, Pakrashi, and Oosterhof

2009; Willis and Todorov 2006). Moreover, the social con-

sequences of these rapid automatic judgments are not re-

stricted to trust ratings in the laboratory. For example, rat-

ings of competence based solely on a 1-second exposure to

faces of unfamiliar political candidates were shown to pre-

dict 69% of the 2004 US Senate races and to be linearly

related to the margins of victory (Todorov et al. 2005).

It is important to note, however, that while at this point

a fairly signiﬁcant body of research has demonstrated that

automatic valenced judgments can be made on the basis of

facial features alone, much of the research to date has (un-

derstandably) been phenomenological in nature, with less

emphasis having been placed on understanding precisely

how these judgments are arrived at. Put simply, while we

know that automatic valence judgments are made, we know

a lot less about the speciﬁc inferential processes that underlie

them. While not the central focus of this article, we note

that there are a number of different possible facial cues that

automatic judgments might rely on. Such cues include facial

expressions (Critchley et al. 2000), the actual physical struc-

ture of the face (Carre, McCormick, and Mondloch 2009),

and resemblance to a speciﬁc known individual (Verosky

and Todorov 2010).

DIGITAL MORPHING RESEARCH

Digital morphing technology, which enables photographs

of two individuals to be blended together into a highly re-

alistic-looking virtual face, has provided researchers with a

new technique with which to explore the implicit aspects of

facial processing. For example, Bailenson et al. (2009) ex-

amined participant responses to composite images created by

morphing the faces of politicians with the faces of participants

themselves (henceforth “self-morphing”) in a ratio of 60%

politician, 40% participant. This ratio is signiﬁcant, as the

authors demonstrate that it produces composite faces within

which individuals could not perceive their own image. Put

differently, participants universally believed they were view-

ing unaltered pictures of the politicians in question. However,

attitudes toward the composites reﬂected an entirely different

story. When the candidate was relatively unfamiliar to them

(e.g., the 2006 Florida gubernatorial candidates for non-Flor-

ida residents), participants consistently demonstrated stronger

relative preference for the morphed images containing ele-

ments of their own faces.

Moving beyond unfamiliar politicians, Bailenson et al.

demonstrated that this self-morphing effect similarly inﬂu-

enced assessments of the 2004 presidential candidates made

by weak partisans and independents (though it had no effect

on strong partisans). These positive assessments once again

accrued to the self-morphed images without any participants

reporting recognition of their own face, suggesting that they

occurred automatically, and that any recognition that did occur

was entirely implicit. Nor is this positive effect of self-morph-

ing limited to simple liking alone. In research that directly

aligns with the neuroscientiﬁc work on automatic trust as-

sessment, DeBruine (2002) used a very similar morphing

manipulation to demonstrate that individuals playing a com-

puter-based sequential trust game were more likely to trust a

player represented by a self-morphed avatar.

While these ﬁndings demonstrate that individuals have a

clear preference for self-morphed images, major questions re-

main about the speciﬁc process underlying this effect, which

makes it hard to fully assess the extent of the potential mar-

keting relevance. In particular, a broad theoretical question is

whether the observed preference for self-morphed images was

driven by narrow perceptions of similarity (i.e., by resemblance

to the self speciﬁcally) or by a broader familiarity-based process

(i.e., by resemblance to a familiar individual, of which the self

is but one example), since the two are essentially confounded

when morphing an individual’s own face into a composite

image. As Bailenson et al. (2009, 938) state, “facial similarity

and familiarity are difﬁcult, if not impossible, to disentangle.”

This theoretical distinction has genuine practical signiﬁcance,

as while technology might enable limited use of individuals’

own faces (or elements of their own faces) in personally-tar-

geted messages, such self-endorsing strategies are highly prob-

lematic to scale to mass media advertising and may in any case

cross the privacy line for many consumers. However, subtle

manipulation of facial familiarity, should it prove possible, is

innately more scalable, assuming that the familiarity manipu-

lation relies on a cue that is likely to be broadly efﬁcacious.

From a purely applied perspective, the self-morphing ﬁnd-

ings raise the possibility that automatic facial processing

may considerably inﬂuence consumer behaviors in ways that

are currently entirely overlooked by marketers. For example,

while digital manipulation of images is certainly nothing

new to advertisers, the default focus is almost exclusively

on attractiveness, with models being “airbrushed and digi-

tally manipulated beyond the human plane” (Ives 2008, 17).

However, both the Bailenson at al. (2009) and the DeBruine

(2002) ﬁndings suggest that attractiveness (for which both

sets of authors controlled) may be too narrow a lens through

which to consider the full implications of how consumers

automatically process faces. As such, we believe that there

are both theoretical and practical reasons for attempting to

deepen our understanding of the roles played by familiarity

and similarity in the automatic evaluation of faces. In the

next section, we use the existing self-morphing ﬁndings to

frame a broader discussion outlining different theoretical

accounts that could explain how consciously imperceptible

facial cues might automatically inﬂuence how faces are per-

ceived.

A Similarity-Based Account

Both Bailenson et al. (2009) and DeBruine (2002) suggest

that automatically perceived similarity to the self underlies the

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

772 JOURNAL OF CONSUMER RESEARCH

positive preference for self-morphed images they observed.

Certainly social psychologists have long observed that indi-

viduals have a preference for people similar to themselves,

being more helpful toward them (Park and Schaller 2005),

judging them as more attractive and persuasive (Shanteau and

Nagy 1979), and even favoring them in adoption decisions

(DeBruine 2004a). Moreover, evolutionary psychology sug-

gests a speciﬁc reason to explain this preference for similarity:

since visual similarity is a cue of potential genetic kinship, then

treating similar others preferentially is one strategy to improve

the likelihood of survival of the broader family group. Con-

sistent with this logic, humans and other primates do appear

capable of recognizing unfamiliar kin (Parr and de Waal 1999;

Wells 1987). This skill likely relies on a visual form of phe-

notype matching, which is the ability to detect the outward

manifestations of a speciﬁc genotype (Hauber and Sherman

2001). This ability to automatically recognize kin certainly

suggests that (automatically) perceived facial similarity could

underlie the positive effects observed to accrue to self-morphed

images.

Familiarity-Based Accounts

While the aforementioned similarity-based account has

the potential to explain the self-morphed ﬁndings, in this

article we explore the possibility of a broader familiarity-

based account. Such an account would suggest that morph-

ing an unknown face with the face of a familiar (but ge-

netically unrelated) individual would have the potential to

similarly inﬂuence how individuals perceive the resulting

composite face. To be absolutely clear, we use the term

“familiarity-based account” broadly to encompass any pro-

cess that involves an individual being inﬂuenced by virtue

of their prior exposure to the speciﬁc face in question. In

our view, the inﬂuence of such familiarity could potentially

occur with at least three distinct levels of speciﬁcity and/

or recognition.

Perceptual Fluency-Based Account. One of the most ro-

bust ﬁndings in social psychology is that repeated exposure

to almost any stimulus increases preference for it. This mere-

exposure effect (Zajonc 1968) has proven to be highly robust

and persists even when the repeated exposures occur outside

of conscious awareness (Kunst-Wilson and Zajonc 1980).

Most accounts of the mere-exposure effect suggest that per-

ceptual ﬂuency (the subjective feeling of ease when pro-

cessing information) increases with repeated exposure to a

stimulus and enhances preference either by being misattrib-

uted to liking (Bornstein and D’Agostino 1994; Schwarz

2004) or by generating positive affect that directly enhances

stimuli evaluations (Winkielman and Cacioppo 2001). As

far as potentially explaining preference for self-morphed

images is concerned, the reality is that since the invention

of the mirror most humans are repeatedly exposed to their

own faces, and as such the positive perceptions observed

to accrue to self-morphed images could potentially be ex-

plained by the increased perceptual ﬂuency that comes from

familiarity with one’s own face.

Valence Accessibility Account. A pure perceptual ﬂuency-

based account for automatic facial processing suggests that an

unknown face morphed with a familiar face would be bene-

ﬁcially evaluated due solely to its increased familiarity, not

because any particular individual-speciﬁc meaning or valence

associated with the familiar face is perceived (i.e., a ﬂuency

process does not rely on speciﬁc recognition). From an evo-

lutionary perspective, however, one might imagine that such a

process could be maladaptive. For example, if an unknown

individual resembled a known threat, then despite their famil-

iarity, the appropriate automatic reaction would presumably be

to not trust and/or avoid this individual. More adaptive, then,

would be for implicit recognition to enable the individual to

automatically perceive some level of approach-avoid motiva-

tion. Indeed, this possibility is consistent with fundamental re-

search on primitive approach-avoid motivations, which, de-

pending on the exigencies of the current situation, are thought

to govern the deployment of predispositions to approach or

withdraw (Bargh 1997; Neumann and Strack 2000). Moreover,

visual cues are one likely trigger of these motivational systems

(Neumann and Strack 2000), which accords with the research

in neuroscience suggesting that the amygdala preconsciously

evaluates the valence of a face (Amaral 2002; Whalen 1998).

This suggests a second familiarity-orientated explanation for

participants’ observed positive preference toward self-morphs

—namely, that some level of implicit recognition of their own

faces may have led them to experience similar approach-avoid

motivations toward the self-morphs as they would with full

conscious exposure to their own image. Since the majority of

(nondepressed) individuals are positively self-disposed, auto-

matically perceived approach motivations could certainly ex-

plain the positive ratings assigned to self-morphed images. We

refer to this possible process as a valence accessibility

account—that implicit recognition of a familiar individual in

a morphed face is sufﬁcient to enable an underlying (and pre-

existing) valence judgment of the familiar individual to be

automatically perceived.

Trait Accessibility Account. A third possibility is that im-

plicit recognition of a familiar individual in a morphed image

is sufﬁcient for an individual not only to perceive a broad

approach/avoid motivation speciﬁc to the known individual but

also to automatically access more speciﬁc trait-based assess-

ments about that person from memory. If implicit recognition

does enable trait assessments to be automatically accessed from

memory, then since most individuals view themselves as being

above average on many dimensions (Alicke et al. 1995), full

implicit recognition could also explain the positive trait ratings

assigned to self-morphed images. We refer to this ﬁnal pos-

sibility as a trait accessibility account—the possibility that im-

plicit recognition of a familiar face actually enables speciﬁc

trait beliefs of the speciﬁc individual to be automatically ac-

cessed from memory.

In summary, at a high level, two broad accounts exist for

the positive preference observed to accrue to self-morphed

images (Bailenson et al. 2009; DeBruine 2002). First, con-

sistent with an innate ability to recognize our own kin, it is

possible that an automatically perceived perception of sim-

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 773

ilarity (to the self) was the primary mediating process. Sec-

ond, implicit perceptions of familiarity (as opposed to sim-

ilarity to the self speciﬁcally) may have driven the observed

effects. Such perceptions of familiarity could operate with

varying levels of actual implicit recognition. First, a ﬂuency-

based account suggests that generally perceived familiarity

(i.e., with no speciﬁc recognition of the familiar individual)

could inﬂuence preference for morphed images via increased

perceptual ﬂuency. Second, a valence accessibility account

suggests that implicit individual-speciﬁc recognition may

occur (i.e., of the self in the self-morph studies reported) at

a level sufﬁcient to enable individuals to access a basic

approach/avoid judgment (henceforth “valence judgment”)

of the individual concerned. Finally, a trait accessibility ac-

count would suggest that this implicit recognition might

actually be sufﬁcient to access speciﬁc trait information from

memory.

Of note, a similarity-based account would suggest that the

positive preference observed to accrue to self-morphed images

(Bailenson et al. 2009; DeBruine 2002) would not extend to

morphed images containing familiar faces. However, all of the

familiarity-based processes discussed above would predict that

similar preferences might also be observed for faces morphed

to contain familiar faces. As such, while in this article we do

not attempt to speciﬁcally rule out a unique similarity-based

process, a primary goal was to establish whether a broader

familiarity-based process might operate. Second, while it is not

clear that the three familiarity-based accounts discussed above

should necessarily be entirely mutually exclusive, a secondary

goal was to begin to understand the relative merits of each,

especially in the context of the facial morphing paradigm.

Research Overview

Four studies explore the effects of morphing unfamiliar faces

with the faces of highly familiar celebrities (Tiger Woods and

George W. Bush). Consistent with a familiarity-driven process

(and inconsistent with a pure similarity-based explanation),

these studies demonstrate that adding imperceptible celeb-

rity facial cues to pictures of unfamiliar people increases

participants’ perceived trustworthiness in (studies 1 and 2A)

and reported likelihood to buy from (study 3) the individuals

pictured. Furthermore, this effect appears consistent with a

valence accessibility account; that is, it involves a level of

implicit recognition that appears to be sufﬁcient to enable

a broad approach/avoid evaluation of the underlying celeb-

rity to be perceived (study 2B). However, it is inconsistent

with accounts based on trait accessibility (study 2A) or per-

ceptual ﬂuency (studies 2B and 3).

STUDY 1: TRUSTING A CELEBRITY-

MORPHED IMAGE

Study 1 was designed to provide an initial test of whether

subtly incorporating elements of a familiar celebrity image

into an unfamiliar face can lead the resulting composite

image to be perceived as more trustworthy than the original

face. To do this we compare the trustworthiness of an un-

familiar face to the same unfamiliar face digitally blended

with the face of Tiger Woods. Tiger was chosen because we

expected him to be highly familiar to our participants. Of

note, all data was collected before the Tiger Woods scandal

materialized. We chose to focus on trust for several reasons.

First, much of the neuroscientiﬁc work on automatic trait

assessment has identiﬁed perceptions of trust as being sig-

niﬁcantly determined by facial properties (e.g., Engell et al.

2007) and to proxy an underlying valence judgment (Oos-

terhof and Todorov 2008). Second, trust has long been iden-

tiﬁed as a critical marketing variable, inﬂuencing ﬁrm met-

rics as diverse as salesman performance (Doney and Cannon

1997), the efﬁcacy of celebrity endorsers (Ohanian 1990),

and ultimately perceptions of the brand itself (Ambler 1997;

Chaudhuri and Holbrook 2001).

Design and Method

Study 1 randomly assigned 109 undergraduate partici-

pants to either a Tiger-morph condition or a control-morph

condition in a single-factor between-subjects design. Par-

ticipants were instructed to spend a few moments looking

at a picture of a face and then to record their impressions

of the trustworthiness of the individual pictured on an 11-

point scale anchored from ⫺5 (well below average) to ⫹5

(well above average). Participants were asked to not over-

think the exercise and to rely on their gut instincts and

immediate reactions. In the Tiger-morph condition, a com-

posite face was created by digitally blending a stock model

face with the face of Tiger Woods. In the control-morph

condition, the same stock model face was digitally blended

with the face of an individual chosen to be of approximately

similar age and attractiveness to Tiger Woods. Using a con-

trol image that has itself been created using a digital blending

process is required to ensure that any observed effects could

not be attributable to the effect of morphing alone, since

composite faces (which are more symmetrical) have been

shown to be perceived as being more attractive than the

source faces from which they are constructed (Langlois and

Roggman 1990). Furthermore, “the what is beautiful is

good” stereotype (Dion, Berscheid, and Walster 1972) would

suggest a possible link between such morph-induced at-

tractiveness and subsequent trait ratings. As such, to ensure

both facial morphs were indeed perceived as being similar

in attractiveness, we conducted a pretest (N p 75) utilizing

an 11-point scale anchored from ⫺5 (very unattractive) to

⫹5 (very attractive) that revealed no differences in the per-

ceived attractiveness of the two faces (M

Tiger_morph

p 3.64,

control_morph

p 3.68; t(73) p .17, p p .86). The original

and composite faces are shown in ﬁgure 1.

Both composite images (henceforth the “Tiger-morph” and

“control-morph” images) were created using Abrosoft Fanta

Morph Pro 4 software and were composed of 65% of the

unfamiliar stock model face and 35% of either Tiger Woods’s

face or the control face. The morphing process involves mark-

ing corresponding points at standard landmarks on both faces

(such as the corners of the eyes and mouth), which allows

the software to generate a composite face from a weighted

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

774 JOURNAL OF CONSUMER RESEARCH

FIGURE 1

CONSTRUCTION OF FACES USED IN STUDY 1

average of landmark-speciﬁc parameters of the source faces.

While prior morphing research has demonstrated that indi-

viduals are unable to consciously recognize themselves in

composite images containing 40% of their own face (Bailen-

son et al. 2009; DeBruine 2004b), we conservatively chose

to use the 65 : 35 ratio in order to minimize the chance of

conscious recognition.

Once participants had provided their trustworthiness rat-

ing, they completed a funnel debrieﬁng (Bargh and Chart-

rand 2000) that ﬁrst asked if there was anything odd or

unusual about the face pictured, then asked if the face re-

minded them of anyone they knew, before ﬁnally providing

an open-ended opportunity to report if the face looked like

anyone famous.

Results and Discussion

First, we considered the funnel debrieﬁng data to assess

possible differences in how the faces were perceived be-

tween conditions. While 16 participants reported that the

face they viewed was odd, there was no difference in the

percentage reporting this between the control-morph con-

dition (13%) and the Tiger-morph condition (17%; x

(1) p

.34, p p .56). Of greater importance, none of the expla-

nations for the faces looking odd betrayed any underlying

suspicion of the morphed nature of the faces pictured. In

fact, the explanations offered were universally entirely un-

related to the experimental manipulation and included such

reasons as “smiling too much” and “eyes a little closed.”

As far as general (conscious) perceptions of familiarity were

concerned, we observed no differences across the two con-

ditions, with 22% of participants in both conditions indi-

cating that the face they viewed reminded them of someone

they knew (x

(1) p .003, p p .96). However, none of the

individuals (in either condition) who suggested the face was

familiar indicated that it resembled Tiger Woods; instead,

they tended to suggest that the face resembled someone they

had a personal relationship with. As such it appears that, at

least as far as conscious recognition is concerned, there was

no difference in either general perceptions of familiarity or

speciﬁc recognition of Tiger Woods between the two com-

posite faces.

However, despite participants in the two conditions re-

porting the two images as being equally familiar, their re-

ported trust ratings revealed a different pattern. In fact par-

ticipants in the Tiger-morph condition rated the face as being

more trustworthy (M p 1.75) than did participants in the

control-morph condition (M p 0.95; t(107) p ⫺2.2, p

.05). Thus it appeared that subtly incorporating elements of

a highly familiar face into an otherwise unfamiliar image

increased perceived trust in the pictured individual. These

initial data thus conceptually mirror the prior research, which

has demonstrated positive changes in perceptions of the

trustworthiness of self-morphed images (DeBruine 2002,

2004b). However, unlike the self, which is both similar and

familiar, Tiger Woods is highly unlikely to be on average

similar in appearance to our participant population. As a

result, these data are more consistent with a broader famil-

iarity-based process as underlying the positive perceptions

of the trustworthiness of the Tiger-morphed image.

While our chosen focus was trust, a natural question to

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 775

FIGURE 2

FACES USED IN STUDY 2

ask is whether the observed effect is unique to trust or might

generalize to other trait assessments. To explore this question

we ran a replicate experiment with 35 MBA students who

rated trustworthiness, liking, and athletic ability of the same

two faces used in the main study. The only difference in

procedure was that 1–10 scales were used (as opposed to

⫺5to⫹5). Not only did these data reveal a (marginally)

signiﬁcant replication of the prior trust result (M

Tiger

p 5.3,

control

p 4.0; t(31) p ⫺1.9, p ! .07) but a similar result

was also observed for reported liking (M

Tiger

p 5.3, M

control

p 3.8; t(31) p ⫺2.1, p ! .05). Furthermore, the trust and

liking ratings were themselves very highly correlated (r p

.96, p

! .01), with a signiﬁcant number of participants ac-

tually reporting the exact same score for both trust and lik-

ing. However, we observed no difference in reported athletic

ability between the two face conditions (M

Tiger

p 7.1, M

control

p 7.4; t(31) p 0.51, p 1 .6), and we also observed no

correlation between the athletic ability ratings and the ratings

of either trust or liking (both p

1 .4).

These follow-up data enable us to draw a number of ad-

ditional conclusions. First, the fact that participants did not

report the Tiger-morph to be more athletic provides initial

evidence against a full trait accessibility account, as presum-

ably if implicit recognition was sufﬁcient to enable full trait

accessibility from memory, then this would have led to higher

athletic ratings of the Tiger-morph. This conclusion is ad-

mittedly a little speculative, however, as we did not measure

our participant’s beliefs about the athletic ability of Tiger

Woods, and it is certainly possible (perhaps due to golf not

being perceived as being a physically intense sport) that our

participants simply did not perceive Woods to be especially

athletic. Second, the reported liking data clearly suggest that

the observed positive effect is not speciﬁc to trust and likely

generalizes to a variety of general personality assessments.

We believe that this result is consistent with a valence ac-

cessibility account, as it appears likely that automatically per-

ceived positive valence would manifest via individuals mak-

ing positive assessment of broad personality-type variables

rather than simply trustworthiness alone (e.g., liking or hon-

esty).

STUDY 2A: EXPLORING A TRAIT

ACCESSIBILITY EXPLANATION

Study 2A was designed to build on study 1 in a number

of ways. First, in order to generalize the results of study 1,

we switch to a different highly familiar ﬁgure: George W.

Bush. Second, in order to address possible concerns that the

results in study 1 might have been driven primarily by the

morph-control image (i.e., by some idiosyncratic negative

feature of the control face), we include two control conditions

in study 2: the original unaltered model face as well as a

control-morph using an individual of similar age and attrac-

tiveness to George Bush. Finally, while no participants ex-

plicitly recognized Tiger Woods in study 1, implicit recog-

nition may well have taken place. As previously discussed,

such implicit recognition might have enabled either a broad

valence judgment of Woods to be accessed or even for speciﬁc

trait beliefs held about him to be perceived in the morphed

image (though the athletic rating data provided preliminary

data suggesting that this is unlikely). To better rule out the

latter possibility, in study 2A we also collected participants’

actual ratings of the trustworthiness of George W. Bush.

Design and Method

Study 2 randomly assigned 179 undergraduate partici-

pants to either a Bush-morph condition, a control-morph

condition, or an unaltered-face condition in a single-factor

between-subjects design. In the two morph conditions, com-

posite faces were created by blending the stock model face

(same face as study 1) with either 35% of the face of George

W. Bush or 35% of a control face chosen to be of similar

age and attractiveness to Bush. Participants in the unaltered

face condition viewed the original unaltered stock model

face. As in study 1, a pretest (N p 66) suggested that the

two morphed faces were equally attractive (M

Bush_morph

3.62, M

control_morph

p 3.73; t(64) p .53, p p .60). The un-

altered face and both composite faces used are shown in

ﬁgure 2.

All other aspects of the experiment were identical to study

1 except that in a supposedly unrelated study participants

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

776 JOURNAL OF CONSUMER RESEARCH

subsequently rated the trustworthiness of six celebrities (in-

cluding George W. Bush) on the same 11-point scale.

Results and Discussion

First, we considered the funnel debrieﬁng data and once

again observed no reliable differences in conscious per-

ceptions of familiarity across the three conditions (per-

centage reporting face resembled someone they knew:

Bush-morph 28%; control-morph 19%; neutral image 15%;

(2) p 2.50, p p .29). As far as speciﬁc recognition of

George Bush was concerned, none of the participants in

any condition suggested that the face they viewed re-

minded them of George Bush.

Turning to our primary dependent measure of interest,

there was once again an overall effect of face condition on

reported trust (F(2, 176) p 4.56, p

! .02). A planned com-

parison revealed that this effect was primarily driven by

participants in the Bush-morph condition rating the face as

being signiﬁcantly more trustworthy (M p 2.16) than either

participants in the unaltered face condition (M p 1.27) or

the control-morph condition (M p 1.30; t(176) p ⫺3.02,

! .01). There was no difference in perceived trust between

the control-morph and unaltered-face conditions (t(176) p

.08, p

1 .90). Next we considered whether participants’ trust

ratings in the Bush-morph condition were related to their

actual beliefs about the trustworthiness of George Bush. This

did not appear to be the case; in fact in the Bush-morph

condition, ratings of trustworthiness of the composite face

were essentially uncorrelated with actual ratings of the trust-

worthiness of George Bush (r p .07, p p .62). Thus, in

addition to not being consciously recognized, any implicit

recognition that may have been taking place was apparently

not sufﬁcient for speciﬁc beliefs about Bush’s trustworthi-

ness to be perceived in the morphed image.

While these data rule out a trait accessibility account, that

is, the possibility that implicit recognition enables speciﬁc

trait assessments to be accessed from memory, they do not

speak to the alternative possibility that an intermediary level

of implicit recognition might exist, one sufﬁcient to perceive

an overall valence judgment of the individual in question (i.e.,

whether to approach or avoid them) but insufﬁcient to access

preexisting beliefs about speciﬁc traits. Recall that, when as-

sessing faces believed to be unknown, trust ratings have been

demonstrated to provide a good approximation of automati-

cally made valence evaluations of faces (Oosterhof and To-

dorov 2008). By this account, the positive trust ratings as-

signed to the Bush-morph might have reﬂected automatically

perceived perceptions of positive valence (i.e., that an ap-

proach motivation was induced). However, it was impossible

for us to assess an overall valence explanation for the study

2A data, as we had no independent measure of whether Bush

might trigger an approach or avoid motivation in our partic-

ipants. Certainly, despite his rather mixed reputation (both

across traits and between individuals), we would also expect

Bush to be inextricably associated with the primary role for

which he is known, that of president of the United States

(McCracken 1989). As such, his close association with the

presidency suggests that implicit recognition of him in a com-

posite image might be expected to automatically cue ap-

proach-consistent concepts such as freedom and safety, thus

leading to a positive valance appraisal of the composite image.

However, rather than simply speculate about the possibility

of a positive approach motivation, we designed study 2B to

explore if we could measure this directly.

STUDY 2B: EXPLORING A VALENCE

ACCESSIBILITY EXPLANATION

Study 2B had two main goals. First, we wanted to in-

vestigate whether the results of study 2A could be attributed

to the Bush-morph image inducing an approach motivation

in our participants. Second, we wanted to rule out a pure

ﬂuency-based explanation for the results of studies 1 and

2A (i.e., that the presence of the Woods or Bush facial cues

led to increased ﬂuency in processing the morphed images,

with this ﬂuency then being misattributed to the trust and

liking ratings). The study utilized a modiﬁed version of a

methodology used by Mehta and Zhu (2009) to evidence

that the color red induces an avoid motivation while the

color blue invokes an approach motivation. Speciﬁcally,

these authors demonstrated that a blue/(red) background fa-

cilitated participants solving anagrams based on approach-

/(avoid-) related words. In this study, instead of manipulat-

ing the background color, we located the appropriate

anagrams directly beneath either the Bush-morph or the con-

trol-morph image from study 2A.

Design and Method

Four hundred and forty-ﬁve participants from a general

online panel participated in this study for payment. The

study was designed as a 2 (face presented: Bush-morph vs.

control-morph) # 2 (solve time: control vs. approach related

words) mixed design, with the second factor being the

within-subject factor. Participants were randomly assigned

to one of the two face-presented conditions, and the study

was administered via computer. Participants were told that

the researchers were interested in how people solve ana-

grams in the presence of another person and that in this ﬁrst

study they would simply be asked to solve a series of an-

agrams presented underneath pictures of faces. It was also

explained that the ﬁrst one or two letters of the target word

would be underlined in the letter string and an example

(ATSKABR

EF for which the solution is BREAKFAST) was

presented. Finally, participants were instructed to write

“don’t know” if they could not solve a given anagram within

2 minutes.

Next, participants proceeded to attempt to solve nine an-

agrams, each of which were presented on a separate screen.

Six of the anagrams represented control words (i.e., count,

ranch), while three were associated with an approach mo-

tivation (i.e., advance, adventure). The presentation order of

the anagrams was randomized. Depending on the condition,

each anagram appeared on the screen directly below either

the Bush-morph or the control-morph image from study 2A.

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 777

Results and Discussion

First, for each participant, we calculated their average

solve time for control and approach words. Only correctly

solved anagrams were included in the average calculations,

and data from participants who got one or fewer words

correct (in either category) were discarded. As a result, data

from 45 participants were excluded from further analyses,

leaving useable data for 397 participants.

These computed average solve times were then used in

a 2 (face presented: Bush-morph vs. control-morph) # 2

(solve time: control words vs. approach words) repeated

measures ANOVA, which revealed a signiﬁcant interaction

(F(1, 395) p 4.50, p

! .04). A planned comparison revealed

that participants who viewed the Bush-morph solved the

approach-related anagrams signiﬁcantly faster (M p 14.2

seconds) than did those who viewed the control-morph (M

p 16.9 seconds, p

! .04) but that there was no difference

in the speed with which the control anagrams were solved

Bush-morph

p 15.2 seconds, M

control-morph

p 15.1 seconds; p

p .95). These data thus build on the ﬁndings from study

2A by revealing that exposure to the Bush-morph image

appeared to induce an approach motivation in participants.

Put differently, participants appeared to implicitly recognize

Bush in the morphed image to an extent sufﬁcient to access

a positive valence judgment. As such, these data support a

valence accessibility account, that is, that the positive trust

ratings observed to accrue to the Bush-morph in study 2A

were attributable to an automatically induced approach mo-

tivation. Moreover, these data are very hard to reconcile

with a pure ﬂuency-based account, as not only is it rather

problematic to explain why increased ﬂuency in processing

an image would inﬂuence the solve times of anagrams po-

sitioned adjacent to that image, but we can think of no reason

why such an effect would only pertain to approach-related

words.

STUDY 3: LIKELIHOOD TO BUY FROM A

TIGER WOODS–MORPHED IMAGE

The ﬁrst three studies demonstrated that blending a fa-

miliar (but not similar) face into an otherwise unfamiliar

face can produce a composite face that is rated as being

more trustworthy and likeable than the original. This effect

on perceived trustworthiness obtained despite a complete

lack of apparent recognition of the presence of the familiar

image in the composite image. The original goal of study

3 was to begin to explore whether this “hijacking” of fa-

miliarity has the potential to affect behaviors of speciﬁc

interest to marketers. As such, in study 3 we again use a

Tiger-morphed image to examine whether the morphing pro-

cedure could inﬂuence perceived likelihood to buy from a

pictured salesperson. Rather serendipitously, data were col-

lected in two time periods: both before and at the peak of

the Tiger Woods scandal. This enabled us to further explore

the valence accessibility account we have argued underlies

the results to date and to rule out explanations based on

facial structure or expression alone.

Design and Method

Study 3 randomly assigned 319 undergraduate partici-

pants to either a Tiger-morph condition or a control-morph

condition. In the Tiger-morph condition, a composite face

was created by digitally blending a stock model face with

the face of Tiger Woods. In the control-morph condition,

the same stock model face was digitally blended with the

face of an individual chosen to be of approximately similar

age and attractiveness to Tiger Woods. As before, Tiger and

the similarly attractive/aged face represented 35% of the

respective morphed images. The stock model face in this

study was a neutral image of a salesperson pictured from

the chest upward. The original salesman image and both

composite faces are shown in ﬁgure 3. Participant instruc-

tions were identical to those of studies 1 and 2A except that

instead of rating trustworthiness they were asked to rate how

likely they would be to buy from the pictured salesman on

an 11-point scale anchored from ⫺5 (very unlikely) to ⫹5

(very likely).

The data were collected in two tranches. The ﬁrst tranche

(167 participants) was collected in the last week of Septem-

ber 2009, before the Tiger Woods scandal emerged. The

second tranche (152 participants) was collected in the second

week of December 2009, when the Tiger Woods scandal

was at its peak and was almost completely dominating the

media. Thus study 3 was a 2 face presented (Tiger-morph

vs. control-morph) # 2 timing (pre-scandal vs. post-scan-

dal) between-subjects design.

Results

Results indicated no main effect of face presented or tim-

ing but a signiﬁcant interaction between the two (F(1, 318)

p 12.01, p

!. 001). As shown in ﬁgure 4, while pre-scandal

participants indicated that they were more likely to buy from

the Tiger-morph than from the control-morph (M

Tiger-morph

1.46, M

control-morph

p 1.04), this pattern reversed itself post-

scandal (M

Tiger-morph

p ⫺.11, M

control-morph

p 1.02). Of note,

given the particular nature of the scandal, there were no

effects of gender.

Discussion

Despite a complete lack of conscious recognition of Tiger

Woods, assessments of the likelihood to buy from the Tiger-

morphed salesman appeared to be signiﬁcantly inﬂuenced by

the precipitous decline in Woods’s real world reputation. This

result is consistent with an associative learning/evaluative con-

ditioning view of the scandal (DeHouwer, Thomas, and Bae-

yens 2001). Speciﬁcally, the (highly negative) scandal appears

to have acted as an affective stimulus, which, since it was paired

with Tiger Woods, was sufﬁcient to change valence judgments

of him. Furthermore, implicit recognition of Tiger was appar-

ently sufﬁcient to access these valence judgments. Of note, the

fact that we observed a valence judgment being inﬂuenced by

behaviors undertaken by the individual in question strongly

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

778 JOURNAL OF CONSUMER RESEARCH

FIGURE 3

FACES USED IN STUDY 3

FIGURE 4

WILLINGNESS TO BUY FROM SALESMAN AS A FUNCTION OF

FACE PRESENTED AND SCANDAL TIMING

suggests that this judgment is related to the actual identity (i.e.,

Tiger Woods) of the known individual in the morphed image.

This distinction is important, as while study 2B revealed

that the Bush-morph appeared to induce an approach moti-

vation in participants, the data were entirely silent as far as

the actual cause of that motivation is concerned. Put differ-

ently, when discussing implicit recognition of the individual

in the morphed images (i.e., Woods or Bush), to this point

we have essentially made an assumption that a valence ac-

cessibility account involves implicit recognition of the celeb-

rities’ facial properties sufﬁcient for the perceiver to access

a valence judgment of that speciﬁc individual in particular.

However, an alternative possibility is that there is something

about the facial structure of these two individuals (or even,

possibly, of their speciﬁc expressions in the photos of them

we used) that in and of itself drove our results to date (i.e.,

independent of their unique identities). For example, the pos-

sibility exists that both Bush’s and Woods’s facial features

ﬁt a genetic pattern generally perceived to appear more trust-

worthy or approachable. However, since in study 3 we utilized

the exact same Tiger-morph image at two points in time, the

observed difference in participant ratings across these two

points in time cannot be solely due to some idiosyncratic

feature of Woods’s facial structure itself or of his speciﬁc

facial expression in the photo we used to create the morph.

Also of note, these results mirror study 2B in that they are

very hard to reconcile with a pure ﬂuency-based process. In-

deed, we note that a ﬂuency-based account might actually pre-

dict the opposite result as, despite Woods being arguably the

world’s most recognizable sports celebrity, the obsessive cov-

erage of the scandal presumably could only have led partici-

pants in the post-scandal condition to be exposed to additional

images of Tiger Woods. This repeated exposure should, all

other things aside, have increased their ﬂuency in processing

the Tiger-morph, leading to enhanced preference. Certainly,

even though one might argue that the relatively short duration

of the scandal might be insufﬁcient to make much positive

difference to an individual’s ﬂuency in processing Woods’s

image, it is problematic for a ﬂuency account to explain a

decline in how Tiger was rated. Thus these data not only suggest

that participants were implicitly recognizing Tiger Woods to a

degree sufﬁcient to access an overall valence judgment of him

but that this implicitly perceived valence was sufﬁcient to over-

come the possible opposing effect of ﬂuency.

GENERAL DISCUSSION

The current research provides the ﬁrst demonstration of

the important ﬁnding that our impressions of unfamiliar

faces can be inﬂuenced by digitally morphing them with

those of highly familiar ﬁgures. Study 1 showed that digi-

tally blending an unfamiliar face with 35% of the face of

Tiger Woods resulted in a composite face that was perceived

as more trustworthy (and in the replicate more likeable) than

the original face. This effect occurred despite 100% of our

participants failing to identify any resemblance between the

Tiger-morphed face and Tiger Woods himself. Study 2A

both replicated this result with a different celebrity (George

W. Bush) and additionally demonstrated that participants’

ratings of the trustworthiness of the Bush-morphed image

were uncorrelated with their consciously held beliefs about

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 779

Bush’s actual trustworthiness. These data were thus incon-

sistent with a trait accessibility account, as it did not appear

that participants were implicitly recognizing Bush to an ex-

tent sufﬁcient to automatically perceive speciﬁc trait as-

sessments of him.

Study 2B built on this result by demonstrating that the

Bush-morph image appeared to induce an approach moti-

vation in participants, a result consistent with a valence ac-

cessibility account, that is, the idea that implicit recognition

is sufﬁcient only to access an overall valence judgment of

the known individual concerned (Todorov and Engel 2008).

Study 3 expanded on this ﬁnding by revealing that a marked

decline in Tiger Woods’s overall reputation appeared to neg-

atively inﬂuence how a Tiger-morphed composite face was

perceived, a result that suggests that overall valence judg-

ments of morphed faces are (at least partially) determined by

actual perceptions of the speciﬁc individual concerned rather

than by some non-individual-speciﬁc aspect of their facial

expression or facial structure. A second contribution of studies

2B and 3 was that they served to rule out a pure ﬂuency-

based explanation, as it is very hard to reconcile a ﬂuency-

based account with the pattern of data we observed in these

studies.

It is important to note that all these results were obtained

without a single participant in any condition (celebrity-morph

or otherwise) ever consciously suggesting resemblance to the

celebrity in the composite image. As such, it is worth high-

lighting that all these results were observed using a manip-

ulation of facial imagery that is subtle enough to universally

elude conscious awareness. Also worthy of note, with the

exception of the post-scandal likelihood-to-buy rating for the

Tiger-morph in study 3, all the ratings our participants pro-

vided (for all the faces presented) were positive in an absolute

sense (i.e., greater than the midpoint of the scale). This result,

which may seem a little probabilistically surprising, is actually

broadly consistent with the work of Cacioppo and colleagues

on the positivity offset (Cacioppo, Gardner, and Bernston

1999). Cacioppo argues that neutral stimuli actually tend to

induce a weak approach motivation, which he suggests serves

to motivate an organism to explore and learn about the en-

vironment (since novel stimuli may otherwise not be ap-

proached). As such, the ﬁnding that truly unknown faces (e.g.,

our control faces) are positively rated is consistent with the

idea that exposure to them induced a weak approach moti-

vation in our participants (and thus is consistent with a valence

assessment process more generally).

Theoretical Contributions

The current results contribute to the nascent facial morph-

ing literature and to the wider facial processing literature in

several important ways. First, recall that both Bailenson et

al. (2009) and DeBruine (2002, 2004b) demonstrated that,

despite a complete lack of conscious recognition, self-mor-

phed images beneﬁted from enhanced preference compared

to the original unaltered source face. Both sets of authors

primarily attributed this result to automatic perceptions of

similarity. In our studies, we conceptually replicate these ﬁnd-

ings when we incorporate celebrity faces rather than partic-

ipants’ own images into composite faces. Since the celebrities

we chose (Tiger Woods and George W. Bush) are unlikely

to be similar to our participants, our results are clearly in-

consistent with a perceived similarity account (or more pre-

cisely with a similarity-to-the-self account). While our data

do not explicitly rule out similarity as being a unique factor

in these prior studies, they do suggest that familiarity-based

accounts may have more potential to parsimoniously explain

preference for both self-morphed and celebrity-morphed im-

ages.

Second, turning to familiarity-based accounts speciﬁcally,

our work suggests that some measure of implicit recognition

likely contributes to how morphed faces are perceived (be they

self-morphed or celebrity-morphed). While study 2A demon-

strated that participants rating a Bush-morph image were unable

to access speciﬁc beliefs about George Bush’s trustworthiness,

study 2B revealed that a broader approach/avoid type evaluation

was possible. Together, studies 2A and 2B combine to sug-

gest that implicit recognition, at least in our facial morph-

ing paradigm, occurs at an intermediary level of speciﬁcity.

It appears sufﬁcient to access an overall valence judgment

of the individual in question (as in study 2B) but insuf-

ﬁcient to access more speciﬁc trait-based ratings that may

be held and that would be reportable with full conscious

recognition (as in study 2A). Study 3 further supported

this valence accessibility account by demonstrating that an

actual behavior undertaken by a speciﬁc individual (i.e.,

Tiger Woods) led to a subsequent change in how an un-

known face (morphed to include Woods’s face) was per-

ceived. As such, these ﬁndings resonate with the neuro-

science literature that suggests that a primary function of

the amygdala is to automatically provide a high-level va-

lence-based evaluation of agents prior to interacting with

them (Amaral 2002; Todorov and Engel 2008; Whalen

1998).

Finally, we believe that our data directly contribute to the

rather thin discussion in the literature concerning precisely

how these valence-based evaluations are constructed. As we

emphasized in our introduction, there is surprisingly little

research on the nature of the informational building blocks

from which the implicit evaluations are presumed to be built.

For example, does a speciﬁc face induce an approach mo-

tivation (and thus garner high trustworthy ratings) due to a

subtle (and hard to control for) aspect of its particular ex-

pression, due to its chronic structural features (which may

ﬁt a genetic pattern indicative of trust), or because it resem-

bles a speciﬁc known individual for whom an overall pos-

itive evaluation exists? While our research is clearly insuf-

ﬁcient to fully characterize what we suspect is a complex

and interdependent set of inferential processes, we do note

that in study 3 the Tiger-morphed face was identical in both

timing conditions. Thus facial expression and idiosyncratic

structural features of the face were both perfectly controlled

for. As such, study 3 suggests that the inferential process

involved appeared to involve sufﬁcient implicit recognition

of a speciﬁc individual (Tiger Woods) for an individual-

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

780 JOURNAL OF CONSUMER RESEARCH

speciﬁc valence judgment of him to be perceived. Thus,

while in no way ruling out the potential for facial expressions

and structural facial features to also contribute to automatic

judgments, our research does identify that the speciﬁc role

resemblance to a known individual can play in determining

automatic valence judgments. Hence, while our experimen-

tal paradigm may have been the digital morphing of faces,

a broader implication is that our automatic attentional re-

sources appear to be extremely sensitive to cues of facial

familiarity. We believe that future research should increas-

ingly focus on better understanding the full range of infer-

ential processes underlying the implicit processing of facial

information. Indeed, this challenge appears to represent a

particularly suitable arena in which neuroscientists and de-

cision scientists could fruitfully collaborate.

Implications for Marketers

At a high level, we believe that the most important im-

plication of our work is that it suggests a general need for

marketers to pay more attention to the full potential con-

sequences of how facial information is automatically pro-

cessed. With facial imagery being such an ever-present cue

in the world of marketing communications, the simple re-

ality is that there is a tremendous variety of circumstances

and contexts where automatic processing of facial stimuli

may inﬂuence consumer behaviors of interest. In our view,

while the inﬂuence of attractiveness and certain emotional

expressions have received some speciﬁc attention, the po-

tentially broader implications of facial familiarity have been

largely unexplored by marketers.

More speciﬁcally, these data suggest a novel way for

marketers to use celebrities and highly familiar ﬁgures. In-

stead of solely placing them front and center as endorsers,

an alternative way to utilize them is to subtly morph their

faces with those of otherwise unfamiliar stock model im-

ages. Indeed, mirroring the success of “forgotten spokes-

persons” (Weisbuch and Mackie 2009), such a strategy may

even have the potential to outperform explicit use of celeb-

rities in some circumstances. Prior research has demon-

strated that the persuasive impact of spokespersons can be

enhanced when prior exposure cannot be recalled, as this

facilitates misattribution of the perceptual ﬂuency experi-

enced while processing the endorser (Weisbuch, Mackie, and

Garcia-Marques 2003). Put differently, when the endorser

is correctly identiﬁed, any effect of ﬂuency is more likely

to be correctly attributed to their fame and familiarity. Sim-

ilarly, we believe it is possible that implicitly perceived

valence judgments of individuals are likely candidates for

misattribution given the absence of conscious awareness of

their source.

As well as this possible valence misattribution, another po-

tential advantage of using celebrity-morphed images is that the

lack of conscious recognition of the celebrity, which we rou-

tinely observed, might also mean persuasion knowledge (Fries-

tad and Wright 1994) is less likely to be activated, which, in

and of itself, might lead to greater potential for successful

persuasion. As such, we believe that there are multiple reasons

to believe that digital morphing may offer marketers a way to

capitalize on celebrities’ familiarity without paying all the costs

(be they to the persuasion process or perhaps ﬁnancial) of the

attendant fame. We note, however, that the legal ramiﬁcations

of using celebrities in morphed images are likely to be rather

complex. While a celebrity owns the rights to the use of his

or her image, it is not entirely clear how these rights might

extend to composite faces containing a minority proportion of

their face, especially when they are not consciously recogniz-

able in the composite (which would also make policing prob-

lematic).

Perhaps, however, some of the beneﬁts of using celebrities

in morphed images might be obtainable without actually

using the actual celebrity in question. Indeed, an intriguing

implication of our work involves the potential for using

celebrity look-alikes in morphed images. If look-alikes are

used directly in an advertising context, then unless their

resemblance to the celebrity of interest is unusually close,

it is likely that consumers will be consciously aware that

the look-alike is not in fact the real deal. This conscious

perception of illegitimacy may interfere with the possible

beneﬁts that might otherwise be expected to (automatically)

accrue from the resemblance. However, if the look-alike is

used in a morphed image, then an interesting possibility is

that this may offer a way to capitalize on the beneﬁts of the

resemblance without the attendant risks of conscious cor-

rection.

Limitations and Directions for Future Research

As we noted earlier, when evaluating unknown faces, rat-

ings of trust have previously been demonstrated to be a good

approximation for an overall valence evaluation of a face

(Oosterhof and Todorov 2008). Put differently, when asking

individuals to rate the trustworthiness of an unfamiliar face,

the answer they provide appears to be signiﬁcantly deter-

mined by an automatically perceived approach/avoid eval-

uation. In essence, then, these trust ratings might be innately

different in nature (they certainly are in terms of their source)

from perceptions of trustworthiness of known people, and

they really are better conceptualized as a proxy for an overall

evaluation of valence. As such, it is important not to over-

state the extent to which such evaluations might inﬂuence

consumer behavior in the way that trust (as traditionally

conceptualized and measured) has been shown to do. How-

ever, we believe that study 3 clearly demonstrates the po-

tential for these overall evaluations to affect more than sim-

ple trust ratings and to directly inﬂuence consumer behavior.

Certainly, however, one fruitful avenue for future research

is to begin to better delineate the circumstances in which

these automatic evaluations of valence can inﬂuence down-

stream consumption behaviors. To give one example, imag-

ine a typical print advertisement that makes some form of

product claim and that includes one or more stock model

images. The question is, would the positive valence that

could be obtained by morphing the model face(s) with a

known ﬁgure inﬂuence the extent to which the product claim

is attended to and/or believed?

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 781

A second (and related) limitation is that we are some dis-

tance from being able to provide a complete theoretical ac-

count of how morphed faces are perceived and processed. In

particular, in the current research, our primary focus has been

on exploring the extent to which individuals can implicitly

recognize celebrity images in composite faces. We have

placed less emphasis on ﬂuency explanations, although we

note that the results of studies 2B and 3 are entirely incon-

sistent with a purely ﬂuency-based process. That said, since

the manipulation in study 3 was naturally occurring, the re-

sulting consequences for internal validity require us to be

conservative in how we interpret the results. We note, how-

ever, that in very recent research, Verosky and Todorov (2010)

demonstrated a conceptually similar result in a laboratory

environment. In particular, these authors found that the pos-

itive or negative behaviors they previously associated with

novel faces appeared to transfer (in broad valence terms) to

composite faces including the aforementioned novel faces. As

such, it seems unlikely that ﬂuency is the primary mechanism

inﬂuencing how composite faces are automatically perceived.

Our view, however, is that it is unlikely to play no role what-

soever and that future research could usefully examine it in

isolation. One idea such research might pursue is to examine

the consequences of digitally combining multiple unfamiliar

faces into one composite face. We believe that this might re-

sult in a composite face that is perceived to be more familiar,

thus enabling familiarity to be studied while perfectly con-

trolling for valence effects (since all the source faces would

be unknown.)

A ﬁnal limitation relates to the valence judgments we

have argued underlie the effects we have observed. We have

argued that our data support the view that when evaluating

an apparently unknown face, if that face subtly resembles

a known individual (which we achieve via morphing), then

a valence judgment of that known individual automatically

colors ratings of the apparently (and consciously perceived

to be) unknown face. Indeed, one of the major contributions

of this work is revealing that such valence accessibility is

possible. What this article does not examine, however, is

the process via which such summary valence judgments are

formed. For example, while study 2B empirically demon-

strated that exposure to the Bush-morph appeared to invoke

an approach motivation in our participants, it had nothing

to say about why this might be. Indeed, the reader might

legitimately wonder why such an apparently polarizing ﬁg-

ure would invoke an approach motivation. While we believe

understanding the origins of summary valence evaluations

is a (highly complex) issue for future research to explore,

one thing we would stress is that the approach and avoid

distinction is a fundamental driver of motivation across all

life forms (Elliot and Covington 2001) and is perhaps best

conceptualized as a primal big picture distinction between

stimuli likely to be beneﬁcial or harmful (Elliot 2006). As

we speculated earlier, our belief is that Bush’s automatic

association with the presidency is a likely candidate for

explaining the approach motivation we observed, but it re-

mains exactly that, speculation. We suggest that future re-

search on valence judgments should focus both on the pri-

mary roles people are associated with (particularly relevant

for celebrities whose fame relates to acting) as well as the

actual behaviors they carry out in their real lives (which

were clearly relevant in study 3). Indeed, it is worth noting

that one beneﬁt of the digital morphing technique we used

is that it actually provides a method via which valence as-

sessments of known individuals can be implicitly measured.

Conclusion

From salesmen to print ads, the human face is almost

ever-present across the broad spectrum of marketing com-

munication efforts. Furthermore, considerable research now

suggests that humans, as beﬁts innately social animals, have

evolved with signiﬁcant mental capabilities designed solely

to facilitate rapid implicit processing of facial information.

However, despite this nontrivial conﬂuence between mar-

keting practice and neural circuitry, little attention has been

paid to understanding the possible ramiﬁcations such au-

tomatic facial evaluations may have in marketing contexts.

While marketers narrowly focus on digitally manipulating

attractiveness, our results suggest that manipulations of fa-

miliarity may have similar potential to inﬂuence consumers.

As such, we hope that these initial ﬁndings may help to

precipitate a broader stream of research in marketing on the

automatic processing of facial imagery. We suspect that this

research will reveal that, rather than being one environ-

mental prime among many, faces are actually the recipients

of a dramatically disproportionate amount of our automatic

attentional resources. Indeed, we believe that faces may ac-

tually be the most primal of primes.

REFERENCES

Adolphs, Ralph, Daniel Trane, and Antonio R. Damasio (1998),

“The Human Amygdala in Social Judgment,” Nature, 393

(June), 470–74.

Alicke, Mark D., M. L. Klotz, David L. Breitenbecher, Tricia J.

Yurak, and Debbie S. Vredenburg (1995), “Personal Contact,

Individuation, and the Better-Than-Average Effect,” Journal

of Personality and Social Psychology, 68 (5), 804–25.

Amaral, David G. (2002), “The Primate Amygdala and the Neuro-

biology of Social Behavior: Implications for Understanding So-

cial Anxiety,” Biological Psychiatry 51 (1), 11–17.

Ambler, Tim (1997), “How Much of Brand Equity Is Explained

by Trust?” Management Decision, 35 (4), 283–92.

Bailenson, Jeremy N., Shanto Iyengar, Nick Yee, and Nathan A.

Collins (2009), “Facial Similarity between Voters and Can-

didates Causes Inﬂuence,” Public Opinion Quarterly,72(5),

935–61.

Bargh, John A. (1997), “The Automaticity of Everyday Life,” in

Advances in Social Cognition, Vol. 10, ed. Robert S. Wyer

Jr., Mahwah, NJ: Erlbaum, 231–46.

Bargh, John A., and Tanya L. Chartrand (1999), “The Unbearable

Automaticity of Being,” American Psychologist 54 (July),

462–79.

——— (2000), “Studying the Mind in the Middle: A Practical

Guide to Priming and Automaticity Research,” in Handbook

of Research Methods in Social Psychology, ed. Harry T. Reis

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

782 JOURNAL OF CONSUMER RESEARCH

and Charles M. Judd, New York: Cambridge University Press,

253–85.

Bornstein, Robert F., and Paul R. D’Agostino (1994), “The At-

tribution and Discounting of Perceptual Fluency: Preliminary

Tests of a Perceptual Fluency/Attributional Model of the Mere

Exposure Effect,” Social Cognition, 12 (Summer), 103–28.

Cacioppo, John T. (2004), “Asymmetries in Affect Laden Infor-

mation Processing,” in Perspectivism in Social Psychology:

The Yin and Yang of Scientiﬁc Process, ed. John T. Jost,

Mahzarin R. Banaji, and Deborah A. Prentice, Washington,

DC: American Psychological Association, 85–95.

Cacioppo, John T., Wendi L. Gardner, and Gary G. Berntson

(1999), “The Affect System Has Parallel and Integrative Pro-

cessing Components: Form Follows Function,” Journal of

Personality and Social Psychology, 76 (5), 839–55.

Carre´, Justin M., Cheryl M. McCormick, and Catherine J. Mond-

loch (2009), “Facial Structure Is a Reliable Cue of Aggressive

Behavior,” Psychological Science, 20 (10), 1194–98.

Chartrand, Tanya L., Joel Huber, Baba Shiv, and Robin J. Tanner

(2008), “Nonconscious Goals and Consumer Choice,” Jour-

nal of Consumer Research, 35 (August), 189–201.

Chaudhuri, Arjun, and Morris B. Holbrook (2001), “The Chain of

Effects from Brand Trust and Brand Affect to Brand Perfor-

mance: The Role of Brand Loyalty,” Journal of Marketing,

65 (April), 81–93.

Chen, Mark, and John A. Bargh (1999), “Consequences of Au-

tomatic Evaluation: Immediate Behavioral Predispositions

to Approach or Avoid the Stimulus,” Personality and Social

Psychology Bulletin, 25 (2), 215–24.

Corwin, G. H. (1921), “The Involuntary Response to Pleasant-

ness,” American Journal of Psychology, 32 (4), 563–70.

Critchley, Hugo, Eileen Daly, Mary Phillips, Michael Brammer,

Edward Bullmore, Steven Williams, Therese Van Amelsvoort,

Dene Robertson, Anthony David, and Declan Murphy (2000),

“Explicit and Implicit Neural Mechanisms for Processing of

Social Information from Facial Expressions: A Functional

Magnetic Resonance Imaging Study,” Human Brain Mapping,

9 (2), 93–105.

DeBruine, Lisa M. (2002), “Facial Resemblance Enhances Trust,”

Proceedings of the Royal Society, Series B, 269 (1498),

1307–12.

——— (2004a), “Resemblance to Self Increases the Appeal of

Child Faces to Both Men and Women,” Evolution and Human

Behavior, 25 (2), 142–54.

——— (2004b), “Trustworthy but Not Lust-Worthy: Context Spe-

ciﬁc Effects of Facial Resemblance,” Proceedings of the Royal

Society, Series B, 272 (1566), 919–22.

De Houwer, Jan, Sarah Thomas, and Frank Baeyens (2001),

“Association Learning of Likes and Dislikes: A Review

of 25 Years of Research on Human Evaluative Condition-

ing,” Psychological Bulletin, 127 (6), 853–69.

Devine, Patricia G. (1989), “Stereotypes and Prejudice: Their

Automatic and Controlled Components,” Journal of Per-

sonality and Social Psychology, 56 (January), 5–18.

Dion, Karen, Ellen Berscheid, and Elaine Walster (1972), “What

Is Beautiful Is Good,” Journal of Personality and Social Psy-

chology, 24 (3), 285–90.

Doney, Patricia M., and Joseph P. Cannon (1997), “An Exami-

nation of the Nature of Trust in Buyer-Seller Relationships,”

Journal of Marketing, 61 (April), 35–51.

Ekman, Paul (1992), “Facial Expressions of Emotion: New Find-

ings, New Questions,” Psychological Science, 3 (1), 34–38.

Elliot, Andrew J. (2006), “The Hierarchical Model of Approach-

Avoidance Motivation,” Motivation and Emotion,30(2),

111–16.

Elliot, Andrew J., and Martin V. Covington (2001), “Approach and

Avoidance Motivation,” Educational Psychology Review,13

(2), 73–92.

Engell, Andew D., James V. Haxby, and Alexander Todorov

(2007), “Implicit Trustworthiness Decisions: Automatic Cod-

ing of Face Properties in Human Amygdala,” Journal of Cog-

nitive Neuroscience, 19 (9), 1508–19.

Fiske, Susan T., Amy J. Cuddy, and Pete Glicke (2007), “Universal

Dimensions of Social Cognition: Warmth and Competence,”

Trends in Cognitive Sciences,11(2),77–83.

Fitzsimons, Grainne M., Tanya L. Chartrand, and Gavan J. Fitz-

simons (2008), “Automatic Effects of Brand Exposure on

Motivated Behavior: How Apple Makes You Think Differ-

ent,” Journal of Consumer Research, 35 (June), 21–35.

Friestad, Marian, and Peter Wright (1994), “The Persuasion Knowl-

edge Model: How People Cope with Persuasion Attempts,”

Journal of Consumer Research, 21 (June), 1–31.

Hansen, Christine H., and Ronald D. Hansen (1988), “Finding the

Face in the Crowd: An Anger Superiority Effect,” Journal of

Personality and Social Psychology, 54 (6), 917–24.

Hassin, Ran R., James S. Uleman, and John A. Bargh, eds. (2005),

The New Unconscious, New York: Oxford University Press.

Hatﬁeld, Elaine, John T. Cacioppo, and Richard L. Rapson (1993),

“Emotional Contagion,” in Review of Personality and Social

Psychology, Vol. 14, Emotion and Social Behavior, Newbury

Park, CA: Sage, 151–77.

Hauber, Mark E., and Paul W. Sherman (2001), “Self-Referent

Phenotype Matching: Theoretical Considerations and Empir-

ical Evidence,” Trends in Neuroscience, 24 (10), 609–16.

Howard, Daniel J., and Charles Gengler (2001), “Emotional Con-

tagion Effects on Product Attitudes,” Journal of Consumer

Research, 28 (September), 189–201.

Ives, Nat (2008), “Despite Talk of Ethics Codes, Airbrushing Is

Here to Stay,” Advertising Age 79 (April 21), 24.

Kanwisher, Nancy, and Galit Yovel (2006), “The Fusiform Face

Area: A Cortical Region Specialized for the Perception of

Faces,” Philosophical Transactions of the Royal Society B:

Biological Sciences, 361 (November), 2109–28.

Kim, Moonja, and Seymour Rosenberg (1980), “Comparison of

Two Structural Models of Implicit Personality Theory,” Jour-

nal of Personality and Social Psychology, 38 (3), 375–89.

Kunst-Wilson, William R., and Robert B. Zajonc (1980), “Affec-

tive Discrimination of Stimuli That Cannot Be Recognized,”

Science, 207 (February), 557–58.

Langlois, Judith H., and Lori A. Roggman (1990), “Attractive

Faces Are Only Average,” Psychological Science, 1 (2),

115–21.

Lewin, Kurt (1935), A Dynamic Theory of Personality, New York:

McGraw-Hill.

McCracken, Grant (1989), “Who Is the Celebrity Endorser? Cul-

tural Foundations of the Endorsement Process,” Journal of

Consumer Research, 16 (3), 310–21.

Mehta, Ravi, and Rui (Juliet) Zhu (2009), “Blue or Red? Exploring

the Effect of Color on Cognitive Task Performances,” Science

323 (5918), 1226–29.

Neumann, Roland, and Fritz Strack (2000), “Approach and Avoid-

ance: The Inﬂuence of Proprioceptive and Extereoceptive

Cues on Encoding of Affective Information,” Journal of Per-

sonality and Social Psychology,79(1),39–48.

Nummenmaa, Lauri, Jukka Hyona, and Jari K. Hietanen (2009),

“I’ll Walk This Way: Eyes Reveal the Direction of Loco-

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

CELEBRITY FACIAL CUES 783

motion and Make Passersby Look and Go the Other Way,”

Psychological Science, 20 (12), 1454–58.

Ohanian, Roobina (1990), “Construction and Validation of a Scale

to Measure Celebrity Endorsers’ Perceived Expertise, Trust-

worthiness, and Attractiveness,” Journal of Advertising,19

(3), 39–52.

Oosterhof, Nikolaas N., and Alexander Todorov (2008), “The

Functional Basis of Face Evaluation,” Proceedings of the Na-

tional Academy of Sciences, USA, 105 (32), 11087–92.

Osgood, Charles E., George J. Suci, and Percy H. Tannenbaum

(1957), The Measurement of Meaning, Urbana: University of

Illinois Press, 1957.

Park, Justin H., and Mark Schaller (2005), “Does Attitude Sim-

ilarity Serve as a Heuristic Cue for Kinship? Evidence of

an Implicit Cognitive Association,” Evolution and Human

Behavior, 26 (2005), 158–70.

Parr, Lisa A., and Frans B. M. de Waal (1999), “Visual Kin Rec-

ognition in Chimpanzees,” Nature, 399 (June), 647–48.

Schwarz, Norbert (2004), “Metacognitive Experiences in Con-

sumer Judgment and Decision Making,” Journal of Consumer

Psychology, 14 (4), 332–48.

Shanteau, James, and Geraldine F. Nagy (1979), “Probability of

Acceptance in Dating Choice,” Journal of Personality and

Social Psychology, 37 (4), 522–33.

Small, Deborah A., and Nicole M. Verrochi (2009), “The Face

of Need: Facial Emotion Expression on Charity Advertise-

ments,” Journal of Marketing Research, 46 (December),

777–87.

Tanner, Robin J., Rosellina Ferraro, Tanya L. Chartrand, James R.

Bettman, and Rick Van Baaren (2008), “Of Chameleons and

Consumption: The Impact of Mimicry on Choice and Pref-

erences,” Journal of Consumer Research, 34 (April), 754–

66.

Todorov, Alexander, and Andrew D. Engell (2008), “The Role of

the Amygdala in Implicit Evaluation of Emotionally Neutral

Faces,” Social, Cognitive, and Affective Neuroscience,3(4),

303–12.

Todorov, Alexander, Anesu N. Mandisodza, Amir Goren, and Crys-

tal C. Hall (2005), “Inferences of Competence from Faces

Predict Election Outcomes,” Science, 308 (5728), 1623–26.

Todorov, Alexander, Manish Pakrashi, and Nikolaas N. Oosterhof

(2009), “Evaluating Faces on Trustworthiness after Minimal

Time Exposure,” Social Cognition, 27 (6), 813–33.

Verosky, Sara C., and Alexander Todorov (2010), “Generalization

of Affective Learning about Faces to Perceptually Similar

Faces,” Psychological Science, 21 (6), 779–85.

Weisbuch, Max, and Diane Mackie (2009), “False Fame, Percep-

tual Clarity, or Persuasion? Flexible Fluency Attribution in

Spokesperson Familiarity Effects,” Journal of Consumer Psy-

chology, 19 (2009), 62–72.

Weisbuch, Max, Diane Mackie, and Teresa Garcia-Marques (2003),

“Prior Source Exposure and Persuasion: Further Evidence for

Misattribution Processes,” Personality and Social Psychology

Bulletin, 29 (6), 691–700.

Wells, Peter A. (1987), “Kin Recognition in Humans,” in Kin

Recognition in Animals, ed. David J. C. Fletcher and Charles

D. Michener, Chichester: Wiley, 395–415.

Whalen, Paul J. (1998), “Fear, Vigilance, and Ambiguity: Initial

Neuroimaging Studies of the Human Amygdala,” Current

Directions in Psychological Science 7 (December), 177–88.

Willis, Janine, and Alexander Todorov (2006), “First Impressions:

Making Up Your Mind after 100 Milliseconds Exposure to a

Face,” Psychological Science, 17 (7), 592–98.

Winkielman, Piotr, and John T. Cacioppo (2001), “Mind at Ease

Puts a Smile on the Face: Psychophysiological Evidence That

Processing Facilitation Increases Positive Affect,” Journal of

Personality and Social Psychology, 81 (6), 989–1000.

Winston, Joel S., Bryan A. Strange, John O’Doherty, and Ray-

mond J. Dolan (2002), “Automatic and Intentional Brain

Responses during Evaluation of Trustworthiness of Faces,”

Nature Neuroscience, 5 (3), 277–83.

Zajonc, Robert (1968), “Attitudinal Effects of Mere Exposure,”

Journal of Personality and Social Psychology Monograph

Supplement, 9 (2, pt. 2), 1–28.

This content downloaded from 129.237.35.237 on Tue, 13 Aug 2013 13:22:12 PM

All use subject to JSTOR Terms and Conditions

Can Digital Face-Morphs Influence Attitudes and Online Behaviors?

Conference Paper

Full-text available

Aug 2018

Self-images are among the most prevalent forms of content shared on social media streams. Face-morphs are images digitally created by combining facial pictures of different individuals. In the case of self-morphs, a person's own picture is combined with that of another individual. Prior research has shown that even when individuals do not recognize themselves in self-morphs, they tend to trust self-morphed faces more, and judge them more favorably. Thus, self-morphs may be used online as covert forms of targeted marketing-for instance, using consumers' pictures from social media streams to create self-morphs, and inserting the resulting self-morphs in promotional campaigns targeted at those consumers. The usage of this type of personal data for highly targeted influence without individuals' awareness, and the type of opaque effect such artifacts may have on individuals' attitudes and behaviors, raise potential issues of consumer privacy and autonomy. However, no research to date has examined the feasibility of using self-morphs for such applications. Research on self-morphs has focused on artificial laboratory settings, raising questions regarding the practical, in-the-wild applicability of reported self-morph effects. In three experiments, we examine whether self-morphs could affect individuals' attitudes or even promote products/services, using a combination of experimental designs and dependent variables. Across the experiments, we test both designs and variables that had been used in previous research in this area and new ones that had not. Questioning prior research, however, we find no evidence that end-users react more positively to self-morphs than control-morphs composed of unfamiliar facial pictures in either attitudes or actual behaviors.

The Impact of Model Pose on Consumer Perceptions of Price: A Perceived-Power Perspective

Article

Sep 2024

An investigation of a quantum probability model for the constructive effect of affective evaluation

Article

Full-text available

Jan 2016
PHILOS T R SOC A

The idea that choices can have a constructive effect has received a great deal of empirical support. The act of choosing appears to influence subsequent preferences for the options available. Recent research has proposed a cognitive model based on quantum probability (QP), which suggests that whether or not a participant provides an affective evaluation for a positively or negatively valenced stimulus can also be constructive and so, for example, influence the affective evaluation of a second oppositely valenced stimulus. However, there are some outstanding methodological questions in relation to this previous research. This paper reports the results of three experiments designed to resolve these questions. Experiment 1, using a binary response format, provides partial support for the interaction predicted by the QP model; and Experiment 2, which controls for the length of time participants have to respond, fully supports the QP model. Finally, Experiment 3 sought to determine whether the key effect can generalize beyond affective judgements about visual stimuli. Using judgements about the trustworthiness of well-known people, the predictions of the QP model were confirmed. Together, these three experiments provide further support for the QP model of the constructive effect of simple evaluations.

Enhancing Sales of Green Agricultural Products through Live Streaming in China: What Affects Purchase Intention?

Article

Full-text available

Mar 2023

Live streaming of agricultural products can reduce farmer losses caused by unsalable agricultural products and increase the income of farmers. Live streaming can be especially effective for green product sales as it can increase viewer understanding of and interest in the production and attributes of green products. Using the stimulus, organization, and response theoretical framework, this paper explores the influence of agricultural products anchors’ characteristics on consumers’ purchasing behavior and the mediating effect of green consumption cognition. The data come from surveys distributed online to Chinese shoppers who had participated in an agricultural livestreaming event. A total of 691 valid responses were received. Anchor characteristics of influence, sales promotion, and interactive entertainment were found to positively impact consumers’ purchasing behavior. Green consumption cognition, composed of environmental problem perception and green consumption awareness, has a positive impact on purchasing intent. Furthermore, anchor characteristics of influence and sales promotion can positively influence consumers’ purchasing behavior indirectly through green consumption cognition.

Role of celebrity credibility on banking service providers

Article

Mar 2023
Market Intell Plann

Ramendra Pratap Singh

Purpose – This study examines the role of celebrity credibility in fostering loyalty and long-term commitment toward the financial service provider (banking services in this study) through the development of trust in the service provider. The relationship combines the commitment-trust theory, the source credibility, and the meaning transfer model. Design/methodology/approach – The proposed model uses online survey data from 855 participants. The data are analyzed using structural equation modeling. The relationship between celebrity credibility, commitment, and loyalty to the service provider is explored, with trust in the service provider acting as a mediator. The profession and gender of celebrities are considered moderators. Findings – Trust in service providers partially mediates the relationship between celebrity credibility, customer loyalty, and commitment. Moderation results across multiple groups show that male and sports celebrities are more persuasive in expertise, and customers find women and movie stars to be more effective endorsers in the attractiveness dimension. Research limitations/implications – This study focuses only on banking services with data collected from a single country to examine a celebrity’s role as a spokesperson. Practical implications – Considering the benefits of having a well-known celebrity as a spokesperson for financial services like banking is intriguing. Given the unique characteristics and significant perceived risk associated with financial services, marketers may consider hiring celebrity endorsers to increase tangibility and reduce risk. Originality/value – The results of this study may provide a deeper understanding of the combined effect of celebrity credibility and trust in financial service providers, which provides tangibility and helps establish a strong customer relationship.

Does source credibility matter in promoting sustainable consumption? Developing an integrated model

Article

Oct 2022

Purpose This paper aims to explore the role of source credibility while purchasing environment-friendly products using Ajzen’s (1991) theory of planned behavior as underpinning model. Design/methodology/approach The proposed theoretical model was empirically tested with the data collected from 334 respondents using structural equation modeling. Findings The results gave empirical support to the addition of source credibility to the original theory of planned. Moreover, consumer attitude was found mediating the effect of corporations’ credibility on purchase intention. Also, attitude and perceived behavioral control were found as the most important predictors of consumer’s intention to purchase environment-friendly products. Practical implications This study provides valuable insights for the marketers engaged in sustainable business practices. Amid, ever-increasing carbon emission, promoting the use of environment-friendly products has become the need of the time. Credibility plays a crucial role while promoting and communicating an organization’s sustainable practices among its stakeholders including consumers. Therefore, the marketer should formulate appropriate marketing communication strategy to communicate the consumer about the green practices and environment-friendly products they produce. The results suggest that corporation’s credibility shapes consumer attitude and influences intention to purchase environment-friendly products. Earning trust of the consumer is pivotal to achieve success in the market. Therefore, results may help the marketers to better understand consumer’s response toward their marketing strategies and further convince and persuade them to buy their products. Social implications The findings of this study may be useful for marketers, strategists, policymakers and government while formulating promotional strategies to make consumer aware, educate and persuade them to purchase products which do not cause harm to the environment. Originality/value The study is novel in terms of exploring role of source credibility and extending theory of planned behavior in the context of sustainable consumption.

Market hinders interpersonal trust: The exposure to market relationships makes people trust less through elevated proportional thinking and reduced state empathy

Article

Full-text available

Sep 2022

In a series of five experiments, we provided evidence that evoking the market mindset negatively affects trust. We found that the market mindset reduces trust compared to the communal mindset (Experiment 1) and a neutral condition (Experiment 2). Next, we examined the psychological mechanisms behind the detrimental effect of the market mindset on trust and found that this effect was mediated by enhanced proportional thinking (Experiments 3 and 4) and reduced state empathy (Experiments 4 and 5). Finally, in a preregistered Experiment 5, we showed that these two psychological mechanisms are relatively independent.

Relation between Self-Focused Attention and Sensitivity to Distortions of Own Face

Article

Full-text available

Oct 2024

p>90 volunteers (62 women), aged from 18 to 38 years participated in an experimental task of own face recognition in morphed images. The aim of the study was to examine the test-retest reliability of self-face recognition test scores, the concurrent validity of test scores with self-focused attention, depression, anxiety, negative expressivity and personality, and a comparison of test scores by participant sex. An increased sensitivity to distortions of own face on retesting was found. Females were more sensitive to distortions of own face. Extraversion was positively correlated with self-face recognition test scores. The scales of public and private self-consciousness and negative expressivity correlated negatively with measures of the self-face recognition test, suggesting that people with high scores on these scales are more sensitive to facial distortions and refuse to recognize themselves in distorted images.</p

Do I Trust you or your Avatar? An Experimental Study

Article

Jan 2023

An Ai Method to Score Celebrity Visual Potential from Human Faces

Article

Jan 2021

Mind at Ease Puts a Smile on the Face: Psychophysiological Evidence That Processing Facilitation Elicits Positive Affect

Article

Full-text available

Dec 2001

The affect system, in its position to monitor organismic-environmental transactions, may be sensitive to the internal dynamics of information processing. Hence, the authors predicted that facilitation of stimulus processing should elicit a brief, mild, positive affective response. In 2 studies, participants watched a series of neutral pictures while the processing ease was unobtrusively manipulated. Affective reactions were assessed with facial electromyography (EMG). In both studies, easy-to-process pictures elicited higher activity over the region of zygomaticus major, indicating positive affect. The EMG data were paralleled by self-reports of positive responses to the facilitated stimuli. The findings suggest a close link between processing dynamics and affect and may help understand several preference phenomena, including the mere-exposure effect. The findings also highlight a potential source of affective biases in social judgments.

Associative Learning of Likes and Dislikes: A Review of 25 Years of Research on Human Evaluative Conditioning

Article

Full-text available

Nov 2001

Evaluative conditioning refers to changes in the liking of a stimulus that are due to the fact that the stimulus has been paired with other, positive or negative stimuli. Although evaluative conditioning appears to be subjected to certain boundary conditions, significant evaluative conditioning effects have been obtained using a large variety of stimuli and procedures. Some data suggest that evaluative conditioning can occur under conditions that do not support other forms of Pavlovian conditioning, and several models have been proposed to account for these differences. In the present article, the authors summarize the available literature, draw conclusions where possible, and provide suggestions for future research.

Stereotypes and Prejudice: Their Automatic and Controlled Components

Article

Full-text available

Jan 1989

Patricia G. Devine

Three studies tested basic assumptions derived from a theoretical model based on the dissociation of automatic and controlled processes involved in prejudice. Study 1 supported the model's assumption that high- and low-prejudice persons are equally knowledgeable of the cultural stereotype. The model suggests that the stereotype is automatically activated in the presence of a member (or some symbolic equivalent) of the stereotyped group and that low-prejudice responses require controlled inhibition of the automatically activated stereotype. Study 2, which examined the effects of automatic stereotype activation on the evaluation of ambiguous stereotype-relevant behaviors performed by a race-unspecified person, suggested that when subjects' ability to consciously monitor stereotype activation is precluded, both high- and low-prejudice subjects produce stereotype-congruent evaluations of ambiguous behaviors. Study 3 examined high- and low-prejudice subjects' responses in a consciously directed thought-listing task. Consistent with the model, only low-prejudice subjects inhibited the automatically activated stereotype-congruent thoughts and replaced them with thoughts reflecting equality and negations of the stereotype. The relation between stereotypes and prejudice and implications for prejudice reduction are discussed.

Asymmetries in affect laden information processing

Chapter

Full-text available

Jan 2004

John T Cacioppo

Emotional Contagion

Book

Sep 1993

When people are in a certain mood, whether elation or depression, that mood is often communicated to others. When we are talking to someone who is depressed it may make us feel depressed, whereas if we talk to someone who is feeling self-confident and buoyant we are likely to feel good about ourselves. This phenomenon, known as emotional contagion, is identified here, and compelling evidence for its affect is offered from a variety of disciplines - social and developmental psychology, history, cross-cultural psychology, experimental psychology, and psychopathology.

An Examination of the Nature of Trust in Buyer-Seller Relationships

Article

Apr 1997
J MARKETING

The authors integrate theory developed in several disciplines to determine five cognitive processes through which industrial buyers can develop trust of a supplier firm and its salesperson. These processes provide a theoretical framework used to identify antecedents of trust. The authors also examine the impact of supplier firm and salesperson trust on a buying firm's current supplier choice and future purchase intentions. The theoretical model is tested on data collected from more than 200 purchasing managers. The authors find that several variables influence the development of supplier firm and salesperson trust. Trust of the supplier firm and trust of the salesperson (operating indirectly through supplier firm trust) influence a buyer's anticipated future interaction with the supplier. However, after controlling for previous experience and supplier performance, neither trust of the selling firm nor its salesperson influence the current supplier selection decision.

Fear, Vigilance, and Ambiguity: Initial Neuroimaging Studies of the Human Amygdala

Article

Dec 1998

Paul J Whalen

The Attribution and Discounting of Perceptual Fluency: Preliminary Tests of a Perceptual Fluency/Attributional Model of the Mere Exposure Effect

Article

Jun 1994
SOC COGNITION

Emotional Contagion

Article

Jun 1993

Emotions have ubiquitous effects in human affairs. Vivian Gornick, in Fierce Attachments,^ recounts a typ-ical exchange with her mother. Gor-nick always begins these encounters with high hopes. "Somehow," de-spite her best intentions, the conver-sations always spiral downward: Today is promising, tremendously prom-ising I go to meet my mother. I'm flying. Flying! I want to give her some of this shiningness bursting in me, siphon into her my immense happiness at being alive. Just because she is my oldest inti-mate and at this moment I love every-body, even her. "Oh, Ma! What a day I've had," I say. "Tell me," she says. "Do you have the rent this month?" "Ma, listen ..." I say. "That review you wrote for the Times," she says. "Ifs for sure they'll pay you?" "Ma, stop it. Let me tell you what I've been feeling," I say. "Why aren't you wearing something warmer?" she cries. "It's nearly winter." The space inside begins to shimmer. The walls collapse inward. I feel breath-less. Swallow slowly, I say to myself, slowly. To my mother I say, "You do know how to say the right thing at the right time. Ifs remarkable, this gift of yours. It quite takes my breath away." But she doesn't get it. She doesn't loiow I'm being ironic. Nor does she Elaine Hatfield is a Professor of Psychology and Richard L. Rapson is a Professor of History at the Uni-versity of Hawaii. John T. Ca-cioppo is a Professor of Psychology at the Ohio State University. Ad-dress correspondence to Elaine Hatfield, 2430 Campus Rd., Honolulu, HI 96822; BITNET: psych@uhunix; INTERNET: psych@uhunix.uhcc.hawaii.edu. know she's wiping me out. She doesn't know I take her anxiety personally, feel annihilated by her depression. How can she know this? She doesn't even know I'm there. Were I to tell her that it's death to me, her not knowing I'm there, she would stare at me out of her eyes crowd-ing up with puzzled desolation, this young girl of seventy-seven, and she would cry angrily, "You don't under-stand! You have never understood!" (pp. 103-104)

Approach and Avoidance Motivation

Article

Jun 2001

In this article, we introduce this special issue by establishing a conceptual foundation for the distinction between approach and avoidance motivation. We do so primarily by explicating several reasons why the approach–avoidance distinction should be viewed as fundamental and basic to the study of human behavior. In addition, we compare and contrast the “approach–avoidance” designation with other designations that have been used in the motivational literature to cover the same or similar conceptual ground. Finally, we conclude by briefly overviewing the other contributions to this special issue, specifically highlighting how they make use of the approach–avoidance distinction.

A Tiger and a President: Imperceptible Celebrity Facial Cues Influence Trust and Preference

Abstract

Recommended publications

Neural substrates of emotionally valenced episodic memory: A PET study using film stimuli

I Can't Keep Your Face and Voice Out of My Head: Neural Correlates of an Attentional Bias Toward Non...

Awareness of Emotional Stimuli Determines the Behavioral Consequences of Amygdala Activation and Amy...

Brand Preference Affects the Threshold for Perceptual Awareness.