A person turns to a close partner after a difficult day, but the partner is unavailable. Does the resulting worry reveal a fixed “attachment style,” the state of this relationship, a reasonable response to one event or some combination? R. Chris Fraley’s 2019 review Attachment in Adulthood: Recent Developments, Emerging Debates, and Future Directions asks researchers to keep those possibilities open.
This article is a critical guide to Fraley’s Annual Review of Psychology paper. The canonical Library title abbreviates the source title; the full title includes “Emerging Debates.” The guide is distinct from general Library entries on attachment or co-regulation. Its job is to explain what this review argues about adult attachment, which disagreements remain active and why a questionnaire result cannot be turned into a verdict about a person or relationship.
The review’s research question
Attachment theory was first developed to explain children’s bonds with caregivers and responses to separation. Adult researchers adapted its questions to close relationships, particularly romantic partnerships. Fraley’s review asks how attachment-related functions shift across development, how security and insecurity arise, whether relationship patterns are stable and how attachment theory should handle forms of intimacy that older models treated poorly.
This is a selective scholarly review, not an assessment manual or a new experiment. Fraley organizes emerging themes and unresolved debates rather than claiming to settle them all. The abstract names developmental change, consensual nonmonogamy, possible evolutionary interpretations of insecurity and models of thriving through relationships. A good reader notices that “attachment” here refers both to a theoretical system and to specific ways researchers measure differences among people and relationships.
Closeness is not enough to identify an attachment relationship. Attachment functions often include seeking proximity under threat, turning to someone as a safe haven and using a relationship as a secure base for exploration. Those functions can be distributed across people and can change through life. The review asks how, and under which conditions, adult partners take on roles once associated primarily with parents.
The paper matters because popular descriptions often reduce the field to a few labels. “Anxious,” “avoidant” and “secure” can become identities with an implied destiny. The research is more conditional. Measures can capture tendencies, but different relationships, experiences and times may show different patterns. The review’s debates are a reason to ask more precise questions, not to abandon the subject.
Anxiety and avoidance as research dimensions
Many self-report studies describe adult romantic attachment with two dimensions. Attachment anxiety concerns worry about availability, rejection or abandonment. Attachment avoidance concerns discomfort with dependence, disclosure or closeness. Lower levels of both are often treated as relative security in that measurement tradition. These dimensions help researchers examine variation without forcing everyone into a single category.
They are not diagnoses. A high score on a questionnaire cannot establish the cause of a person’s behavior, predict every future relationship or prove that a partner caused the pattern. Scores depend on the questions, the relationship named, the time frame and who answers. A person may respond differently about a current partner, a parent or close friends.
Measurement research has tested whether common scales behave similarly across groups. Guzmán-González and colleagues examined several brief Spanish versions of an adult romantic attachment measure across six Chilean samples, including students, community adults and couples. They tested the proposed anxiety and avoidance dimensions and developed a brief version for those study settings. That is useful psychometric work; it does not make every online attachment quiz valid. It also does not show that a score has the same meaning across every language and culture without further testing.
A score can be informative when it is connected to a research question. For example, does worry about a partner’s availability predict support-seeking during a stressful task? Does discomfort with closeness change how support is offered? Even then, the study must measure behavior and context rather than infer them directly from a label.
What childhood does and does not predict
One debate concerns continuity from early caregiving to adult relationships. A simple story says that an infant’s attachment pattern determines adult romantic attachment. Fraley’s review argues for a more complicated developmental account. Early experiences may contribute, but pathways run through many later relationships, contexts and choices. The available longitudinal evidence does not justify an unbroken one-to-one line.
Fraley and Roisman’s review of developmental antecedents makes four points. Associations between early care and adult attachment exist but can be weak and inconsistent across measurement methods. Patterns appear more malleable in childhood and adolescence than in adulthood. Early experiences do not determine adult outcomes. Relationship-specific processes and the distinction between distant and current influences need more research.
That last point has practical significance. Someone with supportive early care can encounter an unreliable adult relationship and feel insecure in it. Someone with difficult early experiences can build a trusted adult bond. Neither observation disproves that development matters. It means development is ongoing, and a current relationship has its own evidence.
Longitudinal studies improve this question by measuring people at more than one time. They still face attrition, differences between interview and self-report methods, changing partners and uncertainty about causal mechanisms. An observed correlation across decades is meaningful, but it cannot tell a particular person that one childhood episode caused their current response.
Change within and between relationships
Adult attachment is sometimes described as stable. Stability can mean a person’s ranking relative to others changes slowly, or that an individual score hardly moves. Those are different claims. Fraley’s review examines evidence for continuity and change, including the possibility that a working model is revised by later experiences.
A study of couples measured partners repeatedly across a year and investigated how their reports of relationship security related to one another. Such designs show why a partner cannot be treated merely as background context. Each person’s responses may shape and be shaped by the other’s, while the same individual can show different patterns with different people. The study does not establish that every change is mutual co-regulation or that a couple can be summarized by one score.
Research on within-person variation found that attachment security toward a particular partner can fluctuate over time. That result challenges the idea that one survey captures a permanent relationship state. It also raises a question about interpretation: is a fluctuation measurement noise, a response to a real event, or evidence of a changing bond? Repeated measures, partner reports and contextual information help distinguish these possibilities.
The Institute’s application is to describe a pattern before naming a trait. When did the person seek support, what response was available and did the pattern repeat? Was the relationship safe? Were there outside pressures such as illness, work hours or discrimination? Those questions respect the research while avoiding the temptation to explain a whole life from a label.
Relationships, stress and support
Attachment theory is especially relevant under threat, uncertainty or separation. Reviews of adult romantic attachment and stress discuss how anxiety and avoidance may be associated with different ways of regulating emotion and using support. Yet the observed behavior depends on the stressor and the partner’s response. A person may appear distant because closeness feels risky, because the partner is unavailable or because solitude is a chosen and effective way to recover.
A systematic review of attachment and social support in romantic dyads takes both partners into view. That design is valuable because support is not simply a property of the person asking for it. One partner may offer help that the other can use, misread what is needed or be constrained by their own capacity. A one-person questionnaire misses that interaction.
Association with mental health is another area needing restraint. A meta-analysis found that higher attachment anxiety and avoidance were associated with poorer mental-health indicators, with anxiety generally showing stronger associations in its pooled results. The study does not show that attachment insecurity is itself a disorder or that it alone causes depression or anxiety. Distress can alter relationships, and shared circumstances can influence both.
These findings support careful relationship research, not moral ranking. A secure relationship may provide a resource, while other supports and material conditions also matter. Therapy or education may help some people understand patterns, but the 2019 paper is not a treatment trial, and clinical decisions require clinical evidence.
Questions older models did not ask well
Fraley explicitly discusses consensual nonmonogamy as a challenge to assumptions that attachment functions belong only in one exclusive romantic pair. If people maintain more than one close consensual relationship, researchers need to ask how safe haven, secure base and commitment are organized rather than assuming the answer. The point is a research question, not a claim that one relationship form is inherently more secure than another.
Similarly, the review considers debates about the possible evolutionary function of insecure attachment. An adaptationist story can sound persuasive while being difficult to test. A pattern that may have served a function in one context is not thereby beneficial in all contexts, nor does a present-day correlation prove a historical evolutionary explanation. Fraley presents this as an area of debate requiring better evidence.
The review also asks how close relationships support thriving, not merely how they protect against distress. A secure base may enable exploration, learning and participation. Researchers should measure those positive outcomes rather than assume they follow automatically from a low-anxiety, low-avoidance score. Relationship quality is expressed in what people can do together and apart.
These questions widen the field. They encourage researchers to examine more kinds of partners, cultures, family structures and life stages. They also make the theory more demanding: constructs must be defined well enough to travel without erasing differences in how people build reliable support.
How to read an adult-attachment study
Start with the relationship being measured. Is the participant answering about romantic partners in general, one current partner, parents or a wider network? A broad tendency and a relationship-specific pattern can both matter, but they do not answer the same question. The study should say which one it needs.
Then inspect the method. Self-report captures a person’s interpretation of experiences. Interviews may assess narratives about earlier relationships. Observation can record an interaction under defined conditions. These methods overlap but are not interchangeable. If two methods yield different results, the difference may reveal that they assess different aspects rather than that one must be worthless.
Inspect timing and direction. A cross-sectional association between attachment and satisfaction cannot tell whether insecurity preceded dissatisfaction, resulted from it or developed alongside another stressor. Repeated measures can improve the picture, especially when they include both partners. Experimental changes in a short task can test a mechanism, but they do not by themselves establish long-term relationship change.
Finally inspect the claim’s scale. A group-average result does not diagnose an individual. A sample from one country or relationship type does not settle questions for all cultures and family forms. A source guide should report whether a claim is a robust association, a proposed explanation or a future research direction.
Uses for the Library and its limits
The Institute’s synthesis is to treat attachment language as a way to ask about access to dependable support and room for exploration. In a learning group or workplace, people may need reliable responses from others in order to take risks and recover from mistakes. That analogy can guide design, but it is not evidence that colleagues are attachment figures in the same sense as intimate partners.
A digital service could use the broader lesson that predictability and repair matter in relationships with institutions. A person who cannot reach help after an error may reasonably lose trust. This is an application of the paper’s concern with availability, not an empirical claim that a website has an attachment style. The service must be assessed through usability, trust and real outcomes.
The same restraint applies in personal advice. A person who recognizes a pattern may find it useful to discuss the pattern and the conditions around it. The review does not license diagnosing a partner at a distance, telling someone to stay in an unsafe relationship or declaring that one childhood event fixed their future. Safety, consent and present evidence come first.
What the evidence does not show
Fraley’s review does not prove that adult attachment is unchangeable, wholly determined by infancy or fully explained by one pair of questionnaire dimensions. It does not establish one ideal family form or relationship structure. Several of its most interesting sections are explicitly debates and research questions.
Attachment measures are not clinical diagnoses, and associations with distress do not show a single causal direction. A person can show concern about availability for reasons rooted in current behavior by a partner. A measured pattern may change when conditions change. Broad theoretical terms should not override a person’s account of safety and need.
The review also does not offer a universal intervention. Research on how relationships can increase security is promising, but a specific practice requires testing for its setting, participants, durability and possible harms. A promise to “heal attachment” from one exercise would outrun the evidence assembled in the paper.
Return to the unanswered message
The partner eventually replies. Perhaps the delay was ordinary, perhaps it was part of a repeated pattern, or perhaps the relationship is becoming less dependable. One episode cannot decide among those readings. Fraley’s review gives a framework for studying how people seek support, how partners respond and how patterns persist or change.
The most faithful use of that framework is curious rather than diagnostic. Ask which relationship, under which conditions, at which time, and with what evidence. An attachment pattern is a question about a relationship over time, not a permanent label for a person.
Research references
- Fraley, R. C. (2019). Attachment in adulthood: Recent developments, emerging debates, and future directions. Annual Review of Psychology, 70, 401–422. View source
- Fraley, R. C., & Roisman, G. I. (2019). The development of adult attachment styles: Four lessons. Current Opinion in Psychology, 25, 26–30. View source
- Hudson, N. W., Fraley, R. C., Brumbaugh, C. C., & Vicary, A. M. (2014). Coregulation in romantic partners’ attachment styles: A longitudinal investigation. Personality and Social Psychology Bulletin, 40(7), 845–857. View source
- Girme, Y. U., Agnew, C. R., VanderDrift, L. E., Harvey, S. M., Rholes, W. S., & Simpson, J. A. (2018). The ebbs and flows of attachment: Within-person variation in attachment undermine secure individuals’ relationship wellbeing across time. Journal of Personality and Social Psychology, 114(3), 397–421. View source
- Guzmán-González, M., Rivera-Ottenberger, D., Brassard, A., Spencer, R., & Lafontaine, M.-F. (2020). Measuring adult romantic attachment: Psychometric properties of the brief Spanish version of the Experiences in Close Relationships. Psicologia: Reflexão e Crítica, 33, 9. View source
- Zhang, X., Li, J., Xie, F., Chen, X., Xu, W., & Hudson, N. W. (2022). The relationship between adult attachment and mental health: A meta-analysis. Journal of Personality and Social Psychology, 123(5), 1089–1137. View source