Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The motivated affective behavior system: a dynamic account of the attachment behavioral system
(USC Thesis Other)
The motivated affective behavior system: a dynamic account of the attachment behavioral system
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Running head: MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 1
The Motivated Affective Behavior System:
A Dynamic Account of the Attachment Behavioral System
Jennifer Rose Talevich
University of Southern California
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 2
Acknowledgements
To my advisor, Dr. Stephen J. Read, I am most thankful. Throughout these five years his breadth
of knowledge, support and guidance for my ideas, and patience when my own was well
exceeded, have been sustaining gifts. Together, we have managed to accomplish what is
generally several years’ worth of work in a 12 months span. Without his willingness to invest so
much of his time in mentoring this year I would certainly not be graduating now. This latter is
also true of Dr. Margaret Gatz who made it possible for me to obtain the funding necessary to
concentrate on my research in this final year. I would also like to thank David Walsh, Norman
Miller, and John Monterosso for their encouragement of my good ideas, invaluable criticism of
those that weren’t so good, and most of all, for their excellent humor and friendship. I would
also like to thank the two people whose enthusiastic support of my academic dreams, and
certainty that I could achieve them, has been unflagging since I first dreamt them: Dr. Charles
Chubb, my undergraduate adviser and first co-author, and my uncle Frederick J. Hickman. Also
deserving thanks for many contributions to my education are my father, James R. Talevich, my
step-mother Laurie, and my step-advisor Lynn Miller. I’d especially like to thank the first for
inspiring Study 4. To quote a seven-year-old me, “Thanks Dad: I win.”
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 3
Table of Contents
Acknowledgements …………………………..……………………….….……………………… 2
Abstract …………………………………………………………………………..……………… 4
Introduction …………………………………..…………………….…………….……………… 5
Study 1 ……………………………………………..…….………………………..…………… 29
Study 2 ……………………………………………….………………………....……………… 38
Study 3 ……………………………………………….………………………………………… 48
Study 4 …………………………………………….………………………..……..…………… 59
General Discussion ……………………………..……………………………....……...….…… 71
References ……………………………………………………………………………………… 87
Tables …………………………………………………….……………………………..……… 95
Figures ………………………………………………………………………………………… 114
Appendix …………………………………………………………………..…..……………… 126
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 4
Abstract
A model of motivated affective behavior system (MABS) is reviewed and validated. Four
human-data studies test a computational model (Talevich, 2012) that integrates attachment
theory, several models of emotion processing, and goal systems theory. Previous work on
attachment models the dynamic experiences that serve to activate and deactivate the attachment
system (Mikulincer & Shaver, 2003; Shaver & Mikulincer, 2002). In contrast, the current work
models the dynamics within the attachment system: the internal mechanisms that underlie
activation, hyperactivation, and deactivation. MABS is a mediational process model. Appraisals
of a situation influence behavior through an intervening motivation sub-system, which, in turn,
triggers emotional responses that call for action (or inaction). The motivation subsystem
includes Approach and Avoidance Motive Systems and a Goal-Situation Congruence system.
The final product is an assessment of the total impact a situation has upon one’s goals (Goal
Impact). Attachment research generally examines emotions as outcomes with regard affect
regulation. The present model may be the first to treated as functional signals for attachment
behavior. This provides a highly detailed account of the emotion and motivation processing that
is missing in attachment research. Furthermore, personality is examined as chronic activations
within in this intervening process.
Keywords: behavioral system, emotion, affect, motivation, goals, motivated behavior,
construals, working models, schemas, attachment, close relationships, dating, courtship,
relationship dissolution, Goal Impact, cheating
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 5
The Motivated Affective Behavior System:
A Dynamic Account of the Attachment Behavioral System
The Motivated Affective Behavior System (MABS) is an intervening process model by
which an objective situation is mediated by construals, a motivation sub-system and emotions, to
evoke behaviors. Traits are envisioned as chronic associations between each element in this
process and, as such, are expected to both influence and be mediated by each intervening step.
Few psychological theories examine these relationships in a single framework. Attachment
theory, which delineates the purpose and process of human bonding, is an interesting exception.
The immense predictive power of attachment orientations has made attachment theory
one of the broadest and most influential theories of personality and its formation, as well as of its
impact on social behavior (Mikulincer & Shaver, 2012). The attachment processes specified by
Bowlby include working models of the self and other, “set goals,” and behaviors. Some
emotions are specified -such as the angry tone underscoring protestations. The patterns among
these elements, as they become rote, are the basis for a trait “attachment style” or “attachment
orientation”.
Despite the centrality of motivation in Bowlby’s theory, surprisingly little research has
examined associations between working models and activation of behavioral goals. Several
authors have noted this short-coming (Collins & Allard, 2004; Gillath et al., 2006). Prominent
attachment researchers have called on the field to move forward from the current form of
attachment theory and its research base to a more comprehensive behavioral systems theory of
motivation and personality (Mikulincer & Shaver, 2012). For, though the role of motivation is
theoretically acknowledged, there is a scarcity of empirical work examining the motivational
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 6
processes involved in the process of human bonding, the development of attachment orientations,
and the ways in which attachment orientations influence behavior.
Emotion processing has received even less attention in the attachment literature than has
motivation. This is a rather large oversight given the advances over the last few decades in our
understanding of the purpose of emotions. The field generally now agrees that emotions serve
the evolutionary adaptive purpose of signaling behavioral responses -like sophisticated reflexes
(Smith & Lazarus, 1990). Emotions are a consistent topic of discussion in the attachment
literature. They are generally discussed as positive or negative, as outcomes in of and of
themselves or in relation to coping (Mikulincer & Shaver, 2003) and affect regulation
(Mikulincer, Shaver, & Pereg, 2003). When discrete emotions, such as anger, are discussed the
focus is on situational variables that evoke emotional reactivity and the consequences this has
upon one’s partner (Mikulincer, Shaver, Gillath, & Nitzberg, 2005). But the functional role of
emotions as a signal for one’s own attachment behaviors has yet to be empirically demonstrated,
to this author’s knowledge. Therefore, although emotions are an important part of attachment
theory and a common theme in attachment research, the functional role of specific emotions in
eliciting specific attachment-related behaviors has not been previously examined and thus will be
a unique contribution of this work.
Despite the centrality of motivation to attachment theory, it has received
disproportionally little attention by empirical researchers. And though emotions have received a
fair amount of attention, as discussed above, their functional role as a mediator between
motivation and behavior has been virtually ignored. To this author’s knowledge, motivation and
emotions have not been empirically examined as mediators, between working models and
behaviors, in the same model. The dynamics described by Bowlby are assumed and sometimes
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 7
descriptively implied but, in empirical work, they are often skipped altogether. For instance, the
dynamics of attachment have been verbally modeled by Mikulincer and Shaver (Mikulincer &
Shaver, 2003; Mikulincer et al., 2003; Shaver & Mikulincer, 2002). They delineate a theoretical
verbal model of the dynamics by which the attachment system is activated and reactivated or
deactivated by repeated experience. Their model goes from construals (“is an attachment figure
available?”) directly to behavior (e.g. “seek proximity”). In contrast, MABS is an empirical
computational model of the dynamics within the attachment behavioral system: the process that
intervenes between a construal that activates the attachment system and the behavioral strategy
that is selected. The current work is intended to provide further empirical evidence with human
data to validate the MABS computational model.
In the current work, Bowlby’s concept of the behavioral system (of which attachment is
but one) is combined with (comparatively more) recent theories from the motivation and emotion
literature. Thus, we will begin with an introduction to attachment theory. Next we will consider
construals and their knowledge structures variously called schemas or working models.
Thereafter, a motivation sub-system will be introduced, which includes four elements: the
approach and avoidance goal systems that are activated by a situation, their congruence with the
that situation, and a computation called Goal Impact which is the product of this subsystem.
This approach is grounded in generations of motivation theory that will be reviewed. Goal
Impact is proposed to determine which emotion is evoked by the situation and, in accordance
with appraisal theories of emotion (Smith & Lazarus, 1990), provokes the behavior most called
for by the environment. Goal Impact is a calculation that is simple, like expectancy-value
assessments, but may more closely map onto the psychological process underlying the appraisal
process. The studies presented here are based upon human data but the theoretical underpinning
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 8
of the experimental and survey designs were pre-tested as a neural network model. The
description of this model will also serve as a summary and integration of the MABS elements.
The Attachment Behavioral System
Attachment is one of five Behavioral System proposed by Bowlby (1969) Other systems
include affiliation, caregiving, exploration, and Shaver et al. have added power (2011). The
attachment system will be the focus of this set of studies but the mechanisms of the current
model are proposed to apply to the other behavioral systems as well. Implications of the current
model for these other systems will be addressed under future directions in the general discussion.
Each behavioral system is proposed to have a biological function and a set of behaviors
that serve this function. The function of the attachment system is to obtain practical, as well as
emotional, security through other people. Bowlby called this function a “set goal” indicating a
homeostasis-like setting of “felt security.” Any deviation motivates one to regain felt security.
At birth, human infants are entirely helpless. They cannot move themselves away from danger
nor toward food. All they can really manage is to cry, squirm, and look at things. Fortunately,
the Caregiving Behavioral System in adults motivates them to respond positively to these
behaviors. Adults respond in helpful ways when an infant cries or looks at them with big eyes:
the first attachment strategies. As the infant grows, his behaviors become more sophisticated.
But they serve the same purpose: to have, at-the-ready, the necessary means by which to quickly
obtain help from others when needed.
The attachment system continues to operate in adulthood –bonding adults together in
couples (called “pair-bonds”) who provide mutual caregiving to fulfill each other’s attachment
needs. This relationship is not a mere byproduct of infant attachment but part of what sustains
it. For the helplessness of the human infant is heavily resource-consuming. For this reason, in
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 9
the Pleistocene hunter-gatherer era of human development, father involvement was critical to the
survival of offspring (Miller & Fishkin, 1997; Miller, Pedersen, & Putcha-Bhagavatula, 2005).
The bond between mother and father became as critical to infant survival as the child’s own bond
to each parent: for it kept the father present, (at least) through gestation so he himself could bond
to the child. Essentially, attachment and caregiving are the ties that bind families together. And
families are the start: they are the most basic social unit of the world’s most social species.
Attachment Patterns in the Motivated Affective Behavioral System. If a person,
adult or child, has a strong bond with their attachment figure (parent, spouse, etc.), such that they
are confident of help in times of need, this is a secure attachment. These bonds are characterized
by the almost exclusive use of the primary attachment strategy: proximity seeking. Getting
closer (physically and, for adults, psychologically) in times of need is the behavioral hallmark of
secure attachment. In babies, the biological starter materials for obtaining proximity are smiling
and vocalizations, which attract adults to bring them nearer. In adults, proximity seeking may
include a phone call, asking for a date, or words of affection.
Help from others cannot always be available. When the primary strategy, proximity
seeking, fails to bring aid, alternative behaviors are employed. These are called secondary
strategies. Secondary strategies come in two orientations: hyperactivation or deactivation of the
system. Each secondary strategy includes a unique pattern of motives, emotions, and behaviors.
Hyperactivation of the attachment system is the first line of defense against neglect or
potential rejection. Like the primary attachment strategy, the goal here is to obtain a feeling of
security through the attachment figure. The emotion and behaviors, however, are very different
from the primary strategy. If an infant’s crying goes unheeded he cries louder. This is an
energetic (often angry or frustrated) effort to obtain a response from an attachment figure. These
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 10
are called protest or reactive behaviors. The emotions that underlie them are fear and anger.
Adult protest behaviors also include crying and yelling, as well as more controlled but
nonetheless hostile, behaviors. This process of hyperactivation of the attachment system is
distracting and worrisome. If efforts to gain care continue to be fruitless, this over-investment of
energy and emotions is hardly worthwhile. It is at this point that the other secondary strategy
becomes better suited to the situation.
Deactivation of the attachment system is motivated by the need to diminish attachment
needs and hopes. Deactivating behaviors are strategies by which to create this distance. If a
child’s cries continue to go unheeded, they will eventually dissipate as he attempts to sooth
himself. He may fall asleep or allow himself to be distracted by toys. Similarly, adults will turn
to distractions. They will work to reduce dangerous desires such as emotional intimacy and
inter-dependence, and suppress thoughts about them. These may be positive experiences in
some ways, but they activate the attachment system, and so run counter to this strategy.
Once the attachment system is deactivated, other systems can motivate behavior and a
person can try to cope through other means. Non-attachment motivations from other systems,
like the Exploration Behavioral System or Power Behavioral System, may be indirectly
associated with deactivation of the attachment system. For instance, if the attachment system is
naturally deactivated (because the set-goal of felt security is attained) then goals from these other
systems would be free to activate. Such goals range from exploring the environment or engaging
in leadership activities (e.g. start a club). Because the attachment system has been deactivated
because it has fulfilled its function, these goals would be associated with secure attachment.
However, when the attachment system is insecurely deactivated (because the set-goal of felts
security is unobtainable) then goals from other systems become activated but with a different
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 11
flavor. Most prominent would be motivations that provide alternate avenues to feelings of safety
and security: like making money (Mikulincer & Shaver, 2008) or dominating others (Mikulincer
& Shaver, 2012). Deactivation is a process of self-preservation and self-regulation. With
deactivation of the attachment system comes a rather sad but calm state, and resources are freed
from useless worries and wishes. The tranquility of being disconnected may suggest a positive
or at least not-negative state. But remember, this strategy is the final response to repeated
failures to obtain caring. The feelings that accompany this are an anger that has turned cold, and
moved into sadness: a lonely state, unjust, and hopeless.
This process is interwoven throughout every life. A daughter hitches a ride on her
father’s leg as he readies for work, perhaps she yells an objection, when he removes her to leave.
And when she inevitably loses the battle: a little pout as she turns off her need to have him there,
in that moment, to go about her own day. A boy loses his first love: he calls, tries to see her and
regain her affection. He is afraid of losing her. Then he protests, fights her leaving: anger. And
finally, he learns to disengage: sadness and bitterness, a cold anger, about love. Maybe he
concentrates on his studies and pretends not to notice the glances of other girls, future
heartbreakers, who look his way. But at the end of the day the daughter will jump into her
father’s arms, and the boy, he will eventually love again.
Every person has in their repertoire, and must at some time in their lives, employ the
secondary strategies of hyper- and deactivation of the attachment system: sometimes all with the
same person as in the examples above, and sometimes with different people. It is when these
strategies stop being responses to actual events, and start being the default response to most
events, that they become chronic traits or attachment orientations. Secure attachment is
characterized by chronic positive expectations which lead to the optimistic construal of events:
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 12
an essential belief that one will be cared for in times of need and dismissal of possible evidence
to the contrary. Proximity seeking is the default behavior in times of need, and assured needs
will be met, happiness is the chronic emotion. It is nice to be secure.
Insecure attachment comes in two forms that correspond to the secondary strategies of
hyper- and deactivation. Anxious attachment is a chronic hyper-activation of the attachment
system. It is to be plagued, preoccupied by worries about relationships. Attachment anxiety is
characterized by the hyper-motivation to obtain reassurances and avoid rejections. The
accompanying emotions are fear and bursts of anger. The behavioral responses to these feelings
are excessive proximity seeking, or clinginess, and protests to construed neglect or rejections.
Conversely, attachment avoidance is chronic deactivation of the attachment system. It is
characterized by attempts to shut down needing and wanting: motivations for intimacy and
dependency. These needs elicit sadness and anger that encourage behaviors that distance one
from others and engages one in activities that are either distracting or which provide alternative
sources of security and self-value.
Just what is a trait, exactly?
Chronic attachment orientations are considered dispositional characteristics. Recent
theories posit chronic goal states and beliefs as the basis of personality variables such as
extraversion or agreeableness (Read et al., 2010; Read & Miller, 2002). Many of the words that
people use to describe personality are emotions words (e.g. “cheerful”). So, in layman’s terms,
traits are often thought of as biases for a particular kind of emotional reaction across diverse
situations – in other words, chronically activated emotions (Plutchik, 1980; Smith & Lazarus,
1990).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 13
The present work considers the dynamics by which repeated experiences “build up”
schemata (sets of construals or “working models”), goals, emotions, and behaviors. Essentially,
it models the formation of personality (Talevich, 2012). These dynamics are captured in
previous work: a neural network model that will be described below. This current work, with
human subjects, seeks to confirm these dynamics. The learned links between certain events and
how they are construed, between some construals and certain goals, and so on in the different
patterns of goals, emotions, and behaviors, that is the composition we call personality. And so, it
is expected that the traits modeled herein, anxious and avoidant attachment, will influence,
sometimes directly and sometimes indirectly, all variables from construals to behaviors.
The Intervening Role of Construals
Personality, as chronically activated motivation, contributes to the perception of an
encounter by biasing what is noticed and what is ignored (Smith & Lazarus, 1990) and what is
liked and what is disliked (Ferguson & Bargh, 2004) about objective conditions. In other words,
people do not experience an event directly but experience the construal of an event. It is this
construal that mediates between the situation and the motives that are activated.
In the attachment literature, several construals are deemed necessary for the goal-
correction of behavior: the degree of threat, the attachment figure’s past responsiveness, and
one’s own inner state. (Mikulincer & Shaver, 2003).
The neural network model described below begins with construals as inputs. With
human data, Study 3 of this work will examine how the situation and traits effect construals and
are then mediated by construals to motivation, emotion, and behavior.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 14
The Intervening Role of Motivation
Goal pursuit is activated by the environment or situation to guide our actions throughout
the day with a constancy and continuity that is beyond conscious awareness. This allows us to
adapt to what is called for by circumstances (Bargh, 1990; Bargh, Gollwitzer, Lee-Chai,
Barndollar, & Trotschel, 2001). Most of this goes on without conscious deliberation of every
action. The activated goals induce a unique state in which the goal remains activated despite
interruption or the passage of time (Ovsiankina, 1928). Thus, when a behavioral sequence fails
to satisfy a goal, such as obtaining a feeling of security, the magnetic power of goal pursuit pulls
the person to re-engage in the face of temporary failure, to re-appraise, and to subsequently alter
behavior.
The biological function of the attachment system is to procure safety and security through
the most powerful source of protection and abundance available: other humans. “Felt security”
is the “set-goal” or over-arching motivation in the service of this biological function. Many
other goals are sub-goals (or means) to this higher-order goal. In other words, felt security is a
highly abstract goal category that would, logically, require the means of more concrete goals
such as gaining (or avoiding) emotional intimacy and avoiding rejection (Talevich, Read, Walsh,
Iyer, & Chopra, 2012).
Bowlby pinpointed the magnetic power of goal pursuit (pursuit of felt security
specifically), and the consequences of repeated failure to achieve it, as the source of insecure
attachment (Bowlby, 1969; Mikulincer & Shaver, 2007). Attachment anxiety is characterized
by near-constant engagement and re-engagement with the goal for felt security. As discussed
previously, this is what characterizes hyperactivation of the attachment system (Cassidy &
Kobak, 1988; Main, 1990; Mikulincer & Shaver, 2003)). Avoidant attachment, on the other
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 15
hand, is characterized by disengagement from intimacy and dependency goals (deactivation of
the attachment system). This failure to re-engage is so persistent, in fact, that goals are
abandoned even while they remain obtainable (Mikulincer & Shaver, 2007).
Lewin (1939) envisioned life as composed of forces, some of which were helping forces,
driving movement towards a goal, while others were hindering forces, driving movement away
from a goal. Similarly, Beach and Mitchel (1987) discuss “fit” of current self-trajectories in the
decision to adopt or decline a new trajectory. And finally, Smith and Lazarus (1990) consider
two primary appraisals to be responsible for the evocation of emotion: the appraised motivational
relevance and the appraised motivational congruence of an event. Working models of
attachment include information about the routes and barriers to goal attainment (Mikulincer &
Shaver, 2007).
In line with all these historical theories, my colleagues and I have developed a simple
calculation. Rather than asking people to rate the value of something and then estimate the
likelihood they will get it, we ask a slightly more specific question. We ask participants for the
importance of their goals (value) and then ask them to rate how much the context under study
helps or hinders the accomplishment of each goal (congruence). We call this Goal Impact
because it measures the impact that the situation has upon one’s most important goals. Note that
because this is a multiplicative term, congruence has no effect if a goal is not important
(chronically or situationally activated).
We have found that Goal Impact will predict up to 40% of the variance in decision-
making situations such as whether or not to leave one’s current job, maintain a healthy weight, or
comply with psychotropic medication (Talevich, Saks, & Read, 2012; Talevich, Walsh, & Read,
2012; Walsh et al., 2012). However, which goal is relevant in any given situation may not be
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 16
obvious or straightforward. Goals may be active because they are motivated by the current
situation, by a previous situation and as yet unsatisfied, or chronically activated by one’s traits.
For this reason, we find that it is important to measure an array of life goals (Chulef, Read, &
Walsh, 2001; Talevich, Read, et al., 2012) – not just those that seem relevant to the foci of study.
This method of measuring Goal Impact clarifies what is meaningful, how (in what direction), and
why.
Once the motivational impact of the situation is apparent, when one senses that they are
or aren’t going to get what they need, this naturally, will be an emotion-evoking piece of
information. Thus, motivation is the evoking precursor of emotion (H
e
).
The Intervening Role of Emotion
Positive and negative emotions reflect the dynamics of approach and avoidance
motivations. What is aversive conjures negative emotions and what is appetitive often conjures
positive emotions in the service of their attainment (Cacioppo, Gardner, & Berntson, 1999). As
liking is for doing (Ferguson & Bargh, 2004), these emotions serve a purpose: to induce
behaviors that will bring the goal closer (increase engagement) in the case of positive emotions.
Anger, though a negative emotion, serves to move one toward something to engage it. But most
negative emotions, such as sadness or disgust, serve to repel or decrease engagement.
Leading theorists view emotions as a solution to a fundamental adaptive problem
(LeDoux, 2003; Scherer, Dan, & Flykt, 2006; Smith & Lazarus, 1990). An organism needs to
respond to the environment quickly and efficiently with the correct, most adaptive, behavior
possible. For simple organisms, innate reflexes are the simplest solution for mobilizing the right
behavior for the circumstances. But reflexes are inadequate to deal with the complexities of
human life. A more flexible mobilization signal is required, and so, reflexes evolved into
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 17
emotion patterns. Smith and Lazarus (1990) implicate two primary appraisals, that of goal
importance and goal congruence, that assess the relationship of the person to the environment
and thereby select the appropriate emotion to signal. The thwarting or facilitation of (important)
goals evokes distinct emotional reactions - each of which has a proposed adaptive function. The
function of anger is to mobilize to defend against, or offensively prevent, harm or loss.
Fear/anxiety is for the purpose of avoiding potential harm, and one feels sadness to help
disengage from a loss. As goal importance and congruence are the factors of Goal Impact, the
neural network model and hypotheses herein, predict that emotion mediates between motivation
and behavior.
Bowlby’s theory is that of a goal-corrected system in which the behavioral strategy
selected is that which will most adaptively move one down the routes and around barriers to the
attainment of felt security (Mikulincer & Shaver, 2003). This is consistent with a motivational
appraisal process. However, the literature on working models suggests motivational implications
can become a part of these knowledge structures (Collins & Allard, 2004). If the motivational
implications of a situation are part of a knowledge structure, such as a schema or working model,
then one may not require a new motivational appraisal to select the appropriate emotional
response. The next question, then, is how emotions are elicited by a situation when motivational
appraisal processing is not undertaken.
Emotional responses can become associated with a specific stimuli. It is possible that
some stimuli have a more direct circuit for emotion elicitation. This circuit is proposed to be
innate and hard-wired. Fear is the most often considered but not the sole emotion proposed to
work in this manner (LeDoux, 2003). Even if the circuit is hardwired as proposed, except for the
association between fear and loud noises, the association of fear with a particular stimuli would
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 18
still require learning. But it may also be that the association is learned more quickly due to this
direct route.
Another conceptualization of emotion processing comes from Scherer’s Component
Process Model (2006). This model includes sequential stimulus checks that are based on
perceptual aspects of a stimulus. Appraisals of sensations, such as novelty or mere pleasantness
(or unpleasantness), as well as motivational appraisals imply relevance (importance). Relevance
appraisals are combined with Implication Appraisals (e.g. goal congruence, urgency). Together,
importance and implications evoke emotions but these may or may not be related to one’s goals.
The emotion-evoking stimuli most prominent in attachment theory are threat and
attachment figure responsiveness. These two aspects of a situation activate and deactivate the
attachment system, respectively. Threat, which activates the attachment system, evokes fear.
As discussed above, fear may be evoked through a hardwired direct route, through appraisals of
motivational importance and congruence, as well as through appraisals of unpleasantness and
urgency.
On the other hand, a caring response from an attachment figure, at a moment of need can
down-regulate fear and thereby provide a soothing comfort. The MABS model was trained to
associate responsiveness with happiness. Without training, the neural network dynamically
learned that happiness inhibits fear (they are negatively associated in simulation results). Thus
responsiveness, which deactivates the attachment system, evokes happiness, and happiness
reduces fear. Happiness may be evoked through a hardwired direct route, through appraisals of
motivational importance and congruence, as well as through appraisals of pleasantness and the
beneficial things this pleasantness implies.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 19
Emotions are functional in nature and their function is the mobilization of behavior. As
discussed earlier, three emotions are particularly important in the mobilization of attachment
behavior. Although fear generally mobilizes an away-from-danger flight response, that flight is
generally in the direction of an attachment figure for protection. This is proximity seeking.
Anger mobilizes protest against possible rejection or abandonment to “fight for” what one is
owed: needed attention from the attachment figure. Sadness is associated with disengagement
from an attachment figure who has been repeatedly unsupportive.
Emotions, as they become chronic associations, are features in the development of a trait
attachment orientation in the MABS model. The progression from fear, through anger, to
sadness signals the progression from security, through anxious insecurity (marked by chronic
fear and anger), into avoidant insecurity (marked by chronic anger and sadness).
The Computational MABS Model
The studies presented herein are based upon human data and can be read and understood
as standard empirical work. But the theoretical underpinning of the experimental and survey
designs were pre-tested as a neural network model of infant attachment behavior. The purpose
of creating the neural network model was to use it as a theory-building tool. A neural network
model requires specifications at an exacting level not required by verbal models. And it must
actually “work” – that is it must learn what it is supposed to learn and produce the theorized
output in simulations. Thus, the theory is rigorously pre-tested in simulated trials before it is
tested on humans. The ability to specify complex structural equation models of human data for
the following studies is due to the fact that these relationships had already been specified tested,
adjusted, and retested using this tool.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 20
Bowlby’s theory of attachment was based on observations made during his work with
orphans: children. This is why attachment theory spent its own infancy under the auspices of
developmental psychology before growing into, first, adult close relationships and, more
recently, into a variety of domains from religion to business. The current studies are designed in
accordance with the neural network of infant attachment but the subjects are all adults. This
assumes that the process by which one bonds to another and the impact of that bond are the same
from infancy through adulthood. It also assumes that the traditional categories of attachment
behaviors (proximity seeking, protest, and dismissal) are enacted by adults as much as children –
although the specific behaviors may be more sophisticated in adults. Essentially, although
construals, motivational contents, and behavioral strategies will be markedly different in adults
and infants, the basic mechanisms and systems remain the same.
A neural network can be viewed as a scaled-down model of a real brain (O’Reilly &
Munakata, 2000). The advantages of using a “scaled-down brain” to model psychological
relationships are many (Read & Monroe, 2009) Like the brain, it learns (Hebbian and error-
corrected human-like learning algorithms) over time. As an on-line computational model, a
neural network has the ability to capture dynamic relationships that verbal models, flowcharts,
and static statistical models can represent in outline but not replicate (Read & Monroe, 2009).
What is most exciting about a neural network model, for a psychologist, is that is allows
you to build virtual subjects by varying the experiences presented to the network for learning.
Essentially, you can manufacture your own participants, specifying precisely their neural
architecture and experiences, and then test them. Theoretical relationships and mediations can
be pre-tested before human data collection. Of course, eventually, human data is necessary to
validate the model and that is the purpose of the studies herein. But, even as a thought
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 21
experiment, a neural network model requires a scientist to think about her subject at an exacting
level of specificity. Elements, parameters of elements, and the relationships between elements
must be not only envisioned, but built. And once built, it will be rebuilt, for every network fails
many times over before it passes the test of biological plausibility: a solution possible within the
constraints of what we currently know about how the human brain learns and builds associations.
Shortly, I will walk through the MABS model. First, it should be understood that
a neural network model, like the brain, is hierarchical. Elements, such as emotions, are
represented as layers. The idea of layers is derived from the fact that the brain has multiple,
hierarchical, layers that feed connections both forward to higher layers as well as back down to
lower layers. Although MABS is a theoretical model abstracted from specific brain regions it
preserves this concept. Thus, there are layers, e.g. the emotions layer, composed of a group of
“units” that represent discreet emotions like fear, sadness, etc. Units are themselves considered
to represent a group of neurons with a common function. For instance, the common function of
the “fear” group of neurons may be to detect (by learned associations) danger to one’s self or
one’s goals and activate the fear unit.
The computational version of the Motivated Affective Behavioral System
(MABS) models the dynamical patterns among knowledge structures (schemas), a motivation
sub-system, emotions and behaviors. These patterns allow for the build-up of chronic activations
that are the very formation of an attachment “style.” The same process is proposed to underlie
the formation of other traits, as well.
Construals. The first layers in the neural network are inputs to the system that represent
event construals. There are six construal inputs layers, specifically, the layers (with units) are 1)
Threat cue (present ,not), the 2) Attachment Figure (AF,Other), 3) AF Status (attentive,
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 22
inattentive, absent, dangerous), 4) AF Past Response (responsive, unresponsive, harmful), 5)
one’s own Past Behavior and 6) one’s own Past Inner State. The first four are construals of the
current situation. The last two are exact duplicates of the Behaviors layer and Current Inner
State Layer, both of which are higher up in the network and will be introduced soon. For now, it
is important to understand that these duplicates are unique layers called “context” or “recursive”
layers because they provide the context of “what happened last time”. These two inputs are both
temporal construals of the self: what one did last and the motivations and emotions that led up to
it. This is critical to targeted learning. That is, in order to change behavior to target a better
outcome next time, there must be a record for reference (or rather, for context, as the layer is
called). This is the basis of goal-corrected behavior.
All these construals, the perceived details of an event, come together in the next layer of the
neural network to form the situation as a whole. The Situation Compute is a hidden layer that
computationally transforms its inputs into a unique pattern that represents them as a whole.
When these patterns, associations between construals, become chronic, the Situation Compute
becomes analogous to a schema or working model. It is a script of events, people, and one’s self
among them. connections and layers of the connectionist computer model. Lines without arrows
indicate bidirectional connectivity.
The Motivation Sub-System. The situation, as it is perceived, is projected from the
Situation Compute throughout the sub-goal system. This is the first call that will eventually
select the action required by the environment. If the Attachment Behavioral System is “on”
then goals that foster bonding and eliciting care from others are most likely to be activated. In
this vein, the goals included in the MABs neural network are those to avoid rejection, avoid
harm, and get help. Two additional goals were selected to represent those that may be activated
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 23
when the attachment system is “off”. Exploration is associated with secure attachment when
there is no threat in the environment. This would be motivated by the goal to explore which,
therefore, is include in the MABs model. However, one may note that these goals can be
grouped according to their approach or avoidance orientation: wanting to avoid rejection or
wanting to approach a person (to get help) or to approach something new and interesting (to
explore). In keeping with this balance of approach and avoidance (to be discussed below), the
system needed an avoidance goal that could be active when the attachment system was “off.” I
chose the goal to avoid effort. This goal is not discussed, to my knowledge, in the attachment
literature but satisfies the requirement of being an avoidance goal unrelated to harm (potential
harm would turn “on” the attachment system). It also fits logically. As discussed in the opening,
those with attachment avoidance must cope using their own resources – so it would be
reasonable to assume they have need to conserve those resources. This may also be true of
those with anxious attachment who are expending so much energy on relationship-specific goals.
If avoiding effort is less important to secures that could contribute to the observation that while
avoidant infants often explore their environment, they do so to a lesser extent and with less
enthusiasm, than secure children (Ainsworth, Blehar, Waters, & Wall, 1978).
The computational MABS model represents these approach and avoidance goals in
separate layers. The approach system governs response to rewarding stimuli. In the literature it
is variously referred to as behavioral facilitation (Depue & Collins, 1999) or the Behavioral
Approach System, aka BAS (Gray, 1991, 2003). The avoidance system, also referred to as the
Behavioral Inhibition System, aka BIS (Gray, 1991, 2003) governs response to punishment and
aversive stimuli. Social goals can be categorized along two dimensions: goals to gain social
rewards (implicating the approach system) and goals to avoid social punishments (implicating
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 24
the avoidance system) (Gable, 2006; Gable & Strachman, 2008). Attachment-relevant social
rewards are attentiveness and responsiveness from the attachment figure. Social punishments
typically include rejection and unresponsiveness. Gable (2006) has found that the approach and
avoidance systems operate via different processes. The approach system is associated with
social exploration. This push to “get out there” and socialize is an exposure process by which
good experiences reinforce the social behaviors that produced them. The avoidance system,
however, operates via a reactivity process that seems to suggest a differential sensitivity to
avoidance goals. This is particularly relevant to insecure attachment as those who are anxious
are particularly sensitive to rejection cues and the avoidance of intimacy is the hallmark of
avoidant attachment.
Each goal layer projects its activations up to the congruence layers. There is a unique
congruence layer for every individual goal. The Situation Compute also projects directly to each
congruence layer. And so, each congruence layer activates “yes” if both its corresponding goal
is activated and the Situation Compute indicates a pattern the layer has learned is favorable to its
goal. On the other hand, if the Situation Compute projects a pattern unfavorable to the goal, its
congruence layer will activate “no”.
Finally, the activations from the avoidance and approach goal layers combine with the
activations from the congruence layers as they converge on the Goal Impact Compute. Like the
Situation Compute, this is a hidden layer that computes a unique pattern, this time to represent
the state of the motivation sub-system.
Emotions. In accordance with appraisal models of emotion, the Goal Impact compute
then projects this pattern up to the Emotion layer to signal what behavior is needed. There are
also direct connections from Threat Cue and from Attachment Figure Responsiveness to the
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 25
emotion layer. This is in keeping with other models of emotions in which stimuli can more
directly affect emotions (LeDoux, 2003) or be elicited by non-motivation appraisals, e.g.
pleasantness (Scherer et al., 2006),
The four most relevant emotions to attachment theory are represented in the MABS
neural network: fear, anger, sadness and happiness. From these are signaled the behaviors of
proximity seeking (from fear), protest (from fear and anger), dismissing (from anger and
sadness), or exploration (from happiness).
Overview of Studies
See Table 1 for an overview of studies and hypotheses. Rather than by number, hypotheses are
designated by the first letter of the intervening element to which they refer – in the hopes of
providing the reader with a mnemonic. The Motivated Affective Behavior System (MABS) is an
intervening process model. The drawing of the computational model (Figure 1) is the more
comprehensive and accurate depiction. However, it is beyond the capacity of statistical models
to represent the full dynamics of the computational model. Therefore, a scaled-down version of
the computational model (Figure 2) will be tested. In this version, construals (H
c
), a motivation
sub-system (H
m
), and emotion (H
e
) all sequentially intervene to produce the behavior called for
by the environment. The objective situation is proposed to predict construals in conjunction with
traits. Traits (H
t
) are hypothesized to be an emergent property of MABS and, therefore, expected
to predict any element in this intervening process be it construals, motivation, emotion, or
behavior. With regard to the motivation sub-system, traits may influence the importance of
individual goals, or the perceived congruence between goals and any particular situation at hand.
I will be examining this model within the framework of attachment. Attachment is a life-
long process from early bonds to parents to pair-bonds between adults. These studies will focus
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 26
on pair-bonds, close relationships between adults, because only adults can articulate the inner-
processes involved in this model. Furthermore, while infant attachment is a critical building
block, most of a human being’s life is spent (and most of their relationships occur) in adulthood.
Study 1 will examine the motivation sub-system (represented by Goal Impact) and how it
mediates between trait attachment and behavioral intentions to seek a (or another) relationship.
Study 2 two will manipulate attachment to predict specific motives, emotions, and traditional
categories of attachment behavior. Study 3 improves on the methodology of Study 2 to predict
construals of an attachment situation, Goal Impact, emotions, and communication behavior
(specifically, the contents of a letter). Study 4 tests the generalizability of the MABS model to
see how well it can predict who pays on a date. Since who pays on a date involves social norms,
this tests the MABS model in the context of social roles to predict norm compliance and
violation.
The neural network model will continue to serve as an organizer of theory in this work.
Each study will be introduced with the model elements it seeks to validate.
Mediation in the New Millennium
This model is unusual in that it is an intervening process model predicting one indirect
effect after another. The majority of effects should be fully mediated from one element (in
neural network terms, “layer”) through another in the chain (e.g. Construals to Emotions to
Motivations). Each element in the model will be composed of multiple variables (e.g. Emotions
include fear, anger, etc). The literature has begun to re-define mediation. Mediation, as defined
by Baron and Kenny (1986), starts when there is a direct relationship between variables X and Y
(See Figure 3).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 27
If there is also a direct relationship between X and M, as well as between M and Y, then
M should be investigated as a mediator of the relationship between X and Y. In a regression
equation with X and M as predictors of Y, full mediation by M is said to occur if the coefficient
for X is no longer a significant predictor. In partial mediation, it is also acceptable that the
relationship between X and Y remain significant but is less so.
This process, as delineated by Baron and Kenny (1986) almost thirty years ago, is a
piecemeal process in which M is only considered to mediate the relationship between X and Y
in the case of an initial direct effect between X and Y. However, updated methodology states
that it is possible to have an indirect effect carried through M despite there being no significant
direct relationship between X and Y (Hayes, 2009; Mathieu & Taylor, 2006). The relationship
between X and Y may not be statistically detectable in one particular model though it is in
another. The total effect of X as a predictor is the sum of many different paths of influence and a
single model may not capture every one. There can be multiple indirect, but opposite signed,
effects in operation simultaneously that suppress one another in the calculation of total effects.
Furthermore, one indirect effect may be represented in the model while the opposite signed
effect is not.
Since the MABS model is a mediational model, the focus is not on direct or total effects
but the indirect effect from one variable to another – the latter of which may be several steps
away in the chain. It is only feasible to model a small representation of the variables, and
effects, in a single structural equation model. Of all the emotions, goals, and behavioral actions
of which we humans are capable (and of which are affecting one another), this model represents
at most four in each category (which still reaches a pretty high level of complexity).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 28
Methods for the Main Analyses.
Analyses will be conducted in Amos using a bootstrap to estimate direct and indirect
effects in the proposed model, as well as providing an overall model fit (Bollen-Stein Bootstrap).
Bootstrapping is considered by Hayes (2009) to be the method of choice in mediational analyses
for its power, control of Type 1 errors, and freedom from assumptions of normality. Freedom
from the frequently violated assumption of normality makes this method superior to the many
other common statistical tests of mediation that are not (e.g, Sobel’s test; Sobel, 1982). For
example, in Study 1 bootstrapping works by generating 2000 samples by resampling the original
N of 182. Each “new” case is randomly drawn and copied to create a new sample of 182. This
is repeated until there are 2000 bootstrapped samples of 182 cases each. The reason the original
cases are copied is so that when any one case is drawn it can be immediately drawn again
(Preacher & Hayes, 2004, p. 722). Effects are computed for each sample, which yields a bias-
corrected confidence interval (Hayes, 2009).
Structural equation modeling (SEM) will be employed to test this multi-step mediator
model, with its many proposed direct and indirect effects, as a whole. While it may not yet be
typical to test numerous indirect steps, Hayes observes that research testing intervening variables
models is frequently out of step with advances made in the statistical methods literature. The
piecemeal employment of multiple independent models to test components of a larger theoretical
model is a limited approach, and structural equation modeling presents a touted alternative
(Brown, 1997; Hayes, 2009).
An SEM model that fits the data well will have a nonsignificant chi-square statistic. This
indicates the proposed model does not significantly differ from the relationships found in the
actual human data. Thus, it fits the data. However, since the chi-square statistic assumes
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 29
normality we will also report the Bollen-Stine bootstrap, which does not (Bollen & Stine, 1992;
Hancock & Mueller, 2006). The Bollen-Stine bootstrap has also been shown capable of
successfully evaluating models with nonnormal distributions that arise from small samples sizes,
e.g. 101 participants (Ievers-Landis, Burant, & Hazen, 2011).
Standardized path coefficients (b) are reported in the model diagrams. These are
analogous to the regression coefficients readers are accustomed to seeing in path diagrams. In
figures, standardized path coefficients (b) are reported for easy comparison of effects across
different measures. In tables, unstandardized path coefficients (B) are also reported. Two of the
model variables are dichotomous, and in such cases, standardized indirect effects are less
meaningful because standardization destroys the interpretation of the indirect effect as the mean
difference between groups on the outcome attributable to the pathway through M (Hayes, 2009).
Study 1: Who Has a Roving Eye
The purpose of Study 1 is to validate the motivation sub-system of the MABS model (See
Figure 4) and its relationship to traits and behavior. The motivation sub-system includes the
elements of goal importance, goal impact, and their product Goal Impact. MABS predicts that
motivation will mediate between traits and behavior: specifically that Goal Impact will mediate
between chronic attachment orientations and relationship decision-making behaviors (Hm). The
behaviors under study include whether or not to stay in or leave a relationship, whether or not to
look for a relationship if one is single or look for a new partner if one already has one. Chronic
attachment orientation is an ideal personality variable by which to examine the relationship
between motivation and behavior because the behaviors associated with avoidance and anxiety
are well defined and markedly different. Attachment anxiety is characterized by a
preoccupation with relationships and worry about being abandoned. This results in notably
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 30
“clingy” behavior. Such individuals, it stands, would be more likely to look for a relationship if
single and less likely to look outside a current relationship for another partner. Attachment
avoidance, on the other hand, is characterized by distancing and detachment. These individuals
would be less likely to want a relationship if single and less protective of a current relationship –
ergo more likely to be tempted by other partners if currently in a relationship. Thus, the Hm
prediction for Study 1 is that attachment avoidance will predict a willingness to leave a
relationship, or at least, keep an eye out for other relationship opportunities. Anxious
attachment, on the other hand, will predict an unwillingness to leave a relationship and a
disinterest in other potential mates –as this could pose a danger to the current relationship.
The impact that leaving the relationship, or seeking a new one, has upon one’s most
important goals (Goal Impact) is expected to mediate between both attachment orientations and
these behaviors. Because anxiously attached individuals are pre-occupied with their
relationships, relationship goals are likely to be ranked among the highest of their life goals.
Conversely, those with avoidant attachment may not consider relationship goals to be among the
most important of their life goals. Or perhaps, relationship goals may still be important to them,
but would have considerable competition from non-relationship goals. For instance, spending
time with their partner may be important but weighted rather evenly with putting in enough
hours at the office to achieve a promotion. Thus, for one who is anxiously attached, leaving
their relationship or considering other options, would pose a threat to their most important goals.
Whereas, for the avoidant, it could as easily be the relationship that poses a threat to other goals,
such as that promotion, and leaving a relationship or at least considering other options, could be
advantageous to goal achievement.
The hypotheses of Study 1 are:
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 31
H
t
: Attachment orientations will directly and indirectly influence intentions to
seek a new relationship, if one is single, or a new partner if one is not.
H
m
: Goal Impact will partially mediate attachment orientation in the prediction of
intentions to seek a relationship, if one is single, or a new partner if one is
currently in a relationship.
Method
Measures.
Behavioral Intentions. Measured twice, once before the goal measures and once after.
Participants were asked to respond to six questions designed to measure the judged likelihood
that they would leave their relationship.
Intention to Leave. Ss were first asked to indicate the extent to which they agreed or
disagreed (7-point scale) with the following three statements concerning their relationship: “I
like it.”, “I want to leave it.”, and “I intend to leave it.” They then responded to three questions
about the likelihood they would leave their relationship over three different time frames: within
the next 6 months, within the next 12 months, and within the next 2 years. Participants
responded by using a sliding scale ranging from 0 to 100. Only those currently in a relationship
(“Coupled” individuals) were asked this set of questions.
Intention to Seek. In order to have a behavioral variable that could be asked of both
singles and couples, I asked about each Ss intentions to seek “a” relationship (if single) or
“another” relationship (if coupled). The first three ratings were “I would like to be in a/another
relationship”, “I want to be in a/another relationship”, and “I intend to seek a/another
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 32
relationship.” Then they were asked to indicate the likelihood (0%-100%) they would actually
seek a/another relationship in 6 months, 12 months, or 2 years.
Original Goal Taxonomy Q-Sort. Study 1 utilizes Chulef, Read, & Walsh’s (2001)
hierarchical taxonomy of 135 human goals, which provides a broad framework for the study and
assessment of human motives and their role in social behavior. This taxonomy starts with 135
individual goals and then organizes them into ever-broader categories.
This taxonomy provides a broad framework for the study of human motives, for mapping
subjective values in social and personality research, and for constructing decision prediction
measures such as will be used in this study.
Generally, if we ask people what is important among their life goals, they say everything
is highly important. Thus, we administered a Q-Sort procedure that forced Ss to sort their life
goals into a quasi-normal distribution. Several procedures have been tested by myself and
colleagues. This method is the first, the original method. The 29 goals for Study 1 were taken
from an earlier version of our Hierarchical Taxonomy of Human Goals (Chulef et al., 2001).
The procedure was conducted as follows: On the first page SS were presented a list of goals and
two boxes into which they were instructed to sort their two most important goals (box 1) and two
least important goals (box 2). These goals would be programmatically assigned importance
values 7 (box 1, most important) and 1 (box 2, least important). On the second page, the unsorted
goals were again presented, with two boxes into which Ss were instructed to sort their four most
important (box 1) and four least important (box 2) goals. Programmatically, these were
assigned importance values 2 and 6 respectively. Finally, on the third page, the unsorted goals
were again presented beside two boxes for sorting the five most important goals remaining (box
1) and five least important goals remaining (box 2). Programmatically, these were assigned
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 33
importance values 3 and 5 respectively. The remaining unsorted goals were assigned an
importance value of 4. The final quasi-normal distribution of importance values was: 1-(two
least important goals), 2-(four), 3-(five), 4-(seven), 5-(five), 6-(four), 7-(two most important
goals)
Goal Congruence. Next I asked about the goal congruence of being IN a relationship
and the congruence of being OUT of a relationship. “Please rate the extent to which being in
[a/your current] romantic relationship would make it harder or easier to pursue each of the listed
goals over the coming year.” (11 point scale Extremely Hard (-5) … Neither (0)…Extremely
Easy (+5)). Note this survey makes use of piped text such that “a” is piped to singles and “your
current” is piped to those who are in a relationship. Each of the 29 goals was presented for
rating. On the next page, this procedure was repeated except it asked for the congruence of
being out of a relationship.
Goal Impact. This was calculated by multiplying the importance of each goal by their
congruence score (which could range from -5 to +5) then taking the sum of all products.
Chronic Attachment. The ECR-R was administered, which is the most commonly used
attachment measure at this time. It has two 18 item scales, one for attachment anxiety, e.g. “I
worry a lot about my relationships” and one to assess avoidance, e.g. “I find it relatively easy to
get close to my partner” (reverse coded). (Fraley, Waller, & Brennan, 2000).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 34
Results
Participants. This online study was taken by 383 undergraduates, 28% of whom are
male, from the psychology subject pool at the University of Southern California. At the time,
55% of the females and 73% of the males, for a total of 58% of the total sample were single –
that is, not in a relationship with anyone.
Preliminary Analyses. The scree plot showed two distinct factors for the ECR-R. There
was some double-loading of items but none greater than .35 and all loaded most strongly on the
expected factor. All items were retained for the final composites of Avoidance ( α=.82, M=3.3,
SD=.83) and Anxiety ( α=.94, M=3.6, μ=, SD=1.19).
Goal Impact (GI) Composites. A total of six goal impact composites were created. The
first two comprise the overall impact of being in a relationship and being out of a relationship.
Then, four finer-grained composites were created in order to separately examine the facilitative
and inhibitive effects for each of these two situations. First, all 29 goal importance ratings were
multiplied by the goal congruence rating. The congruence ratings ranged from -5 (the situation
made achieving the goal extremely hard) to +5 (the situation made achieving the goal extremely
easy) with an intermediate zero to represent neither harder nor easier. This was done twice, once
for each of the two situations presented: 1) being in a relationship (GI IN) and 2) being out of a
relationship (GI OUT):
In order to compare the facilitating aspects of the situation to the inhibiting aspects of a
situation, I separated each goal impact into facilitative and inhibitive goal impacts. To do this, I
copied values 1 through the highest goal impact (5) to create 29 facilitative impacts. Then I
copied the lowest impact score (-5) up through -1 to create 29 inhibitive impacts. Then, the 29
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 35
facilitative goal impacts of the situation were summed to form an aggregate composite and the
29 inhibitive goal impacts of the situation were summed to form another. This procedure was
done twice, once for the impact of being IN a relationship and once for the impact of being OUT
of a relationship.
GI IN Harder = Goal Importance x Harder-Congruence of being IN a relationship
GI IN Easier = Goal Importance x Easier-Congruence of being IN a relationship
GI OUT Harder = Goal Importance x Harder-Congruence of being OUT of a relationship
GI OUT Easier = Goal Importance x Easier-Congruence of being OUT of a relationship
These four composites represent the overall assessment Ss make of whether being in or
out of a relationship makes it easier or harder for them to accomplish their most important life
goals. See Table 2 for means and standard deviations.
Intentions. Both before and after completing the goal measures, participants completed a
set of six items that measured their intentions to leave their relationship (if in one) and their
intentions to seek a new one (asked of all Ss). For each time, the 6 items of each intention
measure were z-scored and then summed to form a composite for each time. The 12 items (6
from Time1 & 6 from Time2) of the seeking measure were inter-correlated (.70-.96), as were the
leaving items inter-correlations (.39-.90). Thus the Time1 & Time 2 items were combined for
each intention measure. The Cronbach’s alpha for all leaving intentions was .848 and .979 for
seeking intentions. Gender failed to influence intentions to seek or leave a relationship.
Surprisingly, intentions to leave a relationship, if currently in one, were not significantly
predicted by either attachment orientation or Goal Impact and was thus dropped from the model.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 36
SEM Model 1. Analyses were conducted in Amos using a bootstrap sample of 2000 to
estimate direct and indirect effects in the proposed model (as described in the introduction under
Methods for the Main Analyses). A Bollen-Stein bootstrap was also conducted to obtain a
model-fit test that did not rely upon assumptions of normality. The model is pictured in Figure 5
with standardized regression coefficients (also in Table 3).
The superior over-all model included two Facilitative Goal Impacts: how goals were
made easier by being in a relationship and how they were made easier by of being out of a
relationship. An earlier model included all four impacts. However, the Inhibitive Impacts of
being in or out of a relationship were not directly related to seeking a/another relationship. The
earlier model also had excellent fit ( χ
2
= 8.4, df=9, p=.5) and indirect effects were significant but
the current model is the better representation of the elements at work. Furthermore, the indirect
effects of the current model are stronger for this simplicity. The current model (Figure 5)
indicates it is the ease of a situation that most matter in predicting relationship and extra-
relationship seeking
Direct Effects. The regression coefficients, or total effects, are depicted in Figure 5 and
Table 4.
Anxious singles seem to think that being in a relationship would make it easier to achieve
their most important goals. However, those who are actually in a relationship have the opposite
opinion: they find their relationships make goal achievement less easy. (This appears to be
distinct from its being harder, for which there are no effects, as discussed above). Avoidant
individuals, whether currently single or not, agree. They too consider goal achievement to be
less easy when in a relationship. The Facilitative Impact of being out of (the current)
relationship was significant only for those currently in one, if they were avoidant. There was no
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 37
such relationship for single Avoidants nor attachment anxiety in any relationship status. To all
insecurely attached participants in a relationship it seems that a different relationship would
make it easier to achieve their most important life-goals. To Insecures, the grass is greener, or at
least easier, on the other side.
Consider now the effects of Goal Impact on intentions to seek a (or another) relationship.
Being single increases the Facilitative Goal Impact of being in a relationship: for singles, the
better they imagine a relationship would be for their goals, the more likely they are to seek one.
The logical converse is that for those in a relationship, the better that relationship is for goal
achievement, the less likely one should be to seek another – or so is the trend, though it did not
reach significance.
There is a positive relationship for coupled individuals between seeking intentions and
the Facilitative Impact of being out of a relationship. This indicates those currently in a
relationship intended to seek a different one if they expect it to make achieving their most
important goals easier. However, for those who are actually single, the perception that their
goals are better served by staying single does not much factor into whether or not they intend to
look for a relationship.
Indirect Effects. In confirmation of hypothesis H
m
, the relationship between traits and
behavior is mediated by motivation. Specifically, Facilitative Goal Impact partially mediates
between chronic attachment and intentions to seek a/nother relationship (Table 5).
In Figure 5, the Facilitative Goal Impact of being “out” of a current relationship
positively mediates between attachment avoidance and seeking. That is, Avoidants in a
relationship seek other partners because, in part, they imagine it would be easier to achieve
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 38
important goals with a new partner. Single Avoidants are not under this impression: these paths
are not significant for them. Nor are they significant for attachment anxiety.
For single Avoidants, their reduced intentions to seek a relationship are mediated by their
perception that life goals would be less easy to achieve if they were in a relationship. The
Facilitative Impact of being in a relationship negatively mediates between avoidance and
intentions to seek a relationship).
Anxious singles, on the other hand, imagine their life goals would be easier if they were
in a relationship and this increases their intentions to seek one: the Facilitative Impact of being in
a relationship positively mediates between attachment anxiety and intentions to seek a
relationship. Anxious individuals currently in a relationship, however, do not find it to be
facilitative of their life goals. As the Facilitative Impact of their relationship decreases, their
intentions to seek a new one increase.
Discussion
This first study has confirmed an important lynchpin in the MABS model: the
mediational role of the motivational subsystem (composed of goal importance, goal congruence,
and their product Goal Impact) in predicting behavior from one’s traits (H
m
). This is a partial
mediation where the direct relationship between traits and behavior remains significant, which
supports the concept of traits as chronic associations throughout the MABS process model (H
t
).
Study 2: Experiment 1
The purpose of Study 2 is to experimentally replicate the findings of Study 1 – that
motivation mediates between traits and behaviors (H
m
). It also seeks to expand the findings of
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 39
Study 1 to include emotions that are proposed to mediate between motivations and behavior
(H
e
). The MABS elements currently under consideration are pictured in Figure 6.
Study 1 relied on correlations between chronic attachment and the variables of interest.
An experimental approach would make a better case for the causal effects of traits on this
intervening process. To implement such an approach, I will prime attachment orientation. There
is much precedence for this (Bartz & Lydon, 2004; Mikulincer & Shaver, 2001). For example,
Mikulincer and Shaver (2001) primed secure attachment in an attempt to mitigate intergroup bias
and found that the prime reduced negative reactions toward out-group members. The current
study also seeks to expand the findings of Study 1 by showing that emotions also partially
mediate between traits and behavior by intervening between motives and behaviors (H
e
).
The hypotheses addressed in Study 2 are as follows:
H
t
: Traits will directly and indirectly influence behaviors. Traits will also directly
influence specific goals and/or emotions that, thereafter, partially mediate between traits
and behaviors.
H
m
: Primed and chronic attachment orientations, through specific motives, will indirectly
predict attachment behavior: proximity seeking, protest, and dismissing.
H
e
: Emotions (happiness, fear, anger, and sadness) will intervene between specific
motives and attachment behaviors.
Method
First will be administered the ECR-R to capture participants chronic attachment
orientations. Next is the attachment prime and situation capture procedure. The remaining
measures are administered after this procedure in the order listed.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 40
Attachment Prime & Situation Capture Procedure. The priming of attachment is a
two part procedure. I used the procedure from Bartz and Lydon (2004) with minor tweaks.
Their procedure has two parts 1) a description of an attachment relationship and 2) a
visualization of one such relationship followed by a short-answer question to cement the visual.
As did Bartz and Lydon, I presented one of four primes, randomly selected for presentation, to
the participant (attachment anxiety, attachment avoidance, attachment security, no-prime). I did
not use Bartz and Lydon’s fourth prime, Fearful Attachment, which describes an anxious-
avoidant amalgamation. Instead, as my fourth condition, I created a description that does not
prime attachment but instead asks Ss to think of an interaction that they would consider “typical”
of their relationship. This will allow me to compare each prime to the control condition. As the
control condition should be a reflection of participant’s own attachment style, this contrast will
allow me to examine each prime condition in comparison to a no-prime condition based upon Ss
own dispositional attachment. In the second step, Bartz & Lydon asked Ss to write a sentence or
two about their thoughts and feelings regarding themselves in relation to their chosen person.
My instructions were slightly more specific. I asked Ss to describe a specific event that occurred
with their chosen person, as described below.
Part I: Attachment Description Primes. For part one, participants were given a
description of one out of three Hazan and Shaver (1987) attachment relationship descriptions or
else they received the control prime. The instructions were modified to refer to a specific person
with whom the Ss had a meaningful relationship (Bartz & Lydon, 2004):
“Please think about an important and meaningful relationship you have had...”
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 41
“…in which you have found it easy to be emotionally close to this person. In this
relationship, you felt comfortable depending on this person and having them depend on you. You
didn't particularly worry about being alone. You felt accepted by this person.” (Secure Prime)
“…in which you have felt like you wanted to be completely emotionally intimate with this
person but felt that this person was reluctant to get as emotionally close as you would have liked.
In this relationship you felt uncomfortable being alone and worried that this person didn’t value
you as much as you valued them.” (Anxious Prime)
“…in which you felt comfortable not being emotionally close to this person. In this
relationship you felt that it was very important to be independent and self-sufficient and you
preferred not to depend on this person or have this person depend on you.” (Avoidant Prime)
“you experienced the amount of independence, emotional closeness, and support that is
typical of the majority of your relationships.” (Control Prime).
The purpose of the control prime is to capture the effects of Ss true dispositional
attachment without alteration.
Then I asked “By what name do you call this person? (e.g. Dad, Kelley. We only ask so
we may refer to him/her by name in the next few questions.)” I also requested the attachment
figure’s gender.
Part II: Visualization & Situation capture. With the name and gender collected from
the previous page, I personalized the primes for the next set by “piping” this information into the
visualization instructions. These instructions are below with a sample piping of “Casey” as the
attachment figure, who has been identified as female:
“Take a moment and try to get a visual image in your mind of Casey. Imagine
what she looks like. What is it like being together? Remember a time you were
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 42
actually with Casey. What would she say to you? What would you say in
return? How do you feel when you are with Casey? How would you feel if she
was here with you now?”
After the visualization, Ss were asked to recall and describe two situations, one with
threat and one without, involving the person. The initial prime was also piped into these
instructions in order to assure the situations recalled were related to the attachment manipulation:
“A situation that typifies your relationship with Casey, where there was no
unusual stress in your life or threat to yourself or your goals, and where [prime
repeated, “you felt…”] What was the situation and how did Casey respond to
you? “ (non-threat)
“A typical situation with Casey, when you DID have some stress in your life or
threat to yourself or your goals, that typifies how [prime repeated, “you
felt…”] What was the stressful situation and how did Casey respond to you?”
(threat)
Measures.
Chronic Attachment. The ECR-R was administered before the Attachment Prime and
Situation Capture discussed above. This will allow me to control for the Ss own chronic
attachment while looking at primed attachment.
Original Goal Taxonomy Q-Sort described in Study 1was used again but this time with
the new and more comprehensive taxonomy of 44 goals recently developed with colleagues
(Talevich, Read, et al., under review). Unfortunately, it was not feasible to administer the goal
congruence measure in this study due to time constraints.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 43
Emotions. As you will recall, during the prime, participants provided one-sentence
descriptions of a situation involving the chosen attachment figure. Participants were presented
with that description and below it was asked, “What did you feel when this was occurring?”
followed by scales for 1) fear/anxiety, 2) angry, 3) sadness, and 4) positive/happiness (four 5-
point scales from “Not my feelings at all” to “Clearly described my feelings.”
Behavior. Below the Ss own one-sentence description of the situation, they were asked
to describe what they did in response to said situation: “How did you respond to this situation,
what did you do?.”
On another page, Ss were asked to code their own self-reported behavioral responses by
checking one of the following four descriptions. Descriptions were composed by this author to
represent archetypal attachment behaviors represented in the literature at large. “What would be
the closest categorization of this behavior?”
“I approached, or reached out to, the person in some positive way to get comfort,
help, or just enjoy being close.” (Proximity Seeking)
“I showed or voiced my objection to the situation.” (Protest)
“I tried to ignore or avoid the situation or person, deal with it on my own, and push
down any negative feelings I had.” (Dismiss).
“I created, explored, or discovered something new.” (Explore)
This procedure most closely mimics the behavior of the Motivated Behavior System
neural network model, which generates categorical output.
Results
Participants: This online experiment was taken by 110 female undergraduates from the
psychology subject pool at the University of Southern California. There was no theoretical or
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 44
practical reason to expect gender to interact with the attachment manipulation or affect the
intervening process model under study. Therefore, it was expedient to funnel the male
participants into Study 1 in which gender effects were plausible. This is a small sample size for
a structural equation model. Power could be a concern. On the other hand, bootstrapping allows
for reduced sample size. Any issues with the model possibly due to sample size will be
addressed in the results section.
Preliminary Analyses.
Chronic Attachment. As with previous studies, the scree plot showed two distinct
factors for the ECR-R. Three items from each scale double load on the other, but to a lesser
degree. All items were retained. Both the Avoidance ( α=.88, M=3.4, μ=3.5, SD=.97) and
Anxiety ( α=.94, M=3.7, μ=3.6, SD=1.19) composite are normal and have no outliers.
Attachment Primes. The four attachment prime conditions were effect coded into three
vectors. The control prime was effect coded -1, the relevant experimental prime (e.g. Secure)
was coded 1, and the remaining experimental primes (e.g. Anxiety and Avoidant primes) were
coded 0 to be ignored. Thus, we created three composites, which are Secure Prime vs. Control
Prime, Anxious Prime vs. Control Prime, and Avoidant vs. Control Prime.
Motivation. The goal clusters selected from the Taxonomy of Human Goals for the
following structural equation model includes Emotional Intimacy, Avoid Rejection, and Avoid
Effort. Since the goal measures were administered after the ECR-R and the attachment prime, it
is reasonable to expect these procedures activated attachment-related goals for our participants.
Attachment theory would suggest that motives that are particularly relevant to attachment are:
emotional intimacy (Collins & Read, 1990) and avoiding rejection (Feldman & Downey, 1994).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 45
Likewise, the Motivated Behavior System model ( Talevich, 2012) includes avoid rejection as
well as the goal to avoid effort.
Behavior Variables. This experiment introduces an original procedure in which
participants self-categorize their own behaviors. For analyses, the categorical behavior variable
was dummy coded into three dichotomous variables: Proximity Seeking, Protest, and Dismissing
SEM Model 2.0. The SEM model is depicted in Figure 7 with standardized beta
estimates. (All tables and in-line results are unstandardized). Table 8 lists the hypothesized
indirect effect results whereas the regression weights (total effects) can be found in Table 7.
In support of the process model as a whole, motivation and emotion together mediate the
relationship between attachment style and behavior (H
me
). Model 2 depicts a process of emotion
by which fear moves into anger and then into sadness. Similarly, among the behaviors, we see
proximity seeking increases protest but protest decreases proximity seeking, as do dismissive
behaviors. Although these relationships were not initially postulated, they replicate the
relationships found in the simulation data generated by the MABS neural network. Consistent
with attachment theory, anxious attachment significantly predicts goals to Avoid Rejection,
explaining 10% of the variance. Avoidance negatively predicts the Emotional Intimacy cluster,
as does the secure prime, explaining a total of 8% of the variance in these goals.
As reviewed in the introduction, there can be multiple indirect effects in operation
together but in opposite directions such that they cancel, or suppress, one another (Hayes, 2009).
In the present model one will observe that the indirect relationship between chronic avoidance
and all three behaviors approach significance. Yet the only path between them is through
Emotional Intimacy and Fear– yet these have a direct relationship that is insignificant (b=.1,
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 46
p=.22). It is a similar story for the prime. The indirect effects for the prime on the three
behaviors are all significant despite only a marginal effect between Avoid Effort and Fear.
Thus, motivation mediates between attachment style and emotions. Specifically, goals to
avoid effort, have emotional intimacy, and avoid rejection are responsible for trait biases in
emotionality: that is, different people respond to the same situation with different emotions
because different goals are triggered (Table 8).
Hypothesis H
e
, in which emotions mediate between motivation and behavior, is also
upheld: here, fear mediates the relationship between protest behaviors and goals. Putting it all
together, primed security satisfies needs for emotional intimacy and seems to be generally
motivating of effort (reduction of avoiding effort.) Having these motives is directly related to
fear, thus with their reduction, fear is abated. Since fear is directly related to protest behaviors,
when it is abated, protest simmers down.
Chronic Avoidance is also involved in the above process. It is, like primed security,
inversely related to the motivation to be emotionally intimate. By reducing needs for emotional
intimacy, Avoidants reduce their fearfulness, and this leads to a lessening of protest behaviors.
The Motivated Behavior System predicts sadness mediates between goals, such as the
need to avoid rejection, and dismissive behavior. In the current analyses, this is accomplished
via an indirect process through anger.
Anxiety, is strongly and directly related to the need to avoid rejection. This goal is
directly related to anger. Attachment theory describes protest behaviors as “angry” but in the
present model, the relationship between anger and protest behaviors is insignificant (though
necessary for the model. However, the indirect relationship of Avoid Rejection to protest
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 47
behaviors, through anger, is significant. The direct relationship between this goal and protest is
also significant but, unexpectedly, negative. I will return to this matter in the discussion.
Discussion & Limitations
Protest is the unique behavioral hallmark of attachment anxiety and proposed to arise
from anger. A well-established line of research by Downey et al (G. Downey & Feldman, 1996;
G. Downey, Mougios, Ayduk, London, & Shoda, 2004) has implicated sensitivity to rejection –
that is, angry feelings about potential rejection spur protestations. This seems to be at odds with
the present model’s direct relationship in which goals to avoid rejection reduce protest. But in
fact, it is consistent with emerging work by Downey et al on “Self-Silencing” (Geraldine
Downey, Romero-Canyas, Reddy, & Rodrigues, 2012; London, Downey, Romero-Canyas,
Rattan, & Tyson, 2012) In their work, individuals high in rejection sensitivity will often put
their own needs and concerns aside for a significant other: reducing expression of true feelings
and desires. (This reprieve may be short-lived if the partner is later perceived as rejecting (so
the goal to avoid rejection has failed). Hostility will be unleashed upon the partner who failed to
appreciate the sacrifice (Geraldine Downey et al., 2012).)
Limitations. There is one seeming inconsistency between the attachment literature and
this model. One would expect to find a relationship between fear and proximity seeking.
However, while Ss may have felt fear during the actual event, their recollection of the event is
not likely to invoke fear retrospectively. Whereas the recollection of a past situation that was
fearful, angry, or sad is more likely to evoke anger or sadness in the present moment, when one
is free from any danger. Also, post-prime emotions may have been mixed since the prime
included recollection of both a threating and non-threatening situation.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 48
In order to experimentally replicate the findings of Study 1 the attachment prime needs be
related to the selected motives. Partial success can be reported: the secure vs. control contrast
prime is relate to the motives to Avoid Effort (negatively) and to have Emotional Intimacy
(positively). Unfortunately, the insecure primes do not predict the importance of the specific
motives theorized. Recall that, right before the goal measure, participants were asked to recall
both a threatening and non-threatening situation involving the attachment figure. Threatening
and non-threatening situations will activate different goals and so the two situations were
confounded at the time of the goal measure. Only the secure situation would activate goals
consistently across both threatening and non-threatening situations. For, if there is no threat, or
if there is threat but with the assurance of help, then the outcome to prepare for is much the
same: everything will probably be okay. Because goals can be activated by the situation and/or
disposition it makes sense that secure prime and dispositional attachment are the variables that
predict these specific goals.
Study 3: Experimental Validation of the Entire MABS Model
Study 2 provided validation for several elements in the MABS Model (those shown in
Figure 6). By manipulating attachment, it provided some causal evidence for the model and trait
formation. Most importantly, Study 2 marked the first foray into developing an experimental
design that could successfully test the intervening processes of the MABS model. Study 3 seeks
to validate the full model (as shown in Figure 8). Study 3 improves upon the procedure
invented for Study 2 as well as including measures omitted in Study 2: goal congruence and
construals. Last, but not least, Study 3 measures actual real-time behavior.
An attachment situation will be induced by separately priming threat and attachment
figure responsiveness. The situation will be a current one (as opposed to recalled or intended).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 49
Participants will rate their most important goals and their congruence with the current situation.
They will rate their emotions. And finally they will compose a letter to the attachment figure
regarding the current situation in which they find themselves.
The hypotheses of Study 3 are:
H
c1
: Construals of threat will be the product of the threat manipulations (which
represents the objective situation) and the participants own chronic attachment
orientation.
H
c2
: Construals of the attachment figure’s behavior will be the product of the
responsiveness manipulation and the participants own chronic attachment
orientation.
In both H
c1
and H
c2
, attachment insecurity should increase construals of threat
and unresponsiveness though effects may be greater for attachment anxiety given
it is characterized by hyperactivation (and sensitivity) whereas avoidance is
marked by deactivation (and insensitivity).
H
m
: The Impact of an ongoing situation upon one’s goals will mediate between
construals and how one feels about that situation.
H
e
: Emotions will mediate between Goal Impact and the proportion of proximity
seeking, protesting, and dismissive sentences that are selected in the composition
of a letter to the attachment figure.
H
t
: Attachment avoidance and anxiety may influence any element in this process
from construals through behaviors and, from that element on, ,the trait will
indirectly influence the elements that follow.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 50
Method
Attachment Situation Manipulation. Rather than manipulate attachment orientation, as
in Study 2, I will manipulate the details of an attachment-related situation. This manipulation
will serve to represent the objective details of the situation. I will then measure how people
construe the objective situation.
This is a two by three design: Attachment Figure Status (attentive, not attentive) X Threat
(none, relationship, non-relationship). The typical attachment primes aim to invoke a particular
trait/chronic attachment style by having participants recall a relationship in which the application
of a particular attachment strategy has become chronic. However, by priming a chronic
attachment style, the researcher loses the ability to tease apart the situational and chronic
responses. Thus, I have devised an attachment situation prime. Rather than being asked to recall
a relationship characterized by responsiveness or unresponsiveness, as attachment primes do,
participants are asked to describe an ongoing situation in which an attachment figure is currently
being responsive (or not). Furthermore, this behavior may be typical, or in fact unusual, for the
relationship. In addition to priming situational responsiveness, the current manipulation will
allow me to prime situational threat.
Threat Manipulation. By random assignment, participants were asked to “Please think
of a situation in your life that is…
1. … going smoothly, well for you” (No Threat)
2. … going poorly and is causing you stress. this should not be a problem involving a loved
one or someone with whom you have a meaningful relationship (e.g. boy/girlfriend,
parents, etc.)” (Threat from without)
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 51
3. … going poorly and is causing you stress. This should be a problem involving a loved
one or someone with whom you have an important, meaningful relationship (e.g.
boy/girlfriend, parents, etc.). (Threat from within)
All Ss were further told “It's important you select a situation that is 1) still ongoing and
that 2) there has been time and opportunity, if you wished, to share it with others - including
someone with whom you have an important relationship. Take a moment to think about it.”
Thereafter they are asked to type a description of the situation.
Threat was manipulated on three levels because a threat involving an attachment figure
may be experienced differently than a threat from outside the relationship. However, given the
complexity of the models of Studies 2 and 3, the two threat conditions were collapsed to one
binary variable to represent threat (1) or no threat (0).
Attachment Figure Status Manipulation. By random assignment,
1) “Whose supportive involvement in this particular situation has been most meaningful to
you: their being there for or with you has been the most impactful? (This should be
someone with whom you have an important or meaningful relationship (e.g. romantic
partner, parent, or other giver of care ).” (Attentive/Responsive)
2) “Whose lack of supportive involvement in this particular situation has been most notable
and meaningful to you? (This should be someone with whom you have an important or
meaningful relationship (e.g. romantic partner, parent, or other giver of care ). Maybe
they are usually there for you, with you, or maybe their being uninvolved is typical of
your relationship. Either way, this time, they aren't there. “ (Inattentive/Unresponsive)
Ss are then asked the name of the attachment figure (AF), which we will then use to
refer to the AF in the future. Then they are asked to type a description of “what has [AF]
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 52
done or said (or not done or not said) that has had the most impact on you in this particular
situation?”
Measures.
Construals. To measure construed responsiveness and harmfulness, the next set of
questions asks how attentive, available, or harmful the attachment figure is being in this situation
(each on an 8-point scale). On the next page participants are asked the same 4 questions except
the introduction is: “So, excluding this particular situation, how is [AF], typically speaking?”
Construed Response. 1) “Available, physically present, or reachable?” 2) “Responsive?”
3) “Attentive?”. And the same 3 questions above are repeated appending “Typical to each”, e.g.
“Typical responsive”. These six items will be combined to create the Responsiveness Past &
Present variable in analyses for the sake of parsimony. Though the effects of past vs. present
responsiveness should differ at a fine-grained level, their functions are expected to be similar
from a birds-eye view and are, thus, combined to avoid further complicating the model.
Construed Harm. 4) Ss are also asked how harmful, excepting being unresponsive or
unavailable, do they find the AF’s behavior currently and, on the second page, typically. These
are then combined, again, for the sake of parsimony in the model.
Construed Threat. 5) Ss are also asked to rate the level of threat they perceive in this
situation on a 6-point scale from “not at all threatening” to “extremely threatening”.
Goal Importance. Participants were presented with a list of 46 goals (Talevich, Read, et
al., 2012) and asked to select the 15 most important to them under the circumstances they had
previously described.
On the next page, they were asked to rank, in order from most to least important (1 to
15), the goals they had just chosen. These rankings were then scored as follows: #1 thru #2 = 7,
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 53
#3 thru #5 = 6, #6 thru #10 = 5, and #11 thru #15 = 4. This replicates the values that would have
been assigned to these top 15 goals if the entire set of 46 goals had been sorted. This procedure
shortens and simplifies the task for participants while retaining maximum predictive power. In
previous work (e.g., Talevich, Walsh, et al., 2012) we have found that the top 11 goals are as
predictive as the entire set.
Goal Congruence. Participants were asked to rate how much harder or easier it is achieve
each of their 15 goals under the circumstances they describe. On one page they were asked how
much harder (1 no harder to 5 extremely hard) it was to achieve each goal, and on the next page,
how much easier (1 no easier to 5 extremely easy) it was to achieve each goal.
Emotions. Emotions are measured after the goal congruence measure. Participants are
asked how happy, sad, angry, and afraid they feel, given the situation and how it affects their
goals, on a 5-point scale.
Behavior Measure. Participants were asked to compose a letter, addressed to the AF, by
selecting sentences from each of three categories: Proximity-Seeking, Protest, and Dismissing.
A pilot was conducted to generate sentences for the Letter Composition behavior measure (See
Appendix).
Results
Dispositional Attachment. The ECR-R was factor analyzed and the scree plot shows
two distinct factors. All items loaded on their expected factor, anxiety or avoidance, with a few
items double loading. Three anxiety items double loaded on avoidance and 4 avoidance items
double loaded on anxiety but none with an absolute value greater than .386, thus all items were
retained. After reverse coding appropriately, the 18 anxiety items had a reliability of .937 and the
18 avoidance items had a reliability of.942. Two composites were created. The Anxiety
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 54
composite had a mean of 2.7 and standard deviation of .75 whereas the Avoidance composite
had a mean of 2.4 and a standard deviation of .718. As is typical of attachment composites,
neither was skewed.
Goal Impact Aggregate Composites. Goal Impact is the product of the importance of
each goal multiplied by its facilitation and inhibiting congruence with the situation. Thus, the
importance of each of the top 15 goals was multiplied by the “easier” congruence ratings. The
facilitative impact composite (M=11.0, SD=5.6) is the mean of these 15 impacts. This procedure
is repeated to create the Inhibitive Impact composite (M=11.4, SD=4.8) such that each
importance was multiplied by the “harder” congruence rating for each of the 15 goals.
In Study 1, only Facilitative Impact factored meaningfully into the decision to seek out a
relationship. Here in Study 3, only Inhibitive Impact appears to be related to this attachment
manipulation. This makes sense since the attachment system is activated in response to threats.
Facilitative Goal Impact, therefore, will be excluded from the models presented below.
Main Analyses:
SEM Model 3.1. All Conditions are included. Regression total effects are reported in
Table 13. All hypotheses from the Overview of Studies (Table 1) are confirmed and reported in
Table 14: Indirect Effects & Hypotheses.
Construals mediate between motivation and both trait attachment and the objective
situation (H
c
). Construed threat is a product of both objective threat and attachment anxiety and
it mediates between them and Goal Impact. The construed responsiveness of the attachment
figure also mediates between the objective attachment figure response (which was manipulated)
and Goal Impact. Surprisingly, there was no relationship between construed responsiveness and
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 55
anxiety –at least that was significant across all conditions. Analyses were broken down to
examine these relationships situation by situation. Attachment anxiety is significantly related to
appraised responsiveness in the no-threat-AF-not-responsive condition. So, when there was no
objective threat but the attachment figure was objectively unresponsive, attachment anxiety
further exaggerated the perception of unresponsiveness (b=.75, B=.41, p<.05) accounting for
16% (r
2
smc)
of the variance of the construal in this condition.
Inhibitive Goal Impact mediates the relationship between both construed threat and
construed responsiveness and fear and anger. Said another way, Goal Impact mediates between
working models and emotions. And it is through construals, or working models, that Goal
Impact mediates between attachment anxiety and emotions.
Motivation, in the form of Inhibitive Goal Impact, mediates between construals &
emotions. The difficulties for one’s goals (Inhibitive Goal Impact) from construed threat, and
the lessening of these difficulties from construed attachment figure responsiveness, determines
the amount of fear or anger one feels.
Next, emotions mediate between motivation & behavior (H
e
). Specifically, when fear is
evoked because goals are thwarted (inhibited), one seeks proximity. But when anger (or perhaps
frustration) arises in response to the thwarting, protests follow.
Anxiety directly influences construed threat, goal impact, and fear. This is consistent
with the idea that traits are a “build up” of chronic associations. The effects of attachment
avoidance are not significant across all conditions (except that it is related to reduced proximity).
This will be addressed in the follow-up model to come. However, there is one more interesting
question to be addressed by the current model.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 56
Which is more predictive of behavior: who you are or the situation in which you find
yourself? To test this question, I created two variations on the current model. One in which only
the situational variables were represented, manipulated threat and responsiveness, but the trait
attachment variables were excluded –and vice versa for the second model. Table 12 shows the
variance accounted for in the dependent variables in each of the three models: the current model,
the trait-only model, and the situation-only model. At the construal level, the situation reigns
supreme –accounting for almost all of the variance explained: the objective situation model
accounts for 44% of the variance in Construed Responsiveness. The trait model accounts for
none. The objective situation model accounts for 19% of the variance in construed threat and
the trait model a mere 2%.
However, the trait model accounts for nearly the same amount of variance in motivation
and emotion such that by the time effects have made their way through the mediational pathways
to predict behavior, there is no (or at least only a negligible) difference between the objective
situation model and the trait model in explaining the variance one finds in behavior.
SEM Model 3.2: The Secure Situation. Attachment avoidance is not significantly
related to the mediational process in Model 3.1 depicted in Figure 9 which fits the data across all
conditions. A possible conclusion might be that this process applies to some personality traits
but not to others. I would suggest, more precisely, that the influence of personality depends on
the nature of the situation. Attachment anxiety is characterized as a chronically hyperactivated
state of the attachment system. In other words, the attachment system is always on –with little
regard for the situation at hand (or experimental condition). Thus we see effects for anxiety
across all conditions. In a sense, having attachment anxiety is like having threat all the time.
Avoidance, on the other hand, is characterized as a chronic deactivation of the attachment
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 57
system. Despite a wish to remain independent and unaffected, threat will activate the attachment
system at least somewhat. The next model will look at a non-threatening situation in the hopes
of revealing the influence of attachment avoidance.
One disadvantage of looking at a single condition is that I cannot also use the
manipulation variables. Otherwise, results for the secure situation Model 3.2 (Figure 10) are
consistent with Model 3.1 (Figure 9). Regression coefficients are reported in Table 16 and
hypothesized indirect effects in Table 17. This time, trait avoidance (b= .277, p=.029), but not
anxiety (b=1.37, p=.627), predicts construals. The perceived harmfulness of the attachment
figure in this secure situation strongly predicts the appraisal that one’s goals are being thwarted
(Inhibitive Goal Impact). Thus, construed harmfulness mediates between attachment avoidance
and motivation (H4a). Motivation, in turn, carries the effects of traits and construals indirectly to
emotions. That is, attachment avoidance effects emotions by way of the perceived impediments
that construed harmfulness has upon one’s most important goals. There is also a direct effect
from construed harmfulness to sadness, possibly indicating a learned association that does not
require an assessment of current goals.
The thwarting of one’s goals induces a feeling of sadness and this, in turn, reduces
Proximity Seeking (He). Emotions mediate between motivation & behavior (He). And so in
confirmation of all hypotheses, motivation and emotions have mediated the relationship between
trait attachment, construals, and behaviors.
Discussion & Limitations
In Model 3.1, Inhibitive Goal Impact predicts fear and anger. But the SEM model does
not indicate why one emotion or the other (or both) would be selected. However, the neural
network version of MABS does suggest why. Fear is elicited when there is threat and
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 58
uncertainty of a response. Naturally, these elements create a situation that is bad for getting what
one wants in the situation: help. Anger, on the other hand, occurs when the situation is more
generally thwarting –or merely not impacting goals positively. Sadness, on the other hand, is
computed as coming from an especially low Goal Impact. Now, let’s take a look at the
standardized regression beta weights and see if they are consistent with the calculation for the
neural network model. Inhibitive Goal Impact in Model 3.1 is predicted by construed threat
(B=.22) and responsiveness (B= -.26). Fear and anger result. But in Model 3.2 where sadness is
produced, construals of harmfulness predict the thwarting of important goals with far more
strength (B=.4). Therefore, construed harmfulness is perceived to make goals much harder to
achieve than threat or responsiveness construals. This would be consistent with the neural
network which models the source of sadness as a situation in which goals are much harder to
achieve then the situation that elicits fear or anger.
Dismissing the Dismissing: In Study 3,the Dismissing behavioral measure appears to
have failed. There are two likely causes for this. First, it may be that the sentences chosen from
the pilot were not the best choices to represent dismissive behavior. In the future, a subject-
driven taxonomical study on the attachment sentences could be illuminating. However, it may
be that dismissing behavior is not what is done but what is not done. I appear to have captured
relationships I expected to find with dismissing behaviors through a decrease in proximity-
seeking. This may also be due, in part, to the interdependence of the behaviors. Ss are limited in
the total number of sentences they may select. If one chooses more dismissing sentences, then
one must select fewer proximity sentences and/or fewer protesting sentences.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 59
Dangling happiness: It should not be concluded from this model that it does not apply to
happiness. In the attachment literature, security enables exploration. This is the behavior that
would be mediated by happiness -a behavior not measured in this design.
Happiness could also lead to proximity seeking, in particular situations. It is intuitive that
when someone is happy they wish to share it with someone special, and get closer, particularly if
that special someone is the source of the happy situation. However, many of the proximity
seeking sentences in this experiment involved asking for help. Thus, I have not modeled all
proximity seeking behaviors but those most critical in the attachment literature: help and
comfort-seeking.
Why is it always the bad stuff?: In these analyses, it has been inhibitive, but not
facilitative Goal Impact involved in these mediations. The contents of the model are all specific
to the domain of attachment security. Activation of the attachment system occurs in response to
threat. Thus, although certain aspects may be facilitating, such as attachment figure
responsiveness, the model as a whole is about the processing of threat, and threats are rather
consistently inhibitive of a person’s life goals - whatever they may be. Recall that in Study 1, it
was only the Facilitative Goal Impacts of being in or out of a relationship that mattered in
predicting intentions to seek a (or another) relationship.
Study 4: Who Pays on a Date
Studies 1-3 have shown how the process of motivations and emotions influence
traditional attachment behaviors. These are examined again (validating the elements depicted in
Figure 11) with real-world behavior. The purpose of Study 4 is to replicate the results of
previous studies with a subject that has not been considered in the attachment literature:
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 60
Who Pays on a Date? It’s a common question among singles as indicated by the
abundance of advice from online dating sites and etiquette experts. Dating etiquette was once
clear cut: the man pays for a date. Part of this was owing to the fact that women did not work
and had no means with which to contribute. The feminist movement of the 70s changed the
game. For the first time, women (in quantity) were able to provide for themselves and their
children. The man’s role as sole provider was not desirable to her and, particularly undesirable,
was the imbalance of power these social roles entailed. Power is money, and when women were
finally able to earn it for themselves, they wanted an even playing field at work and at home.
This changed dating etiquette as women began to assert their independence by contributed to
dating expenses.
Many decades have passed and the positions of women in the workforce, and as financial
contributors at home, are no longer contentious issues but the norm. But though the inequality of
power and money is much changed between men and women, the inequality of parental
investment and risk in dating remain the same. Nearly 20% of all women in the US have been
raped (Kilpatrick, Resnick, Ruggiero, Conoscenti, & McCauley, 2007). This number does not
include attempted rape or other unwanted advances . The majority of assaults upon women are
by their dating partners (22%). And while the most common locations for assault were his home
(31%) or her home (26%) just over 20% of assaults occurred in public dating venues: parties,
cars, outdoors, and bars . If an encounter leads to intercourse, consenting or otherwise, she
alone bears the risk of pregnancy. This is far from an unusual scenario, the 2010 Census
indicates 36% of all births that year were to unmarried women (widowed, divorced, or never
married). And, if a woman decides to raise the child of a man who is not bonded, to her or to the
child, then her minimum 9 months becomes a lifetime investment borne alone.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 61
In bearing these risks, and accepting a date with a man, a woman is investing herself
upfront. Paying on a date is one way for a man to demonstrate his own investment. If she is
wise she will then wait for him to invest. If the gender-specific dating roles of courter (suitor or
pursuer) and the courted (pursued), developed in response to the inequality of minimal
investments, then who pays on a date should be influenced by the extent to which these roles are
adopted and their norms upheld or defied.
According to the MABS model, the extent to which these norms are followed will depend
upon the results of the motivation sub-system: Goal Impact. That is, the extent to which a role is
adopted, and its norms followed or violated, will depend on how it impacts the individual’s most
important goals. For instance, if one’s most important goals are about relationships, and
securing one, then following dating norms may best behoove goal achievement. Conversely, if
one’s most important goals are to accrue money or develop a career then the preference would be
to invest resources in this direction rather than in dating, and the pursuer role could conflict with
these goals.
These different motives have been linked to attachment orientation. Of course, those
who are anxiously attached will have particularly strong relationship goals whereas those who
are avoidant will tend to deemphasize them. Insecure attachment has also been linked to beliefs
about money. This makes sense if one considers that if one is insecure in their attachments they
may seek alternative means of security – such as financial security and security through power
and control over others. Indeed, attachment avoidance has been linked to power-money motives
– the belief that money can purchase power and control over others. Attachment anxiety has
been linked to preoccupation with and the hoarding of money (Mikulincer & Shaver, 2008).
Being the pursuer (and investor) has costs for the accrual of money and, by extension, the accrual
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 62
of power in general. So attachment anxiety and avoidance should both conflict with the role of
pursuer. On the other hand, for those high in attachment avoidance, the role of pursuer could be
consistent with spending money to gain power and control on a date.
But dating is a collaborative dance between two people, and decisions are not just based
on one’s own motivations alone, but in response to those communicated by the other. Social
norms are scripts that can guide behavior but must be adapted to work within the circumstances.
Thus, how much one pays on a date should be in response to how much their partner offers to
pay. Therefore, how much a person’s dates offer to pay will be included as a variable. This
marks the first inclusion of a situation variable in the model.
The hypotheses addressed by Study 4 are:
H
m
: the Goal Impact of the two dating roles (being the active pursuer or the
passively pursued) will intervene to predict how much one pays on a date from
how much their date offered to pay, attachment avoidance, and attachment
anxiety.
H
e
: Specific emotions evoked by their relationship to the Goal Impact of dating
roles, will influence how much one pays on a date.
H
t
: Attachment anxiety & avoidance will directly and indirectly influence
behaviors. These traits will also directly influence specific goals and/or emotions
that, thereafter, partially mediate between traits and behaviors.
Method
Measures. Measures were administered in the order they appear.
Chronic Attachment. The ECR-R was administered as in Studies 1 and 2.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 63
Who Pays on a Date. Participants were asked “When you are on a date, who
most frequently pays for it?” and were asked to rate each of the following on a scale from
1-strongly disagree to 7-strongly agree: “My date pays for both of us”, “I pay for both of
us”, and “We split the bill.”
Q-Sort. A new Q-Sort procedure was created for this study. As in study 2, 46
goals were taken from the Talevich, Read, Walsh, Chopra, and Iyer taxonomy (2012).
As with the original procedure (Study 1), there were 3 pages of sorting tasks each with a
list of goals on the left. On the right of each page were 3 boxes into which Ss were to sort by
importance their top-15, mid-15, and bottom-15 goals. Each box allowed up to 16 goals. Note
this required Ss to sort all 46 goals on the first page. On the second page, the goals that had been
previously dragged into the top-15 box now appeared in the panel to the left. To the right were,
again, three boxes into which Ss were to sort, by importance, their top-5, mid-5, and bottom-5
goals. This procedure was repeated on the third page in which the previous page’s top-5 sorted
goals appeared to the left and, on the right, boxes into which Ss were to sort their two most
important goals and their three second-most important goals (top-2 and bottom-3 of the most
important 5. The final distribution was quasi-normal and duplicated, on average, the values of
the original Q-Sort (See Figure 12 for a comparison of the old and new q-sort distributions).
Goal Congruence. Participants were asked to rate the extent to which two situations
facilitated or inhibited their most-important life goals. To reduce the task for participants, they
were only asked to rate the top 15 most important goals. The two situations represented the
adoption of two different but complimentary social roles: being the romantic pursuer vs. being
romantically pursued: “The following set of questions ask what makes your goals harder or
easier to achieve.” On an 11-point scale Extremely Hard (-5) to Extremely Easy (5) with a
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 64
Neither (0) middle point. Active Pursuer: “To what extent would pursuing the other person,
being the initiator, and provider on dates help you achieve the following goals (you have for
your life as a whole) over the coming year:” Passive Pursued: “Please rate the extent to which
not pursuing but instead being the one pursued and provided for, but also the one waiting on a
date's initiative, would make it harder or easier to achieve each of the listed goals over the
coming year.” All participants were asked to rate both situations for their top 15 goals for a total
of 30 ratings.
Emotions. On a 5-point scale from “Not my feelings (1)” to “Clearly describes my
feelings (5)” Ss rated four emotion categories: Fear: “Afraid, scared, frightened, nervous, jittery,
shaky”, Anger: “Angry, hostile, irritable, scornful, disgusted, loathing”, Sad: “Sad, blue,
downhearted, alone, lonely”, and Happy: “Happy, joyful, delighted, cheerful.” The descriptive
terms were borrowed from the PANAS (Watson & Clark, 1994).
Who Pays administered for a second time, as described above.
Partner Offers to Pay. Ss were asked about the last 7 people with whom they have had
at least one date. A date was defined as any time they’d gone out with a person with a romantic
or sexual intent, including a “coffee date”. They were asked to include dates whether they
“turned into” something or were disappointing. Then they were asked to indicate, from 0 to
100%, “Whomever paid in the end, how much of the expenses did your date offer (or indicate,
perhaps by a wallet reach, etc.) they would like to pay?“
Results
Participants. Recruited from Amazon’s Mechanical Turk (MT), which has an abundance
of individuals under 25 on MT, a pre-screener was used to limit participant participation. We
screened for age and relationship status. Participation of those under 25 was limited to allow
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 65
room for those over 25. We only accepted those who were unmarried and either not in a
relationship or who had been in their current relationship less than two years. The final number
of participants was 234, which excludes one case for having missing data. There were 84
women and 150 men of ages 18-61 who participated. For a breakdown by age and gender, see
Figure 13. Age is not a variable in this study. However, it seemed plausible that dating variables
could be different between college age (18-24), the median marrying years (25-29), and
thereafter, so efforts were taken to ensure they were represented.
Composites.
Attachment composites and preliminary analyses. As with previous studies, the scree
plot showed two distinct factors for the ECR-R. No items double-loaded above .3. All items
were retained. Both the Avoidance ( α=.95, M=3.2, μ=3.3, SD=1.1) and Anxiety ( α=.94, M=3.9,
μ=3.8, SD=1.2) composite are normal and have no outliers. Gender and attachment are both
relevant to the dependent variable of who pays on a date. A recent meta-analysis suggests a
relationship between attachment and gender in community samples but not web studies.
Consistent with other web studies, bootstrapped regression confirms that the relationship
between attachment and gender was totally insignificant (Avoidance b=-.02, p=.8; Anxiety
b=.09, p=.6). A relationship between age and attachment avoidance approached, but failed to
reach, significance whether controlling for gender (b=.07, p=.13) or not.
Who Pays Composites. Time 1 and Time 2 who pays variables were all correlated above
.8 and were, therefore combined. See Table 19 for descriptive statistics.
Goal Impact Composites. A total of six goal impact composites were created. The first
two comprise the overall impact of being in the pursuer or the pursued dating role. Then, four
finer-grained composites were created in order to separately examine the facilitative and
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 66
inhibitive effects for each of these two situations. First, all 46 goal importance ratings were
multiplied by the goal congruence rating. The congruence ratings ranged from -5 (the situation
made achieving the goal extremely hard) to +5 (the situation made achieving the goal extremely
easy) with an intermediate zero to represent neither harder nor easier. This was done twice, once
for each of the two situations presented: 1) being the active pursuer and 2) being the passively
pursued:
In order to compare the facilitating aspects of the situation to the inhibiting aspects of a
situation, I separated each goal impact into facilitative and inhibitive goal impacts. To do this, I
copied values 0 through the highest goal impact (5) to create 46 facilitative impacts. Then I
copied the lowest impact score (-5) up through 0 to create 46 inhibitive impacts. So, in this
study, all of the values below 0 were dropped for the facilitate composites, and all above 0 were
dropped for the inhibitive composites. In Study 1, zero values had been excluded from both
composites to reduce their inter-dependence. But the composites were correlated nonetheless, as
one might expect, since they are really two sides of the same scale. Next, the 46 facilitative goal
impacts of the situation were summed to form an aggregate composite and the 46 inhibitive goal
impacts of the situation were summed to form another. This procedure was done twice, once for
each dating role. The final composites are as follows.
Active Pursuit Impact= Goal Importance x Congruence as active suitor.
Active Pursuit Inhibiting Impact = Goal Importance x Harder-Congruence as the active suitor.
Active Pursuit Facilitating Impact = Goal Importance x Easier-Congruence as the active suitor.
Passive Pursuit Impact = Goal Importance x Congruence if passively pursued.
Passive Pursuit Inhibiting Impact = Goal Importance x Harder-Congruence if passively pursued
Passive Pursuit Facilitating Impact= Goal Importance x Easier-Congruence if passively pursued
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 67
These composites represent the overall assessment Ss make of whether the role of being a
romantic suitor or being romantically pursued makes it easier or harder for them to accomplish
their most important life goals. See for means and standard deviations, see Table 20.
Pay Offers Composite. A composite was computed from the 7 offers to pay the check
(or part of it) made by each person’s most recent 7 dates. Women report their last 7 dates most
often offered to pay the whole bill, which averaged to her 7 dates cover 78% of the expenses
(mode=100, μ=74,M=78 ,SD=23). Men’s dates usually offered to split the check, which
averaged to his 7 dates covering an average of 44% of the expenses (mode=50, μ
=44,M=46,SD=23).
Main Analyses
SEM Model 4.1 Why Women Pay on a Date. Regression Coefficients are reported in
Figure 14 depicting Model 4.1 as well as in Table 21. Hypothesized indirect effects in Table 23.
A woman is less likely to pay on a date when her date offers to pay. This effect is both direct
and indirect. When he offers to pay this (marginally) increases her sense that pursuing him is not
beneficial to her goals. The negative impact of active pursuit increases fear and decreases her
splitting the check. Recall that emotions are measured directly following goal congruence. So
this is a fear for her goals. What seems to be happening is this: when he offers to pay, she
perceives he is wanting to pursue her. (The best strategy for her goals, then, is to allow him to
do so). This produces a cautionary signal (fear) that inhibits her own pursuit, and her offer to
split the check, in order to allow him to court her. Attachment anxiety works along these same
pathways increasing negative impressions of pursuit, fear for goals, and inhibiting her tendency
to split the check. Attachment avoidance also increases fear but not due to the negative impact
of pursuit. Attachment avoidance motives are in the service of increasing distance and
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 68
decreasing dependence. It may be fear of intimacy or dependence of either party that leads
avoidant women to split the check. It is interesting to see fear inhibiting the same behavior but
to opposite ends. Attachment avoidance for women is incompatible with being courted, as
indicated by the negative relationship between avoidance and the total impact of being passively
pursued.
For women in general, the more positive being pursued is for their goals, the happier they
are, and the more likely she is to pick up the check. This may seem at odds –picking up the
check when it is best for one’s goals to be the individual pursued. But, if these happy moments
arise from the feeling she has been treated well then she may be inspired to reciprocate and treat
him (without the fear she will appear to be pursuing him.)
Anxiety is rather fascinating in this model. One indirect path decreases how often she
splits the check because pursuit of him seems like a bad idea (for her goals). The other indirect
path decreases how often she splits the check because her anxiety makes her less happy (which
leads to less reciprocation, if the above interpretation is correct). So both indirect paths are
negative, that is she is less likely to pay on a date the more anxious she is. Yet, the direct path
between anxiety and her paying the whole check remains significant and positive. As attachment
avoidance seeks to decrease dependency and emotional intimacy, perhaps the anxious motives
underlying her picking up the check more often (or perhaps reciprocating earlier in the
relationship) stem from a wish to increase his dependency (be it financial or in gratitude) and
hurry or heighten their emotional intimacy.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 69
SEM Model 4.2 Why Men Don’t Pay on a Date. Regression coefficients are reported in
Figure 15 and Table 22 and hypothesized indirect effects in Table 24. When his date offers to
pay on a date, a man is likely to accept: he increasingly splits the check and will less often pay
the whole bill. At least, the direct relationships are these. But the indirect relationship is
precisely the opposite. Sometimes, when his date offers to pay, his pursuit of her becomes less
in conflict with his goals. That is, pursuing her presents less of a hardship for his most important
goals. This leads to him feeling less sad and less angry. The reduction of negative emotions
here increases his paying on a date.
When he feels less angry he is more likely to pay for the whole check and when he feels
less sad he is less likely to split the check. One must consider the possible implications of a
woman splitting the check. As discussed above, if she splits the check it may indicate she is
unwilling to allow him to pursue her (perhaps from attachment avoidance or perhaps from a wish
to avoid him specifically –and he would not know which). If one of his most important goals is
to find a relationship, then this path is particularly intuitive. Anxiety also increases a man’s
sadness and anger independent of the congruence of his romantic role. Through the increase of
negative emotions, attachment anxiety reduces how much a man pays on a date by increasing the
frequency he splits the check and decreasing how often he pays the whole thing.
Attachment avoidance is related to incongruence between a man’s most important goals
and being pursued by a woman. It would be consistent with attachment theory if many of those
goals were related to maintaining his independence. When being pursued makes his goals
harder to achieve, a man feels more sadness and anger. So, as with anxious attachment,
avoidance indirectly reduces how much a man pays on a date.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 70
Discussion
It is interesting to note how avoidant attachment, in the service of the same purpose,
produces the opposite behavior in men and women due to their complimentary social roles. The
literature suggests that insecurely attached people are generally not compatible with one another.
But there is an exception: if one partner is anxious and the other is avoidant. An avoidant man
finds that both social roles (being the pursued and being the purser) are harder for his goals. A
temporal progression might explain this seeming contradiction. It would account for the data if,
in the beginning, an anxious woman is more likely to let a man do the pursuing because that
social role complements her goals. But the direct relationship contradicts the indirect route.
Anxiety directly encourages her to pick up the whole check on a date. Perhaps she is overly-
eager to reciprocate. This could complement the role of avoidance in the men’s model because,
at first, she will not pursue him (as would be incongruent with her goals and his independence-
related goals). But once emotional intimacy is on the table, pursuit becomes particularly
incongruent with his goals. Fortunately, he no longer needs to do the pursuing because his
anxious woman will maintain the relationship.
This model exemplifies the suppressor effects at work in this intervening process. For
instance, the relationship between Goal Impact and anger is not significant, however the opposite
signed direct and indirect effects involving anger indicate suppressor effects at work and the
indirect effects through anger are significant.
Offers to pay by dating partners could have been considered either an objective situation
variable or a construal variable because it relies on impressions from memory. However, it
seemed a dollar amount would be remembered with less subjectivity than something someone
said or how they acted toward oneself.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 71
General Discussion
Summary
These studies support the Motivated Affective Behavior System (MABS), an intervening
process model by which the objective situation is mediated by construals, a motivation sub-
system and emotions, to evoke behaviors. Furthermore, traits are found to influence, and be
mediated by, each step in this process.
These dynamics were originally modeled as a neural network (Talevich, 2012) pictured
in Figure 1. The current work validates this model with human data and so the model will now
be reviewed in light of the human-data findings.
In the neural network (hereafter called the net), six input layers represented six different
construals: a threat cue, an attachment figure, their current behavior and their past behavior, as
well as one’s own past behavior and one’s past inner state. The latter two construals provide a
“what happened last time” context by copying the previous state of the network and feeding it
back into the next experience. Structural Equation Models 3.1 and 3.2 of this work show how
construals are a product of the situation and personality traits but that the actual situation does
account for most of the variance in what is construed.
The next layer in the net is called the Situation Compute. This is a “hidden” layer that
dynamically computes a unique pattern to represent the construed situation as a whole. This is
theorized to represent schemata or working models of the self and others. In the computational
model, this schemata intervenes between construals and a motivation sub-system.
The motivation sub-system consists of goals represented as separate approach and
avoidance systems as well as a unique congurence layer for each goal. Into each congruence
layer is fed activations from its correspoinding goal and the Situation Compute. The highest
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 72
layer of the motivation sub-system is Goal Impact which receives inputs from all of the
congruence layers, all goals, and the situation compute. In the net, this layer is also a hidden
layer which dynamically computes a unique pattern to represent the impact that the situation has
upon active goals. This system operates in a similar fashion to the primary appraisals of
relevance and congruence specified by appraisal theory and this layer is analogous to the
appraisal outcome (Smith & Lazarus, 1990).
With colleagues, we have developed a simple formula that fairly represents this outcome,
Goal Impact. We ask participants to rate the importance of a goal and then to seperately rate,
on an 11 point scale, how much a particular situation makes that goal easier (+5) or harder (-5) to
achieve. We then multiply this scale value by the importance of the goal. Studies 1, 3, and 4 all
compute Goal Impact in this manner. Study 3 shows that construals do predict Goal Impact and
intervene between Goal Impact and both the situation and traits. (Traits are not a unique element
or layer in the neural network but are, instead, represented by the learned patterns linking the
elements throughtout the model. More on this later). Specifically, Study 3 finds that construals
of threat and responsiveness predicted the Goal Impact of an ongoing situation with which
participants were currently dealing.
In the neural network model, Goal Impact activations travel to the emotion layer which
contains the units fear, anger, sadness, and happiness. This, then, completes a connective path
from the Situation Compute to Goal Impact to Emotions. Study 3 confirmats this path.
Construals were mediated by Goal Impact in their influence upon emotions. Specifically,
construals of both threat and responsiveness were mediated by Inhibitive Goal Impact in the
evocation of both anger and fear.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 73
In addition to the connection from the Goal Impact Compute, the neural network layer
Emotions also received input directly from the construal layers. A few things in the network
predicted which emotion would be activated. Fear has been found to have a lower threshold than
other emotions, making organisms particularly sensitive to possible dangers in the environment,
and is believed to travel via a more direct route than other emotions (LeDoux, 2003). Thus, the
heirarchically lowest connection to the emotion layer comes from Threat Cue in the construal
input layer.
Both the direct connection and indirect path (through Goal Impact) between fear and
threat appraisals are confirmed in Study 3. In Study 3, participants were asked to report a current
and ongoing situation that was either going well or poorly (threat manipulation). There,
participants rated the magnitude of threat they perceived in the situation. This threat construal
was directly related to fear as well as indirectly related to fear through Inhibitive Goal Impact.
Once the emotion layer is activated in the net it sends its activations on to a layer called
the For-Behavior Compute which is another hidden layer that computes a pattern from
activations drawn throughout the network. Connected to this layer are the Emotion, Goal Impact
Compute, and Situation Compute layers. Essentially, the MABS model puts forth that it is the
activated schema, Goal Impact, and emotions that, in combination, elicit behavior. (Traits elicit
behavior to the extent they represent built up associations, or patterns, between all these
elements). As a hidden layer, the unique patterns generated by the For-Behavior Compute are an
emergent property of the network and not directly measured in the present studies with human
data. However, the behavior that is thereafter elicited is measured. The neural network includes
four behavior categories: proximity seeking, protesting, dismissing, and exploring. These
behaviors were measured in Studies 2 and 3. In Study 2, participants reported their own
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 74
behavior in response to their attachment figure’s behavior. Then they self-categorized their own
behavior into one of these four categories. The fear associated with the goals to Avoid Effort
and have Emotional intimacy, and the anger elicited from the motivation to Avoid Rejection
predicted protest behaviors. The goal to avoid rejection both directly, and indirectly through
emotion, predicted protest behaviors. Sadness, which was activated by anger, elicited dismissing
behaviors.
Interestingly, in this model, proximity seeking was only directly predicted by the
reduction of dismissing and protest behaviors, and through them, indireclty related to the
elements that preceded them. In Study 3, proximity seeking, protesting, and dismissing
behaviors represented a count of the number of sentences from each category that participants
selected to include in a letter to their attachment figure regarding an ongoing situation. Fear and
construals of responsiveness predicted proximity seeking. Anger and construals of
responsiveness predicted protest. These were the direct relationships. However, both behaviors
were also indirectly predicted by construals through Inhibitive Goal Impact, which evoked the
fear and anger that called the behaviors into action.
As alluded to previously, traits (attachment orientation) were not represented in the
neural network model as a distinct element but were proposed to arise from the patterns between
elements that, with experience, were made chronic. In the human data studies of this work, trait
attachment was measured using the ECR-R and, in a few studies, also manipulated. Therefore,
these measures are distinct variables in the structural equation models of the human data.
In Study 1, attachment anxiety and avoidance influenced Goal Impact, which mediated
their relationship to behavioral intentions to seek a relationship –or another relationship if the
participants was already in one. In Study 2, Anxiety increased motives to Avoid Rejection
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 75
(which indirectly through anger also increased protest behaviors and, through sadness,
dismissing behaviors). Avoidance reduced Emotional Intimacy motives (which indirectly
through fear increased protest behaviors). Primed security negatively predicted both motives to
Avoid Effort and to have Emotional Intimacy. It seems inducing a security prime satisfies these
needs to some extent. This reduces fear which reduces protest behaviors and, indirectly through
the reduction of protest behaviors increases proximity seeking which decreases dismissing
behaviors. All these indirect effects for the secure prime on behavior are signficant.
In Study 3, Model 1, Anxiety is directly related to threat construals, Goal Impact, and
fear. Anxiety also influences Goal Impact indirectly through construals. Through construals it
indirectly affects fear and anger. And, through all these elements, Anxiety indirectly predicts
protest behaviors. In Model 3.2, Avoidance increased the harm participants perceived in their
attachment figure’s behavior when an ongoing situation was going well for them and the other
person was, currently, being attentive and responsive. These construals of harm increased
sadness directly, as well as indirectly, through Inhibitive Goal Impact. Finally, sadness evoked a
reduction in proximity seeking. Avoidance was also directly related to a reduction in proximity
seeking in both models 3.1 and 3.2.
Finally, Study 4 takes MABS out of the realm of traditional attachment behaviors and
into a more neuanced situation to predict who pays on a date. First is taken the congruence of
different dating roles of suitor (pursuer) and receiver (pursued). These are combined with goal
importance to compute the Goal Impact of assuming (or violating) these roles and their norms.
Anxiety directly encourages women to pay the whole check on a date yet also indirectly
discourages her from doing so through MABS intervening processes. Anxiety increases the
perception that being the active purser would be bad for her most important goals. This
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 76
increases her fear (for her goals and signals inhibition) which decreases how often she splits the
check. Anxiety also works directly on happiness, reducing it, which reduces how often she treats
him to the date.
However, the indirect process results in the opposite behavior and insecure attachment
indirectly predicts women will uphold the social norms of dating roles and not pick up the check
when a violation of the norms would be poor for her goals. For men, only avoidance predicted
the perception that either dating role twarted his goals. When his goals were thwarted, negative
emotions increased. Anxiety, meanwhile, directly increased negative emotions. As negative
emotions increased, either directly from anxiety or indirectly through motivation from
avoidance, the frequency with which men picked up the whole check decreased and how often
they split the check increased.
Personality as Chronic Associations in the MABS Model.
The current work suggests a dynamic way to view the nature of a trait: as chronic
associations between construals, goals states (Read et al., 2010; Read & Miller, 2002), emotions,
and behaviors.
The present work illustrates the dynamic links between certain events, goals, emotions
(e.g. “he’s an angry person”) and behaviors. The neural network version of MABS shows that
the these dynamics are learned and with repeated experiences “build up” schemata (Talevich,
2012). The current work with human data speaks to the role of those schemata once they have
been built and personality formed. Essentially, it models the formation of personality: the
learned links between certain events and how they are construed, between some construals and
certain goals, and so on in the different patterns of goals, emotions, and behaviors (Talevich,
2012). For instance, for Gable (Gable, 2006) social approach goals and social rewards are
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 77
mediated by an exposure process whereby people who “get out there” accumulate more positive
experiences (Gable, 2006). The link between the BIS and social punishments is different. It is
mediated by a reactivity process whereby those with avoidance (BIS) social goals are more
sensitive to negative feedback and events – they react more strongly (Gable, 2006). (Attachment
avoidance should not be confused with the avoidance motivation system (BIS), for instance,
attachment avoidance is related to chronic under-reaction and anxiety to over-reaction).
Emotion in Behavioral System.
In their seminal work Emotion and Adaptation, Smith and Lazarus lay out their
cognitive-motivational-emotive system model (hereafter referred to as the Appraisal model).
The current work contributes to this work by suggesting a new operationalization of the primary
appraisal process.
In the Appraisal model, personality and the objective situation both influence situational
construals. The next step is an appraisal process by which the significance for personal well-
being of the construed situation is assessed. From the appraisal process comes an appraisal
outcome, which specifies particular action tendencies, affect, and physiological responses.
Action tendencies are translated into coping activities that can take one of two forms: emotion-
focused coping or problem-focused coping.
The primary appraisals of significance for personal well-being have two components: an
assessment of motivational relevance and the extent to which goals are thwarted or facilitated.
These assessments are often operationalized in an expectancy-value framework: for instance,
asking for the importance of a goal and the probability it will be attained. Probability estimates
are aimed at assessing the extent to which goals are thwarted or facilitated. However, unlike
expectancy-value, Goal Impact captures this underlying construct directly by asking for the
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 78
inhibition and facilitation of a goal and then multiplying that rating by importance. In so doing, I
argue that Goal Impact, a congruence-value product, is a more direct measure of the appraisal
process than an expectancy-value product, and as such, may be a psychologically more accurate
representation of the appraisal outcome described by Smith and Lazarus (1990).
There is another difference between the Motivated Affective Behavior System model and
the cognitive-motivational-emotive system. This appraisal model puts forth the primary
appraisal as necessary for emotion. In the MABS model, motivation does elicit emotion but is
not a necessary precondition. Keltner, Ellsworth, & Edwards suggest emotions can cause
appraisals (Keltner, Ellsworth, & Edwards, 1993). In the MABS model, the For-Behavior
Compute compiles the state of the network, including emotions. This is then fed back into the
network as an input representing its “previous inner state” - including what was most recently
felt. In this way, emotions can (indirectly) influence future appraisals. But, in the MABS model,
it is not the only way that emotions can influence the appraisal process.
In the MABS neural network, there is a direct connection from two construals: threat cue
and attachment figure responsiveness. Concentrating first on the threat cue, its connection to
emotion is intended to simulate the more direct route, lower perceptual threshold, and greater
speed for potentially dangerous stimuli (LeDoux, 2003). Thus, in MABS, fear can arise in
response to a bad situation for one’s goals (a poor Goal Impact) or when threat is present. The
results of Study 4 (Model 3.1) support these simulation findings with human data. Construed
threat directly predicts fear without “consulting” the appraisal outcome Goal Impact. However,
in the neural network, the threat cue is not connected to fear specifically, but to emotions
generally. The network is trained on situations in which threat cue and fear are always activated
together. So it is through learning that the network comes to associate the construal with the
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 79
specific emotion. Thus, the direct pairing of construals and emotions may not be unique to fear
and threat. Other construals can trigger a specific emotional response directly if they are
preconditioned, as fear is to threat cue, perhaps to the point an appraisal is no longer necessary.
Or, it could be there are other pairings vitally relevant to survival that, like threat cues and fear,
we are hardwired to quickly associate with one another. For instance, the MABS neural network
model also includes a direct link from responsiveness of the attachment figure to the emotion
layer to simulate down-regulation of emotion. The idea is, when there is a protector present,
happiness is directly activated and can act as a counter to fear – providing soothing at the speed
of fear. This relationship was not examined in these studies, because happiness is not related to
attachment behavior (but exploration), though future analyses may do so.
The evidence suggests that emotions, fear most prominently, can be independent of
motivation appraisals. This remains consistent with the concept of emotions as sophisticated
reflexes. Most stimuli are ambiguous. And, even when it is not, the motivational meaning of the
stimuli may be ambiguous. So in most cases, a motivational appraisal process is probably
necessary. But, it is reasonable to conclude that there may be some species-specific stimuli that
are pretty clear – perhaps from a lower species-specific pre-set threshold, or perhaps from
learning that has rendered an association practically automatic (e.g. as a part of personality or a
working model). It would be adaptive for nature to allow a direct relationship between a
construed event and emotion elicitation when the implications of the event are well known and a
fast response (or fast down-regulation of a response) is important to welfare. Stimulus checks
could determine whether the appropriate routing of information to the motivation system or to
emotions directly (Scherer et al., 2006).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 80
The brain is known to have more pathways that feed back to lower levels than feed
forward to higher levels. The hierarchy of the MABS neural network was built with this in
mind. Most of the layers of the MABS net are bidirectionally connected, including the emotion,
the motivation-subsystem layers, and the Situation Compute. What this means is that, in MABS,
emotion activations can travel backwards to provide feedback to influence “earlier” processes. If
Goal Impact is analogous to Appraisal Outcome, this implies emotions can provide feedback to,
and thereby, directly influence the appraisal process.
The ability to feed information backward, in combination with the ability to “skip” the
appraisal process and signal emotion straight from a construal, has interesting implications in
MABS net simulations that involve threat. In these simulations, the Threat Cue activates the
emotion and the Situation Compute (aka Schema) simultaneously. These recipients of the threat
cue are themselves bidirectionally connected. But the situation as a whole is complex. So while
the network has just begun to make sense of the situation, fear, already strongly activated, begins
to feed back to the Situation Compute. Soon, every active unit in the Situation Compute has
received a projection down from fear. By comparison, the Threat Cue (which initially projected
to both fear and the Situation Compute) has influenced only half of the pattern being generated
by the Situation Compute. What this means is that when the situation, as a whole, is at last
perceived, fear is twice as influential as the threat cue that elicited it in the first place. This has
interesting implications for coping. If emotions feedback to alter schema formation, then it can
bias the appraisal process. Some research suggests that fear acts on assessments of probability
and utility – an operationalization of the appraisal process. In their work, Marsella and Gratch
(2009) have found that fear induces wishful thinking, which acts on likelihood estimates, and
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 81
distancing (or rationalization) which operates on utility. The MABs model suggests the process
by which these types of coping may occur, as will be elucidated in the next section.
Contributions to Coping.
In their work on thought systems, McGuire & McGuire (1991) enumerated several
coping postulates that are formulated from the antecedents and consequences of an event. If an
event is goal achievement, then these postulates can be formulated in terms of the antecedents
and outcomes of goal achievement.
The Sufficient-Reason Postulate is a formula for the calculation of expectancy that, when
applied to goal pursuit, bears a striking resemblance to how we measure goal congruence: the
judged likelihood of achieving a goal will go up with the aspects of a situation that facilitate it,
and down with the aspects that inhibit it.
This rational calculation of expectancy is biased by the Wishful-Thinking Postulate,
which predicts that the importance of a goal will influence the judged likelihood it will be
achieved. To make a more gratifying likelihood adjustment, one need only increase the salience
of the facilitating aspects in a situation and make less salient those that inhibit the desired goal.
As described in the preceding section, these adjustments may be in response to feedback from
the emotion system. In the case of fear, a perceived threat may activate fear, from which
spreading activations both move up to behaviors and down to adjust Goal Impact. However,
MABS net simulations suggest it is feedback from emotion to influence perceptions of the
situation that then feed-forward to indirectly influence motivation. Of course, these motivational
adjustments could also be made in response to past activations recorded in the Inner State input
layer. In other words, one’s current feelings (about what has already occurred) are a part of the
situation, as a whole, and thereby influence the next motivational appraisal.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 82
Another postulate by the McGuires (1991) is the Rationalization Postulate by which
desire (value) is adjusted to be in line with expectations. When expectations are poor, a "sour
grapes" rationalization increases the number of undesirable outcomes that goal achievement is
perceived as promoting and the number of desirable outcomes it is perceived as preventing.
Negative emotions are undesirable outcomes and positive emotions desirable outcomes. When
expectations are poor (inhibiting forces are weightier), negative emotions are increased and
positive emotions are decreased, and the goal becomes less desirable. The reverse process,
“sweet lemons”, works in the same way to increase the desirability of a goal for which there are
many facilitating factors.
A future inquiry with the MAB simulations might look at the spreading of activations to
see if the dynamics of appraisal adjustments in response to emotions can be captured. It would
be interesting to know by which path (or paths), feeding forward or backwards, that motivations
are adjusted by emotions, and under what conditions. One might consider the implications for
emotion vs. problem-solving coping in a MABS framework. For instance, problem-solving
coping might be a feed forward process by which emotions elicit behaviors to act upon the
situation. But emotion-focused coping may be a feed-backward process whereby emotions alter
perceptions of the situation against which motivations are appraised.
Contributions to Attachment Theory.
In the last few decades attachment research has centered around the concept of the
working model: memory representations of people and relationships. But Bowlby’s focus was
on the regulatory function of the different behavioral systems, including attachment. In service
of this function is the “set goal” to maintain “felt security”. Bowlby’s theory of attachment was
a theory of motivation (Mikulincer & Shaver, 2007). The current work makes motivation central
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 83
to attachment by demonstrating the role of the motivation sub-system represented by Goal
Impact.
The biasing of the motivation sub-system described in the last section has interesting
implications for attachment theory. As noted in the introduction, researchers have found that
attachment anxiety is associated with an inability to disengage from goal pursuit – not only
attachment related goals but also those attachment-irrelevant such as investment strategies
(Jayson, 2004; Mikulincer & Shaver, 2001). Goal Impact, in light of the Wishful-Thinking
Postulate, suggests a mechanism for this process:
Attachment anxiety is associated with an intense desire for emotional security. Anxious
desires may enhance the salience of goal-facilitating aspects of a situation while down-playing
the salience of the goal-inhibiting aspects. If one perceives a goal to be more easily achieved
than not, then it is only rational to continue trying, thus explaining the failure to disengage from
goal pursuit long after achievement of the goal has become (realistically) unobtainable. This
process may become so rote that it is generalized to apply to goals in general.
Those with avoidant attachment prematurely disengage from goal pursuit and the
opposite-logic of the Wishful-Thinking Postulate is one possible mechanism by which this could
occur. The less desirable is pursuit, the more salient would be the goal-inhibiting aspects of a
situation and less salient the goal-facilitating aspects. No one is going to pursue an undesirable
goal but pursuit itself may be what is undesirable. Pursuit of a goal carries risk and requires the
expenditure of resources. The MABS neural network model and results of Study 2 implicate the
role of Avoiding Effort in attachment dynamics. It suggest that the desire to conserve resources
would make the pursuit of other goals less desirable. But this is rather indirect. Alternatively, or
in conjunction with this process may be another, the Rationalization Postulate described above.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 84
Mikulincer and Shaver suggest disengagement is a defensive maneuver by which Avoidants
prevent further damage to their sense of self-worth. This shuffling of goal importance, to be in
line with congruence, could only be captured by measuring an array of goals, their importance
and their congruence, and taking the aggregate. This is the process of computing Goal Impact.
Finally, MABS captures the dynamics by which motivations, congruencies, and their
product Goal Impact become chronic and, thereby, a building block of personality (Read &
Miller, 2002; Read & Monroe, 2009; Revelle & Scherer, 2009). We think of emotions as
momentary feelings and appraisal theory puts them forth as sophisticated reflexes informing us
of the demands of the current environment (Smith & Lazarus, 1990). So how is it that chronic
attachments become associated with emotions like fear and sadness? And how is it that
emotions are associated with traits e.g. a “cheerful” person or an “angry” person? Here
attachment theory lends insight to appraisal theory: If goal importance and congruence are
chronically adjusted to be in line with traits then the emotions that signal whether to engage or
disengage with those motives would be signaled as often as adjustments were made (Revelle &
Scherer, 2009). .
MABS considers the person as a whole, their personality and the dynamics to which it
gives rise, in the context of the situation (Mikulincer & Shaver, 2003). It was only by way of
these dynamics that Study 4 was able to generalize beyond the traditional attachment behavior
variables to predict who pays on a date. Researchers may be able to make more unexpected
generalizations by thinking not just in terms of attachment behaviors but attachment emotions
and motivations.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 85
Expanding Attachment Theory: Bowlby’s Behavioral Systems.
The predictive power of attachment theory has seen it grow into so many different
domains that it seems near ubiquitous. But these domains may be only indirectly related to the
attachment system.
For instance, the acquisition of money has been linked to avoidant attachment
(Mikulincer & Shaver, 2008). But if avoidant attachment is marked by disengagement of the
attachment system this implies the motive for money is not related to the attachment system
directly. According to Bowlby, goals for exploration and affiliation, sex, and caregiving do not
originate with the attachment system but issue from their own behavioral systems. Bowlby
envisioned these, and attachment, as unique and separate systems. A better suspect for money
motives would be the power system since, in our society, money is power.
Recent work by Mikulincer and Shaver has investigated Bowlby’s other proposed
behavioral systems (exploration, caregiving, and sexual) as well as a power system. They, and
colleagues, have developed four new Behavioral System Scales: Exploration (Doron, 2009),
Caregiving (Shaver, Mikulincer, & Shemesh-Iron, 2009), Power (Shaver et al., 2011), and the
Sexual Behavioral System (Shaver & Mikulincer, 2006). These scales are similar in flavor to
attachment questionnaires (Collins & Read, 1990; Fraley et al., 2000). Each scale has the
essential aim to determine the chronic state of the system: whether it is in chronic overdrive or
hyperactivated (e.g. attachment anxious, power-hungry, rapacious) or chronically deactivated
(e.g. attachment avoidant, obsequious, sexually disinterested). This reflects the belief that that
all of these behavioral systems share the same rules and dynamics (Mikulincer & Shaver, 2012).
If so, each system should operate with the intervening dynamics modeled by MABS. Each
system would have its unique set of relevant construals, motive sets, and behaviors. But these
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 86
would be linked by Goal Impact and emotion in unique patterns. Each model may prove as
expansive in its ability to predict and understand personality and behavior as the current studies.
But what is more, when all combined, these systems would make strides toward a comprehensive
theory of human motivation (Mikulincer & Shaver, 2012).
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 87
References
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A
psychological study of the strange situation. Hillsdale, NJ.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
personality and social psychology, 51(6), 1173.
Bartz, J. A., & Lydon, J. E. (2004). Close Relationships and the Working Self-Concept: Implicit
and Explicit Effects of Priming Attachment on Agency and Communion. Personality and
Social Psychology Bulletin, 30(11), 1389.
Bartz, J. A., & Lydon, J. E. (2008). Relationship-specific attachment, risk regulation, and
communal norm adherence in close relationships. Journal of Experimental Social
Psychology, 44(3), 655–663.
Beach, L. R., & Mitchell, T. R. (1987). Image theory: Principles, goals, and plans in decision
making. Acta Psychologica, 66(3), 201–220. doi:10.1016/0001-6918(87)90034-5
Bollen, K. A., & Stine, R. A. (1992). Bootstrapping Goodness-of-Fit Measures in Structural
Equation Models. Sociological Methods & Research, 21(2), 205–229.
doi:10.1177/0049124192021002004
Bowlby, J. (1969). Attachment and loss: Attachment (vol. 1). New York: Basic Books.
Brown, R. L. (1997). Assessing specific mediational effects in complex theoretical models.
Structural Equation Modeling: A Multidisciplinary Journal, 4(2), 142–156.
doi:10.1080/10705519709540067
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 88
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The affect system has parallel and
integrative processing components: Form follows function. Journal of Personality and
Social Psychology, 76(5), 839.
Cassidy, J., & Kobak, R. R. (1988). Avoidance and its relation to other defensive processes.
Clinical implications of attachment, 300–323.
Chulef, A. S., Read, S. J., & Walsh, D. A. (2001). A Hierarchical Taxonomy of Human Goals.
Motivation and Emotion, 25(3), 191–232. doi:10.1023/A:1012225223418
Collins, N. L., & Allard, L. M. (2004). Cognitive representations of attachment: The content and
function of working models. Social cognition, 75–101.
Collins, N. L., & Read, S. J. (1990). Adult attachment, working models, and relationship quality
in dating couples. Journal of personality and social psychology, 58(4), 644.
Depue, R. A., & Collins, P. F. (1999). Neurobiology of the structure of personality: Dopamine,
facilitation of incentive motivation, and extraversion. Behavioral and Brain Sciences,
22(03), 491–517.
Doron, Y. (2009). Conceptualization and operationalization of individual differences in the
functioning of the exploration behavioral system (Unpublished Doctoral Dissertation).
Bar-Ilan University, Ramat-Gan, Israel.
Downey, G., & Feldman, S. I. (1996). Implications of rejection sensitivity for intimate
relationships. Journal of Personality and Social Psychology, 70(6), 1327–1343.
Downey, G., Mougios, V., Ayduk, O., London, B. E., & Shoda, Y. (2004). Rejection sensitivity
and the defensive motivational system. PSYCHOLOGICAL SCIENCE-CAMBRIDGE-,
15, 668–673.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 89
Downey, Geraldine, Romero-Canyas, R., Reddy, K. S., & Rodrigues, S. (2012, January 27).
After all I have done for you: Relational accommodation fuels post-rejection hostility.
Presented at the SPSP, San Diego, Ca.
Ferguson, M. J., & Bargh, J. A. (2004). Liking Is for Doing: The Effects of Goal Pursuit on
Automatic Evaluation. Journal of Personality and Social Psychology. Vol. 87(5), 87(5),
557–572.
Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-
report measures of adult attachment. Journal of Personality and Social Psychology,
78(2), 350–365.
Gable, S. L. (2006). Approach and Avoidance Social Motives and Goals. Journal of Personality,
74(1), 175–222.
Gable, Shelly L., & Strachman, A. (2008). Approaching social rewards and avoiding social
punishments: Appetitive and aversive social motivation. In J. Y. Shah & W. L. Gardner
(Eds.), Handbook of motivation science. Guilford Press.
Gillath, O., Mikulincer, M., Fitzsimons, G. M., Shaver, P. R., Schachner, D. A., & Bargh, J. A.
(2006). Automatic Activation of Attachment-Related Goals. Personality and Social
Psychology Bulletin, 32(10), 1375.
Gray, J. A. (1991). The neuropsychology of temperament.
Gray, J. A. (2003). The neuropsychology of anxiety: An enquiry into the functions of the septo-
hippocampal system. Oxford Psychology Series (2nd ed.). Great Clarendon Street,
Oxford OX2 6DP: Oxford University Press, UK. Retrieved from
http://stoa.usp.br/vahs/files/-1/16169/Gray%20e%20McNaughton%20-
%20Neuropsychology%20of%20Anxiety.pdf
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 90
Hancock, G. R., & Mueller, R. O. (2006). Non-Normal and Categorical Data in Structural
Equation Modeling. Structural Equation Modeling: A Second Course. IAP.
Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new
millennium. Communication Monographs, 76(4), 408–420.
Ievers-Landis, C. E., Burant, C. J., & Hazen, R. (2011). The Concept of Bootstrapping of
Structural Equation Models with Smaller Samples: An Illustration Using Mealtime
Rituals in Diabetes Management. Journal of Developmental & Behavioral Pediatrics,
32(8), 619–626. doi:10.1097/DBP.0b013e31822bc74f
Jayson, J. (2004). Attachment Theory and Escalation of Commitment: An Evaluation of
Prototypical Goal Engagement Or Disengagement in an Organizational Context
(Unpublished doctoral dissertation). Bar-Ilan University, Ramat-Gan, Israel.
Keltner, D., Ellsworth, P. C., & Edwards, K. (1993). Beyond simple pessimism: Effects of
sadness and anger on social perception. Journal of Personality and Social Psychology,
64, 740–752. doi:10.1037/0022-3514.64.5.740
Kilpatrick, D. G., Resnick, H. S., Ruggiero, K. J., Conoscenti, L. M., & McCauley, J. (2007).
Drug-facilitated, incapacitated, and forcible rape: A national study ( No. 219181) (p.
72). Final report submitted to the U.S. Department of Justice. Retrieved from
http://www.mincava.umn.edu/redirect/http://www.ncjrs.gov/pdffiles1/nij/grants/219181.p
df
LeDoux, J. E. (2003, November 28). Emotion Circuits in the Brain. review-article. Retrieved
February 13, 2009, from
http://arjournals.annualreviews.org.libproxy.usc.edu/doi/abs/10.1146%2Fannurev.neuro.2
3.1.155
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 91
Lewin, K. (1939). Field theory and experiment in social psychology: Concepts and methods.
American Journal of Sociology, 868–896.
London, B., Downey, G., Romero-Canyas, R., Rattan, A., & Tyson, D. (2012). Gender-based
rejection sensitivity and academic self-silencing in women. Journal of Personality and
Social Psychology, 102(5), 961–979. doi:10.1037/a0026615
Main, M. (1990). Cross-cultural studies of attachment organization: Recent studies, changing
methodologies, and the concept of conditional strategies. Human development. Retrieved
from http://psycnet.apa.org/psycinfo/1990-14461-001
Marsella, S., & Gratch, J. (2009). EMA: A computational model of appraisal dynamics.
Mathieu, J. E., & Taylor, S. R. (2006). Clarifying conditions and decision points for mediational
type inferences in organizational behavior. Journal of Organizational Behavior, 27(8),
1031–1056.
McGuire, W. J., & McGuire, C. V. (1991). The content, structure, and operation of thought
systems. In R. S. Wyer & T. K. Srull (Eds.), Advances in Social Cognition (Vol. IV, p. 1–
). Hillsdale, N.J.: Lawrence Erlbaum Associates, Publishers.
Mikulincer, M, & Shaver, P. R. (2001). Attachment theory and intergroup bias: evidence that
priming the secure base schema attenuates negative reactions to out-groups. Journal of
Personality and Social Psychology, 81(1), 97–115.
Mikulincer, M., & Shaver, P. R. (2003). The Attachment Behavioral System in Adulthood:
Activation, Psychodynamics, and Interpersonal Processes. ADVANCES IN
EXPERIMENTAL SOCIAL PSYCHOLOGY, 35, 56–152.
Mikulincer, M., & Shaver, P. R. (2007). Contributions of attachment theory and research to
motivation science. Handbook of Motivation Science. New York: Guilford.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 92
Mikulincer, M., & Shaver, P. R. (2012). Attachment Theory Expanded. The Oxford Handbook of
Personality and Social Psychology, 467.
Mikulincer, M., Shaver, P. R., & Pereg, D. (2003). Attachment theory and affect regulation: The
dynamics, development, and cognitive consequences of attachment-related strategies.
Motivation and emotion, 27(2), 77–102.
Mikulincer, M, & Shaver, P. R. (2008). “Can’t Buy Me Love”: An Attachment Perspective on
Social Support and Money as Psychological Buffers. Psychological Inquiry, 19(3-4),
167–173. doi:10.1080/10478400802631295
Mikulincer, M, Shaver, P. R., Gillath, O., & Nitzberg, R. A. (2005). Attachment, Caregiving, and
Altruism: Boosting Attachment Security Increases Compassion and Helping. Journal of
Personality and Social Psychology, 89(5), 817–839. doi:10.1037/0022-3514.89.5.817
Miller, L.C., & Fishkin, S. A. (1997). On the dynamics of human bonding and reproductive
success: Seeking windows on the adapted-for human–environmental interface.
Evolutionary social psychology. Lawrence Erlbaum Associates.
Miller, Lynn Carol, Pedersen, W. C., & Putcha-Bhagavatula, A. (2005). Promiscuity in an
Evolved Pair-Bonding System: Mating Within and Outside the Pleistocene Box.
Behavioral and Brain Sciences, 28(02), 290–291. doi:10.1017/S0140525X05370051
O’Reilly, R. C., & Munakata, Y. (2000). Computational explorations in cognitive neuroscience:
Understanding the mind by simulating the brain. The MIT Press.
Ovsiankina, M. (1928). Investigations on the psychology of action and affection: The resumption
of interrupted acts.(K. Lewin, Ed). Psychologische Forschung, 11, 302–389.
Plutchik, R. (1980). Emotion, a psychoevolutionary synthesis. Harper & Row New York.
Retrieved from http://www.getcited.org/pub/101971132
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 93
Read, S. J., & Miller, L. C. (2002). Virtual Personalities: A Neural Network Model of
Personality. Personality and Social Psychology Review, 6(4), 357.
Read, S. J., & Monroe, B. M. (2009). Using Connectionist Networks to Understand
Neurobiological Processes in Social and Personality Psychology. In E. Harmon-Jones &
J. Beers (Eds.), Methods in Social Neuroscience. New York: Guilford Press.
Read, S. J., Monroe, B. M., Brownstein, A. L., Yang, Y., Chopra, G., & Miller, L. C. (2010). A
neural network model of the structure and dynamics of human personality. Psychological
review, 117(1), 61–92.
Revelle, W., & Scherer, K. R. (2009). Personality and emotion. Oxford companion to emotion
and the affective sciences, 303–305.
Scherer, K. R., Dan, E. S., & Flykt, A. (2006). What determines a feeling’s position in affective
space? A case for appraisal. Cognition and Emotion, 20(1), 92.
Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment &
Human Development, 4(2), 133–161.
Shaver, P. R., & Mikulincer, M. (2006). Conceptualization and operationalization of individual
differences in the functioning of the sexual behavioral system. Presented at the
International Association for Relationship Research Biannual Conference, Crete, Greece.
Shaver, P. R., Mikulincer, M., & Shemesh-Iron, M. (2009). A behavioral systems perspective on
prosocial behavior. Prosocial motives, emotions, and behavior: The better angels of our
nature, 73–92.
Shaver, P. R., Segev, M., & Mikulincer, M. (2011). A Behavioral Systems Perspective on Power
and Aggression. Human aggression and violence Causes. Retrieved from
http://portal.idc.ac.il/en/Symposium/HSPSP/2009/Documents/ShaverC.pdf
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 94
Smith, C. A., & Lazarus, R. S. (1990). Emotion and adaptation. (L. A. Pervin, Ed.), 609–637.
Talevich, J.R., Read, S. J., Walsh, D. A., Iyer, R., & Chopra, G. (2012). Toward a
Comprehensive, Empirically-based Taxonomy of Human Goals. Journal of Personality
and Social Psychology, under review.
Talevich, J.R., Saks, E. R., & Read, S. J. (2012). Encouragement or Coercion: Medical
adherence as a function of how psychotropic medication, and pressure to take it, impacts
life goals. Journal of Mental Health Law and Policy, under review.
Talevich, J.R., Walsh, D. A., & Read, S. J. (2012). Goal Impact: A Goal Systems, Domain
General Prediction Tool, Applied to Voluntary Job Turnover. Journal of Behavioral
Decision Making, (under review).
Talevich, J.R. (2012, March). A Behavioral Systems Approach Toward a Unified Theory of
Human Motivation: A computational model of attachment theory dynamics (Qualifying
Exam). University of Southern California, Los Angeles.
Walsh, D. A., Talevich, J. R., Read, S. J., Lee, Lee, & Larsen. (2012). Goal Impact as a
predictor of weight management. manuscript in progress, University of Southern
California.
Watson, D., & Clark, L. A. (1994). The PANAS-X: Manual for the Positive and Negative Affect
Schedule-Expanded Form. Unpublished manuscript, University of Iowa.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 95
Tables
Table 1
Overview of Studies
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 96
Table 2
Study 1 Goal Impact Aggregate Composites
Table 3
Model 1 Regression Model Summaries
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 97
Table 4
Model 1 Regression Coefficients
Table 5
Model 1 Partial Mediation
Table 6
Model 1 Indices of Approximate Fit
CMIN
Model NPAR CMIN DF P CMIN/
Default model 42 0.32 3 0.96 0.11
Saturated model 45 0 0
Independence model 15 426.32 30 0 14.21
RMR, GFI
Model RMR GFI AGFI PGFI
Default model 62.42 1 1 0.07
Saturated model 0 1
Independence model 1395.23 0.8 0.7 0.53
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 98
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 1 0.99 1.01 1.07 1
Saturated model 1 1 1
Independence model 0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.1 0.1 0.1
Saturated model 0 0 0
Independence model 1 0 0
NCP
Model NCP LO 90 HI 90
Default model 0 0 0
Saturated model 0 0 0
Independence model 396.32 333.28 466.8
FMIN
Model FMIN F0 LO HI 90
Default model 0 0 0 0
Saturated model 0 0 0 0
Independence model 0.56 0.52 0.44 0.61
RMSEA
Model RMSEA LO 90 HI 90 PCLOS
Default model 0 0 0 1
Independence model 0.13 0.12 0.14 0
AIC
Model AIC BCC BIC CAIC
Default model 84.32 86.74
Saturated model 90 92.59
Independence model 456.32 457.18
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 0.11 0.11 0.11 0.11
Saturated model 0.12 0.12 0.12 0.12
Independence model 0.6 0.52 0.69 0.6
HOELTER
Model
HOELTE HOELTE
0.05 0.01
Default model 18485 26834
Independence model 81 93
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 99
Table 7
Model 2 Regression Coefficients
Table 8
Model 2 Hypothesized Mediation Results
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 100
Table 9
Model 2 Model Fit Summaries
CMIN
Model NPAR CMIN DF P CMIN/
Default model 30 45.331 48 0.583 0.944
Saturated model 78 0 0
Independence model 12 248.422 66 0 3.764
RMR, GFI
Model RMR GFI AGFI PGFI
Default model 0.054 0.93 0.886 0.572
Saturated model 0 1
Independence model 0.132 0.732 0.684 0.62
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 0.818 0.749 1.013 1.02 1
Saturated model 1 1 1
Independence model 0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.727 0.595 0.727
Saturated model 0 0 0
Independence model 1 0 0
NCP
Model NCP LO 90 HI 90
Default model 0 0 16.913
Saturated model 0 0 0
Independence model 182.422 137.875 234.548
FMIN
Model FMIN F0 LO 90 HI 90
Default model 0.458 0 0 0.171
Saturated model 0 0 0 0
Independence model 2.509 1.843 1.393 2.369
RMSEA
Model RMSEA LO 90 HI 90 PCLOS
Default model 0 0 0.06 0.896
Independence model 0.167 0.145 0.189 0
AIC
Model AIC BCC BIC CAIC
Default model 105.331 114.401 183.486 213.486
Saturated model 156 179.581 359.203 437.203
Independence model 272.422 276.05 303.684 315.684
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 101
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 1.064 1.091 1.262 1.156
Saturated model 1.576 1.576 1.576 1.814
Independence model 2.752 2.302 3.278 2.788
HOELTER
Model
HOELTE HOELTE
0.05 0.01
Default model 143 161
Independence model 35 39
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 102
Table 10
Study 3 Letter Sentence Choices
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 103
Table 11
Study 3 Descriptive Statistics
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 104
Table 12
Study 3 Situation vs. Trait Variance Explained.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 105
Table 13
Model 3.1 Regression Coefficients
Table 14
Model 3.1 Indirect Effects & Hypotheses
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 106
Table 15
Model 3.1 Indices of Model Fit
CMIN
Model NPAR CMIN DF P
CMIN/
DF
Default model 40 39.4 37 0.363 1.065
Saturated model 77 0 0
Independence model 22 584.861 55 0 10.634
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 0.933 0.9 0.996 0.993 0.995
Saturated model 1 1 1
Independence model 0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.673 0.627 0.67
Saturated model 0 0 0
Independence model 1 0 0
NCP
Model NCP LO 90 HI 90
Default model 2.4 0
21.72
1
Saturated model 0 0 0
Independence model 529.861 455.882
611.2
92
FMIN
Model FMIN F0
LO
90
HI 90
Default model 0.218 0.013 0 0.12
Saturated model 0 0 0 0
Independence model 3.231 2.927 2.519 3.377
RMSEA
Model RMSEA LO 90 HI 90
PCLOS
E
Default model 0.019 0 0.057 0.895
Independence model 0.231 0.214 0.248 0
AIC
Model AIC BCC BIC CAIC
Default model 119.4 125.081
Saturated model 154 164.935
Independence model 628.861 631.985
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 107
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 0.66 0.646 0.766 0.691
Saturated model 0.851 0.851 0.851 0.911
Independence model 3.474 3.066 3.924 3.492
HOELTER
Model
HOELTE
R
HOELTE
R
0.05 0.01
Default model 240 276
Independence model 23 26
Table 16
Model 3.2 Regression Coefficients
Table 17
Model 3.2 Indirect Effects & Hypotheses in the Secure Situation
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 108
Table 18
Model 3.2 Indices of Model Fit
CMIN
Model NPAR CMIN DF P
CMIN/
DF
Default model 19 11.886 8 0.156 1.486
Saturated model 27 0 0
Independence
model
12 78.863 15 0 5.258
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 0.849 0.717 0.945 0.886 0.939
Saturated model 1 1 1
Independence
model
0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.533 0.453 0.501
Saturated model 0 0 0
Independence
model
1 0 0
NCP
Model NCP LO 90 HI 90
Default model 3.886 0
17.29
7
Saturated model 0 0 0
Independence
model
63.863 39.742
95.50
7
FMIN
Model FMIN F0
LO
90
HI 90
Default model 0.475 0.155 0 0.692
Saturated model 0 0 0 0
Independence
model
3.155 2.555 1.59 3.82
RMSEA
Model RMSEA LO 90 HI 90 PCLOSE
Default model 0.139 0 0.294 0.191
Independence
model
0.413 0.326 0.505 0
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 109
AIC
Model AIC BCC BIC CAIC
Default model 49.886 64.663
Saturated model 54 75
Independence
model
102.863 112.197
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 1.995 1.84 2.532 2.587
Saturated model 2.16 2.16 2.16 3
Independence
model
4.115 3.15 5.38 4.488
HOELTER
Model
HOELTER HOELTER
0.05 0.01
Default model 33 43
Independence
model
8 10
Table 19
Study 4 Who Pays Composite Descriptive Statistics
Table 20:
Study 4 Goal Impact Composite Descriptive Statistics
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 110
Table 21
Model 4.1 Women's Regression Coefficients
Table 22
Model 4.2 Men's Regression Coefficients
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 111
Table 23
Model 4.1 Hypothesized Indirect Effects for Women
Table 24
Model 4.2 Hypothesized Indirect Effects for Men
Table 25
Model 4.1 Fit Summaries (Women)
CMIN
Model NPAR CMIN DF P CMIN/
Default model 25 18.415 20 0.56 0.921
Saturated model 45 0 0
Independence model 9 165.681 36 0 4.602
RMR, GFI
Model RMR GFI AGFI PGFI
Default model 0.66 0.952 0.891 0.423
Saturated model 0 1
Independence model 3.836 0.63 0.537 0.504
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 112
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 0.889 0.8 1.011 1.022 1
Saturated model 1 1 1
Independence model 0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.556 0.494 0.556
Saturated model 0 0 0
Independence model 1 0 0
NCP
Model NCP LO 90 HI 90
Default model 0 0 12.39
Saturated model 0 0 0
Independence model 129.681 93.41 173.4
FMIN
Model FMIN F0 LO HI 90
Default model 0.222 0 0 0.149
Saturated model 0 0 0 0
Independence model 1.996 1.562 1.125 2.09
RMSEA
Model RMSEA LO 90 HI 90 PCLOS
Default model 0 0 0.086 0.766
Independence model 0.208 0.177 0.241 0
AIC
Model AIC BCC BIC CAIC
Default model 68.415 75.264 129.186 154.186
Saturated model 90 102.329 199.387 244.387
Independence model 183.681 186.146 205.558 214.558
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 0.824 0.843 0.993 0.907
Saturated model 1.084 1.084 1.084 1.233
Independence model 2.213 1.776 2.741 2.243
HOELTER
Model
HOELTE HOELTE
0.05 0.01
Default model 142 170
Independence model 26 30
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 113
Table 26
Model 4.2 Fit Summaries (Men)
CMIN
Model NPAR CMIN DF P CMIN/
Default model 27 16.74 18 0.54 0.93
Saturated model 45 0 0
Independence model 9 222.5 36 0 6.18
RMR, GFI
Model RMR GFI AGFI PGFI
Default model 0.71 0.98 0.94 0.39
Saturated model 0 1
Independence model 3.56 0.74 0.67 0.59
Baseline Comparisons
Model NFI RFI IFI TLI CFI
Default model 0.92 0.85 1.01 1.01 1
Saturated model 1 1 1
Independence model 0 0 0 0 0
Parsimony-Adjusted Measures
Model PRATIO PNFI PCFI
Default model 0.5 0.46 0.5
Saturated model 0 0 0
Independence model 1 0 0
NCP
Model NCP LO 90 HI 90
Default model 0 0 12.32
Saturated model 0 0 0
Independence model 186.5 143.15 237.3
FMIN
Model FMIN F0 LO HI 90
Default model 0.11 0 0 0.08
Saturated model 0 0 0 0
Independence model 1.49 1.25 0.96 1.59
RMSEA
Model RMSEA LO 90 HI 90 PCLOS
Default model 0 0 0.07 0.85
Independence model 0.19 0.16 0.21 0
AIC
Model AIC BCC BIC CAIC
Default model 70.74 74.62 152.03 179.03
Saturated model 90 96.47 225.48 270.48
Independence model 240.5 241.79 267.59 276.59
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 114
ECVI
Model ECVI LO 90 HI 90 MECVI
Default model 0.47 0.48 0.57 0.5
Saturated model 0.6 0.6 0.6 0.65
Independence model 1.61 1.32 1.96 1.62
HOELTER
Model
HOELTE HOELTE
0.05 0.01
Default model 257 310
Independence model 35 40
Figures
Figure 1: The Neural Network Model of the Affective Behavior System (MAB). This drawing represents the
connections and layers of the connectionist computer model. Lines without arrows indicate bidirectional
connectivity.
MOTIVA
Figure 2: A
intervening
elements b
motivation
these elem
Figure 3: A
relationship
equation.
ATED AFFE
A simplified de
g relationships
before it to thos
n, and emotion,
ments traits are
Alternative Inte
p. Dashed line
.
ECTIVE BE
epiction of the
between layer
se that follow.
, to influence b
hypothesized t
ervening Mode
e indicates tota
EHAVIOR SY
Motivated Aff
rs (or elements)
The objective
behavior. Beca
to directly influ
els from Mathie
al significant ef
YSTEM
fective Behavio
) in the model.
situation and t
ause traits are c
uence any elem
eu & Taylor (2
ffect that becom
oral System ne
. Each elemen
traits are partia
conceived of as
ment in the pro
2006). Unbrok
mes insignifica
eural network m
nt carries indire
ally mediated b
s learned assoc
ocess.
ken lines indica
ant when M is
model illustrati
ect effects from
by construals,
ciations betwee
ate a significan
entered into th
115
ing the
m those
en
nt
he
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 116
Figure 4: The motivation sub-system of the MABS computational model to be validated by Study 1. Greyed-out
elements are “hidden layers” in the network that are not measured directly in this study.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 117
Figure 5: Structural Model 1 showing that Goal Impact partially mediates the association between traits (e.g chronic
attachment avoidance and anxiety) and intentions to seek a relationship (singles) or seek a different relationship if
already in one (coupled individuals). The fit of this model was superb: χ
2
=0.32, df=3, p=.96 and Bollen-Stine
bootstrap p=.97 indicating the models does not significantly diverge from the data. Path estimates are standardized
(b) regression coefficients. For approximate fit indices see Table 6. Numbers top-right of DVs indicate how much
variance is accounted (r
2
smc
) for by its predictors. *p <.05. **p<.01, ***p<=.001, †=marginally significant, ns=not
significant.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 118
Figure 6: Study 2 elements to be validated in the MABS model. Greyed-out elements are “hidden layers” in the
network that are not measured directly in this study. Dashed lines indicate effects that are mediated by elements in
the MABS model that are not depicted because they are not tested in the current Study.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 119
Figure 7: Study 2 Structural equation model. All tests indicate exceptional fit. χ
2
=45 , df=48 p=.583 and Bollen-
Stine bootstrap p = . 583. Numbers top-right of DVs indicate how much variance is accounted (r
2
smc
) for by its
predictors. Refer to Table 9 for approximate fit indices.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 120
Figure 8: Study 3 elements to be validated in the MABS model. Greyed-out elements are “hidden layers” in the
network that are not measured directly in this study.
Figure 9: Model 3.1 validation of the full MABS model. All Conditions are included. N=182. Χ
2
= 39, df=37, p=.3.
Bollen-Stine bootstrap p = .5. Numbers top-right of DVs indicate how much variance is accounted (r
2
smc
) for by its
predictors. Approximate fit indices are reported in Table 15.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 121
Figure 10: Model 3.2. Secure Situation included participants from the Responsive-No Threat condition. Χ
2
= 12,
df=8 p= .156, Bollen-Stine bootstrap p = .319, N=26. Numbers top-right of DVs indicate how much variance is
accounted (r
2
smc
) for by its predictors. Approximate fit indices are reported in Table 18.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 122
Figure 11: Study 4 elements to be validated in the MABS model. Greyed-out elements are “hidden layers” in the
network that are not measured directly in this study.
Figure 12: Studies 2 & 3 Q-Sort Distributions. The new Q-Sort for Study 3 overlaid the original Q-Sort distribution
(in grey)
MOTIVA
Fi
ATED AFFE
igure 13: Study
0
10
20
30
40
50
60
70
80
90
100
ECTIVE BE
y 4 Participant
18 ‐24
EHAVIOR SY
Gender
25 ‐29
YSTEM
30+
Women
Men
n
123
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 124
Figure 14: Model 4.1 When women pay on a date. Χ2 = .18, df=20, p=.56 and Bollen-Stine bootstrap p = 67.
Numbers top-right of DVs indicate how much variance is accounted (r2smc) for by its predictors. Approximate fit
Indices are reported in Table 25.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 125
Figure 15: When men don’t pay on a date. Χ2 = 17., df=18, p=.54, Bollen-Stine bootstrap p = 66. Numbers top-
right of DVs indicate how much variance is accounted (r2smc) for by its predictors. Approximate fit Indices are
reported in Table 26.
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 126
Appendix: Attachment Sentence Generation Pilot
The sentences selected from the results of this pilot can be found in Table 10. Italics indicate
actual wording presented to participants.
Introduction.
We will ask you what you would have said, or would have liked to say, to three different
people all in a similar type of circumstance. The circumstance will be one in which
you're experiencing a problem. Any problem in which you faced a threat, a great deal of
stress, some fear, or pain. We will ask you for 7 different statements addressed to that
person. It may help you to think of it as an email or phone call 7 sentences long. We will
describe a relationship and ask you to think of a person who fits that description.
Descriptions.
All 3 are presented to each subject, in random order.
Please think about an important and meaningful relationship (e.g. parent, romantic
partner) in which you……have found it easy to be emotionally close to this person. In this
relationship, you felt comfortable depending on this person and having them depend on
you. You didn't particularly worry about being alone. You felt accepted by this person.
(Secure Description, Bartz & Lydon, 2008)
… felt like you wanted to be completely emotionally intimate with this person but felt that
this person was reluctant to get as emotionally close as you would have liked. In this
relationship you felt uncomfortable being alone and worried that this person didn’t value
you as much as you valued them. (Anxious Description, Bartz & Lydon, 2008)
… felt comfortable not being emotionally close to this person. In this relationship you
felt that it was very important to be independent and self-sufficient and you preferred not
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 127
to depend on this person or have this person depend on you. By what name do you call
this person? (e.g. Dad, Kelley). (Avoidant Description, Bartz & Lydon, 2008)
Name Capture.
By what name do you call this person? (e.g. Dad, Kelley). We only ask so we may refer
to him/her by name in the next few questions.
On the next screen, the name provided above is inserted where [AF] indicates below:
Take a moment and try to get a visual image in your mind of [AF]. Imagine what [AF]
looks like. What is it like being with [AF]? How do you feel when you are together?
Sentence Generation.
Now, imagine a time in your relationship with [AF] that you were having a problem. It is
a time when you need help, support, or comforting from [AF]
Imagine it is before the situation was resolved. What might you have said to [AF] to get
what you need in this moment? (Proximity -follows Secure Prime)
… but you are not getting it! This makes you angry or frustrated in addition to the fear or
anxiety you feel about your situation. Imagine it is before the situation was resolved.
What might you have said to [AF] in anger because you are afraid and need help in this
moment? Maybe you said it, maybe you just thought it, but what would you have wished
or felt the urge to say? It's okay to act out or be a little dramatic as emotions are running
high. (Protest -follows Anxious Prime)
… The sad thing was, [AF] couldn't be relied upon. You were going to have to go it
alone. Imagine it is before the situation was resolved. What would you have said to
[AF] in this moment so that you could take care of yourself and the situation?
(Dismissing –follows Avoidant Prime)
MOTIVATED AFFECTIVE BEHAVIOR SYSTEM 128
Text Boxes.
Please address [AF], such as “[AF], I ...." or "You...?" Remember, it may help you to
write this as if it were one email or phone call that is 7 sentences long.
1) ________________________________________________________________________
2) ________________________________________________________________________
3) ________________________________________________________________________
4) ________________________________________________________________________
5) ________________________________________________________________________
6) ________________________________________________________________________
7) ________________________________________________________________________
Abstract (if available)
Abstract
A model of motivated affective behavior system (MABS) is reviewed and validated. Four human-data studies test a computational model (Talevich, 2012) that integrates attachment theory, several models of emotion processing, and goal systems theory. Previous work on attachment models the dynamic experiences that serve to activate and deactivate the attachment system (Mikulincer & Shaver, 2003
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Real to me: Effecting real-world change from a virtual environment through an interaction between presence and immersive tendencies
PDF
The interpersonal effect of emotion in decision-making and social dilemmas
PDF
Context dependent utility: an appraisal-based approach to modeling context, framing, and decisions
PDF
The human element: addressing human adversaries in security domains
PDF
Powerful guts: how power limits the role of disgust in moral judgment
PDF
Adding complexity to the gratitude-affect relationship: deeper impersonal gratitude leads to more contentment
PDF
Reframing parental psychological control: the role of insecure attachment
PDF
Protecting networks against diffusive attacks: game-theoretic resource allocation for contagion mitigation
PDF
Dynamic analysis and control of one-dimensional distributed parameter systems
PDF
Reaching decisions in dynamic environments
PDF
Dynamic modeling and simulations of rigid-flexible coupled systems using quaternion dynamics
PDF
Human adversaries in security games: integrating models of bounded rationality and fast algorithms
PDF
The cost of a poker-face: consequences of self-regulation on emotion recognition and interpersonal perception
PDF
Justification of selfishness: projection as a mechanism of justification
PDF
Bayesian hierarchical and joint modeling of the reversal learning task
PDF
“You move, therefore I am”: the combinatorial impact of kinesthetic motion cues on social perception
PDF
Behavioral signal processing: computational approaches for modeling and quantifying interaction dynamics in dyadic human interactions
PDF
A meditative application inspired by emotional regulation
PDF
System dynamics of in-home family eating behavior: insights from intensive longitudinal data using Ecological Momentary Assessment and wearable sensors
PDF
Black sheep or devil protection?: the effects of cooperative and competitive intergroup goal structures on group members reactions to an offensive act when committed by an ingroup or an outgroup ...
Asset Metadata
Creator
Talevich, Jennifer Rose
(author)
Core Title
The motivated affective behavior system: a dynamic account of the attachment behavioral system
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Publication Date
10/03/2012
Defense Date
09/04/2012
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
affect,attachment,behavioral system,close relationships,construals,Courtship,dating,emotion,Goal Impact,goals,Infidelity,motivated behavior,Motivation,OAI-PMH Harvest,relationship dissolution,schema,working models
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Read, Stephen J. (
committee chair
), Marsella, Stacy C. (
committee member
), Monterosso, John R. (
committee member
), Tambe, Milind (
committee member
), Walsh, David A. (
committee member
)
Creator Email
talevich@usc.edu,uzebra@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-99481
Unique identifier
UC11290284
Identifier
usctheses-c3-99481 (legacy record id)
Legacy Identifier
etd-TalevichJe-1222.pdf
Dmrecord
99481
Document Type
Dissertation
Rights
Talevich, Jennifer Rose
Type
texts
Source
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 a...
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
Tags
affect
behavioral system
close relationships
construals
Goal Impact
goals
motivated behavior
relationship dissolution
schema
working models