University of Southern California Dissertations and Theses

Investigating the effects of Pavlovian cues on social behavior

(USC Thesis Other)

Investigating the effects of Pavlovian cues on social behavior

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INVESTIGATING THE EFFECTS OF PAVLOVIAN CUES ON SOCIAL BEHAVIOR
by
Peter Wang
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
December 2022
Copyright 2022 Peter Wang

ii

Acknowledgments
First, I would like to thank my faculty advisor, Stephen Read. He has given me steady,
tireless support through my years in school, always willing to meet, provide feedback, and
discuss research. His passion for motivation research inspired me to take my research in ways I
never would have, and his openness to letting me explore my own ideas allowed me to foster my
independence as a researcher, giving me skills to think creatively. He was patient as I
experienced setbacks in my personal life, and for that I am grateful as well.
I would also like to thank Daphna Oyserman for her support. She has been a secondary
advisor to me, putting in time and effort to help me grow as a researcher. I am grateful that she
gave me the extra push I needed to refine my thinking and pursue high standards of research and
presentation. Her guidance has been valuable to me as I attempted to improve on my skills in
research. I also want to thank her lab, as they have given me both professional and social support
for much of my time in graduate school. Importantly, I am grateful to my committee members –
John Monterosso, Eva Buechel, and Leor Hackel – for their feedback and patience as I designed
and redesigned my dissertation.

iii

TABLE OF CONTENTS
Acknowledgements..........................................................................................................................ii

List of Tables...................................................................................................................................v

List of Figures.................................................................................................................................vi

Abstract .........................................................................................................................................vii

Chapter 1: Introduction.................................................................................................................... 1
Pavlovian Cues and Behavior.............................................................................................. 3
The Neural Basis of Pavlovian-instrumental Transfer……................................................. 5
Effects of Pavlovian Cues……............................................................................................ 7
Pavlovian Cues and Social Behavior................................................................................... 9

Chapter 2: Using Attractiveness Cues to Increase Willingness to Date........................................ 15
Introduction........................................................................................................................ 15
Study Aim.......................................................................................................................... 20
Study Design...................................................................................................................... 21
Study 1............................................................................................................................... 21
Study 2............................................................................................................................... 27
Study 3............................................................................................................................... 31
Study 4............................................................................................................................... 36
General Discussion............................................................................................................ 39
Conclusion......................................................................................................................... 41

Chapter 3: Situating Pavlovian Processes within a Cultural Framework...................................... 42
Introduction........................................................................................................................ 42
Study Aim.......................................................................................................................... 48
Study Design...................................................................................................................... 48
Study 1............................................................................................................................... 50
Study 2............................................................................................................................... 57
Study 3............................................................................................................................... 63
General Discussion............................................................................................................ 70
Conclusion......................................................................................................................... 72

Chapter 4: Testing the Effects of Approval Cues on Generosity in an Economic Game.............. 73
Introduction........................................................................................................................ 73
Study Aim.......................................................................................................................... 78
Study Design...................................................................................................................... 78
Methods..............................................................................................................................79
Results................................................................................................................................ 83
Discussion.......................................................................................................................... 84
Conclusion......................................................................................................................... 87

Concluding Remarks...................................................................................................................... 89

iv

References...................................................................................................................................... 94

Appendices................................................................................................................................... 106
Appendix A: Chapter 1 Tables and Figures..................................................................... 106
Appendix B: Chapter 2 Tables and Figures..................................................................... 108
Appendix C: Chapter 3 Tables and Figures..................................................................... 118
Appendix D: Chapter 4 Tables and Figures..................................................................... 127
Appendix E: Concluding Remarks Tables and Figures................................................... 131

v

List of Tables
Appendix B.………………………………………………………...………………………….. 110
Table 1 Generalized Linear Multilevel Model Predicting
Willingness to Date (Study 1)..………………………………………………… 110
Table 2 Generalized Linear Multilevel Model Predicting
Willingness to Date, Including Interactions with
Relationship Variables (Study 1)………………………………………………. 111
Table 3 Generalized Linear Multilevel Model Predicting
Willingness to Date (Study 2) ….……………………………………………… 112
Table 4 Generalized Linear Multilevel Model Predicting
Willingness to Date, Including Interactions with
Relationship Variables (Study 2)………………………………………………. 113
Table 5 Generalized Linear Multilevel Model Predicting
Willingness to Date (Study 3) ….……………………………………………… 114
Table 6 Generalized Linear Multilevel Model Predicting
Willingness to Date, Including Interaction with
Relationship Status (Study 3)………………...…………………………………115
Table 7 Generalized Linear Multilevel Model Predicting
Willingness to Date (Study 4) ….……………………………………………… 116
Table 8 Generalized Linear Multilevel Model Predicting
Willingness to Date, Including Interaction with
Relationship Status (Study 4)…………………………………………………...117

Appendix C.……………………………………………………………………………………. 123
Table 1 Generalized Multilevel Poisson Model Predicting
Total Keypresses with Zero-Inflation Model (Study 1)………………………... 123
Table 2 Generalized Multilevel Poisson Model Predicting
Total Keypresses with Zero-Inflation Model (Study 2)………………………... 124
Table 3 Generalized Multilevel Poisson Model Predicting
Total Keypresses with Zero-Inflation Model (Study 3)………………………... 125
Table 4 Generalized Multilevel Poisson Model with
Zero-Inflation Model Predicting Total Keypresses from
the Interaction of Honor Priming and Threat Cue (Study 3)…………………... 126

Appendix D.……………………………………………………………………………………. 129
Table 1 Generalized Linear Multilevel Model Predicting Generosity………………….129
Table 2 Generalized Linear Multilevel Model Predicting Generosity
with Personality Variables Added….………………………………………….. 130

Appendix E.……………………………………………………………………………………. 131
Table 1 Summary of Study Design Differences Relevant to
the Presence of Supporting Evidence……………………..…………………… 131

vi

List of Figures
Appendix A.………………………………………………………...………………………….. 106
Figure 1 Model of Specific PIT………………………………………………………... 106
Figure 2 Model of General PIT……………………………………………………….... 107

Appendix B.………………………………………………………...………………………….. 108
Figure 1 Message Border Cues 1………………………………………………………. 108
Figure 2 Interaction of Cue and Relationship Status (Study 1)………………………... 109

Appendix C.………………………………………………………...………………………….. 118
Figure 1 Sample Instrumental Conditioning Trial……………………………………... 118
Figure 2 Sample PIT Trial………………………………………………………...…… 119
Figure 3 Interaction of Honor Values with Threat Cue in
Predicting Total Keypresses (Study 1) ………………………………………... 120
Figure 4 Interaction of Honor Values with Threat Cue in
Predicting Total Keypresses (Study 2) ………………………………………... 121
Figure 5 Interaction of Honor Priming with Threat Cue in
Predicting Total Keypresses (Study 3) ………………………………………... 122

Appendix D.………………………………………………………...………………………….. 127
Figure 1 Sample Dictator Game Trial………………………………………………….. 127
Figure 2 Predicted Probabilities of Sharing with Interaction
of Cue Condition with Conscientiousness as Predictors………………………..128

vii

Abstract
Although psychologists have extensively studied the role of contextual cues in
influencing social behavior, the role of Pavlovian cues, or cues associated with reward or
punishment, is unknown. This constitutes a major gap in our understanding of both reward
learning and social behavior. In this paper, I first review research on the effects of Pavlovian
cues on general behavior, with a focus on findings that could potentially link Pavlovian cues to
social behavior. I then present findings from three separate research projects testing the role of
Pavlovian cues in social behavior. The first investigated whether cues associated with
attractiveness can influence dating choices. The second investigated whether honor cultural
values sensitize participants to Pavlovian cues associated with interpersonal threat. The third
investigated whether cues associated with social approval can influence altruism. While our
findings overall are mixed, each project is still an ongoing endeavor, and one of them showed
promising evidence that cultural factors can determine how people respond to Pavlovian cues.

1

Chapter 1: Introduction
Portions of this chapter are from:
Wang, P., & Read, S. J. (2022). Pavlovian cues boost self-regulation: Predicting success with
conditioned stimuli. Manuscript in preparation.

We rely on information from the environment to make decisions. Contextual cues
provide crucial information about when and where to get what we want. The smell of pastries
alerts us to treats nearby. The sight of smoke in the air shows us that danger may be approaching.
The lively jingle of a slot machine tells us this could be our chance to win the jackpot. These
sensory stimuli are called Pavlovian cues, because people learn through experience to associate
them with reward or punishment. By helping people anticipate the presence of a rewarding or
aversive outcome, Pavlovian cues help to guide behavior towards goal attainment (Cartoni et al.,
2016).
Pavlovian cues can be a powerful influence on behavior. Research has linked Pavlovian
conditioning to behaviors and disorders, including depression (Nord et al., 2018), drug use
(Hogarth et al., 2015), and overeating (Johnson, 2013). Yet, major gaps remain in our
understanding of Pavlovian effects on human behavior. First, we do not know whether Pavlovian
cues can influence social behaviors. Social behaviors often require a complex decision-making
process that involves operations such as perspective-taking and impression management. Social
behaviors may include substantially more complex decision-making than previously examined
in-lab behaviors, which often involve pressing buttons repeatedly to obtain reward. Second, we
do not know how broader cultural attitudes might contextualize the effects of Pavlovian cues on
social behaviors. People may experience social rewards and punishments differently depending
on their culturally determined understanding of them (Oishi & Diener, 2009). Given this

2

perspective, one’s cultural background and attitudes may guide the effect of a Pavlovian cue to
the extent that they determine the appeal of the outcome associated with the cue.
In this dissertation, I attempt to address these gaps by discussing and empirically testing
the role of Pavlovian cues in three types of social behaviors: generosity, dating, and defensive
aggression. My aim is to gain a richer understanding of the role that Pavlovian cues play in the
world. Do Pavlovian cues have broader implications for social interaction and other social
phenomenon, given that social behaviors are ubiquitous in everyday life? At the same time, I
hope to provide a theoretical foundation for understanding how people make social decisions.
Theories about goal priming and situated cognition in social psychology emphasize the effects of
context on behavior (Aarts et al., 2008; Smith & Semin, 2004). Yet, the nature of the contextual
elements that change behavior are not always specified. Research on Pavlovian cues may help to
bridge the gap in these theories by specifying a mechanism by which context affects behavior –
that Pavlovian cues help people to anticipate social rewards or threats and guide their actions to
obtain or avoid these outcomes.
I begin in this chapter by describing the nature of Pavlovian cues and their effects on
behavior in humans. I detail common paradigms for studying Pavlovian cues as well as previous
findings that relate Pavlovian cues to a variety of consequential behaviors. I also present the
justification for linking social behavior to Pavlovian cues. I discuss the possibility of using
Pavlovian cues to influence dating choices and generosity in Chapters 2 and 4, respectively, and
present relevant studies that I have conducted. In Chapter 3, I examine the role of Pavlovian cues
in defensive aggression, and how culturally-determined values, such as those endorsed in honor
cultures, might amplify the effects of cues on defensive aggression. Finally, I summarize my
findings from the three chapters and discuss directions for future research.

3

Note that, for the purposes of this paper, I focus on human behaviors. While the effects of
Pavlovian cues on behavior in other animals have been studied extensively in a variety of
domains (Adkins-Regan & MacKillop, 2003; Corbit & Balleine, 2005; Hollis, 1984; Robinson &
Berridge, 2013), the primary interest of my paper is to understand the effects of Pavlovian cues
on social behavior in humans.
Pavlovian Cues and Behavior
Pavlovian cues form when presentation of a stimulus (neutral stimulus; NS) is repeatedly
paired with subsequent presentation of reward or punishment (unconditioned stimulus; US), in a
process called Pavlovian conditioning. For example, a red circle (NS) might be shown every time
cookies (US) are given to a participant. After conditioning, subsequent presentations of the
Pavlovian cue (conditioned stimulus; CS) lead the participant to anticipate the US. Hence, the
red circle becomes a CS once it begins to lead the participant to anticipate cookies whenever it is
shown.
Prior work has demonstrated that Pavlovian cues can affect subsequent behavior in a
phenomenon known as Pavlovian-instrumental transfer (PIT). These studies use Pavlovian
conditioning to pair cues with different outcomes, and in separate training sessions, they use
instrumental conditioning to pair different actions with those same outcomes. After instrumental
training, the study moves onto the PIT test phase. During this phase, a CS is presented, and
participants are given the opportunity to respond with either of the instrumental actions. This
happens in nominal extinction – that is, participants are not given any feedback about what
outcomes occurred until the end of the task. During the test phase, participants are more likely to
take the action that corresponds to the outcome that was conditioned to the CS. Hence, after a
participant learns that they can press a left button to obtain cookies and a right button to obtain

4

chocolate, subsequent presentations of the red circle that predicts a cookie will lead them to press
the left button, since the red circle is associated with cookies. This occurs despite the fact that the
CS and the instrumental action were never paired together before the test phase, suggesting that
the PIT effect does not result from a direct association between the CS and the action.
The literature has identified two forms of PIT (Corbit and Balleine, 2005; Talmi et al.,
2008). In one form, specific PIT, the CS motivates an action when the outcome associated with
the action matches the outcome associated with the CS. Thus, Watson et al. (2014) found that
cues associated with chocolate motivated participants to press a button to obtain chocolate, and
cues for popcorn motivated participants to press a button for popcorn. However, in what is called
general PIT, as long as the CS is associated with a reward or punishment, it can motivate an
action whose outcome is not associated with the CS. When participants are exposed to the CS,
they will still make available instrumental actions more vigorously than if a cue associated with
no outcome were presented. Thus, Watson et al. (2014) found that a cue associated with cashews
motivated greater instrumental responding in general (i.e. button-pressing for chocolates and for
popcorn) than a cue associated with no outcome. This occurs only when the outcome associated
with the CS has no corresponding instrumental response; in this case, there was no button to
press to obtain cashews. To explain these phenomena, researchers have theorized that cues in
specific PIT activate a specific representation of the outcome, while cues in general PIT activate
a generalized reward signal (Cartoni et al., 2016). Hence, specific PIT leads people to anticipate
a particular reward, while general PIT leads people to anticipate rewards in general. For
visualizations of the specific PIT and general PIT procedures, see Appendix A, Figures 1 and 2.
To explain how Pavlovian cues influence behavior, researchers have proposed a
hierarchical structure for the elements involved in PIT, in which the CS provides information

5

about the strength of the response-outcome relationship (Rescorla, 1991; Hogarth et al., 2014). In
this account, Pavlovian cues signal the efficacy of instrumental actions by indicating the
likelihood that the instrumental action will produce an outcome (Hogarth et al., 2014). The
presence of a rewarded Pavlovian cue signals that an instrumental action will be rewarded.
Consistent with the hierarchical account, studies have demonstrated that PIT effects correlate
with the explicit belief that the Pavlovian cue signals the likelihood that a response would be
reinforced (Mahlberg et al., 2019; Seabrooke et al., 2016). Experimental instructions that
contradict this belief reduced or eliminated PIT effects (Hogarth et al., 2014; Seabrooke et al.,
2016). Hence, Pavlovian cues seem to signal the efficacy of an action – the probability of success
(Cartoni et al., 2015).
The Neural Basis of Pavlovian-instrumental Transfer
Neurobiologically, Pavlovian cues trigger the release of dopamine, a neurotransmitter
associated with reward anticipation and reward-seeking (Berridge, 2007; Salamone et al., 2007;
Schultz, 2022). Salamone et al. (2007) found in rats that dopamine is associated with exertion of
effort in instrumental responding. Dopamine depletion in rats impedes performance of high-
effort instrumental actions, but not low-effort actions. Hence, dopamine may drive PIT effects by
driving the exertion of effort in instrumental responding. Dopamine is also associated with
reward prediction error; when a received reward has greater value than one expects, there is a
positive spike in dopamine signal (Schultz, 2022). When it has lower value than one expects, or
when one expects reward and receives none, there is a negative dip in dopamine signal (Schultz,
2022). As a signal of reward prediction error, dopamine helps people adjust their expectations
about the world, possibly making it an important component of Pavlovian and instrumental
learning (Schultz, 2022). Berridge (2007) argues that dopamine drives incentive salience, a

6

phenomenon in which dopamine signals attribute motivational significance to reward-related
stimuli, causing the person to want them. In other words, dopamine may motivate the pursuit,
acquisition, and/or consumption of a stimulus by creating an attraction to the stimulus. In this
view, Pavlovian cues, mediated by dopamine signals, can trigger wanting for a reward (Berridge,
2007). One, some, or more likely all of these dopamine functions may be involved in Pavlovian-
instrumental transfer. Indeed, dopamine may serve different functions depending in part on the
timescale of the dopamine signal (Schultz, 2017).
Lesion studies using rats suggest that the nucleus accumbens and amygdala are necessary
for Pavlovian-instrumental transfer to occur (Corbit & Balleine, 2005). The nucleus accumbens
is involved in motivational phenomena such as learning, reward-seeking, avoidance, and
impulsivity, and dopamine released in the nucleus accumbens mediates reward processing
(Salgado & Kaplitt, 2015). The amygdala is also involved in motivation: learning, value
representation, and stimulus salience (Janak & Tye, 2015). Corbit and Balleine (2005) showed
that, within the amygdala, the basolateral complex and central nucleus may be involved in PIT.
Lesions to the basolateral complex eliminate specific PIT effects – rats no longer showed transfer
when the cued outcome matched an outcome of an instrumental action. Lesions to the central
nucleus eliminate general PIT effects – rats only showed transfer when the cued outcome
matched an instrumental outcome. Two regions of the nucleus accumbens show a similar
dissociation when lesioned; lesions to the nucleus accumbens shell eliminate specific PIT effects,
while lesions to the nucleus accumbens core eliminate general PIT effects (Corbit & Balleine,
2011). These findings suggest that two neural pathways are crucial for PIT effects to occur: one
through the basolateral complex and nucleus accumbens shell that maintains information specific
to the outcome associated with the Pavlovian cue, and one through the central nucleus and

7

nucleus accumbens core that is involved in the general motivating effects of the Pavlovian cue
(Corbit & Balleine, 2005, 2011).
Neuroimaging studies using fMRI with humans support these findings. Talmi et al.
(2008), using a PIT paradigm with monetary rewards, found that both nucleus accumbens and
amygdala activity correlate with the strength of PIT effects. Examining the amygdala on a finer
level, Prévost et al. (2012), in a study using food rewards, found that basolateral complex activity
predicts the strength of specific PIT, while central nucleus activity predicts the strength of
general PIT. These findings are congruent with rat lesion studies, supporting the presence of
outcome-specific and general motivational pathways in the brain.
Effects of Pavlovian Cues
Prior research has documented the role of Pavlovian cues in driving behavior in various
domains. While PIT is generally useful for allowing people to track and make use of resources in
their environments, it can also promote dysfunctional reward-seeking. Much of PIT research has
focused on how Pavlovian cues can trigger cravings for powerful, primary reinforcers,
suggesting that these cues might help sustain addiction and overeating (Hogarth et al., 2015;
Johnson, 2013; Watson et al., 2014). In these instances of maladaptive behaviors, people become
increasingly sensitive to Pavlovian cues and respond with increasing impulsivity due to
excessive release of dopamine (Robinson & Berridge, 2001). The excess of dopamine often
results from drugs that alter normal dopaminergic functioning (Robinson & Berridge, 2001) or
possibly chronic engagement in activities that maximize dopamine release (Zack et al., 2020).
Eating Behaviors
Johnson (2013) suggests that PIT may play a role in overeating. He theorizes that
Pavlovian cues associated with food may activate reward signals in the brain that drive eating

8

behavior, downregulating homeostatic satiety signals that inhibit food-seeking. Consistent with
this account, preschool children find foods to be tastier when they are arbitrarily labeled with
McDonald’s packaging (Robinson et al., 2007), indicating that Pavlovian cues may activate
hedonic signals in neural reward circuitry. Food-related Pavlovian cues may lead to food
consumption even beyond the point of satiation. Watson et al. (2014) found that food-related
Pavlovian cues increase food-seeking behavior, even when participants are satiated. People
exposed to food-related Pavlovian cues may be more likely to engage in unnecessary eating.
Drug and Alcohol Use
Prior research has shown that Pavlovian cues can promote drug-seeking behavior.
Manglani et al. (2017) trained participants to associated colored squares with cigarettes, food, a
neutral outcome, or no outcome, finding that cues associated with cigarettes led participants to
increase responding to earn cigarette puffs. The researchers raise concerns that tobacco-related
cues may interfere with intentions to quit smoking. Hogarth et al. (2015) found that images of
branded cigarette packs increased tobacco-seeking response over plain cigarette packs or no
image, possibly because branded cigarette packs are perceived to be more predictive of tobacco
reward. Similarly, Garbusow et al. (2019), in a paradigm using cues and actions associated with
money, found that PIT effects on money-seeking behavior was stronger among high-risk alcohol
drinkers, indicating that PIT strength may be a vulnerability marker for alcohol use disorder.
Moreover, the association between PIT and risky drinking suggests that PIT may be an
underlying mechanism that promotes alcohol consumption. For both tobacco and alcohol,
Pavlovian-to-instrumental transfer may play a role in sustaining drug use by linking
environmental cues to drug-seeking behavior.
Gambling

9

Zack et al. (2020) note that Pavlovian cues play an important role in gambling and
gambling disorder by signaling the presence of reward and triggering dopamine release. They
theorize that gambling games manipulate the uncertainty of reward delivery and reward
magnitude in order to maximize dopamine release, heightening the potential for addiction.
Cherkasova et al. (2018) also suggest that Pavlovian cues influence gambling behavior, finding
that the addition of an audiovisual cue associated with monetary wins leads participants to make
riskier choices on a lottery task.
Cognitive Effort
Wang & Read (2022) found that Pavlovian cues can also influence cognitive effort on
achievement tasks. In a test of general PIT, they found that cues associated with money lead
participants to persist longer on difficult anagram puzzles. Moreover, they found in a separate
experiment that monetary cues can increase participants’ expectations of success on anagram
tasks, suggesting that the effect of the cues on cognitive effort was driven by reward expectancy.
The findings also have implications for Pavlovian cue effects on self-regulation, since they
showed that cues can drive persistence in the face of difficulty.
Pavlovian Cues and Social Behavior
Commonalities Between Social and Non-social Reward
Studies on Pavlovian learning have focused on the pursuit of powerful, relatively
unambiguous rewards, such as food, drugs, and money. On the other hand, the effects of
Pavlovian cues on social behavior is unclear. There is reason to believe that Pavlovian cues can
influence social behavior; histories of Pavlovian learning may account for some of the effects of
situations on social approach and avoidance (Read et al., 2018). For example, one who
previously learned to associate the dance floor with social bonding might accompany some

10

friends to dance, while one who learned to associate the dance floor with social rejection might
avoid the party altogether (Read et al., 2018).
Furthermore, social approval is associated with the activation of the striatum in the brain,
which is associated with reward processing and dopaminergic function (Izuma et al., 2008).
Kohls et al. (2013) found similar results, with both social reward and social punishment
associated with activation in the ventral striatum. The same region includes the nucleus
accumbens, which is involved in Pavlovian-instrumental transfer using other rewards (Corbit &
Balleine, 2005; Talmi et al., 2008), suggesting that social reward may share a common reward
signal with other types of rewards. Indeed, Izuma et al. (2008) found that the reward-related
brain regions associated with social approval were also activated when participants were given
money. Delgado et al. (2005) also found activation of the striatum when participants were
learning and making decisions about the trustworthiness of their partners in a trust game,
indicating that reward-related regions may be involved when people learn and anticipate social
feedback.
Moreover, Pavlovian cues associated with social reward can have general PIT effects on
non-social behavior similar to cue effects associated with other types of reward. Lehner et al.
(2017) trained participants to associate four different cues with money, food, social reward (i.e. a
thumbs-up gesture), and no outcome. For each participant, the rewards were matched on
subjective value. When participants were given the opportunity to pursue monetary or food
rewards, the money, food, and social cues all motivated overall reward-seeking to the same
extent. While the researchers did not examine pursuit of social reward, the finding that a social
reward cue could induce PIT increases our confidence that comparable PIT mechanisms explain
social and non-social behavior.

11

Similarly, research on anxiety have shown effects of Pavlovian cues associated with
aversive social outcomes on subjective experience and autonomic response. Wiggert et al. (2017)
trained participants to associated different faces with social approval, social disapproval, or
neutral evaluations. Participants rated disapproval cues as more arousing than neutral cues even 1
year after conditioning, and the maintenance of this response was positively correlated with
depressive symptoms and trait anxiety. Ahrens et al. (2016) trained participants to associate two
faces with either a loud scream and fearful face, or no outcome. When exposed to the aversive
cue, participants exhibited a heightened skin conductance response. While these studies did not
examine the effects of aversive Pavlovian cues on social behavior, they do suggest that PIT can
operate within social contexts.
Gęsiarz and Crockett (2015) have proposed a Pavlovian basis for understanding prosocial
behavior. They suggest that Pavlovian cues signaling need, such as sad faces, can induce others
to help. Seymour and Dolan (2008) have also proposed the role of Pavlovian mechanisms in
prosocial behavior, suggesting that cues that signal potential reciprocity (and therefore mutual
benefit) may increase cooperation. To our knowledge, these proposals remain untested, but they
raise promising directions for future investigation.
There is also some evidence that Pavlovian cues may trigger defensive behaviors in
interpersonal aggression. Nadler et al. (2011) recruited participants to play a hypothetical
military defense game, in which participants were tasked with defending a country from enemy
invasion. They trained participants to associate different cues with different military threats or no
threat, and taught participants keypresses that could be used to defend against different threats.
Consistent with previous PIT findings, participants responded more strongly to threat-related
cues than to neutral cues. While the authors did not discuss the study as a test involving social

12

behavior or measure social aspects of the task, one can imagine that defense against military
invasion involves a variety of moral and sociocultural interpretations. We will examine this
question further in Chapter 3.
Differences Between Social and Non-social Reward
Nevertheless, there are differences between social and non-social reward that might
suggest that Pavlovian cues play a much smaller role in social phenomena than phenomena
previously examined in PIT. While social and non-social valuation may involve the same brain
regions (Izuma et al., 2008), it is unclear whether they activate the same neurons within those
regions (Ruff & Fehr, 2014). Klein and Platt (2013) recorded the activity of individual neurons
within the medial striatum of macaques. They found that information about juice reward and
information about social reward (i.e. rewarding images of other macaques) activate separate
groups of neurons, suggesting that social and non-social information may be processed in
distinct, separate pathways within the striatum. This raises the possibility that Pavlovian cues
may not have the same effects in social contexts that they do in non-social contexts.
Social reward may also involve additional mechanisms that are not involved when
working with non-social reward. Krach et al. (2010) argue that Theory of Mind may be
necessary to experience social reward. Theory of Mind is the ability to attribute mental states to
others, allowing one to understand the intentions, needs, and experiences of other people
(Adolphs, 2003). Krach et al. (2010) suggest that, because an understanding of others’ intentions
and affective states is necessary to enjoy a positive social interaction, Theory of Mind may be an
important mediator for the experience of social rewards. They also argue that the interpretation
of social reward is highly context-dependent, suggesting that the experience of social rewards
may be more complex than that of non-social rewards. Group identity is another relevant factor.

13

People who care strongly about group identity value benefits to ingroup members more than to
outgroup members, and this preference is reflected in activation of the reward-related ventral
striatum (Hackel, Zaki, & Van Bavel, 2017). With the addition of complex cognitive and
situational components, attempts to use Pavlovian cues to influence the pursuit of social rewards
might “wash out” in the sea of other factors.
Why Social Behavior?
Still, understanding the effects of Pavlovian cues in social contexts can provide important
insights for both Pavlovian mechanisms and social phenomena. The complexity of social reward
means that we can study the interplay of Pavlovian mechanisms and social factors. The takeaway
would not just be that Pavlovian cues can influence social behavior, it would be how Pavlovian
cues work with the rich array of beliefs, attitudes, perceptions, and other elements that shape
social action. The ambiguity of social reward could be a boon to gaining a more complete
understanding of how Pavlovian cues operate. For example, researchers have manipulated the
reward values of outcomes associated with Pavlovian cues (e.g., through satiation or deprivation)
in experiments to induce revaluation, in which animals or participants change their behavior in
response to a Pavlovian cue because of the change in reward value (Dayan & Berridge, 2014). In
social contexts, we could attempt a similar effect by other means, such as by revealing the
ingroup status of a partner to increase the value of their approval, and then measuring responses
to cues associated with social approval from that partner. Studies such as these could provide
methodological diversity to support existing theories, and perhaps even provide new insights.
Studies linking Pavlovian learning to social behavior can offer new ideas for our
understanding of social phenomena as well. Much of the social psychological literature focuses
on the effects of situation on behavior (Aarts et al., 2008; Adams & Marshall, 1996; Mischel et

14

al., 2002; Smith & Semin, 2004; Wilson, 2002). Theories of situated cognition emphasize that
explanations for behavior must account for elements of the social and physical situation (Smith
& Semin, 2004). Theories about goal priming emphasize that aspects of one’s environment
associated with a goal can trigger goal pursuit (Aarts et al., 2008; Papies, 2016). In some cases,
these situational cues may be semantic ones, containing inherent meaning related to the goal,
such as the recipe flyer that reduces snack-buying by mentioning health and diet (Papies et al.,
2014). But there undoubtedly are also cases in which motivational cues develop from a history of
association with reward or punishment, such as the ringtone that alerts you to a message from
your friends. Research on Pavlovian learning can provide a more detailed account of the
situational cues that affect social behavior, complementing existing theories by specifying
mechanisms for the development, maintenance, and operation of these cues.

15

Chapter 2: Using Attractiveness Cues to Increase Willingness to Date
Introduction
People often form judgments based on appearances. We find faces of a significant other
to be rewarding (Scheele et al., 2013) and infant faces to be cute (Glocker et al., 2009). We
extend our judgments to other faces on the basis of facial similarity, in what Hugenberg and
Wilson (2013) have described as overgeneralization. Thus, adult faces with baby-like facial
features, such as large eyes and high eyebrows, are judged to have more baby-like traits
(Zebrowitz & Montepare, 1992), and they elicit spontaneous warmth responses (Hugenberg &
Wilson, 2013). Moreover, faces that structurally resemble happy expressions are judged to be
more trustworthy than faces that resemble angry expressions (Oosterhof & Todorov, 2009).
The effects of appearance are especially salient in the realm of mate selection, based on
theories proposed by evolutionary psychologists (Rhodes et al., 2005; Thornhill & Gangestad,
1999). Given the role of physical attractiveness in finding mates, we suspected that Pavlovian
cues associated with attractiveness can influence dating choices. In this chapter, I first discuss the
theorized evolutionary functions of physical attractiveness, focusing particularly on components
of facial attractiveness, and their effects on romantic attraction. I then explain our reasoning for
studying the effects of attractiveness cues before presenting a series of four studies examining
our research question.
Romantic Attraction
While different researchers may define it differently, romantic love or attraction generally
involves emotional attachment and passionate affection towards a person (Fischer et al., 2002;
Jankowiak & Fischer, 1992). Over time, people in romantic relationships often show strong
commitment to be with one another and act to maintain their relationships and the well-being of

16

their partners (Joel et al., 2017). Romantic attraction is a near-universal phenomenon; out of 166
cultures, Jankowiak and Fischer (1992) found evidence of romantic love in 147 of them, based
on folklore and observations by anthropologists. From an evolutionary standpoint, romantic
attraction may have evolved in the past to optimize reproductive success by focusing mating-
related investment of time and resources on a particular individual or individuals (Fisher et al.,
2002).
Although men seem to place more emphasis on physical attractiveness than women,
attractiveness is an important factor for both sexes in terms of romantic attraction (Feingold,
1990). In a study on first dates, Walster et al. (1966) found that physical attractiveness was the
largest determinant of willingness to continue dating, even when accounting for social skills,
introversion, and other traits. Similarly, Luo and Zhang (2009) found in a speed-dating study that
physical attractiveness most strongly determined interest in a potential partner, compared to three
other factors (similarity in attitudes and personality, reciprocity of liking, and general positive
traits that predict relationship quality). Replicating these findings, Ha et al. (2010) found that
heterosexual adolescents showed more desire to date an attractive person than a less attractive
person in a study examining hypothetical dating choices. This relationship was moderated by
self-perceived mate value, such that adolescents who saw themselves as having higher value
were especially interested in dating attractive people. This is consistent with the matching
hypothesis, in which people are more likely to form a relationship with someone of similar
attractiveness (Berscheid et al., 1971). However, Ha et al. (2010) noted that the interaction was
weak, particularly in comparison to the much larger main effect of attractiveness on willingness
to date.
Physical Attractiveness and Its Function

17

Researchers propose that people tend to favor physical attractiveness when choosing
mates because physical attractiveness indicated the presence of disease and general fitness for
reproduction in ancestral environments (Thornhill & Gangestad, 1999; Weeden & Sabini, 2005).
While the link between attractiveness and health may be less relevant in industrialized societies
with access to modern medicine (Kalick et al., 1998), attractiveness remains a important factor in
romantic attraction (Luo & Zhang, 2009). To explain the functions of physical attractiveness, we
discuss two components of facial attractiveness and what information they may convey.
Facial symmetry contributes to perceptions of attractiveness. Rhodes et al. (1998) edited
photos of faces to create different levels of symmetry. They found that participants preferred
high-symmetry over low-symmetry faces, reflected in ratings of attractiveness and mate appeal.
Likewise, Perrett et al. (1999) also found symmetrical faces to be more attractive.
As with other components of attractiveness, facial symmetry also conveys information
about physical health. Symmetrical faces may have been markers of successful immune systems
in early environments, indicating the absence of pathogens and diseases, as well as greater
reproductive success (Thornhill & Gangestad, 1999). Consistent with this idea, Waynforth
(1998) found that facial symmetry in men from two villages in Belize was associated with fewer
major illnesses and more offspring. Moreover, college students in Austria rated symmetrical
faces more highly on a variety of positive attributes, including attractiveness and physical health
(Fink et al., 2006). These findings suggest that facial symmetry may be an important component
of attractiveness because of it can provide information about immune function and reproductive
fitness.
Facial averageness is another determinant of attractiveness. Langlois and Roggman
(1990) created images of average faces by merging multiple faces of the same sex. Working with

18

United States participants, they found that the composite faces were generally rated as more
attractive than the individual faces that make up the composites, and that attractiveness ratings
increased as the number of faces used to generate the composites increased. Rhodes, Yoshikawa
et al. (2001) replicated these findings among Chinese and Japanese participants. People seem to
judge average faces as more attractive.
Researchers have suggested that facial averageness can convey information about genetic
fitness (Thornhill & Gangestad, 1993). Facial averageness may reflect genetic variability, which
strengthens resistance against pathogens (Thornhill & Gangestad, 1993). Genetic variability
makes it difficult for pathogens to adapt to the host’s environment, thereby improving fitness
(Tooby & Cosmides, 1990). People may find average faces more attractive, because average
faces reflect greater genetic diversity and therefore genetic fitness (Thornhill & Gangestad,
1993). Consistent with this account, Rhodes, Zebrowitz et al. (2001) found that people perceive
average faces as healthier, and people who are rated as having more distinct faces have poorer
health outcomes.
Thus, physical attractiveness, particularly facial attractiveness, may signal abilities to
resist disease and pass on genes. The role of attractiveness in romantic attraction may be an
evolutionary adaptation that originally paired people with genetically healthy partners.
Attractiveness and Pavlovian Conditioning
Given the importance of attractiveness in dating, we decided to use cues associated with
attractiveness to influence dating decisions. There is some evidence that contextual cues can
influence attraction. Watkins et al. (2012) found that priming concerns about pathogens
increased female participants’ self-reported attraction to men with masculine appearances.
Carpentier et al. (2014) primed sexual content through webpage banner ads and then asked

19

participants to judge a dating profile. Participants who were primed with sexual content judged
the target in the profile as more sexually attractive, and attractiveness ratings in turn predicted
overall appeal of the target. In another study, Carpentier et al. (2007) found that priming
participants with sexually provocative music led them to weigh sex appeal more heavily in their
overall evaluations of dating profiles. Additionally, among heterosexual males who believed that
alcohol consumption increases sexual desire, priming with alcohol-related words led them to rate
young women in photographs as more attractive (Friedman et al., 2005). The same manipulation
did not lead them to rate the women as more or less intelligent. While researchers in all of these
studies described their priming effects in terms of semantic associations, their findings suggest
that perceptions of attractiveness and attraction in general are malleable to contextual factors,
increasing our confidence that Pavlovian cues may have an effect on dating choices.
Indeed, one can imagine sensory cues in the real world that can influence attraction.
Clothing brands, accessories, luxury goods, and other elements can affect perceptions of
attractiveness, not just due to aesthetic appeal but perhaps also due to their associations with
attractive individuals seen as seen in mass media. Consistent with this account, Weismueller et
al. (2020) found that the attractiveness of a social media influencer significantly increases
consumer purchase intentions for products that they endorse. Hence, it is possible that Pavlovian
cues associated with physical attractiveness already exist in the world and exert an effect on
dating, social media use, and other social behaviors.
Moreover, prior research has shown that attractiveness has reward value. Hahn et al.
(2016) found that participants will exert greater effort on a keypress task to view more attractive
faces of their preferred sex. This indicates that attractiveness can act as a reward, reinforcing
instrumental actions. Cloutier et al. (2008) found activation of the nucleus accumbens, a reward-

20

processing region of the brain, when participants viewed attractive faces. Liang et al. (2010)
found a similar relationship between attractive faces and activation in the nucleus accumbens.
The activation of the nucleus accumbens also correlates with the strength of Pavlovian-
instrumental transfer (Talmi et al., 2008). Together, the evidence suggests that attractiveness can
act as a form of reward. Assuming that attractiveness is rewarding, we saw good reason to
suspect that Pavlovian cues can form predicting attractiveness, and that such cues can affect
romantic attraction.
Study Aim
The primary aim of our studies was to test whether Pavlovian cues associated with
attractiveness could influence dating choices. Such cues would lead participants to anticipate
attractiveness as reward, and thereby lead them to be more willing to go on a date. The studies
would have the potential to explain, at least in part, real-world phenomena, including why
celebrity and influencer endorsements lead people to buy products. They would open new
investigations into the nature of social behavior; the studies would suggest that Pavlovian cues
are not just context cues for guiding behavior, but also tools that people actively use to influence
others in social interactions. This is particularly relevant to social media use, given the controlled
nature of self-presentation in social media posting.
Additionally, we asked participants to report current relationship status, relationship
status satisfaction, and dating priorities. We anticipated that the effects of attractiveness cues
would be stronger when participants were: single, dissatisfied with their current relationship
status, and prioritizing dating other people. This investigation would show that existing motives
can sensitize people to the effects of attractiveness cues.

21

Study Design
Our studies consisted of two main parts. In the first phase, we trained participants to
associate different cues with either more attractive or less attractive faces. In the second phase,
we asked participants to complete a hypothetical dating task, in which participants are shown
photographs of different people and asked whether they would be willing to date each person.
We tested whether presentation of the conditioned cues during the dating task could affect how
willing participants were to date someone. We predicted that presentation of an attractiveness
cue during the task would lead participants to be more willing to date.
While our research question was inspired by Pavlovian-instrumental transfer, we did not
actually include an instrumental conditioning task in our study. Instead, we taught participants
how to complete the dating task by instruction and gave them practice trials to allow them to
acquaint themselves with it. Although this design did not follow existing PIT paradigms that ask
participants to press buttons to effect an outcome (Hogarth et al., 2014; Prévost et al, 2012;
Watson et al., 2014), it does mimic to some extent a real-world dating app. The binary-choice
task, in which participants are shown a photo and press one button to indicate unwillingness to
date and another button to indicate willingness to date, resembles somewhat the basic structure
of the Tinder dating app, which allows rapid selection with simple binary responses.
Study 1
Methods
Sample Size Determination. We conducted an a priori power analysis in G*Power 3.1 (Faul et
al., 2009), for a within-subjects ANOVA with 2 groups (attractiveness cue condition vs. control
cue condition). Since this is a novel online paradigm, we assumed a small-to-medium effect size
(partial η2 = 0.025). After specifying an alpha level of 0.05 and desired power of 0.9, we found

22

that 106 participants were needed to reliably find an effect. To account for potential attention
check failures and incomplete responding, we aimed to recruit 150 participants.
Sample. The study was designed and administered online using Qualtrics. Participants were
undergraduate students recruited from the University of Southern California psychology
participant pool. Due to lack of sign-ups, we only recruited 139 participants (83 females, 55
males, 1 did not respond).
Piloting Stimuli. In a pilot study, we recruited 68 participant pool students to rate 80
photographs of faces from each sex on attractiveness. Participants rated based on their preferred
sex – those who preferred men in dating rated male faces, and those who preferred women in
dating rated female faces. We took faces from the 10k US Adult Faces Database (Bainbridge et
al., 2013). The faces were selected so that they roughly appeared to be young adults. From the 80
faces, we selected the top 12 most attractive faces of each sex, as well as the 12 least attractive of
each sex; these faces would be used in conditioning. We also picked 34 faces from around the
mean of the attractiveness ratings to be used in the dating task.
Dating Preference. In the main study, participants were told that we were interested in how
people make decisions in social situations. We began by asking participants to indicate their
preferred sex (male or female) in dating. This choice was used to determine the sex of the faces
participants would see in throughout the study. Note that we had mistakenly failed to include a
third option, no preference, which was rectified in Studies 3 and 4.
Phase 1: Pavlovian Conditioning. Participants were told that they would see a series of faces on
the screen. To motivate them to pay attention to the faces, we also told participants that they will
be asked about the faces at the end of the study. Participants then saw 24 trials. Half of these
trials paired the 12 attractive faces with a visual cue (reward cue), and the other half paired the

23

12 less attractive faces with another visual cue (control cue). The order of the reward cue and
control cue trials were randomized within every 4 trials. The visual cues were two different
message borders framing the loading message “Please wait…” (see Appendix B, Figure 1 for the
borders). We selected them because they seemed more natural within the context of the study
than the appearance of an arbitrary pattern. On each trial, a cue was presented for 2 seconds
before the page automatically moved on to show a face. To move on from this page, participants
needed to click the arrow at the bottom of the page. The assignment of cue to attractive or less
attractive face was counterbalanced such that roughly half of the participants saw one cue
associated with attractive faces, and the other half saw the other cue associated with attractive
faces.
Phase 2: Dating Task. Participants were told that the task examines how people make fast
judgments, and that they will see a series of photos. They were told that they should judge
whether they would hypothetically date the person shown in each photo. They were told to
respond as fast as possible by pressing “E” on their keyboard for “No” and “I” for “Yes.”
Participants had 2 seconds to make a decision on each trial before the page automatically moved
on. Participants were then given 4 practice trials. After the practice trials, participants were
shown 30 main trials. On each trial, one of the cues would be shown for around 500
milliseconds, and then a face would appear for around 200 milliseconds. After the face,
participants were prompted to respond with an “E” or an “I” to indicate whether they would be
unwilling or willing to date the person shown. If participants failed to respond within 2 seconds,
task would automatically move onto the next trial. The faces were chosen so that their
attractiveness ratings were close to the mean rating in our pilot study. The order of the faces
shown during the dating task did not vary. However, we counterbalanced the task so that the

24

order of reward cue and control cue presentation flipped between participants. Thus, if one
participant saw a reward cue on the first trial and a control cue on the second trial, another
participant would see a control cue on the first trial and a reward cue on the second trial. This
ensured that no face on the dating task was always shown with one cue and not the other.
Questionnaires. For exploratory purposes, we administered the UCLA Loneliness Scale Version
3 (Russell, 1996), as well as the Social Anxiety Questionnaire for Adults (Caballo et al., 2012).
We included an attention check within the Social Anxiety Questionnaire asking participants to
select the right-most option. We then asked participants to indicate their relationship status,
report satisfaction with their relationship status, and report importance of looking for a new
partner at the moment. We later created a binary variable based on relationship status, indicating
whether a participant was single or not single. To check whether any of the faces we used were
recognizable figures, we asked participants to indicate whether they recognized any celebrities or
public figures from the study, and name anyone they recognized. We also measured
demographic variables: age, race-ethnicity, and year in school.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. We excluded 10 participants who failed the attention check,
resulting in a sample of 129 participants (81 females, 48 males) in our analyses. Overall,
participants were willing to date 24.80% of the faces. Around 58.91% (76 participants)
recognized at least one face in the study. Of the 129 participants, 53 were single and 76 were
either dating, in a casual relationship, or in an exclusive relationship.

25

Main Effect of Cue Condition. Data analysis was performed in R 3.5.1 using the “glmer”
function from the “lme4” package. Since dating choices were binary, we conducted a two-level
cross-classified generalized linear multilevel model with a binomial distribution (logistic), using
maximum likelihood estimation with Laplace approximation. We included participant-level
random intercepts and face-level random intercepts, accounting for differences between
participants and differences between faces, respectively. Level 1 consisted of the observation
level, with a sample size of 3842 out of 3870 possible observations, suggesting that participants
failed to respond on 28 trials. Level 2 consisted of the participant level, with 129 participants,
and the trial level, with 30 faces.
We entered our main predictor of interest: cue condition in Level 1, with 2 conditions:
whether the reward cue or control cue preceded a trial (coded as 1 and 0, respectively). We
controlled for counterbalancing of message border to outcome (Level 2) and counterbalancing of
cue to dating task trial (Level 2). Our dependent variable was reported willingness or
unwillingness to date (coded as 1 and 0, respectively).
The model equation is listed below:
Level 1:
P ( 𝑑𝑎 𝑡 𝑒 𝑖𝑗
= 1 ) = 𝑙 𝑜 𝑔𝑖𝑠𝑡 𝑖 𝑐 ( 𝛽 0 𝑗 + 𝛽 1 𝑗 cu e
𝑖𝑗
) + 𝑒 𝑖𝑗

Level 2:
𝛽 0 𝑗 = 𝛾 00
+ 𝛾 01
cou nt erba lance
𝑖𝑗
+ 𝛾 02
or d er
𝑖𝑗
+ 𝑢 0 𝑗
𝛽 1 𝑗 = 𝛾 10
+ 𝑢 1 𝑗
The intraclass correlations (ICC) for the participant level and trial level were calculated.
Participant-level ICC was 0.305, indicating that 30.5% of the variance in the data was due to

26

differences between participants. Face-level ICC was 0.032, indicating that 3.2% of the variance
in the data was due to differences between faces.
Our results showed that reward cues increased the likelihood of choosing to date, B =
0.208, p = 0.012, 95% CI [0.047, 0.370]. The coefficient indicates that the odds of indicating
willingness to date are multiplied by 1.231 when the reward cue is present compared to when the
control cue is present. There were no significant effects of the counterbalancing variables.
Results are shown in Appendix B, Table 1.
Interaction of Cue Condition with Relationship Variables. We then added interactions
between cue and mean-centered relationship satisfaction, mean-centered importance of finding a
date, and binary relationship status to the model, while removing the counterbalancing variables
and face-level random intercepts to avoid convergence issues. We did not find significant
interactions between cue and relationship satisfaction, and between cue and importance of
finding a date. However, we did find a significant interaction of cue with current relationship
status, B = -0.192, p = 0.035, 95% CI [-0.370, -0.013], such that the effect of the reward cue on
willingness to date was stronger among single people (see Appendix B, Figure 2 for visualization
of the interaction). See Appendix B, Table 2, for the results.
Discussion
We predicted that participants would be more willing to go on a date after seeing a cue
associated with attractive faces. The results confirmed our hypothesis. We also hypothesized that
relationship status, satisfaction, and priorities would interact with attractiveness cues. We only
found an interaction with relationship status, but it was in the direction that we predicted – single
participants were more sensitive to attractiveness cues, such that the effect of the reward cue on
willingness to date was stronger.

27

While our results confirmed our predictions, we needed to address a major issue. Most of
our sample recognized faces as public figures, and some of the faces we used as attractive faces
in conditioning were celebrities. Hence, it is possible that our reward cues had effects on dating
choice not through their association with attractiveness, but rather through their association with
the prestige of being a celebrity.
Study 2
The aim of Study 2 was to replicate Study 1 using non-celebrity faces. To do this, we
piloted more faces from the 10k US Adult Face Database (Bainbridge et al., 2013). This allowed
us to find less recognizable faces.
Moreover, the interaction of relationship status and attractiveness cue in Study 1 led us to
suspect that pre-existing motivation to date might strengthen the effect of attractiveness cues. We
therefore attempted to manipulate dating motivation by randomly assigning half our participants
to watch romantic or “sexy” videos at the beginning of the study, and the other half to watch
videos with the same characters without romantic content. We expected cue effects to be
stronger among participants who saw romantic videos.
Methods
Sample. Since we added a between-subjects manipulation, we increased our target sample size
to 240 USC participants. We met this goal, recruiting 240 participants (151 females, 84 males, 3
other, 2 did not answer). As in Study 1, the study was conducted online in Qualtrics.
Piloting Stimuli. In a pilot study, we recruited 36 participant pool students to rate 80
photographs of faces from each sex on attractiveness. Participants rated based on their preferred
sex – those who preferred men in dating rated male faces, and those who preferred women in
dating rated female faces. Additionally, we asked participants to indicate whether they

28

recognized the each of the faces as a public figure. The faces were from the 10k US Adult Faces
Database (Bainbridge et al., 2013), including some that had been used in Study 1. From the 80
faces, we selected the top 12 most attractive faces of each sex that were not easily recognizable,
as well as the 12 least attractive of each sex that were not easily recognizable. For the dating
task, we selected 34 of the faces that were around average in attractiveness and were not easily
recognizable. In the same pilot study, we also showed participants 4 pairs of video clips from 4
different films or television shows, to verify that the videos could potentially be used to prime
dating motivation. Each pair consisted of a video that contained romantic content and a video
featuring the same characters interacting with each other without romantic content. Participants
were asked to rate the videos on sexiness. As we expected, the romantic videos were rated as
sexier than non-romantic videos.
Dating Preference. In the main study, participants were told that we were interested in how
people make decisions in social situations. As in Study 1, we began by asking participants to
indicate their preferred sex in dating. This choice was used to determine the sex of the faces
participants would see in throughout the study.
Dating Motivation Manipulation. Before this portion, we administered a sound check to ensure
that participants could hear sounds on their computer. We first told participants that we were
testing out stimuli for future studies, and that we wanted ratings of different videos from
participants. We then randomly assigned participants to either watch romantic or non-romantic
videos, and asked participants to rate the videos they saw on positivity, negativity, sexiness, and
visual clarity.
Phase 1: Pavlovian Conditioning. Before the Pavlovian conditioning task began, we told
participants that they would see decorative patterns followed by faces. We told them that some of

29

the faces have been rated as highly attractive, and to pay attention to which of the decorative
patterns is associated with the highly attractive faces. We added these instructions to be
consistent with standard PIT paradigms (Nadler et al., 2011; Watson et al., 2014) that ask
participants to learn the association between the Pavlovian cues and their outcomes. The rest of
the Pavlovian conditioning task was the same as in Study 1, except that less recognizable faces
were used instead.
Phase 2: Dating Task. The dating task was the same as in Study 1, except different, less
recognizable faces were used.
Questionnaires. In addition to the Social Anxiety Questionnaire for Adults (Caballo et al.,
2012), we administered the BIS/BAS Scale, which measures approach and avoidance tendencies
(Carver & White, 1994), for exploratory purposes. We included an attention check within the
Social Anxiety Questionnaire asking participants to select the right-most option. We asked
participants to indicate their relationship status (which we later transformed into a binary
variable). We also asked for age, race-ethnicity, and year in school.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. We excluded 39 participants who reported having issues hearing or
seeing the videos (including 2 who did not answer the question), 1 participant for providing an
incorrect response on the sound check, and 23 participants for failing the attention check,
resulting in a sample of 177 participants (112 females, 63 males, 2 other) in our analyses.
Overall, participants were willing to date 27.95% of the faces. Of the 177 participants, 86 were
single and 91 were either dating, in a casual relationship, or in an exclusive relationship.

30

Main Effect of Cue Condition. Data analysis was performed in R 3.5.1 using the “glmer”
function from the “lme4” package. We conducted a two-level cross-classified generalized linear
multilevel model with a binomial distribution (logistic), using maximum likelihood estimation
with Laplace approximation. We included participant-level random intercepts and face-level
random intercepts. Level 1 consisted of the observation level, with a sample size of 5271 out of
5310 possible observations, suggesting that participants failed to respond on 39 trials. Level 2
consisted of the participant level, with 177 participants, and the trial level, with 30 faces.
We entered our main predictor of interest: cue condition in Level 1, with 2 conditions:
whether the reward cue or control cue preceded a trial (coded as 1 and 0, respectively). We
controlled for counterbalancing of message border to outcome (Level 2) and counterbalancing of
cue to dating task trial (Level 2). Our dependent variable was reported willingness or
unwillingness to date (coded as 1 and 0, respectively).
The model equation is listed below:
Level 1:
P ( 𝑑𝑎 𝑡 𝑒 𝑖𝑗
= 1 ) = 𝑙 𝑜 𝑔𝑖𝑠𝑡 𝑖 𝑐 ( 𝛽 0 𝑗 + 𝛽 1 𝑗 cu e
𝑖𝑗
) + 𝑒 𝑖𝑗

Level 2:
𝛽 0 𝑗 = 𝛾 00
+ 𝛾 01
cou nt erba lance
𝑖𝑗
+ 𝛾 02
or d er
𝑖𝑗
+ 𝑢 0 𝑗
𝛽 1 𝑗 = 𝛾 10
+ 𝑢 1 𝑗
The intraclass correlations (ICC) for the participant level and trial level were calculated.
Participant-level ICC was 0.309, indicating that 30.9% of the variance in the data was due to
differences between participants. Face-level ICC was 0.027, indicating that 2.7% of the variance
in the data was due to differences between faces.

31

We found no effect of cue condition on willingness to date, B = 0.038, p = 0.579, 95% CI
[-0.096, 0.172]. There were also no significant effects of the counterbalancing variables. Results
are shown in Appendix B, Table 3.
Interaction of Cue Condition with Relationship Variables. We then added an interaction
between cue and relationship status to the model. We did not find a significant interaction
between cue and relationship status, B = -0.034, p = 0.620, 95% CI [-0.169, 0.101]. See
Appendix B, Table 4, for the results.
Interaction of Cue Condition with Video Manipulation. We also tested the interaction of cue
condition with video condition in the model. We did not find a significant interaction, B = -
0.165, p = 0.228, 95% CI [-0.434, 0.104].
Discussion
We conducted Study 2 in order to replicate the findings of Study 1 with non-celebrity
faces. We did greatly reduce the recognizability of faces, but we were not able to replicate Study
1. We also attempted to prime dating motivation before conditioning, to investigate whether pre-
existing goals can amplify cue effects. We did not find evidence to support this.
One possibility we considered was that the photos we used in conditioning were older
and perhaps outdated. Many of the photos were from a decade ago; image quality, fashion sense,
hairstyles, and other features may have changed since then, especially with surges in social
media usage. Because of this aesthetic divide, our attractiveness cues may have been weaker.
Study 3
The aim of Study 3 was to replicate Study 1 using more up-to-date, non-celebrity photos.
To do this, we used profile pictures of models from the website Models.com. This also allowed
us to find more attractive faces while limiting the recognizability of the faces.

32

Methods
Sample Size Determination. Using the “powerSim” function from the “simr” package in R
3.5.1, We ran 1000 simulations of a multilevel binomial logistic regression with the parameter
estimate of cue condition from Study 1. Based on the study, we assumed an estimate of 0.20268
for cue condition, with random intercept variance of 1.433 for participant ID. Accounting for
random intercepts of participant ID, 165 participants were needed to obtain a power of 80.70%.
Hence, we aimed to recruit 175 participants.
Sample. Due to lack of sign-ups, we only recruited 170 participants (110 females, 59 males, 1
other).
Piloting Stimuli. In a pilot study, we recruited 39 participant pool students to rate 50
photographs of faces from each sex on attractiveness. Participants rated based on their preferred
sex – those who preferred men in dating rated male faces, and those who preferred women in
dating rated female faces. Those who had no preference were randomly assigned a sex to rate.
Additionally, we asked participants to indicate whether they recognized the each of the faces as a
public figure. The faces were model profile pictures from Models.com. From the 50 faces, we
selected the top 12 most attractive faces of each sex that were not easily recognizable, as well as
the 12 least attractive of each sex that were not easily recognizable. For the dating task, we used
the same faces used in Study 2.
Dating Preference. In the main study, participants were told that we were interested in how
people make decisions in social situations. We began by asking participants to indicate their
preferred sex (male, female, or no preference) in dating. This choice was used to determine the
sex of the faces participants would see in throughout the study. If participants indicated no
preference, they would be randomly assigned to see one of the sexes.

33

Phase 1: Pavlovian Conditioning. The Pavlovian conditioning task was the same as in Study 2,
except model faces were used instead.
Phase 2: Dating Task. The dating task was the same as in Study 2.
Questionnaires. For exploratory purposes, we administered the BIS/BAS Scale (Carver &
White, 1994). To examine how many participants ended up learning the Pavlovian associations,
we asked participants to indicate which cue was associated with highly attractive faces. We also
asked participants to indicate their relationship status (which we later transformed into a binary
variable). We asked participants to indicate whether they recognized any celebrities or public
figures from the study, and name anyone they recognized. We also asked for age, race-ethnicity,
and year in school.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. Out of the 170 participants, only 85 correctly indicated which cue
was associated with attractive faces. We excluded 14 participants who reported recognizing a
face and 1 participant who did not answer the recognition question, resulting in a sample of 155
participants (100 females, 54 males, 1 other) in our analyses. Overall, participants were willing
to date 18.90% of the faces. Of the 155 participants, 73 were single and 82 were either dating, in
a casual relationship, in an exclusive relationship, or in some other relationship arrangement.
Main Effect of Cue Condition. Data analysis was performed in R 3.5.1 using the “glmer”
function from the “lme4” package. We conducted a two-level cross-classified generalized linear
multilevel model with a binomial distribution (logistic), using maximum likelihood estimation
with Laplace approximation. We included participant-level random intercepts and face-level

34

random intercepts. Level 1 consisted of the observation level, with a sample size of 4625 out of
4650 possible observations, suggesting that participants failed to respond on 25 trials. Level 2
consisted of the participant level, with 155 participants, and the trial level, with 30 faces.
We entered our main predictor of interest: cue condition in Level 1, with 2 conditions:
whether the reward cue or control cue preceded a trial (coded as 1 and 0, respectively). We
controlled for counterbalancing of message border to outcome (Level 2) and counterbalancing of
cue to dating task trial (Level 2). Our dependent variable was reported willingness or
unwillingness to date (coded as 1 and 0, respectively).
The model equation is listed below:
Level 1:
P ( 𝑑𝑎 𝑡 𝑒 𝑖𝑗
= 1 ) = 𝑙 𝑜 𝑔𝑖 𝑠 𝑡 𝑖 𝑐 ( 𝛽 0 𝑗 + 𝛽 1 𝑗 cu e
𝑖𝑗
) + 𝑒 𝑖𝑗

Level 2:
𝛽 0 𝑗 = 𝛾 00
+ 𝛾 01
cou nt erba lance
𝑖𝑗
+ 𝛾 02
or d er
𝑖𝑗
+ 𝑢 0 𝑗
𝛽 1 𝑗 = 𝛾 10
+ 𝑢 1 𝑗
The intraclass correlations (ICC) for the participant level and trial level were calculated.
Participant-level ICC was 0.361, indicating that 36.1% of the variance in the data was due to
differences between participants. Face-level ICC was 0.013, indicating that 1.3% of the variance
in the data was due to differences between faces.
We found no effect of cue condition on willingness to date, B = 0.096, p = 0.244, 95% CI
[-0.066, 0.258]. There was a significant effect of the counterbalancing of cue to outcome, B = -
0.485, p = 0.047. This suggests that the message borders, outside of their association with the
outcomes, had a direct effect on willingness to date. Results are shown in Appendix B, Table 5.

35

Interaction of Cue Condition with Relationship Variables. We then added an interaction
between cue and relationship status to the model. We did not find a significant interaction
between cue and relationship status, B = 0.020, p = 0.903, 95% CI [-0.304, 0.345]. See Appendix
B, Table 6, for the results.
Contrast Effects. Given the failure to replicate Study 1, we considered whether attractiveness
cues produced a contrast effect among less attractive faces, such that they decrease willingness to
date for less attractive faces but increase willingness to date for more attractive faces. We took
means of attractiveness ratings for each face in the Dating Task and investigated the interaction
of cue condition with the attractiveness of the target faces. However, we found no significant
interaction, B = 0.022, p = 0.954, 95% CI [-0.731, 0.775].
Discussion
We conducted Study 3 in order to replicate the findings of Study 1 while using more up-
to-date photos of faces. We were not able to replicate Study 1. We did not find a main effect of
the attractiveness cues, nor did we find an interaction with relationship status. Moreover this
failure to replicate was not due to a contrast effect; the effect of attractiveness cues did not differ
depending on whether the target face was more or less attractive. Surprisingly, only half of our
participants were able to indicate the correct Pavlovian contingency, suggesting that they may
have had trouble distinguishing the attractive from the less attractive faces. Nevertheless, we did
not administer a contingency check in Study 1, so we cannot compare contingency awareness
between the two studies. One possibility we considered is that the models we picked may be too
unconventionally attractive. This means that there may be greater variability in whether people
perceive them as attractive or not.

36

Study 4
In Study 4 we attempted again to replicate Study 1, using faces with more conventional
appearances.
Methods
Sample Size Determination. As in Study 3, we aimed to recruit 175 participants.
Sample. Due to lack of sign-ups, we only recruited 132 participants (71 females, 60 males, 1 did
not answer).
Piloting Stimuli. In a pilot study, we recruited 40 participant pool students to rate 80
photographs of faces from each sex on attractiveness. The faces were AI generated faces from
Generated Photos (generated.photos). Since the faces were generated, they did not correspond to
any public figures. From the 80 faces, we selected the top 12 most attractive faces of each sex, as
well as the 12 least attractive of each sex. We also selected 34 faces that somewhat above
average in attractiveness for use in the dating task. We selected them to be above average
because we were concerned about the low rate of choosing to date in Study 2 (18.90%)
compared to Study 1 (24.80%).
Dating Preference. This procedure was the same as in Study 3.
Phase 1: Pavlovian Conditioning. This phase was the same as in Study 3, except that the model
faces were replaced with generated faces.
Phase 2: Dating Task. The dating task was the same as in Study 3, except that the model faces
were replaced with generated faces. Also, due to experimenter error, only 26 trials were
administered correctly; the other 4 were not counterbalanced and are therefore excluded from
analyses.

37

Questionnaires. We did not ask about recognition, since the faces were generated. Otherwise,
this portion of the study was the same as in Study 3.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. Out of the 132 participants, only 51 correctly indicated which cue
was associated with attractive faces. Overall, participants were willing to date 34.33% of the
faces. Of the 132 participants, 49 were single and 80 were either dating, in a casual relationship,
in an exclusive relationship, or in some other relationship arrangement. The remaining 3 did not
respond to the question.
Main Effect of Cue Condition. Data analysis was performed in R 3.5.1 using the “glmer”
function from the “lme4” package. We conducted a two-level cross-classified generalized linear
multilevel model with a binomial distribution (logistic), using maximum likelihood estimation
with Laplace approximation. We included participant-level random intercepts and face-level
random intercepts. Level 1 consisted of the observation level, with a sample size of 3351 out of
3432 possible observations, suggesting that participants failed to respond on 81 trials. Level 2
consisted of the participant level, with 132 participants, and the trial level, with 26 faces.
We entered our main predictor of interest: cue condition in Level 1, with 2 conditions:
whether the reward cue or control cue preceded a trial (coded as 1 and 0, respectively). We
controlled for counterbalancing of message border to outcome (Level 2) and counterbalancing of
cue to dating task trial (Level 2). Our dependent variable was reported willingness or
unwillingness to date (coded as 1 and 0, respectively).
The model equation is listed below:

38

Level 1:
P ( 𝑑𝑎 𝑡 𝑒 𝑖𝑗
= 1 ) = 𝑙 𝑜 𝑔𝑖𝑠𝑡 𝑖 𝑐 ( 𝛽 0 𝑗 + 𝛽 1 𝑗 cu e
𝑖𝑗
) + 𝑒 𝑖𝑗

Level 2:
𝛽 0 𝑗 = 𝛾 00
+ 𝛾 01
cou nt erba lance
𝑖𝑗
+ 𝛾 02
or d er
𝑖𝑗
+ 𝑢 0 𝑗
𝛽 1 𝑗 = 𝛾 10
+ 𝑢 1 𝑗
The intraclass correlations (ICC) for the participant level and trial level were calculated.
Participant-level ICC was 0.347, indicating that 34.7% of the variance in the data was due to
differences between participants. Face-level ICC was 0.142, indicating that 14.2% of the
variance in the data was due to differences between faces.
We found no effect of cue condition on willingness to date, B = 0.087, p = 0.322, 95% CI
[-0.085, 0.258]. There was a significant effect of the counterbalancing of cue on the Dating Task,
B = -0.338, p = 0.016. While the precise details are unclear, this suggests that willingness to date
differed when certain cues appeared with certain faces. Results are shown in Appendix B, Table
7.
Interaction of Cue Condition with Relationship Variables. We then added an interaction
between cue and relationship status to the model. We did not find a significant interaction
between cue and relationship status, B = -0.202, p = 0.261, 95% CI [-0.555, 0.150]. See
Appendix B, Table 8, for the results.
Contrast Effects. As in Study 2, we found no evidence for an interaction of cue with target face
attractiveness, B = -0.091, p = 0.717, 95% CI [-0.584, 0.402].
Discussion
We did not find evidence of a main effect of attractiveness cues or an interaction with
relationship status. As in Study 3, this failure to replicate was not due to a contrast effect.

39

Moreover, only 38.6% of our participants were able to indicate the correct Pavlovian
contingency, suggesting that they may have had trouble distinguishing the attractive from the
less attractive faces.
General Discussion
We conducted three studies with the aim of investigating whether Pavlovian cues
associated with attractiveness could influence dating choices. Additionally, we investigated
whether prior motives about finding a relationship would strengthen the effect of attractiveness
cues on dating choices. We found evidence to support both hypotheses in Study 1. However, we
could not replicate these findings in Studies 2-4.
Study Limitations
Our failure to replicate Study 1 is likely related to changes we made to the face stimuli in
follow-up studies. One possibility is that our attractive faces were not attractive enough for
participants to be able to learn through Pavlovian conditioning. Consistent with this account, we
found low rates of awareness of the cue-outcome relationships in Studies 3 and 4. While we did
not check for contingency awareness in Study 1, we suspect that the attractive faces in Study 1
may have been easier to distinguish from the less attractive faces, particularly if they were
celebrities. Celebrities can occupy a larger-than-life role in the minds of the public, with
disproportionate influence over behaviors such as consumer choice (Silvera & Austad, 2004) and
drug use (Zsila et al., 2020). The salience of celebrity status means that participants may have
learned cue-outcome relationships more easily in Study 1.
Another possibility is that celebrities are more conventionally attractive. That is, they are
more consistently attractive to a larger set of people. People tend to sexualize celebrities,
encouraged by tabloids trying to unearth their sex lives and the general aura of celebrity status

40

that transforms public figures into “sex symbols” (Mercer, 2013). Hence, being a celebrity may
itself contribute to perceptions of attractiveness, over and above the effects of physical
appearance. Moreover, people who are attractive to large segments of the population may be
more likely to become celebrities, since they may appeal to larger audiences.
Finally, since participants may be more familiar with celebrities, their lives, and their
roles in mass media, familiarity may be the working ingredient in Study 1. Since participants
recognized more of the attractive than the unattractive faces, we may have inadvertently trained
them to associate the reward cue with familiarity rather than attractiveness. These familiarity
cues may have then led participants to be more willing to date unfamiliar people. It is unclear if
this effect would have operated through reward mechanisms, since familiar stimuli can be either
rewarding or aversive. That is, it is unclear if familiarity per se would be rewarding.
It is also important to note that our study design departed significantly from the design of
existing Pavlovian-instrumental transfer studies. In particular, prior studies measured persistence
rather than choice as the primary dependent variable, and tested the presence of Pavlovian-
instrumental transfer in nominal extinction (Hogarth et al., 2015; Talmi et al., 2008; Watson et
al., 2014) – that is, only the Pavlovian cues were shown to participants without any other
information about the potential outcome of their responses. Persistence may be more sensitive to
the effects of Pavlovian cues, since participants can exert effort proportional to the motivational
strength of a cue. Moreover, binary choice options statistically have less power than continuous
measures of persistence, such as frequency of key-pressing. We discuss this aspect of study
design further in the Concluding Remarks section.
Moreover, unlike previous studies that test in nominal extinction, our test of Pavlovian-
instrumental transfer – the dating task – may have provided too much information to participants

41

to anchor on when making decisions. Our intention in showing faces to participants during the
task was to ensure that participants would vary their decisions between trials rather than rely on a
strategy of always responding yes or no. Moreover, showing faces during the decision-making
task would mirror real-world dating apps that allow users to see photos before making any
decisions. However, showing the faces may have allowed participants to weigh factors other than
the Pavlovian cues more heavily in their decisions. We plan to use more ambiguous stimuli for
test of PIT in future studies, for example by blurring the faces, presenting silhouettes, or showing
only the backs of people’s heads so participants cannot see their faces.
Conclusion
Our initial study showed promising results, suggesting that Pavlovian cues associated
with attractive faces may increase dating desire, especially among single people. However, we
found that some of the attractive faces we used for conditioning were recognizable celebrity
faces. To avoid confounding celebrity status and attractiveness, we used non-celebrity faces in
Studies 2-4. In doing so, we may have reduced the effectiveness of our conditioning paradigm, as
we failed to replicate results of Study 1. Future efforts should first focus on disentangling aspects
of the attractive photos in Study 1 that may have been responsible for the cuing effect.

42

Chapter 3: Situating Pavlovian Processes within a Cultural Framework
Portions of the Introduction in this chapter are from:
Wang, P., Atari, M., & Oyserman, D. (2022). Who can I count on: Honor and self-reliance
during the COVID-19 pandemic. Manuscript in preparation.

Introduction
In response to the challenges of survival, people develop physical tools, ideas, traditions,
and narratives in order to adapt to and change their environments. These elements are part of
what we call culture, the sets of practices and ways of thinking that characterize a group of
people (Lehman et al., 2004). Culture allows people to adapt to the problems of survival by
creating their own environments, coordinating intragroup behavior, passing down knowledge,
and encouraging innovation (Lehman et al., 2004; Oyserman, 2017). It can influence phenomena
ranging from visual perception (Nisbett & Miyamoto, 2005), to consumer behavior (Kacen &
Lee, 2002), to public health (Wang et al., 2022). Cultures have core themes (e.g., individualism,
collectivism, and honor) that help people make meaning of their experiences; these themes
correspond to mindsets that organize cognition, motivation, and behavior to interpret
information, form values, and attain goals in ways specific to each of the themes (Oyserman,
2017). However, while different cultures may practice different traditions and emphasize
different cultural mindsets, the mindsets are available to all cultures, since all of them are needed
to solve universal problems (Oyserman, 2015, 2017). Thus, although different cultures show
systematic differences on cognitive, behavioral, and even neural measures (Choi et al., 2007;
Kitayama & Park, 2010; Morris & Peng, 1994), the specific mindset that members of a culture

43

adopt at a given moment depends on the contextual cues, affordances, and other situational
factors present in the moment (Oyserman, 2015).
In this chapter, I attempt to integrate what we know about Pavlovian-instrumental
transfer with the perspective of one of the cultural mindsets: honor. To our knowledge, no
studies have tested how PIT operates within the context of culture, and we intend to address this
gap. I will begin with a brief summary of three commonly discussed cultural mindsets. I then
focus on our mindset of interest, honor, and its role in defensive aggression. Finally, I present
results of three studies that examined whether an honor mindset can amplify the effects of
Pavlovian threat cues on defensive aggression in a military defense computer game.
The Cultural Mindsets
Culture-as-situated cognition theory predicts that human culture evolved within ecologies
as a “good enough” solution to basic problems of survival: since people need others to survive,
groups need to be sustained over time and need to organize interactions to reduce conflict while
facilitating individual welfare (Oyserman, 2011, 2017). Sustaining the group implies sensitivity
to issues of group membership, fitting in, and belonging, termed collectivistic mindset.
Organizing interactions implies sensitivity to reputation, loyalty to family and other groups, and
willingness to protect them, termed honor mindset. Facilitating individual welfare implies
sensitivity to individual needs, desires, and preferences, termed individualistic mindset. Each of
these cultural mindsets can be understood as a set of mental representations or cognitive schemas
containing relevant mental content, cognitive procedures, and goals (Oyserman, 2017). Culture-
as-situated cognition theory predicts that each of these cultural mindsets (collectivism, honor,
individualism) is part of human culture, that each society develops within a particular niche, and

44

that situated features within niches matter by affecting which cultural mindset is on the mind
(Oyserman, 2017).
Individualism emphasizes uniqueness and standing out (Oyserman, 2017), and people in
individualistic cultures tend to identify more as independent individuals than as members of
groups (Triandis, 2001). Cognitively, it is associated with analytic thinking, in which people tend
to focus on informational elements as independent of each other (Choi et al., 2007). For example,
Ji et al. (2000) found that students from an individualistic background (European American)
relied less on visual context in a perception task than students from a collectivistic background
(East Asian). By promoting distinctiveness, individualism helps address the need for innovation
in societies (Oyserman, 2017).
Collectivism emphasizes connecting with others and fitting in (Oyserman, 2017), and
people in collectivistic cultures tend to identify more as members of groups than as independent
individuals (Triandis, 2001). Cognitively, it is associated with holistic thinking, in which people
tend to process informational elements in relation to one another (Choi et al., 2007). Thus,
students from a collectivistic background relied more on visual context in a perception task (Ji et
al., 2000). Moreover, East Asians tend to give more weight to contextual factors when making
causal attributions about behavior (Choi et al., 1999). By emphasizing interpersonal
relationships, collectivism helps to coordinate groups and provides social support to group
members (Triandis et al., 1988).
Honor emphasizes social comparison (Oyserman, 2017). People in honor cultures are
attuned to differences in reputation and are strongly motivated to protect their reputations (Leung
& Cohen, 2011). Honor cultures have strong norms of reciprocity, both positive and negative
(Leung & Cohen, 2011). Cognitively, honor may be associated with sensitivity to reputational

45

threats (Osterman & Brown, 2011), although empirical research on cognitive mechanisms of
honor is limited. By emphasizing reciprocity and reputation, honor helps encourage cooperation
within groups and discourage violations of social norms (Harinck et al., 2013; Nowak et al.,
2016).
Honor Culture
Socioecological factors formed the cultural phenotypes of honor in its cultural evolution
(Cao et al., 2021; Nowak et al., 2016; Shackelford, 2005). Historically, in pastoral environments,
an individual’s standing, especially economic standing, is directly related to livestock which is
vulnerable to theft. Pastoral environments often lack any sources of central legal enforcement.
Vulnerability to theft and lack of a central governing body have been theorized to favor
psychological phenotypes characteristic of honor cultures (Leung & Cohen, 2011). In this way,
reputational values and reciprocity norms in honor cultures represent adaptations to interpersonal
threats in lawless environments. Indeed, as Chapter 4 will discuss, reputation systems help
sustain cooperation within groups by allowing people to reward trustworthy members and
withhold benefits from untrustworthy members (Milinski, Semmann, & Krambeck, 2002).
Likewise, reciprocity, both positive (Berg et al., 1995) and negative (Nowak et al., 2016), has
been shown to promote cooperation within groups. McElreath (2003) modeled adaptations to
lawless environments, demonstrating that reputational effects can have a crucial influence on
fitness under these conditions. Focusing on interpersonal threats, McElreath (2003) concluded
that reputation-based strategies may result in strong negative reciprocity in lawless conditions.
Consistent with these accounts, honor culture emerged in places where central authorities were
unreliable (Cohen & Nisbett, 1997) or even absent (see Nowak et al., 2016; Shackelford, 2005),
and resources scarce (Uskul & Cross, 2020). Lacking institutional safeguards, people needed a

46

means to reduce the chances that others might try to exploit them, attack them, or appropriate
their resources.
Cultural psychologists have argued that honor is a cultural mindset which focuses people
on protecting their well-being and that of their families or larger ingroups (Cohen et al., 1996;
Oyserman, 2017). To accomplish this, honor-minded people are motivated to defend their
reputation so that others will be less likely to take advantage of them. People in honor cultures
stay vigilant to slights against their reputations to deter others who might exploit them, their
families, or groups (Leung & Cohen, 2011). Empirically, people who endorse honor values are
more likely to respond aggressively to norm violations such as insults (e.g., Cohen et al., 1996;
Flinkenflogel et al., 2020). This allows them to show that they do not tolerate attacks against
them, making potential offenders less likely to target them.
Honor-related reciprocity features behaviors that may be seen as “irrational.” Cohen et al.
(2018) theorizes that such irrationality is an important mechanism for deterring interpersonal
threats and signaling trustworthiness. When threatened, people in honor cultures commit to risky,
even life-threatening means to retaliate, ignoring the costs of their actions. While seemingly
irrational, these responses signal to others that the actor will not tolerate any attacks and will
spare no effort to ensure that attackers are punished, even if it means hurting themself in the
process (Cohen et al., 2018). A similar principle applies to positive reciprocity – people in honor
cultures will exert disproportionate effort to return a favor or pay a debt. This leads others to see
them as trustworthy in nature, willing to put in every effort to do the right thing (Cohen et al.,
2018).
It should be noted that, in conflict situations, people in honor cultures may actually
attempt to de-escalate the situation first (Harinck et al., 2013). However, they resort to

47

aggression if they feel insulted, since their reputations are threatened (Harinck et al., 2013). It is
precisely because confrontation in honor cultures is costly that people attempt de-escalation, and
politeness norms in honor culture help fulfill this function (Harinck et al., 2013).
Honor Culture and Defensive Aggression
These accounts of honor culture all point to a close relationship between honor mindset
and defensive aggression – aggression against threat. In a classic study, Cohen et al. (1996)
compared men from honor-culture backgrounds (the American South), and from non-honor
backgrounds (the American North). In the experimental condition, a confederate would bump
into a participant and then call them an “asshole.” The participant would then complete a series
of tasks and measures assessing emotional and cognitive responses. The researchers found that,
after the insult, participants from an honor-culture background showed greater anger, exhibited
heightened cortisol levels (indicating stress), exhibited heightened testosterone levels (indicating
readiness for aggression), and were more likely to have violence on the mind. Similarly, Günsoy
et al. (2015) found that, when compared with participants from Ghana, participants from Turkey
(an honor-culture background) tended to respond to interpersonal threats in hypothetical
vignettes with retaliation rather than avoidance. These studies suggest that threats are more likely
to trigger aggression in people of honor-culture backgrounds.
Complementing this account, Cohen and Leung (2012) connect honor to military contexts
by proposing that honor cultures often treat both moral courage and physical courage as aspects
of honor. The two are intertwined, since doing the right thing may entail physical risk and
sacrifice. The authors collected biographical information on United States Presidents,
Congresspeople, and Supreme Court Justices, as well as existing survey ratings of these figures
from scholars. They found that military experience correlated positively with ratings of integrity,

48

character, and moral leadership, but only among Southern political figures. The findings support
their account, suggesting that honor culture may combine moral and physical courage in ways
that other cultures do not. The findings also suggest that honor is closely related to physical
defense and protection of an ingroup.
Given the role of honor mindset in defensive aggression, we sought to examine whether
honor could amplify the effects of Pavlovian cues associated with interpersonal threat on
defensive behavior.
Study Aim
The primary aim of our studies was to examine the role of culture in the Pavlovian-
instrumental transfer. If people with honor mindsets have more salient protection and reciprocity
goals, then they may respond more strongly to cues associated with interpersonal threat. To our
knowledge, if our predictions are confirmed, these would be the first studies to show the role of
culture in Pavlovian-instrumental transfer. The studies have the potential to add to existing
theories of culture. For example, Pavlovian cues could constitute one of the situated factors in
culture-as-situated-cognition theory (Oyserman, 2015) that triggers mindset-relevant goals.
Furthermore, the studies would suggest that sociocultural values could influence how people
perceive and respond to Pavlovian cues. This could potentially vastly expand the relevance of
Pavlovian cues to broader social phenomena, such as political or cultural activities.
Study Design
Our study draws primarily on a prior study by Nadler et al. (2011) examining the effects
of Pavlovian cues in a computer game about military defense, which was in turn adapted from a
paradigm by Paredes-Olay et al. (2002). Participants were tasked with defending a country from
enemy invasion. In the instrumental training phase, participants learned to destroy enemy

49

vehicles using different keyboard buttons that represented different weapons. In the Pavlovian
conditioning phase, participants learned to associate different cues with the different enemy
threats or with no outcome. When compared with the no-outcome cue, subsequent presentation
of threat cues led participants to fire their weapons more vigorously.
While the authors did not discuss the study in terms of social behavior, focusing instead
on demonstrating PIT effects in a quasi-avoidance task (Nadler et al., 2011), we suggest that the
task involves social outcomes and a social premise. The overarching narrative throughout the
task specifies a military invasion – a social phenomenon with clear ingroup-outgroup
distinctions. The enemy targets represent interpersonal threats – agents of a hostile outgroup. If
we frame the task in terms of social behavior, the relevance of the honor mindset becomes
evident. Honor, with its associations with defensive aggression (Cohen et al., 1996), threat
vigilance (Leung & Cohen, 2011), and martial virtue (Cohen & Leung, 2012), may sensitize
participants to effects of the threat cue, such that they respond more strongly with defensive
fighting.
Closely following the design by Nadler et al. (2011), our studies consisted of three main
parts. First, we trained participants to fight off enemy planes and ships with keyboard buttons
that represented either an anti-plane weapon or an anti-ship weapon. Next, we trained
participants to associate different colored boxes with the presence of enemy planes, ships, tanks,
or no outcome. Finally, in the PIT phase, we presented the colored boxes and allowed
participants to respond with the weapons in nominal extinction (i.e. without giving feedback on
enemies destroyed). In the questionnaire portion, we asked participants to report their
endorsement of honor values using the Honor Values Scale (HVS; Novin & Oyserman, 2016).

50

We predicted that endorsement of honor values would strengthen the effect of threat cues on
weapon firing.
In Studies 2 and 3, we attempted to prime honor mindset to examine the causal role of
honor. In Study 2, we did this by manipulating the order of the HVS so that it appeared either
before or after the PIT phase. Participants who completed the honor scale first should be more
likely to have honor on the mind while responding in the PIT phase. We therefore expected
participants who completed the honor scale first to show stronger PIT effects. This method has
been used to prime honor mindset in prior research (Flinkenflogel et al., 2020; Novin &
Oyserman, 2016).
In Study 3, we attempted to prime honor by showing participants reading passages about
honor. Participants who read honor passages before the PIT phase should show stronger PIT
effects than those who did not.
Lastly, we also predicted that our studies would replicate previous findings with the same
design by Nadler et al. (2011). That is, participants should demonstrate specific PIT by
responding to plane cues with anti-plane weapons and to ship cues with anti-ship weapons. They
should also demonstrate general PIT by responding to the tank cue more strongly overall than to
the no-outcome cue.
Study 1
Methods
Sample Size Determination. We conducted an a priori power analysis in G*Power 3.1 (Faul et
al., 2009), for a repeated-measures ANOVA with 2 groups (threat cue condition vs. control cue
condition) and a within-between interaction. Since there is no precedent for our interaction of
interest, we assumed a small-to-medium effect size (partial η2 = 0.02). After specifying an alpha

51

level of 0.05 and desired power of 0.9, we found that 132 participants were needed to reliably
find an effect. We therefore aimed to recruit 160 participants.
Screening. Participants were allowed to participate if they were fluent English, and they lived in
the U.S. Due to recent concerns of gender imbalance on Prolific (Charalambides, 2021), we used
the Prolific screening option to evenly recruit female and male participants.
Sample. The study was designed and administered online using PsyToolkit (Stoet, 2010, 2017).
We recruited 161 participants on Prolific (83 females, 74 males, 2 other, 2 did not answer).
Participant ages ranged from 19 to 74 years old (M = 36.59, SD = 13.60).
Phase 1: Instrumental Conditioning. We told participants that the study was about how people
make decisions. Following procedures from Nadler et al. (2011), we told participants that they
will play a game in which they are a military unit commander defending the country of Viltoma
from enemy attacks. They are then told to fend off air and sea attacks by press the “A” and “L”
keys. One fires missiles at warplanes, while the other fires missiles at warships, but due to a
malfunction, participants will not know which key corresponds to which weapon, and they must
learn the association by trial and error. We also told participants that their missiles will miss, so
they will need to fire multiple times (i.e. press a key multiple times in order to destroy an
enemy). Participants then saw 20 instrumental conditioning trials. On each trial, a photograph of
either a ship or a plane was shown on the screen, and participants would need to press the “A”
and “L” keys multiple times to learn the instrumental contingencies between key and type of
threat. Each trial required a random number of keypresses, 2-8 presses, on the correct key before
the threat would be destroyed. Once a threat was destroyed, participants received feedback
confirming that it was destroyed, and the page moved on to the next trial. The assignment of key
to outcome was counterbalanced, such that the “A” key was anti-plane and “L” was anti-ship for

52

some participants while “A” was anti-ship and “L” was anti-plane for others. The order of the
trials was randomized such that two plane and two ship trials were randomized every four trials.
See Appendix C, Figure 1, for a sample image of an instrumental conditioning trial.
Phase 2: Pavlovian Conditioning. In the next phase, we told participants that they will receive
coded signals in the form of colored boxes from another unit commander, indicating the type of
enemy the commander was facing. We told the participants to observe the progress of the
commander and learn the meaning of the different colors by watching the screen. Participants
saw four different cues (white, purple, red, orange), each consistently paired with one of four
different outcomes (plane, ship, tank, no outcome). The outcomes were represented by photos of
the vehicle, or, in the case of no outcome, by the text: “No enemy.” The assignment of cue to
outcome was randomized between participants. On each trial, participants saw a cue for 1,000
ms, followed by an outcome for 1,000 ms. Participants were then asked to press space to
continue. Trials were separated by a fixation point that appeared on the screen for 750 ms.
Participants saw 56 trials of Pavlovian conditioning (14 plane cue, 14 ship cue, 14 tank cue, 14
no-outcome cue). The order of the trials was randomized such that each cue occurred twice
within each 8-trial block.
Phase 3: Pavlovian-instrumental Transfer. In this portion, we told participants that they will
see the colored boxes on the screen. However, they will not be able to see the enemy or what
happens to the enemy. We told participants that they may press whatever they feel is useful
whenever the “A” and “L” keys appear on the screen. On each trial, one of the four cues was
shown for 3,000 ms, accompanied by the “A” and “L” key icons. We measured the number of
times participants pressed each key in response to the cue. Trials were separated by a fixation
point that appeared on the screen for 750 ms. Participants saw 40 trials overall (10 of each cue).

53

The four cues were randomized for every four trials. See Appendix C, Figure 2, for a sample
image of a PIT trial.
Awareness Check. To ensure that participants learned the Pavlovian contingencies, we showed
them each of the cues and asked them to indicate the outcome with which each cue was
associated. Participants who failed this check were excluded from analyses.
Questionnaires. We then asked participants to complete the 18-item HVS (Novin & Oyserman,
2016), a measure of honor values (e.g., “Reputation matters and should be vigorously
defended.”). For exploratory purposes, we also included the 12-item Support for Diplomacy
Scale (Vail & Motyl, 2010) and the 30-item Moral Foundations Questionnaire (Graham et al.,
2011). Additionally, we asked participants to report any technical issues they may have
experienced during the study. Finally, we asked participants for demographic information: age,
gender, race-ethnicity, and U.S. state of residence.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. We excluded 33 participants who failed the awareness check, as
well as 1 participant who made as many as 440 keypresses on a single trial, indicating that they
may have held the key down instead of pressing it repeatedly. Our final sample had 127
participants (65 females, 59 males, 1 other, 2 did not answer) in our analyses.
For our analyses, we dummy-coded our cue condition variable to create three cue
variables: plane cue, ship cue, and tank cue. A value of 0 on all three cue variables indicates that
the corresponding trial presented a no-outcome cue. This allowed us to compare responses to the
three threat cues to responses to the no-outcome cue.

54

Specific PIT. We calculated the proportion of responding with the anti-plane key to total number
of keypresses. To do this, we added 1 to both the number of anti-plane presses and the number of
anti-ship presses to insure that we did not divide by 0. We then divided the number of anti-plane
presses by the combined number of keypresses to create a variable measuring proportion of anti-
plane keypresses to total keypresses.
Data analysis was performed in R 3.5.1 using the “lmer” function from the “lme4”
package. We conducted a two-level multilevel linear model using restricted maximum
likelihood. We included participant-level random intercepts, accounting for differences between
participants. Level 1 consisted of the observation level, with a sample size of 5080 observations.
Level 2 consisted of the participant level, with 127 participants.
We entered the three cues in Level 1, with 2 values: whether the cue was shown or not
(coded as 1 and 0, respectively). Our dependent variable was the proportion of anti-plane
keypresses to total keypresses.
The intraclass correlations (ICC) for the participant level were calculated. Participant-
level ICC was 0.026, indicating that 2.6% of the variance in the data was due to differences
between participants.
As predicted, participants responded with greater frequency of anti-plane keypresses
when a plane cue appeared, B = 0.236, p < 0.001. Participants responded with lower frequency of
anti-plane keypresses when a ship cue appeared, indicating a preference for anti-ship keypresses,
B = -0.230, p < 0.001. Interestingly, participants also showed a smaller, but statistically
significant, preference for anti-ship keypresses when a tank cue appeared, B = -0.031, p = 0.001.
General PIT. Data analysis was performed in R 3.5.1 using the “glmmTMB” function from the
“glmmTMB” package. The distribution of count data for total keypresses indicated that

55

participants pressed 0 times on a large number of trials. We therefore conducted a two-level
generalized linear multilevel model with a Poisson distribution (logistic) and a zero-inflation
model. By including the zero-inflation model, we could model the zero trials and non-zero trials
separately, with the zero-inflation model predicting the likelihood of participants pressing 0
times. We included participant-level random intercepts in both models, accounting for
differences between participants. Level 1 consisted of the observation level, with a sample size
of 5080 observations. Level 2 consisted of the participant level, with 127 participants.
We entered the three cues in Level 1 in both the Poisson and zero-inflation models. Our
dependent variable in the Poisson model was total number of keypresses on each trial, while the
dependent variable in the zero-inflation model was whether participants pressed 0 times or not on
each trial.
The intraclass correlations (ICC) for the participant level were calculated using the
“icc_counts” function in the “iccCounts” package for a zero-inflated Poisson model. Participant-
level ICC was 0.537, indicating that 53.7% of the variance in the data was due to differences
between participants.
As predicted, the Poisson model indicates that participants responded more frequently
when a plane cue appeared, B = 0.502, p < 0.001, when a ship cue appeared, B = 0.497, p <
0.001, and when a tank cue appeared, B = 0.434, p < 0.001, than when a no-outcome cue
appeared. Moreover, the zero-inflation model indicates that participants were less likely to press
0 times on a trial when a plane cue appeared, B = -8.739, p < 0.001, when a ship cue appeared, B
= -8.607, p < 0.001, and when a tank cue appeared, B = -5.309, p < 0.001, than when a no-
outcome cue appeared.

56

Interaction of Honor and Threat Cue. To avoid convergence issues, we created a single
variable indicating whether a threat cue (i.e. plane, ship, or tank cue) was shown on a trial or not.
We then examined the interaction of honor values with the presence of a threat cue. We used the
General PIT zero-inflated Poisson model, replacing the three cue variables with our threat cue
variable and adding an interaction with scores on the HVS. As predicted, the Poisson model
indicates that participants who value honor showed stronger effects of the threat cue, B = 0.230,
p < 0.001, 95% CI [0.155, 0.306]. Simple effects analysis indicated that the difference in
responding between threat cue and no-outcome cue trials was larger among participants who
scored at or above mean HVS scores, B = 0.716, p < 0.001, than among participants who scored
below mean HVS scores, B = 0.355, p < 0.001. The effect was driven by increased responding to
threat cues and decreased responding to neutral cues among participants who endorse honor
more strongly (see Appendix C, Figure 3, for visualization). We found a main effect of threat
cues as well; the presence of the threat cue predicted greater responding, B = 0.475, p < 0.001.
There was no main effect of honor values. The zero-inflation model only found a significant
main effect of threat cue, B = -6.302, p < 0.001; the presence of a threat cue decreased the
probability that participants would respond 0 times on a trial. See Appendix C, Table 1, for
results.
Discussion
We found evidence for our hypothesis; honor values moderated the effect of the threat
cue, such that cue effects were stronger among participants who scored higher in honor values.
Honor drove greater responding to the threat cue and lower responding to the neutral cue,
mirroring accounts stating that people in honor cultures retaliate against threats (Leung & Cohen,
2011) while avoiding confrontation with non-threatening others (Harinck et al., 2013).

57

Our results are promising, suggesting that culture may influence how people respond to
Pavlovian cues. However, because we examined individual differences in honor values, we could
not establish the causal role of honor in the study.
Study 2
The purpose of Study 2 was to replicate the findings of Study 1 and use priming to
establish the causal role of honor in moderating our PIT effect. To do this, we manipulated the
order of the HVS scale, such that some participants saw the HVS immediately before the PIT
phase while others saw the HVS after the PIT phase. We expected participants who saw the
honor scale first to respond more strongly to the threat cue. Moreover, we expected to replicate
the Study 1 interaction of honor values and cue condition.
Methods
Screening. Participants were allowed to participate if they were fluent in English, and they lived
in the U.S. As in Study 1, we used the Prolific screening option to evenly recruit female and
male participants.
Sample. The study was designed and administered online using PsyToolkit (Stoet, 2010, 2017).
As in Study 1 we aimed to recruit 160 participants. We ended up recruiting 160 participants, but
2 participants did not complete the study, leaving us with 158 Prolific participants (76 females,
80 males, 2 other). Participant ages ranged from 19 to 81 years old (M = 40.43, SD = 13.77).
Phase 1: Instrumental Conditioning. The instrumental conditioning task was the same as in
Study 1.
Phase 2: Pavlovian Conditioning. Pavlovian conditioning was the same as in Study 1.
Honor Priming. We attempted to prime honor mindset by manipulating the order of the HVS
scale. Participants were randomly assigned to see the HVS scale before Phase 3, or at the end of

58

the study (after all other questionnaires were completed). The control group was given the 18-
item Need for Cognition scale (NFC; Cacioppo et al., 1984) before Phase 3 instead of the HVS.
The short-form NFC measures individual differences in the motivation to engage in cognitive
effort, and we selected the scale for our control condition because it matched the number of
items in the HVS and because it was ostensibly unrelated to interpersonal aggression. The
experimental group completed the NFC at the end of the study, in the questionnaire portion.
Phase 3: Pavlovian-instrumental Transfer. This phase was the same as in Study 1.
Awareness Check. As in Study 1, we checked to make sure participants learned the correct
Pavlovian contingencies. Participants who failed this check were excluded from analyses.
Questionnaires. Aside from the HVS and NFC, we also included the 12-item Support for
Diplomacy Scale (Vail & Motyl, 2010) and the 8-item Defensive Fight scale (Corr & Cooper,
2016) for exploratory purposes. Since the study was conducted in late February of 2022,
coinciding with the Russian invasion of Ukraine, we asked participants about their attitudes
towards the invasion. We also asked participants to report their political leanings. We then asked
participants to report any technical issues they may have experienced during the study. Finally,
we asked participants for demographic information: age, gender, race-ethnicity, and U.S. state of
residence.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. We excluded 22 participants who failed the awareness check. We
also systematically excluded anyone who responded more than 30 times on a trial, which would
suggest that they held the keys down instead of pressing them; this resulted in the exclusion of 1

59

additional participant. Our final sample had 135 participants (66 females, 67 males, 2 other) in
our analyses.
We used the same calculations as in Study 1 to dummy-code our cue variables.
Specific PIT. We calculated the proportion of responding with the anti-plane key to total number
of keypresses using the same method as in Study 1.
Data analysis was performed in R 3.5.1 using the “lmer” function from the “lme4”
package. We conducted a two-level multilevel linear model using restricted maximum
likelihood. We included participant-level random intercepts, accounting for differences between
participants. Level 1 consisted of the observation level, with a sample size of 5400 observations.
Level 2 consisted of the participant level, with 135 participants.
We entered the three cues in Level 1, with 2 values: whether the cue was shown or not
(coded as 1 and 0, respectively). Our dependent variable was the proportion of anti-plane
keypresses to total keypresses.
The intraclass correlations (ICC) for the participant level were calculated. Participant-
level ICC was 0.062, indicating that 6.2% of the variance in the data was due to differences
between participants.
As predicted, participants responded with greater frequency of anti-plane keypresses
when a plane cue appeared, B = 0.231, p < 0.001. Participants responded with lower frequency of
anti-plane keypresses when a ship cue appeared, indicating a preference for anti-ship keypresses,
B = -0.237, p < 0.001. Participants also showed a smaller, but statistically significant, preference
for anti-ship keypresses when a tank cue appeared, B = -0.067, p = 0.001.
General PIT. Data analysis was performed in R 3.5.1 using the “glmmTMB” function from the
“glmmTMB” package. The distribution of count data for total keypresses indicated that

60

participants pressed 0 times on a large number of trials. We therefore conducted a two-level
generalized linear multilevel model with a Poisson distribution (logistic) and a zero-inflation
model. We included participant-level random intercepts in both models, accounting for
differences between participants. Level 1 consisted of the observation level, with a sample size
of 5400 observations. Level 2 consisted of the participant level, with 135 participants.
We entered the three cues in Level 1 in both the Poisson and zero-inflation models. Our
dependent variable in the Poisson model was total number of keypresses on each trial, while the
dependent variable in the zero-inflation model was whether participants pressed 0 times or not on
each trial.
The intraclass correlations (ICC) for the participant level were calculated using the
“icc_counts” function in the “iccCounts” package for a zero-inflated Poisson model. Participant-
level ICC was 0.639, indicating that 63.9% of the variance in the data was due to differences
between participants.
As predicted, the Poisson model indicates that participants responded more frequently
when a plane cue appeared, B = 0.499, p < 0.001, when a ship cue appeared, B = 0.495, p <
0.001, and when a tank cue appeared, B = 0.429, p < 0.001, than when a no-outcome cue
appeared. Moreover, the zero-inflation model indicates that participants were less likely to press
0 times on a trial when a plane cue appeared, B = -8.133, p < 0.001, when a ship cue appeared, B
= -8.211, p < 0.001, and when a tank cue appeared, B = -5.588, p < 0.001, than when a no-
outcome cue appeared.
Interaction of Honor Values and Threat Cue. We created a single variable indicating whether
a threat cue (i.e. plane, ship, or tank cue) was shown on a trial or not. We then examined the
interaction of honor values with the presence of a threat cue. We used the General PIT zero-

61

inflated Poisson model, replacing the three cue variables with our threat cue variable and adding
an interaction with scores on the HVS. As predicted, the Poisson model indicates that
participants who value honor showed stronger effects of the threat cue, B = 0.381, p < 0.001,
95% CI [0.317, 0.446]. Simple effects analysis indicated that the difference in responding
between threat cue and no-outcome cue trials was larger among participants who scored at or
above mean HVS scores, B = 0.552, p < 0.001, than among participants who scored below mean
HVS scores, B = 0.282, p < 0.001. Interestingly, the effect was driven largely by decreased
responding to neutral cues among participants who endorse honor more strongly (see Appendix
C, Figure 4, for visualization). We found a main effect of threat cues as well; the presence of the
threat cue predicted greater responding, B = 0.480, p < 0.001. There was also a main effect of
honor values; honor values correlated negatively with overall responding, B = -0.378, p < 0.001,
but this seems to be driven by decreased responding to the neutral cue. The zero-inflation model
only found a significant main effect of threat cue, B = -6.506, p < 0.001; the presence of a threat
cue decreased the probability that participants would respond 0 times on a trial. See Appendix C,
Table 2, for results.
Interaction of Honor Priming and Threat Cue. Using the General PIT zero-inflated Poisson
model, we the examined the interaction of honor priming (honor scale before PIT = 1, honor
scale after PIT = 0) with threat cue. We found a main effect of threat cue; participants responded
more strongly after seeing a threat cue, B = 0.570, p < 0.001. We found a main effect of threat
cue in the zero-inflation model as well; the presence of a threat cue decreased the probability that
participants would respond 0 times on a trial, B = -6.550, p < 0.001. We found a significant
interaction of honor priming and threat cue, B = -0.160, p = 0.018, 95% CI [-0.293, -0.027].
However, the interaction was in the opposite direction of what we expected. Based on simple

62

effects analysis, we found that the threat cue was weaker when participants completed the honor
scale first, B = 0.410, p < 0.001, than when participants completed the honor scale at the end of
the study, B = 0.576, p < 0.001. This led us to suspect that the NFC scale that we used for our
control group may have had its own effect on cue strength.
Interaction of Need for Cognition and Threat Cue. We examined the interaction of honor
priming and NFC scores in the zero-inflated Poisson model. However, convergence issues led us
to remove the interaction term from the zero-inflation portion of the model, so the zero-inflation
model only included main effects of NFC and the threat cue. We found a significant interaction
of NFC and threat cue, B = 0.215, p = 0.011. Simple effects analysis indicated that the difference
in responding between threat cue and no-outcome cue trials was larger among participants who
scored at or above mean NFC scores, B = 0.531, p < 0.001, than among participants who scored
below mean NFC scores, B = 0.367, p < 0.001.
Discussion
We replicated findings from Study 1; honor values moderated the effect of the threat cue,
such that cue effects were stronger among participants who scored higher in honor values.
Interestingly, unlike in Study 1, the interaction was driven primarily by lower responding to the
neutral cue among high-honor participants. While the reason is unclear, it is possible that our
honor priming manipulation may have influenced the interaction in some way. Nevertheless, the
interaction suggests that high-honor participants discriminated between threat and non-threat
cues more clearly.
Manipulating the order of the honor scale did moderate the effect of the threat cue on
responding, but in the opposite direction of what we expected. This may be because participants
who saw the honor scale first saw the need for cognition scale last, and participants who saw the

63

honor scale last saw the need for cognition scale first. Hence, we primed need for cognition,
which we initially thought was unrelated to the task, in our attempt to prime honor. Our findings
suggest that need for cognition amplifies the cue effects, much like honor values, so the need for
cognition priming may have lead participants to respond more strongly to the threat cue. In
hindsight, need for cognition may have led participants to be more engaged in the task, and put
in greater effort to learn how to play the game.
Study 3
In Study 3, we attempted a different method of priming honor. We showed participants
honor-related reading passages before or after the PIT phase. We had two aims: first, we sought
to replicate again the interaction of honor values and threat cues that we found in Studies 1 and
2. Second, we sought to establish the causal role of honor by priming honor using the reading
passages.
Methods
Screening. Participants were allowed to participate if they were fluent English, and they lived in
the U.S. We used the Prolific screening option to evenly recruit female and male participants.
Sample. The study was designed and administered online using PsyToolkit (Stoet, 2010, 2017).
We added attention checks in the study to ensure that participants read and understood the
reading passages. To account for potential exclusions due to attention check failures, we
increased our target sample size to 200. We ended up recruiting 201 participants, but 1
participant did not complete the study, leaving us with 200 Prolific participants (101 females, 95
males, 3 other, 1 did not answer). Participant ages ranged from 18 to 73 years old (M = 38.50,
SD = 12.52).

64

Phase 1: Instrumental Conditioning. We made one change to the instrumental conditioning
task from Study 1. We increased the range of button presses required to destroy an enemy from
2-8 presses to 4-10 presses. We did this to increase the difficulty of the task and to require more
effort from participants during the PIT phase. We suspected that increasing effort requirements
would increase the interaction effects of honor and threat cue, with high- and low-honor
participants diverging such that high-honor participants would exert more effort to meet the
challenge, while low-honor participants would give up sooner.
Phase 2: Pavlovian Conditioning. Pavlovian conditioning was the same as in Study 1.
Honor Priming. We attempted to prime honor mindset by showing participants reading
passages related to honor. Participants were told that this portion of the study was about reading
comprehension. Participants were randomly assigned to see the passages before Phase 3, or at the
end of the study (after all questionnaires were completed). Thus, participants in the control group
completed the PIT phase directly after Pavlovian conditioning instead of reading the passages
first. The three short passages were adapted from three sources: an article from The Atlantic
about honor in inner-city Black communities (Anderson, 1994), codes of conduct suggested to be
from the American Old West (Weiser, 2020), and an anecdote about a bar fight used in a prior
study on honor (Cohen & Nisbett, 1997). The passages were edited or rewritten to ensure that
they were easier to agree with and more relatable to a broader segment of Americans. The first
passage discussed using honor and reputation to deter interpersonal threats. The second passage
listed unwritten codes of conduct that emphasized trustworthiness, reciprocity, and other honor-
related virtues. The last passage discussed a story about a man who got into a fight in order to
defend his pride. After finishing each passage, participants would see a question that assessed
reading comprehension for the preceding passage. The three questions also acted as an attention

65

check – participants who could not answer the three questions correctly were excluded from
analyses.
Phase 3: Pavlovian-instrumental Transfer. This phase was the same as in Study 1.
Awareness Check. As in Study 1, we checked to make sure participants learned the correct
Pavlovian contingencies. Participants who failed this check were excluded from analyses.
Questionnaires. We used the same measures as in Study 2, with the honor scale administered at
the beginning of the questionnaire portion.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained trial-level observations nested
within participant-level data. We excluded 35 participants who failed the awareness check, as
well as 24 participants who failed the reading-passage attention checks. We also systematically
excluded anyone who responded more than 30 times on a trial, which would suggest that they
held the keys down instead of pressing them; this resulted in the exclusion of 3 additional
participants. Our final sample had 138 participants (68 females, 68 males, 2 other) in our
analyses.
We used the same calculations as in Study 1 to dummy-code our cue variables.
Specific PIT. We calculated the proportion of responding with the anti-plane key to total number
of keypresses using the same method as in Study 1.
Data analysis was performed in R 3.5.1 using the “lmer” function from the “lme4”
package. We conducted a two-level multilevel linear model using restricted maximum
likelihood. We included participant-level random intercepts, accounting for differences between

66

participants. Level 1 consisted of the observation level, with a sample size of 5520 observations.
Level 2 consisted of the participant level, with 138 participants.
We entered the three cues in Level 1, with 2 values: whether the cue was shown or not
(coded as 1 and 0, respectively). Our dependent variable was the proportion of anti-plane
keypresses to total keypresses.
The intraclass correlations (ICC) for the participant level were calculated. Participant-
level ICC was 0.045, indicating that 4.5% of the variance in the data was due to differences
between participants.
As predicted, participants responded with greater frequency of anti-plane keypresses
when a plane cue appeared, B = 0.239, p < 0.001. Participants responded with lower frequency of
anti-plane keypresses when a ship cue appeared, indicating a preference for anti-ship keypresses,
B = -0.221, p < 0.001. Participants also showed a smaller, but statistically significant, preference
for anti-ship keypresses when a tank cue appeared, B = -0.046, p < 0.001.
General PIT. Data analysis was performed in R 3.5.1 using the “glmmTMB” function from the
“glmmTMB” package. The distribution of count data for total keypresses indicated that
participants pressed 0 times on a large number of trials. We therefore conducted a two-level
generalized linear multilevel model with a Poisson distribution (logistic) and a zero-inflation
model. We included participant-level random intercepts in both models, accounting for
differences between participants. Level 1 consisted of the observation level, with a sample size
of 5520 observations. Level 2 consisted of the participant level, with 138 participants.
We entered the three cues in Level 1 in both the Poisson and zero-inflation models. Our
dependent variable in the Poisson model was total number of keypresses on each trial, while the

67

dependent variable in the zero-inflation model was whether participants pressed 0 times or not on
each trial.
The intraclass correlations (ICC) for the participant level were calculated using the
“icc_counts” function in the “iccCounts” package for a zero-inflated Poisson model. Participant-
level ICC was 0.653, indicating that 65.3% of the variance in the data was due to differences
between participants.
As predicted, the Poisson model indicates that participants responded more frequently
when a plane cue appeared, B = 0.427, p < 0.001, when a ship cue appeared, B = 0.436, p <
0.001, and when a tank cue appeared, B = 0.366, p < 0.001, than when a no-outcome cue
appeared. Moreover, the zero-inflation model indicates that participants were less likely to press
0 times on a trial when a plane cue appeared, B = -10.793, p < 0.001, when a ship cue appeared,
B = -10.175, p < 0.001, and when a tank cue appeared, B = -6.641, p < 0.001, than when a no-
outcome cue appeared.
Interaction of Honor Values and Threat Cue. We created a single variable indicating whether
a threat cue (i.e. plane, ship, or tank cue) was shown on a trial or not. We then examined the
interaction of honor values with the presence of a threat cue. We used the General PIT zero-
inflated Poisson model, replacing the three cue variables with our threat cue variable and adding
an interaction with scores on the HVS. However, there was no significant interaction of honor
values with threat cue, B = -0.040, p = 0.170, 95% CI [-0.097, 0.017]. We found a main effect of
threat cues; the presence of the threat cue predicted greater responding, B = 0.425, p < 0.001.
There was no main effect of honor values. The zero-inflation model only found a significant
main effect of threat cue, B = -6.917, p < 0.001; the presence of a threat cue decreased the

68

probability that participants would respond 0 times on a trial. See Appendix C, Table 3, for
results.
Interaction of Honor Priming and Threat Cue. Using the General PIT zero-inflated Poisson
model, we the examined the interaction of honor priming (honor passages before PIT = 1, honor
passages after PIT = 0) with threat cue. We found a significant interaction of honor priming and
threat cue, B = 0.278, p < 0.001, 95% CI [0.166, 0.391]. Simple effects analysis indicated that
the difference in responding between threat cue and no-outcome cue trials was larger among
participants who saw the reading passages first, B = 0.602, p < 0.001, than among participants
who saw the passages after, B = 0.323, p < 0.001. See Appendix C, Figure 5, for visualization.
Additionally, we found a main effect of threat cue; participants responded more strongly after
seeing a threat cue, B = 0.323, p < 0.001. We found a main effect of threat cue in the zero-
inflation model as well; the presence of a threat cue decreased the probability that participants
would respond 0 times on a trial, B = -6.503, p < 0.001. We also found a main effect of priming;
participants who saw the passages first generally responded less than participants who saw the
passages after, B = -0.640, p < 0.001. See Appendix C, Table 4, for results.
Interaction of Honor Priming, Honor Values, and Threat Cue. Given that there was no
interaction of honor value scores with threat cue, we tested whether our priming manipulation
influenced the moderation effect of honor values on the relationship between threat cues and
responding. We did not find a three-way interaction, either in the Poisson model, B = 0.095, p =
0.138, or in the zero-inflation model, B = -0.218, p = 0.705.
Influence of Honor Priming on Honor Values. The reading passages and the honor scale were
separated by the PIT task, and hence the reading passages were unlikely to have an effect on

69

honor values scores. However, we checked whether there was an effect anyway. In a linear
regression, we found that honor priming did not influence honor values, B = 0.051, p = 0.718.
Discussion
We found an interaction of honor priming with threat cue. Participants who saw the
honor passages first showed stronger effects of the threat cue than participants who saw the
honor passages later. This provides some evidence for the causal role of honor in amplifying the
effects of threat cues. Methodologically, the study also shows that reading passages can be used
to make honor mindset accessible. At the time of writing, there are very few such primes
(Flinkenflogel et al., 2020; Novin & Oyserman, 2016).
Although we replicated specific and general PIT effects, we surprisingly did not replicate
the interaction honor values with threat cue in predicting total keypresses. While the reasons are
unclear, we consider two possibilities. First, the increase in keypresses required to defend against
an enemy may have introduced some confusion into the task. Although participants were told in
the instructions that they would need to press the keys multiple times, many participants reported
at the end of the study that they were confused by the instrumental learning task, that they
thought the task was broken, or that they thought their keypresses were not registering. We
counted 11 participants who reported this as a technical issue at the end of the study. Thus,
participants may have perceived the instrumental actions as unreliable in Study 3. Second, the
honor priming manipulation may have weakened PIT effects overall in the study. Participants
who saw the honor passages first were more prone to fatigue effects than participants who saw
the passages after; this is supported by the finding that participants who saw the passages first
responded less overall than participants who saw the passages later. The weakened PIT effect
may have made it less likely for us to find an interaction with honor values. It is unlikely that the

70

honor priming prevented replication by influencing honor values scores, since we did not find a
relationship between honor priming and honor values. Moreover, we did not find a three-way
interaction between honor priming, honor values, and threat cue, indicating that the mere
introduction of the honor passages did not weaken the influence of honor values on PIT effects.
General Discussion
Although the evidence from Study 3 is mixed, our findings across our three studies reveal
some promising evidence that cultural factors influence responding to Pavlovian cues. While we
could not replicate this in Study 3, Studies 1 and 2 showed that endorsement of honor values
strengthened the effect of threat cues on defensive behavior. Study 3 showed that priming
participants with honor leads them to respond more strongly to threat cues, suggesting that the
role of honor in amplifying cue effects may be causal.
Our findings situate Pavlovian-instrumental transfer effects within the broader framework
of cultural motivation and behavior. They suggest that Pavlovian cues may trigger culturally
learned motivations, pointing to a way by which situational factors could activate cultural
mindsets. They also indicate how Pavlovian mechanisms work with higher-level elements of
cognition, such as cultural values, to influence behavior. From the standpoint of reward learning,
cultural mindsets may work as revaluation mechanisms, changing the significance of cue-
associated outcomes and thereby modifying the effects of Pavlovian cues.
Our studies have implications for a variety of phenomena in the real world. For example,
elements of social media, such as notifications or ringtones, may act as Pavlovian cues by
signaling the potential for social interaction and approval. With its emphasis on reputation and
impression management, could honor influence the way these cues affect social media behavior?
One possibility is that, when honor is on the mind, these cues could lead people to control their

71

online image more carefully. Studies could also examine the social and physical cues that people
in honor cultures look for to anticipate threats. In natural settings, these cues may help prime
people for confrontations.
Limitations
There are some limitations with our studies. The failure in Study 3 to replicate the
interaction between honor values and threat cue suggests that we made design changes that were
suboptimal. Follow-up studies should minimize the potential effects of irrelevant factors, such as
fatigue effects and task confusion. For example, to minimize fatigue effects, a follow-up study
could try shortening the passages to only include the most relevant honor-related information.
This would shorten the amount of time participants need to spend on the study before completing
the PIT phase. To reduce task confusion, follow-up studies could return to lower requirements
for button-pressing (as in Studies 1 and 2) or make it more clear to participants that they need to
press the keys multiple times.
Moreover, while the reading passages seemed to be effective in moderating the influence
of the threat cue, they may be priming other, relevant concepts in additional to honor. For
example, violence and physical defense are common themes in the three passages, and these
concepts could have influenced PIT effects separately from honor. After all, violence is not
inherently related to honor, although honor may sometimes involve violence. Follow-up studies
could address this by adding passages for the control group. These would be similar to the honor
passages, except that they discuss violence without referencing honor-specific values, allowing
researchers to control for the mere presence of violent content in the honor passages.
Finally, while the instructions in our studies were based on ones used in prior research
(Nadler et al., 2011; Paredes-Olay et al., 2002), they may have also facilitated non-Pavlovian

72

mechanisms that could explain our effects. In particular, the instructions provide participants
with a causal model of the Pavlovian conditioning phase, presenting the colored boxes as coded
signals indicating the type of enemy threat. While Pavlovian contingencies in the real world can
often be causal, participants may have acquired a purely informational association by learning
the code without learning the motivational relevance of the cues. We plan to address this issue by
reducing narrative details in the instructions for the Pavlovian learning phase, and by increasing
the motivational salience of the outcomes with additional features that enhance immersion (e.g.,
alarm sounds that occur with enemy images).
Conclusion
We conducted three studies to show that honor culture can influence how people respond
to Pavlovian cues associated with interpersonal threat. Although more work needs to be done to
refine our study design and replicate our results, our findings show promising evidence that
culture factors can influence Pavlovian-instrumental transfer effects. The implications point to
greater integration of theories from reward learning and cultural psychology.

73

Chapter 4: Testing the Effects of Approval Cues on Generosity in an Economic Game
Introduction
Humans are remarkable in their tendencies to cooperate, share, and help one another
(Fehr & Fischbacher, 2003). As individuals, we give resources to charities and follow
unenforced etiquette out of consideration for others. We even help complete strangers. While
cultures differ in specific norms and traditions, all cultures evolved to facilitate cooperation
(Oyserman, 2017). In this chapter, I explore altruism and the factors that promote the expression
of altruism before hypothesizing and testing the potential role of Pavlovian cues in triggering
altruistic behavior.
Altruism
Cooperative and altruistic behaviors, behaviors that appear to primarily benefit others
rather than oneself, likely evolved in response to the survival needs of our early ancestors (Hill,
2002; Fehr & Fischbacher, 2003). These behaviors allowed communities to share resources,
coordinate group hunting, and divide labor such that different members of the community make
different contributions to the group (Hill, 2002; Fehr & Fischbacher, 2003). To the extent that
everyone contributes, altruism collectively increases benefits to groups and allows the highly
complex societies that we have today (Fehr & Fischbacher, 2003). Yet, not everyone contributes.
Only a handful of free-riders, people who make no contribution to the group but cash in on the
group benefits of others’ altruism, are sufficient to cause cooperation within the group to break
down (Fehr & Schmidt, 1999). Hence, there must be motivational mechanisms in place to
increase altruism and decrease free-riding, systems that, in a sense, reward altruistic behavior
while punishing selfish behavior.

74

Factors that Increase Altruism. An important mechanism for sustaining group
cooperation is altruistic reward (Fehr & Fischbacher, 2003). Reciprocity norms create situations
in which altruistic behaviors promote long-term benefits, since initial acts of generosity are
rewarded with future acts of generosity from the recipient, encouraging a chain of altruistic
reciprocation. Trust games are a way to study altruistic reward. In the investment game (Berg et
al., 1995), participants are given an amount of money, and they are asked to decide how much of
it (if any) to entrust to an anonymous partner. The amount they send would be tripled, allowing
the partner to benefit more from the investment. On the next round, the partner would then
decide how much money to return. Since the amount was tripled, there is the potential for both
partners to benefit from reciprocity, if they are willing to cooperate. The game showed that the
vast majority of participants in the investor role were willing to entrust money to complete
strangers, likely in the expectation that the act would be reciprocated (Berg et al., 1995). This
suggests that altruistic reward can motivate cooperation. Moreover, Flynn (2003) found that the
frequency with which employees at a telecommunications firm exchange favors is positively
related to their reputations and productivity; employees who helped one another more often
ended up being more productive and were more highly respected by other employees. The latter
outcome, reputational gain, is part of another important mechanism for sustaining group
cooperation.
By helping to track personal histories of altruism, reputation can also promote altruistic
behavior (Fehr & Fischbacher, 2003). Reputational mechanisms allow people to confer benefits
to others known to be trustworthy, and punish or withhold benefits from those known to be
untrustworthy. This phenomenon, known as indirect reciprocity, creates another motivational
pressure to act altruistically (Nowak & Sigmund, 1998). Prior research suggests that people do

75

engage in indirect reciprocity when reputational information is available, and that indirect
reciprocity can help sustain cooperation for collective benefit (Milinski et al., 2002). Milinski et
al. (2002) found that participants in an indirect reciprocity game were more willing to give
resources to players who had previously contributed to benefit the group than players who
contributed nothing. Moreover, individual contributions for collective benefit could be sustained
only after such reputational information was known. Reputation motivates potential free-riders to
cooperate, since cooperators benefit from having a positive reputation.
Theorization from cultural psychology seem to complement findings regarding the roles
of altruistic reward and reputation in promoting cooperation. As discussed in Chapter 3, people
value reputation and reciprocity strongly in places that historically lacked strong governments,
places where cooperative and other norms were more difficult to enforce (Cohen et al., 1996;
Leung & Cohen, 2011). Emphases on reputation and reciprocity in these societies may have
provided a motivational impetus to cooperate (Leung & Cohen, 2011).
Altruistic punishment is another mechanism for promoting altruistic behavior – people
are willing to bear a cost to themselves in order to punish others for not following cooperative
norms (Güth et al., 1982). Participants will spend money to punish other participants who do not
split money fairly between themselves and a partner, even when there is no clear benefit to the
punisher (Fehr & Fischbacher, 2004). In addition, Fehr and Schmidt (1999) show that
cooperation within a group can be maintained when members are given opportunities to punish
free-riders.
Cuing Altruism
Given the importance of reciprocity, reputation, and altruistic punishment in maintaining
cooperation, it is possible that cues signaling outcomes related to any of these three factors may

76

influence altruistic behavior. Consistent with this possibility, Pichon et al. (2007) found that the
presentation of positive religious words primed participants to help more with the distribution of
charity pamphlets. Positive religious words also made words related to prosociality more
accessible. Given that religions tend to promote values associated with reciprocity and reputation
(Sosis, 2005), the use of religious words may have influenced helping behavior through religious
associations with reciprocity and reputation. Relevant to this discussion is the costly signaling
theory of religion – being in a demanding, strict religious group involves a commitment to group
cooperation that is “costly-to-fake,” and hence membership in such a group is a signal to others
that one will reliably cooperate (Sosis & Bressler, 2003). It may therefore be possible to prime
altruism using religious identity symbols, since such membership symbols represent
trustworthiness. Other, nonreligious ways of signaling trustworthiness, such as reputation scores
on shopping sites, exist as well (Przepiorka & Berger, 2017), although the religious method is
perhaps a more salient example.
The preceding theories about costly signaling suggest that contextual stimuli can promote
altruism by indicating trustworthiness. They suggest that people use symbols associated with
trustworthiness to communicate to others their willingness to cooperate. While these theories
suggest that contextual cues can prime altruistic behavior, , studies have not explicitly tested the
effects of Pavlovian cues on altruism, although researchers have hypothesized that Pavlovian
cues may be relevant (Gęsiarz and Crockett, 2015; Seymour & Dolan, 2008). Gęsiarz and
Crockett (2015) suggest that a Pavlovian cue may promote altruism by indicating that someone is
in need. Someone else’s distress may act as punishment in Pavlovian learning if the observer
feels empathic concern for them. Since empathic concern increases altruistic helping
(FeldmanHall et al., 2015), cues associated with empathized suffering may drive people to use

77

whatever instrumental actions they can employ to relieve the distress of the afflicted person.
Interestingly, this hypothesis suggests an interaction between mechanisms responsible for
Pavlovian-instrumental transfer and affective empathy to shape behavior.
Seymour and Dolan (2008) hypothesize that Pavlovian cues may indicate trustworthiness
and the likelihood that a person will reciprocate cooperation. They argue that, through Pavlovian
learning, people can detect key features of persons and situations that predict reciprocity, making
it unnecessary for people to explicitly engage in the complex search for markers of
trustworthiness. Their account seems to suggest that reciprocity can be conceptualized as an
epistemic reward, one that contains information about the likelihood of receiving a more
concrete reward. This is congruent with findings showing that the opportunity to gain knowledge
about a future, positive outcome can elicit reward prediction error signals in the nucleus
accumbens, just as primary rewards do (Charpentier et al., 2018).
While both of these hypotheses identify interesting directions for research, we chose to
focus on social approval as a motivator of altruism. We based our decision on empirical evidence
showing that social approval can be used in Pavlovian learning and Pavlovian-instrumental
transfer (Lehner et al., 2017), and that social approval activates reward-related circuitry in the
brain (Izuma et al., 2008; Kohls et al., 2013). Social approval reflects reputational gains, thereby
tying our research to the role of reputation systems in group cooperation. The existing evidence
gave us greater confidence that Pavlovian cues could influence altruism through social approval.
Moreover, we considered reputation to be a powerful motivator, since we believe that it can
strengthen the effects of the other motivators of altruism. People who enjoy altruistic reward can
acquire a reputation for playing fair and reciprocating, while people who engage in altruistic
punishment can acquire a positive reputation for keeping free-riders in line.

78

Study Aim
The primary aim of our study was to test whether Pavlovian cues associated with social
approval could influence altruistic behavior. The study would tell us not only about the viability
of using Pavlovian cues to influence altruism, but also about how reputational concerns arise
during cooperative exchange in the “real world.” Indeed, people may have extensive learning
histories associating specific symbols, such as stars, “thumbs-up”, and hearts, with social
approval, and the presence of these symbols may encourage further pursuit of reputation gain,
either through prosocial behavior or some other means. Understanding the role of Pavlovian cues
would allow us to understand the concrete, proximal triggers that prompt cooperation. To our
knowledge, prior research has not investigated the effects of Pavlovian cues on altruism, so our
study would help clarify these issues.
Additionally, our study would address our broader aim of integrating research on
Pavlovian cues with our knowledge of social behavior.
We also measured personality traits from the Big Five Inventory (Rammstedt & John,
2007). We anticipated that the effectiveness of social approval cues would depend in part on how
much participants cared about reputation. We therefore identified three traits of interest, based on
their relevance to social approval: extraversion, agreeableness, and conscientiousness. Our aim
was to examine whether participants who scored higher on these traits showed greater altruism in
response to social approval cues.
Study Design
Our study consisted of two main parts. In the first phase, we trained participants to
associate different cues with either social approval or no outcome. In the second phase, we asked
participants to complete repeated trials on a dictator game, in which participants are given an

79

endowment of money and must choose how much of the money to share with another
participant. Dictator games have been commonly used as measures of charitable giving (Nettle et
al., 2013; Rigdon et al., 2009). We tested whether presentation of the conditioned cues during the
game could affect the amount of money participants chose to give to other participants. We
predicted that presentation of an approval cue during the game would lead participants to allocate
more money to another participant.
Note that, while our research question was inspired by Pavlovian-instrumental transfer,
we did not actually include an instrumental conditioning task in our study. We assumed that
participants knew giving more money would altruistically benefit other participants. While this
means that our design did not strictly imitate a PIT paradigm, it did have the advantage of
imitating real-world behavior, at least in comparison to existing PIT paradigms that commonly
ask participants to press buttons to effect an outcome (Hogarth et al., 2014; Prévost et al, 2012;
Watson et al., 2014). Outside the lab, people do make decisions about how much money to give
to charities and people in need.
Methods
Sample Size Determination
We conducted an a priori power analysis in G*Power 3.1 (Faul et al., 2009), for a within-
subjects ANOVA with 2 groups (approval cue condition vs. neutral cue condition). Since this is
a novel online paradigm, we assumed a small-to-medium effect size (partial η2 = 0.03). After
specifying an alpha level of 0.05 and desired power of 0.85, we found that 76 participants were
needed to reliably find an effect. We therefore aimed to recruit 80 participants.
Screening

80

Participants were allowed to participate if they were fluent English, and they lived in the
U.S. Due to recent concerns of gender imbalance on Prolific (Charalambides, 2021), we used the
Prolific screening option to evenly recruit female and male participants. Moreover, to ensure
quality of responses, we recruited participants who had completed at least 10 submissions and
had an approval rate of at least 90%.
Sample
The study was designed and administered online using PsyToolkit (Stoet, 2010, 2017).
We recruited 80 participants on Prolific, but 3 participants did not complete the study, so we
obtained data for 77 participants (40 females, 36 males, 1 other). Participant ages ranged from 19
to 72 years old (M = 37.05, SD = 13.15).
Phase 1: Pavlovian Conditioning
Participants were told that that the first portion of the study is to understand how people
learn new information. They were told that they would see either a purple or orange box on the
screen, followed by either a person holding a thumbs-up gesture, or a blank screen. The use of a
thumbs-up gesture as social reward in PIT has been previously demonstrated by Lehner et al.
(2017). Following protocol from PIT studies (Nadler et al., 2011; Watson et al., 2014),
participants were asked to learn which color preceded the thumbs-up gesture. Participants were
then shown 28 trials of Pavlovian conditioning. On 14 of these trials, participants saw a colored
box that was always associated with the thumbs-up gesture. On the other 14 trials, participants
saw a different box that was always associated with a blank screen. The order of the reward and
neutral trials were randomized within every 4 trials. Trials began with a fixation point shown on
the center of the screen for 500 ms. A cue was then shown for 500 ms, followed by the outcome
for 1,000 ms. At the end of each trial, participants were asked to press the space bar to continue

81

onto the next trial. The assignment of reward or neutral outcome to a purple or orange box was
counterbalanced between participants, such that some participants saw the purple box as a
reward cue, while others saw the orange box as a reward cue.
Phase 2: Dictator Game
Participants then received instructions about the dictator game. They were told that they
would make a series of decisions about how to allocate money. They were told that, on each
round, they would be given a specific endowment of money and be paired with another
participant, and they would decide whether to share a portion of the money with the partner, or
keep the entire amount. To encourage participants to share, we told them that whatever amount
they chose to share would be doubled, so that the partner would get twice the amount that was
given. Participants were told that the partners have previously participated in a separate study
that offered no extra bonuses, and they will only be identified by participant number. Participants
were also told that one of the rounds would be randomly selected and enacted, so they should
treat each round as if it could really happen. They were told to ignore any colored boxes they see
on the screen during the task.
On each round, participants were paired with a different participant, and received a
random endowment between $0.50 and $1.00, selected by the experimenter using a random
number generator during the design of the study. They were shown the participant ID of the
partner for the round (a preset, randomly selected number from 1 to 150) and offered a binary
choice between keeping the full endowment, or giving a specific, preset amount to the partner
(See Appendix, D, Figure 1, for a screenshot of a sample trial). This binary design was inspired
by a similar design in prior research (Hackel & Zaki, 2018), and we believed that the rapid
decision-making it allowed was more susceptible to Pavlovian priming. The preset amount that

82

participants were allowed to give were based on a proportion of the endowment that was
randomized on every trial. The proportions were also selected using a random number generator
(from 10% to 50%) during study design. Both the endowments and the proportions were varied
from trial to trial to reduce the likelihood that participants would commit to a particular strategy
in the game (e.g., always choosing to share).
Before each trial, either the approval or the neutral cue were presented on the screen.
Subsequently, the screen showed them: the amount of endowment, the partner’s ID, the amounts
they would give and keep from sharing, and the amounts they would give and keep from keeping
the entire endowment. Participants could press “A” on their keyboards to share, and “L” to keep
the entire endowment, with 8 seconds to respond. Before beginning the main trials, participants
were given 4 practice trials. Data from these practice trials were not used in analyses.
Participants completed 4 practice trials and 36 main trials overall. Half of the trials were
preceded by the approval cue, and the other half were preceded by the neutral cue. Moreover, the
pairing of cue condition with decision trial was counterbalanced, such that, for each partner ID,
roughly half the participants saw an approval cue beforehand, and the other half saw a neutral
cue beforehand.
In actuality, participants were not paired with actual people, and every participant saw the
same trials in randomized order. Participants were debriefed at the end of the study, and were
given a bonus of $1.00 (the maximum endowment they could have received on a trial).
After the dictator game, participants were shown the purple and orange boxes, and asked
to indicate which color was associated with the thumbs-up gesture. Following PIT paradigm
procedures (Nadler et al., 2011; Watson et al., 2014), we excluded participants who could not
correctly report the outcome associated with the approval cue.

83

Big Five Inventory
We then administered the 10-item short form of the Big Five Inventory (Rammstedt &
John, 2007), measuring the Big Five personality traits: extraversion, agreeableness,
conscientiousness, openness to experience, and neuroticism. We also measured demographic
variables: age, gender, and race-ethnicity.
Results
Since we conducted a repeated-measures study, we determined that multilevel analyses
were appropriate for the structure of our data, which contained face-level observations nested
within participant-level data. We excluded 2 participants who could not correctly choose the
approval cue, resulting in a sample of 75 participants in our analyses. Overall, participants chose
to share on 46.16% of the trials.
Main Effect of Cue Condition
Data analysis was performed in R 3.5.1 using the “glmer” function from the “lme4”
package. Since generosity choices were binary, we conducted a two-level generalized linear
multilevel model with a binomial distribution (logistic), using maximum likelihood estimation
with Laplace approximation. Level 1 consisted of the observation level, with a sample size of
2680 out of 2700 possible observations, suggesting that participants failed to respond on 20
trials. Level 2 consisted of the participant level, with 75 participants.
We entered our main predictor of interest: cue condition in Level 1, with 2 conditions:
whether the reward cue or neutral cue preceded a trial (coded as 1 and 0, respectively). We
controlled for counterbalancing of color to outcome (Level 2) and counterbalancing of cue to
trial (Level 2). Our model included participant-level random intercepts. Our dependent variable
was the choice to share or keep the entire endowment (coded as 1 and 0, respectively).

84

The model equation is listed below:
Level 1:
P ( 𝑠 ℎ 𝑎 𝑟 𝑒 𝑖𝑗
= 1 ) = 𝑙 𝑜 𝑔𝑖𝑠𝑡 𝑖 𝑐 ( 𝛽 0 𝑗 + 𝛽 1 𝑗 cu e
𝑖𝑗
) + 𝑒 𝑖𝑗

Level 2:
𝛽 0 𝑗 = 𝛾 00
+ 𝛾 01
cou nt erba lance
𝑖𝑗
+ 𝛾 02
or d er
𝑖𝑗
+ 𝑢 0 𝑗
𝛽 1 𝑗 = 𝛾 10
+ 𝑢 1 𝑗
The intraclass correlation (ICC) for the participant level was calculated. Participant-level
ICC was 0.891, indicating that 89.1% of the variance in the data was due to differences between
participants.
Our results showed that approval cues did not affect generosity, B = 0.001, p = 0.992,
95% CI [-0.245, 0.247]. Results are shown in Appendix D, Table 1. There were no significant
effects of the counterbalancing variables.
Interaction of Cue Condition with Personality Variables
We then added interactions between cue and mean-centered extraversion, agreeableness,
and conscientiousness scores to the model, while removing the counterbalancing variables to
avoid convergence issues. While we did not find any main effects of the personality variables,
we did find a significant interaction of cue with conscientiousness, B = 0.449, p = 0.011, 95% CI
[0.102, 0.796], such that more conscientious participants responded with greater generosity to
approval cues (see Appendix D, Figure 2 for visualization of the interaction). The other
interactions were not significant (see Appendix D, Table 2, for results).
Discussion
Our primary hypothesis predicted that participants exposed to cues associated with social
approval would act more generously. If the hypothesis were confirmed, our results would show

85

that Pavlovian cues can influence altruistic behavior. We also explored whether personality
variables related to social approval would strengthen the effect of approval cues on generosity.
We based this exploration on the expectation that participants who cared more about reputation,
and hence scored higher on these variables, would be more sensitive to approval cues. While we
did find an interaction between cues and conscientiousness, we did not find support for our
primary hypothesis.
Study Limitations
While it is unclear why approval cues did not influence generosity, there are several
possibilities. One possibility is that social approval simply does not matter in our dictator game.
Participants acted anonymously in the game without any prospect of additional interactions with
their partners. This means that the reputations of the participants were not truly at stake.
Approval cues may be more relevant to games that involve repeated interactions with the same
partner, or ones that make participant decisions visible to other participants.
Another possibility is that decisions in the dictator game were not sensitive enough to
contextual cues. The intraclass correlations suggest that a large majority of the variance in the
data was due to differences between participants, rather than differences between trials. Although
we attempted to encourage participants to respond differently between trials by varying different
parameters, participants overall did not differ much in their responses much from trial to trial.
Since cue presentation is intended to change behavior between trials, the dictator game may not
have been an appropriate task to use for testing the effects of Pavlovian cues.
It is also possible that images of a thumbs-up gesture are not a clear and powerful enough
reward. While a thumbs-up does signify approval in American culture, the decontextualized
presentation of a thumbs-up during Pavlovian learning may have been too ambiguous in

86

meaning. Elements of approval, such as who is the approver, and what is it that they approve of,
are missing, making our outcome less relevant to the dictator game. Overall, any future iterations
of this study should make major changes to its design, such that the role of reputation is clearer,
the task more sensitive to contextual changes, and the outcome more relevant to the task.
Moreover, as in our dating studies, our study design here departed from prior PIT studies
that tested persistence as opposed to choice (Hogarth et al., 2015; Talmi et al., 2008; Watson et
al., 2014). Persistence may be more sensitive to the motivational strength of a Pavlovian cue and
may provide greater statistical power than binary choices. One solution is to use an effort-based
test of PIT. For example, participants could be asked to press a button repeatedly to benefit
another participant.
Finally, the dictator game may measure fairness in additional to generosity. It is possible
that some participants perceived the task in terms of splitting resources fairly, rather than
generous giving. In this case, the role of social approval is more ambiguous, since “sharing” is
restoring resources to their rightful owners rather than providing extra resources to benefit
others. In future studies, we plan to use tasks that are more clearly related to generosity.
Conscientiousness and Social Approval
Our findings showed that conscientious people were more sensitive to approval cues in
our study. We were somewhat surprised, as we expected agreeableness to be more relevant to
approval cues because of its emphasis on being cooperative and altruistic (John & Srivastava,
1999). It is possible that, because of its brevity (2 items per factor), the short-form Big Five
Inventory was not reliable enough to find interactions with agreeableness. Follow-up studies
should use the full version of the Big Five Inventory (John, Donahue, & Kentle, 1991) to

87

examine interactions with Big Five traits more closely. Moreover, 75 participants may not have
been sufficient to find an interaction with an individual difference variable.
While more work is needed to replicate and understand the interaction with
conscientiousness, we considered two possible explanations for the interaction. First,
conscientiousness may be related to social approval insofar as the approval is a reward for norm
and rule adherence. Indeed, Furnham and Coveney (1996) found that conscientiousness had the
highest positive correlation with need for social approval within a customer service context. If
we conceptualize the dictator game not as a test of generosity, but as a test of fairness, the
interaction may make more sense. Participants high in conscientiousness may see sharing on the
dictator game as a fairer division of resources, and approval cues as an indicator for the
opportunity to perform well in adhering to fairness norms. Hence, approval cues triggered in
conscientious participants a motivation to be fairer. This account would also predict a main
effect of conscientiousness, but unfortunately, our study was likely underpowered to find a
relationship between an individual difference measure and a binary dependent variable.
The other explanation is that conscientious participants, in the hopes of conforming to the
expectations of an authority figure (i.e. the experimenter), responded to the cues in ways that
they perceived were expected. However, one could also argue that conscientious participants
were more honest in their responses than other participants, by following the instructions to
ignore the cues more closely. Future versions of this study could ask participants to report
whether they knew the hypothesis or not and exclude those participants.
Conclusion
We found only very weak evidence to support our predictions that Pavlovian cues for
social approval can increase generosity. Our results show no main effect of approval cues, but

88

also suggest that approval cues may influence generosity among conscientious participants.
Overall, the results are inconclusive, and the study’s design requires improvement before we can
pursue this our research question again.

89

Concluding Remarks
The past three chapters explored different possibilities for integrating our understanding
of Pavlovian-instrumental transfer with our understanding of social behavior. The presented
studies saw varying degrees of success.
Chapter 2 discussed the possibility of using cues associated with attractiveness to
influence dating behavior. This could potentially shed light on the nature of contextual cues that
influence romantic attraction. Initial results were promising, indicating that cues associated with
attractive faces led participants to be more willing to date hypothetical targets. Moreover,
relationship status moderated this effect, as the effects of the cues were stronger among those
who were single. However, subsequent follow-up studies could not replicate the findings,
suggesting that changes in the stimuli used may have weakened our manipulation.
Chapter 3 discussed the potential role of cultural values in determining how people
respond to Pavlovian cues. To the extent that cued outcomes are culturally relevant, cultural
values may strengthen or weaken cue effects. Our aim was to situate the role of Pavlovian cues
within theoretical frameworks emphasizing the influence of contextual factors in activating
culture-based motivation. Results supported our predictions; in Studies 1 and 2, people who
endorse honor cultural values respond more strongly to cues associated with interpersonal threat.
Moreover, by experimentally manipulating the accessibility of honor values in Study 3, we also
showed that the role of honor values may be causal.
Chapter 4 investigated the possibility of using cues associated with social approval to
influence altruistic behavior. However, we found no evidence for the direct impact of Pavlovian
cues in an initial study. Although, we did find that conscientiousness interacted with approval
cues to influence generosity; approval cues increased generosity among participants who were

90

highly conscientious. Nevertheless, more work is needed to replicate and understand this effect.
For example, future studies should use a more reliable measure of the Big Five traits, since the
short-form inventory we used only contained two items for each factor. Moreover, disadvantages
in the study’s design, such as the potential ambiguity of the outcome used in conditioning and
the measurement of choice as opposed to persistence, should be addressed in future attempts.
Interestingly, all three of our projects showed some evidence that individual differences –
life experience, cultural values, and personality – may interact with Pavlovian cues to influence
behavior. In our dating studies, we found that the effects of attractiveness cues were stronger
among single people. In the honor studies, we found that the effects of threat cues were stronger
among people who endorsed honor values more strongly. In our generosity study, we found that
the effects of approval cues were stronger among people who were more conscientious. These
results suggest that higher-order cognitions, such as attitudes and beliefs, may determine the
salience of a motivational outcome, thereby influencing sensitivity to a given Pavlovian cue.
Addressing Study Design Problems
Our studies met varying levels of success, which may be due to variations in aspects of
study design. We list three potential factors that may have obscured Pavlovian cue effects: the
measurement of choice as opposed to persistence, the ambiguity of outcomes used in learning,
and the clarity of Pavlovian contingencies. In Appendix E, Table 1, we summarize the presence
of these factors in each of our three projects.
The measurement of choice as opposed to persistence as our dependent variable in our
studies may be one reason why we could not find Pavlovian cue effects. We focused on choice in
the dating and generosity studies because salient real-world examples of dating and generous
behavior often involve choice rather than effort. However, persistence is widely measured in

91

existing studies on Pavlovian-instrumental transfer (Hogarth et al., 2015; Talmi et al., 2008;
Wang & Read, 2022; Watson et al., 2014). Persistence may be more sensitive to cue effects,
since participants can exert effort proportionate to the strength of the cue, as opposed to choosing
between binary options. Finally, measures of persistence, such as frequency of key-pressing,
allow analyses of a continuous dependent variable, providing more statistical power than
analyses of a binary variable. In our studies, we find that our project on honor, which showed the
most promising results, was the only project that tested cue effects on persistence as opposed to
choice. We plan to redesign our other studies to measure persistence as opposed to choice.
The ambiguity of outcomes used in Pavlovian learning may also be relevant. Our project
on honor used clear, unambiguous outcomes (images of wartime enemies). Our project on dating
used potentially ambiguous outcomes, since the perceived attractiveness of a face may vary
between perceivers. Our project on generosity may also have used an ambiguous outcome –
while a thumbs-up may signal social approval, a thumbs-up devoid of context may lack the
specific meaning necessary to understand it as indicating social approval of altruistic behavior.
The absence of clear, unambiguous outcomes means that cue effects may be weaker or that
participants may fail to learn the motivational relevance of the cue.
Related to this, the clarity of Pavlovian contingencies may be another issue in study
design. While the generosity and honor studies saw high rates of contingency awareness, with
large proportions of participants correctly reporting the Pavlovian contingencies, the dating
studies saw low rates of awareness. This suggests that participants in the dating studies were not
able to learn the associations between cue and outcome. The faces we used may have varied on
other dimensions in addition to attractiveness, making it difficult for participants to perceive
attractiveness as the main distinguishing feature between reward and non-reward trials. Another

92

possibility is that participants varied in their judgments of attractiveness – the same face could be
more or less attractive to different people. Thus, the difference between reward and non-reward
trials would be less salient to participants who found the reward faces to be less attractive or the
neutral faces to be more attractive, making the cue-outcome contingency unclear. Future studies
should make Pavlovian contingencies easier to learn by, for example, increasing the visual
salience of the cues used or using less ambiguous outcomes.
Pavlovian Cues and Social Behavior
Overall, we are still in the initial stages of testing whether Pavlovian cues can be used to
influence social behavior. One of the greatest challenges of this endeavor is establishing a valid
and reliable paradigm to use. Social outcomes are complex, ambiguous, and subject to divergent
interpretations from different people. Even basic social rewards such as approval and
attractiveness are difficult to implement without careful consideration of extensive set of factors
that determine their meaning. Yet, as we saw in the studies on honor culture, this is also the
value of situating Pavlovian cues within the realm of social behavior. By understanding the role
of meaning-making and interpretation in cue-triggered behavior, we can understand how the
broader, big-picture factors, like cultural ecology or moral belief, are concretized into the
tangible elements of everyday situations.
Although our findings were mixed, our studies do provide initial evidence that Pavlovian
cues may be involved in social behavior. Our initial findings from the dating studies suggest that
attractiveness cues may influence dating decisions, and our findings throughout the honor studies
suggest that cultural values may determine the strength of Pavlovian threat cues. These results
have powerful implications for our understanding of the role of Pavlovian cues. Pavlovian cues
may not only influence pursuit of outcomes such as drugs, food, and money, but also higher-

93

order outcomes such as social approval and ingroup protection. The findings also provide a
mechanistic explanation of how context influences social action. As elements in the physical
environment, Pavlovian cues can trigger motives to pursue social goals, essentially acting as goal
primes. While our present findings require some further investigation, future work can explore
how Pavlovian cues coordinate with other mechanisms, such as perspective-taking, lay theories,
social norms, and other factors that might strengthen, weaken, or mediate the effects of
Pavlovian cues on social behavior. Together, work examining the influence of cues on social
behavior can provide a more complete understanding of social motivation and supplement
existing theories about how social goals are triggered by context.

94

References
Aarts, H., Custers, R., & Veltkamp, M. (2008). Goal priming and the affective-motivational
route to nonconscious goal pursuit. Social Cognition, 26(5), 555-577.

Adams, G. R., & Marshall, S. K. (1996). A developmental social psychology of identity:
Understanding the person-in-context. Journal of Adolescence, 19(5), 429-442.

Adkins-Regan, E., & MacKillop, E. A. (2003). Japanese quail (Coturnix japonica) inseminations
are more likely to fertilize eggs in a context predicting mating opportunities. Proceedings
of the Royal Society of London. Series B: Biological Sciences, 270, 1685-1689.

Adolphs, R. (2003). Cognitive neuroscience of human social behaviour. Nature Reviews
Neuroscience, 4(3), 165-178.

Ahrens, L. M., Pauli, P., Reif, A., Mühlberger, A., Langs, G., Aalderink, T., & Wieser, M. J.
(2016). Fear conditioning and stimulus generalization in patients with social anxiety
disorder. Journal of Anxiety Disorders, 44, 36-46.

Anderson, E. (1994, May). The code of the streets. The Atlantic.
https://www.theatlantic.com/magazine/archive/1994/05/the-code-of-the-streets/306601/

Bainbridge, W.A., Isola, P., & Oliva, A. (2013). The intrinsic memorability of face
images. Journal of Experimental Psychology: General. Journal of Experimental
Psychology: General, 142(4), 1323-1334.

Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and
Economic Behavior, 10(1), 122-142.

Berridge, K. C. (2007). The debate over dopamine's role in reward: The case for incentive
salience. Psychopharmacology, 191(3), 391-431.

Berscheid, E., Dion, K., Walster, E., & Walster, G. W. (1971). Physical attractiveness and dating
choice: A test of the matching hypothesis. Journal of Experimental Social Psychology,
7(2), 173-189.

Caballo, V. E., Salazar, I. C., Irurtia, M. J., Arias, B., & Hofmann, S. G. (2012). The
multidimensional nature and multicultural validity of a new measure of social anxiety:
The Social Anxiety Questionnaire for Adults. Behavior Therapy, 43(2), 313-328.

Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). The efficient assessment of need for
cognition. Journal of Personality Assessment, 48(3), 306-307.

Cao, Y., Enke, B., Falk, A., Giuliano, P., & Nunn, N. (2021). Herding, Warfare, and a Culture of
Honor: Global Evidence (No. w29250). National Bureau of Economic Research.

95

Carpentier, F. D., Knobloch-Westerwick, S., & Blumhoff, A. (2007). Naughty versus nice:
Suggestive pop music influences on perceptions of potential romantic partners. Media
Psychology, 9(1), 1-17.

Carpentier, F. R. D., Northup, C. T., & Parrott, M. S. (2014). Revisiting media priming effects of
sexual depictions: Replication, extension, and consideration of sexual depiction
strength. Media Psychology, 17(1), 34-54.

Cartoni, E., Balleine, B., & Baldassarre, G. (2016). Appetitive Pavlovian-instrumental transfer:
A review. Neuroscience & Biobehavioral Reviews, 71, 829-848.

Cartoni, E., Moretta, T., Puglisi-Allegra, S., Cabib, S., & Baldassarre, G. (2015). The
relationship between specific Pavlovian instrumental transfer and instrumental reward
probability. Frontiers in Psychology, 6, 1697.

Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective
responses to impending reward and punishment: The BIS/BAS scales. Journal of
Personality and Social Psychology, 67(2), 319-333.

Charalambides, N. (2021, August 24). We recently went viral on TikTok – here's what we
learned. Prolific. https://www.prolific.co/blog/we-recently-went-viral-on-tiktok-heres-
what-we-learned

Charpentier, C. J., Bromberg-Martin, E. S., & Sharot, T. (2018). Valuation of knowledge and
ignorance in mesolimbic reward circuitry. Proceedings of the National Academy of
Sciences, 115(31), E7255-E7264.

Cherkasova, M. V., Clark, L., Barton, J. J. S., Schulzer, M., Shafiee, M., Kingstone, A., Stoessl,
A. J., & Winstanley, C. A. (2018). Win-concurrent sensory cues can promote riskier
choice. The Journal of Neuroscience, 38(48), 10362-10370.

Choi, I., Koo, M., & An Choi, J. (2007). Individual differences in analytic versus holistic
thinking. Personality and Social Psychology Bulletin, 33(5), 691-705.

Choi, I., Nisbett, R. E., & Norenzayan, A. (1999). Causal attribution across cultures: Variation
and universality. Psychological Bulletin, 125(1), 47-63.

Cloutier, J., Heatherton, T. F., Whalen, P. J ., & Kelley, W. M. (2008). Are attractive people
rewarding? Sex differences in the neural substrates of facial attractiveness. Journal of
Cognitive Neuroscience, 20(6), 941-951.

Cohen, D., Hernandez, I., Gruschow, K., Nowak, A., Gelfand, M., & Borkowski, W. (2018).
Rationally irrational?: The ecologies and economics of honor. In A. K. Uskul & S. Oishi
(Eds.) Socioeconomic environment and human psychology: Social, ecological, and
cultural perspectives (pp. 77-102). New York: Oxford University Press.

96

Cohen, D., & Leung, A. K.-Y. (2012). Virtue and virility: Governing with honor and the
association or dissociation between martial honor and moral character of US presidents,
legislators, and justices. Social Psychological and Personality Science, 3(2), 162-171.

Cohen, D., & Nisbett, R. E. (1997). Field experiments examining the culture of honor: The role
of institutions in perpetuating norms about violence. Personality and Social Psychology
Bulletin, 23(11), 1188-1199.

Cohen, D., Nisbett, R. E., Bowdle, B. F., & Schwarz, N. (1996). Insult, aggression, and the
southern culture of honor: An “experimental ethnography”. Journal of Personality and
Social Psychology, 70(5), 945-960.

Corbit, L. H., & Balleine, B. W. (2005). Double dissociation of basolateral and central amygdala
lesions on the general and outcome-specific forms of Pavlovian-instrumental transfer.
Journal of Neuroscience, 25(4), 962-970.

Corbit, L. H., & Balleine, B. W. (2011). The general and outcome-specific forms of Pavlovian-
instrumental transfer are differentially mediated by the nucleus accumbens core and shell.
Journal of Neuroscience, 31(33), 11786-11794.

Corr, P. J., & Cooper, A. J. (2016). The Reinforcement Sensitivity Theory of Personality
Questionnaire (RST-PQ): Development and validation. Psychological Assessment,
28(11), 1427-1440.

Dayan, P., & Berridge, K. C. (2014). Model-based and model-free Pavlovian reward learning:
Revaluation, revision, and revelation. Cognitive, Affective & Behavioral Neuroscience,
14(2), 473-492.

Delgado, M. R., Frank, R. H., & Phelps, E. A. (2005). Perceptions of moral character modulate
the neural systems of reward during the trust game. Nature Neuroscience, 8(11), 1611-
1618.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using
G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods,
41, 1149-1160.

Fehr, E., & Fischbacher, U. (2003). The nature of human altruism. Nature, 425, 785-791.

Fehr, E., & Fischbacher, U. (2004). Third-party punishment and social norms. Evolution and
Human Behavior, 25(2), 63-87.

Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. The
Quarterly Journal of Economics, 114(3), 817-868.

97

Feingold, A. (1990). Gender differences in effects of physical attractiveness on romantic
attraction: A comparison across five research paradigms. Journal of Personality and
Social Psychology, 59(5), 981-993.

FeldmanHall, O., Dalgleish, T., Evans, D., & Mobbs, D. (2015). Empathic concern drives costly
altruism. NeuroImage, 105, 347-356.

Fink, B., Neave, N., Manning, J. T., & Grammer, K. (2006). Facial symmetry and judgements of
attractiveness, health and personality. Personality and Individual Differences, 41(3), 491-
499.

Fisher, H. E., Aron, A., Mashek, D., Li, H., & Brown, L. L. (2002). Defining the brain systems
of lust, romantic attraction, and attachment. Archives of Sexual Behavior, 31(5), 413-419.

Flinkenflogel, N., Novin, S., van der Meulen, A., & Krabbendam, L. (2020). Where to draw the
line: Honor mindset increases retaliation in response to unfair behavior. Culture and
Brain, 1-16.

Flynn, F. J. (2003). How much should I give and how often? The effects of generosity and
frequency of favor exchange on social status and productivity. Academy of Management
Journal, 46(5), 539-553.

Friedman, R. S., McCarthy, D. M., Förster, J., & Denzler, M. (2005). Automatic effects of
alcohol cues on sexual attraction. Addiction, 100(5), 672-681.

Furnham, A., & Coveney, R. (1996). Personality and customer service. Psychological Reports,
79(2), 675-681.

Garbusow, M., Nebe, S., Sommer, C., Kuitunen-Paul, S., Sebold, M., Schad, D. J., Friedel, E.,
Veer, I. M., Wittchen, H. U., Rapp, M. A., Ripke, S., Walter, H., Huys, Q. J. M.,
Schlagenhauf, F., Smolka, M. N., & Heinz, A. (2019). Pavlovian-to-instrumental transfer
and alcohol consumption in young male social drinkers: Behavioral, neural and polygenic
correlates. Journal of Clinical Medicine, 8(8), 1188.

Gęsiarz, F., & Crockett, M. J. (2015). Goal-directed, habitual and Pavlovian prosocial behavior.
Frontiers in Behavioral Neuroscience, 9, 135.

Glocker, M. L., Langleben, D. D., Ruparel, K., Loughead, J. W., Gur, R. C., & Sachser, N.
(2009). Baby schema in infant faces induces cuteness perception and motivation for
caretaking in adults. Ethology, 115(3), 257-263.

Graham, J., Nosek, B. A., Haidt, J., Iyer, R., Koleva, S., & Ditto, P. H. (2011). Mapping the
moral domain. Journal of Personality and Social Psychology, 101(2), 366-385.

98

Günsoy, C., Cross, S. E., Uskul, A. K., Adams, G., & Gercek-Swing, B. (2015). Avoid or fight
back? Cultural differences in responses to conflict and the role of collectivism, honor,
and enemy perception. Journal of Cross-Cultural Psychology, 46(8), 1081-1102.

Güth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental analysis of ultimatum
bargaining. Journal of Economic Behavior & Organization, 3(4), 367-388.

Ha, T., Overbeek, G., & Engels, R. C. M. E. (2010). Effects of attractiveness and social status on
dating desire in heterosexual adolescents: An experimental study. Archives of Sexual
Behavior, 39(5), 1063-1071.

Hackel, L. M., & Zaki, J. (2018). Propagation of economic inequality through reciprocity and
reputation. Psychological Science, 29(4), 604-613.

Hackel, L. M., Zaki, J., & Van Bavel, J. J. (2017). Social identity shapes social valuation:
Evidence from prosocial behavior and vicarious reward. Social Cognitive and Affective
Neuroscience, 12(8), 1219-1228.

Hahn, A. C., Fisher, C. I., DeBruine, L. M., & Jones, B. C. (2016). Sex-specificity in the reward
value of facial attractiveness. Archives of Sexual Behavior, 45(4), 871-875.

Harinck, F., Shafa, S., Ellemers, N., & Beersma, B. (2013). The good news about honor culture:
The preference for cooperative conflict management in the absence of
insults. Negotiation and Conflict Management Research, 6(2), 67-78.

Hill, K. (2002). Altruistic cooperation during foraging by the Ache, and the evolved human
predisposition to cooperate. Human Nature, 13, 105-128.

Hogarth, L., Maynard, O. M., & Munafo, M. R. (2015). Plain cigarette packs do not exert
Pavlovian to instrumental transfer of control over tobacco ‐seeking. Addiction, 110(1),
174-182.

Hogarth, L., Retzler, C., Munafo, M. R., Tran, D. M., Troisi II, J. R., Rose, A. K., Jones, A., &
Field, M. (2014). Extinction of cue-evoked drug-seeking relies on degrading hierarchical
instrumental expectancies. Behaviour Research and Therapy, 59, 61-70.

Hollis, K. L. (1984). The biological function of Pavlovian conditioning: The best defense is a
good offense. Journal of Experimental Psychology: Animal Behavior Processes, 10(4),
413-425.

Hugenberg, K., & Wilson, J. P. (2013). Faces are central to social cognition. In D. E. Carlston
(Ed.), The Oxford handbook of social cognition (pp. 1-33). Oxford, UK: Oxford
University Press.

Izuma, K., Saito, D. N., & Sadato, N. (2008). Processing of social and monetary rewards in the
human striatum. Neuron, 58(2), 284-294.

99

Janak, P. H., & Tye, K. M. (2015). From circuits to behaviour in the amygdala. Nature,
517(7534), 284-292.

Jankowiak, W. R., & Fischer, E. F. (1992). A cross-cultural perspective on romantic
love. Ethnology, 31(2), 149-155.

Ji, L.-J., Peng, K., & Nisbett, R. E. (2000). Culture, control, and perception of relationships in
the environment. Journal of Personality and Social Psychology, 78(5), 943-955.

Joel, S., Eastwick, P. W., & Finkel, E. J. (2017). Is romantic desire predictable? Machine
learning applied to initial romantic attraction. Psychological Science, 28(10), 1478-1489.

John, O. P., Donahue, E. M., & Kentle, R. L. (1991). Big Five Inventory (BFI) [Database
record]. APA PsycTests.

John, O. P., & Srivastava, S. (1999). The Big Five Trait taxonomy: History, measurement, and
theoretical perspectives. In L. A. Pervin & O. P. John (Eds.), Handbook of personality:
Theory and research (pp. 102-138). Guilford Press.

Johnson, A. W. (2013). Eating beyond metabolic need: How environmental cues influence
feeding behavior. Trends in Neurosciences, 36(2), 101-109.

Kacen, J. J., & Lee, J. A. (2002). The influence of culture on consumer impulsive buying
behavior. Journal of Consumer Psychology, 12(2), 163-176.

Kalick, S. M., Zebrowitz, L. A., Langlois, J. H., & Johnson, R. M. (1998). Does human facial
attractiveness honestly advertise health? Longitudinal data on an evolutionary
question. Psychological Science, 9(1), 8-13.

Kitayama, S., & Park, J. (2010). Cultural neuroscience of the self: Understanding the social
grounding of the brain. Social Cognitive and Affective Neuroscience, 5(2-3), 111-129.

Klein, J. T., & Platt, M. L. (2013). Social information signaling by neurons in primate striatum.
Current Biology, 23(8), 691-696.

Kohls, G., Perino, M. T., Taylor, J. M., Madva, E. N., Cayless, S. J., Troiani, V., Price, E., Faja,
S., Herrington, J. D., & Schultz, R. T. (2013). The nucleus accumbens is involved in both
the pursuit of social reward and the avoidance of social punishment. Neuropsychologia,
51(11), 2062-2069.

Krach, S., Paulus, F. M., Bodden, M., & Kircher, T. (2010). The rewarding nature of social
interactions. Frontiers in Behavioral Neuroscience, 4, 22.

Langlois, J. H., & Roggman, L. A. (1990). Attractive faces are only average. Psychological
Science, 1(2), 115-121.

100

Lehman, D. R., Chiu, C.-y., & Schaller, M. (2004). Psychology and culture. Annual Review of
Psychology, 55, 689-714.

Lehner, R., Balsters, J. H., Herger, A., Hare, T. A., & Wenderoth, N. (2017). Monetary, food,
and social rewards induce similar Pavlovian-to-instrumental transfer effects. Frontiers in
Behavioral Neuroscience, 10, 247.

Leung, A. K.-Y., & Cohen, D. (2011). Within-and between-culture variation: Individual
differences and the cultural logics of honor, face, and dignity cultures. Journal of
Personality and Social Psychology, 100(3), 507-526.

Liang, X., Zebrowitz, L. A., & Zhang, Y. (2010). Neural activation in the “reward circuit” shows
a nonlinear response to facial attractiveness. Social Neuroscience, 5(3), 320-334.

Luo, S., & Zhang, G. (2009). What leads to romantic attraction: Similarity, reciprocity, security,
or beauty? Evidence from a speed-dating study. Journal of Personality, 77(4), 933-964.

Mahlberg, J., Weidemann, G., Hogarth, L., & Moustafa, A. A. (2019). Cue-elicited craving and
human Pavlovian-to-instrumental transfer. Addiction Research & Theory, 27(6), 482-488.

Manglani, H. R., Lewis, A. H., Wilson, S. J., & Delgado, M. R. (2017). Pavlovian-to-
instrumental transfer of nicotine and food cues in deprived cigarette smokers. Nicotine &
Tobacco Research, 19(6), 670-676.

McElreath, R. (2003). Reputation and the evolution of conflict. Journal of Theoretical Biology,
220(3), 345-357.

Mercer, J. (2013). Introduction: Sex and the celebrity. Celebrity Studies, 4(1), 1-3.

Milinski, M., Semmann, D., & Krambeck, H.-J. (2002). Reputation helps solve the ‘tragedy of
the commons’. Nature, 415, 424-426.

Mischel, W., Shoda, Y., & Mendoza-Denton, R. (2002). Situation-behavior profiles as a locus of
consistency in personality. Current Directions in Psychological Science, 11(2), 50-54.

Morris, M. W., & Peng, K. (1994). Culture and cause: American and Chinese attributions for
social and physical events. Journal of Personality and Social Psychology, 67(6), 949-
971.

Nadler, N., Delgado, M. R., & Delamater, A. R. (2011). Pavlovian to instrumental transfer of
control in a human learning task. Emotion, 11(5), 1112-1123.

Nettle, D., Harper, Z., Kidson, A., Stone, R., Penton-Voak, I. S., & Bateson, M. (2013). The
watching eyes effect in the Dictator Game: It's not how much you give, it's being seen to
give something. Evolution and Human Behavior, 34(1), 35-40.

101

Nisbett, R. E., & Miyamoto, Y. (2005). The influence of culture: Holistic versus analytic
perception. Trends in Cognitive Sciences, 9(10), 467-473.

Nord, C. L., Lawson, R. P., Huys, Q. J., Pilling, S., & Roiser, J. P. (2018). Depression is
associated with enhanced aversive Pavlovian control over instrumental behaviour.
Scientific Reports, 8(1), 1-10.

Novin, S., & Oyserman, D. (2016). Honor as cultural mindset: Activated honor mindset affects
subsequent judgment and attention in mindset-congruent ways. Frontiers in Psychology,
7, 1921.

Nowak, M. A., & Sigmund, K. (1998). Evolution of indirect reciprocity by image
scoring. Nature, 393, 573-577.

Nowak, A., Gelfand, M. J., Borkowski, W., Cohen, D., & Hernandez, I. (2016). The evolutionary
basis of honor cultures. Psychological Science, 27(1), 12-24.

Oishi, S., Diener, E. (2009). Goals, culture, and subjective well-being. In E. Diener (Ed.),
Culture and well-being. Social indicators research series, vol 38 (pp. 93-108). Springer,
Dordrecht.

Oosterhof, N. N., & Todorov, A. (2009). Shared perceptual basis of emotional expressions and
trustworthiness impressions from faces. Emotion, 9(1), 128-133.

Osterman, L. L., & Brown, R. P. (2011). Culture of honor and violence against the
self. Personality and Social Psychology Bulletin, 37(12), 1611-1623.

Oyserman D. 2015. Culture as situated cognition. In R. Scott & S. Kosslyn (Eds.), Emerging
trends in the social and behavioral sciences: An interdisciplinary, searchable, and
linkable resource. Hoboken, NJ: John Wiley & Sons.

Oyserman, D. (2011). Culture as situated cognition: Cultural mindsets, cultural fluency, and
meaning making. European Review of Social Psychology, 22(1), 164-214.

Oyserman, D. (2017). Culture three ways: Culture and subcultures within countries. Annual
Review of Psychology, 68, 435-463.

Papies, E. K. (2016). Goal priming as a situated intervention tool. Current Opinion in
Psychology, 12, 12-16.

Papies, E. K., Potjes, I., Keesman, M., Schwinghammer, S., & van Koningsbruggen, G. M.
(2014). Using health primes to reduce unhealthy snack purchases among overweight
consumers in a grocery store. International Journal of Obesity, 38(4), 597-602.

102

Paredes-Olay, C., Abad, M. J., Gámez, M., & Rosas, J. M. (2002). Transfer of control between
causal predictive judgments and instrumental responding. Animal Learning & Behavior,
30(3), 239-248.

Perrett, D. I., Burt, D. M., Penton-Voak, I. S., Lee, K. J., Rowland, D. A., & Edwards, R. (1999).
Symmetry and human facial attractiveness. Evolution and Human Behavior, 20(5), 295-
307.

Pichon, I., Boccato, G., & Saroglou, V. (2007). Nonconscious influences of religion on
prosociality: A priming study. European Journal of Social Psychology, 37(5), 1032-1045.

Prévost, C., Liljeholm, M., Tyszka, J. M., & O'Doherty, J. P. (2012). Neural correlates of
specific and general Pavlovian-to-instrumental transfer within human amygdalar
subregions: A high-resolution fMRI study. The Journal of Neuroscience, 32(24), 8383-
8390.

Przepiorka, W. & Berger, J. (2017). Signaling theory evolving: Signals and signs of
trustworthiness in social exchange. In B. Jann & W. Przepiorka (Eds.), Social dilemmas,
institutions, and the evolution of cooperation (pp. 373-392). Berlin: De Gruyter.

Rammstedt, B., & John, O. P. (2007). Measuring personality in one minute or less: A 10-item
short version of the Big Five Inventory in English and German. Journal of Research in
Personality, 41(1), 203-212.

Read, S., Brown, A., Wang, P., & Miller, L. (2018). The Virtual Personalities Neural Network
Model: Neurobiological underpinnings. Personality Neuroscience, 1, 10.

Rescorla, R. A. (1991). Associative relations in instrumental learning: The eighteenth Bartlett
memorial lecture. The Quarterly Journal of Experimental Psychology, 43(1), 1-23.

Rhodes, G., Proffitt, F., Grady, J. M., & Sumich, A. (1998). Facial symmetry and the perception
of beauty. Psychonomic Bulletin & Review, 5(4), 659-669.

Rhodes, G., Simmons, L. W., & Peters, M. (2005). Attractiveness and sexual behavior: Does
attractiveness enhance mating success? Evolution and Human Behavior, 26(2), 186-201.

Rhodes, G., Yoshikawa, S., Clark, A., Lee, K., McKay, R., & Akamatsu, S. (2001).
Attractiveness of facial averageness and symmetry in non-Western cultures: In search of
biologically based standards of beauty. Perception, 30(5), 611-625.

Rhodes, G., Zebrowitz, L. A., Clark, A., Kalick, S. M., Hightower, A., & McKay, R. (2001). Do
facial averageness and symmetry signal health? Evolution and Human Behavior, 22(1),
31-46.

Rigdon, M., Ishii, K., Watabe, M., & Kitayama, S. (2009). Minimal social cues in the dictator
game. Journal of Economic Psychology, 30(3), 358-367.

103

Robinson, M. J., & Berridge, K. C. (2013). Instant transformation of learned repulsion into
motivational “wanting”. Current Biology, 23(4), 282-289.

Robinson, T. E., & Berridge, K. C. (2001). Incentive-sensitization and addiction. Addiction,
96(1), 103-114.

Robinson, T. N., Borzekowski, D. L. G., Matheson, D. M., & Kraemer, H. C. (2007). Effects of
fast food branding on young children's taste preferences. Archives of Pediatrics &
Adolescent Medicine, 161(8), 792-797.

Ruff, C. C., & Fehr, E. (2014). The neurobiology of rewards and values in social decision
making. Nature Reviews Neuroscience, 15(8), 549-562.

Russell, D. W. (1996). UCLA Loneliness Scale (Version 3): Reliability, validity, and factor
structure. Journal of Personality Assessment, 66(1), 20-40.

Salamone, J. D., Correa, M., Farrar, A., & Mingote, S. M. (2007). Effort-related functions of
nucleus accumbens dopamine and associated forebrain circuits. Psychopharmacology,
191(3), 461-482.

Salgado, S., & Kaplitt, M. G. (2015). The nucleus accumbens: A comprehensive review.
Stereotactic and Functional Neurosurgery, 93(2), 75-93.

Scheele, D., Wille, A., Kendrick, K. M., Stoffel-Wagner, B., Becker, B., Güntürkün, O., Maier,
W., & Hurlemann, R. (2013). Oxytocin enhances brain reward system responses in men
viewing the face of their female partner. Proceedings of the National Academy of
Sciences, 110(50), 20308-20313.

Schultz, W. (2022). Dopamine reward prediction error coding. Dialogues in Clinical
Neuroscience, 18(1), 23-32.

Schultz, W. (2017). Electrophysiological correlates of reward processing in dopamine neurons.
In J.-C. Dreher & L. Tremblay (Eds.), Decision neuroscience: An integrative perspective
(pp. 21-31). Elsevier Academic Press.

Seabrooke, T., Hogarth, L., & Mitchell, C. J. (2016). The propositional basis of cue-controlled
reward seeking. Quarterly Journal of Experimental Psychology, 69(12), 2452-2470.

Seymour, B., & Dolan, R. (2008). Emotion, decision making, and the amygdala. Neuron, 58(5),
662-671.

Shackelford, T. K. (2005). An evolutionary psychological perspective on cultures of honor.
Evolutionary Psychology, 3, 381-391.

Silvera, D. H., & Austad, B. (2004). Factors predicting the effectiveness of celebrity
endorsement advertisements. European Journal of Marketing, 38(11-12), 1509-1526.

104

Smith, E. R., & Semin, G. R. (2004). Socially situated cognition: Cognition in its social context.
In M. P. Zanna (Ed.), Advances in experimental social psychology, vol. 36 (pp. 53-117).
Elsevier Academic Press.

Sosis, R. (2005). Does religion promote trust?: The role of signaling, reputation, and
punishment. Interdisciplinary Journal of Research on Religion, 1, 7.

Sosis, R., & Bressler, E. R. (2003). Cooperation and commune longevity: A test of the costly
signaling theory of religion. Cross-Cultural Research: The Journal of Comparative
Social Science, 37(2), 211-239.

Stoet, G. (2010). PsyToolkit – A software package for programming psychological experiments
using Linux. Behavior Research Methods, 42(4), 1096-1104.

Stoet, G. (2017). PsyToolkit: A novel web-based method for running online questionnaires and
reaction-time experiments. Teaching of Psychology, 44(1), 24-31.

Talmi, D., Seymour, B., Dayan, P., & Dolan, R. J. (2008). Human Pavlovian-instrumental
transfer. The Journal of Neuroscience, 28(2), 360-368.

Thornhill, R., & Gangestad, S. W. (1999). Facial attractiveness. Trends in Cognitive Sciences,
3(12), 452-460.

Thornhill, R., & Gangestad, S. W. (1993). Human facial beauty: Averageness, symmetry, and
parasite resistance. Human Nature, 4(3), 237-269.

Tooby, J., & Cosmides, L. (1990). On the universality of human nature and the uniqueness of the
individual: The role of genetics and adaptation. Journal of Personality, 58(1), 17-67.

Triandis, H. C. (2001). Individualism-collectivism and personality. Journal of Personality, 69(6),
907-924.

Triandis, H. C., Bontempo, R., Villareal, M. J., Asai, M., & Lucca, N. (1988). Individualism and
collectivism: Cross-cultural perspectives on self-ingroup relationships. Journal of
Personality and Social Psychology, 54(2), 323-338.

Uskul, A. K., & Cross, S. E. (2020). Socio-ecological roots of cultures of honor. Current
Opinion in Psychology, 32, 177-180.

Vail, K. E. III, & Motyl, M. (2010). Support for diplomacy: Peacemaking and militarism as a
unidimensional correlate of social, environmental, and political attitudes. Peace and
Conflict: Journal of Peace Psychology, 16(1), 29-57.

Walster, E., Aronson, V., Abrahams, D., & Rottman, L. (1966). Importance of physical
attractiveness in dating behavior. Journal of Personality and Social Psychology, 4(5),
508-516.

105

Wang, P., Atari, M., & Oyserman, D. (2022). Who can I count on: Honor and self-reliance
during the COVID-19 pandemic. Manuscript in preparation.

Wang, P., & Read, S. J. (2022). Pavlovian cues boost self-regulation: Predicting success with
conditioned stimuli. Manuscript in preparation.

Watkins, C. D., DeBruine, L. M, Little, A. C., Feinberg, D. R., & Jones, B. C. (2012). Priming
concerns about pathogen threat versus resource scarcity: Dissociable effects on women’s
perceptions of men’s attractiveness and dominance. Behavioral Ecology and
Sociobiology, 66(12), 1549-1556.

Watson, P., Wiers, R. W., Hommel, B., & de Wit, S. (2014). Working for food you don’t desire.
Cues interfere with goal-directed food-seeking. Appetite, 79, 139-148.

Waynforth, D. (1998). Fluctuating asymmetry and human male life-history traits in rural
Belize. Proceedings of the Royal Society of London. Series B: Biological
Sciences, 265(1405), 1497-1501.

Weeden, J., & Sabini, J. (2005). Physical attractiveness and health in Western societies: A
review. Psychological Bulletin, 131(5), 635-653.

Weiser, K. (2020, September). The code of the West. Legends of America.
https://www.legendsofamerica.com/we-codewest/

Weismueller, J., Harrigan, P., Wang, S., & Soutar, G. N. (2020). Influencer endorsements: How
advertising disclosure and source credibility affect consumer purchase intention on social
media. Australasian Marketing Journal, 28(4), 160-170.

Wiggert, N., Wilhelm, F. H., Boger, S., Georgii, C., Klimesch, W., & Blechert, J. (2017). Social
Pavlovian conditioning: Short- and long-term effects and the role of anxiety and
depressive symptoms. Social Cognitive and Affective Neuroscience, 12(2), 329-339.

Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin & Review, 9(4),
625-636.

Zack, M., George, R. S., & Clark, L. (2020). Dopaminergic signaling of uncertainty and the
aetiology of gambling addiction. Progress in Neuro-Psychopharmacology & Biological
Psychiatry, 99, 109853.

Zebrowitz, L. A., & Montepare, J. M. (1992). Impressions of babyfaced individuals across the
life-span. Developmental Psychology, 28(6), 1143-1152.

Zsila, Á., Orosz, G., McCutcheon, L. E., & Demetrovics, Z. (2020). A lethal imitation game?
Exploring links among psychoactive substance use, self-harming behaviors and celebrity
worship. Addictive Behaviors Reports, 12, 100319.

106

Appendices
Appendix A: Chapter 1 Tables and Figures
Figure 1
Model of Specific PIT

107

Figure 2
Model of General PIT

108

Appendix B: Chapter 2 Tables and Figures
Figure 1
Message Border Cues

109

Figure 2
Interaction of Cue and Relationship Status (Study 1)

Note. The effect of the reward cue on willingness to date was stronger among single people than
among people who were not single.

0
5
10
15
20
25
30
35
Single Not Single
% Trials Choosing to Date
Reward Cue Control Cue

110

Table 1
Generalized Linear Multilevel Model Predicting Willingness to Date (Study 1)
Random Intercepts
Variance SD
Participants 1.514 1.230
Faces 0.159 0.399
Fixed Effects
Predictor B SE p
Intercept -1.092 0.524 0.037*
Cue Condition 0.208 0.083 0.012*
Counterbalance (Conditioning) -0.292 0.236 0.216
Counterbalance (Task) -0.012 0.236 0.961
*p < 0.05, **p< 0.01, ***p < 0.001.

111

Table 2
Generalized Linear Multilevel Model Predicting Willingness to Date, Including Interactions with
Relationship Variables (Study 1)
Random Intercepts
Variance SD
Participants 1.389 1.179
Fixed Effects
Predictor B SE p
Intercept -1.502 0.125 0.000***
Cue Condition 0.228 0.083 0.006**
Importance of Finding Date -0.030 0.126 0.811
Relationship Status Satisfaction -0.096 0.084 0.248
Relationship Status 0.004 0.135 0.977
Cue x Importance -0.066 0.085 0.439
Cue x Satisfaction -0.018 0.057 0.750
Cue x Status -0.192 0.091 0.035*
*p < 0.05, **p< 0.01, ***p < 0.001.

112

Table 3
Generalized Linear Multilevel Model Predicting Willingness to Date (Study 2)
Random Intercepts
Variance SD
Participants 1.533 1.238
Faces 0.133 0.364
Fixed Effects
Predictor B SE p
Intercept -1.114 0.192 0.000***
Cue Condition 0.038 0.068 0.579
Counterbalance (Conditioning) 0.026 0.201 0.897
Counterbalance (Task) -0.322 0.201 0.110
*p < 0.05, **p< 0.01, ***p < 0.001.

113

Table 4
Generalized Linear Multilevel Model Predicting Willingness to Date, Including Interactions with
Relationship Variables (Study 2)
Random Intercepts
Variance SD
Participants 1.512 1.230
Faces 0.133 0.364
Fixed Effects
Predictor B SE p
Intercept -1.116 0.191 0.000***
Cue Condition 0.037 0.069 0.587
Counterbalance (Conditioning) 0.007 0.201 0.974
Counterbalance (Task) -0.288 0.201 0.153
Relationship Status -0.139 0.107 0.194
Cue x Status -0.034 0.069 0.620
*p < 0.05, **p< 0.01, ***p < 0.001.

114

Table 5
Generalized Linear Multilevel Model Predicting Willingness to Date (Study 3)
Random Intercepts
Variance SD
Participants 1.899 1.378
Faces 0.069 0.263
Fixed Effects
Predictor B SE p
Intercept -1.829 0.219 0.000***
Cue Condition 0.096 0.083 0.244
Counterbalance (Conditioning) -0.485 0.244 0.047*
Counterbalance (Task) 0.151 0.244 0.536
*p < 0.05, **p< 0.01, ***p < 0.001.

115

Table 6
Generalized Linear Multilevel Model Predicting Willingness to Date, Including Interaction with
Relationship Status (Study 3)
Random Intercepts
Variance SD
Participants 1.898 1.378
Faces 0.069 0.263
Fixed Effects
Predictor B SE p
Intercept -1.718 0.266 0.000***
Cue Condition 0.086 0.120 0.476
Counterbalance (Conditioning) -0.506 0.246 0.039*
Counterbalance (Task) 0.148 0.244 0.543
Relationship Status -0.188 0.260 0.470
Cue x Status 0.020 0.166 0.903
*p < 0.05, **p< 0.01, ***p < 0.001.

116

Table 7
Generalized Linear Multilevel Model Predicting Willingness to Date (Study 4)
Random Intercepts
Variance SD
Participants 2.231 1.494
Faces 0.813 0.902
Fixed Effects
Predictor B SE p
Intercept -0.512 0.301 0.089
Cue Condition 0.087 0.087 0.322
Counterbalance (Conditioning) -0.338 0.278 0.225
Counterbalance (Task) -0.670 0.279 0.016*
*p < 0.05, **p< 0.01, ***p < 0.001.

117

Table 8
Generalized Linear Multilevel Model Predicting Willingness to Date, Including Interaction with
Relationship Status (Study 4)
Random Intercepts
Variance SD
Participants 2.146 1.465
Faces 0.808 0.899
Fixed Effects
Predictor B SE p
Intercept -0.459 0.344 0.181
Cue Condition 0.211 0.140 0.132
Counterbalance (Conditioning) -0.228 0.277 0.409
Counterbalance (Task) -0.573 0.276 0.038*
Relationship Status -0.223 0.299 0.455
Cue x Status -0.202 0.180 0.261
*p < 0.05, **p< 0.01, ***p < 0.001.

118

Appendix C: Chapter 3 Tables and Figures
Figure 1
Sample Instrumental Conditioning Trial

119

Figure 2
Sample PIT Trial

120

Figure 3
Interaction of Honor Values with Threat Cue in Predicting Total Keypresses (Study 1)

Note. Note that predictive intervals, such as the ones shown here, are generally wider than
confidence intervals.

121

Figure 4
Interaction of Honor Values with Threat Cue in Predicting Total Keypresses (Study 2)

Note. Note that predictive intervals, such as the ones shown here, are generally wider than
confidence intervals.

122

Figure 5
Interaction of Honor Priming with Threat Cue in Predicting Total Keypresses (Study 3)

Note. Note that predictive intervals, such as the ones shown here, are generally wider than
confidence intervals.

123

Table 1
Generalized Multilevel Poisson Model Predicting Total Keypresses with Zero-Inflation Model
(Study 1)
Random Intercepts
Poisson Model
Variance SD
Participants 0.597 0.773
Zero-Inflation Model
Variance SD
Participants 3.655 1.912
Fixed Effects
Poisson Model
Predictor B SE p
Intercept 1.571 0.075 0.000***
Threat Cue 0.475 0.030 0.000***
Honor Values -0.176 0.099 0.075
Threat Cue x Honor 0.230 0.039 0.000***
Zero-Inflation Model
Predictor B SE p
Intercept 2.291 0.209 0.000***
Threat Cue -6.302 0.209 0.000***
Honor Values 0.002 0.265 0.995
Threat Cue x Honor 0.003 0.231 0.988
*p < 0.05, **p< 0.01, ***p < 0.001.

124

Table 2
Generalized Multilevel Poisson Model Predicting Total Keypresses with Zero-Inflation Model
(Study 2)
Random Intercepts
Poisson Model
Variance SD
Participants 0.871 0.933
Zero-Inflation Model
Variance SD
Participants 6.463 2.542
Fixed Effects
Poisson Model
Predictor B SE p
Intercept 1.286 0.087 0.000***
Threat Cue 0.480 0.034 0.000***
Honor Values -0.378 0.089 0.000***
Threat Cue x Honor 0.381 0.033 0.000***
Zero-Inflation Model
Predictor B SE p
Intercept 2.224 0.254 0.000***
Threat Cue -6.506 0.227 0.000***
Honor Values -0.224 0.256 0.382
Threat Cue x Honor 0.254 0.208 0.222
*p < 0.05, **p< 0.01, ***p < 0.001.

125

Table 3
Generalized Multilevel Poisson Model Predicting Total Keypresses with Zero-Inflation Model
(Study 3)
Random Intercepts
Poisson Model
Variance SD
Participants 0.858 0.926
Zero-Inflation Model
Variance SD
Participants 6.213 2.493
Fixed Effects
Poisson Model
Predictor B SE p
Intercept 1.469 0.084 0.000***
Threat Cue 0.425 0.027 0.000***
Honor Values 0.152 0.101 0.135
Threat Cue x Honor -0.040 0.029 0.170
Zero-Inflation Model
Predictor B SE p
Intercept 2.353 0.250 0.000***
Threat Cue -6.917 0.254 0.000***
Honor Values -0.374 0.300 0.213
Threat Cue x Honor -0.282 0.280 0.313
*p < 0.05, **p< 0.01, ***p < 0.001.

126

Table 4
Generalized Multilevel Poisson Model with Zero-Inflation Model Predicting Total Keypresses
from the Interaction of Honor Priming and Threat Cue (Study 3)
Random Intercepts
Poisson Model
Variance SD
Participants 0.910 0.954
Zero-Inflation Model
Variance SD
Participants 6.056 2.461
Fixed Effects
Poisson Model
Predictor B SE p
Intercept 1.743 0.122 0.000***
Threat Cue 0.323 0.032 0.000***
Honor Priming -0.640 0.173 0.000***
Threat Cue x Honor Priming -0.278 0.057 0.000***
Zero-Inflation Model
Predictor B SE p
Intercept 1.934 0.345 0.000***
Threat Cue -6.503 0.305 0.000***
Honor Priming 0.804 0.509 0.114
Threat Cue x Honor Priming -0.795 0.465 0.087
*p < 0.05, **p< 0.01, ***p < 0.001.

127

Appendix D: Chapter 4 Tables and Figures
Figure 1
Sample Dictator Game Trial

128

Figure 2
Predicted Probabilities of Sharing with Interaction of Cue Condition with Conscientiousness as
Predictors

Note. As conscientiousness increases, approval cues lead to a greater increase in the predicted
probability of sharing. Note that predictive intervals, such as the ones shown here, are generally
wider than confidence intervals.

129

Table 1
Generalized Linear Multilevel Model Predicting Generosity
Random Intercepts
Variance SD
Participants 26.276 5.126
Fixed Effects
Predictor B SE p
Intercept 0.788 1.076 0.464
Cue Condition 0.001 0.126 0.992
Counterbalance (Conditioning) -1.581 1.231 0.199
Counterbalance (Game) -0.727 1.233 0.555
*p < 0.05, **p< 0.01, ***p < 0.001.

130

Table 2
Generalized Linear Multilevel Model Predicting Generosity with Personality Variables Added
Random Intercepts
Variance SD
Participants 25.790 5.078
Fixed Effects
Predictor B SE p
Intercept -0.496 0.666 0.457
Cue Condition -0.072 0.138 0.598
Agreeableness 0.856 0.638 0.179
Extraversion -0.791 0.669 0.237
Conscientiousness 1.417 0.896 0.114
Cue x Agreeableness -0.245 0.130 0.060
Cue x Extraversion -0.296 0.156 0.057
Cue x Conscientiousness 0.449 0.177 0.011*
*p < 0.05, **p< 0.01, ***p < 0.001.

131

Appendix E: Concluding Remarks Tables and Figures
Table 1
Summary of Study Design Differences Relevant to the Presence of Supporting Evidence
Project Persistence or
Choice
Ambiguity of
Outcomes
Awareness of
Contingencies
Supportive
Evidence
Attractiveness
Cues and Dating
Choice Ambiguous Low awareness Initial evidence;
failures to
replicate
Honor and Threat
Cues
Persistence Unambiguous High awareness Consistent
support
Approval Cues
and Generosity
Choice Ambiguous High awareness No supporting
evidence

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Using Pavlovian cues to boost self-control

PDF

Examining Pavlovian-to-instrumental transfer effect in human mate selection behavior

PDF

Dopamine dependent: examining the link between learning and treatment-resistant depression

PDF

How misinformation exploits moral values and framing: insights from social media platforms and behavioral experiments

PDF

Generalization of social rejection in social networks

PDF

Beyond active and passive social media use: habit mechanisms are behind frequent posting and scrolling on Twitter/X

PDF

Neural and behavioral correlates of fear processing in first-time fathers

PDF

Testosterone’s effect on reward, cognition, and myelination: rat brain and behavior

PDF

Endocrinology of social behavior and reward

PDF

Facebook habits: rewards, cues, and automaticity

PDF

Neural and behavioral mechanisms of oxytocin in the regulation of food intake

PDF

Homeostatic imbalance and monetary delay discounting: effects of feeding on RT, choice, and brain response

PDF

Reward substitution: how consumers can be incentivized to choose smaller food portions

PDF

The motivated affective behavior system: a dynamic account of the attachment behavioral system

PDF

Social exclusion decreases risk-taking

PDF

Effect of mindfulness training on attention, emotion-regulation and risk-taking in adolescence

PDF

Reward immediacy and subjective stress modulate anticipation of primary and secondary rewards in temporarily-abstinent cigarette smokers

PDF

The acute impact of glucose and sucralose on food decisions and brain responses to visual food cues

PDF

A self-knowledge model of social inference

PDF

“You move, therefore I am”: the combinatorial impact of kinesthetic motion cues on social perception

Asset Metadata

Creator Wang, Peter (author)

Core Title Investigating the effects of Pavlovian cues on social behavior

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Degree Conferral Date 2022-12

Publication Date 09/09/2022

Defense Date 08/10/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag attraction,culture,generosity,Learning,Motivation,OAI-PMH Harvest,reward

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Read, Stephen J. (committee chair), Buechel, Eva (committee member), Hackel, Leor (committee member), Monterosso, John (committee member)

Creator Email peterpwa@usc.edu,pwang203@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111939846

Unique identifier UC111939846

Legacy Identifier etd-WangPeter-11197

Document Type Dissertation

Format application/pdf (imt)

Rights Wang, Peter

Type texts

Source 20220909-usctheses-batch-979 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu