University of Southern California Dissertations and Theses

Hard feelings: friendship, betrayal, and outrage in international relations

(USC Thesis Other)

Hard feelings: friendship, betrayal, and outrage in international relations

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content HARD FEELINGS:
FRIENDSHIP, BETRAYAL, AND OUTRAGE IN INTERNATIONAL RELATIONS
by
Pongkwan Sawasdipakdi
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
(POLITICAL SCIENCE AND INTERNATIONAL RELATIONS)
August 2024
Copyright 2024 Pongkwan Sawasdipakdi

Acknowledgments
My interest in the topic of this dissertation began with a term paper for the seminar in
International Relations and Foreign Policy with Professor Patrick James. Motivated by my
experiences in a small country, Thailand, I became curious about how such nations acheive
their goals despite facing constraints imposed by more powerful countries. During an office
hours meeting with Pat, I was confronted with a question for which I had no immediate
answer: “What case are you going work on?” In a spontaneous response, I suggested,
“Perhaps I could explore how Thailand escaped harsh punishments and obtained a favorable
deal from the Allies after World War II, despite algining with Japan durign the conflict.”
As I delved more into the case, I was struck by the differing responses Thailand faced
from the United States and Great Britain, both being parts of the Allies. At the core of
these differences was the reaction to Thailand’s declaration of war on the United States
and Great Britain: while Great Britain responded in kind, the United States chose to ignore
Thailand’s war declaration. Struck by these differentiations, I pondered whether there might
be insights to be gleaned from comparing how Thailand managed to escape such a response
from the United States while failing to do so with the British.
During the spring break of 2019, I hopped on a plane to Washington D.C., where I visited
the archives at the Library of Congress and took a shuttle bus to the U.S. National Archives at
College Park, Maryland, despite staying at my friend’s apartment in Virginia. In the summer
of the same year, I flew to New York City to visit the Franklin D. Roosevelt Presidential
Library, not realizing that it was actually in Hyde Park, not in New York City. Too broke to
stay at a hotel in Hyde Park, I decided to continue staying at my friend’s apartment in NYC
and began a train ride to the FDR Library at 5:00 a.m. every day for a week. From these
trips, I took hundreds of pictures of U.S. diplomatic communications related to Thailand
during World War II in the hope of finding answers to my lingering questions.
However, the more I delved into these episodes of events, the more I began to question
whether Thailand truly managed to secure a favorable deal with the United States. I started
to ponder whether the British reaction should be treated as a baseline, with that of the
Americans seen as a deviation, or vice versa. The more I scrutinized these documents, the
more I wondered why the British reacted so vehemently to Thailand’s declaration of war
and alliance with Japan. What was it about the mentions of friendship and betrayal that
seemed to permeate British private diplomatic documents everywhere?
ii

I took this puzzle to Professor Brian Rathbun, my advisor and academic hero, who
has been tremendously helpful in shaping the success of this dissertation. He suggested
that I delve deeper into moral psychology and the theory of moral foundations, which has
profoundly influenced my understanding of friendship, betrayal, and outrages in internatioanl
relations up to the completion of this dissertation and beyond, as I embark on my first book
project. Brian has also been a meticulous advisor, who thoroughly read every chapter and
paper draft I sent him and provided detailed comments and solutions for how I should address
those issues. His advisory style and attentiveness to my work inspired me to be an advisor like
him to my future students, although I know I will only be able to match half of his intelligence.
During the writing of my dissertation, I was grateful to receive support from other dissertation committee members, professors, family, partner, and friends, and thankfully, without
experiencing any betrayal in between. Notably, I thank Professor Jacques Hymans and
Professor Daphna Oyserman, two of my dissertation committee members, who also closely
reviewed my dissertation prospectus and its final product, providing valuable comments
and advice on how to further improve it. During my qualifying exam oral defense, I felt
nervous when Proffessor Hymans questioned why I had not incorporated primary British
diplomatic documents into my earlier draft. Thankfully, his question was not intended to
fail me but to suggest the availability of the British diplomatic documents about Thailand
during World War II at the closest location – the University of California, Los Angeles.
This greatly enriched my case study chapter.
I thank Professor David Kang, Dr. Gloria Koo, Jay Lee, and other staff at the University of Southern California Korean Studies Institute (KSI) for their support, intellectually,
financially, and mentally. The funding provided by KSI played a crucial role in making the
experiments in this dissertation possible, and I am sincerely thankful for their assistance. I
am especially appreciative of Professor Kang’s kind invitations to his annual Thanksgiving
and Lunar New Year Parties, where I always felt welcomed despite not being his direct
advisee. These gatherings provided me with a sense of belonging and community, particularly
during times when I was far away from home. Moreover, the close-knit environment at
KSI fostered a culture of collaboration and support, where I felt comfortable sharing my
incomplete work, seeking feedback, and asking for help with issues I deemed unresolvable.
The individuals I had privilage to know and spend time with at KSI – Nayoung Lee,
Kyuri Kim, Alex Lin, Jackie Wong, Han-Hui Hsieh, Noelle Troutman, Miriam Barnum,
Caleb Pomeroy, Amoz Hor, Gaea Morales, Joshua Leung, Yuji Idomoto, Ciara Sterbenz,
and Sabreena Croteau – will always hold a special place in my heart. Even if one day they
forget me, I will force my friendship upon them. In particular, Nayoung, despite our paths
growing apart, will forever remain my wifey, my most sincere friend, and the individual with
whom I shared countless memorable moments in Los Angeles with. As for Alex, let me
iii

borrow his own words from his dissertation acknowledgment, “[Alex Lin] and I remained
good friends even after [Alex] left LA, and we talked deeply about work and life.”
I also wish to extend my gratitute to Professor Saori Katada, Matt Stevens, and Linda
Kim at the Center for International Studies (CIS) for providing me with the opportunity
to have my drafts read at their working paper series. The comments, both harsh and soft,
I received from the workshop were invaluable and greatly contributed to the improvement
of my dissertation chapters and papers. I must also knowledge the financial support from
CIS, which enabled me to conduct the earlier versions of my survey experiment. Without their assistance, I would not have been able to identify the issues with the survey
and rectify them in time.
Above all, my heartfelt thanks must be extended to the Political Science and Internatinal
Relations (POIR) department at the University of Southern California for being my supportive home department for seven years and for providing every possible support to complete
this dissertation and my studies. My archival research in College Park, Hyde Park, and
Bangkok would not have been possible without their funding. I also thank Danielle Ballard,
Karin Amundsen, and Verianda Chavarin for their constant reminders of administrative
tasks, despite my absent-mindedness.
For seven years, I have been grateful to be surrounded by individuals with great personalities, who consistently listened to my complaints, offered comfort when needed, and
sometimes gently, sometimes not so gently, scolded me when I became “too much.” At POIR,
I am thankful for the opportunity to get to know and befriend with the most empathetic
person, Victoria Chonn Ching, the ever-travelling Dat Nguyen, and friends from my cohort –
Edward Gonzalez, Joey Saraceno, Anne Van Wijk, Steffi Neumeier, Kyle Reed, Ayana Best,
Jarred Cuella, and Nayoung Lee, whom I have already mentioned above.
Beyond USC, I had the absolute pleasure of meeting a group of politically like-minded
individuals dedicated to strengthening democracy in our respective countries. What began as political activities evolved into genuine friendship. I have cherished spending time
with Eakpravee “Tum” Pattanaeakpinyo, Eins Achiraya, Phakpicha “Aom” Matartananchai,
Nopposon “Chain” Kokilawatee, Pakpoom “Pai” Buabthong, and Jatuporn “Toy” Leksut.
Beyond LA, I fondly remember all the political discussions, personal consultations, and unwavering support from Chalida “Ooh” Chararnsuk, Thiti “Jam” Jamkajornkeiat, Cholpisit “Ice”
Kiattisewee, Kamin “Bell” Manatpon, Sanpak “Pat” Kuhacharoen, and Nalin Sindhuprama.
In addition to the Thai community, I had the opportunity to work with other Asian diasporas, including Ken Wu, Charles Lam, Jerry Raburn, Alfred Tun, and Ethan Myint. Without these people, I would not have been able to relieve the stress of feeling helpless in supporting democracy fighters at home. Their friendship and support have been invaluable to me.
With the limited space here, I want to extend my heartfelt thanks to Tanangarn “Punch”
Soiongart and Tossapol “Dodge” Nopsuwanchai for travelling all the way from Chicago and
iv

San Francisco to experience the vibrancy of LA and help me fulfill the guest list for my
graduation day. A special thinks to Punch, who expected a Europe-like train experience and
took the Amtrak all the way from Chicago, enduring two technical disruptions that delayed
her arrival by six hours. I also want to express my deepest gratitude to Thitirat “Ying”
Thipsamritkul for being my virtual shoulder to cry on, despite the physical distance between
us. When it comes to my mental health, my deepest appreciation goes to my therapist, Darrin
Dennis, for listening to me, comforting me, and equipping me with the skills to cope with
emotional difficulties and stress, allowing me to graduate with my sanity (or close to it) intact.
I must mention Kamin “Army” Wongpracha, my other half and moral supporter, who
has been with me throughout the second half of my Ph.D. journey. He has been an endless
source of food and snacks, making the conclusion of this dissertation possible. Without him,
I might have finished this dissertation and graduated a year ago. But such a derailment is
not always a bad thing. I am glad we have shared this journey together, and I cherish every
moment with him. Without him, life would be boring and colorless.
Most importantly, I must express my heartfelt gratitude to my mom and dad for their
unwavering emotional support during times of hardship, for comforting me at night when
I needed it most, and for their infinite patience with the slow pace of my PhD studies. I
am profoundly thankful for their unconditional support and understanding, even when it
meant being apart from them for extended periods due to my studies. I am truly blessed
to be their daughter. This dissertation is dedicated to them as a token of appreciation for
how well they have raised and groomed me.
v

Table of contents
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1 Theorizing Betrayal in International Relations . . . . . . . . . . . . . . . 7
Chapter 2 Is Friendship in IR a Myth? . . . . . . . . . . . . . . . . . . . . . . . . . 28
Chapter 3 Fractured Ties: Betrayal’s Impacts on Trust and Cooperation . . . . . . 66
Chapter 4 How Dare You? The British Reactions to Thailand Declaring War against
the Allies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Chapter 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Appendix A: Survey Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Appendix B: Survey Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
vi

List of Tables
2.1 Nearest neighbors to “friend” and “ally” . . . . . . . . . . . . . . . . . . . . 37
2.2 Loyalty, Betrayal, Fairness, and Cheating Words from the extended Morality
Foundation Dictionary (eMFD) . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.3 Rationality and Emotion Dimension . . . . . . . . . . . . . . . . . . . . . . 45
2.4 Rationality words within each moral domain . . . . . . . . . . . . . . . . . . 48
2.5 Emotion words within each moral domain . . . . . . . . . . . . . . . . . . . 49
3.1 Randomly assigned vignettes varying the relationship between the United
States and Country A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2 The expected outcomes of the pre-test measurements in Scenario A . . . . . 79
3.3 Randomly assigned vignettes varying Country A’s response to the U.S. request
in ScenarioA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.4 The expected outcomes of the post-test measurements in Scenario A . . . . . 86
3.5 Randomly assigned vignettes varying the relationship between the United
States and Country B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.6 The expected outcomes of the pre-test measurements in scenario B . . . . . 91
3.7 Randomly assigned vignettes varying Country B’s response to the U.S. request
in scenario B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.8 The expected outcomes of the post-test measurements in Scenario B . . . . . 95
3.9 (Scenario A) The effects of prior relationship and Country A’s decision on
the reductions of trust, positive attitude, and willingness to collaborate with
Country A in the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.10 (Scenario B) The effects of prior relationship and Country B’s decision on
the reductions of trust, positive attitude, and willingness to collaborate with
Country B in the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.11 The logistic models showing the effect of prior relations and Country A’s and
B’s decision on respondents’ willingness to punish Country A and B . . . . . 115
3.12 (Scenario A) The estimated mediation effects on the reduction of trust and
willingness to collaborate with Country A from bootstrap samples using Generalized Linear Models (GLM) with the relationship control group as the
reference group (90% confidence level) . . . . . . . . . . . . . . . . . . . . . 125
vii

3.13 (Scenario B) The estimated mediation effects on the reduction of trust and
willingness to collaborate with Country B from bootstrap samples using Generalized Linear Models (GLM) with the relationship control group as the
reference group (90% confidence level) . . . . . . . . . . . . . . . . . . . . . 126
3.14 (Scenario A) The estimated mediation effects on the inclination to punish
Country A from bootstrap samples using Generalized Linear Models (GLM)
with the relationship control group as the reference group (90% confidence
level) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
3.15 (Scenario B) The estimated mediation effects on the inclination to punish
Country B from bootstrap samples using Generalized Linear Models (GLM)
with the relationship control group as the reference group (90% confidence
level) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
B1 Basic demographics of the survey sample compared to the U.S. Census data 191
viii

List of Figures
2.1 Projecting fairness words onto the rationality/emotion dimension with rationality on the left pole and emotion on the right pole . . . . . . . . . . . . . 46
2.2 Projecting cheating words onto the rationality/emotion dimension with rationality on the left pole and emotion on the right pole . . . . . . . . . . . . . 46
2.3 Projecting loyalty words onto the rationality/emotion dimension with rationality on the left pole and emotion on the right pole . . . . . . . . . . . . . 47
2.4 Projecting betrayal words onto the rationality/emotion dimension with rationality on the left pole and emotion on the right pole . . . . . . . . . . . . . 47
2.5 Projecting rational fairness words onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole . . . . . . . . . . . . . . . . . 52
2.6 Projecting emotional fairness words onto the relationship dimension with
“ally” on the left pole and “friend” on the right pole . . . . . . . . . . . . . . 52
2.7 Projecting rational cheating words onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole . . . . . . . . . . . . . . . . . 53
2.8 Projecting emotional cheating words onto the relationship dimension with
“ally” on the left pole and “friend” on the right pole . . . . . . . . . . . . . . 53
2.9 Projecting rational loyalty words onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole . . . . . . . . . . . . . . . . . 55
2.10 Projecting emotional loyalty words onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole . . . . . . . . . . . . . . . . . 55
2.11 Projecting rational betrayal words onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole . . . . . . . . . . . . . . . . . 56
2.12 Projecting emotional betrayal words onto the relationship dimension with
“ally” on the left pole and “friend” on the right pole . . . . . . . . . . . . . . 56
2.13 The percentage of words that are located closer to either the friend or ally
poles, categorized by the morality domain to which they belong . . . . . . . 58
2.14 The projection of rational fairness terms (left) and emotional fairness terms
(right) onto a relationship dimension in FRUS and quotidian texts, where the
top represents the “friend” pole and the bottom represents the “ally” pole . 59
ix

2.15 The projection of rational cheating terms (left) and emotional cheating terms
(right) onto a relationship dimension in FRUS and quotidian texts, where the
top represents the “friend” pole and the bottom represents the “ally” pole . 60
2.16 The projection of rational loyalty terms (left) and emotional loyalty terms
(right) onto a relationship dimension in FRUS and quotidian texts, where the
top represents the “friend” pole and the bottom represents the “ally” pole . 61
2.17 The projection of rational loyalty terms (left) and emotional loyalty terms
(right) onto a relationship dimension in FRUS and quotidian texts, where the
top represents the “friend” pole and the bottom represents the “ally” pole . 62
2.18 The projection of rational loyalty terms (left) and emotional loyalty terms
(right) onto a relationship dimension in FRUS and quotidian texts, where the
top represents the “friend” pole and the bottom represents the “ally” pole . 63
3.1 The 3x3 factoral design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2 The variations on the two dimensions . . . . . . . . . . . . . . . . . . . . . . 81
3.3 The extent to which respondents perceive Country A/B as a friend of the
United States, rated on a scale of 0 (lowest) to 5 (highest) . . . . . . . . . . 101
3.4 The extent to which respondents trust Country A/B, rated on a scale of 0
(lowest) to 5 (highest) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.5 The extent to which respondents feel positive towards Country A/B, rated on
a scale of 0 (lowest) to 5 (highest) . . . . . . . . . . . . . . . . . . . . . . . . 104
3.6 The extent to which respondents desire the United States to continue collaborating with Country A/B, rated on a scale of 0 (lowest) to 5 (highest) . . . 105
3.7 The shift in respondents’ trust towards Country A/B after learning about its
response to the U.S. requests, rated on a scale of 0 (lowest) to 5 (highest) . . 108
3.8 Scenario A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.9 Scenario B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.10 The shift in respondents’ desire for the United States to continue collaborating
with Country A/B after learning about its decision, rated on a scale of 0
(lowest) to 5 (highest) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.11 The percentage of respondents who were willing to punish Country A/B after
learning about its decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.12 Primary purpose of punishing Country A/B (as a percentage) . . . . . . . . 119
3.13 Respondents’ reported emotions when learning about Country A/B’s decision,
rated on a scale of 0 (lowest) to 5 (highest). . . . . . . . . . . . . . . . . . . 121
B1 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
B2 Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
B3 Race . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
x

B4 Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
B5 Ideology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
B6 Party affiliation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
B7 Party leaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
xi

Abstract
Is there such a thing as betrayal in international relations? Conventional wisdom suggests that because interstate relations are transactional, betrayaland the feelings and outcomes associated with itdoes not have a significant place in international politics. In this
realm, there are no permanent friends, only permanent interests. Therefore, emotional
bonds that lead to emotional reactions when a country reneges on a commitment or defects
are thought not to exist.
However, in this dissertation, I argue that betrayal is indeed significant in international
relations because friendship plays a crucial role in this realm. Due to friendship, countries,
including their policymakers and the public, apply the moral codes of friendship to their
relationships with other states. Just like in interpersonal relationships, countries expect
friends to remain loyal and refrain from actions that violate the relationship’s rules. Specifically, they expect friends not to abandon or betray them despite having the ability to do so.
Because of these moral expectations, when a friend reneges on a commitment, defects, or
abandons, it is perceived as betrayal. Betrayal is judged as morally wrong, leading to strong
negative emotional reactions, punitive actions, and the termination of the relationship.
This dissertation utilizes a multi-method approach, comprising word embeddings analysis,
a survey experiment, and a case study of British-Thai relations during the Second World War,
compared to the shadow case of U.S.-Thai relations during the same period. The findings
suggest that international relations are not solely governed by self-interests but are deeply
intertwined with emotional and moral considerations. This research opens new avenues for
understanding state behavior, emphasizing the need to consider both rational and emotional
factors in explaining interstate relations and international politics.
xii

Introduction
Is there such a thing as betrayal in international relations? From the Vietnam War to
the conflict in Afghanistan, the term “betrayal” has been used to depict instances where
the United States withdrew its troops from territories it had pledged to defend, leaving its
allies feeling abandoned (for example, see Gamson, 2021; Packer, 2022; Young, 2023; Shaw
and Schall, 2015). However, labeling these occurrences as betrayal seems contradictory to
conventional wisdom in the field. Without a central authority in the international system,
states—the primary actors—must prioritize self-interest and national security for survival
(Waltz, 1979). With this goal in mind, switching sides or abandoning others is seen as a
viable, if not preferable, option for survival when the configuration of power distribution
in the international system shifts. Being too committed to other parties for ideological,
cultural, or moral reasons can spell doom for a country. In international relations, it is
often said that there are no permanent friends, only permanent interests, implying that
there are really no friends at all.
However, if betrayal is a reasonable practice in international politics, why do countries
outcry when they feel betrayed by another country? Should policymakers not interpret these
incidents as reflections of the other country’s changing national interests, especially when
the seemingly betrayal action does not shift the balance of power out of their favor?
1

Unfortunately, reality often falls short of this expectation. A recent example is the creation of a new security pact between the United States, the United Kingdom, and Australia
(AUKUS), which led Canberra to withdraw a multibillion-dollar submarines contract with
France. French Foreign Minister Jean-Yves Le Drian described the move as a “stab in
the back” (Dalton, 2011). While Australia’s decision reflects its evolving strategic interests,
France viewed it as a betrayal of a long-standing friendship. The French did not merely voice
their discontent; they took action by recalling ambassadors from Canberra and Washington,
as well as canceling a gala at their embassy in Washington and on their battleship in
Baltimore in protest against what they termed a “brutal, unilateral, and unpredictable
decision” (Roger and Shear, 2021).
This incident is not an anomaly. Politicians and the media frequently utilize the term
“betrayal” to characterize perceived breaches of commitment between states. However,
despite its widespread usage, there exists a notable lack of clear understanding regarding
the precise meaning of “betrayal” in the context of international relations and its potential
ramifications. This dissertation aims to bridge this gap by examining the relevance and
significance of betrayal in international relations, despite the seemingly paradoxical nature
of applying such an emotionally and morally charged concept in a domain thought to be
driven by rationality and self-interset.
This dissertation also represents an original contribution to the field. It delves into a
novel concept in international politics that has received scant attention. To my knowledge,
there is only one existing work on betrayal, conducted by Eznack (Eznack, 2012). In
contrast to her argument, which emphasizes the ability of countries that consider themselves
2

friends to mend the damage of betrayal out of affection, I contend that there are instances
where betrayal leaves a lasting and difficult-to-repair mark on the relationship between two
countries. Particularly when two countries share deep ties and are perceived by policymakers
as friends, the pain from being betrayed by a friend leaves a scar that damages the willingness
of the two countries to collaborate again. For example, after Thailand had joined an alliance
with Japan during World War II, Thailand was no longer a friend of Great Britain but an
opportunist deserving of punishment and unworthy of trust.
Structure of the Dissertation
This dissertation adopts a multi-methodological approach, incorporating both quantitative
and qualitative research methods to investigate the central research question: Does the
concept of betrayal, particularly within the context of perceived friendship between states,
carry significant meaning in international relations?
The next chapter (Chapter 1) delves into the theoretical underpinnings of betrayal in
international relations. It examines how various schools of thought, particularly realism
and rationalism, approach this complex phenomenon. The chapter explores the conventional understanding of betrayal and its consequences within these frameworks, followed by
counterarguments that emphasize the significance of morality and normative expectations
in international relations. Additionally, this chapter introduces the concept of friendship
in international relations and how it engenders the expectation of loyalty, thereby making
betrayal a potentially costly action in the realm of international politics.
Chapter 2 lays the groundwork for subsequent analysis by investigating whether the
3

concept of friendship holds significance in international relations because without friendship,
there is no betrayal. It employs a novel approach based on empirical evidence. While existing
studies on friendship in this field often rely on philosophical interpretations or rhetorical
analyses of speeches and treaties by state leaders (for example, see Oelsner, 2007; Oelsner and
Koschut, 2014; Roščin, 2017; Berenskoetter, 2007), a systematic exploration of the concept
itself is lacking. This chapter aims to fill this gap by analyzing extensive collections of
private discussions involving policymakers from the United States. Additionally, it examines
the language used by the general public in everyday discourse. Through the use of word
embeddings analysis, the chapter investigates how policymakers deploy the term “friend”
relative to “ally” within the discourse of international relations. This analysis seeks to
illuminate the underlying assumptions and expectations associated with friendship versus
alliance in foreign policy discourse.
Chapter 3 delves into the causal impact of betrayal in international relations through
a survey experiment involving the American public. This method offers the possibility of
precisely identifying the presence and impact of a sense of betrayal. And since relationships
of friendship are bigger than simple relationships along leaders, the public is a relevant site to
explore the phenomenon. The experiment manipulates scenarios depicting betrayal by countries with varying degrees of closeness, such as a regular partner with diplomatic relations, an
ally, or a friend. By examining how perceptions of betrayal influence foreign policy decisions
in the controlled setting, the chapter aims to elucidate causal relationships that might be
challenging to establish through real-world observations alone due to their complexity.
The most significant consequence of betrayal is observed in the friend condition, wherein
4

respondents exhibited the greatest decrease in trust, positive attitude, and willingness to
cooperate with the friend compared to those in the ally and control groups. The findings
further revealed that anger partially mediated the inclination to punish, while feelings of hurt
partially mediated the decrease in the willingness to cooperate with the other country in the
future. Tellingly, the feeling of hurt was more strongly correlated with friends than alliances.
This suggests that friends are more likely to experience hurt when betrayed by another friend,
which has implications for their trust and willingness to maintain the friendly relationship.
Chapter 4 applies the theoretical concept to a real-world case by examining the relationship between Thailand and Great Britain, as well as between Thailand and the United States
during World War II (1939 to 1945). Despite having signed non-aggression agreements with
France, Great Britain, and Japan, Thailand, with informal commitments to aid the Allies,
entered into an offensive and defensive pact with Japan on December 21, 1941, and declared
war on Great Britain and the United States on January 25, 1942. However, the responses
from the United States and Great Britain towards the Thai government differed significantly.
British policymakers and diplomats, perceiving the Thai government as a friend of the
British Empire, responded vehemently to Thailand’s decision to switch sides. This resulted
in punitive measures that persisted until the end of the war and significantly reshaped
perceptions of Thailand. In contrast, American policymakers, who regarded Thailand as a
distant country, were more inclined to formulate their policy responses based on strategic
needs and self-interests. Once again, friendship engenders an expectation of loyalty, which
elicits emotional reactions when that loyalty commitment is violated.
The conclusion chapter Chapter 5 of this dissertation will explore various avenues for
5

extending and building upon the original research presented in the preceding chapters. After
all, this dissertation has not addressed all aspects of betrayal in international relations.
Questions remain, such as how countries would react if a betraying nation were to apologize
and provide explanations for their actions, how smaller countries with limited capacity would
fare if betrayed by stronger nations, what roles ideologies and generalized trust play in
the reaction to an act of betrayal, and so on. Overall, this dissertation paves the way
for a more advanced understanding of the complex dynamics of betrayal, friendship, and
loyalty in the realm of global politics.
6

Chapter 1
Theorizing Betrayal in International Relations
Betrayal, trust violation, and breaches of loyalty are ubiquitous aspects of human interaction. These phenomena typically trigger a cascade of responses in interpersonal relations,
ranging from vengeance (Bies and Tripp, 1996) to profound emotional upheavals such as
betrayal, anger, and hopelessness (Robinson and Rousseau, 1994; Turnley and Feldman,
1998; Kramer and Tyler, 2001). Despite the significant impact of betrayal in personal
contexts, its relevance in international relations is debatable. The field of IR, characterized
by rational calculations and self-interest, seemingly marginalizes the emotional connotations
associated with betrayal.
7

This chapter begins by examining the concept of betrayal as it is portrayed within the
mainstream International Relations literature, which is predominantly shaped by classical
realism, structural realism, and rationalism. According to these theoretical frameworks,
betrayal—particularly the emotional reactions and consequences typically associated with
it—holds little relevance in the realm of international politics.
Realists argue that since states prioritize self-help, excessive focus on interstate relationships that could lead to feelings of betrayal is dangerous. Instead, realists contend
that violations of commitments are viewed as routine elements of international relations,
significant only if they impact the balance of power within the system. They further assert
that states should not only expect but also engage in such betrayals if necessary. From the
rationalist perspective, a state’s failure to honor commitments is interpreted merely as an
indication of its true preferences, which is pragmatically assessed when making decisions,
devoid of any emotional considerations.
As this chapter progresses, I argue that the presence of moral imperatives in international
relations necessitates a reconsideration of betrayal not merely as a revealed preference or a
common practice devoid of emotional significance. I present the concept of friendship in
international relations, positing that where there is friendship, there inevitably arises the
potential for betrayal. While realists may view non-assistance or switching of allegiances as
standard practices in international relations, I contend that the existence of friendship creates
expectations that make acts of self-help emotionally painful. When countries perceive others
as “friends,” they expect support; the absence of such support can provoke moral judgment,
leading to feelings of surprise, hurt, and anger. Such emotional responses can result in a
8

reduction in trust, termination of relationships, or the imposition of penalties.
The Irrelevance of Betrayal in IR
The conventional wisdom in the field of International Relations, therefore, implies that
betrayal, or at least the feelings and outcomes associated with it, do not have a significant
place in international relations. This does not imply that incidents of betrayal, such as
the reneging of commitments, are absent in international relations. To suggest so would do
injustice to the history of international politics. On the contrary, defection is a common
feature of international politics, to the extent that it should not trigger feelings of “betrayal”
and any emotional, disproportionate, and long-lasting responses to it.
The prevalence of betrayal in international politics can be attributed, in part, to the
inherent nature of anarchy in the international system. Unlike domestic politics, which
typically operate under a central authority, the international arena lacks a governing body
to enforce rules or settle disputes (Waltz, 1979; Mearsheimer, 1994). In this environment,
states rely on self-help mechanisms for their own survival. Consequently, realpolitik, which
often involves reneging on commitments made to others, is considered an inevitable, if not
rational, course of action. In the realm of politics, especially international politics, acts of
deceit, manipulation, and betrayal are often deemed acceptable and exempt from the same
moral scrutiny applied to private matters (Morgenthau, 1945).
The same logic applies to states other than our own. Just as we can renege on commitments made to others, so too can other states violate commitments made to us. This reality
underscores the nature of international politics, where self-interest often takes precedence
9

over allegiance to agreements. Consequently, states should remain vigilant, recognizing
that betrayal by other states is a constant possibility and can lead to severe consequences,
including the potential for devastating military defeats (Mearsheimer, 1994), 19]. To mitigate
the costs of betrayal, which could result in destruction for a country, states must prioritize
their military capability, thereby enhancing their capacity to withstand betrayals by others.
Placing excessive emphasis on relationships with other nations and reacting to their decisions,
particularly when those decisions are not directly tied to shifts in relative power, distracts
state leaders from the core principles of international politics.
While realists advocate for countries to amass sufficient military capacity to defend themselves against instances of betrayal and to betray others when deemed necessary, rationalists
suggest that betrayal can be viewed as a piece of information about the other country’s
genuine preferences regarding their foreign policy. According to the rationalist explanation,
one of the central bargaining problems stems from the lack of information and uncertainty
about the preferences of others, which arises from the propensity for state leaders to bluff
or lie (Fearon, 1995, 1998). In this context, a breach of commitment or an act of betrayal
should not be interpreted solely as “betrayal” but rather as a disclosure of another party’s
true preferences. It serves as a signal that other states should consider before making any
decisions. State leaders should be grateful that the true colors of the other state are revealed
before it becomes difficult to resolve.
Indeed, states may sometimes rationally react to commitment violations through punishment or negative reciprocity. However, for these reactions to be deemed rational, they
must serve a purpose beyond mere retribution. These purposes, primarily, aim to maintain
10

the order of the international system or uphold the general rule of cooperation. As noted
by Lang, “Punishment differs from vengeance because it is a response to a general rule and
not a single act of harm. . . For an infliction of harm to count as punishment, it must be
intended to support, in some way, a general rule of behavior for a society” (2008, 11).
In the context of international relations, punishment appears to function effectively
in deterrence or compliance contexts. Indeed, punishment and negative reciprocity are
fundamental elements of the theory of cooperation. As exemplified by the tit-for-tat strategy,
where actors mirror the preceding action of the other actor, positive and negative reciprocity
have been proven effective as models of cooperation in international relations (Axelrod, 2009;
Leng and Goodsell, 1974). Nonetheless, it is crucial to ensure that negative reciprocal
action is proportionate to the preceding action committed by the other party. Punishing
excessively or for too long can have detrimental effects on further cooperation, thus being
irrational (Axelrod, 2009).
In summary, while acknowledging the prevalence of commitment violations in international affairs, the prevailing wisdom in International Relations asserts that the notion of
“betrayal,” as understood in interpersonal relationships, has no rightful place in the realm of
international politics. Realists contend that due to the inherent lack of trust in international
affairs, a breach of commitment does not amount to a breach of trust and betrayal as
commonly perceived. Consequently, such breaches should not evoke the hurt, anger, and
fear that often lead to irrational reactions. From the perspective of rationalists, commitment
violations or defections offer valuable insights into the foreign policy preferences of other
states and thus should be interpreted primarily as informative rather than emotive. While
11

incidents of betrayal may warrant punitive measures, such actions are undertaken with the
intention to deter similar behavior by other states and maintain order in the system rather
than driven by emotion or retributive motives.
Embracing Morality in International Relations
However, we now understand that international politics is not simply a realm of amorality,
where state leaders are solely driven by their calculations of costs and benefits. Even
punishments, typically seen as a means to enforce compliance and uphold rules or norms
in international relations, are not devoid of moral considerations. Punishments are often
accompanied by moral condemnation; they are imposed not solely to enforce compliance or
sustain the rule but rather as reactions towards actions perceived as unjust (Welch, 1993)
or unfair (Rathbun et al., 2022; Rathbun, 2023). We hate when others act egoistic. As
Rathbun (2023) argues, “What we generally regard as the most striking manifestations of
the lack of ethics, the use of violence for political purposes and other aspects of ‘power
politics,’ are more often than not the very expression of moral principles, just not the
ones we are used to” (para. 10).
In his book, Rathbun delineates two manifestations of morality in international relations.
Moral condemnation involves moral judgment and punitive measures aimed at curbing others’
excessive egoistic behaviors. It serves as a response to morally bad and selfish actions,
such as cheating and dishonesty, committed by others. Binding moralities, which include
respect for authority and loyalty to the ingroup, work to foster cohesiveness within a group
by prioritizing duties owed to the group over individual needs. One of the most critical
12

components of binding moralities is ingroup loyalty, which requires members to prioritize
their group over other groups and the group’s interests over their own. Under this moral
imperative, individuals are expected to sacrifice their interests for the collective good of
the group (Rathbun, 2023, chapter 2).
Considering the observations of morality in international relations, it is perplexing why
betrayal, with its associated emotions and ramifications, is not recognized as a relevant
concept in the field. If binding moral principles are fundamental in binding the cohesiveness
of groups at the national level, the same requirement for ingroup loyalty should also manifest
at the international level. Of course, some may argue that groups and demands for ingroup
loyalty do not exist beyond borders since states prioritize their own survival above all else.
Nevertheless, research has demonstrated the presence of ingroup and outgroup dynamics,
even at the international level (Alexander et al., 1999; Mercer, 1996).
Existing studies on the ingroup-outgroup distinctions at the international level demonstrate multiple ways in which ingroup identity can be constructed. One way to conceptualize
this is by examining whether countries share a regime type, as proposed in the democratic
peace theory (Hagan, 1994). Another approach is to consider whether the two countries share
cultural, ethnic, and language similarities (Wyszomirski et al., 2003). However, in addition
to focusing on the similarities of domestic characteristics, an ingroup identity can also be constructed based on the relationship between a group of countries, particularly whether these
countries are allies or perceived as aligned with one another. Certainly, these factors often
overlap, and when multiple factors converge, they may form friendship, which represents a
higher level of ingroup identity among states than that observed among allies (Wendt, 1999;
13

Berenskoetter, 2014; Oelsner, 2007; Berenskoetter, 2007). With friendship, there should
come an expectation for ingroup loyalty and also betrayal. Therefore, according to this logic,
betrayal should indeed be considered a relevant concept in international relations.
Friendship in International Relations
Friendship remains an underexplored type of relationship in international relations. Due to
the egoistic nature of international politics, debates persist regarding the extent to which
friendship exists and matters in the realm of international relations. Furthermore, due to the
field’s overwhelming interest in analyzing enemy and rivalry images, this type of relationship
in international relations remains undertheorized. However, research on friendship in international relations does exist. Scholars investigating friendship in international relations, mostly
rooted in the constructivist tradition, contend that friendship between states is not only
conceivable but already evident in specific regions of the world (Wendt, 1999; Deutsch, 1968;
Oelsner and Koschut, 2014; Oelsner, 2007; Berenskoetter, 2007; Haugevik, 2024). Friendships
are not just rhetoric but rather a type of relationship that involves changes in shared identity,
mutual trust, and emotional bonds between states.
Alexander Wendt (1999) conceptualizes friendship as a role structure wherein states
expect each other to abide by two fundamental rules: settling disputes without resorting to
war or the threat of war (the rule of non-violence) and collectively defending any member
whose security is threatened by a third party (the rule of mutual aid) (Wendt, 1999, 298-99).
However, unlike allies, friends adhere to these principles not out of fear of punishment or
potential future abandonment; rather, they act in unity because they perceive themselves
14

as a cohesive entity. This unity is demonstrated through shared characteristics such as
“we-feeling, solidarity, plural subject, common ingroup identity, thinking like a team, [and]
loyalty” (Wendt, 1999, 305). This shared identity gives rise to altruistic behavior, which
is pivotal in understanding collective action in the real world. In this respect, friends
are qualitatively different from allies in their temporal expectations. “Allies engage in
the same basic behavior as friends, but they do not expect their relationship to continue
indefinitely” (Wendt, 1999, 299).
At the core of friendship lies non-reflective trust and generalized reciprocity. As emphasized by Koschut and Oelsner, such close bonds are more likely to form among actors who
share strong ideational and emotional connections, enabling mutual identification and trust
(Oelsner and Koschut, 2014, 14). Unlike the calculative trust that forms the foundation
of transactional relationships like alliances, non-reflective trust does not require regular
reevaluation of the other party’s trustworthiness. Instead, friends “genuinely trust each
other because their relationship is not based on instrumental, rational thought processes
and utility-based cost-benefit calculations, but is manifested as an emotional bond and
moral disposition” (Oelsner and Koschut, 2014, 14).
As demonstrated in Chapter 2, the word embedding analysis of substantial corpora
encompassing both private and public discussions among policymakers indicates a notable
distinction between how policymakers conceptualize friends versus allies. Terms denoting the
assessment of trustworthiness and transactional calculation, such as transparency, incentive,
and information, are positioned more closely to the term “ally.” In contrast, terms reflecting
emotions and interpersonal connections, such as scared, guilty, and disappointed, are situated
15

in proximity with the term “friend.” This analysis underscores that policymakers perceive
friends and allies as distinct relational constructs within foreign policy discourse, akin to
how individuals understand friendship and alliances in their interpersonal relations.
Scholars researching the topic acknowledge the existence of rhetorical friendship. Koschut
and Oelsner (2014) label this as strategic international friendship. In strategic friendships,
actors may label each other as “friends” in political discourse and treaties. However, this
labeling does not necessarily result in significant long-term changes in behavior or mutual
perceptions among these actors. For instance, Roschin (2017) discusses in his book how the
term “friendship” was employed by imperial powers to extend their influence and impose
indirect rule, similar to the way the British fostered their relationship with Thailand as
shown in Chapter 4. However, here, the rule of reciprocity implicit in the term “friendship”
was used to overcome obstacles posed by these local states that may hinder the imperial
powers rather than investing in a genuine reciprocal relationship. For example, in the
case of the British colonization of India, the rule of reciprocity was invoked to limit the
power of local princely states.
Nevertheless, the utilization of the term “friendship” by these imperial powers underscores
the existence of friendship in international relations rather than rejecting it. It demonstrates
a common understanding of how friendship functions and the expectations associated with
such a relationship, including what one may give and receive within it. With their power,
they could have resorted to brute force, but they chose not to. Regardless of whether the
friendship is genuine or rhetorical, once we label the other party as “friend,” we expect
them to adhere to the principles of friendship. We feel disappointed if the friendship is not
16

honored, and we are hurt if the other side calls us friends but does not truly consider us as
such. These instances highlight that friendship comes with certain moral expectations. The
violation of such expectations, such as betrayal, can lead to moral judgment, condemnation,
and even more severe outcomes, such as punishment or the termination of the relationship.
The Question of Unit of Analysis
However, a question persists regarding the appropriate unit of analysis for applying the
concept of friendship among states. After all, states do not possess physical bodies or
emotions. Yet, as Hymans suggests (2010), states are not merely empty vessels; they are
run by emotionally driven individuals within hierarchical organizations. Hymans argues that
“attempts to calculate the national interest objectively are practically doomed in the face
of the daunting complexities of politics at this highest level of social aggregation (Hymans,
2010, 462). Therefore, the people governing these states can perceive other nations and
their leaders as friends, imbuing their interactions with emotions and moral expectations.
This perspective is illustrated by the historical example of British-Thai relations discussed
in Chapter 4, where British diplomats who viewed Thailand through a lens of friendship
initially advocated for a policy of restraint and amity.
Even with these arguments and evidence, questions remain about how interpersonal
relationships at the individual level aggregate to influence national foreign policy. There
are primarily two pathways through which such aggregation might occur. The first pathway
is evident during times of crisis when foreign policymaking demands urgency and decisions
must be made swiftly. Under such circumstances, the policy-making process is less likely
17

to involve lengthy debates among a wide range of stakeholders and is more susceptible to
the influence of individual perceptions and relationships.
In scenarios where time is of the essence, the personal feelings and perceptions of key
decision-makers can significantly shape the direction of national policy. This influence is
particularly evident in cases such as British-Thai and U.S.-Thai relations during the Second
World War. Here, the British policy was largely shaped by the British Minister’s favorable
personal feelings towards the Thai people. In stark contrast, U.S. policy towards Thailand
during the same period was influenced by the American minister’s perception of unpopularity
and disinterest towards Thailand. This demonstrates how individual relationships and
subjective experiences can directly impact how countries interact at the diplomatic level,
ultimately shaping the broader landscape of international relations.
The second pathway involves the development of interpersonal friendships between leaders into broader diplomatic narratives adopted by the entire population. This pathway
involves a transformation of private friendships into public discourse, where the personal
rapport between leaders becomes a model for national attitudes towards another country.
As these personal interactions are mediated through media and official communications,
they can set a tone that encourages the public to view the bilateral relations through a
similar lens of friendship.
Kristin Haugevik’s (2018) exemplifies the second pathway by exploring the dynamic between the United States and Great Britain. She investigates the evolution of the “special relationship” between these two nations, arguing that its foundational elements were significantly
shaped by the personal rapport between President Franklin D. Roosevelt and Prime Minister
18

Winston Churchill. This interpersonal connection gradually evolved into a broader political
discourse as British politicians mentioned and debated the relationship in the parliament.
Despite some domestic criticisms of this relationship, the portrayal of the United States
as a “friend” of Great Britain has remained relatively stable. Its enduring characteristics are
attributed to diplomatic rituals and everyday diplomacy that, in turn, reinforce the close
bonds between the two nations. This dynamic underscores how states, though not living
entities, can indeed experience friendship and express emotional responses to international
events. Furthermore, just as we can gauge friendship between countries by examining public
sentiment, it is also possible to assess the sentiment of betrayal by observing public reactions.
The Rules of Friendship and the Risk of Betrayal
Now that we recognize the role of friendship in international relations, some may ask
what defines friendship and what rules govern it. Scholars have long examined the
essential characteristics of friendship and how they differ from other relationships (Afifi
and Metts, 1998; Argyle and Henderson, 1984). The common consensus among scholars
is that friendships are voluntary, reciprocal, and characterized by mutual care (Silver,
1989; King, 2007). They entail trust, openness, honesty, acceptance, reciprocity, solidarity,
and loyalty (Berenskoetter, 2014; Eznack, 2012). With these characteristics, friendship
transcends mere transactional interactions.
Clark and Mill (2011) categorize friendship within communal relationships, aligning them
with family members and romantic partners, though friendships are often viewed as lower
in the hierarchy of this category. This classification distinguishes them from exchange
19

relationships, which are defined by the norms that regulate the giving and acceptance of
benefits. In exchange relationships, receiving a benefit creates an obligation—or debt—to
reciprocate with a comparable benefit (Clark and Mills, 2011, 233). Essentially, the benefit
in these relationships comes with a price tag.
In communal relationships, reciprocal obligations are not strictly enforced, yet there is
often a “hope” for reciprocation. For example, parents are expected to feed their children
without a strict obligation for the children to feed them in return, though they may “hope”
that their children will reciprocate this care in the future. In friendship, which ranks lower in
the hierarchy of communal relations, this “hope” for reciprocation is more pronounced than
in relationships like those between parents and children. While there is no rigid expectation
for friends to reciprocate exactly—for instance, by spending the exact amount on dinner
that was previously spent on them—there is an understanding that generosity will not
be one-sided indefinitely.
Despite a general agreement on the foundational aspects of friendship, the specific rules
governing this type of relationship remain largely implicit, to the extent that some “may
be discovered or acknowledged only when they are violated” (Eznack, 2012, 33). Argyle
and Henderson (1984) sought to identify these universal rules of friendship by conducting
experiments across various cultures. They identified the most crucial rule of friendship
as defending one another in their absence. This is followed by rules that include sharing
successes, providing emotional support, trusting and confiding in each other, volunteering
help in times of need, and striving to enhance happiness when together, as well as repaying
debts and favors. This last aspect of friendship rules lends support to the argument that
20

friendship, while deeply communal, ranks lower in the hierarchy and sometimes can border
on exchange relationships.
One of the most essential elements in friendship is trust, which is crucial for maintaining
the relationship. However, the type of trust in this context differs significantly from the
notion commonly used by rationalists in the fields of international relations and economics.
Rationalists contend that trust requires constant reassessment of the other party’s trustworthiness and ongoing monitoring of their behaviors. In contrast, non-calculative trust,
as described by Williamson, is marked by an absence of oversight or scrutiny, includes a
tendency to forgive, and encompasses the entire relationship (Williamson, 1993). This type
of trust is profound; it involves trusting the entire character of others so deeply that their
trustworthiness in specific aspects is seldom questioned. For example, when a friend fails to
honor a commitment, it is often attributed to situational factors, a concept demonstrated
in Mercer’s seminal work (1996).
This non-calculative trust is particularly important for friendships as it helps individuals manage the cognitive anxiety that comes with uncertainty, including the potential
for betrayal. However, constantly monitoring a friend’s behaviors and reassessing their
trustworthiness would itself violate contravene the very principles of friendship. Therefore,
humans choose to deepen their trust in their friends, believing that their friends will meet
their relational expectations despite having the capacity to breach them in order to maintain
the relationship. Humans act “as if another will under no condition harm us, or slight
our welfare” (Silver, 1989, 276).
Despite its role in managing anxiety, it is precisely because of non-calculative trust that
21

betrayal remains a possibility. Without monitoring and enforcement mechanisms, one can
only “hope” that friends will reciprocate concerns for well-being and affection and that they
will never break their trust or the bonds of their friendship. With betrayal unimaginable,
when actions that are considered betrayal and breaches of trust occur, they always come as
a surprise, making their impact more shocking and profound.
As demonstrated in Chapter 3 of this dissertation, an experiment testing public responses
to an act of betrayal in international politics revealed that respondents who were exposed to a
friendship prime—making them believe that the hypothetical country they were interacting
with was a friend—expressed heightened feelings of surprise when the country refused to
help upon request or switched sides. This finding illustrates how the high trust inherent
in friendships leads to more intense feelings of surprise when faced with violations of the
relationship’s expectations. The feeling of surprise is often accompanied by other negative
emotions, such as hurt and anger, which may lead to a willingness to impose punitive actions
and terminate the relationship.
The Immorality of Betrayal
The question then becomes, given the unmanageable risk of betrayal and its potentially
damaging consequences, how do people react to betrayal, and how do they cope with it?
While individuals manage their cognitive anxiety in the face of the uncertainty inherent in
these relationships by trusting their friends, society has developed a code of conduct that,
over time, has evolved into a moral code. This code helps govern the relationship between
friends, setting expectations such as loyalty, honesty, and support (Fitness, 2006). When
22

these expectations are not met, the betrayal is not just a personal disappointment but also
a violation of these broader social and moral expectations.
The moral foundations literature identifies ingroup loyalty as one of the five universal
moral foundations (Haidt, 2007). These give foundations include harm/care, fairness/reciprocity, ingroup/loyalty, authority/respect, and purity/sanctity (Haidt and Joseph, 2004,
2008; Haidt and Bjorklund, 2007). These moral foundations have their origins in the
early stages of societal development. Initially created as solutions to social challenges
such as cheating and group competition, these social guidelines have gradually evolved
into core aspects of morality.
It is noteworthy that after the process of moralization, these social solutions do not merely
serve as guidelines to tackle social challenges: They become virtues that must be followed
and vices that must be avoided. Over time, people start to forget the original reasons for
following these social guidelines; they follow them because it is inherently right to do so. In
contrast, failing to adhere to these guidelines is considered sinful. In this context, being loyal
and caring for the well-being of friends is deemed moral, while betrayal is inherently wrong.
The success of these moral foundations in deterring unfavorable behaviors lies in two
mechanisms. First, they instigate feelings of guilt within the would-be violator, prompting
them to rethink their actions to avoid neglecting their sense of morality. Second, it triggers
moral judgment and moral condemnation by the victim and other members of society. Several
experiments in moral psychology found that people people make a moral judgment in a
blink of an eye, without a deliberative reasoning process (Skitka and Mullen, 2002; Skitka,
2002; Skitka and Mullen, 2008; Skitka, 2010; Haidt, 2007). This automatic response is an
23

“evaluative feeling of good-bad or like-dislike (about the actions or character of a person)
[that] appears in consciousness without any awareness of having gone through steps of search,
weighing evidence, or inferring a conclusion” (Haidt, 2007, 1). In this sense, betrayal, if
committed, will be met with strong moral judgment, condemnation, and possible punitive
actions by the victim and others in society.
However, it is noteworthy that betrayal can mean different things in different relational
contexts. Scholars define betrayal as a violation of relational expectations or codes of
conduct of a specific type of relationship (Finkel et al., 2002; Fitness, 2006; Jones et al.,
1997; Joskowicz-Jabloner and Leiser, 2013; Rao and Lee, 2007; Jones et al., 1997). This
variability highlights the complexity of betrayal, as what constitutes a betrayal in one type
of relationship may not be seen as such in another. For example, a child may feel betrayed
if his/her parents present them with bills for the costs of their upbringing, expecting them
to pay back. The feeling of betrayal arises when a communal relationship, which should be
based on love and care, is treated as an exchange for a market-pricing relationship. This
“trade-off taboo” can create a sense of betrayal, as it undermines the foundational principles
of the relationship (Finkel et al., 2002; Fiske and Tetlock, 1997).
Friendship can be more complex than other types of relationships. As discussed earlier, it
is often characterized as a low-ranking communal relationship, bordering those of exchange
relationships. This positioning makes the roles of mutual caring and exchange relevant to
the code of conduct governing friendship. Additionally, there are different levels of closeness
among friends, each requiring different intensities of rule application. Consequently, one
may not realize they have crossed a line until the other person feels wronged by their
24

action. The worst-case scenario occurs when the two parties in the friendship disagree on
the level of closeness, leading them to apply different expectations and codes of conduct to
the relationship. This misalignment can result in misunderstandings and feelings of betrayal,
which complicate the dynamics of the friendship even further.
Given the complexity of the notion of betrayal, this dissertation focuses on the clearcut situation of betrayal—switching sides. An act of siding with a third party instead of a
friend not only violates the expectation of loyalty but denounces the friendship altogether.
It also indicates that the other party in the relationship has been dishonest about how
much they care about the relationship. As Fitness (2006) notes, “Lying is, by definition,
inauthentic communication; thus, if my relationship partner lies to me, I may assume that
he is promoting his own interests over mine and that he cares more about protecting himself
than about caring for me or our relationship” (p. 6).
In Chapter 3, a survey experiment was conducted to test whether individuals were more
likely to respond negatively when a hypothetical country with a long history of friendship
switched sides, compared to when an allied country or a hypothetical country with regular
diplomatic relations committed the same action. In Chapter 4, I substantiate the results
from the survey experiment with a case study of the British-Thai relations during the
Second World War, when Thailand broke its treaty of neutrality with Britain and signed
a treaty of alliance with Japan.
The findings in these two chapters are similar: Betrayal reduces trust and decreases the
willingness to continue the relationship with the betrayer. In addition, as shown in Chapter 4,
the reaction to betrayal was not a result of rational and calculative thought but rather an
25

intuitive response. As in the other cases of moral violations, betrayal triggers an automatic,
intuitive, and emotional-based judgment that leads to a reaction even before a person
consciously understands the circumstance (Fitness, 2006). When moral violations occur,
people often make moral judgments without a deliberative reasoning process (Haidt, 2007;
Haidt et al., 1993; Mullen and Skitka, 2006; Skitka and Mullen, 2002; Skitka, 2002, 2010).
However, besides the consensus that betrayal acute emotional reactions that precede the
comprehension of the situation, it remains inconclusive as to which specific feelings dominate
after betrayal. While a number of studies find that betrayal in close relationships leads to
“hurt” and “pain” when experiencing betrayal (Leary et al., 1998; Planalp and Fitness, 1999),
some found betrayal to be associated with anger (Fehr and Baldwon, 1996; Morrison and
Robinson, 1997). The latter emotional response arises from perceptions of injustice and
moral wrongdoing, which elicit anger in these relational contexts. The survey experiment
in Chapter 3 found support for both emotions. Respondents who were primed with the
friendship condition reported high feelings of both anger and hurt upon learning about the
act of betrayal by a friend. However, respondents in the friendship condition reported the
highest feelings of hurt compared to respondents in other conditions.
These varying emotions have significant implications for actions, as partially revealed
by the experiment. Different emotional responses can lead to different behaviors. For
example, anger has been found to motivate confrontation and engagement with the betrayer,
potentially driving the need for retribution (Fitness and Fletcher, 1993). In contrast, hurt,
similar to hatred, may lead to emotional withdrawal and avoidance. As demonstrated in
Chapter 3, anger was found to partially mediate the desire to punish the betrayer, while
26

hurt was found to partially reduce the willingness to collaborate further with the betrayer.
Empirical Strategy of This Dissertation
This dissertation aims to provide empirical evidence for the phenomenon of friendship and
betrayal in international politics. It accomplishes this by utilizing a mixed-method approach
to address the central question of whether betrayal is a relevant concept in international
relations. Given that both the concepts of friendship and betrayal are thinly studied in
the field of international relations, it is necessary to first present empirical evidence of their
existence in international politics before proceeding to the study of their causes and effects.
For this reason, the subsequent (Chapter 2) uses word embeddings analysis to examine
whether friendship is a concept distinct from alliance in international relations. With the
findings showing that friendship is indeed a distinct concept in international politics, I then
turn to a survey experiment in Chapter 3 to study the causes and effects of betrayal in foreign
policy decisions. Additionally, as Haugevik (2018) suggested, we might be able to see the
sentiment of betrayal by examining public opinion, as they are the consumers and embracers
of political discourse. In Chapter 4, I examine the real-world case of British-Thai relations
during the Second World War, as it is a prime example of countries that considered each
other friends but ended up switching sides during the conflict. Throughout this dissertation,
I will demonstrate that betrayal is a significant phenomenon in international relations and
can bring substantial consequences in international affairs.
27

Chapter 2
Is Friendship in IR a Myth?
Betrayal would not be relevant in international relations if friendship does not exist within
this realm. From general observations, we can see that some countries feel closer to certain
groups of countries than others. We can also see that policymakers often make references to
“friends” in their foreign policy-related speeches. For instance, in her remarks to the ASEAN
nations in 2009, Secretary of State Hilary Clinton identified North Korea as “a threat to our
friends and allies” (Clinton, 2009). Similarly, in his inaugural speech, President John F.
Kennedy also mentioned friends: “To those old allies whose cultural and spiritual origins we
share, we pledge loyalty to faithful friends” (Kennedy, 1961).
28

Despite regular references to friends in addition to allies in their speeches, we still do
not know if policymakers understand these two types of relationships as similar or distinct
constructs in international relations. Additionally, we do not know if policymakers, when
making references to friendship in their foreign policy-related speeches, genuinely consider
countries they mentioned as friends or if it is merely a rhetorical act.
Qualitative observations yield little insight into these questions, in part because, with
small numbers of observations, there are inconsistencies in the application of the term “friend”
when compared to “ally” by politicians. On the one hand, politicians often mention a
country’s obligations to allies, not friends. For example, in the same speech to the ASEAN
nations, Clinton emphasized the U.S. responsibility to “protect our allies” against unwelcome
behaviors by North Korea, thereby excluding friends from the U.S. sphere of protection
(Clinton, 2009). On the other hand, when Trump’s threat to withdraw from NATO loomed
large, politicians like Vice President Kamala Harris emphasized the U.S.’ obligation to keep
one’s word and be loyal to friends (Page, 2020).
Amidst these confusions, This chapter aims to present a systematic study of how policymakers perceive “friends” as opposed to other well-known types of relationships in international relations, such as alliances. This analysis utilizes the word embeddings method
to examine private discussions among U.S. foreign policymakers documented in the Foreign
Relations of the United States (FRUS) collection. Word embeddings function by positioning words that share similar contexts closer together. If “friend” and “ally” are indeed
synonymous in foreign policy discourse, the contexts in which these terms appear should be
indistinguishable. Conversely, if the words clustering around these two terms are qualitatively
29

different, this would indicate that policymakers conceptualize and use these terms differently.
Finally, I compare the use of “friend” and “ally” in the realm of foreign policy with their
everyday use by the public. This comparison aims to illuminate how these terms are
conceptualized and applied differently in personal versus political contexts, providing a
deeper insight into the nuances of relational terminology in international relations.
This chapter proceeds as follows. The subsequent section provides a justification for
comparing friends with allies, as opposed to other types of relationships in international
relations. This is followed by the introduction to the word embeddings method and its
function in enhancing our understanding of words and their contexts. I then present the
results of the word embeddings analysis, which support the assumption that “friends” and
“allies” are different constructs in the eyes of policymakers.
Why Comparing Friends with Allies?
A question may arise as to why I compare friends to allies and not other relational concepts.
The reason for this comparison is due to the extensive knowledge and understanding the field
of international relations has about alliances in international politics. Voluminous literature
exists in the field attempting to disentangle alliance formations, alliance politics, the dynamics between alliance members, and effective management of alliances (for example, see Leeds,
2003, 2018; Leeds et al., 2000; Morrow, 1991, 1994; Niou and Ordeshook, 1994; Morrow, 1991;
Snyder, 1984). Due to this extensive understanding of alliances in international relations,
we can establish some baseline expectations about the behaviors of and between allies.
From the realist perspective, we expect alliances to be formed to either restore the balance
30

of power (Waltz, 1979) or to balance against threat (Walt, 2019). These actions reflect
the interests of states in maintaining their security and ultimate survival. The balance of
power theory suggests that states form alliances to prevent any one state from becoming
too powerful, thereby maintaining the equilibrium in the international system. Conversely,
the balance of threat theory posits that states align themselves against perceived common
threats to their security, not just power per se.
Nevertheless, because alliances are formed to address changes in the structure of the
international system, they are not expected to continue perpetually. In fact, they are
inherently temporary. Mearsheimer aptly characterizes alliances as “temporary marriages
of convenience,” highlighting the fluid nature of these partnerships (Mearsheimer, 1994,
11). Once the shared interests dissipate, so too do the alliances. Maintaining long-term
relationships and loyalty in this self-help world is deemed perilous and costly for states,
echoing the realist belief that “it pays to be selfish” (Mearsheimer, 1994, 11).
Rationalists agree with realists on the assumption of anarchy and the self-interested
nature of the state. However, rationalists diverge from realists on the presumed temporary
nature of alliances. Unlike the realist expectation for alliances to be dissolved once the shared
interests or threats no longer exist, rationalists posit that these cooperative arrangements can
endure beyond immediate shared interests (Morrow, 2000). A genuine alliance, according
to Niou and Ordeshook (1994), involves a formal commitment with specified obligations,
thereby enhancing the long-term security of alliance partners despite the absence of immediate threats (Lalman and Newman, 1991).
At the core of a state’s commitment to alliance obligations lie the perceived benefits of
31

continued participation and the costs associated with defection. Remaining in the alliance
enhances the state’s security while reducing its defense burden (Leeds, 2003). By adhering
to alliance commitments, states signal to external observers their readiness to intervene
in conflicts involving their allies, thereby bolstering the alliance’s credibility and deterring
potential threats (Morrow, 2000; Leeds, 2003).
In addition to signaling to actors outside the alliance, by honoring alliance commitments,
states also reassure other alliance partners of their security-seeking motivations and cooperative attention (Kydd, 2007; Snyder, 1984). This dynamic within alliances illustrates each
member’s endeavor to demonstrate reliability, hoping to receive the same from other partners.
The reassurance function contributes to building trust and solidarity among alliance partners,
fostering cooperation that transcends immediate self-interests.
Despite the incentives to cooperate, this body of literature also acknowledges the possibility of states defecting or abandoning alliances. The likelihood of such actions may increase
when the costs of alliance formation and violation are low, or when significant factors undergo
changes, prompting leaders to reassess their interests (Bennett, 1997; Morrow, 1991; Silver,
1989). Given these potential scenarios, states must devise strategies to mitigate the risk
of abandonment, including increasing the costs of defection, even though these efforts may
create negative externalities on other issues (Snyder, 1984).
As we observe, the rationalist approach to studying the formation, maintenance, and
abandonment of alliances places significant emphasis on the costs and benefits associated
with such alliances, which ultimately shape the state’s structure of interests. States engage
in alliances with the expectation of deriving benefits and reciprocity from their partners,
32

maintaining their commitment through a careful calculation of these costs and benefits.
Consequently, alliances in international relations resemble those in other contexts, characterized by a transactional and calculative nature. Within this framework, commitment to
alliances is driven not by ethics or morality but by a rational assessment of the costs and
benefits of such alliances, devoid of emotional bonds. This gives rise to the first hypothesis:
H1: The term “ally” is more likely to associated with transactional relationships, driven
by rational cost-benefit analysis.
Unlike our extensive understanding of alliances, friendship remains a less explored type of
relationship in this field. Debates persist regarding the extent to which friendship exists and
matters in the realm of international relations. Existing research on friendship argues that
friendship is indeed a distinct construct in international relations, not merely rhetorical but
a type of relationship that involves changes in shared identity, mutual trust, and emotional
bonds between states (Berenskoetter, 2014; Wendt, 1999; Haugevik, 2024; Oelsner, 2007;
Oelsner and Koschut, 2014).
Instead of relying on the logic of rationality and cost-benefit calculations, friends in
international relations are bound by shared characteristics such as “we-feeling, solidarity,
plural subject, common ingroup identity, thinking like a team, [and] loyalty) (Wendt, 1999,
305). Just like in interpersonal relations, friendship between states is predicated on noncalculative trust—the belief that the other side would never betray us even if they have the
ability to do so. It is precisely because of this non-calculative trust that routine assessments
of the other side’s trustworthiness and enforcement mechanisms become unnecessary. In fact,
such routine assessments and expressions of distrust could undermine the very foundation
33

of friendship, even at the international level.
Since friendship at the interstate level operates much similarly to friendship at the
interpersonal level, we can expect the same relational expectations and moral codes of
conduct to function in international relations as well. Because of the prevalence of emotional
bonds and moral obligations associated with friendship, it is likely that policymakers, when
referring to their friends at the international level, think of them in the context of morality
and emotions. This gives rise to the second hypothesis:
H2: The term “friend” is more likely to be linked to relational dynamics, emphasizing
emotional connections and moral considerations.
Because these two types of relationships in international relations imply different
relational expectations towards the other party, we anticipate the terms “allies” and
“friends” will reflect distinct relational constructs within foreign policy discourse.
This raises the last hypothesis:
H3: Allies and friends are qualitatively different.
Introducing Word Embeddings
Even though there is some theoretical groundwork for distinguishing between friends and
allies, it remains unclear whether policymakers practically differentiate between the two
terms. After all, it is uncertain whether friendship is merely rhetoric used in foreign
policy discourse, devoid of any particular meaning. To examine whether policymakers make
practical distinctions between friends and allies, I turn to text analysis, particularly word
34

embedding models as utilized in previous works such as the study by Kozlowski et al. (2019)
and by Rathbun and Pomeroy (2022).
These works highlight the value of word embedding models in understanding the contextual nuances of specific words. For example, as demonstrated in Kozlowski, Taddy,
and Evans’ study, occupational identifying words are closely associated with words that
signify specific gender, race, and class. This result exposes the biases linked to gender,
race, and class that are often associated with various occupations. Similarly, the findings
from Rathbun and Pomeroy’s work indicate that policymakers’ perceptions of threats and
harms in international relations are deeply influenced by moral considerations. As a result,
if policymakers indeed differentiate between allies and friends, the two terms should be
associated with different sets of words.
How do we make sense of word embedding models? “Word embeddings operationalize
the intuition that we can know a word by the company it keeps” (Rathbun and Pomeroy,
2022, 18). These models assign words specific positions in a geometric space, determined
by how words share contextual similarity with other terms in the corpus. Words with
analogous contexts are positioned closer together in the vector space, while less-related words
are spaced further apart. As illustrated in the research by Kozlowski, Taddy, and Evan,
traditionally feminine occupations like “nurse” and “nanny” are located near words denoting
female gender, such as “she,” “her,” and “woman.” On the contrary, traditionally masculine
professions like “engineer” and “lawyer” appear on the opposite end, surrounded by terms
connoting male gender like “his,” “him,” and “man.” It is noteworthy that these words do
not require direct co-occurrence to be placed close in the vector space; a shared context is
35

sufficient for proximity in position (Kozlowski et al., 2019).
For my investigation, I rely on 16,000 documents from the Foreign Relations of the
United States (FRUS) collection. This compilation consists of private communications
among U.S. policymakers, carefully collected by the Department of State’s historian’s office.
Specifically, I utilize the corpus created from the documents spanning the years 1952 to
1977, as aggregated by Rathbun and Pomeroy (2022). The rationale for selecting this
particular collection stems from FRUS’ inclusion of private communications rather than
public statements. This aspect renders it a hard test for my hypothesis. If friendship is
indeed merely rhetoric in international relations, it is less probable that policymakers will
make practical distinctions between the two relational concepts in private settings.
The FRUS documents were pre-trained by Rathbun and Pomeroy (2022), Words
extracted from these pre-processed digitized texts will be assigned positions within a
N-dimensional vector space in relation to other terms. Essentially, word embeddings
analysis creates a map of words where words closer to each other share semantic contexts.
This proximity is measured through “cosine similarity,” as the map is not a simple
two-dimensional representation but rather composed of hundreds of dimensions. Cosine
similarity helps determine how similar two words are by measuring the cosine of the angle
between their vector representations in this high-dimensional space. In other words, terms
that share analogous contexts will be placed in proximity, while words with no significant
relation will find themselves at opposite ends of the vector space.
For the case of friends and allies, if policymakers perceive the two terms as different
relational constructs in foreign policy, the terms proximate to the word “friend” should
36

show minimal overlap with those near the word “ally.” Additionally, the terms surrounding
the word “ally” should denote a more cold, rational, and calculative mode of thinking.
Conversely, the terms surrounding the word “friend” should indicate a warm and relational mode of thinking.
Allies and Friends are Indeed Different
In the initial analysis, I adopt an exploratory and inductive approach to investigate how
policymakers practically differentiate between friends and allies within the realm of international relations. To accomplish this, I identify the 40 nearest neighbors to the terms “friend”
and “ally.” The words that exhibit high cosine similarities with the target terms “friend”
and “ally” are considered to share similar contexts with the target words. The intuition
is that if the two words share an analogous meaning and context, their nearest neighbors
should somewhat overlap. Conversely, if the two terms appear in unrelated contexts, their
nearest neighbors should be notably different.
Table 2.1: Nearest neighbors to “friend” and “ally”
Top 40 nearest
neighbors to “friend”
Cosine
similarity
Top 40 nearest
neighbors to “ally”
Cosine
similarity
neighbor 0.470
(0.423 - 0.510)
western 0.657
(0.639 - 0.691)
good 0.470
(0.436 - 0.497)
nato 0.583
(0.556 - 0.609)
countri 0.459
(0.434 - 0.480)
berlin 0.573
(0.535 - 0.609)
relat 0.449
(0.432 - 0.466)
west 0.563
(0.537 - 0.603)
peopl 0.448
(0.421 - 0.476)
access 0.561
(0.541 - 0.582)
our 0.430
(0.400 - 0.452)
our 0.542
(0.520 - 0.584)
37

Top 40 nearest
neighbors to “friend”
Cosine
similarity
Top 40 nearest
neighbors to “ally”
Cosine
similarity
attitud 0.427
(0.410 - 0.447)
power 0.510
(0.481 - 0.547)
hope 0.426
(0.396 - 0.452)
should 0.507
(0.493 - 0.534)
relationship 0.424
(0.386 - 0.468)
posit 0.507
(0.485 - 0.524)
with 0.421
(0.365 - 0.455)
forc 0.491
(0.475 - 0.518)
frank 0.419
(0.375 - 0.461)
us 0.489
(0.473 - 0.525)
veri 0.419
(0.396 - 0.449)
consult 0.478
(0.443 - 0.500)
your 0.416
(0.373 - 0.446)
german 0.476
(0.445 - 0.517)
maintain 0.416
(0.387 - 0.453)
their 0.470
(0.433 - 0.493)
and 0.416
(0.374 - 0.462)
other 0.464
(0.436 - 0.482)
person 0.414
(0.376 - 0.437)
soviet 0.463
(0.429 - 0.492)
cooper 0.412
(0.380 - 0.465)
germani 0.460
(0.446 - 0.485)
interest 0.409
(0.372 - 0.434)
would 0.459
(0.424 - 0.479)
helpful 0.407
(0.350 - 0.467)
east 0.455
(0.436 - 0.472)
other 0.407
(0.364 - 0.444)
unit 0.452
(0.417 - 0.484)
my 0406
(0.385 - 0.421)
if 0.452
(0.420 - 0.480)
most 0.406
(0.369 - 0.429
to 0.451
(0.416 - 0.475)
cordial 0.406
(0.376 - 0.454)
respons 0.445
(0.404 - 0.476)
who 0.405
(0.381 - 0.443)
command 0.444
(0.418 - 0.477)
close 0.403
(0.379 - 0.428)
with 0.443
(0.415 - 0.483)
him 0.403
(0.374 - 0.416)
have 0.442
(0.404 - 0.478)
38

Top 40 nearest
neighbors to “friend”
Cosine
similarity
Top 40 nearest
neighbors to “ally”
Cosine
similarity
mani 0.400
(0.341 - 0.445)
millitari 0.441
(0.421 - 0.468)
feel 0.400
(0.362 - 0.428)
all 0.441
(0.400 - 0.458)
both 0.397
(0.362 - 0.421)
such 0.440
(0.408 - 0.484)
help 0.395
(0.362 - 0.416)
maintain 0.439
(0.400 - 0.430)
wish 0.395
(0.358 - 0.428)
be 0.439
(0.419 - 0.458)
welcom 0.387
(0.354 - 0.418)
gdr 0.439
(0.408 - 0.476)
confid 0.387
(0.347 - 0.417)
as 0.438
(0.401 - 0.475)
like 0.387
(0.350 - 0.426)
we 0.436
(0.420 - 0.449)
but 0.386
(0.360 - 0.410)
consid 0.436
(0.406 - 0.479)
understand 0.386
(0.333 - 0.409)
will 0.434
(0.396 - 0.472)
unit 0.386
(0.363 - 0.402)
ani 0.432
(0.401 - 0.468)
desir 0.384
(0.372 - 0.395)
right 0.431
(0.405 - 0.450)
way 0.383
(0.340 - 0.410)
there 0.431
(0.394 - 0.463)
want 0.380
(0.360 - 0.406)
the 0.427
(0.408 - 0.448)
As depicted in Table 2.1, the lists of top 40 nearest neighbors to the words “friend”
and “ally” show minimal overlap, indicating a distinct differentiation in their usage. Furthermore, neither “friend” nor “ally” appears in proximity to the other. This disparity
strongly suggests that policymakers do not use these two terms interchangeably. Moreover,
this initial examination reveals how policymakers place “friend” and “ally” in different
types of relationship categories and establish different sets of expectations for each, thus
39

confirming the third hypothesis.
In the perspective of policymakers, the terms “friend” and “ally” imply different types
of relationships, with “ally” being more instrumental and “friend” being more relational.
For instance, in proximity to terms such as “power,” “force,” and “access,” allies appear
to assume a more instrumental role in a state’s pursuit of its interests. In contrast, the
term “friend,” surrounded by nearest neighbors such as “relate,” “relationship,” “people,”
“cooperate,” “wish,” and “desire,” takes on a more social and interpersonal connotation.
Moreover, because policymakers categorize “friend” and “ally” into different relationship
types, they evoke distinct expectations for each. As evident in the top 40 closest words to
“friend,” policymakers tend to hold “hopes” and “wishes” for friends to be “helpful,” “frank,”
and “cordial.” In contrast, the list of words surrounding the term “ally,” which includes terms
like “should,” indicates an expectation of obligation from allies.
Allies are Instrumental, but Friends are Relational
However, despite the insights gained from identifying the 40 nearest words to both terms,
such examinations have limitations. First, these surrounding words may not fully illustrate the nuances within different relationship categories, including the associated relational
expectations. Second, and more importantly, examining only 40 nearest words may not
provide enough evidence to confidently conclude that the terms “friend” and “ally” are
indeed distinct in the eyes of policymakers.
If my argument is correct, friendship should have moral connotations that allyship
does not, and friendship should possess an emotional character that distinguishes it from
40

the rational and transactional nature of alliances. As mentioned earlier in this chapter,
allies, characterized by a transactional relationship, involve a process of calculation, strategic cooperation, and mutual benefit. Conversely, friends are more emotionally connected,
characterized by trust, empathy, and supportiveness in interpersonal interactions. If allies
and friends in international politics are understood similarly to allies and friends in the
interpersonal realm, the same patterns should also emerge in word embeddings. That is, the
word “ally” should be positioned closer to rationality terms, while the word “friend” should
be positioned closer to terms suggesting emotions and feelings.
To further examine the distinctive nature of friendship and allyship, I utilize the extended
Moral Foundation Dictionary (eMFD) developed by Hopp (2021).1 Conceptually, despite
the refusal by structural realists, international relations has proven to be a field filled
with morality (Rathbun, 2023; Pomeroy and Rathbun, 2023; Rathbun and Pomeroy, 2022;
Rathbun et al., 2022). I focus specifically on loyalty/betrayal and fairness/cheating as these
are the two domains most relevant to the issue of friendship and betrayal. As demonstrated
by Rathbun et al. (2022), countries are, in fact, less concerned about the benefit of club goods
and more about being treated fairly by members of the international community. Similarly,
the alliance literature is concerned with how to make members of alliances loyal to their
commitments. Because of their relevancy to international relations, using these two moral
domains should be appropriate for our investigation.
1
Instead of promoting the “winner take all” approach in which one word belongs to a discrete, binary
category, the eMFD adopts a multifoundational, probabilistic, and continuous word weighting approach.
Under this approach, each word is assigned a probability of being associated with the virtues and vices of
each moral foundation category. This allows words to be highly associated with more than one moral domain.
However, for research with an interest in selecting words based on their representativeness of a particular
moral domain, such as my current work, Hopp et al. (2021) provides an alternative classification scheme in
which they assign each word to specific foundations based on the highest probability score.
41

Practically, however, the extended Moral Foundation Dictionary includes many terms
that are not terribly well-described as truly moral but are better described as being rational
and calculative. For example, the domain of fairness includes both rationality words such
as “benefit,” “competitive,” and “advantage,” as well as words conveying emotion such as
“enjoy,” honor,” and “surprise.” I make use of this broad spectrum of “moral” words to
explore the semantic associations of friend and ally in this corpus of texts. First, I use word
embeddings to distinguish between more rational and emotional terms in subdictionaries
of the eMFD. Having classified those terms as either rational or emotional, I then judge
their association to conceptions of friendship and alliance to test whether, as expected,
friendship has more emotional connotations. The inclusion of both rationality and emotion
words allows us to identify how policymakers process information and arrive at decisions
regarding friends compared to allies. In addition, the extended Moral Foundation Dictionary
distinguishes between positive and negative terms. Loyalty is positive, whereas betrayal is
negative, for instance. Table 2.2 displays words listed in the domain of loyalty/betrayal
and fairness/cheating in the eMFD.
Table 2.2: Loyalty, Betrayal, Fairness, and Cheating Words from the extended Morality Foundation
Dictionary (eMFD)
Loyalty Betrayal Fairness Cheating
liked prejudice fresh litigation
successor fear accepting destroying
agreement abandoned swiftly sentenced
playing uncertainty reached charged
gain bitter justice violations
passionate injury supporting charges
support failure clearly leave
outstanding unfortunately free arrests
committed forget entitied negative
praised struggle gains lawsuits
42

Loyalty Betrayal Fairness Cheating
advanced questioned engagement violated
pride warning active broken
rich killer wealthy guilty
important fearful grant problem
endorsed stealing competitive victims
promise critical favorite detained
unified leaked benefits inability
confident doubt true threaten
admit hateful ethical low
join reject share pay
thank rejected resolve hide
improvement bias giving scared
improved suspected benefit bad
loved ignored admitted lies
confidence ban attract detention
united conspiracy verdict steal
intelligence blame freedom worries
hopes doubts opportunity denied
loyalty angry motivated debt
parties argument credit accused
excuse offenses sentences resigned
easily accusing freed violation
peaceful hatred advantage complaints
beneficial warnings boost punished
wins disappointed fair deny
fulfill affected interests fraud
certainly forgotten enjoy refuse
strong argued winners illegal
challenges havoc surprise desperate
devoted attacker assets threatened
willingness aggressive honored impose
successful hard protect delayed
remarkable rejection awarded killings
solidarity unclear freedoms prison
success criticism wealth criminal
winning argue clear injustice
happy burden matters accusations
shares betrayal agreed struggles
celebrate win fails
solve recommend painful
friends peacefully punishment
sharing supports weak
43

Loyalty Betrayal Fairness Cheating
liberty responsible shake
friend grants lose
loves ensure crime
positive commit confront
definitely legal undermining
careful promises unfair
surprised secure mistakes
want honest devastating
hero importance fake
commitment winner critics
agreements sentence rigged
dedicated opportunities cuts
supported freely dump
supporter ensuring seriously
agree worth propaganda
proud solutions lawsuit
promised defender suspended
challenging ill
integrity suspect
prevented sentencing
easy disagree
alive blocked
honor fail
legally violating
substantial
accept
For the purpose of this study, which aims to explore the rationality/emotional dichotomy
relevant to friendship and allyship, it is necessary to categorize these morality terms into a rationality/emotion dimension, which I can construct using word embeddings. Rather than simply picking and choosing terms I believe to be relevant, which might suffer from bias, I instead
make use of word embeddings itself to categorize these terms as more rational and emotional.
The analysis involves two major steps. Firstly, I construct a rationality/emotion dimension using words closely associated with the two concepts. For the rationality pole, I
44

compile a dictionary comprising terms closely linked to rational choice theory in international relations, assuming that these words primarily represent the rational decision-making
process. Specifically, the terms selected for the rationality dimension are mostly derived from
Keohane’s work on hegemony, cooperation, and discord (1984), which explores mechanisms
that make states realize the benefits of cooperation. For the emotion pole, I came up with
words related to emotion and each specific emotion itself, Table 2.3 displays the rationality
and emotion dictionaries I compiled. The vector position of “rationality” and “emotion” are
the averages of the locations of their component words in this n-dimensional vector space.
Table 2.3: Rationality and Emotion Dimension
Rationality Dimension Emotion Dimension
rationality emotion
transparency feel
iteration happiness
incentive sadness
interest anger
information fear
preference surprise
calculation disgust
cost anxiety
benefit embarrassment
welfare humiliation
strategy
enforcement
institution
utility
maximization
defect
The result of projecting the moral words into the rationality/emotion dimension successfully groups each term into two categories, which I will call rational “moral” words and
emotional moral words. Figure 2.1, Figure 2.2, Figure 2.3, Figure 2.4 illustrate the position45

benefit
grant
ensur
protect
legal
asset
promis
integr
competit
attract
resolv
engag
fair
boost
sentenc
entitl
challeng
wealth
defend
share
ethic
worth
easi
winner
enjoy
freeli
verdict
award
fresh
freed
honor
motiv
aliv
swift
favorit
win
honest
admit
true
wealthi
surpris
−0.2 0.0 0.2 0.4
Projecting Fairness to Rationality/Emotion Dimension
Figure 2.1: Projecting fairness words onto
the rationality/emotion dimension with rationality on the left pole and emotion on the
right pole
suspend
block
debt
deni
low
destroy
lie
sentenc
undermin
crimin
illeg
arrest
punish
struggl
prison
fraud
steal
inabl
crime
devast
litig
desper
detain
dump
unfair
victim
kill
rig
hide
resign
shake
pain
disagre
broken
complaint
suspect
guilti
accus
ill
mistak
scare
worri
−0.2 0.0 0.2
Projecting Cheating to Rationality/Emotion Dimension
Figure 2.2: Projecting cheating words onto
the rationality/emotion dimension with rationality on the left pole and emotion on the
right pole
46

improv
support
import
advanc
intellig
agreement
gain
unit
agre
peac
promis
certain
success
commit
care
posit
parti
benefici
outstand
unifi
solv
willing
fulfil
definit
endors
challeng
hope
play
like
share
successor
want
loyalti
liberti
join
excus
devot
easili
solidar
rich
friend
strong
win
celebr
dedic
love
admit
confid
prais
pride
thank
remark
passion
hero
proud
happi
surpris
−0.2 0.0 0.2 0.4
Projecting Loyalty to Rationality/Emotion Dimension
Figure 2.3: Projecting loyalty words onto the
rationality/emotion dimension with rationality on the left pole and emotion on the right
pole
question
affect
offens
hard
ban
burden
abandon
aggress
failur
critic
argument
reject
unclear
struggl
argu
attack
prejudic
steal
ignor
bias
doubt
injuri
warn
blame
conspiraci
forgotten
hate
uncertainti
suspect
unfortun
forget
accus
leak
betray
hatr
fear
angri
bitter
disappoint
−0.2 0.0 0.2 0.4
Projecting Betrayal to Rationality/Emotion Dimension
Figure 2.4: Projecting betrayal words onto
the rationality/emotion dimension with rationality on the left pole and emotion on the
right pole
47

ing of each moral word in relation to the rationality and emotion dimensions. Upon analysis,
the results show words indicating emotions positioned closer to the emotion pole on the right
and those indicating strategic calculation situated closer to the rationality pole on the left.
For example, words like “surprised,” “worried,” “astonished,” and “disappointed” are
situated on the right end of each moral domain, indicating their emotional connotations.
Similarly, words like “benefit,” “suspend,” “improve,” and “question,” which are associated
with calculative methods, are situated on the far left side of the four moral domains. This
clear alignment highlights the accuracy and efficiency of the categorization process, contrasting with the subjective method manually done by researchers.
Based on the projection, I categorize the terms into eight additional dictionaries for use in
the main analysis. These include the rational fairness, emotional fairness, rational cheating,
emotional cheating, rational loyalty, emotional loyalty, rational betrayal, and emotional
betrayal dictionaries. Using the information from the four figures, any words positioned to
the left of the zero line is added to the rationality dictionary of each moral domain, specifically
the rational fairness, rational cheating, rational loyalty, and rational betrayal dictionaries.
Conversely, any words positioned to the right of the zero line will be categorized in the emotion dictionary for each moral domain, namely the emotional fairness dictionary, emotional
cheating, emotional loyalty dictionary, and emotional betrayal dictionaries. Table 2.4 and
Table 2.5 show the rationality and emotion words within each domain.
Table 2.4: Rationality words within each moral domain
Rational Fairness
Rational
Cheating Rational Loyalty Rational Betrayal
benefit suspend improv question
grant block support affect
48

Rational Fairness
Rational
Cheating Rational Loyalty Rational Betrayal
ensur debt import offens
protect deni advance hard
legal low intellig ban
asset destroy agreement burden
promi lie gain abandon
integr sentenc unit aggress
competit undermin agre failur
attract crimin peac
resolv illeg promis
engag arrest certain
fair punish success
boost commit
sentenc care
entitl posit
challeng parti
wealth benefici
defend outstand
share unifi
ethic solv
worth willing
easi fulfil
definit
endors
challeng
hope
play
like
share
successor
want
loyalti
Table 2.5: Emotion words within each moral domain
Emotional
Fairness
Emotional
Cheating
Emotional
Loyalty
Emotional
Betrayal
surpris worri surpris disappoint
wealthi scare happi bitter
true mistak proud angri
admit ill hero fear
49

Emotional
Fairness
Emotional
Cheating
Emotional
Loyalty
Emotional
Betrayal
honest accus passion hatr
win guilti remark betray
favorit suspect thank leak
swift complaint pride accus
aliv broken prais forget
motiv disagree confid unfortun
honor pain admit suspect
freed shake love uncertainti
fresh resign dedic hate
award hide celebr forgotten
verdict rig win conspiraci
freeli kill strong blame
enjoy victim friend warn
winner unfair rich injuri
dump solidar doubt
detain easili bias
desper devot ignor
litig ecus steal
devast join prejudic
crime liberti attack
inabl argu
steal struggl
prison unclear
struggl reject
argument
critic
In the next step, I project each of the terms from these dictionaries onto a relationship dimension, constructed using the word “friend” on one end and “ally” and “alliance”
on the other. Considering that friends typically embody a more relational nature, while
alliances often showcase a more transactional aspect, I hypothesize that words from the
eMFD with a rational connotation will be positioned closer to the ally pole on the left side.
Conversely, more emotional-laden words in the eMFD are expected to be situated nearer
to the friend pole on the right side.
50

The results unequivocally support both H1 and H2. Policymakers consistently distinguish
between allies and friends, assigning them to distinct relational categories and employing
different judgment processes when discussing the moral qualities of states. Regarding the
domain of fairness in Figure 2.5 and Figure 2.6, 59 percent of the terms in the rational
category are positioned closer to the ally pole on the left, whereas only 41 percent of terms
are situated closer to the friend pole on the right. Conversely, in emotional fairness, 65
percent of the terms are located closer to the friend pole on the right, while only 35 percent
of the terms are located closer to the rational pole on the left.
In the cheating domain, as depicted in Figure 2.7 and Figure 2.8 cheating, 54 percent
of the rational terms are located closer to the rational pole on the left, while 46 percent
of the rational terms are situated closer to the friend pole on the right. Furthermore, even
though some terms may be closer to the friend dimension, the cosine similarity tends to be
smaller compared to when they are situated closer to the rational dimension. In contrast,
66.5 percent of the emotional cheating terms are located closer to the friend dimension on
the right, while 34.5 percent of the emotional cheating terms are situated closer to the ally
dimension on the left. Analogous to what happened with the rational cheating words, the
cosine similarity of words situated closest to the friend domain is twice as large as that of
words situated closest to the ally domain.
In the loyalty domain, which is arguably one of the most important aspects of alliances
and friendships, the results point to a consistent trend. As demonstrated in Figure 2.9
and Figure 2.10, the rational terms for loyalty, such as “commitment” and “agreement,”
are situated closer to the ally dimension. Specifically, 55 percent of the rational loyalty
51

integr
legal
challeng
asset
entitl
ensur
resolv
defend
protect
boost
sentenc
grant
ethic
share
competit
attract
worth
wealth
engag
easi
benefit
fair
−0.2 −0.1 0.0 0.1
Projecting Rational Fairness Words to Relationship Dimension
Figure 2.5: Projecting rational fairness
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
winner
fresh
verdict
award
freed
swift
favorit
motiv
aliv
freeli
win
admit
wealthi
honor
enjoy
true
honest
−0.1 0.0 0.1 0.2 0.3
Projecting Emotional Fairness Words to Relationship Dimension
Figure 2.6: Projecting emotional fairness
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
52

block
crimin
undermin
illeg
sentenc
suspend
deni
punish
lie
destroy
arrest
debt
low
−0.2 −0.1 0.0 0.1
Projecting Rational Cheating Words to Relationship Dimension
Figure 2.7: Projecting rational cheating
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
pain
struggl
disagr
crime
inabl
rig
steal
devast
scare
victim
prison
dump
shake
fraud
unfair
resign
desper
mistak
detain
worri
accus
broken
suspect
complaint
ill
hide
litig
guilti
kill
−0.2 −0.1 0.0 0.1 0.2
Projecting Emotional Cheating Words to Relationship Dimension
Figure 2.8: Projecting emotional cheating
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
53

terms are located closer to the ally dimension, while only 45 percent are located closer
to the friend dimension. For the emotional terms, 67 percent of the terms are located
closer to the friend domain, while 33 percent are located closer to the ally domain. It is
also noteworthy that emotional terms that are located closer to the ally dimension, such as
“block” and “suspend,” are not entirely representative of emotional and moral considerations.
In contrast, terms that are situated closer to the friend dimension are clearly representative
of emotional constructs, suggesting that when policymakers mention friends, they switch
to a more emotional and moral-based mindset.
The expectations also hold in the case of betrayal, as depicted in Figure 2.11 and
Figure 2.12. For the rational terms associated with betrayal, only one term is situated
closer to the “friend” dimension, while the rest are skewed towards the “ally” dimension.
However, when it comes to emotional terms, the distribution of words located closer to the
“ally” dimension versus the “friend” dimension is less clear. In fact, for the emotional betrayal
domain, the location of the words is more skewed towards the “ally” pole. Specifically, 47
percent of the emotional terms are clearly situated closer to the “friend” dimension, while
53 percent are closer to the “ally” dimension.
There might be two explanations for the failure to meet the expectation. First, the
complexity and emotional weight of betrayal might prompt policymakers to respond more
emotionally, irrespective of whether the betrayal was committed by an ally or a friend.
Betrayal, by its very nature, involves a significant breach of trust, which can evoke strong
emotional reactions that transcend the specific type of relationship. Second, it is possible that
some words categorized into the emotional betrayal domain are not definitively emotional.
54

commit
agreement
challeng
unifi
definit
posit
advanc
fulfil
import
parti
certain
success
will
support
gain
play
intellig
liberti
unit
loyalti
solv
care
share
peac
outstand
improv
want
successor
like
benefici
hope
−0.2 −0.1 0.0 0.1 0.2
Projecting Rational Loyalty Words to Relationship Dimension
Figure 2.9: Projecting rational loyalty words
onto the relationship dimension with “ally”
on the left pole and “friend” on the right pole
solidar
join
strong
pride
celebr
hero
easili
passion
win
admit
devot
rich
confid
excus
dedic
remark
proud
love
happi
thank
friend
0.00 0.25 0.50
Projecting Emotional Loyalty Words to Relationship Dimension
Figure 2.10: Projecting emotional loyalty
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
55

failur
ban
abandon
question
affect
aggress
burden
hard
−0.2 −0.1 0.0 0.1
Projecting Rational Betrayal Words to Relationship Dimension
Figure 2.11: Projecting rational betrayal
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
reject
warn
struggl
attack
argument
uncertainti
ignor
forget
injuri
betray
steal
critic
leak
doubt
argu
fear
blame
prejud
forgotten
unclear
conspiraci
disappoint
bias
accus
unfortun
suspect
bitter
hatr
hate
angri
−0.1 0.0 0.1 0.2
Projecting Emotional Betrayal Words to Relationship Dimension
Figure 2.12: Projecting emotional betrayal
words onto the relationship dimension with
“ally” on the left pole and “friend” on the
right pole
56

Words such as “argument,” “doubt,” “leak,” or “critic” could be easily categorized in the
rational betrayal domain. Without these ambiguous words, it is possible that the emotional
betrayal words would cluster more towards the “friend” pole.
Figure 2.13 helps visualize the breakdown of rational-laden words and emotional-laden
words in the eMFD that are situated closer to the “friend” pole and the “ally” pole. Across
the four moral domains, the majority of eMFD terms with rational connotations are located
closer to the ally pole. In contrast, the emotional-laden words within the eMFD are located
closer to the “friend” pole in almost all moral domains, except for the betrayal moral domain,
for reasons already indicated above. Specifically, emotional words in the loyalty domain are
highly skewed towards the friend pole. This suggests the importance of loyalty and the
strong emotions associated with it in the context of friendship.
The Everyday Use of “Friend” and “Ally”
In addition to examining how U.S. policymakers discuss friends and allies, it is also important
to consider how their discussions deviate from the everyday discourse of laypeople. Laypeople
typically perceive and discuss friendships and alliances through a more personal and less
strategic lens, while policymakers might frame these relationships in terms of national
interest, security, and strategic benefits.
To capture how normal people discuss friends and allies, I rely on a corpus of Wikipedia
and Gigaword 5 data pretrained by Pennington, Socer, and Manning (Pennington et al.,
2014). This data consists of encyclopedia entries and newswire texts, representing everyday,
nonpolitical language. Then, I project the same eight dictionaries, consisting of the
57

0
10
20
30
40
50
60
70
80
90
100
Rational words Emotional words
Percentage
Fairness domain
0
10
20
30
40
50
60
70
80
90
100
Rational words Emotional words
Percentage
Cheating domain
0
10
20
30
40
50
60
70
80
90
100
Rational words Emotional words
Percentage
Loyalty domain
0
10
20
30
40
50
60
70
80
90
100
Rational words Emotional words
Percentage
Betrayal domain
Relationship Ally Friend
Figure 2.13: The percentage of words that are located closer to either the friend or ally poles,
categorized by the morality domain to which they belong
58

FRUS (1952−77) Quotidian Texts
challeng
share
wealth
boost
entitl
asset
easi
attract
competit
worth
sentenc
ensur
defend
fair
benefit
resolv
engag
legal
integr
grant
protect
ethic
share
legal
worth
protect
fair
boost
benefit
grant
defend
wealth
attract
asset
ethic
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
win
admit
winner
verdict
freed
favorit
swift
award
fresh
aliv
honest
freeli
honor
enjoy
motiv
true
wealthi
win
admit
award
winner
true
fresh
honor
enjoy
freed
verdict
honest
swift
favorit
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
Rational Fairness and Emotional Fairness
Figure 2.14: The projection of rational fairness terms (left) and emotional fairness terms (right)
onto a relationship dimension in FRUS and quotidian texts, where the top represents the “friend”
pole and the bottom represents the “ally” pole
59

FRUS (1952−77) Quotidian Texts
sentenc
punish
illeg
crimin
suspend
arrest
block
undermin
lie
debt
low
deni
destroy
low
debt
arrest
block
destroy
lie
suspend
punish
deni
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
steal
suspect
accus
struggl
rig
fraud
devast
unfair
detain
dump
hide
scare
guilti
crime
shake
victim
desper
broken
pain
inabl
complaint
disagr
resign
ill
prison
kill
mistak
worri
litig
suspect
steal
prison
crime
kill
broken
fraud
pain
ill
victim
complaint
resign
hide
unfair
shake
scare
dump
rig
detain
desper
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
Rational Cheating and Emotional Cheating
Figure 2.15: The projection of rational cheating terms (left) and emotional cheating terms (right)
onto a relationship dimension in FRUS and quotidian texts, where the top represents the “friend”
pole and the bottom represents the “ally” pole
60

FRUS (1952−77) Quotidian Texts
loyalti
benefici
liberti
successor
fulfil
unifi
outstand
challeng
solv
definit
share
gain
play
advanc
care
intellig
success
commit
improv
certain
parti
peac
hope
like
want
support
import
agreement
posit
unit
will
will
like
support
play
want
agreement
share
care
hope
success
certain
unit
gain
successor
import
commit
fulfil
parti
improv
agr
posit
solv
endor
unifi
liberti
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
hero
passion
rich
proud
celebr
pride
dedic
excus
love
solidar
easili
devot
happi
win
admit
thank
join
confid
friend
remark
strong
win
strong
love
friend
join
rich
hero
proud
pride
thank
passion
admit
remark
prai
dedic
happi
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
Rational Loyalty and Emotional Loyalty
Figure 2.16: The projection of rational loyalty terms (left) and emotional loyalty terms (right) onto
a relationship dimension in FRUS and quotidian texts, where the top represents the “friend” pole
and the bottom represents the “ally” pole
61

FRUS (1952−77) Quotidian Texts
burden
abandon
ban
failur
aggress
affect
hard
question
hard
question
ban
affect
burden
abandon
offen
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
steal
bias
conspiraci
hatr
betray
prejud
angri
injuri
forgotten
hate
forget
bitter
blame
suspect
unclear
uncertainti
leak
ignor
accus
disappoint
unfortun
struggl
warn
argu
reject
argument
fear
critic
doubt
attack
attack
fear
suspect
doubt
blame
unclear
argument
forget
bitter
critic
hate
reject
warn
forgotten
steal
leak
bias
disappoint
betray
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
Rational Betrayal and Emotional Betrayal
Figure 2.17: The projection of rational loyalty terms (left) and emotional loyalty terms (right) onto
a relationship dimension in FRUS and quotidian texts, where the top represents the “friend” pole
and the bottom represents the “ally” pole
62

FRUS (1952−77) Quotidian Texts
burden
abandon
ban
failur
aggress
affect
hard
question
hard
question
ban
affect
burden
abandon
offen
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
steal
bias
conspiraci
hatr
betray
prejud
angri
injuri
forgotten
hate
forget
bitter
blame
suspect
unclear
uncertainti
leak
ignor
accus
disappoint
unfortun
struggl
warn
argu
reject
argument
fear
critic
doubt
attack
attack
fear
suspect
doubt
blame
unclear
argument
forget
bitter
critic
hate
reject
warn
forgotten
steal
leak
bias
disappoint
betray
−0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
Relationship Dimension
Rational Betrayal and Emotional Betrayal
Figure 2.18: The projection of rational loyalty terms (left) and emotional loyalty terms (right) onto
a relationship dimension in FRUS and quotidian texts, where the top represents the “friend” pole
and the bottom represents the “ally” pole
63

rational fairness dictionary, emotional fairness dictionary, rational cheating dictionary,
emotional cheating dictionary, rational loyalty dictionary, emotional loyalty dictionary,
rational betrayal dictionary, and emotional betrayal dictionary, onto a relationship
dimension in this corpus.
The results in Figure 2.14, Figure 2.15, Figure 2.16, and Figure 2.17 indicate that, in
general, both rational and emotional terms are located closer to the “friend” pole in quotidian,
non-political texts compared to their locations in the FRUS documents. This suggests that
people are more inclined to discuss friendship rather than allyship in everyday conversations.
Alliances are simply not a common part of everyday vocabulary. Despite this, the differences
from the politicians’ discussions in the FRUS documents are not extensive, indicating that
policymakers and the general public do not perceive friends and allies very differently.
Conclusion
This chapter represents the first systematic investigation into the differentiation between
friends and allies. The findings reveal that real decision-makers discuss friends and allies differently. Friendships are associated with more emotionally-laden moral language,
while alliances are discussed with more rationalistic terminology, as distinguished and categorized through an embedding analysis. Additionally, the comparison between the embeddings of the FRUS documents and the non-political texts shows that policymakers,
though expected to think differently from ordinary people, use the terms “friends” and
“allies” in a similar manner.
These insights lay the groundwork for deeper exploration in the following chapters. While
64

this analysis highlights the distinct nature of “friends” and “allies” in foreign policy discourse,
it also underscores the limitations of word embeddings in capturing the full spectrum of
these relationships. The next chapter will move beyond these limitations by employing a
survey experiment to investigate the processes of friendship and betrayal, providing a richer
understanding of their causal effects in international relations.
65

Chapter 3
Fractured Ties: Betrayal’s Impacts on Trust
and Cooperation
In the previous chapter, I demonstrated that policymakers perceive allies and friends as
distinct relational constructs, associating each type with different informational processing
mechanisms and expectations. The term “ally” often aligns with rational terms, suggesting
associations with rational calculations and transactional expectations, whereas “friends”
is typically closer to emotional terms, indicating emotional responses or considerations
when discussing friendships.
66

However, the word-embeddings approach has limitations. It provides context and general associations of words and terms, revealing how different communities use these words
similarly or differently and how their meanings evolve over time (Kozlowski et al., 2019).
Yet, this analysis does not reveal the mechanisms or types of relationships that link the
terms to their contextual words. Thus, it remains unclear whether these relationships are
merely correlational or causal.
To address the shortcomings of the word embeddings analysis, this chapter employs a
survey experiment to examine public responses to scenarios where an ally or friend either
declines to assist the United States in crisis or switches sides. Participants were assigned
various conditions and asked to report their levels of trust, positive attitude, willingness to
continue collaboration, and inclination to penalize the ally or friend based on their decision.
This experiment aims to capture the causal impact of perceiving another country as a friend
on individuals’ attitudes and reactions, especially when that country fails to provide support
or switches allegiance. If friendship holds a distinct significance in international relations,
differing fundamentally from other constructs such as alliances, we might expect a more
intense response to incidents where a friend fails to aid or betrays compared to an ally.
The survey experiment confirms that friends are more likely to experience a sharper
decline in trust, positive feelings, and collaboration willingness after incidents where a
friend fails to assist, reneges on a promise, or switches sides. Nonetheless, there is no
significant difference in the likelihood of wanting to penalize a friend over an ally. Yet,
if punishment is considered necessary, respondents showed a greater propensity to penalize a
friend, motivated by a belief that “they deserve it,” highlighting distinct motivations for
67

punishing friends versus allies.
Moreover, this experiment also unveils potential emotional mediators influencing responses to betrayal incidents. Although the results remain inconclusive, the findings suggest
that decisions to terminate collaboration and impose sanctions may be influenced by distinct
emotions. Anger appears to drive a stronger inclination towards punishment, while feelings
of hurt prompt a reassessment of future cooperation. Interestingly, hurt is more closely
associated with friendship than with alliance relationships. Although not overwhelmingly
significant, hurt emerged as a mediator between friendship and the decline in trust, between
friendship and the reduction of positive attitude, as well as between friendship and the
reduced willingness to collaborate.
This chapter is organized as follows: It begins by revisiting the discussion on friendship,
trust, and loyalty, which are recognized as foundational qualities of these relationships. The
next section introduces the survey experiment, detailing the sample used, the experimental
design, the procedure followed, and the findings. Initially, the analysis focuses on the effects
of different types of prior relationships on the reduction of trust, positive attitude, and
willingness to continue collaboration. It then examines how these relationship types influence
the propensity to punish the other country. The final section delves into potential emotional
mediators that mediate the effects of various types of prior relationships on trust reduction,
positive attitude, willingness to collaborate, and the inclination to administer punishment.
68

Friendship Loyalty and Betrayal of Trust
Recalling the discussion about friendship from Chapter 1, friendship, much like other types
of social relationships, is ascribed with social rules about its favorable behaviors and moral
codes of conduct. The maintenance and quality of friendship are predicated upon the rules
of reciprocity, mutual care, openness, honesty, and loyalty (King, 2007; Silver, 1989; Berenskoetter, 2014; Eznack, 2012). However, without the enforcing mechanisms and constant
monitoring of behaviors that exist in other types of contractual relationships, the “health”
of friendship depends on the “hope” and “trust” that the other person in the relationship
will adhere to these rules and avoid breaking them.
One way these codes of conduct can deter rule-breaking behaviors is by instigating feelings
of right and wrong as well as guilt, which may discourage individuals from committing
actions deemed wrong by society. Frank (1988) explains how emotions and passions can lead
to actions that ultimately serve our long-term interests. Similarly, scholars who study the
role of morality in society found that, despite these moral codes being implicit, people often
“feel” that something is wrong and immoral, even if they cannot explain why. Haidt (Haidt,
2007) calls this “morally dumbfounding,” where people know intuitively that something is
wrong, even when they cannot articulate the reason.
Yet, even with these moral codes of conduct, there always exists a possibility that these
relational codes will be violated. After all, these codes of conduct, like other types of morality,
are intended to deter unfavorable behaviors but cannot entirely prevent them. However,
given that moral codes exist to govern these relationships, the violator cannot escape the
consequences of the violation. The breach of relational rules can lead to reduced trust, neg69

ative attitudes, and the deterioration of the relationship. In the worst-case scenario, such a
violation can lead to confrontation and punishment by the offended party. Interestingly, since
these plausible reactions stem from moral judgment, they are more likely to be associated
with moral condemnation or justification in terms of morality.
Since these relational codes of conduct are specific to each type of relationship, reactions
to notions of right and wrong are triggered when the rules of those relationships are violated.
However, given the implicit nature of these rules, we often do not know where the line between
morality and immorality is drawn. As a result, this study focuses only on the extreme case
of violation—switching sides—which can be understood as a violation of the “we” feeling
between friends and the rejection of the relationship altogether. Since the expectations for
trust, mutual care, honesty, and loyalty are specific to friendship, I expect that the negative
reactions to friends will be more intense than reactions to non-friends. However, given that
switching sides may also affect the relationship between allies, I also expect that they will
react negatively to an act of betrayal, although to a lesser degree than betrayal between
friends. This gives rise to my first two hypotheses:
H1a: Friends are more likely than non-friends to experience the highest reduction in
trust, positive attitude, and willingness to remain in the relationship with the other party
after an act of betrayal.
H1b: Allies are likely to experience a reduction in trust, positive attitude, and willingness
to remain in the relationship after an act of betrayal, but to a lesser degree than friends.
In addition, different degrees of violation can result in varying levels of responses. A
violation that does not break the fundamental value of the relationship, though may invite
70

some reactions, may not result in an extreme deterioration of the relationship. For example,
a friend not reciprocating a favor, not following through with their words, or siding with
another party during an argument may yield different degrees of reactions. These actions,
though all violating some rules of friendship, differ in their extremity. Not following through
with words and not immediately reciprocating goodwill may not result in a significant reaction, especially given that friendship borders between communal and exchange relationships,
making the rule of strict reciprocity questionable. However, switching sides, which amounts
to the denouncement of friendship, may be construed as a severe betrayal and lead to a
higher intensity of negative reaction. This raises additional hypotheses:
H2a: Major violations of the moral codes of friendship, such as switching sides, will result
in the most severe reactions, including a significant reduction in trust, positive attitude,
willingness to maintain the relationship with the other party.
H2b: Minor violations of the moral codes of friendship, such as not following
through with words or not reciprocating a favor, will result in less severe reactions
compared to major violations.
Since moral violations are usually “felt” before a conscious realization, emotions may
play a role in mediating the effect of a prior relationship on the ultimate reactions. Recalling
the discussion of emotional reactions in friendship and other relational violations in Chapter
1, the immediate reaction to breaches of trust and violations of relational moral codes is
often an acute negative response, such as anger, pain, or hatred [Fitness 2006]. These
different feelings may produce different outcomes following betrayal. Research has shown
that hurt is more associated with withdrawal and avoidance, while anger often triggers more
71

confrontational and provocative actions. This raises the final two hypotheses:
H3a: The feeling of hurt is more likely to mediate the impact of a prior relationship on
the reduction of trust and willingness to maintain the relationship with the betrayer.
H3b: The feeling of anger is more likely to mediate the impact of a prior relation on
the desire to punish the betrayer.
Why Survey Experiment and Why the Public?
This survey experiment is designed to examine the causal effects of prior relationships,
particularly friendships, on foreign policy decisions. Given that this chapter focuses on
identifying these causal effects, the survey experiment is deemed the most appropriate
method of inquiry. Since the last decades of the twentieth century, experiments have
gained prominence as a robust method suitable for analyzing causal effects and mechanisms
(Druckman et al., 2011; Mcdermott, 2011). Experiments can “provide precise methodological
control, unparalleled causal insight, and innovative theoretical clarification and direction”
(Mcdermott, 2011, 504). Particularly in the field of international relations, where decisionmaking processes are complex, survey experiments provide an opportunity to test variables
that are difficult to randomize in real-world settings using the language of experimentation.
Additionally, experiments allow IR scholars to delve into human cognition and psychology,
which are crucial for understanding foreign policy decision-making (Hyde, 2015).
However, given that this work aims to understand foreign policy decision-making, one
might question the suitability of using a survey experiment on the general public for this
purpose. Concerns may arise regarding the external validity of survey experiments in studies
72

on foreign policy and international relations. Samples recruited from online platforms may
broadly reflect the characteristics of a country’s population. However, much of the research
on foreign policy and international relations focuses on the behavior of policymakers and
political elites, who may behave differently from the general public (Hafner-Burton et al.,
2013). If cognitive and behavioral differences exist between experienced and less experienced
individuals, using convenience samples of inexperienced subjects might be inappropriate.
Despite these concerns, a growing body of literature in international relations employs
survey experiments with convenience samples to explore decision-making processes within
the realm. Studies that substantiate the behavioral and cognitive differences between experts
and non-experts are predominantly conducted in fields other than international relations and
politics. Moreover, even if such differences exist within international relations, their impact
might not be particularly significant for two reasons.
First, international politics is inherently complex, involving numerous actors whose interactions can lead to a variety of outcomes. Even if experts are more adept at applying
heuristics than non-experts, the complexity of international relations might still impede their
ability to predict unforeseeable events. Numerous studies demonstrate that policymakers’
predictive abilities are often constrained by biases and emotions (for example, see Jervis,
2017; Khong, 1992; Johnson and Tierney, 2019, 2011; Rapport, 2013). Thus, while experts
may be more informed, they are not infallible, suggesting that findings from non-expert
samples can still provide valuable insights into the dynamics of international relations.
Second, not every policymaker possesses expertise in the field of international relations.
An elected official’s proficiency in domestic politics does not necessarily translate to skills
73

in international politics. While some policymakers may have more experience and a better
ability to interpret international situations (Hafner-Burton et al., 2013), even these individuals are suceptible to the overconfidence trap, often overestimating the accuracy of their
predictions (Tetlock, 2006). Consequently, they may be less inclined to seek new information.
In situations where emotions are heightened, these experts might rely on intuition, trusting
their own judgments despite potential inaccuracies.
To substantiate the appropriateness of non-experts sample further, Hafner-Burton et.
(2014) conducted a survey experiment using a convenience sample of university students and
compared it with a unique sample of actual U.S. policy elites. The treatments in both experiments were identical, enabling the researcher to assess the external validity of the student
samples. The results showed similar effects of behavioral traits on cooperation preferences
among both students and U.S. policy elites. This indicates that non-expert samples can provide valuable insights into the behaviors and decisions of policymakers, suggesting that such
studies can meaningfully contribute to our understanding of foreign policy decision-making.
Even if findings derived from a convenience sample of the inexperienced public do not
closely match the behaviors and cognitive functions of policy elites, there remains significant
value in studying public behaviors. In scenarios where the public strongly supports a
particular policy, elected officials may face serious consequences for opposing these popular
views. As suggested by the literature on domestic audience costs, governments that do
not follow through on their threats to use force against an opponent will be punished by
their domestic constituencies. In democracies, the penalty for saying one thing and doing
another often manifests as failure in re-election. While the form of punishment might vary
74

in autocracies, it does not imply that domestic audience costs are exclusive to democratic
systems (Weeks, 2008). Furthermore, in times when a society experiences strong shared
emotions, the public may exert even greater pressure on the government to undertake specific
actions. Thus, using a convenience sample to explore the causal relationships between
friendship, betrayal, and the reduction of trust and desire to punish is justified.
The Experiment
Survey Sample
I conducted a survey with 501 American adults recruited through the online survey service
provider Prolific, utilizing quota sampling to ensure representation across genders and political ideologies. This approach yielded a gender-balanced sample comprising 50% male
(n=253), 49% female (n=243), and 1 percent third-gender respondents (n=5). In terms of
political ideologies, the sample includes 23% ideologically left (n=114), 20% left-leaning
(n=101), 23% center (=119), 19% right-leaning (n=93), and 15% ideologically right respondents (n=74). Regarding party affiliation, 40% of respondents (n=200) reported that
they considered themselves Democrats, 28% of respondents (n=142) considered themselves
Republicans, and 32% (n=159) considered themselves independents. However, among those
who considered themselves independent, 60% leaned more towards the Democratic Party,
while 40% (n=64) leaned more towards the Republican Party. Complete demographic
information of the sample can be found in the Appendix.
75

Experimental Design
This experiment employs a 3x3 factoral design, wherein the relationship between the United
States and another country constitutes one dimension, and the other country’s response to
the United States forms the other dimension (Figure 3.1). Each respondent is randomly
assigned to a relationship factor and a response factor twice, as they are assigned to read
two international scenarios involving the United States and hypothetical countries. The
first scenario pertains to a hardcore international security situation involving a military
intervention by a third country. The second scenario, though still within the security
domain, focuses on U.S. arms trade with another country. By exposing respondents to
both scenarios, we can assess reactions to both hardcore and non-hardcore security issues.
The order of the scenarios is randomized to eliminate the possibility of exposure to one
scenario affecting the impression of the other.
Scenario A: Military Intervention
Each respondent is randomly assigned to read one of the three prime vignettes, which vary
the relationship between the United States and Country A. In vignette 1, respondents are
informed that Country A has diplomatic relations with the United States, but they neither
share a long history with nor are they treaty allies of the United States (“control prime”).
As Country A neither has a contractual relationship nor friendship with the United States,
respondents primed with this condition are expected to be more neutral in terms of their trust
in and their feelings towards Country A. Regarding whether they want the United States to
continue collaborating with Country A, they are expected to be neutral about the question.
76

Relationship dimension
R
esp
o
nse
dim
e
nsio
n
Figure 3.1: The 3x3 factoral design
In vignette 2, respondents are informed that Country A and the United States have
diplomatic relations with a limited shared history. However, they are treaty allies with a
commitment to help defend each other in times of attack (“ally prime”). This condition
represents a contractual, transactional relationship in international relations. Consequently,
respondents primed with this condition are more likely to have greater trust in and feel more
positively towards Country A compared to respondents who received the control prime.
Additionally, they are expected to be more likely to express a more favorable view of the
continuity of cooperation between the United States and Country A.
77

Lastly, vignette 3 informs respondents that the United States shares diplomatic relations
and a long history of friendship with Country A. Respondents are also informed that they
are treaty allies with a commitment to help defend each other in times of attack (“friend
prime”). Given that this condition introduces a historical friendship atop a contractual
and transactional relationship, respondents primed with this condition are expected to
demonstrate the highest levels of trust and the most positive attitudes towards Country A
compared to respondents who received the other two primes. Additionally, they are expected
to be more inclined towards the United States continuing collaboration with Country A. The
exact wordings of the vignettes are presented in Table 3.1.
Table 3.1: Randomly assigned vignettes varying the relationship between the United States and
Country A
Intro
(all)
In this section, we are going to describe the relationship between the United
States and another country. After describing the relationship, we will ask you a
few questions.
Control
prime
Country A and the United States have diplomatic relations, but they lack a long
history together, and they are not treaty allies. They are considered regular
partners without any special or deep connection.
Ally
prime
Country A and the United States have diplomatic relations, with a limited
shared history. However, they are treaty allies with a commitment to help defend
each other in times of attack.
Friend
prime
Country A and the United States share diplomatic relations and a long history of
friendship, having fought in many wars together. The United States can always
count on Country A in times of hardship. They are also treaty allies with a
commitment to help defend each other in times of attack.
After reading the first vignette, respondents are tasked with two essential checks: an
attention check and a manipulation check. The attention check requires respondents to
identify the nature of the relationship between the United States and Country A. Failure to
78

accurately identify this relationship prompts respondents to revisit the vignette and retry
the attention check. Progression beyond this stage is contingent upon a correct response to
the attention check. Subsequently, respondents are instructed to undergo the manipulation
check, where they are prompted to evaluate the extent to which they perceive Country A
as a friend of the United States. The intuition is that respondents in the friend condition
will express the highest level of this statement, whereas those in the control condition will
exhibit the lowest level of agreement.
Following the completion of the attention and manipulation checks, respondents will
proceed to answer a series of pre-test measurements. These inquiries are designed to gauge
the level of trust and positive feelings towards Country A among respondents primed with
varying relationship conditions. Furthermore, respondents are prompted to express their
inclination towards continued collaboration between the United States and Country A. These
pre-test measurements serve a dual purpose. Firstly, they enable the comparison of respondents’ responses across different relationship conditions. Secondly, they establish a
baseline for subsequent pre- and post-test comparisons once the second dimension of the
experiment is introduced to respondents. A summary of the expected outcomes of the pretest measurements is outlined in Table 3.2.
Table 3.2: The expected outcomes of the pre-test measurements in Scenario A
Pre-test Expectation
Trust in country A • Respondents in the ally condition are more likely to express
a higher level of trust in Country A than those in the control
condition.
• Respondents in the friend condition are more likely to
express a higher level of trust than those in the control and
ally conditions.
79

Pre-test Expectation
Positive feelings
towards Country A
• Respondents in the ally condition are more likely to express
a higher level of positive feelings towards Country A than
those in the control condition.
• Respondents in the friend condition are more likely to
express a higher level of positive feelings towards Country
A than those in the control and ally conditions.
Willingness for the
United States to
continue collaborating
with Country A
• Respondents in the ally condition are more likely to express
a higher level of willingness for the United States to continue
collaborating with Country A than those in the control
condition.
• Respondents in the friend condition are more likely to
express a higher level of willingness for the United States
to continue collaborating with Country A than those in the
control and ally conditions.
The experiment then proceeds to its second part, where respondents are presented with
an international event involving the United States and Country A. Scenario A depicts a
situation in which a territory hosting a U.S. military facility has been invaded by a foreign
country. Consequently, the U.S. has formally requested Country A to join efforts in defending
the invaded territory. Respondents are randomly assigned to one of three prime vignettes,
which vary in Country A’s response to the U.S. request. Figure 3.2 shows the variations
on the two dimensions.
In vignette A, respondents are informed that Country A is considering the request from
the United States but has not yet provided a definite answer (“control prime”). Since this
vignetted does not present respondents with a scenario where Country A refuses to assist the
United States, it is not anticipated that respondents’ levels of trust in and feelings towards
Country A will be significantly different from their assessments in the pre-test. Similarly,
80

Relationship dimension
R
esp
o
nse
dim
e
nsio
n
C
o
ntrol R
efusal Betra
y
al
Control Ally Friend
Figure 3.2: The variations on the two dimensions
the willingness of the United States to continue collaborating with Country A across all
relationship conditions is also not expected to be significantly different from the pre-test level.
In vignette B, respondents are informed that Country A has declined to participate in
the U.S. efforts to protect the invaded territory (“refusal prime”). This condition presents
respondents with a situation where Country A rejects the U.S. request, potentially leading
respondents primed with this condition to exhibit lower levels of trust and less positive feelings towards Country A compared to those in the control condition. Moreover, respondents
in this condition are less likely to express a desire to continue cooperation with Country
81

A in the future. However, the extent of the shift in trust levels, positive feelings, and
willingness to cooperate with Country A depends on the relationship condition assigned to
respondents in the initial stage of randomization. The most significant shift in trust levels,
positive feelings, and willingness to cooperate is expected among respondents exposed to the
“friend prime,” followed by those who received the “ally prime.” Conversely, the smallest
shift is anticipated among respondents in the control group, as no specific expectations for
loyalty were presented in this condition.
Vignette C presents a scenario in which respondents are informed that Country A has
rejected the U.S. request and has instead joined the invading force (“betrayal prime”). This
Scenario embodies the pinnacle of betrayal in international relations, where an ally or friend
not only declines assistance but also aligns with the enemy. Consequently, respondents exposed to this condition are expected to express the lowest levels of trust, positive feelings, and
willingness for the United States to continue cooperation with Country A. The magnitude
of this shift depends on the relationship condition randomly assigned to respondents in the
initial stage. Similarly to the anticipated shift in the refusal condition, respondents assigned
to the friend condition in the first stage are expected to exhibit the most significant negative
shift in levels of trust, positive feelings, and willingness for the United States to continue
cooperating with Country A. Those in the ally condition are anticipated to demonstrate a
smaller shift compared to those in the friend condition but still larger than respondents in
the control group. However, the extent to which the level of reduced trust, positive feelings,
and willingness to continue collaboration with Country A among respondents in the friend
condition is significant enough to bring their post-test levels of trust, positive sentiments,
82

and willingness to continue collaboration on par with the responses from respondents form
the other relationship remains an empirical question. The exact wordings of the three
vignettes are presented in Table 3.3.
Table 3.3: Randomly assigned vignettes varying Country A’s response to the U.S. request in
ScenarioA
Intro
(all)
Next, we will introduce you to a Scenario facing the United States and Country
A. After describing the scenario, we will ask you about your preferred foreign
policy options.
Control
prime
A foreign country has invaded a territory hosting a United States military base.
The U.S. is interested in safeguarding the sovereignty of the affected territory
and its military installation. Consequently, the U.S. has formally requested
Country A to join the effort to defend the invaded territory.
Country A is considering the request from the United States but has not given a
definite answer yet.
Refusal
prime
A foreign country has invaded a territory hosting a United States military base.
The U.S. is interested in safeguarding the sovereignty of the affected territory
and its military installation. Consequently, the U.S. has formally requested
Country A to join the effort to defend the invaded territory.
Country A has declined to participate in the joint effort to protect the invaded
territory.
Betrayal
prime
A foreign country has invaded a territory hosting a United States military base.
The U.S. is interested in safeguarding the sovereignty of the affected territory
and its military installation. Consequently, the U.S. has formally requested
Country A to join the effort to defend the invaded territory.
Country A has declined to participate in the joint effort to protect the invaded
territory. Instead, it has decided to collaborate with the invading force.
After reviewing the response from Country A, participants are prompted to undergo
another attention check. This check assesses their comprehension of the scenario assigned
to them. If respondents are unable to accurately identify Country A’s response to the
U.S. request, they are directed to revisit the vignette and undergo the attention check
83

once more. Progress beyond this point is contingent upon successfully passing the check.
Subsequently, respondents are tasked with answering a series of post-test questions. These
questions aim to gauge shifts in respondents’ levels of trust and positive feelings towards
Country A, as well as their inclination for continued collaboration between the United
States and the aforementioned country.
In addition to the three questions utilized for pre-test measurement, respondents are also
prompted to report the intensity of their feelings of happiness, anger, betrayal, surprise, and
hurt in response to Country A’s decision on a 6-point scale. Additionally, they are asked
whether they desire to impose punishment on Country A for its decision (yes/no). Furthermore, respondents are asked to indicate their level of agreement (on a 6-point scale) with
various means of punishment, such as publicly shaming Country A, suspending diplomatic
relations with Country A, imposing economic sanctions against Country A, and withdrawing
military cooperation with Country A.
Lastly, they are asked to indicate the purpose behind punishing Country A. In this last
question, four options are provided to respondents: (a) to stabilize the international system;
(b) to make Country A pay for it; (c) to protect other countries; (d) to serve as a deterrent
to others. These options, adapted from Mooijman et al. (2015), aim to capture respondents’
motives for punishment. Option (c) and particularly (d) represent a deterrence motive.
According to this rationale, punishment is administered to dissuade the other country from
engaging in further destructive actions. Conversely, option (b) represents the motive of just
deserts, wherein punishment is carried out to seek retribution against the other country and
rectify perceived injustices. Since friendships entail strong relational bonds, respondents in
84

the friend condition are expected to be more inclined to pursue punishment for the just
deserts motive. In contrast, allies, bound by contractual and transactional relationships, are
more likely to pursue punishment for deterrence. This is because there exists an incentive
to prevent a counterpart from violating contractual obligations in the future. A summary of
post-test measurement expectations is provided in Table 3.4, while the survey instruments
utilized in this experiment are located in Appendix A.
85

Table 3.4: The expected outcomes of the post-test measurements in Scenario A
Post-test Expectation
Trust in
country A
Control prime
• The post-test levels of trust in Country A among respondents in all
three relationship conditions is not expected to differ significantly
from the pre-test levels.
• The shifts in the level of trust in Country A among respondents across
the three relationship conditions are not expected to be significantly
different.
Refusal prime
• The post-test level of trust in Country A among respondents in the
control condition is not expected to differ significantly from the pretest level, while the post-test levels of trust among respondents in the
ally and friend conditions are expected to be lower than the pre-test
levels.
• The shifts in the level of trust in Country A among respondents in
the friend condition are expected to be higher than the shift in the
level of trust in Country A among respondents in the ally condition.
Betrayal prime
• The post-test level of trust in Country A among respondents in the
control condition is not expected to differ significantly from the pretest level, while the post-test level of trust among respondents in the
ally and friend condition is expected to be lower than the pre-test
level.
• The shifts in the level of trust in Country A among respondents in
the friend condition are expected to be greater than the shift in the
level of trust in Country A among respondents in the ally condition.
• The shifts in the level of trust in Country A among respondents subjected to the betrayal prime under both ally and friend conditions are
expected to be greater than the shifts observed among respondents
subjected to the refusal prime with the same conditions.
86

Post-test Expectation
Positive
feelings
towards
Country A
Control prime
• The post-test level of positive feelings towards Country A among
respondents in all three relationship conditions is not expected to
differ significantly from the pre-test level.
• The shifts in the level of positive feelings towards Country A among
respondents across the three relationship conditions are not expected
to be significantly different.
Refusal prime
• The post-test level of positive feelings towards Country A among
respondents in the control condition is not expected to differ significantly from the pre-test level, while the post-test levels of positive
feelings among respondents in the ally and friend conditions are
expected to be lower than the pre-test levels.
• The shifts in the level of positive feelings towards Country A among
respondents in the friend condition are expected to be greater than
the shift in the level of positive feelings in Country A among respondents in the ally condition.
Betrayal prime
• The post-test level of positive feelings towards Country A among
respondents in the control condition is not expected to differ significantly from the pre-test level, while the post-test levels of positive
feelings among respondents in the ally and friend condition are
expected to be lower than the pre-test level.
• The shifts in the level of positive feelings towards Country A among
respondents in the friend condition are expected to be greater than
the shift in the level of positive feelings towards Country A among
respondents in the ally condition.
• The shifts in the level of positive feelings towards Country A among
respondents subjected to the betrayal prime under both ally and
friend conditions are expected to be greater than the shifts observed
among respondents subjected to the refusal prime with the same
conditions.
87

Post-test Expectation
Willingness for
the United
States to
continue
collaborating
with Country
A
Control prime
• The post-test levels of desire for the United States to continue
collaborating with Country A among respondents across the three
relationship conditions are not expected to be significantly different
from the pre-test levels.
• The shifts in the levels of desire for the United States to continue
collaborating with Country A among respondents across the three
relationship conditions are not expected to be significantly different.
Refusal prime
• The post-test level of desire for the United States to continue collaborating with Country A among respondents in the control condition
is not expected to differ significantly from the pre-test level, while
the post-test levels of such desire among respondents in the ally and
friend conditions are expected to be lower than the pre-test levels.
• The shifts in the level of desire for the United States to continue
collaborating with Country A among respondents in the friend condition are expected to be greater than the shift in the level of such
desire in Country A among respondents in the ally condition.
Betrayal prime
• The post-test desire level for the United States to continue collaborating with Country A among respondents in the control condition
is not expected to differ significantly from the pre-test level, while
the post-test levels of such desire among respondents in the ally and
friend condition are expected to be lower than the pre-test level.
• The shifts in the level of desire for the United States to continue
collaborating with Country A among respondents in the friend condition are expected to be greater than the shift in the level of such
desire among respondents in the ally condition.
• The shifts in the level of desire to continue collaborating with Country
A among respondents subjected to the betrayal prime under both
ally and friend conditions are expected to be greater than the shifts
observed among respondents subjected to the refusal prime with the
same conditions.
88

Post-test Expectation
Willingness to
impose
punishment on
Country A
• Respondents in the ally condition and friend condition are more likely
than respondents in the control condition to impose punishment on
Country A.
Reason for
punishment
• Respondents in the ally condition are more likely to exhibit the
deterrence motive as the rationale for punishment, whereas those in
the friend condition are more inclined to demonstrate the just deserts
motive as the rationale for punishment.
Scenario B: Arms Deal
The arms deal scenario was crafted to mirror the controversy surrounding the submarine deal
between France and Australia. In 2011, the Australian government terminated a submarine
contract with France in favor of acquiring more advanced nuclear submarines from the United
States. This cancellation infuriated the French government and triggered a diplomatic rift
not only between France and Australia but also between France and the Untied States. The
Australian government’s decision was labeled by the Macron administration as a “stab in
the back”-- a term indicating their fury caused by betrayal. This Scenario illustrates that
even events not directly related to hardcore security issues, which are typically considered
the primary national interest in international relations, can still be perceived as acts of
betrayal, resulting in negative consequences.
This scenario commences similarly to the other scenario. Each respondent is randomly
assigned to read one of the three prime vignettes, which varies the relationship between
the United States and Country B. In vignette 1, respondents are informed that Country
B has diplomatic relations with the United States but does not share a long history of
89

friendship nor are they treaties allies of the United States (“control prime”). In vignette
2, respondents are informed that Country B is a treaty ally of the United States, but they
did not share a long history of friendship with the United States (“ally prime”). In vignette
3, respondents are informed that the United States shares diplomatic relations and a long
history of friendship with Country B. Additionally, the two countries are also treaty allies
with a commitment to help defend each other in times of attack (“friend prime”). The exact
wordings of the vignettes are presented in Table 3.5.
Table 3.5: Randomly assigned vignettes varying the relationship between the United States and
Country B
Intro
(all)
In this section, we are going to describe the relationship between the United
States and another country. After describing the relationship, we will ask you a
few questions.
Control
prime
Country B and the United States have diplomatic relations, but they lack a long
history together, and they are not treaty allies. They are considered regular
partners without any special or deep connection.
Ally
prime
Country B and the United States have diplomatic relations, with a limited
shared history. However, they are treaty allies with a commitment to help defend
each other in times of attack.
Friend
prime
Country B and the United States share diplomatic relations and a long history of
friendship, having fought in many wars together. The United States can always
count on Country B in times of hardship. They are also treaty allies with a
commitment to help defend each other in times of attack.
After exposure to the relationship vignette, respondents are presented with an attention
check and a manipulation check akin to Scenario A. The attention check requires respondent
to specify the type of relationship between the United States and Country B assigned to
them. Progress beyond this point is contingent upon a correct response to this question.
In the event of an incorrect identification of the relationship type between the United
90

States and Country B, respondents are prompted to reread the assigned vignette. The
manipulation check then prompts respondents to indicate the extent to which they perceive
Country B as a friend of the United States. Subsequently, they are asked to complete pretest measurements similar to those in Scenario A, designed to gauge respondents’ level of
trust in Country B, positive feelings towards Country B, and desire for the United States to
sustain cooperation with Country B. A summary of the expected outcomes of the pre-test
measurements is presented in Table 3.6.
Table 3.6: The expected outcomes of the pre-test measurements in scenario B
Pre-test Expectation
Trust in country B • Respondents in the ally condition are more likely to express
a higher level of trust in Country B than those in the control
condition.
• Respondents in the friend condition are more likely to
express a higher level of trust than those in the control and
ally conditions.
Positive feelings
towards Country B
• Respondents in the ally condition are more likely to express
a higher level of positive feelings towards Country B than
those in the control condition.
• Respondents in the friend condition are more likely to
express a higher level of positive feelings towards Country
B than those in the control and ally conditions.
Willingness for the
United States to
continue collaborating
with Country B
• Respondents in the ally condition are more likely to express
a higher level of willingness for the United States to continue
collaborating with Country B than those in the control
condition.
• Respondents in the friend condition are more likely to
express a higher level of willingness for the United States
to continue collaborating with Country B than those in the
control and ally conditions.
Upon completion of the pre-test measurements, respondents are then randomized again.
91

The second set of vignettes present respondents with Country B’s varying responses to
an arms deal it had previously agreed upon with the United States. Vignette A informs
respondents that Country B is sticking to the arms deal and is waiting for the weapons
to be transferred (“control”). Since a violation of commitment is not presented under
this scenario, respondents are less inclined to exhibit a lower level of trust in Country B,
positive feelings towards Country B, and the willingness of the United States to continue
collaborating with the United States.
In vignette B, respondents are informed that Country B has changed its mind and
cancelled the purchase of weapons from the United States (“cancellation prime”). This
scenario bears resemblance to the refusal prime in Scenario A, where Country A declines
to assist the United States in defending a territory with the U.S. military installments.
Consequently, respondents assigned to this scenario are more inclined to exhibit diminished
levels of trust in Country B, reduced positive feelings towards Country B, and a decreased
desire for the United States to continue collaborating with Country B compared to their
responses in the pre-test measurements. However, the extent of the shifts in the posttest measurements from the pre-test measurements depends on the relationship condition
previously assigned to respondents. The degree of shifts among respondents in the friend
condition is expected to be most pronounced compared to the other two conditions, while
the degree of the shifts among respondents in the ally condition is expected to be greater
than those in the control group.
Vignette C informs respondents that, in addition to Country B’s changing its mind
and cancelling the purchase from the United States, it has agreed to buy the same type
92

of weapons from another country (“betrayal prime”). Given that Australia’s decision to
purchase more advanced weapons from the United States could cause diplomatic disruptions
between France and Australia, it is even more likely that Country B’s decision to change
its mind and opt for the same type of weapons from an alternative source would elicit
dissatisfaction among respondents. Consequently, respondents assigned to this condition are
more likely to report negative shifts in the level of trust in Country B, reduced positive
feelings towards Country B, and decreased willingness for the United States to continue
collaborating with Country B compared to their responses to the pre-test measurements.
These shifts among respondents in this condition are expected to be greater than the shifts
among those in the cancellation condition. However, the extent of the shifts is expected to
vary among respondents assigned to different relationship condition with the shifts among the
respondents in the friend condition being the greatest. The exact wordings of the vignettes
varying Country B’s response to the arms deal are illustrated in Table 3.7.
Table 3.7: Randomly assigned vignettes varying Country B’s response to the U.S. request in scenario
B
Intro
(all)
Next, we will introduce you to a scenario facing the United States and
Country B. After describing the scenario, we will ask you about your
preferred foreign policy options.
Control
prime
A few years ago, Country B agreed to purchase $60 billion worth of
advanced weapons from the United States.
This year, Country B is sticking to the plan and is waiting for the weapons
to be transferred.
Refusal
prime
A few years ago, Country B agreed to purchase $60 billion worth of
advanced weapons from the United States.
This year, Country B has changed course and canceled the weapons purchase.
93

Betrayal
prime
A few years ago, Country B agreed to purchase $60 billion worth of
advanced weapons from the United States.
This year, Country B has changed course and canceled the weapons purchase.
Instead, it has purchased the same type of weapons from another country.
Following the review of Country B’s response, respondents are instructed to complete an
additional attention check, which requires them to accurately identify Country B’s stance
on the arms deal. If respondents fail to correctly identify Country B’s response, they are
prompted to revisit the vignette and repeat the attention check until they can accurately
identify the correct scenario. Respondents are unable to proceed until they correctly identify
Country B’s response. Subsequently, respondents are asked to answer a series of post-test
questions. These questions emcompass the three questions utilized as pre-test measurements
and additional inquiries aimed at assessing respondents’ willingness to punish Country B for
its action. Analogous to the questions in Scenario A, respondents are requested to report
the intensity of their feelings of happiness, anger, betrayal, surprise, and hurt in response to
Country B’s decision using a 6-point scale. Furthermore, they are asked to indicate whether
they desire to impose punishment on Country B for its decision (yes/no), their agreement
with particular means of punishment (6-point scale), and the rationale behind their decision
for punishment. The expected outcomes of the post-test measurements are anticipated to
be rather akin to those in Scenario A and is shown in Table 3.8.
94

Table 3.8: The expected outcomes of the post-test measurements in Scenario B
Post-test Expectation
Trust in
country B
Control prime
• The post-test levels of trust in Country B among respondents in all
three relationship conditions is expected to improve from the pre-test
levels.
• The shifts in the level of trust in Country B among respondents across
the three relationship conditions are not expected to be significantly
different.
Refusal prime
• The post-test level of trust in Country B among respondents in the
control condition is not expected to differ significantly from the pretest level, while the post-test levels of trust among respondents in the
ally and friend conditions are expected to be lower than the pre-test
levels.
• The shifts in the level of trust in Country B among respondents in
the friend condition are expected to be higher than the shift in the
level of trust in Country B among respondents in the ally condition.
Betrayal prime
• The post-test level of trust in Country B among respondents in the
control condition is not expected to differ significantly from the pretest level, while the post-test level of trust among respondents in the
ally and friend condition is expected to be lower than the pre-test
level.
• The shifts in the level of trust in Country B among respondents in
the friend condition are expected to be greater than the shift in the
level of trust in Country B among respondents in the ally condition.
• The shifts in the level of trust in Country B among respondents subjected to the betrayal prime under both ally and friend conditions are
expected to be greater than the shifts observed among respondents
subjected to the refusal prime with the same conditions.
95

Post-test Expectation
Positive
feelings
towards
Country B
Control prime
• The post-test level of positive feelings towards Country B among
respondents in all three relationship conditions is expected to improve
from the pre-test level.
• The shifts in the level of positive feelings towards Country B among
respondents across the three relationship conditions are not expected
to be significantly different.
Refusal prime
• The post-test level of positive feelings towards Country B among
respondents in the control condition is not expected to differ significantly from the pre-test level, while the post-test levels of positive
feelings among respondents in the ally and friend conditions are
expected to be lower than the pre-test levels.
• The shifts in the level of positive feelings towards Country B among
respondents in the friend condition are expected to be greater than
the shift in the level of positive feelings in Country B among respondents in the ally condition.
Betrayal prime
• The post-test level of positive feelings towards Country B among
respondents in the control condition is not expected to differ significantly from the pre-test level, while the post-test levels of positive
feelings among respondents in the ally and friend condition are
expected to be lower than the pre-test level.
• The shifts in the level of positive feelings towards Country A among
respondents in the friend condition are expected to be greater than
the shift in the level of positive feelings towards Country B among
respondents in the ally condition.
• The shifts in the level of positive feelings towards Country B among
respondents subjected to the betrayal prime under both ally and
friend conditions are expected to be greater than the shifts observed
among respondents subjected to the refusal prime with the same
conditions.
96

Post-test Expectation
Willingness for
the United
States to
continue
collaborating
with Country
B
Control prime
• The post-test levels of desire for the United States to continue
collaborating with Country B among respondents across the three
relationship conditions are expected to improve from the pre-test
levels.
• The shifts in the levels of desire for the United States to continue
collaborating with Country B among respondents across the three
relationship conditions are not expected to be significantly different.
Refusal prime
• The post-test level of desire for the United States to continue collaborating with Country ABamong respondents in the control condition
is not expected to differ significantly from the pre-test level, while
the post-test levels of such desire among respondents in the ally and
friend conditions are expected to be lower than the pre-test levels.
• The shifts in the level of desire for the United States to continue
collaborating with Country B among respondents in the friend condition are expected to be greater than the shift in the level of such
desire in Country B among respondents in the ally condition.
Betrayal prime
• The post-test desire level for the United States to continue collaborating with Country B among respondents in the control condition
is not expected to differ significantly from the pre-test level, while
the post-test levels of such desire among respondents in the ally and
friend condition are expected to be lower than the pre-test level.
• The shifts in the level of desire for the United States to continue
collaborating with Country B among respondents in the friend condition are expected to be greater than the shift in the level of such
desire among respondents in the ally condition.
• The shifts in the level of desire to continue collaborating with Country
B among respondents subjected to the betrayal prime under both
ally and friend conditions are expected to be greater than the shifts
observed among respondents subjected to the refusal prime with the
same conditions.
97

Post-test Expectation
Willingness to
impose
punishment on
Country B
• Respondents in the ally condition and friend condition are more likely
than respondents in the control condition to impose punishment on
Country B.
Reason for
punishment
• Respondents in the ally condition are more likely to exhibit the
deterrence motive as the rationale for punishment, whereas those in
the friend condition are more inclined to demonstrate the just deserts
motive as the rationale for punishment.
Procedure
The experiment was conducted online on February 18, 2024. Respondents recruited through
Prolific were required to provide consent for participation in the study before proceeding.
Subsequently, upon providing consent, respondents were asked to furnish basic demographic
information, including their gender, year of birth, race, highest level of education attained,
political ideology, party affiliation, the region of the United States from which they originate.
After completing the demographic information, they were directed to complete a survey
battery regarding their foreign policy preferences, foreign policy attentiveness, and generalized trust. The foreign policy preferences battery aimed to measure respondents’ inclination
towards isolationism, cooperative internationalism, military internationalism and trust in the
international system. The generalized trust scale was included to capture respondents’ level
of generalized trust that might influence their reactions to an act of betrayal. It is noteworthy
that some participants were randomized to complete the foreign policy preferences, foreign
policy attentiveness, and generalized trust scale questions at this stage, while others were
randomized to complete it at the end of the survey. This randomization aims to prevent the
98

effects of completing the questions from influencing the experiment and vice versa.
Respondents then proceeded to the first part of the experiment, where they were randomized to be exposed to either Scenario A or scenario B in the first round. Within each
scenario, respondents were randomly assigned to one of the three relationship vignettes.
Upon exposure to the prime vignette, respondents were asked to complete the attention
check, manipulation check, and pre-test measurement questions. They were then randomly
assigned to one of the three response vignettes and were asked to complete another set of
attention check and post-test measurement questions.
After completing the questions for the first scenario, respondents were presented with
the other scenario. Given the randomization of the order of the scenarios, some would read
Scenario A, while others would read scenario B in this second round. Then, the experiment
proceeded similarly to the first round. Respondents were randomly assigned to one of
three relationship vignettes and asked the attention check, manipulation check, and pretest measurement questions. Subsequently, they were randomly assigned to one of the three
response vignettes and were asked with another set of attention check questions and the posttest measurement questions. Upon the completion of two rounds of the study, the survey
ended. Respondents received a completion code which qualified them for $3 compensation.
Manipulation Checks
Before delving into the experiment’s findings, it is important to verify whether the vignettes
effectively manipulated the respondents as intended. The findings indicate that the manipulations were effective in both Scenario A and Scenario B. In Scenario A, the difference
99

in how respondents assigned to each relationship group perceived Country A as a friend,
on average, was statistically significant using an ANOVA test (p-value < 0.001). As we
moved from the control group to the ally group, the degree to which respondents perceived
Country A as a friend of the United States, on average, increased by 0.53 points (p-value <
0.001). Similarly, as we moved from the ally group to the friend group, the degree to which
respondents perceived Country A as a friend of the United States, on average, increased
by 0.95 points (p-value < 0.001).
The same pattern holds for scenario B. An ANOVA test reveals that the degree to which
respondents in each relationship condition perceived Country B as a friend of the United
States was statistically different (p-value < 0.001). In this scenario, the perception of Country
B as a friend of the United States shifted by 0.33 points as we transitioned from the control
group to the ally group (p-value < 0.001), and by 0.95 points as we moved from the ally
group to the friend group (p-value < 0.001). Figure 3.3 show the mean difference across the
three relationship conditions in both Scenario A and Scenario B.
Findings
Pre-test Results: Friends Trust Each Other More
The results of the pre-test confirm the experiment’s expectations. Respondents randomly
assigned to the friend condition indicated the highest level of trust towards Country A/B, the
highest level of positive feelings towards Country A/B, and the highest level of willingness to
continue collaborating with Country A/B. Compared to respondents in the friend condition,
those randomly assigned to the ally condition demonstrated a lower level of trust towards
100

1
2
3
4
5
Control Ally Friend
(a) Scenario A
1
2
3
4
5
Control Ally Friend
(b) Scenario B
Figure 3.3: The extent to which respondents perceive Country A/B as a friend of the United States,
rated on a scale of 0 (lowest) to 5 (highest)
Country A/B, a lower level of positive feelings towards Country A/B, and a lower level of
willingness to continue collaborating with Country A/B, although still higher than those in
the control condition. These findings suggest a distinct degree of closeness between allies
and friends in international relations, as detailed below.
In terms of trust, respondents randomly assigned to the friend condition exhibited the
highest level of trust towards Country A compared to respondents in the two other conditions.
Specifically, respondents in the friend condition tended to trust Country A more than those
in the ally condition by 1.17 points (p-value < 0.001), and more than those in the control
condition by 1.65 points (p-value < 0.001). Additionally, respondents in the ally condition
reported a higher level of trust in Country A than those in the control group by 0.48 points
(p-value < 0.001). An ANOVA test confirms that the mean difference among the three
101

groups was statistically significant (p-value < 0.001).
The same pattern holds true for scenario B as well. Respondents randomly assigned to
the friend condition reported trusting Country B more than those assigned to the control and
ally conditions. Specifically, respondents in the friend condition, on average, exhibited a trust
level 1.15 points higher than respondents assigned to the ally condition (p-value < 0.001),
and 1.49 points higher than those in the control condition (p-value < 0.001). Moreover,
respondents randomly assigned to the ally condition by 0.34 points (p-value < 0.001). An
ANOVA test confirms that the mean difference among the three relationship groups was
statistically significant (p-value < 0.001). Figure 3.4 shows the level of trust towards Country
A and B across all three relationship groups in both Scenario A and Scenario B.
1
2
3
4
5
Control Ally Friend
(a) Scenario A
1
2
3
4
5
Control Ally Friend
(b) Scenario B
Figure 3.4: The extent to which respondents trust Country A/B, rated on a scale of 0 (lowest) to
5 (highest)
A similar trend can be observed for respondents’ positive feelings towards Country A
102

(Scenario A) and Country B (Scenario B). In Scenario A, respondents randomly assigned
to the friend condition reported the highest positive feelings towards Country A compared
to respondents assigned to the other two groups. Specifically, respondents in the friend
condition reported their positive feelings towards Country A 0.72 points higher than respondents in the ally condition (p-value < 0.001) and 1.15 points higher than respondents in
the control condition (p-value < 0.001). Additionally, respondents randomly assigned to the
ally condition felt 0.42 points more positively towards Country A than those assigned to
the control condition (p-value < 0.001). An ANOVA test confirms that the mean difference
between the three groups was statistically significant (p-value < 0.001).
Scenario B reveals a similar finding. The difference in respondents’ positive feelings
towards Country B among the three groups was statistically significant (p-value < 0.001).
Respondents in the friend condition expressed the highest positive feelings towards Country
B, with 0.72 points higher than those in the ally condition (p-value < 0.001) and 1.05 points
higher than those in the control condition (p-value < 0.001). Moreover, respondents in the
ally condition felt more positively towards Country B than those in the control condition
by 0.32 points (p-value < 0.001). Figure 3.5 shows respondents’ positive feelings towards
Country A/B in both scenarios across all three relationship conditions.
Lastly, regarding the willingness for the United States to continue collaborating with
Country A in Scenario A and Country B in scenario B, respondents randomly assigned to
the friend condition expressed the highest level of willingness to continue collaborating with
Country A/B. In Scenario A, respondents in the friend condition indicated their willingness
to continue collaborating with Country A in the future 0.60 points higher than respondents
103

1
2
3
4
5
Control Ally Friend
(a) Scenario A
1
2
3
4
5
Control Ally Friend
(b) Scenario B
Figure 3.5: The extent to which respondents feel positive towards Country A/B, rated on a scale
of 0 (lowest) to 5 (highest)
in the ally condition (p-value < 0.001) and 0.88 points higher than respondents in the control
condition (p-value < 0.001). Similarly, respondents randomly assigned to the ally condition
expressed a 0.30-point higher desire for the United States to continue collaborating with
Country A compared to those in the control condition (p-value < 0.001).
The findings in Scenario B, however, present a slight deviation. In this scenario, the mean
difference between the ally and control groups did not achieve statistical significance (p-value
= 0.83). This outcome suggests a potential indifference among respondents regarding their
willingness for the United States to continue collaborating with a country that is either a
regular partner or an ally. Nevertheless, the finding from the friend group still confirms the
expectation that friends are more willing to stick together. Respondents randomly to the
friend condition expressed their willingness for the United States to continue collaborating
104

with Country B 0.71 points higher than those randomly assigned to the ally condition
(p-value < 0.001) and 0.76 points higher than those randomly assigned to the control
condition (p-value < 0.001).
1
2
3
4
5
Control Ally Friend
(a) Scenario A
1
2
3
4
5
Control Ally Friend
(b) Scenario B
Figure 3.6: The extent to which respondents desire the United States to continue collaborating
with Country A/B, rated on a scale of 0 (lowest) to 5 (highest)
Post-test Results: The Impact of Betrayal on Trust, Cooperation, and Willingnesss to Cooperation
Post-test measurements assess respondents’ attitudes towards Country A and B after learning
these countries’ decisions in response to the the U.S. request in Scenario A and Scenario B.
As anticipated, respondents in the refusal and betrayal conditions, on average, demonstrated
greater negative shifts in the level of trust toward both Country A and Country B compared
to respondents in the control condition. Specifically, respondents in the betrayal condition
exhibited a greater reduction in the level of trust towards Country A and Country B than
105

respondents in the refusal condition.
In Scenario A, respondents in the refusal condition, irrespective of their relationship
condition, on average, exhibited a decrease in their trust by 1.05 points. Respondents in the
betrayal condition, regardless of their relationship condition, demonstrated an even greater
decrease in their trust toward Country A, with their trust reduced by 1.17 points. This
contrasts with the control condition, wherein respondents, irrespective of their relationship
condition, experienced a reduction in their trust by just 0.5 points.
Similarly, in Scenario B, the greatest reduction of the level of trust also occurred among
respondents in the betrayal condition, with the level of trust decreasing by 1.11 points from
the pre-test measurement. This was followed by a reduction n the level of trust among
respondents in the refusal condition, which decreased by 0.73 points. Lastly, respondents in
the control condition, regardless of the relationship condition they were in, did not exhibit
a reduction in their level of trust toward Country B. On the contrary, their level of trust
increased. This is because the prompt for the control condition in this scenario indicates
that Country B maintained its decision to buy weapons from the United States and is
awaiting the weapons transfers. As a result, for the control condition, the level of trust
was shown to increase by 0.19 points.
Moreover, when considering the effect of relationship on the changing level of trust
towards Country A and Country B, the trust towards Country A and Country B among
respondents in the friend group experienced the greatest reduction. This confirms our
expectation that respondents in the friend group would elicit the greatest reduction of trust
after an incident of betrayal. Figure 3.7 shows the comparison of the pre- and post-test results
106

of the level of trust towards Country among respondents in the three relationship groups. In
Scenario A, respondents in the friend condition experienced the greatest reduction of trust
in both the refusal and betrayal conditions. Specifically, respondents in the friend condition,
upon learning that Country A declined to help the U.S. defend the invaded territory, on
average, reported a 1.67-point reduction of trust towards Country A compared to a 0.96-
point reduction in the ally condition (p-value < 0.001) and a 0.52-point reduction in the
control condition (p-value < 0.001). Similarly, respondents in the friend condition, upon
learning that Country A had switched sides, on average, reported a 2.67-point reduction of
trust towards Country A compared to a 1.70-point reduction among respondents in the ally
condition (p-value < 0.001) and a 1.27-point reduction among respondents in the control
condition (p-value < 0.001).
The same pattern can also be observed in Scenario B, although it is less pronounced
due to the nature of the issue. Upon learning that Country B refused to follow through
with its weapon order from the United States, respondents in the friend condition, on
average, reported a decrease in their trust level towards Country B by 1 point. In contrast,
respondents in the ally and control conditions reported reductions of 0.57 points (p-value <
0.1) and 0.64 points (p-value < 0.05), respectively. Similarly, upon learning that Country B
had declined to follow through with the order and decided to buy the same weapons from
another country, respondents in the friend condition reported a reduction in trust as high
as 1.46 points, compared to a 0.85 point reduction in the control condition (p-value < 0.01)
and 1.02 points in the ally condition (p-value < 0.1).
Regarding respondents’ positive feelings towards Country A and Country B, the findings
107

Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
(a) Scenario A
Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
(b) Scenario B
Figure 3.7: The shift in respondents’ trust towards Country A/B after learning about its response
to the U.S. requests, rated on a scale of 0 (lowest) to 5 (highest)
108

demonstrated a similar pattern.
Table 3.9: (Scenario A) The effects of prior relationship and Country A’s decision on the reductions
of trust, positive attitude, and willingness to collaborate with Country A in the future
Trust Positive attitude Continue collaboration
(Intercept) -0.129 -0.452*** -1.194***
(0.120) (0.130) (0.129)
Ally -0.507** -0.330+ -0.243
(0.176) (0.189) (0.188)
Friend -0.577** -0.411* 0.017
(0.179) (0.193) (0.192)
Refusal -0.389* -0.281 -0.306
(0.175) (0.188) (0.187)
Betrayal -1.136*** -1.406*** -1.031***
(0.181) (0.195) (0.194)
Ally x Refusal 0.065 -0.458+ -0.017
(0.255) (0.275) (0.273)
Friend x Refusal -0.572* -0.573* -0.150
(0.251) (0.271) (0.269)
Ally x Betrayal 0.068 -0.157 -0.074
(0.253) (0.273) (0.271)
Friend x Betrayal -0.825** -0.504+ -0.565*
(0.257) (0.278) (0.276)
Num.Obs. 501 501 501
R2 0.388 0.364 0.231
Note: + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
The results of the differences-in-differences analysis presented in Table 3.9 and Table 3.10
reveal a significant association, particularly evident in Scenario A, between membership
in ally or friend conditions and substantial reductions in both trust and positive attitude.
However, this correlation lacks statistical significance regarding the decrease in willingness
to continue collaboration with the U.S. in the future. Conversely, the betrayal conditions
exhibit a strong connection with greater reductions in trust, positive attitude, and willingness
to continue collaboration with the country that switched sides. Notably, when the friend
109

Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
Figure 3.8: Scenario A
Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
Figure 3.9: Scenario B
The shift in respondents’ positive feelings towards Country A/B after learning about its
response to the U.S. requests, rated on a scale of 0 (lowest) to 5 (highest)
110

Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
(a) Scenario A
Pretest Posttest
1
2
3
4
5
Control Ally Friend
Control condition
1
2
3
4
5
Control Ally Friend
Refusal condition
1
2
3
4
5
Control Ally Friend
Betrayal condition
(b) Scenario B
Figure 3.10: The shift in respondents’ desire for the United States to continue collaborating with
Country A/B after learning about its decision, rated on a scale of 0 (lowest) to 5 (highest)
111

condition interacts with the refusal and betrayal conditions, the reductions in trust, positive
attitude, and willingness to continue collaboration are amplified. Interestingly, when the
refusal and betrayal conditions interact with the friend condition do reductions in trust,
positive feelings, and willingness to continue collaboration increase; this effect is not observed
when the two conditions interact with the ally condition.
Table 3.10: (Scenario B) The effects of prior relationship and Country B’s decision on the reductions
of trust, positive attitude, and willingness to collaborate with Country B in the future
Trust Positive attitude Continue collaboration
(Intercept) 0.411*** 0.089 -0.893***
(0.122) (0.120) (0.137)
Ally -0.254 -0.214 0.315+
(0.167) (0.164) (0.188)
Friend -0.411* -0.210 0.514**
(0.171) (0.168) (0.192)
Refusal -0.977*** -0.856*** -0.340+
(0.170) (0.167) (0.190)
Betrayal -1.260*** -1.146*** -0.503*
(0.175) (0.172) (0.196)
Ally x Refusal 0.180 0.038 -0.496+
(0.240) (0.236) (0.269)
Friend x Refusal -0.023 -0.023 -0.345
(0.248) (0.243) (0.278)
Ally x Betrayal 0.084 -0.089 -0.539+
(0.246) (0.241) (0.276)
Friend x Betrayal -0.199 -0.078 -0.576*
(0.243) (0.238) (0.272)
Num.Obs. 501 501 501
R2 0.301 0.257 0.138
Note: + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
112

The Intolerable Consequences of Switching Sides
The preceding section highlighted how betrayal can significantly reduce the willingness to
collaborate with friends in the future. However, does betrayal also incite a desire to punish a
friend? The experimental results (see Figure 3.11) revealed that, across both Scenario A and
Scenario B, respondents in the friend condition were not more likely to express a willingness
to punish Country A or Country B for refusing help, declining a purchase, or switching
sides, compared to respondents in the ally and control conditions. Notably, in the refusal
condition, respondents in the friend condition were significantly less likely than those in the
ally condition to seek punishment for Country A or Country B. This finding is contrary to
our initial expectation that respondents in the friend condition would show a heightened
willingness to punish the other country following a refusal of help or an act of betrayal.
In Scenario A, only 16.7 percent of respondents in the friend condition expressed willingness to punish Country A upon learning that it had declined the U.S. request to help
protect the invaded territory, compared to 44 percent in the ally condition and 16.1 percent
in the control condition. Similarly, in scenario B, only 13 percent of respondents in the
friend condition expressed willingness to punish Country B after learning about Country B’s
refusal to follow through with its weapon order. This percentage contrasted with 22.6 percent
of respondents in the ally condition and 11.7 percent in the control condition. Notably,
being friends with another country moderated the willingness to punish that country when
it refused to help defend a territory or follow through with a deal, as opposed to the
willingness to punish allies.
However, the moderation effect significantly diminished or even disappeared in cases
113

Relationship prime Control Ally Friend
0
10
20
30
40
50
60
70
80
Control Refusal Betrayal
Percentage
Scenario A
0
10
20
30
40
50
60
70
80
Control Refusal Betrayal
Scenario B
Figure 3.11: The percentage of respondents who were willing to punish Country A/B after learning
about its decision
of betrayal. For instance, in Scenario A, 63.2 percent of respondents in the friend condition reported a willingness to punish Country A after discovering it had switched sides,
compared to 68.9 percent in the ally condition and 69.4 percent in the control condition.
Remarkably, the willingness to punish Country A in the friend condition increased by
47.5 percentage points from the refusal condition, signifying a more pronounced reaction
to betrayal by switching sides.
In Scenario B, 26.2 percent of respondents in the friend condition expressed a willingness
to punish Country B after learning that it had purchased the same type of weapons from
another country, compared to 28 percent in the ally condition and 11.3 percent in the
control condition. Here again, the willingness to punish in the friend condition rose by 13.2
114

percentage points from the refusal condition. Notably, respondents in the friend condition
showed a greater propensity to punish a friend than respondents in the control condition,
who had no ally treaty or shared history of friendship when the friend opted to buy weapons
from another source. This may suggest an expectation of loyalty that is distinctively strong
in close relationships, as opposed to those without such a background.
According to the logistic models for Scenarios A and B (see Table 3.11), having a prior
relationship seems to enhance the likelihood of respondents punishing another country. The
models for Scenario B also reveal that interactions between the ally or friend conditions with
the refusal or betrayal conditions increase the likelihood of punishing for either Country A
or Country B. Although these correlations are not statistically significant, they challenge
Mercer’s theory, which posits that ingroup members are more likely to forgive the negative
behaviors of other ingroup members by attributing such actions to situational factors (Mercer,
1996). Particularly in Scenario B, the willingness to punish Country B was higher among
respondents in the ally and friend conditions than those in the control condition, underscoring
how prior relationships can intensify reactions when expectations are not met.
Table 3.11: The logistic models showing the effect of prior relations and Country A’s and B’s
decision on respondents’ willingness to punish Country A and B
Scenario A Scenario B
(Intercept) 0.032 0.018
(0.051) (0.043)
Ally 0.131+ -0.002
(0.074) (0.059)
Friend 0.066 -0.001
(0.076) (0.060)
Refusal 0.128+ 0.099+
(0.074) (0.060)
Betrayal 0.662*** 0.095
115

Scenario A Scenario B
(0.077) (0.062)
Ally x Refusal 0.148 0.112
(0.108) (0.084)
Friend x Refusal -0.060 0.014
(0.106) (0.087)
Ally x Betrayal -0.137 0.169+
(0.107) (0.086)
Friend x Betrayal -0.128 0.150+
(0.109) (0.085)
Num.Obs. 501 501
R2 0.295 0.089
Note: + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
While friendship did not significantly impact the likelihood of respondents punishing
Country A or Country B, it is noteworthy that respondents under various relationship
conditions cited different motives for penalizing these countries. Punitive motives were
assessed with the question, “If Country A/B is punished, what should be the primary
purpose of punishment?” Respondents could choose from four options: (a) to stabilize
the international system, (b) to make Country A/B pay for it, (c) to protect other countries,
and (d) to act as a deterrent to others.
The inquiry into the motivations behind the inclination to punish is crucial, as the
willingness to impose penalties can stem from various motivations. Existing literature in
international relations identifies at least two explanations for why states punish others.
The first explanation related to deterrence, which has been a predominant doctrine in
international relations (Howard and Paret, 2008; Schelling, 2008). The logic of deterrence
emphasizes the necessity for states to demonstrate resolve and employ coercive measures
to dissuade adversaries from engaging in further destructive actions, thereby transforming
116

their aggressive intentions into more favorable behaviors.
The second logic of punishment expands on traditional deterrence, focusing not only
on altering the behavior of the targeted adversary but also on influencing other states
with similar aggressive tendencies. Bas, Coe, and Gheorghe (2024) argue that states are
more inclined to inflict punishment when multiple audiences with comparable aggressive
intentions are watching. The primary objective here extends beyond merely deterring the
direct aggressor; it also aims to serve as a deterrent to other nations, discouraging them
from embarking on similar aggressive paths.
The third logic of punishment, referred to as the “just deserts” motive by Carlsmith,
Darley, and Robinson (2002), stands in stark contrast to the previous two. This perspective
diverges from deterrence-focused logic, whether aimed at the targeted state or broader audiences. In the context of just deserts, the motivation for punishment is to exact retribution
on the aggressor as a means to rectify a perceived injustice. The justification for this form
os punishment is rooted in righting a wrong, rather than achieving some future benefit, as
highlighted by (Carlsmith et al., 2002, 285). This rationale is deeply intertwined with moral
outrage (Fiske and Tetlock, 1997; Lerner and Keltner, 2001).
Based on these logics, respondents in the alliance condition, which is a relationship based
on transactional calculations, are expected to punish in order to deter the other side from
committing the negative behavior again. This aligns with the traditional deterrence theory,
where the focus is on preventing future undesirable actions through the demonstration of
resolve and coercive capacity. In contrast, respondents in the friend condition, which is
bound by the logic of loyalty and non-calculative trust, should feel a sense of injustice when
117

the other side violates the bond of friendship. This violation not only breaches trust but
also offends the moral and ethical standards that underpin the relationship. Therefore,
these respondents would likely want to punish to right the wrong, adhering more closely
to the “just deserts” motive.
The findings offer partial support for these expectations. Although deterrence emerged
as the predominant motive for punishment across all relationship types and response conditions, a distinct pattern was observed once betrayal occurred. Respondents in the friend
condition were notably more likely to indicate that their motivation for punishment was
to make Country A and Country B “pay for it,” compared to respondents in the other
two relationship conditions.
For Scenario A, 21.1 percent of respondents in the friend condition reported that the
punishment was to make Country A “pay for it,” compared to 14 percent in the ally condition
and 10.2 percent in the control condition. Furthermore, this represented an increase of 12.77
percentage points from the refusal condition. As for Scenario B, 19.7 percent of respondents
in the friend condition who were exposed to the betrayal condition cited the “just deserts”
motive for punishment, an increase of 4.5 percentage points from the refusal condition. In
contrast, respondents in the ally condition cited the “just deserts” motive for only 8 percent,
and those in the control condition for 7.55 percent.
These results indicate that friendship intensifies the inclination towards “just deserts”
as a punishment motive, suggesting that violations of friendship bonds trigger a stronger
desire for retributive justice. This aligns with the expectation that the violation of a close,
trust-based relationship leads to a more intense emotional response and a corresponding
118

Control Ally Friend
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Control
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Refusal
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Betrayal
(a) Scenario A
Control Ally Friend
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Control
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Refusal
0
10
20
30
40
50
60
70
80
90
100
To act as a deterrent
to others
To make Country A
pay for it
To protect
other countries
To stabilize
the international system
Percentage
Betrayal
(b) Scenario B
Figure 3.12: Primary purpose of punishing Country A/B (as a percentage)
119

desire to rectify the perceived wrong.
Anger, Hurt, and Surprise: The Mediation Effects of Emotions
In the preceding sections, we observed that friends experience the most significant decline in
trust towards one another when confronted with betrayal. Similarly, friendship results in the
greatest reductions in positive attitudes towards the other country and willingness to cooperate with them in the future. However, interestingly, friends are not more likely to punish
each other than they are to punish allies. What might explain these differences? Could the
variations in trust, positive attitude, willingness to collaborate, and inclination to punish be
mediated by specific emotions that differ depending on the nature of the prior relationships?
Here, I focus more on the feelings of betrayal, anger, surprise, and hurt, as they have been
shown to be associated with betrayal and the physical reactions that follow. Existing research
on betrayal has demonstrated the relevance of hurt, pain, anger, and hatred as emotional
responses to violations of relational expectations and as determinants of subsequent physical
reactions (Leary et al., 1998; Planalp and Fitness, 1999; Fitness, 2006; Fehr and Baldwon,
1996; Morrison and Robinson, 1997). While anger leads to confrontation and a desire
to punish the betrayer, hurt can result in a reduction of trust and withdrawal from the
relationship. These negative emotions may play an important role in mediating the effects
of relational expectations and physical reactions.
Generally, Figure 3.13 shows that respondents in the friend and ally conditions were
equally likely to feel betrayed and angry toward Country A and Country B after learning that
they had refused to help the United States or had switched sides. In the refusal condition
120

Control Ally Friend
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Emotion
Control condition
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Refusal condition
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Betrayal condition
(a) Scenario A
Control Ally Friend
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Emotion
Control condition
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Refusal condition
Angry
Betrayed
Happy
Hurt
Surprised
0 1 2 3 4 5
Betrayal condition
(b) Scenario B
Figure 3.13: Respondents’ reported emotions when learning about Country A/B’s decision, rated
on a scale of 0 (lowest) to 5 (highest).
121

of scenario A, respondents in the ally condition, on average, reported feeling moderately
betrayed (mean = 2.8) and angry (mean = 2.32) by Country A, while respondents in the
friend condition reported similar levels of feeling betrayed (mean = 2.65) and angry (mean
= 2.33) toward Country A. This stands in contrast to respondents in the control condition,
who reported significantly lower levels of feeling betrayed (mean = 1.54) and angry (mean
= 1.36). The differences between the control and ally group for the feelings of betrayal and
anger were both significant at the 90 and 95 percent confidence levels, respectively. Similarly,
the differences in the feelings of betrayal and anger between the control and friend group
were also both significant at the 95 percent confidence level.
However, once Country A switched sides, respondents in the control condition reported an
increased level of feelings of anger (mean = 3.22) and betrayal (mean = 3.71). These reported
feelings of anger and betrayal were not significantly different from the feelings of anger (mean
= 3.46) and betrayal (mean = 3.97) among respondents in the ally condition, nor from the
reported feelings of anger (mean = 3.30) and betrayal (mean = 3.86) among respondents
in the friend condition. This suggests that an incident of betrayal has an independent
impact on individuals’ feelings of anger and betrayal regardless of their country’s prior
relations with the other country.
When it comes to surprise and hurt, respondents in the friend condition reported significantly higher levels of these two emotions upon learning that Country had declined the
U.S.request or switched sides compared to respondents in the other relationship conditions.
Inthe refusal condition, respondents in the friend condition, on average, reported a feeling
ofsurprise at 3.07 points and feelings of hurt at 2.47 points. These numbers were significantly
122

higher than those among respondents in the control condition, who reported their feelings
of surprise at 1.41 points (p-value < 0.001) and hurt at 1.23 points (p-value < 0.001), and
among respondents in the ally condition,who reported their feelings of surprise at 2.46 points
(p-value < 0.1) and hurt at 1.88 points (p-value = 0.12).
In the betrayal condition, even though the differences in reported feelings of surprise
and betrayal among respondents in the three conditions diminished, the reported feelings of
surprise (mean = 3.63) and hurt (mean = 3.28) were still the highest among respondents
in the friend condition compared to those in the other two conditions. Specifically, the
reported level of surprise among respondents in the friend condition was higher than those
in the control condition by 0.65 points (p-value < 0.1) and the ally condition by 0.40 points
(p-value = 0.3), respectively. Similarly, the reported feeling of hurt among respondents in
the friend condition was higher than those in the control condition by 0.77 points (p-value
< 0.1) and higher than those in the ally condition by 0.59 points (p-value = 0.17).
In Scenario B, however, the disparities in reported emotions among respondents in
different relationship conditions were less conspicuous. This can be attributed, once more, to
the domain of interaction, which holds less existential significance for the country’s security.
Certainly, losing a contract to sell weapons to another country may harm the country’s
financial gains, but it does not affect its national security, which is a prime concern in
international relations. Nevertheless, respondents in the betrayal condition, irrespective of
their relationship conditions, reported the highest levels of surprise, hurt, betrayal, and anger.
The intensity of these emotions was notably higher than among respondents in the refusal
condition. This intriguingly suggests that respondents were more emotionally sensitive when
123

incidents not only impact their country’s interests but also shift advantages to a third party.
After all, with Country B buying the same type of weapons from the third party, the U.S.
financial loss was not notably higher than the loss in the refusal condition.
Despite the findings that respondents in the betrayal condition were more likely to feel
surprised, hurt, angry, and betrayed towards Country A and Country B, it remains unclear
whether these emotions affected respondents’ reduced trust, positive attitude, willingness to
collaborate, and inclination to punish Country A and Country B. It is plausible that, while
respondents reported higher levels of negative feelings, these emotions did not influence
their decisions on whether to trust, pursue further collaboration, or punish the two countries. To determine whether emotions have mediation effects on our outcomes of interest,
I use the R package mma developed by Yu (2017) to estimate mediation effects and make
inferences. Unlike other R packages used to estimate mediation effects, mma allows for
the estimation of multiple mediators simultaneously, making it useful for phenomena with
complicated relationships.
In examining the direct effects of prior relationship and mediation effects of emotions
on reduced trust in, decreasing willingness to collaborate with, and inclination to penalize
Country A and Country B for their decisions, I subsetted the data based on whether
respondents were assigned to the refusal condition or betrayal condition. Within each
condition, I then estimated the direct and indirect effects of ally and friend against responses
of the respondents in the relationship control group, who were informed that Country A only
had diplomatic relations with the U.S. and neither an ally nor a friend of the United States.
The number of observations shown in Table 3.12, Table 3.14, Table 3.13, and Table 3.15
124

reflects the subsetted data. For example, the number of observations for the ally x refusal
condition is 106, which includes respondents who were assigned to the ally and refusal
condition and the control and refusal condition, with the latter used as a reference group.
Table 3.12: (Scenario A) The estimated mediation effects on the reduction of trust and willingness
to collaborate with Country A from bootstrap samples using Generalized Linear Models (GLM)
with the relationship control group as the reference group (90% confidence level)
Ally x
Refusal
Friend x
Refusal
Ally x
Betrayal
Friend x
Betrayal
Trust
Total effect -0.46*** -1.16*** -0.45 -1.41
(-0.71, -0.20) (-1.51, -0.86) (-0.69, -0.18) (-1.78, -1.10)
DE: Prior relationship -0.12 -0.60*** -0.40 -1.30
(-0.39, 0.17) (-0.94, -0.25) (-0.63, 0.16) (-1.68, -0.95)
IE: Angry 0.08 -0.21*** -0.03
(-0.11, 0.25) (-0.51, -0.03) (-0.13, 0.042)
IE: Betrayed -0.29
(-0.50, -0.11)
IE: Surprised -0.11 -0.03
(-0.29, 0.05) (-0.10, 0.02)
IE: Hurt -0.32*** -0.11
(-0.60, -0.05) (-0.24, 0.00)
Willingness to
collaborate
Total effect -0.21 -0.12 -0.32+ -0.53**
(-0.60, 0.16) (-0.47, 0.19) (-0.60, -0.06) (-0.85, -0.19)
DE: Prior relationship 0.10 0.24 -0.23 -0.38+
(-0.17, 0.46) (-0.09, 0.56) (-0.55, 0.04) (-0.66, -0.10)
IE: Angry -0.27*
(-0.52 - -0.03)
IE: Betrayed -0.36*** -0.10 -0.05
(-0.65, -0.13) (-0.26, 0.04) (-0.19, 0.05)
IE: Surprised 0.10 -0.04
(-0.09, 0.29) (-0.16, 0.03)
IE: Hurt 0.04 -0.08 -0.15*
(-0.6, 0.19) (-0.29, 0.13) (-0.33, -0.03)
Num.Obs. 106 116 110 106
As depicted in Table 3.12, the impact of prior relationships on the reduction of trust in
125

Country A in both the ally and friendship conditions in Scenario A was partially mediated
by various negative emotions, such as anger, betrayal, surprise, and hurt. However, these
mediation effects were marginal; being an ally or a friend still significantly influenced the
reduction of trust towards Country A and the willingness to collaborate with Country A
in most cases. Only in the refusal condition were the mediation effects more pronounced
than the direct effects of being an ally on trust, and being an ally or friend on willingness to
collaborate. Particularly in the refusal condition, emotions effectively reversed the positive
sign of the direct effects of alliance and friendship, discouraging individuals from wanting
the U.S. to collaborate with Country A in the future.
Table 3.13: (Scenario B) The estimated mediation effects on the reduction of trust and willingness
to collaborate with Country B from bootstrap samples using Generalized Linear Models (GLM)
with the relationship control group as the reference group (90% confidence level)
Ally x
Refusal
Friend x
Refusal
Ally x
Betrayal
Friend x
Betrayal
Trust
Total effect -0.09 -0.43* -0.18 -0.60***
(-0.29 - 0.19) (-0.68 - 0.16) (-0.41 - 0.01) (-0.82 - -0.30)
DE: Prior relationship -0.17 -0.49*** -0.09 -0.54***
(-0.36 - 0.09) (-0.76 - -0.27) (-0.34 - 0.17) (-0.83 - -0.26)
IE: Angry -0.02 -0.38
(-0.11 - 0.01) (-0.19 - 0.01)
IE: Betrayed 0.08 0.09 0.03
(-0.03 - 0.21) (-0.01 - 0.26) (-0.03 - 0.12)
IE: Surprised 0.03 -0.08
(-0.00 - 0.10) (-0.17 - 0.00)
IE: Hurt -0.02 -0.03 -0.05
(-0.09 - 0.03) (-0.09 - 0.03) (-0.112 - 0.01)
Willingness to
collaborate
Total effect -0.17 0.18 -0.17 -0.05
(-0.47 - 0.28) (-0.15 - 0.56) (-0.45 - 0.18) (-0.34 - 0.22)
DE: Prior relationship -0.20 0.05 -0.19 -0.06
(-0.54 - 0.21) (-0.25 - 0.39) (-0.48 - 0.19) (-0.35 - 0.23)
IE: Angry
126

Ally x
Refusal
Friend x
Refusal
Ally x
Betrayal
Friend x
Betrayal
IE: Betrayed 0.06 0.10 0.04
(-0.4 - 0.20) (-0.06 - 0.32) (-0.06 - 0.16)
IE: Surprised -0.04 -0.03
(-0.09 - 0.03) (-0.11 - 0.03)
IE: Hurt -0.05 0.06
(-0.16 - 0.05) (-0.03 - 0.16)
Num.Obs. 113 106 103 114
Despite the evidence that emotions mediated and sometimes even reversed the direct
effects of prior relationships in Scenario A, it remains unclear which specific emotions predominantly mediate the effects of these relationships on trust and willingness to continue
collaboration. Generally, the feeling of anger was found to slightly, though inconsistently,
mediate the direct effect of prior relationships on trust towards Country A. The feeling of
betrayal, on the other hand, did not seem to mediate the direct effect on trust, but it did
influence the willingness to continue collaborating with Country A in the future.
Notably, in the context of friendship, the feeling of hurt appeared to mediate the effect
of prior relationships on both the reduction of trust towards Country A and the decrease in
willingness for the United States to collaborate with Country A in the future. Conversely,
hurt did not appear to play a significant role in alliances. This suggests that the dynamics
of trust reduction and the decision to withdraw collaboration from friends are mediated
more by feelings of hurt, contrasting with the effects of anger, betrayal, or surprise that
are more evident in alliances.
However, these mediation effects were only evident in Scenario A. For Scenario B, as
shown in Table 3.13, the mediation effects were both substantially and statistically insignif127

icant, indicating a different dynamic in how relationships and emotions interacted in the
non-security related scenario.
Table 3.14: (Scenario A) The estimated mediation effects on the inclination to punish Country
A from bootstrap samples using Generalized Linear Models (GLM) with the relationship control
group as the reference group (90% confidence level)
Ally x
Refusal
Friend x
Refusal
Ally x
Betrayal
Friend x
Betrayal
Inclination to
punish
Total effect 2.18*** -0.30 0.02 -0.39
(1.00 - 3.66) (-2.34, 1.67) (-0.97, 1.22) (-152, 0.62)
DE: Prior relationship 0.67 1.71* -0.23 -0.48
(0.57, 1.90) (-3.27, -0.38) (-1.14, 0.55) (-1.47, 0.37)
IE: Angry 0.39 0.70+ 0.13
(-0.08, 1.10) (0.13, 1.40) (-0.12, 0.50)
IE: Betrayed 0.81+
(0.11, 2.17)
IE: Surprised 0.31 0.72 0.09 0.09
(-0.26, 0.97) (-0.15, 1.88) (-0.05, 0.40) (-0.16, 0.45)
IE: Hurt
Num.Obs. 106 116 110 106
As for the inclination to punish Country A shown in Table 3.14, hurt ceased to play a
role as a mediator between prior relationships and the inclination to penalize the betrayer.
Instead, the feelings of anger and surprise appeared to mediate the effects of prior relationships on the willingness to punish the other country.
128

Table 3.15: (Scenario B) The estimated mediation effects on the inclination to punish Country
B from bootstrap samples using Generalized Linear Models (GLM) with the relationship control
group as the reference group (90% confidence level)
Ally x
Refusal
Friend x
Refusal
Ally x
Betrayal
Friend x
Betrayal
Inclination to
punish
Total effect 1.19* 0.30 1.35+ 1.04+
(0.18 - 2.90) (-0.90 - 1.46) (-0.00 - 2.87) (0.12 - 2.49)
DE: Prior relationship 1.51* 0.57 1.15 1.12*
(0.50 - 3.63) (-0.49 - 1.85) (-0.00 - 2.87) (0.12 - 2.49)
IE: Angry -0.21 -0.11 0.46
(-0.62 - 0.15) (-0.54 - 0.26) (-0.10 - 1.60)
IE: Betrayed -0.14 -0.11 -0.06
(-0.56 - 0.19) (-0.81 - 0.11) (-0.38 - 0.17)
IE: Surprised 0.15
(-0.9 - 0.47)
IE: Hurt -0.15 -0.22
(-0.50 - 0.13) (-0.67 - 0.04)
Num.Obs. 113 106 103 114
Conclusion
The findings from this experiment provide significant insights into the effects of betrayal
in international politics. The results confirm that friends are a distinct relational construct
from allies in international relations and show that friends react more strongly to incidents of
betrayal than non-friends. Specifically, friends experience a sharper decline in trust, positive
feelings, and willingness to collaborate after incidents where a friend fails to assist, reneges on
a promise, or switches sides. The experiment also reveals that the negative effects of switching
sides are more pronounced than those of not assisting in times of need or reneging on
commercial contracts among friends, highlighting loyalty as a foundational rule of friendship.
129

However, contrary to expectations, friends are not more likely than non-friends to impose
punishment on another friend when an incident of betrayal has occurred. In fact, the experiment found that allies are more likely to impose a penalty on another alliance member when it
fails to provide support in times of need or switches sides. This suggests that the transactional
nature of alliances, with their emphasis on reciprocal support and formal agreements, may
lead to a greater propensity for punitive measures compared to responses in friendships.
This is supported by the reasons for punishment given by respondents. Those in the
ally condition were more likely to state that the punishment was carried out to deter
future aggression or stabilize the international system. However, respondents in the friend
condition, once they had experienced an incident of betrayal, increasingly indicated that the
punishment, if it was to happen, would be carried out because the other party “deserved” it.
This suggests that when punishment is administered by friends amid betrayal, it is intended
to right a wrong rather than being motivated by other rational intentions.
Finally, the findings also indicate a marginal mediation effects of emotions on physical
reactions. Though neither statistically nor substantially significant, hurt was found to
marginally mediate the impact of the relationship on trust and willingness to continue
collaboration with the betrayer. On the other hand, anger was found to marginally mediate
the effect of the relationship on the increased inclination to penalize the betrayer.
Despite the insights gained from this experiment, several questions remain that require
further research into this critical issue in international politics. For instance, how longlasting are these reactions? Are they momentary, allowing for reconciliation after the initial
deterioration of the relationship, or are they enduring, leading to irreparable breakdowns?
130

Although there are areas for further exploration, these results highlight the complex dynamics
of betrayal in international relations. These dynamics will be further examined using a realworld case study in the next chapter.
131

Chapter 4
How Dare You? The British Reactions to
Thailand Declaring War against the Allies
The survey experiment in the previous chapter reveals the effects of betrayal in the international realm. Based on the experiment, we now understand that the public experiences
a sharper decline in trust, positive feelings, and willingness to continue collaboration with
another country that once considered a friend after an act of abandonment or switching
sides. Such actions are not seen merely as a revelation of interests; they are perceived as
betrayals, deeply impacting trust, positive feelings, and willingness to continue the collab132

oration. Additionally, the survey experiment shows that while individuals are less likely
to impose punitive actions on another country they consider a friend, when they do, it is
driven by retribution rather than deterrence—the belief that the other country “deserves”
the punishment for having wronged them.
However, while the survey experiment helped identify the causal relationship between
friendship, betrayal, and their consequences in the foreign policy realm, we do not yet
understand how these dynamics materialize in the more complex real-world setting. This
chapter addresses this gap by illustrating how betrayal unfolds and its resulting consequences
in life. Through a detailed case study of British-Thai relations and a shadow case of U.S.-
Thai relations during the Second World War when Thailand turned its back on the Allies in
favor of Japan, we will explore real-world instances of betrayal between nations.
As we will see in this chapter, switching sides can be interpreted differently based on
the relational closeness between the switcher and those impacted by the decision. In this
case, the Thai government’s declaration of war against the Allies and its conclusion of the
offense-defense pact with the Japanese government were viewed differently by the U.S. and
British governments. The Americans, who did not consider Thailand a close friend or even
ally, saw this act of switching sides as new information about the switcher, leading to a
reassessment of the situation and decisions made accordingly. When the situation changed
after the end of the War, they were ready to update their stance and worked with the new
Thai government for mutual benefits and strategic needs.
In contrast, British policymakers, who considered Thailand a friend or even a “stepchild”
of the British Empire, were furious at the Thai government’s decision to side with the
133

Japanese. The action was seen as a betrayal. Despite the friendship, goodwill, and assistance
provided by the British, how could the Thai government allow the Japanese passage, sign
an offense-defense pact with Japan, and declare war on the Allies? The British government
was particularly outraged when Thailand declared war on them, effectively denouncing the
friendship between the two countries. The British anger did not end with the war; they
sought to impose penalties on Thailand that the country could hardly bear. Even when
the U.S. government opposed such severe penalties, the British government only slightly
adjusted their demands. For the British, these penalties were the only way for the Thai
government to “pay back” for its actions and restore the justice it had once broken.
This chapter is structured as follows: The next section discusses the case selection,
explaining why I investigated British-Thai relations during the Second World War as a
case study. This is followed by a brief discussion of the relationship between the Thai and
British governments prior to the Second World War. The rest of the chapter is dedicated to
a detailed account of the British reactions that took place after Thailand had switched sides.
Why the British-Thai Case?
The British-Thai relations from 1939 to 1945 present an ideal case for studying the effects
of betrayal for two primary reasons. First, the events occurred during a time of crisis,
where information from specific individuals with ties to Thailand became a crucial source for
decision-making. As previously discussed, decision-making processes during crises are less
likely to involve lengthy debates, so decision-makers rely heavily on a few people with prior
experience or knowledge of the country or situation. This reliance on limited sources of infor134

mation highlights how perceptions of betrayal can shape responses during critical moments.
Second, this case involves a relationship between a more powerful country (Britain) and a
less powerful country (Thailand), which simplifies the observation of outrage and emotional
responses. In scenarios where a more dominant nation is betrayed by a smaller one, the effects
of betrayal are more readily observable as their actions are less restrained by factors such
as power. Conversely, cases where larger nations commit betrayal to smaller countries may
be more complex, as smaller countries often lack the means to exert significant physical
or policy-related consequences.
Moreover, this case facilitates a comparison with U.S.-Thailand relations during the
same period. Both the British and Americans were part of the Allies and were targeted
by Thai declarations of war. The key difference lay in the prior relationships between
the Thai government and each of these major powers. By examining these contrasting
relationships, we can investigate the effect of friendship on emotional reactions and the desire
for retribution. This comparative analysis enriches our understanding of how friendship
influences the dynamics of betrayal and the resulting responses.
In order to understand that the decisions of these involved actors are not purely driven
by strategic calculations but also their emotions, I closely examined more than 1,500 primary
documents from both the American and British Governments. These documents are collected
from the National Archives and Records Administration (NARA), the Library of Congress,
the Franklin D. Roosevelt Presidential Library, and the Southern Regional Library at the
University of California, Los Angeles. The story of the twist-and-turn relations between
Great Britain and Thailand begins in the following section.
135

The British-Thai Relations Amid the Rise of Japan
The rise of Japanese militarism in the 1930s and its vision for the New Order in East
Asia challenged the stability of colonial rule in Southeast Asia. Following the conquest of
Manchuria and China, the Japanese government set its eyes on Southeast Asia as a next
destination to implement the “Asia for Asia” project. Geographically, Indochina would serve
as the first target, followed by Burma and Malaya. The plan, though unannounced and
unconfirmed by the Japanese government, posed some concerns for the French and Burma
about their colonies’ security, especially when both of them were concurrently facing German
and Italian aggression in Europe. Another battle in a different region may become unbearable
for the two colonial powers whose capabilities were exhausted on the European front.
The British and French sought assurance from the Thai government – the only independent nation in Southeast Asia – that they would remain neutral in the event of
Japanese aggression. If Thailand were to cooperate with the Japanese, Thailand would
provide tremendous strategic value for the advance of Japanese troops into the British and
French territories. The French thus proposed a non-aggression agreement with Thailand in
1939. Seeing the agreement negotiation as an opportunity to renegotiate border adjustment
with the French, Prime Minister Phibun Songkhram quickly responded to the French offer
(Reynolds, 1994). In addition, the Thai government concurrently pursued non-aggression
agreements with Great Britain and Japan. Despite some prolonged discussions on Thailand’s
territorial restoration effort, Thailand eventually signed non-negotiation agreements with
Great Britain, France, and Japan on June 12, 1940. The three pacts were signed on the same
day to avoid the impression that the Thai government prioritized one country over the other.
136

Throughout the period mentioned above, the British government consistently approached
Thailand sympathetically. In contrast to the U.S. government, which strictly preferred the
policy of territorial status quo, the British did not oppose Thailand’s attempt to reclaim
its territory from the French. Not to mention, the British government refused to include
in the British-Thai pact a clause specifying that “the pact would cease to operate in the
event of hostilities between France and Thailand.”1 Even at the risk of creating a misunderstanding with the French, the British government pressed Paris to accept that the
British would conclude a pact with the Thai government, “irrespective of the success of the
French government’s negotiations.”2 The British government went so far as to argue that
if Indochina were to be abandoned by the French, it would be better to let it fall into the
hands of the Thais rather than the Japanese.
The fact that the British tolerated and even supported Thailand’s attempt to restore the
territory appeared somewhat puzzling. For the British, who had, like the French, seized some
pieces of territory from Thailand before, Thailand’s territorial restoration effort should raise
some concerns about the future of British colonies. After all, there was no guarantee that
the Thai government would never seek to restore the territory previously lost to the British
Empire in the early twentieth century. Despite the risk, why did the British government
insist on maintaining the policy of friendship towards Thailand? Why did it not act tough
to deter the Thai government from reclaiming the territory lost to the British in the future?
This puzzle will be disentangled in the subsequent subsection.
1British Foreign Office. (1940, February 8). [Cypher telegram to Sir J. Crosby]. British intelligence on
Siam (Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 21 BURMA 988/1940). Southern
Regional Library Facility at UCLA, Los Angeles, CA.
2
Ibid.
137

Is This Just a Story About Material Interests?
Undoubtedly, the British government had a multitude of reasons to be concerned about
the status of Indochina and Thailand. If the Japanese armed forces successfully established
bases in these two territories, the advancement to Burma and Malaya through Thailand
was not unimaginable. More critical than the survival of Burma and Malaya, however,
was the security of Singapore – Great Britain’s most prominent economic port and biggest
military base in the region. The fall of Singapore would be a disaster to the British
presence in Southeast Asia.
Despite Great Britain’s apparent interest in keeping Indochina and Thailand out of the
Japanese hands, interest is not determinant of how the British should approach the situation.
Interests do not mandate a country to act a certain way. In order to prevent Thailand
from collaborating with the Japanese against British interests, the British had an option
of pursuing either a deterrence or appeasement strategy. While the two may lead to the
same outcome, one strategy may be deemed more appropriate than the other, depending on
the British’s prior interactions with Thailand and its perception of the Thai government’s
intention. For the British government, who were inclined to view the Thais as friends, the
appeasement strategy seemed preferable.
In contrast to the standard IR expectation for the stronger to do what they can, the
British government chose to pursue the policy of friendship and sympathy towards Thailand
amidst the threat of Japanese expansion. In contrast to the French, who requested formal
reassurance from the Thai government in the form of a non-aggression treaty, the British
declined to push Thailand along the same path as this would have suggested the lack of
138

trust between friends. Instead, the British government decided to trust the sincerity of the
Thai prime minister to be “one hundred [percent] neutral in face of existing international
alignments.”3 As indicated by Sir Josiah Crosby, the British Minister in Thailand, verbal
reassurance from the Thai government was adequate to guarantee the Thai government’s
intention to remain neutral in any event of Japanese aggression.
Yesterday I was with the Prime Minister for over an hour and said the same
thing. He gave me definite assurances which if they can be taken at their face
value are satisfactory enough. He declared there had not been, was not now and,
as far as he was concerned, never would be a secret understanding between the
Siamese and Japan. He assured me that the Siamese was too proud to accept
foreign instructions in any of their services.
Moreover, Sir Crosby suggested that the British, with the qualities of “balanced judgment
and spirit of fair play.” should maintain a policy of sympathy and refrain from “exasperating them by a policy of suspicion and nagging” as the French did. He described the
French method of pushing for formal reassurance from the Thai government as driven by
“exaggerated and distorted ideas” about the Japanese influence in Thailand.4
Another evidence of the British’s friendly attitude towards Thailand was their reluctance
to take a firm stance on Thailand’s territorial restoration effort. On top of that, the
British government was willing to conclude a non-aggression pact with the Thai government
regardless of the status of the territorial disputes between France and Thailand. While it is
true that a non-aggression pact between Thailand and Great Britain would work “to diminish
3Crosby, J. (1939, January 2). [Letter to Viscount Haifax]. British intelligence on Siam (Thailand) and
Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 21 P.Z. 1214/1939). Southern Regional Library Facility
UCLA, Los Angeles, CA.
4Crosby, J. (1939, June 29). [Telegram to British Foreign Office]. British intelligence on Siam (Thailand)
and Mainland Southeast Asia, 1887-1948 (BIM-2 Reel 19 P.Z. 4325/1939). Southern Regional Library Facility
UCLA, Los Angeles, CA.
139

Japanese influence [in Thailand] and to increase the [British] influence in proportion,”5
the
British’s real motivation was to reassure Thailand of Great Britain’s honesty6 – the essential
quality of friendship. In contrast, if the British refused to trust the Thai government’s
honesty and sincerity by delaying the conclusion of the pact, this would increase the chance
of “seeing [the Thai Prime Minister] turn in dudgeon of the side of Japan”7 out of fear.
The British’s trust in the Thai government’s sincerity and honesty in their friendship
continued throughout the hostilities between France and Thailand that began around the
end of 1940 over the territorial issue. Suspecting that the French may have initiated the
hostilities along the frontier,8
the British Minister was unwilling to give a harsh warning
to the Thai government. Even on the verge of Germany getting involved in the FrancoThai territorial conflict, Sir Crosby persistently refused to convey a more specific warning to
the Thai government out of fear that more substantial pressure would “disturb the friendly
atmosphere”9 between the two countries. In addition, when the British had an agenda to
influence the Thai government’s decision, they chose to remind the Thai government of their
friendship instead of opting for a standard tool of carrots and sticks. A conversation between
the Thai Deputy Minister of Thailand, Direk Jayanama, and Sir Crosby about the threat of
5Crosby, J. (1940, January 6). [Letter to VIscount Halifax]. (BIM-6 Reel 21 BURMA 1073/1940).
Southern Regional Library Facility UCLA, Los Angeles, CA.
6Crosby, J. (1940, July 15). [Letter to Air Vice Marshal]. British intelligence on Siam (Thailand) and
Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 19 P.Z. 4939/1940). Southern Regional Library Facility
UCLA, Los Angeles, CA.
7Crosby, J. (1940, January 6). [Letter to Viscount Halifax]. British intelligence on Siam (Thailand)
and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 21 BURMA 1073/1940). Southern Regional Library
Facility UCLA, Los Angeles, CA.
8Crosby, J. (1941, January 21). [Telegram to British Foreign Office]. British intelligence on Siam
(Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 BURMA 10521/1941). Southern
Regional Library Facility UCLA, Los Angeles, CA.
9Crosby, J. (1940, October 8). [Telegram to British Foreign Office]. British intelligence on Siam (Thailand)
and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 P.Z. 5648/1940). Southern Regional Library Facility
UCLA, Los Angeles, CA.
140

German and Japanese intervention in the Thai-Indochina territorial dispute.
I asked whether the Thai Government would be prepared to strike such a sordid
bargain and to go back thus cynically upon their declared policy of complete
neutrality to say nothing of their treaty of non-aggression with ourselves, refused
to believe that Thailand would consent to turn against Britain who was her oldest
and best friend for the sake of obtaining territory.
10
While interests informed the British government about the need to prevent Thailand
from falling into the hands of the Japanese, it was their prior relationship that determined
how the British chose to approach the Thai government. In contrast to a common strategy
of deterrence, the British preferred to persuade the Thai government with a reminder of their
friendship. As a result, the British refused to push the Thai government because doing so
would indicate a lack of trust. After all, friends do not push friends around. However, when
in need, the British were willing to step in to help their friend, even when it meant that the
British had to put extra work into convincing the U.S. government to support Thailand.
Helping Is Proof of Friendship
Despite the British confidence in the strength of the British-Thai friendship, there remained
one problem: Thailand’s military weakness would not permit the country to withstand the
Japanese invasion. However, the possibility of the British supplying weapons to the Thai
government was out of the question. Impacted by the ongoing battles on the European front
and shortage of aircraft and vessels in Malaya and Singapore, the British were incapable of
providing meaningful military assistance to the Thai government in any way. The British
10Crosby, Josiah. (1940, October 15). [Telegram to British Foreign Office]. British intelligence on Siam
(Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 P.Z. 6767/1940). Southern Regional
Library Facility UCLA, Los Angeles, CA. Emphasized by author.
141

Foreign Service realized that “the time may come before long when words [would] not be of
much avail against pressure from the Axis Powers,”11 and the Thais may have no choice but
to side with the Japanese because the British were unable to help them.
Before having to suffer “the penalties imposed upon [themselves] by [thier] own military weakness in Malaya and the Far East,”12 the British government turned to the U.S.
government for help. Because Thailand’s decision to side with the Allies was contingent
on its ability to defend itself against the Japanese aggression, by the end of 1940, the
British Foreign Office instructed the British Embassy in Washington to make every effort
possible to convince the U.S. government of Thailand’s importance (Reynolds, 1994, p. 54-
55). Among other requests, the British asked that the U.S. government unfreeze the export
of bombers and fighters to Thailand. These planes had been purchased and paid for by
the Thai government but were later withdrawn by the U.S. government.13 However, out of
strategic necessity, the U.S. government insisted on leaving these planes at its Manila base
and refused to concede to the British request.
For the British, the ability to provide the Thai government with war materials was
essential for both its interests in the region and the preservation of its friendship with
Thailand. Increasing arm supplies to the Thai government would undoubtedly enhance
Thailand’s defense capability against Japanese aggression. However, on top of that, the
11British Foreign Office. (1940, October 15). [Cypher telegram to Sir J. Crosby (Bangkok)]. British
intelligence on Siam (Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 P.Z. 5767/1940).
Southern Regional Library Facility UCLA, Los Angeles, CA.
12Crosby, Josiah. (1940, July 15). [Letter to Air Vice Marshal]. British intelligence on Siam (Thailand)
and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 19 P.Z. 4939/1940). Southern Regional Library Facility
UCLA, Los Angeles, CA.
13Crosby, J. (1940, November 13). [Telegram to British Foreign Office]. British intelligence on Siam
(Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 19 P.Z. 6291/1940). Southern Region
Library Facility University of California, Los Angeles.
142

British considered the ability to arm Thailand as proof of their friendship.
The question of supplying war material to the country forces, what for the Thais,
is the main issue and the acid test of friendship namely who is prepared to trust
them and who is unwilling to do so.14
Unfortunately, Great Britain’s requests were declined by the U.S. government. Instead,
the U.S. government insisted on leaving the planes purchased by the Thai government at
its Manila base to strengthen U.S. military presence in Southeast Asia. In addition to the
geopolitical reason, it is noteworthy that the U.S. dismissal of the British request was driven
by the U.S. suspicion of the Thai government’s trustworthiness. Hugh Grant, the American
Minister in Bangkok, repeatedly demonstrated this suspicion in multiple reports sent to the
State Department. He indicated that because of the Thai government’s untrustworthiness,
any aid provided to the Thai government would eventually benefit the Japanese(Reynolds,
1994, p. 54). In fact, Grant even blamed the British policy for giving the Thai government
too much leeway for its misbehavior during the territorial conflicts. In a way, this is also
evidence that there were other options for dealing with Thailand other than appeasement.
Without material supports from the United States, the British government turned to
financial and economic measures to discourage the Thai government from siding with the
Japanese. This was when the British understanding of their relationship with the Thais
played a part in determining their policy preference, again. The British had an option of
seizing Thailand’s funds held in London to coerce Bangkok to side with the Allies openly
or providing the Thai government with conditional economic assistance. The former was
strongly objected to by Sir Crosby, who viewed the seizure of Thailand’s fund as immoral.
14Ibid.
143

By choosing this option, the British would suffer “irreparable moral damage with the Thai
who would regard it as dishonest and inconsistent with British reputation for probity.”15
Consequently, he actively advocated for the latter option, which would preserve the policy
of friendship and sympathetic understanding towards Thailand – “the only principles that
British government should adhere to.”16
Accordingly, the British government chose to offer Bangkok economic and financial
assistance based on suitable conditions instead of putting economic and financial pressures
on the Thai government.17 An example of this assistance was when the British government
authorized fuel supplies to be shipped from Singapore in light of an oil shortage in Thailand
in 1941. Although this foremost assistance came without any attached conditions, the British
government requested that Bangkok sell rubber and tin – the two essential war commodities
– to them in exchange for any subsequent assistance (Jayanama, 2017, p. 135). Later on, the
U.S. government agreed to cooperate with the British to purchase these two commodities
from Thailand with the purpose of diverting them from the Japanese hands (Martin, 1963,
457). In addition, the U.S. government also promised Bangkok that it would provide more
economic assistance to the country if the Thai government could proportionately redirect
rubber and tin intended to sell to the Japanese to them.
15Crosby, J. (1941, January 15). [Telegram to British Foreign Office]. BBritish intelligence on Siam
(Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 BURMA 10404/1941). Southern
Regional Library Facility UCLA, Los Angeles, CA.
16Ibid.
17British Foreign Office. (1941, March 28). [Cypher Telegram to Viscount Halifax]. British intelligence
on Siam (Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 P.Z. 16909/1941). Southern
Regional Library Facility UCLA, Los Angeles, CA.
144

The Japanese Invasion of Indochina and Thailand
Although the British were doing everything within their power to encourage the Thais to
withstand the Japanese force, they could not reason with the Thai government when it
came to the issue of territorial reclamation.18 In early 1941, with the hostilities along the
Thai-Indochina border at the backdrop, the Japanese government expressed its willingness
to mediate the Franco-Thai conflict to the Thai and Vichy governments. Realizing that the
Thai government would never back down on its request for territorial restoration, the British
capitulated to the idea of Japanese mediation. This was not because rejection could lead to
a more catastrophic outcome but because the British seemed to believe the Thai government
was not pro-Japanese at its core. In addition, if the British continued to be sympathetic to
Thailand, they could still prevent Thailand from betraying their friendship.
[T]he fact that she has refrained from doing so up to now indicates that her
sympathies are not fundamentally pro-Japanese. I still think that by continuing
with a policy of tact and friendliness we may keep her from selling herself utterly
to Japan.19
With Japanese mediation, Franco-Thai territorial disputes finally ended in April 1941
with the Vichy government agreeing to return the pieces of territories to Thailand.
However, despite the success in mediating the conflict in favor of their Asian fellows,
the Japanese desired more.
As the threat of Japanese invasion in Indochina and Thailand became more prominent,
the panicking Thai Prime Minister approached the British and U.S. governments for help.
18Crosby, J. (1941, January 27). [Telegram to British Foreign Office]. British intelligence on Siam
(Thailand) and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 22 BURMA 10662/1941). Southern
Regional Library Facility UCLA, Los Angeles, CA.
19Ibid.
145

Due to the urgency of the situation, Prime Minister Phibun appealed to the British to act
alone if they could not get the Americans immediately onboard (Reynolds, 1994). Unable
to provide Thailand with much material assistance, the British sought to deter Japanese
aggression through a verbal deterrence. As a result, in August 1941, British Secretary
of Foreign Affairs Anthony Eden announced in the House of Commons that any threats
to Thailand’s security and territorial integrity would be considered an attack on British
interests (Jayanama, 2017, p. 139). The urge to deter Japanese aggression and assure the
Thai government led the British to consider presenting a written ultimatum to the Japanese
government.20 Nevertheless, the British plan to present the ultimatum was overturned by
the U.S. government out of fear that it would produce further complications.
In comparison to the British passion for saving Indochina and Thailand, the U.S. government’s reaction to the threat of Japanese aggression in Southeast Asia was more moderate. While agreeing on the importance of Indochina and Thailand in the larger scheme
of Southeast Asian security, the Americans were unwilling to go beyond providing verbal
reassurance to the Thai government. Besides proposing the neutralization of Indochina and
Thailand and freezing Japanese assets in the United States in response to Japan’s request to
establish air and naval bases in southern Indochina (Martin, 1963, p. 457), the Americans did
not show any interest in providing military assistance to Thailand. This includes constant
rejections of Thailand and Great Britain’s enduring requests for the delivery of the planes,
believing that having the planes in Manila would better deter the Japanese (Martin, 1963,
p. 457). Consequently, the most solid response Bangkok received from Washington was
20Foreign Relations of the United States States Diplomatic Papers, 1941, The Far East, Volume IV,
eds. John G. Reid, Louis E. Gates, and Ralph R. Goodwin (Washington: Government Printing Office, 1956),
Document 297
146

reassurance that “if Thais withstood Japanese, United States would give them all possible
help as they had been giving to China.”21
.
However, without adequate military assistance from Great Britain and the United States,
Thailand was defenseless against the Japanese invasions. On December 7, 1941, the Japanese
Ambassador in Bangkok called the official residence of the Prime Minister to inform about
Japan’s declaration of war upon the United States and Japan and requested passage through
Thai territory (Martin, 1963, p. 459). The passage was denied due to the Prime Minister’s
absence in Bangkok. Consequently, the Japanese invaded Thailand from the south by sea
and from the east by land. After a few hours of resistance along the coast, Prime Minister
Phibun ordered a cease-fire to prevent additional loss to the Thai armed forces. At this
point, Churchill’s message of reassurance to the Thai Prime Minister indicating the British
willingness to regard “an attack on [Thailand] as an attack on [Great Britain]” if Thailand
could preserve “the true independence and sovereignty”22 carried no more weight. To avoid
further destruction, The Thai government, after a long debate, agreed to allow the Japanese
passage. On December 8, 1941, Direk Jayanama, the Thai Foreign Minister, signed an
agreement to allow the Japanese armed forces passage through Thailand in exchange for
assurances of Thailand’s independence, sovereignty, and honor (Martin, 1963, p. 460).
Nevertheless, by the end of the month, the Thai cabinet reexamined the situation and
decided to ally with the Japanese. On December 21, 1941, the Thai government signed an
offensive and defensive pact with Japan, thus implicitly declaring that Thailand would fight
21Crosby, J. (1941, August 7). [Telegram to British Foreign Office]. British intelligence of Siam (Thailand)
and Mainland Southeast Asia, 1887-1948 (BIM-6 Reel 23 BURMA 15902/1941). Southern Regional Library
Facility at UCLA, Los Angeles, CA
22Churchill, W. (1941, December 8). Prime Minister Winston Churchill’s Broadbast on War with Japan
[Speech]. The Public’s Library and Digital Archive. http://www.ibiblio.org/pha/timeline/411208ewp.html
147

alongside the Japanese against the Western powers. In response to the conclusion of the
alliance, the U.S. government was prepared to “regard the action of the Thai government
unfriendly to the United States,” since the other party of the alliance was in a state of war
with the U.S.23 However, the United States was willing to continue recognizing the Thai
Minister in Washington as the representative in the United States of the free people of
Thailand since he had explicitly denounced the Thai government’s decision to ally with the
Japanese as unrepresentative of the true will of the Thai people.24
While welcoming the U.S. decision to recognize the Thai Minister in Washington as
the representative of the free people of Thailand, the British government warned against
treating Thailand as an alliance of the enemy. Inclined to believe that the Thai government concluded the agreement with the Japanese under duress, the British cautioned
that any harsh treatment imposed upon Thailand would be counterproductive as it might
alienate the Thais even more.
As regards general policy the conclusion of a Treaty of Alliance which His
Majesty’s Government have good reason to believe commits Thailand to give
assistance of every kind to Japan would normally justify a declaration of war
upon Thailand by the powers at war with Japan. His Majesty’s Government have
however received convincing indications that the majority of Thai opinion is antiJapanese if not pro-Ally, and it is likely to become increasingly anti-Japanese as
the Japanese proceed to apply their usual arrogant methods and to infringe their
agreement to respect the sovereignty of Thailand. His Majesty’s Government
[feels] that a declaration of war might change this movement of opinion for the
worse and have the effect of throwing Thailand into the arms of Japan.25
Consequently, the British urged the U.S. government to adopt a similar policy towards
23Department of State. (1941, December 18). [Aide-Memoire]. Department of State Central Files (RG 59
Box 2236 711.92/30A). The National Archives and Records Administration (NARA), College Park, MD.
24Pramoj, S. (1941, December 21). [Memorandum] Department of State Central Files (RG 59 Box 2236
711.92/31). The National Archives and Records Administration (NARA), College Park, MD.
25Department of State Divison of Far Eastern Affairs. (1941, December 27). [Memorandum] Department
of State Central Files (RG 59 Box 2236 711.92/33). The National Archives and Records Administration
(NARA), College Park, MD.
148

Thailand, that is, to treat Thailand as an enemy-occupied territory instead of an alliance
of the enemy. This policy would remain in place as long as the Thai armed forces did
not obstruct British operations, threaten the security of British troops, or strengthen their
forces along the Thai-Burmese border.26 For the time being, the British government would
refrain from provoking the Thai government and Thai armed forces as this may accelerate
Thailand’s decision to fully side with the Japanese.
War Declaration as a Turning Point
Convinced by the British argument, the U.S. State Department agreed to adopt the policy
recommended by London. However, while the U.S. government continued to implement
the policy throughout the war, the British attitude towards Thailand turned around after
Thailand declared war on Great Britain and the United States. In the period that followed
the war declaration, anger dominated the atmosphere of British decision-making towards
Thailand while sympathy and forgiveness waned.
On January 25, 1942, the Thai government declared that a state of war existed between
Thailand, on the one hand, and Great Britain and the United States, on the other. The
statement referred to the British and U.S. successive acts of aggression, such as sending
troops and airplanes into the Thai territory, as the conditions necessitating a war declaration.
From this period onwards, the British and American stances began to reverse. While the U.S.
government continued to treat Thailand as an enemy-occupied country,27 the British were
26Pramoj, S. (1941, December 21). [Memorandum] Department of State Central Files (RG 59 Box 2236
711.92/31). The National Archives and Records Administration (NARA), College Park, MD.
27Department of State Divison of Far Eastern Affairs. (1942, January 27). [Memorandum] Department
State Central Files (RG 59 Box 2236 711.92/36). The National Archives and Records Administration
(NARA), College Park, MD.
149

furious at the Thai government’s decision to turn their back on one of their oldest friends
and thus immediately recognized the state of war to exist between the two countries.
Why did the declaration of war matter, one could ask? While the declaration of war
served as a formal recognition of a state of war between Thailand, Great Britain, and the
United States, it was a purely symbolic action. After all, the Thai government had concluded
the offensive and defensive pact with the Japanese since December 1941, declaring that the
Thai armed forces would fight alongside the Japanese against the Western powers, including
the British. The declaration of war, therefore, did not add much practical value to the
existing alliance between the Thais and the Japanese. On the contrary, what it did was
explicitly sever the friendship between Thailand and Great Britain. It was arguably this
symbolic act – a denouncing of friendship – that enraged the British because friends. British
diplomatic records repeatedly pointed out that the Thai government’s decision to declare
war on the British, in and of itself, was morally wrong. The British moral judgment of
the Thai government’s decision seems to indicate that the field of international relations is
diffused with moral expectations, as opposed to the conventional wisdom of amorality of
international relations (Rathbun and Pomeroy, 2021).
Arguably, Great Britain’s furious responses towards the Thai government could be driven
by the fear of having more destruction done to them by the Japanese and the Thais. By
responding to Thailand’s declaration of war, the British perhaps attempted to deter the
Thai government from taking control of some British territories that originally belonged to
the Thais. Nonetheless, British diplomatic records never indicated why and how Thailand’s
declaration of war could impact British interests in the region. Even though, in reality, the
150

Thai government later assumed control over Burma’s Kayah and Shan States, to say that the
British foresaw that this would happen with the absence of British deterrence would be an
after-the-fact justification. Instead of having a concrete reason, the British’s shifting attitude
towards Thailand seemed to be affected by the feeling of anger. Mercer (2010) calls this an
emotional belief – how emotion influences the way one sees a situation and interprets evidence.
A document most demonstrative of how the British used anger to inform their decision about
Thailand was their statement about the status of postwar Thailand proposed to the Allies.
The position of Thailand is in some respects unique in the Far East though not
without parallel in Europe. A country with a long traditional friendship with
us has though admittedly under pressure from Japan, betrayed that friendship.
Not content with collaborating with our enemy and despite her treaty of nonaggression with us this quisling government of Luang [Phibun] took the initiative
in declaring war upon us. For these acts Thailand is already paying the price and
will undoubtedly pay a yet heavier price as the war reaches her territories. It is
still possible for the people of Thailand to do something to save themselves from
the worst consequences of their betrayal, and they will be judged by the effects
that they make to redeem themselves from the position in which the action of
their present régime has placed them.28
The proposed statement was met with strong objection by the U.S. State Department.
Inclined to see Thailand’s declaration of war as a decision taken under duress, the Americans
continued to treat Thailand as enemy-occupied territory – the policy once advocated by
the British.29 Accordingly, the U.S. government considered the Phibun administration a
puppet government controlled by the Japanese. As a result, the U.S. government recognized
that actions taken by the puppet government might not be representative of the true
will of the Thai people.30
28Department of State. (1944, March 8). Attitude towards Thailand [Memorandum for the President].
Official Files (339-339b Siam, Government). Franklin D. Roosevelt Presidential Library, Hyde Park, NY.
29Ibid.
30Peck, W. R. (1942, August 22). Political, Economic, Financial and Military Conditions in Thailand,
151

If the Thai people truly repudiated the Thai government’s decision to ally with the
Japanese, this public opinion could be advantageous to the Allies. However, since the
British’s proposed statement blamed both the Thai government and the Thai people for
betraying the long traditional friendship between Great Britain and Thailand, it could
discourage the Thai people from resisting the Japanese. Even worse, it could be exploited
by the Japanese to manipulate Thailand’s public opinions against the Western powers.31
Consequently, the U.S. government recomended that the British “[made] no declaration
rather than [making] the proposed one under consideration.”32 However, if the British
insisted on making the declaration anyway, the U.S. government suggested that the British
“at least [made] an unequivocal commitment that Great Britain [had] no territorial ambitions
in Thailand.”33 This would bring the British closer to the position previously announced by
President Roosevelt and Generalissimo Chiang Kai-Shek on the status of postwar Thailand.
The British government outright rejected the U.S. recommendations concerning the
proposed declaration, unwilling to make any definite commitment on questions affecting
international boundaries prior to the end of the war. More importantly, the British government was anxious to publicly set forth its attitude toward Thailand. John M. Allison,
the Second Secretary of the U.S. Embassy in London, observed that in a conversation with
Ashley Clark, the Head of the Far Eastern Department of the British Foreign Office, Clark
December 1941 to June 1942 [Summarized Report]. Department of State Central Files (RG 59 Box 5834
892.00/233). The National Archives and Records Administration (NARA), College Park, MD
31Department of State. (1944, March 8). Attitude towards Thailand [Memorandum for the President].
Official Files (339-339b Siam, Government). Franklin D. Roosevelt Presidential Library, Hyde Park, NY.
32Ballantine, J. W. (1944, March 17). Thailand [Memorandum of Conversation between George Sansom
and Joseph W. Ballantine]. Department of State Central Files (RG 59 Box 5835 892.01/54). The National
Archives and Records Administration (NARA), College Park, MD.
33Ibid.
152

seemed to put more emphasis on Thailand’s betrayal of friendship as a reason to make their
attitude known to the Thai government and the public.
[T]he British [felt] that the declaration of war by Thailand against Great Britain
and the United States was gratuitous and that this fact should be brought home
to the Thai people. If the Thai government had merely bowed to the superior
strength of the Japanese, and reluctantly allowed their country to be used as
a base for operations against Great Britain, what would have been one thing –
but, for a government which had been in close and friendly relations with Great
Britain to take the action it did, with apparent enthusiasm, was an act not easy
to forgive.
34
Whether the Thai government decided to cooperate with the Japanese with “apparent
enthusiasm” was, in fact, inconclusive. As the situation progressed, the U.S. government
became more inclined to believe that the Thai government yielded to the Japanese due to
the logic of force. In the general report about Thailand written in August 1942, Willys R.
Peck, the new American Minister in Bangkok who arrived with a mandate to improve the
relations with Thailand, points out that “Japanese troops across the border outnumbered
the Thai army 3 to 1. Numerous Japanese airfields were constructed near Thailand and
Japanese planes in Indochina probably outnumbered Thai planes 4 to 1.”35 On the brink of
the Japanese invasion, the Thai Prime Minister “instituted serious, even desperate inquiries
in London and Washington, to discover whether military measures were afoot to curb Japan’s
further expansion, particularly Japan’s threatened invasion of Thailand.”36 Nevertheless, his
request was evidently turned down by both powers, leaving the Thai Prime Minister to
34Gallman, W. J. (1944, April 22). British Attitude Toward Thailand [Letter to the Secretary of State].
Department of State Central Files (RG 59 Box 5835 892.01/4-2244). The National Archives and Records
Administration (NARA), College Park, MD.
35Peck, W. R. (1942, August 22). Political, Economic, Financial and Military Conditions in Thailand,
December 1941 to June 1942 [Summarized Report]. Department of State Central Files (RG 59 Box 5834
892.00/233). The National Archives and Records Administration (NARA), College Park, MD.
36Ibid.
153

decide whether “to resist Japan’s attempt to dominate his country or to capitulate and
escape the devastation of war.” 37
Perhaps, the most telling evidence that the Thai Prime Minister did not predetermine
the abandonment of his country’s relations with the Western powers was the country’s recent
purchase of the British America Tobacco Company’s interests in Thailand at the price of
2,000,000 USD. “Delay of a week would have made the payment unnecessary.”38 These pieces
of evidence do not seem to suggest that the Thai government had a preconceived intention
to sever ties with the U.S. and British governments. However, the evidence could not change
the British belief about Thailand’s abandonment of their friendship.
Punishment Is What a Betrayer Deserves
As the Allies came closer to winning the war in the pacific theater, Great Britain, the United
States, and China began to lay down a design for postwar Southeast Asia. The status of
postwar Thailand continued to be a point of contention between the three countries due
to the British government’s desire to disproportionally punish the Thai government. On
August 20, 1945, a few days after the Japanese were defeated, British Foreign Secretary
Ernest Bevin addressed to the House of Commons the British government’s will to right
the wrongs done by the Thai government.
It will be remembered that Thailand, a country whose relations had been particularly cordial with Great Britain before the war, a country with whom we had
been closely associated in the attainment of her full emancipation on a Sovereign
State, declared war upon us in January 1942. It came as a disagreeable shock
that when Thailand was invaded by the Japanese she immediately entered into
an alliance with Japan and later on accepted British territory at the hands of the
37Ibid.
38Ibid.
154

Japanese. It is pleasing to note, however, that last year’s [g]overnment which took
these measures was replaced; and that there has been a growth of the resistance
movement in Thailand. . . Thailand’s association with Japan inevitably leaves
many practical questions for settlement. These will be examined and our attitude
will depend on the way in which the Siamese meet the requirements of our troops
now about to enter their country: the extent to which they undo the wrongs done
by their predecessors and make restitution for the injury and loss and damage
caused to British and Allied interests and the extent of their contribution to
restitution of peace, good order and economic rehabilitation in Southeast East
Asia.39
Knowing that their determination to punish Thailand would be opposed by the U.S.
government, the British urged that the Americans recognize a state of war between Great
Britain and Thailand as a justification for any punishments imposed by the former on the
latter. The British reminded that the fact that the Americans did not regard themselves
to be at war with Thailand should not make them entitled “to ask that other governments
who [were] in a state of war with Siam [to] forego their right.”40 Furthermore, the U.S.
decision to ignore Thailand’s declaration of war should not “justifiably be adduced a reason
for the mitigation of the conditions upon which those countries [were] prepared to liquidate
the state of war.” This accusation by the British raised a strong note of protest by the
U.S. State Department, who believed that “British [had] no right to make fact they were
at war with Siam and [the Americans] were not excuse for unilateral settlement.”41 In any
case, the U.S. contribution to victory in Southeast Asia, in the U.S. State Department’s
39Bevin, E. (1945, August 20). Text of Passage Relating to Thailand in a Speech Delivered by the Secretary
of State for Foreign Affairs, Mr. Bevin, in the House of Commons on August 20th, 1945 [Speech]. Department
of State Central Files (RG 59 892.01/8-2045). The National Archives and Records Administration (NARA),
College Park, MD.
40Winant, J. G. (1945, September 8). [Telegram to Secretary of State]. Department of State Central Files
(RG 59 741.92/9-845). The National Archives and Records Administration (NARA), College Park, MD.
41American Legation in Bangkok. (1945, November 21). [Telegram to Secretary of State]. Department of
State Central Files (RG 59 741.92/11-2145). The National Archives and Records Administration (NARA),
College Park, MD.
155

view, “was certainly sufficiently great to warrant [the Americans] substantial voice in peace
settlement” in the region.42
With the debate on Great Britain’s right to unilateral policy in Thailand in the background, the British demanded that the Thai government be punished for damages caused
to the Allies and Thailand’s neighboring countries.43 One area that the British considered
unjust was Thailand’s ability to “accumulate a very large surplus of a commodity essential
to the life of neighboring territories.”44 While other countries had to make extra efforts to
restore the pre-war standards of living, the Thais could sustain their living due to their excess
of rice and potentially increase their claims upon the production of the rest of the world if
left unpunished. Leaving the Thai government unaccountable for the decision undertaken
during the war would be “contrary to all principles of justice.”45
To restore justice, the British government requested the Thai government 1,500,000 tons
of Thai rice – the quantity that exceeded the U.S. assessment of Thai rice surplus. This
large amount of rice would be distributed to Thailand’s neighboring countries that suffered
from Thai restrictions on rice exports during the war. Not only could such unrealistic
demand for rice impose an intolerable burden on the Thai economy, but it also deviated
from the common practice towards countries that lost a war. Principally, countries were
usually held accountable for direct damages caused by them during the time of conflict.
42Ibid.
43Department of State. (1945, August 25). [Telegram to American Embassy in London] Department of
State Central Files (RG 59 892.01/8-2045). The National Archives and Records Administration (NARA),
College Park, MD.
44Bevin, E. (1945, August 20). Text of Passage Relating to Thailand in a Speech Delivered by the Secretary
of State for Foreign Affairs, Mr. Bevin, in the House of Commons on August 20th, 1945 [Speech]. Department
of State Central Files (RG 59 892.01/8-2045). The National Archives and Records Administration (NARA),
College Park, MD.
45Ibid.
156

However, Thailand’s restriction on rice exports was different. It did not cause direct damages;
only an indirect inconvenience to other countries. The fact that a country was unable to
purchase a commodity during the war does not entitle that country to receive that commodity
free of charge.46 If such a punishment was placed upon the Thai government, the U.S.
government noted that “a similar demand could logically be made against Indochina and
indeed against every country whose normal exports were out off by the exigencies of war.”47
By the same token, the demand, if successfully imposed on Thailand, “would establish a
new principle in the distribution of the assets of a country with which any of the United
Nations [had] been at war.”48
However, inclined to see the Thais as immoral, the British were willing to go beyond
the existing principle and undergo lengthy arguments with their ally only to seek moral
compensation from Thailand. No punishments were too harsh for the betrayer of friendship,
and no measures were deemed sufficient to repair the relationship damage. Even though
the British government recognized the existence of the Free Thai Movement – Thailand’s
resistant movement against the Japanese – but because the movement did not involve “loss
of life and sacrifice similar to such movements in other countries,”49 its practical value was,
therefore, not particularly high.50 At one point, an American diplomat noted that the British
46British Foreign Office. (1945, December). Proposed Teletype Conversation with American Embassy,
London [Memorandum]. Department of State Central Files (RG 59). The National Archives and Records
Administration (NARA), College Park, MD.
47Department of State. (1945, November 29). [Aide-Memoire]. Department of State Central Files (RG 59
741.92/11-2845). The National Archives and Records Administration (NARA), College Park, MD.
48British Foreign Office. (1945, December). Proposed Teletype Conversation with American Embassy,
London [Memorandum]. Department of State Central Files (RG 59). The National Archives and Records
Administration (NARA), College Park, MD.
49Department of State. (1945, December 22). [Telegram to American Political Advisor in Bangkok]. Department of State Central Files (RG 59 741.92/12-1545). The National Archives and Records Administration
(NARA), College Park, MD.
501944-09-27 NARA RG 59 Box 5835 892.01/9-2744
157

seemed to “attach great importance to acceptance by Siamese of heavy penalties as [a] public
admission of guilt.”51 The severity of the British terms “have had a very adverse effect on the
British position in Siam and on further Siamese friendship and cooperation which a liberal
policy would have cemented.”52 As confessed by Esler Dening, British head negotiator, “the
more realistic American approach might lead to popularity in postwar Thailand, while a
grudge [was] borne against [Great Britain]” (Reynolds, 2010, p. 373).
Conclusion
This chapter adds depth to the theory of friendship and betrayal in international relations by
demonstrating how the British and U.S. governments responded to the Thai government’s
decisions to allow passage for Japanese troops, sign an offense-defense pact with Japan,
and eventually declare war on the Allies.
Without the bond of friendship, the U.S. government was able to react to these incidents
more rationally. This approach allowed them to recognize that the Thai government was
acting under Japanese influence and enabled them to collaborate with the Free Thai Movement, which included individuals who opposed their government, without harboring hard
feelings. These rational decisions, devoid of moral judgment and harsh emotional reactions,
facilitated U.S. collaboration with the Thai government after the war, allowing them to
leverage Thailand’s strategic location when the Cold War began. This cold, calculative
51American Legation in Bangkok. (1945, November 21). [Winant, J. G. (1944, September 27). [Telegram
to Secretary of State]. Department of State Central Files (RG 59 Box 5835 892.01/9-2744). The National
Archives and Records Administration (NARA), College Park, MD.]. Department of State Central Files (RG
59 741.92/11-2145). The National Archives and Records Administration (NARA), College Park, MD.
52British Foreign Office. (1945, December). Proposed Teletype Conversation with American Embassy,
London [Memorandum]. Department of State Central Files (RG 59). The National Archives and Records
Administration (NARA), College Park, MD.
158

response ultimately proved beneficial for U.S. interests in the long run.
In contrast, the British government, hurt and furious by the betrayal of the Thai government, could not think beyond their anger and desire for retribution. Despite strong
objections from the U.S. government against imposing harsh penalties, the British insisted
on demanding reparations from the Thai government in the form of rice at levels so high that
the country could hardly bear it. These penalties were not meant to deter Thailand from such
behavior in the future or to set an example for other countries. Instead, they were carried
out purely to restore “justice” and to make Thailand “pay” for the wrong they had done.
Even after the end of the Second World War, the relationship between the British
and Thai governments never returned to its pre-war peak. Although the two countries
have maintained “friendly” diplomatic relations, British policymakers have continued to
view Thailand as an “opportunist” nation that acts in its own interest without regard
for the goodwill of others.
159

Chapter 5
Conclusion
We now know that betrayal is a thing because friendship is a thing in international relations.
In Chapter 2, I have demonstrated that friendship is a distinct construct in foreign policy
discourse, different from allyship or alliance—a concept we know extensively about. From
the word embeddings analysis, it is evident that policymakers tend to use emotionallyladen moral terms when discussing friendship, in contrast to the more rationalistic terms
used when discussing allyship. Furthermore, the analysis reveals that policymakers discuss
friendship in ways that are strikingly similar to how ordinary people discuss friendship in their
interpersonal relationships. Friendship, whether in the international realm or interpersonal
160

context, is a type of relationship bound by emotional bonds rather than strategic thinking.
Unlike transactional relationships such as alliances, which are underpinned by mutual
interests, friendships are a type of communal relationship characterized by trust, openness,
honesty, reciprocity, mutual caring, and loyalty. Friends inherently trust in the quality and
disposition of one another. They believe that their friends will consider their well-being and
interests, even when those interests may not exactly align with their own. They trust and
hope that friends will remain loyal to the friendship and not abandon them in times of need.
It is precisely because of this inherent, non-calculative trust that monitoring mechanisms
and constant reassessment of the other’s trustworthiness are unnecessary. In fact, such
actions are even opposed, as they manifest distrust and violate one of the core rules of
friendship. Without these mechanisms, friends can only “hope” that their friends will not
betray them despite having the ability to do so. However, it is also this very inherent, noncalculative trust that exposes friendship to the possibility of betrayal. Without constant
monitoring and reassessment, a friend can always turn their back on another at any moment.
Because of the possibility of betrayal, society has developed mechanisms to deter individuals from violating relational expectations in close relationships through the establishment
of moral codes of conduct. These moral codes are specific to each type of relationship, but
they usually fall under five universal moral foundational categories: harm/care, fairness/reciprocity, ingroup/loyalty, authority/respect, and purity/sanctity. These moral foundations
served as solutions to social problems in the early stage of societal development. However,
over time, as members of society forgot their origins and functions, these foundations became
embraced as morality, guiding perceptions of what is good and bad, right and wrong.
161

Betrayal is the antithesis of the morally good behavior that society expects from its
members, just as a friend expects loyalty from another friend. It is a vice that should be
avoided and punished if committed. This strong violation of relational expectations or codes
of conduct typically invites moral judgment and condemnation from the victim and others
in society. One of the qualities of moral foundations is that these judgments are made in the
blink of an eye because humans intuitively “feel” and “know” what behavior is considered
right or wrong. When we witness a betrayal, we instinctively “feel” bad about it, “know” it is
wrong, and immediately judge and condemn both the behavior and the person committing it.
When a victim experiences betrayal, it often instigates negative feelings towards both
the action and the betrayer. We do not initially understand and comprehend the action
consciously; instead, negative emotions guide our judgment. Psychologists have shown that
feelings of hurt, anger, and hatred are commonly associated with betrayal and violations
of trust. In Chapter 3, I found that these emotional reactions, especially pain and anger,
also arose when the public experienced betrayal by another country they considered a friend.
Moreover, in Chapter 4, through the case study of the British-Thai relations during the
Second World War, we see that British policymakers were deeply saddened and furious by
the Thai government’s decision to declare war on them, thus leading to outrage against the
Thai government that lasted beyond the War.
Most importantly, betrayal does not end with emotional reactions. Instead, these emotional responses often lead to outrage, confrontation, or withdrawal from the relationship.
As the survey experiment in Chapter 3 shows, betrayal leads to reductions in trust, positive
attitudes, and willingness to continue collaboration, as well as an increase in the desire to
162

punish the betrayer, mediated by feelings of anger and hurt. The case study in Chapter 4
also illustrates how anger and associated grudges led the British government to impose
penalties on Thailand at levels the country could hardly bear. These decisions, guided by
emotions, led the British to disproportionately punish Thailand—the betrayer—beyond what
was reasonable for retributive purposes. All of this occurred because the British believed
that Thailand deserved to pay the cost of betrayal.
Ways Forward
This dissertation has demonstrated that friendship is a vital and distinct aspect of international relations, with significant implications for how states interact. It reveals that the field
of international relations is not solely characterized by rationalistic self-interests; rather, it is
enriched by relationships, morality, emotions, and sometimes irrational decisions. However,
despite the insights gained from this research, the study on this topic is far from complete.
There are many more avenues through which this research can be pursued further.
One of the many directions that can be explored further is to examine the temporality of
the consequences of betrayal. Essentially, how long does it take before two countries begin
to rebuild their relationship? Do they hold grudges and hatred, and how long do these
emotions last? What does it take for the victim to forgive the betrayer? Studies in the field
of communication suggest that counterreactions by the betrayer play a significant role in
whether there is a chance for reconciliation (Fitness, 2006, 11). Betrayers may concede that
an offense has been committed and apologize, make excuses, provide justifications, or deny
wrongdoing. Do these behaviors occur at the international level, and what are the resulting
163

outcomes? Understanding the timeline and process of reconciliation can provide valuable
insights into the dynamics of forgiveness and the restoration of trust in international relations.
Another avenue for further exploration is to examine other case studies involving countries
with different combinations of power disparity. In the case of British-Thai relations, it was
arguably easier for the British to exert a policy of retribution on the Thai government, as
Britain was a much more powerful partner. However, not many countries have this luxury.
Investigating scenarios where the power dynamics are more balanced or where the betrayed
country is more powerful than the betrayer can provide a richer understanding of how power
influences the response to betrayal and the possibilities for reconciliation.
For example, in 1979, Taiwan felt deeply betrayed by the United States when it officially
recognized the People’s Republic of China (PRC) as the legitimate government of China
and severed formal diplomatic ties with Taiwan. Since Taiwan was a much smaller partner,
it could not express outrage at the United States in the same way Great Britain did to
Thailand. However, it is possible that Taiwan’s feelings of hurt eventually made the United
States feel guilty about its actions, leading to greater commitments to Taiwan than might
have otherwise been expected.
The case study of U.S.-Taiwan relations also opens up opportunities for further comparison. The U.S. recognition of the PRC set a precedent for other countries to recognize the
PRC instead of Taiwan. By examining the reactions of Taiwan to these countries, depending
on their prior closeness of relationship, we can gain a deeper understanding of how betrayal
and subsequent actions vary based on the dynamics of international friendships and alliances.
By addressing these areas, future research can build on the foundation laid by this disser164

tation, enhancing our understanding of the complex interplay between friendship, betrayal,
and state behavior in international relations. Recognizing and addressing the moral and
emotional dimensions of international relations can lead to a deeper understanding in the
field, ultimately contributing to more effective and nuanced diplomatic strategies.
At the very least, this dissertation challenges the notion that international politics is a
field devoid of morality and emotions. As demonstrated in this research, rationality does
not always best describe policymakers’ and the public’s understanding of friendship and
reactions to betrayal in international politics. Instead, emotions and moral considerations
play a significant role in shaping state behavior and international interactions.
165

References
Primary Documents
American Legation in Bangkok. (1945, November 21). [Telegram to Secretary of
State] (RG 59 741.92/11-2145). The National Archives and Records Administration
(NARA), College Park, MD.
Ballantine, J. W. (1944, March 17). [Memorandum of Conversation between George
Sansom and Joseph W. Ballantine]. Thailand (RG 59 Box 5835 892.01/54) The National
Archives and Records Administration (NARA), College Park, MD.
Bevin, E. (1945, August 20). Text of Passage Relating to Thailand in a Speech
Deliveredby the Secretary of State for Foreign Affairs, Mr. Bevin, in the House of
CommonsonAugust 20th, 1945. [Speech]. (RG 59 892.01/8-2045). The National
Archives and Records Administration (NARA), College Park, MD.
British Foreign Office. (1940, February 8). [Cypher telegram to Sir J. Crosby]. (BIM-6
Reel 21 BURMA 988/1940). Southern Region Library Facility University of California,
Los Angeles, Los Angeles, CA.
British Foreign Office. (1940, October 15). [Cypher telegram to Sir J. Crosby (Bangkok)]
(BIM-6 Reel 22 P.Z. 5767/1940). Southern Regional Library Facility University of
California, Los Angeles, Los Angeles, CA.
British Foreign Office. (1941, March 28). [Cypher Telegram to Viscount Halifax] (BIM-6
Reel 22 P.Z. 16909/1941). Southern Regional Library Facility University of California,
Los Angeles, Los Angeles, CA.
British Foreign Office. (1945, December). Proposed Teletype Conversation with
American Embassy, London (RG 59) [Memorandum]. The National Archives and
Records Administration (NARA), College Park, MD.
Churchill, W. (1941, December 8). Prime Minister Winston Churchill’s Broadcast on War
166

with Japan [Speech]. The Public’s Library and Digital Archive. http://www.ibiblio.org
/pha/timeline/411208ewp.html.
Crosby, J. (n.d.). [Telegram to British Foreign Office] (BIM-6 Reel 22 P.Z. 5648/1940).
Southern Regional Library Facility University of California, Los Angeles, Los
Angeles, CA.
Crosby, J. (1939, January 2). [Letter to Viscount Haifax] (BIM-6 Reel 21 P.Z.
1214/1939). Southern Regional Library Facility University of California, Los
Angeles, Los Angeles, CA.
Crosby, J. (1939, June 29). [Telegram to British Foreign Office] (BIM-2 Reel 19
P.Z. 4325/1939). Southern Regional Library Facility University of California, Los
Angeles, Los Angeles, CA.
Crosby, J. (1940, January 6). [Letter to Viscount Halifax] (BIM-6 Reel 21 BURMA
1073/1940). Southern Regional Library Facility University of California, Los
Angeles, Los Angeles, CA.
Crosby, J. (1940, July 15). [Letter to Air Vice Marshal] (BIM-6 Reel 19 P.Z. 4939/1940).
Southern Regional Library Facility University of California, Los Angeles, Los
Angeles, CA.
Crosby, J. (1940, July 15). [Letter to Air Vice Marshal] (BIM-6 Reel 19 P.Z. 4939/1940).
Southern Regional Library Facility University of California, Los Angeles, Los
Angeles,CA.
Crosby, J. (1940, October 15). [Telegram to British Foreign Office] (BIM-6 Reel 22
P.Z. 6767/1940). Southern Regional Library Facility University of California, Los
Angeles, Los Angeles, CA.
Crosby, J. (1940, November 13). [Telegram to British Foreign Office] (BIM-6 Reel 19
P.Z. 6291/1940). Southern Region Library Facility University of California, Los
Angeles, Los Angeles, CA
Crosby, J. (1941, January 15). [Telegram to British Foreign Office] (BIM-6 Reel 22
BURMA 10404/1941). Southern Regional Library Facility University of California,
Los Angeles, Los Angeles, CA.
Crosby, J. (1941, January 21). [Telegram to British Foreign Office] (BIM-6 Reel 22
BURMA 10521/1941). Southern Regional Library Facility University of California,
Los Angeles, Los Angeles, CA.
Crosby, J. (1941, January 27). [Telegram to British Foreign Office] (BIM-6 Reel 22
167

BURMA 10662/1941). Southern Regional Library Facility University of California,
Los Angeles, Los Angeles, CA.
Crosby, J. (1941, August 7). [Telegram to British Foreign Office] (BIM-6 Reel 23
BURMA 15902/1941). Southern Region Library Facility University of California,
Los Angeles, Los Angeles, CA
Department of State. (1941, December 18). [Aide-Memoire] (RG 59 Box 2236 711.92/30A).
The National Archives and Records Administration (NARA), College Park, MD.
Department of State. (1944, March 8). Attitude towards Thailand (339-339b Siam,
Government) [Memorandum for the President]. Franklin D. Roosevelt Presidential
Library, Hyde Park, NY.
Department of State. (1945, August 25). [Telegram to American Embassy in London]
(RG 59 892.01/8-2045). The National Archives and Records Administration (NARA),
College Park, MD.
Department of State. (1945, November 29). [Aide-Memoire] (RG 741.92/11-2845). The
National Archives and Records Administration (NARA), College Park, MD.
Department of State. (1945, December 22). [Telegram to American Political Advisor in
Bangkok] (RG 59 741.92/12-1545). The National Archives and Records Administration
(NARA), College Park, MD.
Department of State Divison of Far Eastern Affairs. (1941, December 27). [Report]
(RG 59 Box 2236 711.92/33). The National Archives and Records Administration
(NARA), College Park, MD.
Department of State Divison of Far Eastern Affairs. (1942, January 27). [Memorandum]
(RG 59 Box 2236 711.92/36). The National Archives and Records Administration
(NARA), College Park, MD.
Gallman, W. J. (1944, April 22). British Attitude Toward Thailand [Letter to the
Secretary of State]. (RG 59 Box 5835 892.01/4-2244). The National Archives and
Records Administration (NARA), College Park, MD.
Peck, W. R. (1942, August 22). Political, Economic, Financial and Military Conditions
in Thailand, December 1941 to June 1942 [Summarized Report]. (RG 59 Box
5834 892.00/233). The National Archives and Records Administration (NARA),
College Park, MD.
Pramoj, S. (1941, December 21). [Memorandum] (RG 59 Box 2236 711.92/31). The National
Archives and Records Administration (NARA), College Park, MD.
168

U.S. Legation in Bangkok. (1949, February). [Bangkok Monthly Political Report,
February 1941] (RG 59 Box 5835). The National Archives and Records Administration
(NARA), College Park, MD.
Winant, J. G. (1944, September 27). [Telegram to Secretary of State] (RG 59 Box
5835 892.01/9-2744). The National Archives and Records Administration (NARA),
College Park, MD.
Winant, J. G. (1945, September 8). [Telegram to Secretary of State] (RG 59 741.92/9-845).
The National Archives and Records Administration (NARA), College Park, MD.
Secondary Documents
Afifi, W. A. and Metts, S. (1998). Characteristics and Consequences of Expectation
Violations in Close Relationships. Journal of Social and Personal Relationships, 15(3):365–
392.
Alexander, M. G., Brewer, M. B., and Hermann, R. K. (1999). Images and affect: A
functional analysis of out-group stereotypes. Journal of Personality and Social Psychology,
77(1):78–93.
Argyle, M. and Henderson, M. (1984). The Rules of Friendship. Journal of Social and
Personal Relationships, 1(2):211–237.
Axelrod, R. M. (2009). The Evolution of Cooperation: Revised Edition. Basic Books, New
York.
Bas, M. A., Coe, A. J., and Gheorghe, E. (2024). Playing to the Audience: Responses to
Violations of International Order. The Journal of Politics, 86(1):113–125.
Bennett, D. S. (1997). Testing Alternative Models of Alliance Duration, 1816-1984. American
Journal of Political Science, 41(3):846.
Berenskoetter, F. (2007). Friends, There Are No Friends? An Intimate Reframing of the
International. Millennium: Journal of International Studies, 35(3):647–676.
Berenskoetter, F. (2014). Friendship, Security, and Power. In Koschut, S. and Oelsner, A.,
editors, Friendship and International Relations, pages 51–71. Palgrave Macmillan, London.
Bies, R. J. and Tripp, T. M. (1996). Beyond Distrust: "Getting Even" and the Need for
Revenge", pages 246–260. Sage Publications, Newsbury Park.
Carlsmith, K. M., Darley, J. M., and Robinson, P. H. (2002). Why do we punish? Deterrence
and just deserts as motives for punishment. Journal of Personality and Social Psychology,
83(2):284–299.
169

Clark, M. S. and Mills, J. R. (2011). A Theory of Communal (and Exchange) Relationships.
In Lange, P. A. M. V., Kruglanski, A. W., and Higgins, E. T., editors, Handbook of Theories
of Social Psychology. Volume 2. SAGE, Credo Reference, Los Angeles [California], Boston,
Massachusetts, [enhanced credo edition] edition.
Clinton, H. (2009). Secretary of State Clinton Says North Korea is “A Threat to Our Friends
and Allies”. Foreign Policy Bulletin, 19(4):16–23.
Dalton, M. (2011). France Calls U.S.-Australia Submarine Deal a Betrayal. Wall Street
Journal.
Deutsch, K. W. (1968). Political Community and the North American Area: International
Organization in the Light of Historical Experience. Princeton Legacy Library. Princeton
University Press, Princeton, New Jersey.
Druckman, J. N., Green, D. P., Kuklinski, J. H., and Lupiad, A. (2011). Experiments: An
Introduction to Core Concepts. In Druckman, J. N., Green, D. P., Kuklinski, J. H., and
Lupiad, A., editors, Cambridge Handbook of Experimental Political Science. Cambridge
university press, Cambridge.
Eznack, L. (2012). Crises in the Atlantic Alliance. Palgrave Macmillan US, New York.
Fearon, J. D. (1995). Rationalist explanations for war. International Organization, 49(3):379–
414.
Fearon, J. D. (1998). Bargaining, Enforcement, and International Cooperation. International
Organization, 52(2):269–305.
Fehr, B. and Baldwon, M. (1996). Prototype and script analyses of laypeople’s knowledge
of anger. In Fletcher, G. J. O. and Fitness, J., editors, Knowledge Structures in Close
Relationships: A Social Psychological Approach, pages 219–245. Lawrence Erlbaum, New
Jersey.
Finkel, E. J., Rusbult, C. E., Kumashiro, M., and Hannon, P. A. (2002). Dealing with
betrayal in close relationships: Does commitment promote forgiveness? Journal of
Personality and Social Psychology, 82(6):956–974.
Fiske, A. P. and Tetlock, P. E. (1997). Taboo Trade-offs: Reactions to Transactions That
Transgress the Spheres of Justice. Political Psychology, 18(2):255–297.
Fitness, J. (2006). Betrayal, Rejection, Revenge, and Forgiveness:An Interpersonal Script
Approach. In Leary, M. R., editor, Interpersonal Rejection, pages 72–103. Oxford
University Press.
Fitness, J. and Fletcher, G. J. O. (1993). Love, hate, anger, and jealousy in close relationships:
A prototype and cognitive appraisal analysis. Journal of Personality and Social Psychology,
65(5):942–958.
170

Frank, R. H. (1988). Passions within Reason: The Strategic Role of the Emotions. Norton,
New York, NY, 1. ed edition.
Gamson, W. A. (2021). Anger, betrayal, and humiliation: How veterans feel about the
withdrawal from Afghanistan. https://www.brookings.edu/articles/anger-betrayal-andhumiliation-how-veterans-feel-about-the-withdrawal-from-afghanistan/.
Hafner-Burton, E. M., Hughes, D. A., and Victor, D. G. (2013). The Cognitive Revolution
and the Political Psychology of Elite Decision Making. Perspectives on Politics, 11(2):368–
386.
Hafner-Burton, E. M., LeVeck, B. L., Victor, D. G., and Fowler, J. H. (2014). Decision Maker
Preferences for International Legal Cooperation. International Organization, 68(4):845–
876.
Hagan, J. D. (1994). Domestic Political Systems and War Proneness. Mershon International
Studies Review, 38(2):183.
Haidt, J. (2007). The New Synthesis in Moral Psychology. Science, 316(5827):998–1002.
Haidt, J. and Bjorklund, F. (2007). Social Intuitionists Answer Six Questions about Moral
Psychology. Moral Psychology, 2.
Haidt, J. and Joseph, C. (2004). Intuitive ethics: How innately prepared intuitions generate
culturally variable virtues. Dædalus, Fall:55–66.
Haidt, J. and Joseph, C. (2008). The Moral Mind: How Five Sets of Innate Intuitions
Guide the Development of Many Culture- Specific Virtues, and Perhaps Even Modules.
In Carruthers, P., Laurence, S., and Stich, S., editors, The Innate Mind, Volume 3. Oxford
University Press.
Haidt, J., Koller, S. H., and Dias, M. G. (1993). Affect, culture, and morality, or is it wrong
to eat your dog? Journal of Personality and Social Psychology, 65(4):613–628.
Haugevik, K. (2024). Friendship in International Politics. Oxford Research Encyclopedia of
International Studies.
Haugevik, K. M. (2018). Special Relationships in World Politics: Inter-State Friendship
and Diplomacy after the Second World War. New International Relations. Routledge,
Abingdon, Oxon ; New York, NY.
Hopp, F. R., Fisher, J. T., Cornell, D., Huskey, R., and Weber, R. (2021). The extended
Moral Foundations Dictionary (eMFD): Development and applications of a crowd-sourced
approach to extracting moral intuitions from text. Behavior Research Methods, 53(1):232–
246.
Howard, M. and Paret, P., editors (2008). On War. Princeton University Press.
171

Hyde, S. D. (2015). Experiments in International Relations: Lab, Survey, and Field. Annual
Review of Political Science, 18(1):403–424.
Hymans, J. E. (2010). The arrival of psychological constructivism. International Theory,
2(3):461–467.
Jayanama, D. (2017). Thai kap songkhr¯am l¯ok khrang th¯i 2 [Thailand and World War II].
S¯ipany¯a, Nonthabur¯i, phim khrang th¯i 5 edition.
Jervis, R. (2017). Perception and Misperception in International Politics. Princeton
University Pres, Princeton, New Jersey, new edition edition.
Johnson, D. D. and Tierney, D. (2011). The Rubicon Theory of War: How the Path to
Conflict Reaches the Point of No Return. International Security, 36(1):7–40.
Johnson, D. D. and Tierney, D. (2019). Bad World: The Negativity Bias in International
Politics. International Security, 43(3):96–140.
Jones, W. H., Couch, L., and Scott, S. (1997). Trust and Betrayal. In Handbook of Personality
Psychology, pages 465–482. Elsevier.
Joskowicz-Jabloner, L. and Leiser, D. (2013). Varieties of trust-betrayal: Emotion and
relief patterns in different domains: Varieties of trust-betrayal. Journal of Applied Social
Psychology, 43(9):1799–1813.
Kennedy, J. F. (1961). Inaugural address.
Keohane, R. O. (1984). After Hegemony: Cooperation and Discord in the World Political
Economy. Princeton University Press, Princeton, N.J.
Khong, Y. F. (1992). Analogies at War: Korea, Munich, Dien Bien Phu, and the Vietnam
Decisions of 1965. Princeton Paperbacks. Princeton University Press, Princeton, N.J.
King, P. (2007). Friendship in Politics. Critical Review of International Social and Political
Philosophy, 10(2):125–145.
Koschut, S. and Oelsner, A. (2014). A Framework for the Study of International Friendship.
In Koschut, S. and Oelsner, A., editors, Friendship and International Relations, pages
3–34. Palgrave Macmillan UK, London.
Kozlowski, A. C., Taddy, M., and Evans, J. A. (2019). The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings. American Sociological Review,
84(5):905–949.
Kramer, R. M. and Tyler, T. (2001). Developing and maintaining trust in work relationships.
In Kramer, R. M. and Tyler, T. R., editors, Trust in Organizations: Frontiers of Theory
and Research. Sage, Thousand Oaks, nachdr. edition.
172

Kydd, A. H. (2007). Trust and Mistrust in International Relations. Princeton University
Press, Princeton, NJ Oxford, third printing and first paperback printing edition.
Lalman, D. and Newman, D. (1991). Alliance formation and national security. International
Interactions, 16(4):239–253.
Lang, A. F. (2008). Punishment, Justice and International Relations: Ethics and Order after
the Cold War. Contemporary Security Studies. Routledge, London ; New York.
Leary, M. R., Springer, C., Negel, L., Ansell, E., and Evans, K. (1998). The causes,
phenomenology, and consequences of hurt feelings. Journal of Personality and Social
Psychology, 74(5):1225–1237.
Leeds, B. A. (2003). Alliance Reliability in Times of War: Explaining State Decisions to
Violate Treaties. International Organization, 57(4):801–827.
Leeds, B. A. (2018). Alliance Treaty Obligations and Provisions (ATOP) Codebook.
Leeds, B. A., Long, A. G., and Mitchell, S. M. (2000). Reevaluating Alliance Reliability:
Specific Threats, Specific Promises. Journal of Conflict Resolution, 44(5):686–699.
Leng, R. J. and Goodsell, R. (1974). Behavioral indicators of war proneness in bilateral
conflicts. In Sage International Yearbook of Foreign Policy Studies, volume 2, pages 191–
226. SAGE, Beverly Hills, CA.
Lerner, J. S. and Keltner, D. (2001). Fear, anger, and risk. Journal of Personality and Social
Psychology, 81(1):146–159.
Martin, J. V. (1963). Thai-American Relations in World War II. The Journal of Asian
Studies, 22(4):451–467.
Mcdermott, R. (2011). New Directions for Experimental Work in International Relations:
New Directions for Experimental Work. International Studies Quarterly, 55(2):503–520.
Mearsheimer, J. J. (1994). The False Promise of International Institutions. International
Security, 19(3):5.
Mercer, J. (1996). Reputation and International Politics. Cornell Studies in Security Affairs.
Cornell Univ. Press, Ithaca, NY.
Mercer, J. (2010). Emotional Beliefs. International Organization, 64(1):1–31.
Mooijman, M., Van Dijk, W. W., Ellemers, N., and Van Dijk, E. (2015). Why leaders punish:
A power perspective. Journal of Personality and Social Psychology, 109(1):75–89.
Morgenthau, H. J. (1945). The Evil of Politics and the Ethics of Evil. Ethics, 56(1):1–18.
Morrison, E. W. and Robinson, S. L. (1997). When Employees Feel Betrayed: A Model of
How Psychological Contract Violation Develops. The Academy of Management Review,
22(1):226.
173

Morrow, J. D. (1991). Alliances and Asymmetry: An Alternative to the Capability
Aggregation Model of Alliances. American Journal of Political Science, 35(4):904.
Morrow, J. D. (1994). Alliances, Credibility, and Peacetime Costs. Journal of Conflict
Resolution, 38(2):270–297.
Morrow, J. D. (2000). Alliances: Why Write Them Down? Annual Review of Political
Science, 3(1):63–83.
Mullen, E. and Skitka, L. J. (2006). Exploring the psychological underpinnings of the
moral mandate effect: Motivated reasoning, group differentiation, or anger? Journal
of Personality and Social Psychology, 90(4):629–643.
Niou, E. M. S. and Ordeshook, P. C. (1994). Alliances in Anarchic International Systems.
International Studies Quarterly, 38(2):167.
Oelsner, A. (2007). Friendship, Mutual Trust and the Evolution of Regional Peace in the
International System. Critical Review of International Social and Political Philosophy,
10(2):257–279.
Oelsner, A. and Koschut, S. (2014). A Framework for the Study of International Friendship.
In Friendship and International Relations, pages 3–34. Palgrave Macmillan, Hampshire.
Packer, G. (2022). The Betrayal. https://www.theatlantic.com/magazine/archive/2022/03/bidenafghanistan-exit-american-allies-abandoned/621307/.
Page, S. (2020). Read the full transcript of vice presidential debate between Mike Pence and
Kamala Harris. USA Today.
Pennington, J., Socher, R., and Chirstopher D., M. (2014). GloVe: Global Vectors for Word
Representation. In Proceedings of the 2014 Conference on Empirical Methods in Natural
Language Processing, pages 1532–43.
Planalp, S. and Fitness, J. (1999). Thinking/feeling about social and personal relationships.
Journal of Social and Personal Relationships, 16:731–750.
Pomeroy, C. and Rathbun, B. C. (2023). Just business? Moral condemnation and virtuous
violence in the American and Russian mass publics. Journal of Peace Research, page
002234332211497.
Rao, V. S. and Lee, S. J. (2007). Responses to Trust Violation: A Theoretical Framework.
Journal of Computer Information Systems, 48(1):76–84.
Rapport, A. (2013). The Long and Short of It: Cognitive Constraints on Leaders’
Assessments of “Postwar” Iraq. International Security, 37(3):133–171.
174

Rathbun, B. C. (2023). Right and Wronged in International Relations: Evolutionary Ethics,
Moral Revolutions, and the Nature of Power Politics. Number 163 in Cambridge Studies in
International Relations. Cambridge University Press, Cambridge, United Kingdom ; New
York, NY.
Rathbun, B. C. and Pomeroy, C. (2021). See No Evil, Speak No Evil? Morality, Evolutionary
Psychology, and the Nature of International Relations. International Organization, pages
1–34.
Rathbun, B. C. and Pomeroy, C. (2022). See No Evil, Speak No Evil? Morality, Evolutionary
Psychology, and the Nature of International Relations. International Organization,
76(3):656–689.
Rathbun, B. C., Rathbun, N. S., and Pomeroy, C. (2022). No Fair! Distinguishing
Between the Pursuit of Status and Equity in International Relations. International Studies
Quarterly, 66(1):sqac002.
Reynolds, E. B. (1994). Thailand and Japan’s Southern Advance, 1940-1945. St. Martin’s
Press, New York.
Reynolds, E. B. (2010). Thailand’s Secret War: OSS, SOE and the Free Thai Underground
during World War II. Cambridge Military Histories. Cambridge University Press, Cambridge.
Robinson, S. L. and Rousseau, D. M. (1994). Violating the psychological contract: Not the
exception but the norm. Journal of Organizational Behavior, 15(3):245–259.
Roger, C. and Shear, M. D. (2021). Furious Over Sub Deal, France Recalls Ambassadors to
U.S. and Australia. The New York Times.
Roščin, E. (2017). Friendship among Nations: History of a Concept. Manchester University
Press, Manchester.
Schelling, T. C. (2008). Arms and Influence. Yale University Press, New Haven, CT.
Shaw, G. D. T. and Schall, J. V. (2015). The Lost Mandate of Heaven: The American
Betrayal of Ngo Dinh Diem, President of Vietnam. Ignatius Press, San Francisco.
Silver, A. (1989). Friendship and trust as moral ideals: An historical approach. European
Journal of Sociology, 30(2):274–297.
Skitka, L. J. (2002). Do the Means Always Justify the Ends, or Do the Ends Sometimes
Justify the Means? A Value Protection Model of Justice Reasoning. Personality and
Social Psychology Bulletin, 28(5):588–597.
Skitka, L. J. (2010). The Psychology of Moral Conviction. Social and Personality Psychology
Compass, 4(4):267–281.
175

Skitka, L. J. and Mullen, E. (2002). The Dark Side of Moral Conviction. Analyses of Social
Issues and Public Policy, 2(1):35–41.
Skitka, L. J. and Mullen, E. (2008). Moral Convictions Often Override Concerns About
Procedural Fairness: A Reply to Napier and Tyler. Social Justice Research, 21(4):529–
546.
Snyder, G. H. (1984). The Security Dilemma in Alliance Politics. World Politics, 36(4):461–
495.
Tetlock, P. E. (2006). Expert Political Judgment: How Good Is It? How Can We Know?
Princeton Univ. Press, Princeton, N.J., 6. pr., and 1. paperback pr edition.
Turnley, W. H. and Feldman, D. C. (1998). Psychological contract violations during
corporate restructuring. Human Resource Management, 37(1):71–83.
Walt, S. M. (2019). The Origins of Alliances. Cornell University Press.
Waltz, K. N. (1979). Theory of International Politics. Addison-Wesley Series in Political
Science. Addison-Wesley Pub. Co, Reading, Mass.
Weeks, J. L. (2008). Autocratic Audience Costs: Regime Type and Signaling Resolve.
International Organization, 62(01).
Welch, D. A. (1993). Justice and the Genesis of War. Number 29 in Cambridge Studies in
International Relations. Cambridge University Press, Cambridge [England] ; New York,
NY, USA.
Wendt, A. (1999). Social Theory of International Politics. Cambridge University Press, 1
edition.
Williamson, O. E. (1993). Calculativeness, Trust, and Economic Organization. The Journal
of Law and Economics, 36(1, Part 2):453–486.
Wyszomirski, M. J., Burgess, C., and Peila, C. (2003). International cultural relations: A
multi-country comparison. Technical report, Arts Policy and Administration Program The
Ohio State University, Columbus, OH.
Young, S. (2023). Kissinger’s Betrayal: How America Lost the Vietnam War. RealClear
Publishing, Herndon, VA.
Yu, Q. and Li, B. (2017). Mma: An R Package for Mediation Analysis with Multiple
Mediators. Journal of Open Research Software, 5(1):11.
176

Appendices
Appendix A: Survey Instruments
Demographic battery
• First, let’s learn about you. What is your gender?
– Male
– Female
– Non-binary / third gender
• What year were you born? Please enter the full, four digit year.
• Are you of Spanish, Hispanic, or Latino origin?
– Yes
– No
• What race do you consider yourself to be?
– White
– Black or African American
– American Indian or Alaska Native
– Asian
177

– Native Hawaiian or Pacific Islander
– Mixed race
– Other
• What is the highest level of education you have completed?
– Some high school or less
– High school graduate
– Some college, but no degree
– Associates or technical degree
– Bachelor’s degree
– Graduate or professional degree (MA, MS, MBA, PhD, JD, MD, DDS etc.)
• Thinking about politics these days, how would you describe your political viewpoints?
– Left
– Left leaning
– Center
– Right leaning
– Right
• In politics, as of today, do you consider yourself a Republican, a Democrat, or an
independent?
– Republican
– Democrat
– Independent
178

• (If answered independent) As of today, do you lean more to the Democratic Party or
the Republican Party?
– Democratic Party
– Republican Party
• In which region of the United States are you from?
– Northeast (ME, NH, VT, MA, RI, CT, NY, NJ, and PA)
– Southeast (DE, DC, MD, VA, WV, KY, NC, SC, TN, GA, FL, AL, MS, AR, LA)
– West (MT, ID, WY, CO, UT, NV, CA, OR, WA, AK, HI)
– Southwest (AZ, NM, TX, OK)
– Midwest (OH, MI, IN, WI, IL, MO, IA, MN, ND, SD, NE, KS)
Foreign policy preferences battery
Isolationism
• Our allies are perfectly capable of defending themselves, and they can afford it. This
allows the United States to focus on internal threats to its well-being rather than
external ones.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
179

– Strongly agree
• The United States needs to play an active role in resolving conflicts around the world.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• We should not focus so much on international matters but concentrate more on our
own national problems.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• American’s conception of its leadership role in the world needs to be scaled down.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
180

• The United States should focus on its own business internationally and allow other
countries to manage at best they can.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
Cooperative internationalism
• The United States needs to cooperate more with the United Nations.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• It is essential for the United States to work with other nations to solve problems such
as overpopulation, hunger, and pollution.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
181

– Strongly agree
• Promoting and defending human rights in other countries is of utmost importance.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Helping to improve the standard of living in less developed country is of utmost
importance.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Protecting the global environment is of utmost importance.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
182

Military internationalism
• The United States should take all necessary measures, including the use of force, to
prevent aggression by any expansionist power.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Instead of simply countering our opponents’ thrust, it is necessary to strike at the
heart of our opponent’s power.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• There is considerable validity in the theory that when one national falls to aggressor
nations, others nearby will soon follow a similar path.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
183

– Somewhat agree
– Strongly agree
• American military strength is not the best way to ensure world peace.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
Trust in the international system
• Generally speaking, international relations is rule-driven.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
Generalized trust scale
• Most people tell a lie when they benefit by doing so.
– Strongly disagree
– Somewhat disagree
184

– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Those devoted to unselfish causes are often exploited by others.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Some people do not cooperate because they pursue only their own short-term selfinterest. Thus things that can be done well if people cooperate often fail because of
these people.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• Most people are basically honest.
– Strongly disagree
– Somewhat disagree
185

– Neither agree nor disagree
– Somewhat agree
– Strongly agree
• There will be more people who will not work if the social security system is developed
further.
– Strongly disagree
– Somewhat disagree
– Neither agree nor disagree
– Somewhat agree
– Strongly agree
Foreign policy attentiveness
• In a typical week, how often do you consume international news?
– More than 5 times
– 4-5 times
– 2-3 times
– 1 time
– Never
Attention Checks
• What is Country (A/B)’s relationship with the United States?
186

– Regular partner
– Treaty ally
– Treaty ally and longstanding friend
• (Scenario A) What was Country A’s response to the U.S. request?
– Country A has not given any definite answer.
– Country A has declined to participate in the joint effort.
– Country A has declined to participate in the joint effort and declared collaboration
with the invading force.
• (Scenario B) What was Country B’s response to the U.S. request?
– Country B is waiting for the weapons transfer.
– Country B has canceled the purchase from the United States.
– Country B has canceled the purchase from the United States and has concluded
a deal for the same type of weapons with another country.
Manipulation checks
• Do you think Country (A/B) is a friend of the United States?
– Definitely yes
– Probably yes
– Might or might not be
– Probably not
187

– Definitely not
Pre-test measurements
• How much do you trust Country (A/B)?
– A great deal
– A lot
– A moderate amount
– A little
– Not at all
• How positive do you feel towards Country A/B?
– Extremely positive
– Somewhat positive
– Neither positive nor negative
– Somewhat negative
– Extremely negative
• Do you want the United States to continue to collaborate with Country A/B?
– Definitely yes
– Probably yes
– Might or might not
– Probably not
188

– Definitely not
Post-test measurements
• Upon learning about Country (A/B)’s decision, how much do you trust Country
(A/B)?
– Not at all
– A little
– A moderate amount
– A lot
– A great deal
• How positive do you feel towards Country (A/B)?
– Extremely negative
– Somewhat negative
– Neither positive nor negative
– Somewhat positive
– Extremely positive
• Please rate the intensity of the following feelings you experience towards Country
(A/B)’s decision on a scale of 0-5, where 0 indicates the least intensity, and 5 indicates
the most intensity?
– Happy
189

– Angry
– Betrayed
– Surprised
– Hurt
• Do you want to punish Country (A/B) for its decision?
– Yes
– No
• How strongly do you feel on a scale of 0-5 about the following possible policy reactions
to Country (A/B)’s behavior, where 0 indicates the least agreement, and 5 indicates
the highest agreement?
– Shame country (A/B) in public
– Suspend diplomatic relations with Country (A/B)
– Launch economic sanctions against Country (A/B)
– Withdraw military cooperation with Country (A/B)
• If Country (A/B) is punished, what should be the primary purpose of the punishment?
– To stabilize the international system.
– To make Country (A/B) pay for it.
– To protect other countries.
– To act as a deterrent to others.
190

• Do you want the United States to continue to collaborate with Country (A/B) in the
future?
– Not at all
– A little
– A moderate amount
– A lot
– A great deal
Appendix B: Survey Sample
Basic demographics
Our survey sample is gender balanced (see Figure B1) and somewhat representative of the
demographics of the United States, as indicated by the 2020 Census data. Table B1 in the
appendix illustrates that, aside from the overrepresentation of respondents aged between 25
to 44 years old and those with bachelor’s degrees, the other dimensions of the demographics
of the survey sample align closely with the breakdown of basic demographics indicated
in the U.S. Census data.
Table B1: Basic demographics of the survey sample compared to the U.S. Census data
U.S. Census
(Percent)
Survey Sample
(Percent)
Gender
Male 49.1 50.5
Female 50.9 48.5
Non-binary / third-gender N/A 1.0
191

U.S. Census
(Percent)
Survey Sample
(Percent)
Age
18 to 24 years 9.4 8.0
25 to 44 years 26.3 53.9
45 to 54 years 25.4 18.2
55 to 64 years 12.3 13.2
65 to 74 years 16.8 5.8
75 to 84 years 10.0 1.0
85 years and over 1.9 0.0
Race
White 75.5 71.3
Black 13.6 12.8
American Indian and Alaska Native 1.3 0.4
Asian 6.3 9.6
Native Hawaiian and Other Pacific
Islander
0.3 0.4
Mixed race 3.0 4.0
Education
Some high school or less NA 1.4
High school or equivalent degree 26.1 14.0
Some college, no degree 19.1 17.6
Associate’s degree 8.8 13.2
Bachelor’s degree 21.6 35.9
Graduate or professional degree 14.0 18.0
Political Ideology and Party Affiliation
For ideology, which was employed as a quota sample alongside gender, the survey sample
indicates a rough balance between left (22.8 percent) and left-leaning (20.2 percent) populations and right (18.6 percent) and right-leaning (18.6 percent) populations, with the
majority of respondents identifying at the center (23.8 percent). Nevertheless, the sample
slightly tilts towards liberals on the political spectrum, suggesting the liberal-leaning nature
192

0
50
100
150
200
250
Female Male Third gender
Count
Figure B1: Gender
0
50
100
<20 20−29 30−39 40−49 50−59 60−69 >70
Count
Figure B2: Age
0
100
200
300
American Indian
or Alaskan Native
Asian
Black or
African American
Mixed race
Native Hawaiian
or Pacific Islander
White
Other
Count
Figure B3: Race
0
50
100
150
Some high school
or less
High school graduate
Some college,
but no degree
Associates or
technical degree
Bachelor's degree
Graduate or
professional degree
Count
Figure B4: Education
193

of the survey platform.
In terms of party affiliation, the sample is, once again, roughly balanced but slightly
tilted towards respondents who are affiliated with the Democratic Party (39.9 percent) as
opposed to the Republican Party (28.3 percent). Even among independent respondents (31.7
percent), the majority of them lean towards the Democratic Party (59.7 percent) compared
to the Republican Party (40.3 percent). This further suggests a leaning towards liberals
on the platform. However, since the bias was not significant, this should not affect the
validity of the experiment.
0
25
50
75
100
Left Left−leaning Center Right−leaning Right
Count
Figure B5: Ideology
0
50
100
150
200
Democrat Independent Republican
Count
Figure B6: Party affiliation
0
25
50
75
Democratic Party Republican Party
Count
Figure B7: Party leaning
194

Abstract (if available)

Abstract Is there such a thing as betrayal in international relations? Conventional wisdom suggests that because interstate relations are transactional, betrayal and the feelings and outcomes associated with it do not have a significant place in international politics. In this realm, there are no permanent friends, only permanent interests. Therefore, emotional bonds that lead to emotional reactions when a country reneges on a commitment or defects are thought not to exist.

However, in this dissertation, I argue that betrayal is indeed significant in international relations because friendship plays a crucial role in this realm. Due to friendship, countries, including their policymakers and the public, apply the moral codes of friendship to their relationships with other states. Just like in interpersonal relationships, countries expect friends to remain loyal and refrain from actions that violate the relationship’s rules. Specifically, they expect friends not to abandon or betray them despite having the ability to do so. Because of these moral expectations, when a friend reneges on a commitment, defects, or abandons, it is perceived as betrayal. Betrayal is judged as morally wrong, leading to strong negative emotional reactions, punitive actions, and the termination of the relationship.

This dissertation utilizes a multi-method approach, comprising word embeddings analysis, a survey experiment, and a case study of British-Thai relations during the Second World War, compared to the shadow case of U.S.-Thai relations during the same period. The findings suggest that international relations are not solely governed by self-interests but are deeply intertwined with emotional and moral considerations. This research opens new avenues for understanding state behavior, emphasizing the need to consider both rational and emotional factors in explaining interstate relations and international politics.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

When passions run high: emotions and the communication of intentions in face-to-face diplomacy

PDF

The Japan-South Korea history issue: the emotional outcome of trusting acts and met/unmet alignment security expectations

PDF

International politics and domestic institutional change: the rise of executive war-making autonomy in the United States

PDF

Classrooms are game settings: learning through and with play

PDF

Can ideologically relevant threat shift group-oriented values? Relevant threatening tweets cause Progressives to be as prejudiced as Conservatives

PDF

Habits and policy: the socio-cognitive foundations of foreign policy stability

PDF

Socio-ecological psychology of moral values

PDF

Explaining the variation in the extent of China's hegemony, 1279-1840 — multiple sources of ideational power beyond Confucianism

PDF

The politics of transformation: mass media and the Northern Ireland peace process

PDF

The neuroscience of ambivalent and ambiguous feelings

PDF

Perspectives on state capacity and the political geography of conflict

PDF

Difficulty-as-importance and difficulty-as-impossibility: unpacking the context-sensitivity and consequences of identity-based inferences from difficulty

PDF

Intuitions of beauty and truth: what is easy on the mind is beautiful and true

PDF

Like father, like son? A succession-based explanation for conflict initiation by authoritarian regimes

PDF

Thawing rivalries and fading friendships: a multi-method approach to rapprochement and alienation

PDF

Perceiving and coping with threat: explaining East Asian perceptions toward China’s rise

PDF

Public opinion and international affairs: a multi-method approach to foreign policy attitudes

PDF

Selflessness takes time: altruistic (but not cooperative) prosocial behavior increases with decision time

PDF

Lesser evils? WMD pursuit beyond nuclear weapons

PDF

Identity and regional integration: case studies on central Asia and southeast Asia

Asset Metadata

Creator Sawasdipakdi, Pongkwan (author)

Core Title Hard feelings: friendship, betrayal, and outrage in international relations

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science and International Relations

Degree Conferral Date 2024-08

Publication Date 06/20/2024

Defense Date 06/05/2024

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

Tag Alliance,Betrayal,emotion,Foreign policy,friendship,International Relations,moral psychology,Morality,OAI-PMH Harvest,political psychology,trust

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Rathbun, Brian (committee chair), Hymans, Jacques (committee member), Oyserman, Daphna (committee member)

Creator Email psawasdipakdi@gmail.com,sawasdip@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113996WT0

Unique identifier UC113996WT0

Identifier etd-Sawasdipak-13132.pdf (filename)

Legacy Identifier etd-Sawasdipak-13132

Document Type Dissertation

Format theses (aat)

Rights Sawasdipakdi, Pongkwan

Internet Media Type application/pdf

Type texts

Source 20240620-usctheses-batch-1172 (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.

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