Abstract
Nursing has a long history of attending to the importance of early attachment experiences to later development. Attachment strategies formed in infancy and early childhood can have lifelong effects on an individual’s behavior and health. Advances in neuroimaging technology allow us to understand how these early experiences map onto the structure and function of the brain and ultimately behavior and health. Previous reviews have discussed the findings of studies observing correlations between attachment strategy and neural function and structure in romantic partners and parents, but far less has been said about nonparenting adults. This article reviews the relationship between attachment strategies developed in childhood and brain structure and function in nonparenting adults. A total of 14 studies met inclusion criteria. Results showed adult attachment patterns of nonparenting adults are pervasively correlated with brain structure and function, with most associations observed in executive regions, followed by affective and reward processing. Notably, no studies found associations between attachment pattern and stress response, in contrast with studies of mothers. These brain regions are linked to the many behavioral, mood and substance abuse disorders observed in adults with insecure attachment patterns. Nurses can use these findings to help prevent, assess and address these health risks in nonparenting adults, as well as provide the brain-based evidence to support the utility of nursing interventions designed to further promote healthy parent–child relationships and secure parent–child attachment.
Introduction
As a discipline, Nursing has a long history of research in attachment and the related concept, parent1 –child relationships (Barnard, Eyres, Lobo, & Snyder, 1983; Bee et al., 1982). Nurses’ interest has historically attended to the importance of parenting (especially mothering; Mercer, 1985, 1995; Rubin, 1967) and understanding how early relationships with caregivers shape developmental outcomes for better or worse (Barnard et al., 1988; Oxford & Finlay, 2013). This understanding has underpinned public health nursing interventions for young childbearing families for decades (Booth, Spieker, Barnard, & Morisset, 1992; Kang, Barnard, & Oshio, 1994; Mitchell et al., 1988; Olds et al., 2007).
Nurses and nurse researchers have long recognized that early experiences shape development and have examined behavioral outcomes in children and adults associated with early experiences in relationships (Letourneau, Dennis, et al., 2012; Letourneau & Joschko, 2013; Letourneau, Tramonte, & Willms, 2012). Now, modern neuroscience permits the examination of underlying brain structures that correlate with nurses’ and others’ observations of human development and health outcomes. Nursing curricula today tend to keep pace with this science by educating nurses about the emerging neuroscientific understanding of the brain (e.g., Bader & Littlejohns, 2010; Bay, Binder, Lint, & Park, 2015; Couns, 2013; Martins & Mello-Carpes, 2014). In that spirit, this article seeks to connect nurses’ rich history promoting healthy parent–child relationships and attachment with the neuroscience linked to those early experiences. To achieve that aim, this article reviews attachment, neuroscience, and ultimately the research linking adult attachment to brain structure and function through modern neuroimaging techniques, with the primary focus on nonparenting adults.
Attachment Theory
When first developed by John Bowlby, attachment theory derived its principles largely from observable behaviors exhibited in interactions between parents and children (Bowlby, 1953, 1969/1982, 1988). Parent–infant attachment is regarded as an evolutionary adaptive survival strategy, allowing infants to form powerful and long-lasting bonds with their attachment figures (usually primary caregiving parent), whose protection, comfort, and support they need in order to survive (Bowlby, 1969/1982, 1988). Originally defined by Mary Ainsworth, attachment is categorized as either secure (Type B) or insecure pattern of behavior, with insecure attachment subdivided into anxious avoidant (Type A) and anxious resistant (sometimes called ambivalent, Type C; Ainsworth & Bell, 1970; Beauchaine, 2003).
Typically measured via observation in the Strange Situation Procedure (Ainsworth, Blehar, Waters, & Wall, 1978) or via the Attachment Q-sort (Waters & Deane, 1985), securely attached children display affection toward their attachment figures (usually primary caregiving parent) but are also comfortable exploring their environment on their own for short periods of time. They use their caregiver as a secure base from which to undergo these explorations and a source of comfort if they become frightened, hurt, or overwhelmed (Sroufe, 2005). Avoidantly attached infants display coldness or indifference toward their caregivers. They regularly exhibit self-soothing behavior and are neither outwardly bothered by their caregiver’s absence nor comforted by their presence (Ainsworth et al., 1978). In contrast, resistantly attached infants become highly agitated when separated from their caregivers for even brief periods. However, they do not appear comforted upon their caregiver’s return, and may act indifferent or even hostile toward them (Ainsworth et al., 1978). Insecure attachment patterns are considered organized as they function to ensure care from parents but operate in a maladaptive way (e.g., “insecurely or maladaptively attached infants may need contact even when environmental stress is minimal…”; Sroufe & Waters, 1977, p. 3).
When children experience consistently sensitive parenting (e.g., caring attention to infant cues such as crying) children become: (a) securely attached to their caregivers and (b) display behaviors that are well balanced between exploratory behavior and seeking proximity to and comfort from their caregiver. When children experience consistently neglectful parenting (e.g., ignoring infant cries that connote need), children become avoidantly attached and become their own source of comfort, as children have learned that their caregivers cannot be relied upon and may reject their bids for nurturance or soothing. When children experience inconsistently neglectful parenting behavior (e.g., sometimes responding quickly to infant cries, but other times ignoring them for long periods of time), children become resistently attached and cling to their caregivers to avoid the neglect they occasionally and unpredictably experience (Ainsworth & Bell, 1970; Egeland & Farber, 1984). Avoidant and resistant insecure attachment patterns act as opposite poles on a continuum of behavior, with secure attachment landing squarely in the middle of the two (Bretherton, 1992).
Two separate schools of thought (Dynamic Maturational Model [DMM] vs. ABCD Model) regarding attachment theory emerged from Ainsworth’s work. One adheres to protection from danger as the foundation of attachment (DMM; Crittenden, 2005), while the other adheres to security via maintaining proximity to and comfort from an attachment figure as the foundation of attachment (ABCD; Main & Solomon, 1986). Both adopt Ainsworth’s three original patterns of attachment (Types A, B and C), but differ in the organizing principle. In Crittenden’s (2005) conceptions of attachment, all infant behavior is considered an “organized” self-protective strategy that elicits predictable contingent responses from caregivers, which can include extreme examples of Type A or Type C patterns and A/C mixed patterns of attachment. In contrast, Main and Solomon (1986) claim to have discovered a fourth type of attachment strategy they describe as “disorganized” (Type D), which is characterized as disoriented behavior (e.g., wandering, confused expressions, freezing, undirected or contradictory movements when separated and reunited with the caregiver; Howe, 2011) and has been supported by extensive research (Bakermans-Kranenberg, van Ijzendoorn, & Juffer, 2005; Barkermans-Kranenburg, Van Ijzendoorn, & Juffer, 2005; Groh, Roisman, van Ijzendoorn, Bakermans-Kranenberg, & Fearon, 2012; Juffer, Bakermans-Kranenberg, & Van Ijzendoorn, 2005; Lyons-Ruth, Alpern, & Repacholi, 1993; Madigan et al., 2006; van Ijzendoorn, Schuengel, & Bakermans-Kranenberg, 1999a, 1999b).
Adult Attachment
While attachment patterns established in childhood can change over the life course, attachment patterns in childhood are relatively stable over time, especially under conditions of low stress and high social support (Sroufe, 2005). Adult attachment patterns are assumed to be learned in childhood via the emotional bonds that are developed between parents and their children (Feldman & Downey, 1994). There are a number of psychometrically sound approaches to measuring and classifying adult attachment patterns, with all instruments differentiating between security and insecurity and subtypes (Ravitz, Maunder, Hunter, Sthankiya, & Lancee, 2010). A recent 25-year review of adult attachment measures reported that of the 29 instruments typically used to assess adult attachment, 11 demonstrated strong psychometric properties (Ravitz et al., 2010). We will comment on the two most common measures.
Considered the gold standard, the Adult Attachment Interview (AAI) is a narrative-based interview that is used to classify adult attachment via psycholinguistic qualities that reveal interviewees’ state of mind, for example, coherence versus noncoherence when adults are asked to describe how their parents behaved toward them as young children (Main, Kaplan, & Cassidy, 1985). Three main adult attachment patterns can be yielded, including secure or autonomous (George, Kaplan, & Main, 1984, 1988, 1996; Pearson, Cohn, Cowan, & Cowan, 1994), dismissing, and preoccupied, which are analogous to the childhood attachment patterns (i.e., secure Type B, anxious avoidant Type A, and anxious resistant Type C respectively; Ainsworth et al., 1978). A fourth pattern called unresolved, parallels disorganization (i.e., Type D). Adults classified as secure (continuous or earned; Type B) provide clear, direct, balanced (i.e., positive and negative accounts of their early childhood experiences with their parents), and coherent responses regarding their childhood attachment experiences, whereas adults classified as dismissing (Type A) undervalue attachment relationships and provide restricted and avoidant responses; adults classified as preoccupied (Type C) ruminate extensively over past attachment experiences and respond with psychological jargon, nonsense words, and childlike speech (Paley, Cox, Burchinal, & Payne, 1999; Westen, Nakash, Thomas, & Bradley, 2006). A significant relationship exists between early experiences with primary caregivers and later attachment-related representations (Steele et al., 2014).
Attachment patterns in childhood are linked with the quality of intimate relationships in adulthood (Bartholomew, 1993). Adults who are classified as having a secure pattern of attachment are more satisfied in their romantic relationships when compared with their insecure counterparts (Gleeson & Fitzgerald, 2014). Secure attachment is also negatively correlated with perceptions of stress, and positively correlated with resiliency, proactive coping strategies (Marriner, Caciolli, & Moore, 2014), self-reliance, self-regulation, and social competence in adulthood (Sroufe, 2005). Additionally, individuals who adopt a secure attachment pattern display fewer signs of emotional and behavioral disorders and are less likely to suffer from depression, anxiety, or excessive aggression in adulthood (Steele & Steele, 2013).
In contrast, insecure attachment strategies in adulthood are associated with problems in social, emotional, and mental domains (Fraley, 2002; Madigan, Atkinson, Laurin, & Benoit, 2013; van Ijzendoorn et al., 1999a), as well as health problems, especially inflammatory conditions over the life span (Puig, Englund, Simpson, & Collins, 2013). Insecurely attached adults are disproportionately more likely to exhibit an array of internalizing and externalizing behavioral problems including depression, anxiety, aggression, and attention-deficit hyperactivity disorder (Steele & Steele, 2013), which appear to also have negative consequences for adult intimate relationships (Paetzold, Rholes, & Kohn, 2015). These findings are in accordance with past research that indicates insecure attachment in adulthood is linked with a host of problems including the inability to establish long-term trusting relationships (Barlow, 2013).
Attachment and Neurodevelopment
During the period of establishment of the initial attachment pattern (typically, first 3 to 6 years of life), the human brain generates as many as 2 million synapses per second. This prolific production of synaptic connections enables the child’s brain to be especially adaptive to experience (Huttenlocher, 2009). Gradually, neuronal connections undergo systematic organization associated with experience including caregiving, wherein the most used pathways are retained and strengthened and the least used ones are eliminated or weakened (Pascual-Leone, Amedi, Fregni, & Merabel, 2005). Given the primary role of experiences with parents in infancy and indeed the evolutionary role of attachment in promoting infants’ survival (Strathearn, Fonagy, Amico, Montague, 2009), social-emotional and behavioral adjustment, experiences with parents are likely to contribute significantly to retention and elimination of synaptic connections across many brain regions. Indeed, we conjecture that the attachment patterns that children and adults demonstrate (secure vs. insecure) likely correlate with underlying changes to brain structure and function that arise from early formative experiences that may be observed in adulthood.
Improvements in neuroimaging technology have allowed researchers to explore correlations between attachment patterns and changes in brain structure (morphology) and function (physiology) in great detail. Several prominent procedures exist for monitoring brain activity, but the most common for attachment studies has been functional magnetic resonance imaging (fMRI). A high resolution and noninvasive procedure, fMRI uses the differing magnetic signals of oxygenated and nonoxygenated hemoglobin to map brain activity localized to the nearest millimeter. By monitoring this activity while exposing participants to certain stimuli, researchers can determine the importance of different brain regions in reacting to and processing different types of information. Thus, neuroimaging technologies, with fMRI chief among them, have been used by researchers to develop a far more detailed understanding of the effects of attachment behaviors on specific regions of the brain than in the past (Ulmer & Jansen, 2013). Another technique used to assess brain structure (not function) is called voxel-based morphometry (VBM), used to assess whole brain structural brain differences by quantifying gray matter concentrations in the whole brain (Ashburner & Friston, 2000; Mechelli, Price, Friston, & Ashburner, 2005).
The lobes of the brain, including frontal, parietal, occipital, temporal, and limbic regions, were originally a purely anatomical classification but have been shown to be related to different brain functions (Martin, 2008). Key brain areas of interest include those associated with executive (cortical) function, affective (emotion) processing and arousal, reward, sensory, social and cerebral functions, and stress. While no single brain region is solely responsible for a specific function, various brain regions do moderate functions, specifically:
Executive function is centered in the frontal lobes, including the right and left inferior frontal cortices, orbitofrontal cortex, superior frontal gyrus, dorsolateral prefrontal cortex, anterior cingulate cortex, medial frontal gyrus, left lateral orbital gyrus, left anterior gyrus, right middle frontal gyrus, prefrontal cingulated, and supplementary motor area. This region is the last to develop and is the most influenced by the environment. Affective processing and arousal are associated with the temporal lobe and specifically the amygdala, hippocampus, anterior temporal pole, and left parahippocampal cortex. Reward is associated with the subcortical structures, specifically the ventral caudate, ventral putamen, striatum, ventral tegmental area, left claustrum, and left globus pallidus. Sensory functions are linked with both the occipital and parietal lobes and include the left occipital cortex, occipital-temporal areas, and somatosensory cortices. Social functions are associated with the insula cortex. Cerebral functions, responsible for thought and action are associated with the right cerebellum. The stress response is associated with the hypothalamus and pituitary structures that connect with the adrenal gland as part of the hypothalamic-pituitary-adrenal (HPA) axis. (See Figure 2 for summary of regions and associated behaviors and functions.)
Animal and Human Attachment Behaviors and Neurodevelopment
Although animal studies do not directly correlate with human experience, reviews of research have shown strong links between rodent and human maternal caregiving behaviors, linked to attachment and brain structure and function (Swain, Loberbaum, Kose, & Strathearn, 2007; Swain, 2008, 2011). Both humans and rodent species calibrate their HPA axes, an endocrine system responsible for regulating their response to stressful stimuli, through qualities of early maternal caregiving behavior and many neural regions affected by maternal behaviors in mice have human counterparts shown to react in a similar fashion (Swain, 2008). Likewise, the detrimental effect of poor early maternal caregiving behavior on emotional and behavioral development—particularly involving the HPA stress response system has been well documented in both rodent (Caldji, Diorio, & Meaney, 2000; Szyf, Weaver, Champagne, Diorio, & Meaney, 2005; Weaver et al., 2005; Weaver, Diorio, Seckl, Szyf, & Meaney, 2004) and human cases (Kidd, Hamer, & Steptoe, 2011; McGowan et al., 2009).
Numerous rodent studies have observed a clear connection between the frequency at which dams engage in affectionate maternal behaviors, linked to attachment, and the development of their pups’ HPA axis. Rodent pups whose mothers perform more nurturing behaviors (licking, grooming, and arched-back nursing) demonstrate a more moderate endocrinal response to stress and return more promptly to a prestress state than do their less nurtured peers (Szyf et al., 2005). Further, reviews of nonhuman animal research has also provided evidence of associations between attachment behaviors and neural development, particularly in basal forebrain structures, which are located to the front of and below the striatum, include the nucleus accumbens and link to reward (Swain et al., 2007). Damage to these forebrain structures, associated with reward, is linked to elimination of attachment behaviors among rodents, including nursing, breastfeeding, and nest building (Numan & Insel, 2003). Other areas involved in regulating rodent maternal behavior include the medial preoptic area of the hypothalamus, the ventral part of the bed nucleus of the stria terminalis (serves as major output pathway for the amygdala associated with affective processing), and the lateral septum (connected to the hippocampus, amygdala, hypothalamus, and cingulate gyrus linked to executive function, and affective and stress processing) (Leckman & Herman, 2002; Numan, 2007; Numan & Stolzenberg, 2009; Swain, 2008, 2011). With the exception of evidence from animal models on associations between early maternal behavior and offspring HPA axis function, much of the extant literature focuses on associations between brain structure or function and maternal behavior. Perhaps in animals, it is difficult to assess attachment behavior linked to neuroanatomic function, in the absence of parenting. Nonetheless, we require a review of evidence of associations between brain structure or function and nonparenting adults’ attachment behavior, linked to their own early maternal caregiving experience.
In humans, research has focused largely on brain structure and function and adult attachment behaviors in parents, but also among romantic partners. In addition to influencing the development of the HPA axis, attachment patterns have been associated with reward regions of the brain (including the basal ganglia, ventral striatum or nucleus accumbens, amygdala, hypothalamus, and hippocampus) and emotional and sensory processing (including the anterior cingulate, insula medial frontal, and orbitofrontal cortices) in parents (Swain, 2011). Further, reviews of the extant research has focused on the influence of becoming a parent on both attachment and brain structure and function (Strathearn et al., 2009; Strathearn, Li, Fonagy, & Montague, 2008), revealing the role of various reward circuits in reinforcing feelings of both romantic and maternal love (Vrtička, Sander, & Vuilleumier, 2012; Weisman, Feldman, & Goldstein, 2012). These feelings are associated with heightened functional responses in reward centers to various attachment stimuli (e.g., an image of a newborn baby, the sound of infant crying) among new parents, romantic partners, or other individuals experiencing strong attachment bonds (Swain, 2008; Weisman et al., 2012). While informative, well described brain structural and functional changes associated with the onset of pregnancy and parenthood (Strathearn et al., 2009; Strathearn et al., 2008) confound the evidence of linkages between attachment pattern and adult brain structure and function.
In summary, the association between human nonparenting adults’ attachment patterns and brain structure and function has not been systematically reviewed. Moreover, as changes have been observed coincident with becoming a parent (Leuner, Glasper, & Gould, 2010; Mayes, Swain, & Leckman, 2005), it is surprising that reviews have not examined the influence of early attachment experiences on nonparents’ brain structures and functions, as a baseline.
Purpose of This Review
Neuroimaging research on the correlation between attachment strategies and brain function and structure has only begun over the last decade. During that time, much of it has focused on human and nonhuman parents and human romantic partners. Thus, the goal of this article is to review and synthesize the literature examining the relationship between attachment patterns emerging from childhood and brain function and structure in nonparenting human adults.
Methods
This review followed the methodology recommended by the PRISMA guidelines (Moher, Liberati, Tetzlaff, & Altman, 2009; see Figure 1) and the Cochrane Collaboration (Higgins & Green, 2011). In March 2016, researchers searched PubMed, EBSCO, ProQuest, and Google Scholar using the following search terms: (Magnetic Resonance Imaging OR fMRI OR Positron-Emission Tomography OR Spectroscopy) AND (Brain Mapping OR Neuroimaging OR Functional Neuroimaging) AND (Attachment OR Attachment Behavior OR Object Attachment OR Early Child Adversity). Searches yielded a total of 1,604 studies.
PRISMA 2009 Flow Diagram (Moher et al., 2009). Summary of regions and associated behaviors and functions.

To meet inclusion criteria, studies had to (a) focus on brain structure or brain function outcomes in adults, (b) use neuroimaging technology to generate findings, and (c) compare securely attached versus nonsecurely attached groups. Adult studies were included if participants underwent attachment-related measures such as an AAI establishing patterns of attachment for adults based on childhood experiences. Adult studies focusing on parents or romantic partners were excluded.
A total of 1,604 abstracts were screened. Of these, 20 met the criteria and the full-text articles were retrieved and assessed for eligibility. Only 11 of the 20 studies were ultimately determined to meet all the eligibility criteria and included in this review. All 20 studies were subject to ancestry searches (i.e., searching the articles’ reference list), as were any relevant review articles found during the initial search. An additional three articles were retrieved in this manner, all of which met full-text inclusion criteria. Thus, this review includes 14 studies in the qualitative synthesis.
Results
The 14 studies that met inclusion criteria were subjected to review. As fMRI studies performed on children are rare (i.e., only one identified, to date, Tottenham, Shapiro, Telzer, & Humphreys, 2012), it was necessary to utilize adult studies. Nevertheless, the included studies focus on attachment patterns (e.g., DeWall et al., 2012; Galynker et al., 2012; Lemche et al., 2006; Lenzi et al., 2013; Riem, Bakermans-Kranenburg, van IJzendoorn, Out, & Rombouts, 2012; Vrtička, Andersson, Grandjean, Sander, & Vuilleumier, 2008; Vrtička, Bondolfi, Sander, & Vuilleumier, 2012), attachment-related feelings about close relationships (Benetti et al., 2010; Donges et al., 2012; Gillath, Bunge, Shaver, Wendelken, & Mikulincer, 2005; Quirin, Gillath, Pruessner, & Eggert, 2010; Suslow et al., 2009), narratives about pictures that activate the attachment system or (e.g., illness, abuse; Buchheim et al., 2006), “secure-base attachment scripts” (e.g. Warren et al., 2010) presumed to have developed in childhood. Nine of the studies included both male and female participants (Benetti et al., 2010; DeWall et al., 2012; Donges et al., 2012; Lemche et al., 2006; Lenzi et al., 2013; Quirin et al., 2010; Suslow et al., 2009; Vrtička et al., 2008; Warren et al., 2010), four studies focused only on female participants (Buchheim et al., 2006; Galynker et al., 2012; Gillath et al., 2005; Vrtička, Bondolfi, et al., 2012), and one study focused on twins of both sexes (Riem et al., 2012).
The ages of participants ranged from 18 years (Gillath et al., 2005; Vrtička et al., 2008) to 36 years (Donges et al., 2012; Warren et al., 2010) to unknown (DeWall et al., 2012). Most of the studies focused on right-handed participants (Buchheim et al., 2006; DeWall et al., 2012; Donges et al., 2012; Gillath et al., 2005; Lemche et al., 2006; Quirin et al., 2010; Riem et al., 2012; Vrtička, Bondolfi, et al., 2012; Warren et al., 2010) and some explicitly studied Caucasians (Galynker et al., 2012; Lenzi et al., 2013; Quirin et al., 2010; Suslow et al., 2009). Other studies specified that participants have normal vision (Lemche et al., 2006; Vrtička et al., 2008; Warren et al., 2010) or normal to corrected vision (Donges et al., 2012; Galynker et al., 2012; Vrtička, Bondolfi, et al., 2012) and normal hearing (Riem et al., 2012). Other studies only included participants with no history of psychiatric illness (Benetti et al., 2010; Buchheim et al., 2006; Donges et al., 2012; Gillath et al., 2005; Lenzi et al., 2013; Quirin et al., 2010; Riem et al., 2012; Vrtička et al., 2008; Vrtička, Bondolfi, et al., 2012) and who do not have claustrophobia (DeWall et al., 2012; Warren et al., 2010). Two studies examined mental health as part of the assessment including anxiety (Lemche et al., 2006), and anxiety and depression (Warren et al., 2010). Galynker (2012) recruited both depressed and healthy participant and excluded participants who were not raised from birth to at least 14 years of age in a household with their biological mother.
A variety of scales and surveys were used to describe attachment strategies in adulthood, including the Relationship Scales Questionnaire (Donges et al., 2012; Griffin & Bartholomew, 1994; Suslow et al., 2009; Vrtička, Bondolfi, et al., 2012), the AAI (Galynker et al., 2012; George, Kaplan, & Main, 1996; Lenzi et al., 2013; Riem et al., 2012), the Adult Attachment Questionnaire (Simpson, Rholes, & Phillips, 1996; Vrtička et al., 2008), the Attachment Style Questionnaire (DeWall et al., 2012; Feeney, Noller, & Hanrahan, 1994), the Adult Attachment Projective (Buchheim et al., 2006; George & West, 2001), the Attachment Script Assessment (Warren et al., 2010; Waters & Rodrigues-Doolabh, 2004), and the Experience in Close Relationships scale (Benetti et al., 2010; Brennan, Clark, & Shaver, 1998; Gillath et al., 2005; Quirin et al., 2010). Lemche et al. (2006) incorporated attachment categorization into the behavioral study itself: Participants were asked to respond as quickly as possible to a series of text prompts displayed in between rounds of subliminal messages containing either gibberish or descriptions of unpleasant attachment situations. Galynker’s (2012) team used the AAI but derived the Coherence of Mind index (George et al., 1996) as a measure of attachment security with values ranging from 1 to 9. Scores of 6 to 9 indicate secure attachment, scores of 1 to 3 indicate insecure attachment, and 4 to 5 were “indeterminant.” Thus, all but three (Brennan et al., 1998; Galynker et al., 2012; Warren et al., 2010) of the studies examined in the current review used one of the 11 adult attachment measures deemed psychometrically sound by Ravitz et al. (2010).
Attachment Classification
As previously mentioned, the original attachment categories established by Ainsworth et al. (1973) are secure (Type B) and insecure with two subcategories, namely insecure anxious avoidant (Type A) and insecure anxious resistant or ambivalent (Type C). Over the years, nomenclature relating to attachment research has expanded. For example, some researchers use the term anxious to describe either a specific type of insecure attachment (e.g., avoidant or dismissing or resistant/ambivalent) or both forms of insecure attachment, which can cause confusion when attempting to review attachment research. Originally, Ainsworth et al. (1978) described both subcategories of insecure attachment as “anxious,” (e.g., avoidant or dismissing and resistant or preoccupied); however, many refer to avoidant and anxious types, suggesting that anxious refers to the resistant category. Thus, for the purposes of this review, any study referring to insecure attachment as only “anxious” can be viewed as being interchangeable with Ainsworth’s “anxious resistant/ambivalent” classification or preoccupied in adults. The reason for this is that researchers’ descriptions of “anxious” attachment typically resemble similar descriptions Ainsworth provided for “anxious resistant/ambivalent” insecure attachment. That said, in order to preserve the integrity of this review article, the language used in each individual study as it relates to attachment will not be altered.
Reviewed studies employed participants’ category of attachment security as a variable with which to analyze brain structure and function. Seven studies treated “avoidant/dismissing” and “resistant/preoccupied” (anxious) attachment categories as separate variables (Benetti et al., 2010; DeWall et al., 2012; Gillath et al., 2005; Quirin et al., 2010; Vrtička et al., 2008; Vrtička, Bondolfi, et al., 2012) with two of those studies including an additional variable representing secure attachment (Quirin et al., 2010; Vrtička et al., 2008). Five studies used a continuous scale to assess degrees of security or insecurity, thus insecure attachment is represented as only one primary category not two (Buchheim et al., 2006; Galynker et al., 2012; Lemche et al., 2006; Riem et al., 2012; Warren et al., 2010). Two studies focused solely on participants’ levels of attachment avoidance or dismissing (Lenzi et al., 2013; Suslow et al., 2009), and another focused solely on attachment anxiety, that is insecurity (Donges et al., 2012). None focused on attachment disorganization (Type D) or A/C mixed attachment patterns.
Stimuli and Design
Studies included used varying stimuli to trigger neural regions associated with attachment. The majority of studies measured participants’ response to evocative images, most commonly faces expressing positive or negative emotions and masked by neutral expressions (Donges et al., 2012; Galynker et al., 2012; Suslow et al., 2009). Faces were typically of strangers (Donges et al., 2012; Galynker et al., 2012; Suslow et al., 2009), although one study used faces of individuals familiar to the participants (e.g., mother, close friend) as well as those of strangers (Galynker et al., 2012). Two studies (Donges et al., 2012; Suslow et al., 2009) employed standardized faces such as Ekman’s (Ekman & Friesen, 1976). Participants were asked to rank images by pressing buttons (Donges et al., 2012; Suslow et al., 2009). Other images besides faces were sometimes used. One study used images of children aged 6 to 12 months and asked participants at various points to either imitate the children’s expression or attempt to empathize with them (Lenzi et al., 2013). Two studies had participants view images of scenes with connotations considered positive (e.g., a mother playing with her baby, a tropical island scene), negative (e.g., a couple arguing, a bird covered in oil), or neutral (e.g., children playing in a sandbox; Buchheim et al., 2006; Vrtička, Bondolfi, et al., 2012). Participants were asked to rank images as positive or negative on a 4-point scale (Vrtička, Bondolfi, et al., 2012) or create a narrative based on the image displayed (Buchheim et al., 2006).
Four studies used alternative means of aural or visual stimuli (Gillath et al., 2005; Lemche et al., 2006; Riem et al., 2012; Warren et al., 2010). Gillath et al. (2005) described relationship scenarios to participants and asked them to either reflect on or suppress thoughts of these scenarios according to visual cues rendered in the form of a traffic light. Lemche et al. (2006) showed participants a series of subliminal sentences containing either nonsense or descriptions of unpleasant attachment experiences, followed by plainly displayed statements to which the participants responded by agreeing or disagreeing. Riem et al. (2012) measured the neural response of participants to sounds of infant crying. Finally, Warren et al. (2010) showed patients a series of pleasant and unpleasant words, each written in one of four colors, and had them indicate which color they saw via a control pad.
Two studies used simulations in which the participants thought they were playing a game with or against other players (DeWall et al., 2012; Vrtička et al., 2008). The games’ outcomes were manipulated to trigger attachment responses. In DeWall (2012), participants took part in a virtual ball-tossing game. Three players passed a “cyberball” back and forth. Unbeknownst to the participant, the other two “players” were part of the computer program. After the first round, the participant was excluded from the game by the other players, triggering feelings of social distress. In Vrtička’s et al. (2008), participants played a counting game. The program actively calibrated the game’s difficulty to the player’s level of success, assuring that each player both won and lost some rounds. The results of each round were followed by the words “won” or “lost” and an image of a smiling or frowning face, symbolizing an opponent. Both smiling and frowning faces could appear over either “won” or “lost” captions, allowing for a wider variety of emotional stimuli. Finally, two studies that were not fMRI studies, but rather VBM studies of brain structure, did not have participants engage in any sort of activity (Benetti et al., 2010; Quirin et al., 2010). Instead, they assessed participants’ attachment pattern (i.e., secure, avoidant, or anxious) and examined results for attachment-based differences in neural structure.
fMRI or VBM Data Acquisition
Five studies employed 1.5 Tesla scanners (Benetti et al., 2010; Buchheim et al., 2006; Gillath et al., 2005; Lemche et al., 2006; Vrtička et al., 2008) with Buchheim et al. (2006) using a Magnetom Symphony Scanner, Lemche et al. (2006) and Gillath et al. (2005) using a General Electric Neurovascular Signa MRI. Benetti et al. (2010) used a General Electric MRI. Gillath et al. (2005) indicate the use of a birdcage head-coil T2-weighted gradient with echo-planar imaging procedure and Lemche et al. (2006) reported using whole brain T1-weighted structural imaging. Vritikca et al. (2008) reported using whole-body INTERA system using standard head coil.
Six studies employed three Tesla scanners (DeWall et al., 2012; Galynker et al., 2012; Quirin et al., 2010; Riem et al., 2012; Vrtička, Bondolfi, et al., 2012; Warren et al., 2010), with Vrtička, Sander, et al. (2012) employing whole-body scanner and standard head coil and Quirin et al. (2010) indicated the use of a birdcage head coil. Warren et al. (2010) and Quirin et al. (2010) employed Siemens Allegra head-only scanner and Galynker (2012) used Phillips Intera and SENSE head coil. Riem et al. (2012) employed Phillips Achieva MRI and Dewall et al. (2012) employed Siemens Trio scanner. Three additional studies (Donges et al., 2012; Lenzi et al., 2013; Suslow et al., 2009) did not indicate the nature of their three Tesla MRI; however, Donges et al. (2012) and Suslow et al. (2009) reported using a single shot echo-planar sequence with parameter selection to minimize distortion and retain adequate signal to noise ratio and T2 sensitivity. Lenzie et al. (2013) only indicated that T2 functional data were collected, while Dewall et al. (2012) reported using T2-weighted graduated echo. Riem et al. (2012) reported T1-weighted anatomicals. Finally, T1 weighted three dimensional magnetization prepared rapid gradient echo were reported by Quirin et al. (2010). When indicated, the number of slices varied with 16 (Lemche et al., 2006), 24 (Gillath et al., 2005), 25 (Buchheim et al., 2006), 30 (Vrtička et al., 2008), 32 (Lenzi et al., 2013), 34 (Donges et al., 2012), 38 (Riem et al., 2012; Warren et al., 2010), 40 (DeWall et al., 2012; Suslow et al., 2009), and 180 in their VBM study (Benetti et al., 2010).
Neural Outcomes
All studies provided neuroimaging data describing the effects of attachment on brain structures and function. However, inconsistencies in attachment terminology were problematic. Although avoidant and anxious attachment behaviors are overlapping subsets of insecure attachment, several articles treated them as distinct categories. As such, the results of this study have been divided into overall insecure, avoidant, and anxious (resistant or ambivalent) attachment strategies. Again, neither disorganized attachment (Type D) nor A/C mixed patterns were mentioned in any of the studies that met inclusion criteria, so these categories were ignored. In brackets after brain areas, we indicate the associated function.
Secure attachment
Only two studies explicitly explored associations between secure attachment and brain function. Participants, who were securely attached, activated the perigenual anterior cingulate (executive) and the medial orbitofrontal cortex (executive) more than insecure or dismissing participants (Lenzi et al., 2013). Secure participants activated the striatum (reward) in response to smiling faces and amygdala (affect) in response to angry faces (Vrtička et al., 2008).
Overall insecure attachment
Participants with insecure attachment displayed greater levels of activation in the right inferior frontal cortex (executive), the left occipital cortex (sensory), and the medial temporal regions (affect), including the amygdala and the hippocampus (Buchheim et al., 2006). When exposed to stimuli designed to evoke a stressful attachment-based response (e.g., listening to infant crying), participants showed above average activity in the amygdala (affect; Lemche et al., 2006; Riem et al., 2012). Unpleasant words triggered greater activity in the orbital frontal cortex (executive) and superior frontal gyrus (executive) in participants with insecure attachment, while pleasant words activated the left dorsolateral prefrontal cortex (executive) and the anterior cingulate cortex (executive; Warren et al., 2010). Insecurely attached individuals displayed a markedly greater activation of the bilateral ventral putamen (reward), left medial thalamus (sensory), right inferior orbitofrontal cortex (executive), and bilateral lateral-prefrontal cortex (executive) when viewing pictures of an attachment figure (i.e., mother) than when viewing images of strangers (Galynker et al., 2012). Viewing images of maternal attachment figures (vs. strangers) also correlated to deactivation of the right middle orbitofrontal cortex (executive) and middle temporal gyri (affect; Galynker et al., 2012). Insecurely attached individual also displayed strong activation in the left superior medial frontal cortex, as well as the left superior frontal gyrus and insula bilaterally and deactivations in the left posterior cingulate and orbitofrontal cortex (Galynker et al., 2012).
Avoidant or dismissing attachment
Participants displaying avoidant attachment behavior experienced increased activation in a number of areas, including the subcallosal cingulate cortex (executive; Gillath et al., 2005), bilateral sensory-motor cortex (sensory), right hippocampus (affective), anterior cingulate cortex (executive), right superior frontal gyrus (excutive; Lenzi et al., 2013), amygdala (affect), insula (social), occipital-temporal areas (sensory), and somatosensory cortices (sensory; Suslow et al., 2009), and decreased activation in the orbitofrontal cortex (executive) and anterior cingulate cortex (executive; Lenzi et al., 2013). They exhibited greater than average activity in the prefrontal cingulate (executive), anterior cingulate (executive), dorsolateral prefrontal cortex (executive), and left amygdala (affect) when asked to make emotional judgments regarding images designed to elicit negative emotions (Vrtička, Bondolfi, et al., 2012). When asked to suppress positive emotions, they displayed high levels of activation in supplementary motor area (executive control of motor systems) and ventral caudate (reward; Vrtička, Bondolfi, et al., 2012). When asked to imitate or empathize with infants, they experienced greater activation of the left inferior frontal gyrus (executive), right middle frontal gyrus (executive), and right superior temporal sulcus (affect), and reduced activation in the left inferior occipital gyrus (sensory), when compared with securely attached participants (Lenzi et al., 2013).
In response to positive stimuli, namely images of smiling faces participants with avoidant attachment showed reduced activation in the striatum (reward) and ventral tegmental area (reward), indicating relative indifference to social reward (Vrtička et al., 2008). Avoidant attachment was negatively correlated with activity in the dorsal anterior cingulate cortex (executive) and bilateral anterior insula (social) in response to social rejection (DeWall et al., 2012). Avoidant individuals showed reduced gray matter concentration in the left hippocampus (affect; Quirin et al., 2010).
Anxious (resistant or ambivalent) attachment
Association between brain regions, functions and attachment pattern.
Attachment and functional associations
Five articles identified associations between attachment pattern and aspects of social-emotional perception and cognitive control that were mediated by brain structures. First, Riem et al. (2012) found that insecure women had heightened amygdala (affect) activation and more irritation to infant crying, compared with secure women. Vrtička et al. (2008) found that activation of the striatum and ventral tegmental areas (reward) were enhanced in secure participants who viewed smiling faces, but reduced in insecure avoidant participants, indicating insecure participants’ impassiveness to social reward. In participants with anxious (resistant or ambivalent) attachment, left amygdala (affect) response was evoked by angry faces. While not examining social rejection specifically, DeWall et al. (2012) suggested that anxious (resistant or ambivalent) attachment correlated with the anterior cingulate cortex (executive) and anterior insula (social) linked to rejection-related distress and avoidant attachment related to less activity in these regions. In contrast, Donges et al. (2012) observed that anxiously (resistantly) attached women were more responsive to smiling faces in multiple areas associated with emotion processing. Warren et al. (2010) found that more insecure participants demonstrated more activity in the orbitofrontal cortex (executive) and in the dorsolateral anterior cingulate cortex, and superior frontal gyrus (executive), consistent with a greater need for cognitive control when exposed to attachment-relevant information.
With respect to associations among attachment pattern, brain, and mood, Galynker et al. (2012) found that the same brain structures, that is the right orbitofrontal cortex (executive function) and bilateral putamen (reward) were more activated and the left superior frontal gyrus (executive) was less activated in women with both depression and attachment anxiety insecurity. Similarly, Gillath et al. (2005) found that compared with women who were more secure, women with an anxious (resistant or ambivalent) attachment pattern showed less deactivation of the orbitofrontal cortex (executive) and anterior temporal pole (affect), while women who were more avoidant, failed to show as much deactivation in the cingulate cortex and lateral prefrontal cortex (executive) when asked to stop thinking about negative relationship scenarios.
Discussion
Results strongly indicate pervasive links between nonparenting adults’ attachment patterns and brain structures and functions. Securely attached participants have more activation in the executive region (anterior cingulate and medial orbitofrontal cortex; Lenzi et al., 2013). They also responded to smiling faces with reward activation (striatum) and to negative faces with affect (amygdala) activation (Vrtička et al., 2008). In summary, 13 regions of the brain associated with executive functions, 6 with affect and 6 with reward, 3 with sensory, 1 with social and 1 with cerebellar function were linked to insecurity or to a subtype. No associations were found with structures associated with stress functions.
Consistent with research on mothers, in nonparenting adults, associations were found between attachment pattern and superior, middle frontal and orbital gyri (executive), and superior temporal gyrus and hippocampus (affect; Coan, Schaefer, & Davidson, 2006; Kim et al., 2010). Both mothers and nonparenting adults with secure attachment showed greater activation of the prefrontal cortex (executive), ventral striatum (reward; Strathearn et al., 2009; Vrtička et al., 2008), and insula (social; Coan et al., 2006). However, associations between attachment pattern and HPA axis were only observed in mothers (Strathearn et al., 2009).
In nonparenting adults, support for these findings in the affective processing region may be found in studies of institutionalized children. These children, presumed to have an insecure attachment pattern, present with atypically large amygdala volumes as well as difficult emotion regulation and anxiety (Tottenham et al., 2010). Similarly, in a study of children and adolescents, an attachment measure that examined the child’s tendency to approach the mother when stressed (consistent with attachment security) was positively associated with amygdala (affect) response when viewing images of their mothers. No amygdala activation was observed when these more secure-type children viewed images of strangers (Tottenham et al., 2012). Amygdala activity is critical to fear learning (LeDoux, 2003) and likely children raised in institutions with multiple caregivers and stressful unresponsive environments experience frequent amygdala activation, resulting in larger volumes.
For the remaining brain regions, extensive literature links insecure attachment to behaviors associated with executive, affect, reward, sensory, and social brain regions. The pervasive differences between securely attached adults and insecurely attached adults may be linked to the many behaviors noted in insecurely attached individuals over the life span. Executive function changes may help explain the high prevalence of externalizing behaviors (e.g., aggression, behavior problems, antisocial behavior; Fearon, Bakermans-Kranenberg, van Ijzendoorn, Lapsley, & Roisman, 2010; Lyons-Ruth et al., 1993), while changes in affective processing may help explain the high prevalence of internalizing behavioral disorders (e.g., anxiety, depression, suicidality; Bos et al., 2011; Colonnesi et al., 2011; Groh et al., 2012) in insecurely attached individuals. Changes in the reward regions may help explain the higher prevalence of substance abuse among insecurely attached individuals (Andersen & Teicher, 2009). Changes in social circuits may help explain the high prevalence of excessive help seeking and dependency and excessive social isolation (Cook et al., 2005) in insecurely attached adults.
Moreover, these findings offer support for associations observed between attachment pattern and social-emotional perception, mood, and cognitive control. For example, Gillath et al. (2005) observed that insecurely (resistantly) attached women are less able to marshal brain resources to stop negative thoughts, consistent with descriptions of Type C people as excessively emotional and ruminating; while women with avoidant patterns maintained activation of areas associated with executive function, consistent with descriptions of Type A people as excessively intellectualizing in their thoughts (Crittenden, 2008). Similarly, Warren et al. (2010) found that insecure adults required more of their executive control functions when faced with attachment-relevant stimuli. This excess burden could relate to the high prevalence of personality disorders among insecurely attached people (Fonagy, Gergely, Jurist, & Target, 2002; Fonagy & Target, 1997).
Surprisingly, insecure attachment was not linked to stress functions via the hypothalamus, in spite of research linking attachment insecurity to posttraumatic stress disorder (Lupien, McEwen, Gunnar, & Heim, 2009; Neigh, Gillespie, & Nemeroff, 2009), and increased susceptibility to stress (Cook et al., 2005; Lemche et al., 2006). In addition to influencing the development of the HPA axis, attachment behaviors have been associated with reward regions of the brain (including the basal ganglia, ventral striatum/nucleus accumbens, amygdala, hypothalamus, and hippocampus) and emotional and sensory processing (including the anterior cingulate, insula medial frontal, and orbitofrontal cortices) in parents (Swain, 2011). These findings contrast with our review of nonparents’ studies. Of the two subtypes of insecure attachment, namely avoidant and resistant or ambivalent, avoidant individuals show poorer memory of attachment-related events (Fraley & Brumbaugh, 2007), linked to the affective and reward brain regions, a higher prevalence of conduct disorders (Sroufe, 2005) linked to the executive brain region, downregulation of emotion, and a low intensity of emotionality (Mikulincer, Shaver, & Pereg, 2003; Pietromonaco, Barrett, & Powers, 2006) and greater emotional distance from others (Vrtička, Bondolfi, et al., 2012; Vrtička, Sander, et al., 2012), linked to the affective brain region.
Resistant individuals show many outcomes linked to the social and affective brain regions. These include more intense behavioral responses to rejection and social conflict (Campbell, Simpson, Boldry, & Kashy, 2005), lower self-esteem (Besser & Priel, 2009), sense of self-worth dependent on other’s approval (Park, Crocker, & Mickelson, 2004), lower self-reliance (Sroufe, 2005), worry about relationships, and heightened fear of rejection (Vrtička, Bondolfi, et al., 2012). These adults also have difficulties in inhibiting negative thoughts and emotions (Vrtička, Bondolfi, et al., 2012), linked to social, affective, and executive regions.
Unfortunately, the nature of the studies’ definition of insecure attachment mostly prevented the examination of direct linkages between brain regions and attachment subtypes. Some studies dichotomized attachment along the dimension of insecure versus secure, while other studies examined subtypes of insecurity including avoidant and resistant or ambivalent. Other limitations of this review include the various measures and range of definitions of attachment employed. Only three studies used the gold standard of the AAI; however, most used measures deemed valid (Ravitz et al., 2010). Nonetheless, this is the first review of the association between attachment patterns and brain structure and function in nonparenting adults. Future research would benefit from examination of more clearly delineated subtypes of attachment as well as longitudinal trajectories of attachment vis-à-vis brain structure and function.
Nurses will be able to use this research to inform their practice by improving nurses’ understanding of how patients’ relationship histories may impact present functioning. By having a better understanding of adult attachment and brain structure and function in nonparenting adults, nurses can better understand and integrate attachment and neuroscience research into their clinical practice. For example, nurses can use these findings to help prevent, assess, and address health risks associated with the various adult attachment patterns in nonparenting adults and understand how adult attachment patterns and brain function impact health and behavior.
Footnotes
Acknowledgement
Thanks to the Palix (formerly Norlien) Foundation and Alberta Children’s Hospital Foundation for funding this work and to Justin Joschko for contributions to and Dr. Nancy Suchman for comments on an early draft of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Palix (formerly Norlien) Foundation and the Alberta Children's Hospital funded this work.
