What Attachment Theory Describes
Attachment theory begins with a simple observation: human infants cannot survive without a caregiver. John Bowlby, drawing on ethological research in the 1950s and 1960s, argued that the behaviors infants use to maintain proximity to their caregivers — crying, clinging, following, reaching — are not random distress signals but an evolved behavioral system shaped by natural selection precisely because infants who maintained proximity to a responsive adult were more likely to survive. The attachment behavioral system, as Bowlby called it, evolved to regulate proximity under threat.
What makes attachment theory more than a developmental observation is Bowlby's second claim: that early caregiving experiences become organized into internal working models — cognitive and emotional representations of the self in relation to others. A child whose distress reliably brought a responsive caregiver forms a model of the self as worthy of care and others as available sources of support. A child whose distress was ignored, inconsistently met, or met with distress of the caregiver's own forms a different model — one that shapes expectations about relationships long after the original caregiving relationship has ended. Internal working models are not beliefs people consciously hold; they are implicit templates that operate outside awareness, structuring how attachment-relevant information is perceived and processed.
Bowlby's original work focused on infants and children. The application of attachment theory to adult romantic relationships came in 1987, when Cindy Hazan and Philip Shaver proposed that romantic love is, at its psychological core, an attachment process. The same behavioral system that organized infants around caregivers, they argued, organizes adults around romantic partners — the same felt security when the partner is available, the same distress when separated, the same protest behaviors when proximity is threatened. Hazan and Shaver's 1987 paper in the Journal of Personality and Social Psychology initiated the adult attachment research program that now encompasses thousands of studies and multiple validated instruments.
Two Dimensions, Four Styles
Adult attachment is best understood as continuous variation along two underlying dimensions rather than as four discrete categories.
The first dimension is attachment anxiety — the degree to which a person worries about whether their partner is available, responsive, and genuinely committed. High anxiety is characterized by hyperactivation of the attachment system: heightened vigilance for signs of rejection, preoccupation with relationship security, and escalating proximity-seeking under perceived threat. Low anxiety reflects confidence that partners will be responsive when needed.
The second dimension is attachment avoidance — the degree to which a person is uncomfortable with emotional closeness and prefers to rely on themselves rather than on others. High avoidance is characterized by deactivation of the attachment system: suppression of attachment needs, discomfort with vulnerability and dependency, and strategic emotional distance. Low avoidance reflects comfort with interdependence and closeness.
The four attachment style labels — Secure, Anxious-Preoccupied, Dismissive-Avoidant, and Fearful-Avoidant — correspond to quadrant positions in this two-dimensional space:
- Secure: low anxiety, low avoidance
- Anxious-Preoccupied: high anxiety, low avoidance
- Dismissive-Avoidant: low anxiety, high avoidance
- Fearful-Avoidant: high anxiety, high avoidance
This quadrant structure, formalized by Kim Bartholomew and Leonard Horowitz in their landmark 1991 paper in the Journal of Personality and Social Psychology, has a further layer of precision. The anxiety dimension maps onto the model of self in internal working models: high anxiety reflects a negative model of self (uncertainty about one's worthiness of care). The avoidance dimension maps onto the model of others: high avoidance reflects a negative model of others (uncertainty about others' availability and goodwill). The four styles thus represent four combinations of these two internal working model dimensions.
This structure explains something popular accounts of attachment style often miss: Dismissive-Avoidant and Fearful-Avoidant are not simply two kinds of avoidance. Dismissive-Avoidant individuals tend to hold a positive model of self alongside a negative model of others — they minimize the importance of attachment needs and typically report high self-sufficiency and low distress. Fearful-Avoidant individuals hold negative models of both self and others — they want closeness but simultaneously fear it, producing an approach-avoidance conflict that neither of the other insecure styles experiences in the same way. The avoidance in these two styles comes from different places and looks different in close relationships.
The Four Styles
Secure Attachment characterizes people who are comfortable with both closeness and autonomy. They tend to view themselves as worthy of care and others as generally trustworthy and responsive. Under relationship stress, they are more likely to communicate needs directly, tolerate conflict without catastrophizing, and return to emotional equilibrium relatively quickly. Secure attachment is associated with higher relationship satisfaction, more effective co-regulation of emotion, and — critically for the intergenerational transmission of attachment — more sensitive caregiving toward children. Approximately 55–60% of adults in Western populations score in the secure range across studies, though this proportion varies with methodology and population.
Anxious-Preoccupied Attachment characterizes people with high anxiety and low avoidance — they want closeness intensely and worry substantially about whether they will get it. The attachment system is chronically activated: small signs of partner inattentiveness or emotional distance can feel like evidence of abandonment. This produces patterns of reassurance-seeking, heightened emotional reactivity in relationship conflicts, and difficulty self-soothing when the partner is unavailable. Preoccupied individuals often describe their relationships as all-consuming. The pattern is associated with lower self-complexity outside of relationships and with a tendency to organize significant portions of identity around relational roles.
Dismissive-Avoidant Attachment characterizes people with low anxiety and high avoidance. They typically report comfort with independence, skepticism about the importance of emotional closeness, and low distress in response to relationship disruption. The low anxiety is genuine on self-report measures but some research suggests physiological activation (elevated cortisol, skin conductance) under attachment threat that is not reflected in self-report — consistent with the theoretical account that avoidant individuals suppress rather than lack attachment feelings. Dismissive individuals often describe themselves as self-sufficient and may find a partner's emotional needs burdensome or difficult to understand.
Fearful-Avoidant Attachment characterizes people with high anxiety and high avoidance — the combination that produces the most visible internal conflict about closeness. Fearful-avoidant individuals typically want connection but anticipate rejection or harm, generating an approach-avoidance dynamic that can look confusing from the outside: pursuing closeness and then retreating from it, oscillating between merger and distance. Originally associated with histories of abuse or disorganized early caregiving (Main & Solomon, 1990), fearful-avoidant attachment is overrepresented in clinical samples and is the category most closely associated with difficulties in emotional regulation and relationship stability. It is the smallest of the four populations by prevalence.
How This Site Measures Attachment
The assessment on this site uses an instrument derived from the Experiences in Close Relationships — Revised (ECR-R), developed by Fraley, Waller, and Brennan (2000). The ECR-R is a 36-item self-report scale that yields continuous scores on the anxiety and avoidance dimensions, not a categorical assignment. It is one of the most widely used and validated instruments in adult attachment research.
Your result page shows where you fall on each dimension — not just which quadrant you occupy. This matters because the boundary between styles is not a sharp line. A person who scores moderately on both dimensions is categorically "secure" by the quadrant rule but is not the same as someone who scores near zero on both. The continuous scores capture this; the categorical label does not.
The ECR-R was developed by factor-analyzing a large pool of attachment-related items and selecting those that best captured the two underlying dimensions. Its reliability is high (Cronbach's Alpha typically .90+ for both subscales), and its validity has been established across diverse populations including adults from multiple countries, clinical samples, and populations outside traditional undergraduate study samples.
Stability and Change
A persistent question in attachment research is how stable attachment patterns are and whether they can change meaningfully in adulthood.
The evidence suggests moderate stability. Test-retest correlations over periods of months to a few years are typically in the .50–.70 range — meaningful consistency, but far from the stability of basic personality traits like Big Five Conscientiousness or Extraversion, which tend to correlate above .70 over comparable periods. Over longer periods, stability decreases further. Longitudinal research suggests that significant positive relationship experiences — particularly a sustained relationship with a secure partner — can shift dimensional scores toward lower anxiety and lower avoidance. Negative relationship events (betrayal, loss, trauma) can shift scores in the opposite direction.
The concept of earned-secure attachment is particularly important here. Earned-secure refers to adults who score in the secure range on self-report or interview measures but who describe a history of adverse early caregiving — parents who were neglectful, abusive, or highly inconsistent. Research using the Adult Attachment Interview shows that earned-secure adults can develop parenting behaviors and relationship patterns that closely resemble those of continuously-secure adults. The pathway appears to involve developing a coherent, reflective narrative about early experiences — being able to describe difficult caregiving without either idealizing it or remaining overwhelmed by it. Earned-secure status is associated with therapy, significant supportive relationships, and the kind of reflective elaboration of early experience that the AAI specifically assesses. Importantly, some research suggests that earned-secure adults show slightly higher rates of depression than continuously-secure adults — the pathway is real, but it does not entirely erase the residue of insecure early experience.
What this means in practice: attachment style is not a personality trait that was set in childhood and will never change, but it is also not an attitude that someone can simply decide to revise. Change tends to happen through sustained relational experience and, for many people, through explicitly reflective work — therapy, deliberate examination of relationship patterns, or both.
Attachment and Other Personality Systems
Attachment style has meaningful relationships with the other frameworks on this site, though none of these relationships are so strong as to make the systems redundant.
The closest structural overlap is between the anxiety dimension of attachment and Big Five Neuroticism — both capture something about chronic negative affect and sensitivity to interpersonal threat. The correlation between attachment anxiety and Neuroticism is typically in the .40–.55 range across studies: substantial, but leaving substantial unique variance in each. High Neuroticism without high attachment anxiety looks different from the combination — Neuroticism captures general emotional reactivity across contexts, while attachment anxiety is specifically relational and specifically about abandonment and rejection.
Attachment avoidance has a weaker and less consistent relationship with Big Five traits, with modest negative correlations with Agreeableness and Extraversion but nothing approaching the Neuroticism-anxiety link.
Among Enneagram types, the preoccupied pattern is most common in types organized around emotional intensity and relational concern (Type 2, Type 4, Type 6), while the dismissive pattern is more common in types organized around self-sufficiency and emotional distance (Type 5, Type 8, some Type 1). But these are tendencies, not rules — Enneagram type and attachment style are partially independent, and the combination of the two systems often describes people with more precision than either alone.
For a deeper look at the research behind attachment, including cross-cultural evidence, the AAI's predictive record, and the therapy-and-change literature, see the Attachment Style research page. For the history of the theory from Bowlby's original formulation through the adult attachment program, see the history page.