University of Massachusetts Amherst University of Massachusetts Amherst

ScholarWorks@UMass Amherst ScholarWorks@UMass Amherst

Doctoral Dissertations Dissertations and Theses

July 2016

Infants' & Toddlers' Social Evaluations of Trustworthy and Infants' & Toddlers' Social Evaluations of Trustworthy and

Untrustworthy Faces Untrustworthy Faces

Ashley Lyons University of Massachusetts Amherst

Follow this and additional works at: https://scholarworks.umass.edu/dissertations_2

Part of the Developmental Psychology Commons

Recommended Citation Recommended Citation Lyons, Ashley, "Infants' & Toddlers' Social Evaluations of Trustworthy and Untrustworthy Faces" (2016). Doctoral Dissertations. 709. https://doi.org/10.7275/8409294.0 https://scholarworks.umass.edu/dissertations_2/709

This Open Access Dissertation is brought to you for free and open access by the Dissertations and Theses at ScholarWorks@UMass Amherst. It has been accepted for inclusion in Doctoral Dissertations by an authorized administrator of ScholarWorks@UMass Amherst. For more information, please contact scholarworks@library.umass.edu.

INFANTS’ & TODDLERS’ SOCIAL EVALUATIONS OF TRUSTWORHTY AND

UNTRUSTWORTHY FACES

A Dissertation Presented

by

ASHLEY B. LYONS

Submitted to the Graduate School of the

University of Massachusetts Amherst in partial fulfillment

of the requirements for the degree of

DOCTOR OF PHILOSOPHY

May 2016

Developmental Psychology

© Copyright by Ashley B. Lyons 2016

All Rights Reserved

INFANTS’ & TODDLERS’ SOCIAL EVALUATIONS OF TRUSTWORTHY AND

UNTRUSTWORTHY FACES

A Dissertation Presented

by

ASHLEY B. LYONS

Approved as to style and content by:

________________________________________

Erik Cheries, Chair

________________________________________

Louise Antony, Member

________________________________________

Ronnie Janoff-Bulman, Member

________________________________________

Joonkoo Park, Member

________________________________________

Hal Grotevant, Department Head

Department of Psychology

DEDICATION

To my Father: Simply, thank you.

v

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my advisor, Erik Cheries, for an

enormous amount of support and guidance over the past years. I would not be where I am

today if it were not for his patience and mentorship. Thanks are also due to my committee

members, Louise Antony and Joonkoo Park, for their helpful comments and suggestions

throughout this process. A special thank you to Ronnie Janoff-Bulman for serving as a

member of not only my dissertation committee, but also many others throughout my

graduate career. A big thanks to the entire Psychology department, especially the

professors and graduate students in the Developmental division.

I wish to express my appreciation to all the past and present graduate and

undergraduate research assistants in the Infant Cognition Lab. These projects would not

have been completed if it were not for their help and master puppetry skills.

Thank you to my friends and family who supported me throughout this processes.

They have kept me grounded and constantly push me to continue, even in tough times.

Finally, a big thank you to my biggest cheerleaders, my parents. To my mother,

who has taught me to never give up, and constantly reminds me that this too shall pass

when times get hard. To my father, who never doubted my abilities. Thank you for

believing in me, as now I believe in myself.

vi

ABSTRACT

INFANTS’ & TODDLERS’ SOCIAL EVALUATIONS OF TRUSTWORTHY AND

UNTRUSTWORHTY FACES

MAY 2016

ASHLEY B. LYONS, B.S., UNIVERSITY OF ILLINOIS

M.S., UNIVERSITY OF MASSACHUSETTS AMHERST

Ph.D., UNIVERSITY OF MASSACHUSETTS AMHERST

Directed by: Professor Erik W. Cheries

Our understanding of the social world is highly influenced by the fast and automatic

evaluations we make about others based on their facial appearance. The goal of the

current studies is to explore the developmental origins of the particular face-trait

evaluation of ‘trustworthiness.’ Experiment 1 tested whether 10-month-old infants

differentiate between faces that adults rate as trustworthy and untrustworthy, and if they

have a preference for one over the other in a crawling task. Experiment 2 tested whether

10-month-olds have implicit expectations about the social behavior of characters with

trustworthy of untrustworthy faces in a looking-time task that presents infants with

congruent or incongruent trait-action pairings. Finally, Experiment 3 explored the

development of more explicit face-trait judgments by testing 2-year-old’ performance in

a pointing task that required them to match a helpful or unhelpful action with characters

that possess ‘trustworthy’ or ‘untrustworthy’ faces. Taken together, these studies provide

preliminary evidence for a rudimentary form of face-based social evaluation in infancy.

vii

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS…………..………………………………………...…….…….v

ABSTRACT ….... ........... ………………………………………………………………..vi

LIST OF FIGURES………..…………………………………………..……………......viii

CHAPTER

I. INTRODUCTION……………………………………………………………….....…...1

Face-Trait Inferences in Adults…………………………………………………….3

Adults’ Judgments of ‘Trustworthiness’……………………………………………5

The Development of Face-Trait Inferences………………………………………...8

Face-Trait Inferences in Infancy?............................................................................10

Face Perception in Infancy………………………………………………….11

Social Evaluations in Infancy……………….………………………………12

Current Experiments………………………………………………………………14

II. EXPERIMENT 1……………………...………………………………………..……15

Experiment 1a: Discrimination Task………….…………...………………………15

Experiment 1b: Discrimination Task Control……………………………………..19

III. EXPERIMENT 2 …..………………………………………………………………...23

Experiment 2a: Behavioral Inference Task (infants)……………………………...23

Experiment 2b: Behavioral Inference Task Control (infants)……………………..28

IV. EXPERIMENT 3…………………………………………………………………….31

Experiment 3a: Behavioral Inference Task (toddlers)…………………………….31

Experiment 3b: Behavioral Inference Task Control (toddlers)……………………40

V. GENERAL DISCUSSION ...……………………………………………....…………44

BIBLIOGRAPHY…………….……..………...…………………………………………53

viii

LIST OF FIGURES

Figure Page

1. The Untrustworthy (a) and Trustworthy (b) faces as presented in test trials………….17

2. Number of infants who crawled to the trustworthy and untrustworthy buckets during

test trials………………………………………………………………………………….19

3. The Untrustworthy vs. Neutral faces (a) and the Trustworthy vs. Neutral faces (b)

presented to infants during the test trial………………………………………………….21

4. Number of infants who crawled to the trustworthy vs. neutral and untrustworthy vs.

neutral buckets during test trials…………………………………………………………22

5. Looking time (s) between the Consistent and Inconsistent feature-action pairings…..28

6. Looking time (s) between the consistent and inconsistent trials in Experiment 2b…...31

7. The trustworthy and untrustworhty characters without their visors (a) and with the

visor (b)…………………………………………………………………………………..37

8. Number of correct vs. incorrect choices by gender, age 17-19 months……………….38

9. Number of correct vs. incorrect overall choices, age 2-2.5 years……………………..40

10. Number of correct vs. incorrect choices by helping or hindering trial, age 2-2.5

years……………………………………………………………………………………...40

11. Number of correct vs. incorrect choices by gender, age 2-2.5 years………………...41

12. The neutral and untrustworthy puppets (a) and the neutral and trustworthy puppets

(b)………………………………………………………………………………………...44

13. Number of correct vs. incorrect overall choices, by condition………………………46

1

CHAPTER I

INTRODUCTION

Infants’ & Toddlers’ Social Evaluations of Trustworthy and Untrustworthy Faces

Adults automatically form social evaluations of others based upon very brief

observations of their appearance and behavior (e.g. Ambady, 2010; Ambady, Hallahan, &

Connor, 1999; Ambady & Rosenthal, 1993; Rule & Ambady, 2008, 2009). Even though

such “thin slice” sampling only represents a small fraction of what a person is like, the

judgments based on these impressions are surprisingly reliable and highly predictive of

both people’s current dispositions, behaviors, and future outcomes (e.g. Ambady et al.,

1999; Rule & Ambady, 2009).

Much of the “Thin Slice” research on adults has focused on identifying the

particular cues that support our automatic social evaluations of others. For example,

adults viewing silent 10-second video clips use non-verbal behavior to accurately predict

teachers’ effectiveness (Ambady & Rosenthal, 1993). Adults have also been shown to

accurately judge another’s sexual orientation from viewing silent 1-second video clips of

their behavior (Ambady et al., 1999). Although features such as movement and posture

play a role in these evaluations, static images of a person can also be used to accurately

predict that person’s future outcome in life, as well as certain biological traits of that

person. Multiple studies have shown that both male and female sexual orientation can be

accurately, automatically, and quickly perceived from static images with as little as 40 ms

of exposure (Rule & Ambady, 2008; Rule, Ambady, and Hallett, 2009; Rule, Macrae, &

Ambady, 2009). And when shown photos of successful males and females who were

Fortune 500 CEOs and top law firm partners, participants accurately predicted the

2

success of that individual with respect to the success of their company (Rule & Ambady,

2009, Rule & Ambady, 2011). Rule and Ambady (2011) extended this finding by

showing participants college yearbook photos of the same males who were successful

partners in top law firms. Not only did the participant’s evaluations of the person

correspond with later success, results also showed that the relationship between how we

perceive someone and their performance is evident very early.

What type of cognitive process allows adults to make such reliable judgments on

the basis of such sparse evidence? Although the “thin slices” of behavior may be

extremely brief, the formation of the evaluation might proceed in a relatively slow and

deliberate fashion, relying to a large extent upon conscious reflection. On the contrary,

research suggests that thin slice judgments are actually more accurate when they are

made without deliberation (Ambady, 2010). In one such demonstration, participants were

assigned to one of four different conditions, two experimental conditions (cognitive

distraction or cognitive reasoning) and two control conditions (delayed ratings or non-

delayed ratings), while viewing video clips of teachers (same as Ambady & Rosenthal,

1993). In the Cognitive Distraction Condition, participants were asked to count

backwards from 1000 by 9s while viewing the videos. Those in the Cognitive Reasoning

Condition were asked to write down all the possible reasons for making their judgment of

the teacher’s effectiveness after viewing the stimuli. Those in the control conditions did

not have any task during or after viewing the stimuli; they were just asked to make their

ratings. Results showed that participants in both control conditions were the most

accurate in their judgments, and those in the Cognitive Distraction Condition were more

accurate than the Cognitive Reasoning Condition. The fact that thin slice judgments are

3

computed separately and at the same time as other cognitive tasks suggests that they

result from a largely automatic and intuitive cognitive process.

Face-Trait Inferences in Adults

The studies described above all demonstrate how properties such as movement,

posture, and clothing can support automatic evaluations in adults. However, the richest

and most easily accessible source to gather socially relevant information from is a

person’s face. Indeed, many of our most basic social responses result from the automatic

way in which we extract emotion from another’s face. In one demonstration of this idea,

participants were asked to either pull or push a lever while viewing either fearful or angry

faces. It has been documented that extension (pushing a lever) happens quicker when

presented with an aversive stimulus, and flexion (pulling a lever) happens quicker when

presented with an appeasing stimulus (Chen & Bargh, 1999, Solarz, 1960). Angry and

fearful faces elicited opposite responses—quicker pushing when we see angry faces (an

avoidance behavior), and quicker pulling of the lever (an approaching behavior) with

fearful faces (Marsh, Ambady, & Kleck, 2005). It is possible that when viewing a fearful

face, adults feel the need to help, resulting in approaching behavior. Adults are also faster

at detecting angry faces that are moving towards them over fearful faces (Adams,

Ambady, Macrae, & Kleck, 2006). This approach-avoidance response also happens when

personally responding to someone prosocially; those who are able to identify a fearful

facial expression more easily and accurately are more likely to respond to someone in a

helpful manner (Marsh & Ambady, 2007; Marsh, Kozak, & Ambady, 2007).

Interestingly, even adults who observe faces that are emotionally neutral have

been shown to make preconscious, extremely rapid, and automatic social evaluations

4

(Stewart et al., 2012; Todorov, Pakrashi, & Oosterhof, 2009; Willis & Todorov, 2006). In

one early study looking at how facial features are perceived in conjunction to demeanor,

adults were shown faces that varied on different social dimensions—trustworthiness,

likeability, attractiveness, aggressiveness, and competence—and asked to make yes or no

judgments about the person (Willis & Todorov, 2006). First, one group of subjects was

asked to rate photographs of people’s faces along a variety of dimensions on a scale of 1

to 9. These same photos were then presented to a separate group of adults who were

asked specific trait questions such as, “is this person competent?,” and were allowed to

answer yes or no. After each question they were also asked to rate how confident they felt

about their answer. Results showed that adults were able to reliably and accurately rate

faces on all five of the dimensions. This study also included a time manipulation, which

involved varying the exposure time to the faces (100 ms, 500 ms, 1000 ms). The

consistency of the ratings was compared with adults who did not have a time constraint

when viewing the faces. After only 100 ms of exposure, adults made consistent decisions

on all five social dimensions (Willis & Todorov, 2006). Recent research has also shown

that adults do not need the entire face to make evaluations along these dimensions, but

they can also do so when given a few coordinated points on a face, mainly the eyes and

the mouth region (Rojas, Masip, Todorov, & Vitria, 2011).

Not only are adults able to make these trait decisions very quickly, they are also

able to use specific social dimensions to accurately predict future outcomes. For example,

adults’ ratings on competence have been shown to accurately predict election winners.

Participants viewed photos for 1-second of unfamiliar political candidates who were

running against each other in an upcoming election. Participants’ ratings of the

5

candidates’ competence accurately predicted the winner (Todorov, Mandisodza, Goren,

& Hall, 2005). This finding has also been replicated using negative trait ratings—

candidates with faces that were perceived as more threatening or deceitful were more

likely to lose elections (Mattes et al., 2010). Replications have also shown that these

results are not due to a familiarity with American faces. The same study was completed

producing similar results, with American subjects rating Canadian political candidates

(Rule & Ambady, 2010 for review).

Adults’ Judgments of ‘Trustworthiness’

While adults are capable of evaluating faces on a variety of traits, one of the most

basic social dimensions appears to be ‘trustworthiness.’ Judging a face as trustworthy or

untrustworthy has been shown to happen at an even faster exposure time then the other

dimensions, with adults able to reliably decide whether or not a face is trustworthy after

an exposure as brief as 33ms (Todorov et al. 2009). Trustworthy judgments have also

been shown to strongly correlate with the overall positive and negative valence of a face

(Todorov, 2008), one of two orthogonal dimensions (along with “dominance”) that are

sufficient to describe face evaluations in statistical models of adults’ impressions

(Oosterhof & Todorov, 2008). These statistical models have been used to create

computer-generated faces that represent the features that elicit reliable trait judgments of

‘trustworthiness’. To create these faces, researchers obtained trait judgments on sets of

human faces with neutral emotional expressions (i.e. not smiling). Researchers then

created computer faces defined by dozens of vertices on a polygonal mesh that could be

morphed to reflect that changes that elicit certain trait judgments in adults who had

viewed photos of natural faces (Oosterhof & Todorov, 2008). By systematically

6

morphing and exaggerating these features, researchers have been able to investigate the

dimensions that predict adults’ trait judgments. Although results showed that participants

view trustworthy faces at the extreme ends of the spectrum as more feminine and as more

baby-faced (morphing the faces to be 8 standard deviations away from a neutral face),

this correlation is not as strongly obtained with the more moderate exemplars that are

typically used in these studies (i.e., faces 3 standard deviations away from neutral).

While previous work on the perception of trustworthiness has focused on making

explicit trait judgments, it has also been found that the perception of trustworthiness in

emotionally neutral faces operates implicitly and can be altered by priming manipulations

(Todorov et al., 2009). Participants were primed with either a trustworthy or

untrustworthy face before seeing a neutral face, and were more likely to classify the

neutral face as untrustworthy after seeing the untrustworthy prime (Todorov et al., 2009).

This suggests that judgments about neutral faces can be altered by viewing an

untrustworthy face, even when the face is not explicitly labeled as untrustworthy. The

perception of trustworthiness can also be altered by adding an emotional expression to

trustworthy and untrustworthy faces (Oosterhof & Todorov, 2009). Both trustworthy and

untrustworthy faces were morphed to include both happy and angry expressions. Results

showed trustworthy faces that were morphed to include happiness were rated as happier

than untrustworthy faces that were morphed to include happiness, and untrustworthy

faces morphed to include anger were rated as angrier than trustworthy faces that were

morphed to include anger, impacting how we perceive them and what emotional state we

attribute to them. Results from this study suggest that emotions such as anger and

7

happiness correlate with trustworthy and untrustworthy traits (Oosterhof & Todorov,

2009).

The research described above suggests that face-trait inferences appear to be

extremely reliable. However, studies examining the validity of inferred facial traits have

produced mixed results (Berry & Brownlow, 1989; Berry, 1991; Penton-Voak, Pound,

Little, & Perrett, 2006; Pound, Penton-Voak, & Brown, 2007; Todorov, Said, Engell, &

Oosterhof, 2008 for review; Zebrowitz, Voinescu, & Collins, 1996). For example,

comparisons between participants’ face-trait inferences and self-reported measures of

personality have shown a correlation between self-report and facial ratings for power and

warmth (Berry, 1991), weakness and approachability (Berry & Brownlow, 1989), and

extraversion (Penton-Voak et al., 2006). However, studies that have used measures other

than self-report, such as behavioral or observational methods, have shown no correlation

between facial ratings and personality. In one study looking at honesty, no correlation

was found between trait ratings (made by the participants of the study) and how honest

someone actually is (as rated by a clinician) (Zebrowitz et al., 1996). The mixed results

from these studies weaken the interpretation that one’s facial characteristics are somehow

causally related to expressed behaviors.

Results from these behavioral studies have also been replicated using

neuroimaging techniques. Research using functional magnetic resonance imaging (fMRI)

has looked at which areas of the brain are involved in encoding trustworthiness and the

specific face features that trigger it (Engell, Haxby, & Todorov, 2007; Said, Baron, &

Todorov, 2009; Winston, Strange, O’Doherty, & Dolan, 2002). Investigations using

fMRI in adults show that the amygdala responds to viewing facial features that

8

correspond to trustworthy and untrustworthy faces (Engell et al., 2007; Said et al., 2009;

Winston et al., 2002). The greatest amygdala response resulted from viewing highly

untrustworthy faces, and is better predicted by ratings averaged across participants then

by each individual’s judgments of the traits (Engell et al., 2007). This stronger activation

of the amygdala is now thought to be due to an attention bias in which more attention is

directed to a stimuli that is motivationally significant to the viewer (Cunningham, Van

Bavel, & Johnsen, 2008).

The Development of Face-Trait Inferences

One big open question is where these thin slice judgments about facial appearance

come from. Although some types of judgments (e.g., ones based upon motion or posture

cues) may develop fairly late in life (Balas, Kanwisher, & Saxe , 2012), it is possible that

we may see the emergence of certain automatic face-based social evaluations at a much

earlier stage. In fact, recent work has explored whether young children are capable of

inferring social traits from faces (Cogsdill & Banaji, 2015; Cogsdill, Todorov, Spelke, &

Banaji, 2014; Ewing, Caulfield, Read, & Rhodes, 2015; Ma & Xu, 2015). For example,

using the same face stimuli that have been validated with adults, a recent finding shows

that children as young as 3 years old are able to make inferences about a face based upon

perceived facial traits (Cogsdill et al., 2014). In this particular study, participants between

3 and 12 years of age viewed pairs of computer-generated faces both high and low on

scales of trustworthiness, dominance, and competence. They were then asked questions

regarding the pairs of faces. For example if viewing a highly trustworthy and highly

untrustworthy face, they were asked, “Which of these people is very nice?” Children

responded by pointing to the screen. Results showed that 3-year olds were able to

9

respond significantly above chance on all three dimensions, with children performing at

adult levels by 5 years of age. Also by 3 years of age, children attributed niceness and

meanness to faces on all three dimensions, not just trustworthiness and untrustworthiness.

By such a young age, children are already making automatic judgments about a person

based upon how they look (Cogsdill et al., 2014). More recently, this finding has been

extended from the computer-generated faces to pictures of adults, children, and monkeys.

Employing the same paradigm, researchers showed children as young as 3 years old pairs

of faces on a computer screen and asked, “Which of these adults/children/monkeys is

very nice/mean?” Children correctly responded well above chance when viewing all types

of stimuli: monkey, adult, and child faces (Cogsdill & Banaji, 2015).

These judgments of trustworthiness can also be found cross-culturally. Chinese

children between the ages of 8- and 12-years old viewed emotionally neutral East Asian

faces and were asked to judge how trustworthy or untrustworthy the face was. One month

later, children were brought back and asked to judge how attractive or unattractive the

same faces were. Results from this study showed that Chinese children used similar cues

to judge trustworthiness as found in previous studies (i.e. Todorov, et al., 2008), and they

also use similar features to judge trustworthiness and attractiveness (Ma et al., 2015).

Taking it one step further, by 5 years children selectively place their trust in those

who look more trustworthy then those who look more untrustworthy. When playing a

game called Token Quest, both 5- and 10-year-old children used facial trustworthiness as

an indicator of whom to invest with. On some trials children were not given access to

face-trait information when making their decision. However, when this information was

10

not available, children were more likely to “pay” tokens to access the information (Ewing

et al., 2015).

Young children are also able to predict election results based on facial

appearance. Swedish children as young as 5 years old were asked to pick who they would

prefer as the captain of their ship. The faces used in this task were pairs of French

politicians running against each other in an election. The correlation between the child’s

choice and the actual election results was greatly above chance, and strongly predicted

adult selections which were completed as a control (Antonokis & Dalgas, 2009).

Face-trait Inferences in Infancy?

The extensive research on face-trait inferences in adults coupled with the

emerging literature investigating the process in children raises a question about how

developmentally primitive this phenomenon might be. Identifying this phenomenon in

infants is important not just for expanding our knowledge of how the infant mind works

but also for constraining theories that explain automatic social evaluations in adults. One

possibility is that adults’ judgments are built slowly from many years of experience

observing that people who behave in a certain way possess a certain type of appearance.

This correlation eventually becomes internalized such that the trait judgment is

automatically triggered upon observing a particular type of face. The development of

specific trait terms in language (e.g., “mean” and “nice,” “trustworthy” and

“untrustworthy”) may further accentuate the strength of such face-trait attributions.

Alternatively, the fast and automatic social evaluations that adults engage in may be

based on very primitive cognitive and emotional processes that develop in relative

isolation to specific experiences matching trait to appearance.

11

Do infants in the first year of life make face-based social inferences? In order for

this to be possible infants would at a minimum need to possess two different cognitive

abilities: 1) an ability to differentiate faces based upon subtle changes in the physical

characteristics that carry meaning in adults and 2) possess trait-related concepts that they

use in their evaluations of others’ behavior. In the next section I review evidence for both

of these abilities.

Face Perception in Infancy

From birth, infants have a preference for happy over fearful faces (Farroni,

Menon, Rigato, & Johnson, 2007). After a few months, this preference begins to develop,

and by 3-4 months infants are able to differentiate faces based on emotions such as happy

and sad faces vs. surprised faces (Young-Browne, Rosenfeld, & Horowitz, 1977) and

smiling faces vs. frowning faces (Barrera & Maurer, 1981). For example, infants in one

early study were habituated to a particular emotional facial expression (such as a happy

face), and once habituated, were presented with a different face (such as a sad face).

Infants in this study showed an increase in looking at the novel face, suggesting they

could differentiate between the two expressions (Young-Browne et al., 1977). Results

from these early studies have also been validated using ERPs, showing that infants

differentiate between positive and negative emotions at a neurological level (e.g.

Grossmann et al., 2011; Grossmann, Striano, & Friederici, 2007; Kobiella, Grossmann,

Reid, & Striano, 2007). Infants are also able to discriminate happy from fearful faces

(Nelson & Ludemann, 1986), and like adults can do so unconsciously (Jessen &

Grossmann, 2015). By 7 months, infants are able to differentiate between fearful and

angry faces (Kobiella et al., 2007). Infants are able to categorize happy and fearful faces

12

by 7 months, and recognize when the faces cross the category boundary (Kotsoni, de

Haan, & Johnson, 2001).

Infants are also able to categorize faces based on non-emotional cues such as

attractiveness (Ramsey, Langlois, Hoss, Rubenstein, & Griffin, 2004) and race, holding

perceptual preferences towards familiar race faces (Bar-Haim, Ziv, Lamy, & Hodes,

2006; Kelly et al., 2007; Kelly et al., 2005). For instance, by 3 months of age, infants

prefer to look at a face belonging to their own race than to a face that is of another race

(Kelly et al., 2005). However, this effect seems to be a perceptual preference and not a

social preference; although 10-month-old infants find race to be salient, they do not use

race as the basis for making social decisions (e.g., whom to take a toy from; Kinzler &

Spelke, 2011).

Social Evaluations in Infancy

Multiple studies have shown that infants from a very young age make social

evaluations of other characters that approximate attributions of trustworthiness, such

“mean” and “nice”. By 12 months of age infants expect a character to respond differently

to an agent who has helped it achieve their goal rather than one who has hindered or

stopped it (Kuhlmeier, Wynn, & Bloom, 2003). Infants have also been shown to exhibit a

personal preference for agents who exhibit “nice” over “mean” actions towards another.

Reaching tasks with infants 6 months old and above and looking time measurements with

infants as young as 3 months have demonstrated a reliable preference to play with or

observe characters who have helped another complete its goal rather than characters who

have hindered another’s goal (Hamlin & Wynn, 2011; Hamlin, Wynn, & Bloom, 2007;

Hamlin, Wynn, & Bloom, 2010). Not only do infants prefer to touch or look at prosocial

13

agents, but they are also more likely to match the food preferences of helpers vs.

hinderers (Hamlin & Wynn, 2012). By 9 months of age, infants evaluate others based

upon shared attributes or similar preferences, preferring an individual who treats a similar

agent positively and a dissimilar agent negatively (Hamlin, Mahajan, Liberman, & Wynn,

2013). The same pattern of results is found when the agents are prosocial or antisocial

instead of similar or dissimilar to the infant—infants prefer individuals who treat

prosocial agents positively, and antisocial agents negatively (Hamlin, Wynn, Bloom, &

Mahajan, 2011). Intentions and outcomes of goals are also evaluated by infants in the

first year of life, with infants as young as 8 months considering an agent’s intentions to

help or hinder another (Hamlin, 2013b). Taken together, these results show that infants

are able to make a variety of social evaluations related to the concept of trustworthiness

based upon others’ behavior within the first year of life (for full review, see Hamlin,

2013a).

The studies described above all demonstrate how infants are capable of making

social evaluations after they observe a particular behavior, such as agents helping or

hindering each other. Of course, many of the automatic social judgments adults make that

we have discussed occur in advance of witnessing actual behavior. Instead, adults’ social

evaluations are based off of extremely minimal observations of some aspect of a person’s

appearance. Do infants translate thin slices of others’ appearance into evaluations of how

that person will behave? In a recent study 12-month-old infants were habituated to two

different characters, one with more positive properties (a soft, fluffy appearance and a

happy baby laugh) and one with more negative properties (a sharp, jagged appearance

and a deep, ominous laugh; Lyons & Cheries, under review). Infants then saw the

14

characters engage in social actions that were consistent or inconsistent with their

observable properties. For example, infants may have observed a “nice” looking and

sounding character perform a nice action of helping, followed by a “mean” looking and

sounding character perform that same helping action. Results showed that infants are

biased to relate these observable properties to the behaviors the agent is going to engage

in, evidenced by longer looking toward a mismatch (Lyons & Cheries, under review).

Current Experiments

We have discussed how infants are able to differentiate faces based on emotion,

discriminate between nice and mean actions, and make some inferences about social

behavior based upon the auditory information they observe of other agents. However, no

work has examined whether infants, like older children and adults, automatically infer

social dispositions like “trustworthy” and “untrustworthy” from faces. The following

studies will investigate this ability in infants. To do this, we used the faces that have been

validated in toddler and adult studies in combination with a helping-hindering social

evaluation paradigm to see if infants perceive faces in a similar manner. Do infants

believe that those who look trustworthy (by adult ratings) should behave in nice ways,

and those who look untrustworthy should behave in mean ways? The earliest related

evidence comes from 3-year-olds’ ability to use linguistic labels to categorize such faces

as “nice” and “mean” (Cogsdill et al., 2014). However, this only reflects a general

association between facial appearance and specific labels and not necessarily the ability

automatically generating expectations about specific social behaviors. It might be that

children apply social labels to particular face types before they use them to infer social

behavior. Indeed, hearing these labels applied to various appearance cues may be an

15

essential part of the developmental process. Alternatively, even pre-linguistic infants may

show some ability to match facial characteristics with observable social behaviors and

dispositions.

One set of studies used an approach-avoidance crawling task to determine if infants

discriminate faces along the same social trait dimensions as adults, such as whether a face

appears trustworthy or untrustworthy (Experiment 1). Another set of studies explored

when over development we become capable of connecting a person’s facial appearance to

their social behavior towards others. We explored whether this ability exists in some

implicit way in infancy (Experiment 2) while also identifying when more explicit forms

of these evaluations emerge in toddlerhood (Experiment 3).

16

CHAPTER II

EXPERIMENT 1

Experiment 1a: Discrimination Task

Preliminary work

Our first goal was to investigate whether young infants have the ability to

differentiate between faces that are high and low on the trustworthiness dimension, and if

they prefer one face to another. To test this, infants completed an approach-avoidance

task where they viewed an equal number of graham crackers being placed in buckets that

had either a trustworthy or untrustworthy face on the front. We hypothesized that if

infants are sensitive to faces on this dimension they would be more likely to approach the

bucket with the trustworthy face than the one with the untrustworthy face.

Methods

Participants

We tested thirty-two 10- to 13-month-old-infants (16 males) from the greater

Amherst area (range=9 months 18 days to 12 months 3 days; mean=10 months 26 days).

Sixteen additional infants were tested but excluded from the sample due to an

unwillingness to crawl (11) and experimenter error (5).

Materials

Infants were placed in a crawling position 5 feet from the experimenter. During

the warm up, infants were shown a small orange bucket placed 2.5 feet away from them,

as well as a rubber duck. During the test trial, infants were shown two small orange

buckets (3 x 5 in.) placed 1.5 feet apart from each other and 2.5 feet from the baby. Each

bucket had one face measuring 4.75 x 7.5 inches attached to the front, one Trustworthy

17

and one Untrustworthy (see Figure 1; Oosterhof & Todorov, 2008). The faces used in the

following studies are each three standard deviations from a neutral face along the

trustworthiness dimension (Todorov et al., 2013). Each graham cracker measured 2 x 1

inches.

a. b.

Figure 1. The Untrustworthy (a) and Trustworthy (b) faces as presented in test trials.

Procedure

Warm-Up Phase. Infants were placed in a crawling position opposite to the

experimenter, and the parent was told to hold on to the infant until the experimenter

prompted the parent. Infants were shown an empty bucket by the experimenter and were

then shown a rubber duck bouncing in to the bucket. Once the duck was in the bucket the

baby was encouraged to crawl to retrieve the duck. After this was done, the bucket and

duck were removed, and the parent was instructed to put the baby back in crawling

position for the test phase.

Test Phase. Infants were shown two empty buckets in the experimenters’ hands.

The buckets were placed on the ground simultaneously, 1.5 feet apart from each other

and 2.5 feet away from the baby. When placing the buckets on the ground, the

experimenter made sure to keep the buckets at an equal distance from the infant to

18

prevent one being closer to the infant. When the experimenter placed each bucket down,

the infant was able to see that one had a trustworthy face and one had an untrustworthy

face attached to the outside. The experimenter took a graham cracker out of a concealed

container placed between their legs and showed the infant it was edible by exclaiming

“Look what I have. I have a yummy graham cracker” before eating it. The experimenter

then took a new graham cracker out of the container and said “Look!” while placing the

graham cracker in the bucket. This was repeated one more time with the same bucket,

before the experimenter placed two more graham crackers in the opposite bucket. Once

the graham crackers were in the buckets, the experimenter looked down and infants were

allowed to crawl to either bucket. A choice was defined as the infant approaching and

touching the bucket or the face. Face side and cracker placement order were

counterbalanced across infants.

Results

A significant majority of infants (22 of 32 infants; p=.05, two-tailed sign test)

crawled to the trustworthy bucket (see Figure 2). There was no difference between males

and females, bucket side, or the order in which the crackers were placed in the bucket.

19

Figure 2: Number of infants who crawled to the trustworthy and untrustworthy buckets

during test trials.

Discussion

Infants in Experiment 1a successfully discriminated between the trustworthy and

untrustworthy faces, and showed a preference for the trustworthy face over the

untrustworthy face in an approach-avoidance task. This pattern of performance is

consistent with infants making a social interpretation of the face stimuli similar to that of

toddlers and adults. Of course, crawling to the bucket with the trustworthy-looking face is

consistent with three different explanations. The preference could have been driven

primarily by an approach response to the trustworthy face, an avoidance response to the

untrustworthy face, or a combination of both. In order to help characterize the nature of

infants’ responses, a different group of subjects were assigned to one of two follow-up

conditions: one where they were presented with a trustworthy face and a neutral face,

versus one where they were presented with a choice between an untrustworthy face and a

neutral face. If the pattern found in Experiment 1a was driven solely by an avoidance

0

6

11

17

22

28

Trustworthy Untrustworthy

*

20

response to the untrustworthy face, infants in the untrustworthy vs. neutral condition

should show a preference for the neutral face and infants in the trustworthy vs. neutral

condition should show no preference. If the pattern was driven solely by a preference to

the trustworthy face, infants in the untrustworthy vs. neutral condition should show no

preference, and infants in the trustworthy vs. neutral condition should show a preference

to the trustworthy face. However, if the pattern found in Experiment 1a was driven by

both an avoidance to the untrustworthy and a preference to the trustworthy, then we

hypothesize a preference to the trustworthy in the trustworthy vs. neutral condition and a

preference to the neutral in the untrustworthy vs. neutral condition.

Experiment 1b

The goal of this study condition is to determine if the pattern found in Experiment

1a was driven by an avoidance to the untrustworthy face only, or by both an avoidance to

the untrustworthy face and a preference for the trustworthy face. To assess this, infants

were separated in to two groups, one given the choice of the trustworthy face vs. a

neutral face, and one group given the choice of the untrustworthy face vs. a neutral face.

Previous research shows that by 8 months of age, infants respond to both the positive and

negative behavioral dispositions of an agent compared to neutral actions (Hamlin et al.,

2011). If infants imbue these face stimuli with similar social dispositions we hypothesize

that infants will show a similar effect; in the trustworthy vs. neutral condition infants will

prefer the trustworthy face, and infants in the untrustworthy vs. neutral condition will

prefer the neutral.

21

Methods

Participants

We tested twenty-two 10- to 13-month-old-infants (9 males) from the greater

Amherst area (range=10 months 19 days to twelve months 4 days; mean=11 months 11

days). Four additional infants were tested but excluded from the sample due to an

unwillingness to crawl.

Materials

The materials used in Experiment 1b were identical to the materials used in

Experiment 1a, however infants now viewed a neutral face (see Figure 3) along with

either the trustworthy or untrustworthy face. The neutral face was three standard

deviations away from both the trustworthy and untrustworthy face on the trustworthiness

dimension.

a. b.

Figure 3: The Untrustworthy vs. Neutral faces (a) and the Trustworthy vs. Neutral faces

(b) presented to infants during the test trial.

Procedure

The procedure of Experiment 1b was identical to the procedure of Experiment 1a,

except infants were split into one of two conditions. Infants in the Trustworthy vs.

Neutral Condition were allowed to crawl to either the trustworthy or neutral face in the

22

test phase, and infants in the Untrustworthy vs. Neutral Condition were allowed to crawl

to either the untrustworthy or neutral face in the test phase.

Results

Trustworthy vs. Neutral. The main results from Experiment 1b are depicted in

Figure 4. No significant difference was found in crawling preference to the trustworthy

bucket (8 of 12 infants; p=.39, two-tailed sign test). There was no difference between

males and females, bucket side, or the order in which the crackers were placed in the

bucket.

Untrustworthy vs. Neutral. The main results from Experiment 1b are depicted in

Figure 4. No significant difference was found in crawling preference to the neutral bucket

(5 of 10 infants; p=1.25, two-tailed sign test). There was no difference between males and

females, bucket side, or the order in which the crackers were placed in the bucket.

Figure 4: Number of infants who crawled to the trustworthy vs. neutral and untrustworthy

vs. neutral buckets during test trials.

0

1

2

3

4

5

6

7

8

9

Trustworthy vs. Neutral Untrustworthy vs. Neutral

Social

Neutral

23

Discussion

Infants in Experiment 1b did not successfully discriminate between the trustworthy

and neutral face, or the neutral and untrustworthy faces, approaching each face equally.

Results from this study suggest that infants may need to view both the trustworthy and

untrustworthy faces together to make a decision on who is approachable. One possibility

for our lack of significant findings in the two conditions is that the difference between the

neutral and ‘social’ face may not be strong enough. While the social faces are each 3SD

from the neutral face, when viewed in conjunction with the neutral face, the social face

may not give off strong enough ‘approach’ or ‘avoid’ cues. Taken together, results from

Experiment 1 imply infants have an ability to distinguish between the trustworthy and

untrustworthy faces, but may need to view both faces to do so. One caveat about

interpreting the current results is that our sample requires an additional 10 infants in order

to complete the counterbalance. Until we include the remaining subjects these results can

only be viewed as preliminary.

24

CHAPTER III

EXPERIMENT 2

Experiment 2a: Behavioral Inference Task (infants)

While previous research (e.g., Cogsdill et al., 2014) as well as Experiment 1 have

shown that infants and children discriminate between trustworthy and untrustworthy

faces, it is not clear if they believe that facial appearance is related to social behavior. For

example, should a character who displays trustworthy features act in a nice or positive

way, and should a character who displays untrustworthy features act in a mean or

negative way? The goal of Experiment 2 was to investigate the ability to evaluate the

social behavior of agents based on how trustworthy or untrustworthy the agents’ facial

features are. In Experiment 2a, infants were shown two characters with facial features on

opposite ends of the trustworthiness dimension interacting with a neutral animal puppet

character in either a helpful or unhelpful way. We hypothesized that if infants are capable

of inferring social behavior from perceived trustworthiness, then they should look longer

when presented with a mismatch between an agent’s facial appearance and social

behavior (i.e. when viewing a trustworthy agent performing a mean action).

Methods

Participants

We tested sixteen 10-month-old infants (8 males) from the greater Amherst area

(range= 9 months 18 days to 10 months 15 days; mean=10 months 6 days). Five

additional infants were tested but excluded from the sample due to experiment error (2)

and fussiness (3).

25

Materials

Infants sat in their parent’s lap 42 inches in front of a stage (60x42x20 in.) flanked

by curtains on all sides. During the familiarization events, infants were presented with

three small cloth hand puppets (a cat, a puppy, and a cow) that each measured

approximately 3 x 10 inches. One of these puppets (the cat) was the ‘struggler’ and was

shown in the test and familiarization phase attempting to open a box (8x14x5 in.) with an

orange rubber block (2x2x2 in.) inside. Throughout habituation and test, infants viewed

two human puppets, measuring 13 inches tall. We used facial stimuli similar to those

used with toddlers and adults attached to the bodies of human-like puppets whose heads

were replaced with the trustworthy and untrustworthy faces, which measured 4.75 x 7.5

inches (Cogsdill et al., 2014). The faces used in Experiment 2 were the same faces used

in Experiment 1.

Procedure

Familiarization Phase. The study began with a Familiarization phase that

included one example of both the helping and hindering events. In the Helping

Familiarization Event, participants witnessed the cat puppet (the struggler) attempt to

open a clear box with an orange block inside. A puppy puppet entered and helped the cat

open the box. In the Hindering Familiarization Event, the actions were the same except

the cow puppet entered and stomped on the lid of the box. Looking time for both of these

trials were recorded as a baseline measurement. A trial ended if the infant looked away

from the stage for 2 consecutive seconds or if 60 seconds elapsed. Order of events was

counterbalanced across infants.

26

Habituation Phase. Following familiarization, infants were habituated to the

trustworthy and untrustworthy puppet characters placed 15 in. away from each other on

the stage. Each puppet was similar except for the color t-shirt they were wearing (red or

white), which was counterbalanced across babies. Each habituation trial began with one

of the puppets jumping quickly in place two times, followed by a one second pause,

followed by the same puppet jumping quickly in place two more times. After another one

second pause, the other puppet performed the same actions. Each trial took approximately

12 seconds. Once both characters had been manipulated all motion on the stage stopped

and infants’ looking times were recorded. A trial ended if the infant looked away from

the stage for 2 consecutive seconds or if 60 seconds elapsed. At the conclusion of each

trial, a black curtain was lowered to occlude the entire stage, the position of each

character was switched, and the curtain was lifted to commence the next trial. Infants

were presented with 6 habituation trials.

Test Phase. Infants were presented with 6 test trials consisting of 3 pairs of both

the helping and hindering events, with one puppet (either trustworthy or untrustworthy)

performing the actions. Each trial begun with the cat puppet attempting to open a box to

retrieve a block in the presence of the two characters from the Habituation phase who

were located on either side of the box event. Just as in the Familiarization phase, the

struggler was seen attempting to open his box. On Helping Trials, infants saw one of the

two characters move from the side of the stage towards the box located in the middle, and

along with the struggler, helped to open the box for the struggler to retrieve his toy. The

character then moved back to its starting position, all motion on the stage stopped, and

infants’ looking time was recorded. A trial ended when the infant looked away for 2

27

seconds or if 60 seconds elapsed. At the end of each trial the curtain lowered, the stage

was reset, and the trial was repeated with the same character performing the hindering

event. On Hindering Trials, infants saw one of the two characters move from the side of

the stage towards the box located in the middle, and as the struggler was trying to open

his box, jumped on the lid hindering him from retrieving his toy. The character then

moved back to its starting position, all motion on the stage stopped, and infants’ looking

time was recorded. At the end of each trial the curtain lowered, the stage was reset, and

the trial repeated with the same character performing the helping event. The side the

characters were placed on and the order in which the characters helped or hindered was

counterbalanced across participants.

Looking time was measured by an on-line observer placed in a separate room who

observed the infant via a closed circuit camera and recorded the duration of their looks

using the jHAB coding program (Casstevens, 2007). Following the live coding, a

different observer who was blind to the condition completed a second set of off-line

measurements and obtained an inter-observer reliability score of .90.

Results

Baseline Trials. Infants showed no preference for the helping or hindering events

during familiarization trials (14.26 s vs. 17.68 s, t(15)= -.924, p=.370).

Test Trials. The main result for Experiment 1a is depicted in Figure 5. Overall,

infants successfully discriminated inconsistent from consistent face-action pairings. A

pair by type (consistent vs. inconsistent) ANOVA revealed a significant difference

between pairs, F(2,30)=4.14, p=.026, with a decrease in looking between the first and last

pair. Critically, we found a significant difference between the consistent and inconsistent

28

face-action pairing, F(1,15)=6.08, p=.026 with infants looking longer at the inconsistent

(M=17.02, SD=6.77) vs. consistent (M=13.00, SD=4.13) pairing. A significant majority

of infants followed this pattern (13 of 16 infants; p=.021, two-tailed sign test). No other

significant effects or interactions were found.

Figure 5: Looking time (s) between the Consistent and Inconsistent feature-action

pairings.

Discussion

The results of Experiment 2a suggest that by 10 months, infants are able to match

facial appearance along the dimension of ‘trustworthiness’ to a characters’ social and

behavioral disposition. Overall, infants looked longer at test events that were inconsistent

with the characters’ face-traits. While this pattern of results may be due to infants’ ability

to infer social disposition from facial features, infants may be showing this pattern due to

a mismatch between the positive/negative valence of the agent and the positive/negative

0

2

4

6

8

10

12

14

16

18

20

Consistent Inconsistent

*

29

valence of the action itself (i.e. closing the box). In this scenario, the agent does not need

to cause the positive or negative behavior, infants are simply responding to matching or

mismatching valances of the facial features and actions. Results from a previous study

suggest infants are able to go beyond simple valence matching and believe that agents

must cause the behavior for the agent to be considered positive or negative (Lyons &

Cheries, under review.). Experiment 2b was designed to rule out this possibility in the

current study.

Experiment 2b: Behavioral Inference Task Control (infants)

Experiment 2b was designed to determine if infants were simply associating the

valence of the facial features to the valence of the action without any regard to whether or

not the puppets were causing the action. To test this, we replicated the procedure from

Experiment 2a, but instead of the agent helping or hindering the neutral character, the

agent simply stood next to the test event while it occurred. For instance, the agent was

next to the box while the neutral character was successful and opened it, or next to the

box while the neutral character was not successful and could not open the box. We

hypothesized that if infants were simply associating the valence of the agent to the

valence of the action, we should observe the same pattern of results as Experiment 2a.

However, if infants’ response was due to the agent causing an expected or unexpected

behavior, we hypothesize no difference in looking time between the consistent and

inconsistent actions.

Methods

30

Participants

We tested a total of ten 10-month-old infants (4 males) from the greater Amherst

area (range= 9 months 18 days to 10 months 15 days; mean=10 months 2 days). We will

recruit six more infants to complete the counterbalance.

Materials

The materials used in Experiment 2b were identical to the materials used in

Experiment 2a.

Procedure

The procedure and measurements of Experiment 2b were similar to Experiment

2a. During familiarization, infants viewed the cat puppet (the ‘struggler’) attempting to

open the clear box to retrieve the toy inside. The puppet was either successful in opening

the box, or unsuccessful and unable to open the box. The habituation phase was identical

to the habituation in Experiment 2a. The test trials of Experiment 2b followed the same

pattern of the familiarization trials. During test trials, the agents moved in from the side

of the stage and paused 3 inches from the box while the struggler either successfully

opened the box or was unsuccessful. Once the positive or negative action was completed,

the agent moved back to its starting position.

The coding measures used in Experiment 2b were identical to the measures used

in Experiment 2a. Following the live coding, a different observer blind to the condition

completed a second set of off-line measurements and was required to obtain an inter-

observer reliability score of .90.

Results

31

Familiarization Trials. Infants showed no preference for the helping or hindering

events during familiarization trials (17.8s vs. 17.06s, t(9)=.208, p=.840).

Test Trials. The main result for Experiment 1b is depicted in Figure 6. Overall,

infants did not successfully discriminate inconsistent from consistent face-action pairings.

A pair by type (consistent vs. inconsistent) ANOVA revealed a marginally significant

difference between pairs, F(2,18)=3.49, p=.052, with a decrease in looking from Pair 1 to

Pair 3. Critically, we found no significant difference between the consistent and

inconsistent face-action pairing, F(1,9)=1.33, p=.278 (See Figure 6).

Once the data has been completely collected, we will investigate any differences

between condition.

Figure 6: Looking time (s) between the consistent and inconsistent trials in Experiment

2b.

Discussion

0

2

4

6

8

10

12

14

16

18

20

Consistent Inconsistent

32

Unlike Experiment 2a, infants did not successfully discriminate between

consistent and inconsistent action and face-trait pairings. That is, infants did not look

longer at test trials where the social face trait (trustworthy or untrustworthy) did not

match the general valence of the outcome of the event (the struggler opening or closing

the box). This falls in line with results from a previous study, suggesting that infants are

able to go beyond simple valence matching as well as believe that an agent must cause

the behavior for the agent to be considered positive or negative (Lyons & Cheries, under

review.). Taken together, the findings from Experiment 2 suggest that by 10 months of

age, infants possess the ability to use the facial trait of trustworthiness to determine a

character’s behavior.

33

CHAPTER IV

EXPERIMENT 3

Experiment 3a: Behavioral Inference Task (Toddlers)

The results of Experiments 1 and 2 suggest that infants may draw some relatively

implicit connection between facial appearance and social disposition. In order to examine

whether this early ability is continuous with later competence that may rely upon more

explicit inferences we adapted the infant task for testing toddlers. The question is whether

or not toddlers can correctly choose between characters with trustworthy- and

untrustworthy-looking faces (by adult judgment) when asked to guess who has performed

a helpful or unhelpful action. Previous research with older 3-year-old toddlers has shown

that they associate faces that are high and low on the trustworthy scale to the terms “nice”

and “mean”(Cogsdill et al., 2014). However, this previous work did not determine

whether this inference is connected to actual social behavior or only the linguistic label.

Our study tested younger toddlers in a non-verbal response task that required them to

match a particular social behavior with a character that possessed either a trustworthy- or

untrustworthy-looking face.

Methods

Participants

Two separate age groups of infants were tested, including sixteen 17- to 19-

month-old infants (range=17 months, 3 days to 19 months, 1 day; mean=18 months 4

days), and 16 (8 males) 2- to 2.5-year-old infants (range=2 years, 9 days to 2 years, 6

months, 19 days; mean=2 years, 3 months, 26 days) all from the greater Amherst area.

An additional eight infants were tested but excluded from the analysis due to

34

experimenter error (2), a language barrier (2), and an unwillingness to choose/fussiness

(4).

Materials

Infants sat in their parents’ lap on one side of a card table (34 by 34 in.) with the

experimenter sitting directly opposite. The two puppets used in the test trials were

identical to the ones used in Experiment 2. Additional materials were used in both the

warm-up phases as well as the test trials. The following materials were used in the warm-

up phases: a white sheet of paper and a blue crayon, a small box filled with a variety of

animal magnets, one sheep puppet, one puppy puppet, and two lime green foam visors.

The following materials were used in the test phases, which consisted of four different

test trials (Ball, Box, Platform, and Blocks). In the Ball trial, a 3-inch spherical orange

ball was used. In the Box trial, the same box and orange block from Experiment 2 was

used. In the Platform trial, a red block (2x2x2) and a platform (27x10x11.5 in.) were

used. In the Block trial, six multicolored blocks (each 2x2x2) were used. Across all four

test trials, a cat puppet was shown to the infants, as well as one human puppet whose face

was occluded by a lime green foam visor.

Procedure

Warm-Up Phase. The study began with two choice warm up trials followed by a

warm up trial to the puppets and their foam visors (see Figure 7). In the Crayon Warm-

Up, infants were shown a puppet drawing swirls on a piece of paper. The experimenter

said “Look, this puppet is drawing, and he has a blue crayon. He’s going to draw some

pretty swirls with his crayon.” After the action, the experimenter removed the puppet

from their hands, placed both hands on the table in view of the infant, and then reached

35

down and put one puppet on each hand. The experimenter then asked the infant “Which

puppet did you see drawing?” Infants were given 10 seconds to choose a puppet before

being prompted a second time. In the Magnet Warm-Up, infants were given three animal

magnets from the other puppet. The experimenter said “Look, this puppet wants to give

you some animal magnets.” After the magnets were given to the infant, the experimenter

performed the same sequence of events but instead asked the infant “Which puppet gave

you animals?”

After each choice warm up task, infants were shown the two characters they

would be choosing from during test events, the untrustworthy and trustworthy character.

Each character looked identical, except for the valence of trustworthiness of their faces.

Infants were shown each character and were told that they liked to wear silly hats. Infants

were then given the opportunity to play with one of the hats for a few seconds, before the

experimenter showed each of the characters putting their hats on. The experimenter then

pointed out that the hats covered the characters faces.

Test Phase. Infants were presented with 4 test trials, the Box task, Ball task,

Platform task, and Blocks task. Across each trial, infants viewed George the cat

attempting to complete a goal. Also on the table was one of the human puppets with his

face occluded by a green visor. Infants saw this character either help or hinder George the

cat. After each trial, the experimenter removed all puppets from the stage, placed their

uncovered hands on the table, and then reached down to pick up the trustworthy and

untrustworthy puppets. Infants were asked “Who did you see perform (task)?” and were

given 10 seconds to point to or reach towards one of the two puppets. The order of the

36

tasks, whether it was a helping event or hindering event, and the side the trustworthy and

untrustworthy puppets were presented on were counterbalanced.

The four tasks are described below. In the Box Task, infants were introduced to a

cat puppet named George. This task was similar to the task used in Experiment 2, only

the experimenter narrated the scenario while acting it out with the puppets. Infants

viewed the cat struggling to open the box to retrieve his block from inside. In the left

corner of the card table, infants saw one of the human puppets with his face occluded by

a green visor. Infants saw this character either help George the cat or hinder him, before

moving back to the corner of the table. In the Ball Task, infants were shown George the

cat playing with an orange ball. The infant was told that George wanted to play a ball

game with his friend. They see George roll the ball to the puppet, which either rolls it

back to him or steals it away. In the Platform Task, infants see a red block in the center of

a tall platform, and the human puppet sitting on the left side of the platform. George the

cat is shown trying to retrieve the red block from the top of the platform, but he is unable

to reach. The puppet either comes over and gives George the block, or steals it away from

him. In the Block Task, infants view George the cat trying to build a block pyramid. As

he struggles to place the last block on top of the pyramid, the human puppet comes over

and either helps him place his block on the top or knocks the tower down.

After viewing a scenario, infants were given a choice of the trustworthy and

untrustworthy puppet. A choice was defined as a point, a touch, or a combination of one

of those and a verbal statement from the infant. Toddlers were only excluded if they were

fussy and made it clear (verbally) that they did not want to choose. The four subjects that

37

were excluded due to an unwillingness to choose did not want to make choices on more

than one trial.

a. b.

Figure 7: The trustworthy and untrustworhty characters without their visors (a) and with

the visor (b).

Results

17-19 Month-Old Infants. Overall, infants did not significantly discriminate

between the trustworthy and untrustworthy character during test trials (36 of 60 trials;

p=.16, two-tailed sign test). Infants in this age group made correct choices above or at

chance across all four trials. Infants did not perform differently when viewing the helping

(17 of 28 trials; p=.34, two-tailed sign test) and hindering (19 of 32 trials; p=.37, two-

tailed sign test) actions. Infants did not perform differently across tasks. Overall, males

did perform significantly better than females (Males: 21 of 30 trials; p=.04, two-tailed

sign test; Females: 15 out of 30; p=1.14; see Figure 8).

38

Figure 8: Number of correct vs. incorrect choices by gender, age 17-19 months.

24-30 Month-Old Infants. Across all four test trials, infants did not discriminate

between the trustworthy and untrustworthy characters during test trials (37 of 67 trials;

p=.45). A chi-square test of goodness-of-fit was performed to determine if the four trials

had an equal number of correct responses. Correct responses were not equally distributed,

X2 (3, N=16)=9.81, p=.02. A closer inspection revealed that infants chose at or above

chance during the first three test trials, but well below chance for the last test trial, with

only two of the 16 infants choosing correctly. This confirmed observer’s reports that

infant’s attention seemed to decline throughout the study. Based on this, the following

results are reported for the first three trials.

Overall, infants significantly discriminated between the trustworthy and

untrustworthy characters during test trials (34 of 48 trials; p=.005, two-tailed sign test;

see Figure 9). When viewing a helping action, infants significantly chose the trustworthy

0

5

10

15

20

25

Male Female

Correct

Incorrect

*

39

character over the untrustworthy character (18 of 24 trials; p=.02, two-tailed sign test),

however, infants did not significantly choose the untrustworthy character when viewing a

hindering action (16 of 24 trials; p=.15, two-tailed sign test; see Figure 10).

Infants did not perform differently across any individual tasks. Overall, males did

perform significantly better than females (Males: 19 out of 24 trials; p=.006, two-tailed

sign test; Females: 15 out of 24 trials; p=.31; see Figure 11).

During the familiarization phase, 12 out of 16 infants had at least one correct

choice between the two tasks. An analysis of the first three trials of data from these 12

participants revealed a similar level of overall discrimination between the trustworthy and

untrustworthy characters during test trials (27 out of 36 trials; p=.004, two-tailed sign

test). Infants discriminated between the trustworthy and untrustworthy characters during

the helping event (13 out of 17 trials; p=.04, two-tailed sign test) but not the hindering

event (14 out of 19 trials; p=.06, two-tailed sign test). There was no difference in any of

the individual tasks. We did find a difference in performance based on gender, with males

having more successful trials (17 out of 21 trials; p=.007) than females (10 out of 15

trials; p=.30, two-tailed sign test), however three of the four infants that were excluded

due to not choosing during the familiarization trials were females.

40

Figure 9: Number of correct vs. incorrect overall choices, age 2-2.5 years.

Figure 10: Number of correct vs. incorrect choices by helping or hindering trial, age 2-2.5

years.

0

5

10

15

20

25

30

35

40

Correct Incorrect

*

0

2

4

6

8

10

12

14

16

18

20

Help Hinder

Correct

Incorrect

*

41

Figure 11: Number of correct vs. incorrect choices by gender, age 2-2.5 years.

Experiment 3a Discussion

The results from Experiment 3a suggest that by two years, toddlers and are able to

match the different facial characteristics to the correct social actions. Their responses

indicated that they believe that characters with trustworthy looking faces (by adult

judgment) should behave in helpful ways, and untrustworthy-looking characters should

behave in unhelpful ways. While we did not find a significant overall effect with our 17-

19 month age group, we did see a gender effect, with males discriminating between the

two faces. Infants who fell between 2-2.5 years of age successfully discriminated

between the trustworthy and untrustworthy character when viewing helping and

hindering actions committed by a character whose face was occluded. Similar to the

pattern found with the 17-19 month old infants, males but not females in this age group

were also more successful at discriminating between the faces and actions. Additionally,

the pattern of results was stronger when infants viewed an occluded character performing

0

5

10

15

20

25

Male Female

Correct

Incorrect

*

42

a helping action. This lack of discrimination found in the hindering condition alone was

unexpected since prior results with adults have suggested the possibility of a ‘negativity

bias’ in this domain (e.g. Baumeister, Bratslavsky, Finkenauer & Vohs, 2001). One

possible explanation for the current results may be due to the particular hindering actions

not being as strongly negative compared to the degree that the helpful actions are

construed as positive. This possibility would need to be explored in future versions of

both scenarios.

As discussed in Experiment 1a, while infants were successful at discriminating

between the faces, we are unsure if this discrimination occurred only in the helping

condition due to an aversion to the untrustworthy face. If infants were avert to the

untrustworthy face, we would expect the same pattern of results as we observed, being

higher discrimination in the helping task only. To test if infants are only responding due

to an aversion to the untrustworthy face, an experiment similar to Experiment 1b was

conducted. Toddlers were placed into one of two conditions, either having to make a

choice between the trustworthy face and a neutral face, or between the untrustworthy face

and a neutral face. If toddlers are only averse to the untrustworthy face, we hypothesize a

majority of correct choices in the untrustworthy vs. neutral condition and random choice

in the trustworthy vs. neutral condition. However, if toddlers are thinking of these faces

as trustworthy and untrustworthy, then we hypothesize a majority of correct choices in

both conditions.

Experiment 3b: Behavioral Inference Task Control (toddlers)

Similar to the logic of the proposed control for the infants’ crawling task

(Experiment 1b), the goal of Experiment 3b was to determine the underlying cause of

43

toddlers’ choice between the trustworthy and untrustworthy faces. That is, do toddlers

choose the correct puppet by reasoning about the trustworthy face, the untrustworthy

face, or by evaluating the social action in relation to both? To assess this, toddlers were

placed in one of two conditions, one group was given the choice between the trustworthy

and a neutral character, and the other group was given the choice between the

untrustworthy and a neutral character. Previous research has shown that by the age of 3,

toddlers believe that those who look trustworthy are “nice”, and those who look

untrustworthy are “mean” (Cogsdill et al., 2014). Therefore if toddlers do in fact

represent both the positive and negative associations with these faces equally then we

hypothesized toddlers will be equally successfully at choosing the correct face when it is

contrasted with a neutral-faced puppet.

Methods

Participants

We recruited thirty-two 24-30 month-old toddlers (range=2 years, 8 days to 2

years 5 months 28 days; mean=2 years, 2 months, 9 days), (16 males) from the greater

Amherst area . Half of this sample participated in the trustworthy vs. neutral condition,

and the other half participated in the untrustworthy vs. neutral condition. An additional

five toddlers were tested but excluded from the sample due to unwillingness to

choose/fussiness.

Materials

The materials used in Experiment 3b were identical to the materials used in

Experiment 3a, however subjects now viewed a neutral face along with either the

trustworthy or untrustworthy face. The neutral face was three standard deviations away

44

from both the trustworthy and untrustworthy face on the trustworthiness dimension (See

Figure 12).

Procedure

The procedure of Experiment 3b was identical to the procedure of Experiment 3a,

except subjects were in one of two conditions. Subjects in the Trustworthy vs. Neutral

Condition were allowed to make a choice between the trustworthy or neutral face in the

test phase, and subjects in the Untrustworthy vs. Neutral Condition were allowed to

choose between either the untrustworthy or neutral face in the test phase.

a. b.

Figure 12: The neutral and untrustworthy puppets (a) and the neutral and trustworthy

puppets (b).

Results

Trustworthy vs. Neutral. Overall, toddlers did not significantly discriminate

between the trustworthy and neutral characters during test trials (34 of 64 trials; p=.71,

two-tailed sign test; see Figure 13), nor did they discriminate based on condition viewed

(Helping: 16 of 32 trials; p=1.14, two-tailed sign test; Hindering: 18 of 32 trials; p=.6,

two-tailed sign test).

45

Toddlers did not perform differently across any individual tasks. Overall, males

and females did not perform differently, with both groups choosing correctly the same

amount of times (17 of 32 trials; p=.86, two-tailed sign test).

During the familiarization phase, 12 out of 16 toddlers had at least one correct

choice between the two tasks. An analysis of data from only these 12 participants

revealed a similar pattern of overall non-discrimination between the trustworthy and

neutral characters during test trials (24 of 48 trials; p=1.11, two-tailed sign test). There

was no difference between helping and hindering tasks, type of task, or gender.

Untrustworthy vs. Neutral. . Overall, toddlers did not significantly discriminate

between the untrustworthy and neutral characters during test trials (37 of 63 trials; p=.21,

two-tailed sign test; see Figure 13), nor did they discriminate based on condition viewed

(Helping: 17 of 32 trials; p=.86, two-tailed sign test; Hindering: 20 of 31 trials; p=.15,

two-tailed sign test).

Toddlers did not perform differently across any individual tasks. Overall, males

(19 of 31 trials; p=.28, two-tailed sign test) and females (18 of 32 trials; p=.6, two-tailed

sign test) did not perform differently.

During the familiarization phase, 15 out of 16 toddlers had at least one correct

choice between the two tasks. An analysis of data from only these 15 participants

revealed a similar pattern of overall non-discrimination between the untrustworthy and

neutral characters during test trials (35 of 59 trials; p=.19, two-tailed sign test). There was

no difference between helping and hindering tasks, type of task, or gender.

46

Figure 13: Number of correct vs. incorrect overall choices, by condition.

Experiment 3b Discussion

The results from Experiment 3b suggest that toddlers need both the trustworthy

and untrustworthy faces paired together to connect them to helping and hindering

behavior. In Experiment 3b, toddlers in both the trustworthy vs. neutral condition and the

untrustworthy vs. neutral condition did not significantly discriminate between the faces

when choosing who should have completed the nice or mean action. While the

trustworthy and untrustworthy faces are +/-3SD from the neutral face, it is possible that

when one of the faces is paired with the neutral face, the distinction between the two is

not strong enough for toddlers to pick up on.

0

5

10

15

20

25

30

35

40

Trustworthy vs. Neutral Untrustworthy vs. Neutral

Correct

Incorrect

47

CHAPTER V

GENERAL DISCUSSION

Taken together, the current studies suggest an early sensitivity to the facial

features that convey the trait of trustworthiness in adults. By 10 months of age infants

exhibit a preference to crawl towards a food reward placed behind a trustworthy rather

than an untrustworthy face. Infants’ willingness to crawl towards the trustworthy-looking

face suggests that these feature combinations elicit positive evaluations that may be

similar to those observed in adults. Although infants’ immediate goal in this task was to

obtain a desirable graham cracker treat from one of two containers (that contained equal

amounts), we found that the valenced associations elicited by the face stimuli

significantly biased subjects’ ultimate approach behavior.

Infants’ crawling behavior was likely due to a largely implicit and general

positive/negative social evaluation based upon the trustworthy/untrustworthy facial cues.

Of course, the specific evaluation driving infants’ approach/avoidance behavior may not

be based on the attribution of trustworthiness, per se, but instead on some further

downstream or related judgment. For instance, their ultimate crawling may have been

biased on a preference for the face they found more attractive rather than then one they

found more trustworthy. That is, they may view more trustworthy faces as more

attractive, as has been documented in studies with adult subjects (Dion, 1973; Ma & Xu,

2015; Willis & Todorov, 2006). Infants have been shown to prefer attractive over

unattractive faces (Griffey & Little, 2014; Hoss & Langlois, 2003; Langlois, Ritter,

Roggman, & Vaughn, 1991; Langlois, et al., 1987). Accordingly, infants may have a

48

preference to the trustworthy face because they view it as the more attractive face,

overall.

An additional possibility is that infants associate the facial features in our stimuli

with ‘mean’ and ‘nice’ actions. Infants may expect that those who look nice could help

them in the future, and vice versa. Of course, because the face stimuli in the current

experiment were attached to buckets, and were therefore disembodied, infants’ crawling

responses should not be due to any expectation of being helped/hindered or being the

recipient of some positive/negative action during the task. However, prior research has

shown that infants do have a preference to approach those who have performed nice

actions in the past (based on preferential reaching) and outside of any context where the

infant’s preference could be explained by expectations of immediate usefulness (i.e.

Hamlin & Wynn, 2011; Hamlin, Wynn, & Bloom, 2007). Instead, the

approach/avoidance behaviors towards mean- and nice-looking individuals may be based

upon positive associations driven by prior observations of usefulness in real-world

situations. Adults perform approach and avoidance behaviors after viewing various facial

stimuli, such as angry and fearful faces (Adams et al., 2006; Marsh & Ambady, 2007;

Marsh et al., 2005; Marsh et al., 2007 ). Finding similar patterns in infants might suggest

that deciding whom to approach or avoid from their facial features is an adaptive

response that could have increased the likelihood of infants experiencing positive social

outcomes over development or evolutionary time.

The preliminary results from Experiment 1b are somewhat inconclusive. So far

these data suggest that the crawling behavior we originally observed requires the contrast

of both the trustworthy and untrustworthy faces together. When compared with only the

49

neutral face, the trustworthy or untrustworthy face stimuli we used may not be strong

enough to elicit a clear evaluation in infants this age.

We also show that by 10 months of age, infants seem to use facial appearance to

form expectations about a characters’ social behavior towards others. Specifically, in

Experiment 2a, infants attended longer to events involving characters who exhibited

facial features that were inconsistent with them helping or hindering another’s goal. This

builds upon existing results demonstrating that infants form expectations about an agents’

social behavior in advance based upon certain audio-visual features (Lyons & Cheries,

under review). Additionally, the results from Experiment 2b suggests that infants only

associate a character’s features to their behavior when the character causes the behaviors

themselves; it is not enough for the character to merely be near the outcome for infants to

associate the character with the behavior.

This pattern of associating facial features to an agents’ behavior continues into

early toddlerhood, as evidenced in Experiment 3. By 2 years of age, infants infer that

when a nice action is performed, it is more likely that someone with trustworthy features

was involved, and vice versa. While the effect is not as robust in 17-19 month old infants,

we did find this pattern in male subjects. Our lack of an overall effect could be attributed

to the level of difficulty of the task, possibly due to the older toddlers having increased

language as well as attention. Male subjects age 2-2.5 years were also more likely to

make correct choices between the characters when presented with the trustworthy and

untrustworthy faces together. Recent research with adults has shown a similar gender

difference between males and females when viewing male faces. Researchers found that

gender (specifically hyper-masculinity) is correlated to judgments of trustworthiness of

50

faces (Macapagal, Rupp, & Heiman, 2011). In line with Study 1, it seems that toddlers

also need to view both the trustworthy and untrustworthy faces to correctly choose who

was more likely to perform the social action.

While these results suggest that infants infer a connection between one’s

appearance and behavior, it is important to stress that this does not suggest a causal

relationship between behavior. As previously mentioned, mixed results have been

obtained in adult studies investigating the connection between appearance and personality

(e.g. Berry, 1991; Penton-Voak et al., 2006; Zebrowitz et al., 1996). On the one hand,

when using self-report personality measures in conjunction with face-trait inferences,

results suggest a correlation between facial features and personality traits such as warmth,

approachability, and extroversion (e.g. Berry, 1991; Penton-Voak et al., 2006). On the

other hand, studies using behavioral or observational methods show no correlation

between facial ratings and personality (e.g. Zebrowitz et al., 1996). Future studies with

infants should investigate to what extent they believe facial appearance causes a

particular behavior.

It is also important to investigate the extent to which infants are using facial

appearance to attribute a trait, per se. That is, it is possible that infants perceive a face as

being trustworthy/nice in a manner where it represented as a stable trait that cannot be

changed. Alternatively, infants could think of such face-based attributes as properties that

are more similar to transient emotions, which may change from one moment to the next.

One way of distinguishing these possibilities is to use a task that tests infants’

expectations about the number of faces involved in an event. For example, when

someone varies their emotions from one moment to the next (e.g., from happy to sad),

51

they are still represented as the same person. It is an open question in the current studies

whether infants’ infer social dispositions from facial appearance in a similarly flexible

way or in a way where a face should not change from trustworthy to untrustworthy-

looking. We are currently investigating this distinction in a study asking whether infants

perceive an appearance of a trustworthy-looking face followed by an untrustworthy-

looking face as two distinct individuals. This and other individuation manipulations will

help determine the nature of infants’ social inferences.

Results from these studies suggest an early developing understanding of the facial

trait of trustworthiness. These results present us with a few possibilities of where this

ability comes from, as well as what the purpose of the ability may be. When thinking

about where this ability comes from, one possibility is that it is strictly experience

dependent. This would mean that our response to facial appearance has been shaped by

early experiences, such as seeing people with specific combinations of facial features

perform nice and mean actions. Applying this interpretation to our current results would

require that by 10 months infants have had sufficient exposure to people who consistently

engage in trustworthy or untrustworthy actions who also exhibit consistent facial features

to one another. The incidence of untrustworthy or even ‘mean’ actions towards 10 month

old infants is hopefully infrequent, but we can further test this hypothesis by reducing the

age of the infants we test in this paradigm. We can also explore the degree to which face-

trait inferences are biased by previous facial experience by examining individual

differences in the type of faces they are exposed to. If facial experience is required for the

type of effects we have observed in the current study than we might predict an interaction

52

between infants’ performance in our tasks and the face-trait ratings their caregiver’s faces

elicit from other adults.

Regardless of the initial origins, the ability to differentiate faces is something that

increases with age. Many studies investigating multiple age groups have shown an

increase in reliable trait-based inferences over development (Cogsdill & Banaji, 2015;

Cogsdill et al., 2014; Ma & Xu, 2015). This developmental increase could be related to

brain maturation (Baron, Gobbini, Engell. & Todorov, 2011; Giedd et al., 1996) as well

as the increase of face perception abilities (Mondloch, Geldart, Maurer, & Le Grand,

2003). This developmental change may also be due to the type of task that is used with

subjects. In this set of experiments, we found evidence that infants at 10 months and

toddlers at 2 years have the ability to determine probably social behavior from faces,

however we did not find that the same held true for 17- month-old infants. This u-shaped

developmental curve we found in subjects’ performance may be due to the difficulty of

the choice task for our 17-19 month old infants, more specifically the difficulty this age

group may have faced using the explicit measures we put forth. This pattern is not

unusual in the field, in fact, many studies have shown a difference in performance when

using implicit over explicit measures. For example, differing results have been found in

tasks investigating reasoning about hidden objects. In these tasks, infants and toddlers

view a ball rolling down an incline, and also view a wall being placed in the path of the

ball, which would stop the ball from continuing to roll. Using explicit measures (reaching

to the location of the ball after the wall has been placed), results suggest that 3 year olds,

but not 2 or 2.5-year-old toddlers, can reason about the location of the ball when it is

blocked by the wall (Berthier, DeBlois, Poirier, Novak, & Clifton, 2000). However,

53

looking time studies (an implicit task) with 2.5-month-old infants suggests that infants

can reason about the location of the ball, and expect the ball to stop rolling when blocked

by the wall (Spelke, Breinlinger, Macomber, & Jacobson, 1992). One future possibility is

to investigate the ability by creating a task that may be easier for toddlers in that age to

perform. This holds true for other studies looking at developmental differences in trait-

based inferences throughout childhood. The same task that is used with older children

may not be appropriate for three year olds.

Future research should investigate which other facial traits are salient for infants

and whether infants make social inferences based on these perceived facial traits as well.

By the age of 3 years, toddlers can differentiate faces based on competence and

dominance, in addition to trustworthiness (Cogsdill et al., 2014). Toddlers attribute the

qualities of ‘smart’ and ‘not smart’ to competent and incompetent faces, as well as ‘strong’

and ‘not strong’ to dominant and submissive faces. By 10 months, infants have the ability

to evaluate social dominance based upon a character’s size (Thomsen, Frankenhuis,

Ingold-Smith, & Carey, 2011) and a character’s actions, such as competing to collect

objects or occupy a space (Mascaro & Csibra, 2012; 2014). Recent results also suggest

infants are able to use the transitive property to determine social dominance hierarchies

(Gazes, Hampton, & Lourenco, 2015). Looking at these results overall as well our

aforementioned studies, it is possible that dominance qualities will elicit the same

responses using a similar paradigm. Studies looking at both dominance and competence

and the social evaluations that infants and toddlers make regarding these dimensions are

currently being conducted.

54

The current set of studies are the first to examine the development of face-trait

attributions across infancy. Our results suggest that infants in the first year of life are

biased to ‘judge a book by its cover.’ Similar to recent arguments that infants connect a

character’s social behavior to its sound and general physical appearance (Lyons &

Cheries, under review), the current results suggest that facial appearance may be a

particularly salient indicator for inferring another’s social disposition early in

development.

55

BIBLIOGRAPHY

Adams, R. R., Ambady, N., Macrae, C., & Kleck, R. E. (2006). Emotional expressions

forecast approach-avoidance behavior. Motivation and Emotion, 30(2), 179-188.

doi:10.1007/s11031-006-9020-2

Ambady, N. (2010). The perils of pondering: Intuition and thin slice judgments.

Psychological Inquiry, 21(4), 271-278. doi:10.1080/1047840X.2010.524882

Ambady, N., Hallahan, M., & Connor, B. (1999). Accuracy of judgments of sexual

orientation from thin slices of behavior. Journal of Personality and Social

Psychology, 77(3), 538-547. doi: 10.1037/0022-3514.77.3.538

Ambady, N., & Rosenthal, R. (1993). Half a minute: Predicting teacher evaluations from

thin slices of nonverbal behavior and physical attractiveness. Journal of

Personality and Social Psychology, 64(3), 431-441. doi:10.1037/0022-

3514.64.3.431

Antonokis, J., & Dalgas, O. (2009). Predicting elections: Child’s play! Science,

323(5918), 1183.

Balas, B., Kanwisher, N., & Saxe, R. (2012). Thin-slice perception develops slowly.

Journal Of Experimental Child Psychology, 112(2), 257-264.

doi:10.1016/j.jecp.2012.01.002

Bar-Haim, Y., Ziv, T., Lamy, D., & Hodes, R. M. (2006). Nature and nurture in own-race

face processing. Psychological Science, 17(2), 159-163. doi:10.1111/j.1467-

9280.2006.01679.x

Baron, S.G., Gobbini, M.I., Engell, A.D., & Todorov, A. (2011). Amygdala and

prefrontal cortex responses to appearance-based and behavior-based person

impressions. Social Cognitive & Affective Neuroscience, 6, 572-581. doi:

10.1093/scan/nsq086

Barrera, M. E., & Maurer, D. (1981). Discrimination of strangers by the three-month-old.

Child Development, 52(2), 558-563. doi:10.2307/1129174

Baumeister, R.F., Bratslavsky, E., Finkenauer, C., & Vohs, K.D. (2001). Bad is stronger

than good. Review of General Psychology, 5(4), 323-370. doi: 10.1037/1089

2680.5.4.323

Berry, D. S. (1991). Accuracy in social perception: contributions of facial and vocal

information. Journal Of Personality And Social Psychology, 61(2), 298-307.

doi:10.1037/0022-3514.61.2.298

56

Berry, D. S., & Brownlow, S. (1989). Were the physiognomists right? Personality

correlates of facial babyishness. Personality And Social Psychology Bulletin,

15(2), 266-279. doi:10.1177/0146167289152013

Berthier, N. E., DeBlois, S., Poirier, C. R., Novak, M. A., & Clifton, R. K. (2000).

Where's the ball? Two- and three-year-olds reason about unseen events.

Developmental Psychology, 36(3), 394-401. doi:10.1037/0012-1649.36.3.394

Casstevens, R.M. (2007). jHab: Java Habituation Software (Version 1.0.0) [Computer

Software]. Chevy Chase, MD.

Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Immediate

behavioral predispositions to approach or avoid the stimulus. Personality and

Social Psychology Bulletin, 25(2), 215-224. doi:10.1177/0146167299025002007

Cogsdill, E.J., and Banaji, M.R. (2015). Face-trait inferences show robust child-adult

agreement: Evidence from three types of faces. J. Exp. Soc. Psychol.60, 150-156.

doi:10.1016/j.jesp.2015.05.007.

Cogsdill, E. J., Todorov, A. T., Spelke, E. S., & Banaji, M. R. (2014). Inferring character

from faces: A developmental study. Psychological Science, 25(5), 1132-1139.

doi:10.1177/0956797614523297

Cunningham, W. A., Van Bavel, J. J., & Johnsen, I. R. (2008). Affective flexibility:

Evaluative processing goals shape amygdala activity. Psychological Science,

19(2), 152-160. doi:10.1111/j.1467-9280.2008.02061.x

Dion, K. K. (1973). Young children's stereotyping of facial attractiveness. Developmental

Psychology, 9(2), 183-188. doi:10.1037/h0035083

Engell, A. D., Haxby, J. V., & Todorov, A. (2007). Implicit trustworthiness decisions:

Automatic coding of face properties in the human amygdala. Journal of Cognitive

Neuroscience, 19(9), 1508-1519. doi:10.1162/jocn.2007.19.9.1508

Ewing, L., Caulfield, F., Read, A., and Rhodes, G. (2015). Perceived trustworthiness of

faces drives trust behaviour in children. Dev. Sci. 18, 327-334. doi:

10.1111/desc.12218.1

Farroni, T., Menon, E., Rigato, S., & Johnson, M. H. (2007). The perception of facial

expressions in newborns. European Journal Of Developmental Psychology, 4(1),

2-13. doi:10.1080/17405620601046832

Gazes, R. P., Hampton, R. R., & Lourenco, S. F. (2015). Transitive inference of social

dominance by human infants. Developmental Science, doi:10.1111/desc.12367

Giedd, J.N., Vaituzis, A.C., Hamburger, S.D., Lange, N., Rajapakse, J.C., Kaysen, D.,

Vauss, Y.C., & Rapoport, J.L. (1996). Quantitative MRI of the temporal lobe,

57

amygdala, and hippocampus in normal human development: Ages 4-18 years.

Journal of Comparative Neurology, 366(2), 223-230.

Griffey, J. A., & Little, A. C. (2014). Infant's visual preferences for facial traits

associated with adult attractiveness judgments: Data from eye-tracking. Infant

Behavior & Development, 37(3), 268-275. doi:10.1016/j.infbeh.2014.03.001

Grossmann, T., Striano, T., & Friederici, A. D. (2007). Developmental changes in infants'

processing of happy and angry facial expressions: A neurobehavioral study. Brain

And Cognition, 64(1), 30-41. doi:10.1016/j.bandc.2006.10.002

Grossmann, T., Johnson, M. H., Vaish, A., Hughes, D. A., Quinque, D., Stoneking, M., &

Friederici, A. D. (2011). Genetic and neural dissociation of individual responses

to emotional expressions in human infants. Developmental Cognitive

Neuroscience, 1(1), 57-66. doi:10.1016/j.dcn.2010.07.001

Hamlin, J. (2013a). Moral judgment and action in preverbal infants and toddlers:

Evidence for an innate moral core. Current Directions in Psychological Science,

22(3), 186-193. doi:10.1177/0963721412470687

Hamlin, J. (2013b). Failed attempts to help and harm: Intention versus outcome in

preverbal infants’ social evaluations. Cognition, 128(3), 451-474.

doi:10.1016/j.cognition.2013.04.004

Hamlin, J., Mahajan, N., Liberman, Z., & Wynn, K. (2013). Not like me = bad: Infants

prefer those who harm dissimilar others. Psychological Science, 24(4), 589-594.

doi:10.1177/0956797612457785

Hamlin, J. K., & Wynn, K. (2011). Young infants prefer prosocial to antisocial others.

Cognitive Development, 26(1), 30-39. doi: 10.1016/j.cogdev.2010.09.001

Hamlin, J., & Wynn, K. (2012). Who knows what's good to eat? Infants fail to match the

food preferences of antisocial others. Cognitive Development, 27(3), 227-239.

doi:10.1016/j.cogdev.2012.05.005

Hamlin, K., Wynn, K., & Bloom, P. (2007). Social evaluation in preverbal infants.

Nature, 450(7169), 557-559. doi: 10.1038/nature06288

Hamlin, J. K., Wynn, K., & Bloom, P. (2010). Three-month-olds show a negativity bias

in their social evaluations. Developmental Science, 13(6), 923-929. doi:

10.1111/j.14677687.2010.00951.x

Hamlin, J., Wynn, K., Bloom, P., & Mahajan, N. (2011). How infants and toddlers react

to antisocial others. PNAS Proceedings of the National Academy of Sciences of

58

the United States of America, 108(50), 19931-19936.

doi:10.1073/pnas.1110306108

Hoss, R. A., & Langlois, J. H. (2003). Infants prefer attractive faces. In O. Pascalis, A.

Slater, O. Pascalis, A. Slater (Eds.) , The development of face processing in

infancy and early childhood: Current perspectives (pp. 27-38). Hauppauge, NY,

US: Nova Science Publishers.

Jessen, S., & Grossmann, T. (2015). Neural signatures of conscious and unconscious

emotional face processing in human infants. Cortex: A Journal Devoted To The

Study Of The Nervous System And Behavior, 64260-270.

doi:10.1016/j.cortex.2014.11.007

Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Ge, L., & Pascalis, O. (2007). The

other-race effect develops during infancy: Evidence of perceptual narrowing.

Psychological Science: A Journal of the American Psychological Society / APS,

18(12), 1084–1089. doi:10.111/j.1467-9280.2007.02029.x

Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Gibson, A., Smith, M., et al. (2005).

Three-month-olds, but not newborns, prefer own-race faces. Developmental

Science, 8(6), F31–6. doi:10.1111/j.1467-7687.2005.0434a.x

Kinzler, K. D., & Spelke, E. S. (2011). Do infants show social preferences for people

differing in race?. Cognition, 119(1), 1-9. doi:10.1016/j.cognition.2010.10.019

Kobiella, A., Grossmann, T., Reid, V. M., & Striano, T. (2008). The discrimination of

angry and fearful facial expressions in 7-month-old infants: An event-related

potential study. Cognition And Emotion, 22(1), 134-146.

doi:10.1080/02699930701394256

Kotsoni, E., de Haan, M., & Johnson, M. H. (2001). Categorical perception of facial

expressions by 7-month-old infants. Perception, 30(9), 1115-1125.

doi:10.1068/p3155

Kuhlmeier, V., Wynn, K., & Bloom, P. (2003). Attribution of dispositional states by 12-

month olds. Psychological Science, 14(5), 402-408. doi: 10.1111/1467-

9280.01454

Langlois, J. H., Ritter, J. M., Roggman, L. A., & Vaughn, L. S. (1991). Facial diversity

and infant preferences for attractive faces. Developmental Psychology, 27(1), 79-

84. doi:10.1037/0012-1649.27.1.79

Langlois, J. H., Roggman, L. A., Casey, R. J., Ritter, J. M., Rieser-Danner, L. A., &

Jenkins, V. Y. (1987). Infant preferences for attractive faces: Rudiments of a

59

stereotype?. Developmental Psychology, 23(3), 363-369. doi:10.1037/0012-

1649.23.3.363

Lyons, A.B., & Cheries, E. (under review). Predicting social behavior by sound and

surface appearance in infancy.

Ma, F., & Xu, F. (2015). Children's and adults’ judgments of facial trustworthiness: The

relationship to facial attractiveness. Perceptual And Motor Skills, 121(1), 179-

198. doi:10.2466/27.22.PMS.121c10x1

Macapagal, K.R., Rupp, H.A., & Heiman, J.R. (2011). Influences of observer sex, facial

masculinity, and gender role identification on first impressions of men’s faces.

Journal of Social, Evolutionary, and Cultural Psychology, 5(1), 92-105.

doi:10.1037/h0099273

Marsh, A. A., & Ambady, N. (2007). The influence of the fear facial expression on

prosocial responding. Cognition and Emotion, 21(2), 225-247.

doi:10.1080/02699930600652234

Marsh, A. A., Ambady, N., & Kleck, R. E. (2005). The effects of fear and anger facial

expressions on approach- and avoidance-related behaviors. Emotion, 5(1), 119-

124. doi:10.1037/1528-3542.5.1.119

Marsh, A. A., Kozak, M. N., & Ambady, N. (2007). Accurate identification of fear facial

expressions predicts prosocial behavior. Emotion, 7(2), 239-251.

doi:10.1037/1528-3542.7.2.239

Mascaro, O., & Csibra, G. (2012). Representation of stable social dominance relations by

human infants. PNAS Proceedings Of The National Academy Of Sciences Of The

United States Of America, 109(18), 6862-6867. doi:10.1073/pnas.1113194109

Mascaro, O., & Csibra, G. (2014). Human infants’ learning of social structures: The case

of dominance hierarchy. Psychological Science, 25(1), 250-255.

doi:10.1177/0956797613500509

Mattes, K., Spezio, M., Kim, H., Todorov, A., Adolphs, R., & Alvarez, R. (2010).

Predicting election outcomes from positive and negative trait assessments of

candidate images. Political Psychology, 31(1), 41-58. doi:10.1111/j.1467-

9221.2009.00745.x

Mondloch, C. J., Geldart, S., Maurer, D., & Le Grand, R. (2003). Developmental changes

in face processing skills. Journal Of Experimental Child Psychology, 86(1), 67-

84. doi:10.1016/S0022-0965(03)00102-4

Nelson, C.A., & Ludemann, P.M. (1986). The discrimination of intensity changes of

emotion by 4 & 7 month old infants. Paper presented at the Midwest

Psychological Association, Chicago, IL.

60

Oosterhof, N. N., & Todorov, A. (2008). The functional basis of face evaluation. PNAS

Proceedings Of The National Academy Of Sciences Of The United States Of

America, 105(32), 11087-11092. doi:10.1073/pnas.0805664105

Oosterhof, N. N., & Todorov, A. (2009). Shared perceptual basis of emotional

expressions and trustworthiness impressions from faces. Emotion, 9(1), 128-133.

doi:10.1037/a0014520

Penton-Voak, I. S., Pound, N., Little, A. C., & Perrett, D. I. (2006). Personality

judgments from natural and composite facial images: More evidence for a 'kernel

of truth' in social perception. Social Cognition, 24(5), 607-640.

doi:10.1521/soco.2006.24.5.607

Pound, N., Penton-Voak, I. S., & Brown, W. M. (2007). Facial symmetry is positively

associated with self-reported extraversion. Personality And Individual

Differences, 43(6), 1572-1582. doi:10.1016/j.paid.2007.04.014

Ramsey, J. L., Langlois, J. H., Hoss, R. A., Rubenstein, A. J., & Griffin, A. (2004).

Origins of a stereotype: Categorization of facial attractiveness by 6-month-old

infants. Developmental Science, 7, 201-211. doi: 10.1111/j.1467-

7687.2004.00339.x.

Rojas Q., M., Masip, D., Todorov, A., & Vitria, J. (2011). Automatic prediction of facial

trait judgments: Appearance vs. structural models. Plos ONE, 6(8),

doi:10.1371/journal.pone.0023323

Rule, N. O., & Ambady, N. (2008). Brief exposures: Male sexual orientation is accurately

perceived at 50ms. Journal of Experimental Social Psychology, 44(4), 1100-1105.

doi:10.1016/j.jesp.2007.12.001

Rule, N. O., & Ambady, N. (2009). She’s got the look: Inferences from female chief

executive officers’ faces predict their success. Sex Roles, 61(9-10), 644-652.

doi:10.1007/s11199-009-9658-9

Rule, N., & Ambady, N. (2010). First impressions of the face: Predicting success. Social

and Personality Psychology Compass, 4(8), 506-516. doi:10.1111/j.1751-

9004.2010.00282.x

Rule, N.O., & Ambady, N. (2011). Judgments of power from college yearbook photos

and later career success. Social Psychological and Personality Science, 2(2), 154-

158. doi: 10.1177/1948550610385473

Rule, N. O., Ambady, N., & Hallett, K. C. (2009). Female sexual orientation is perceived

accurately, rapidly, and automatically from the face and its features. Journal of

61

Experimental Social Psychology, 45(6), 1245-1251.

doi:10.1016/j.jesp.2009.07.010

Rule, N. O., Macrae, C., & Ambady, N. (2009). Ambiguous group membership is

extracted automatically from faces. Psychological Science, 20(4), 441-443.

doi:10.1111/j.1467-9280.2009.02314.x

Said, C. P., Baron, S. G., & Todorov, A. (2009). Nonlinear amygdala response to face

trustworthiness: Contributions of high and low spatial frequency information.

Journal of Cognitive Neuroscience, 21(3), 519-528. doi:10.1162/jocn.2009.21041

Solarz, A. (1960). Latency of instrumental responses as a function of compatibility with

the meaning of eliciting verbal signs. Journal of Experimental Psychology, 59,

239-245.

Spelke, E. S., Breinlinger, K., Macomber, J., & Jacobson, K. (1992). Origins of

knowledge. Psychological Review, 99(4), 605-632. doi:10.1037/0033-

295X.99.4.605

Stewart, L. H., Ajina, S., Getov, S., Bahrami, B., Todorov, A., & Rees, G. (2012).

Unconscious evaluation of faces on social dimensions. Journal of Experimental

Psychology: General, 141(4), 715-727. doi:10.1037/a0027950

Thomsen, L., Frankenhuis, W. E., Ingold-Smith, M., & Carey, S. (2011). Big and mighty:

Preverbal infants mentally represent social dominance. Science, 331(6016), 477-

480. doi:10.1126/science.1199198

Todorov, A. (2008). Evaluating faces on trustworthiness: An extension of systems for

recognition of emotions signaling approach/avoidance behaviors. In A.

Kingstone, M. B. Miller, A. Kingstone, M. B. Miller (Eds.) , The year in cognitive

neuroscience 2008 (pp. 208-224). Malden: Blackwell Publishing.

Todorov, A., Mandisodza, A.N., Goren, A., & Hall, C.C. (2005). Inferences of

competence from faces predict election outcomes. Science, 308, 1623-1626. doi:

10.1126/science.1110589

Todorov, A., Pakrashi, M., & Oosterhof, N. N. (2009). Evaluating faces on

trustworthiness after minimal time exposure. Social Cognition, 27(6), 813-833.

doi:10.1521/soco.2009.27.6.813

Todorov, A., Said, C.P., Engell, A.D., & Oosterhof, N.N. (2008). Understanding

evaluation of faces on social dimensions. Trends in Cognitive Sciences, 12(12).

455-460. doi: 10.1016/j.tics.2008.10.001

62

Willis, J., & Todorov, A. (2006). First impressions: Making up your mind after a 100ms

exposure to a face. Psychological Science, 17(7), 592-598. doi: 10.1111/j.1467

9280.2006.01750.x

Winston, J.S., Strange, B.A., O’Doherty, J., & Dolan, R.J. (2002). Automatic and

intentional brain responses during evaluation of trustworthiness of faces. Nature

Neuroscience, 5(3), 277-283. doi: 10.1038/nn816

Young-Browne, G., Rosenfeld, H. M., & Horowitz, F. D. (1977). Infant discrimination of

facial expressions. Child Development, 48(2), 555-562. doi:10.2307/1128653

Zebrowitz, L. A., Voinescu, L., & Collins, M. A. (1996). 'Wide-eyed' and 'crooked

faced': Determinants of perceived and real honesty across the life span.

Personality And Social Psychology Bulletin, 22(12), 1258-1269.

doi:10.1177/014616729622