# When Love Narrows the Future
## כשהאהבה מצמצמת את המחר

### A bounded qualitative review for Flow Hijacked

**Conceptual progression:** When Care Becomes Control → The Borrowed Self → Two Nervous Systems, One Attractor → Reopening the Relationship  
**Evidence cutoff:** 22 September 2026  
**Review type:** Critical, mechanism-focused qualitative synthesis; not a prevalence estimate, diagnosis proposal, or exhaustive systematic review

---

## Reading key: what kind of statement is being made?

This report maintains four levels of inference.

- **Directly demonstrated:** the stated result was measured in the cited study or pooled in the cited review.
- **Strongly inferred:** multiple convergent findings support the interpretation, but the complete proposed mechanism was not directly tested.
- **Theoretically plausible:** compatible with established findings, but presently indirect, population-bound, or dependent on untested links.
- **Flow Hijacked synthesis:** a proposed integrative formulation. It may organize evidence without itself being an established scientific construct.

The word **cause** is reserved for designs that warrant causal inference. “Associated with,” “predicts,” “precedes,” and “may maintain” are not treated as interchangeable.

## Review question and boundaries

The central question is: **How can people remain deeply connected without becoming responsible for keeping another person psychologically possible?** The technical question is when interpersonal help reduces immediate distress yet progressively reduces autonomous coping, behavioral flexibility, differentiation, and the futures each person can imagine or enact.

The review therefore includes only evidence that materially clarifies one of five decisions:

1. whether *codependency* is a defensible unitary construct;
2. how supportive behavior can become accommodation, substitution, or control;
3. how a person's self-worth or role can become excessively contingent on another person's state;
4. whether recurrent patterns are demonstrably reciprocal and dynamically rigid; and
5. what distinguishes capacity-building interdependence from agency-replacing regulation.

General literatures on trauma, attachment, addiction neuroscience, family therapy, personality disorder, romantic relationships, emotion regulation, or complex systems were excluded unless a source resolved one of these decisions. Evidence involving abuse, coercive control, acute danger, overdose risk, children, dependent adults, or impaired decision-making is not used to recommend tolerating harm or withholding necessary protection.

## Search and selection approach

### Design and search routes

This is a **critical, mechanism-focused qualitative synthesis**, not a systematic review, prevalence review, or bibliometric census. The objective was to adjudicate a bounded set of propositions, not to retrieve every paper that has used *codependency* or every adjacent relationship construct. Searches were updated through **22 September 2026**. No protocol for this review was preregistered, no exhaustive database export was created, and record counts at identification and screening were not prospectively logged. A PRISMA flow diagram would therefore imply a level of search comprehensiveness and auditability that this review does not have.

Evidence was located iteratively through: targeted scholarly-web searches; PubMed and PubMed Central; publisher journal and book-chapter pages; DOI-resolver and Crossref-like bibliographic metadata; institutional repositories where available; reference lists of systematic reviews and meta-analyses; and backward and targeted forward citation chaining from landmark measures, theories, and primary studies. Publisher or repository full text was sought whenever interpretation depended on methods, sample construction, temporal ordering, factor structure, numerical results, or author qualifications; where it was not obtained, the claim was restricted as described below. PubMed/PMC, DOI, publisher, and stable institutional records were also used to reconcile titles, years, journal details, identifiers, and duplicate reports. This process did **not** constitute a comprehensive search of PsycINFO, Scopus, Web of Science, or all non-English databases, and absence from the final set should not be read as proof that no such evidence exists.

Searches were organized into six bounded streams. The strings below are representative Boolean stems; terms and syntax were adapted iteratively to each platform and to the vocabulary found in anchor papers. They are reported to make the conceptual search boundaries transparent, not as a claim that a single archived query produced the corpus.

| Stream | Representative Boolean stem |
|---|---|
| **1. Codependency: history and construct status** | (`codependency` OR `co-dependency` OR `co-alcoholic`) AND (`history` OR `definition` OR `construct validity` OR `psychometric*` OR `measure*` OR `scale` OR `diagnos*` OR `critique`) |
| **2. Care, accommodation, reassurance, and control** | (`family accommodation` OR `reassurance seeking` OR `overprotect*` OR `rescu*` OR `enabling` OR `monitoring` OR `support substitution`) AND (`relationship*` OR `partner*` OR `family` OR `OCD` OR `anxiety`) AND (`experimental` OR `longitudinal` OR `prospective` OR `meta-analysis` OR `maintenance`) |
| **3. The relational organization of self** | (`differentiation of self` OR `enmeshment` OR `relationship-contingent self-esteem` OR `unmitigated communion` OR `self-silencing` OR `self-sacrifice` OR `interpersonal dependency` OR `compulsive caregiving`) AND (`couple*` OR `partner*` OR `relationship*` OR `attachment` OR `identity`) |
| **4. Dyadic regulation and dynamical evidence** | (`dyadic regulation` OR `coregulation` OR `co-regulation` OR `affective dynamics` OR `physiological linkage` OR `synchrony` OR `coupling` OR `state space grid` OR `recurrence` OR `rigidity` OR `attractor*` OR `metastab*`) AND (`couple*` OR `partner*` OR `family` OR `relationship*`) |
| **5. Addiction and concerned significant others** | (`addiction` OR `alcohol` OR `substance use`) AND (`family` OR `partner` OR `concerned significant other`) AND (`CRAFT` OR `Al-Anon` OR `enabling` OR `accommodation` OR `reinforcement` OR `natural consequences` OR `tough love` OR `family-involved treatment`) |
| **6. Capacity-building support** | (`autonomy support` OR `responsive support` OR `secure base` OR `dependency paradox` OR `goal support` OR `scaffolding` OR `self-efficacy`) AND (`couple*` OR `partner*` OR `relationship*` OR `caregiving` OR `interdependence`) |

### Eligibility and bounded use of adjacent literatures

A source was eligible for the **core set** only if it materially clarified at least one of the review's five decisions: the scientific status of codependency; the conditions under which help becomes accommodation, substitution, or control; the organization of self-worth or role around another person's state; reciprocal or dynamically rigid interaction; or a capacity-building alternative. Eligible evidence comprised systematic reviews and meta-analyses; randomized, quasi-experimental, prospective, longitudinal, intensive-repeated-measures, and dyadic studies; large or methodologically informative observational studies; psychometric work; focused qualitative studies that clarified lived processes or construct meaning; and seminal theoretical or methodological work that created an indispensable, testable construct. Foundational papers were retained for provenance and construct definition even when their designs were weak by contemporary standards; contemporary reviews and stronger prospective or experimental studies were used to judge whether the originating claim survived.

Because direct codependency research is methodologically thin, evidence from OCD/anxiety accommodation, excessive reassurance seeking, partner goal support, and parent–child dynamic-systems research was admitted only when it operationalized a proposed mechanism more precisely than the umbrella literature did. Such studies were treated as **bounded analogues**, not proof that the same process generalizes to addiction partnerships or all adult relationships. Child, clinical, student, heterosexual-couple, and Western samples were never silently generalized beyond their sampled context.

Sources were excluded from the core set when they concerned trauma, attachment, addiction neuroscience, personality disorder, family therapy, romantic relationships, emotion regulation, or complex systems only in general; used dynamical or neuroscientific language without an operational measure relevant to the proposed loop; repeated another source's argumentative job without adding stronger design or a necessary contradiction; or supported only a popular maxim. Intervention studies were retained only when they bore on a relevant interpersonal process or outcome, not merely because they involved families. Popular books, mutual-aid histories, diagnostic-system pages, and recovery materials were used only to document historical influence, cultural language, or classification status; they were not counted as empirical validation. Citation prominence identified possible landmarks but was never an inclusion criterion by itself.

The search and synthesis were English-language-emphasized. International populations and studies conducted outside English-speaking countries were eligible, but screening and close interpretation required an English full text or an interpretable English abstract/record. No systematic multilingual search or translation procedure was conducted. Paywall status alone was not an exclusion rule: when full text could not be consulted, the source could support only what was explicit in a verified abstract or authority record, unless the result was independently checked in a full systematic review or author manuscript. Sources requiring unreported methodological interpretation were replaced, triangulated, or used only as ancillary records.

### Qualitative quality appraisal and permissible inference

Quality was appraised against design-specific questions rather than collapsed into one numerical score. The hierarchy of evidence was adjusted to the claim: a psychometric paper may be the best evidence about a scale but poor evidence about causality; a seminal theory may define a construct but cannot establish its prevalence or effects.

| Evidence type | Appraisal questions emphasized | Inference permitted in this review |
|---|---|---|
| **Systematic review / meta-analysis** | Search and eligibility transparency; preregistration where applicable; overlap among primary samples; risk-of-bias appraisal; heterogeneity and moderator analysis; publication bias; whether pooled studies were mostly cross-sectional | Scope and average association, with the causal ceiling set by the primary designs |
| **Randomized or experimental study** | Randomization and allocation; active versus inert comparator; manipulation fidelity; analysis-specific sample size; attrition; dyadic clustering; ecological validity; follow-up duration | Causal language only for the manipulated contrast, measured outcome, population, and timescale |
| **Longitudinal, prospective, or intensive dyadic study** | Temporal spacing; baseline adjustment; within-person and cross-partner measurement; missingness and attrition; stable-trait and common-method confounding; whether the model tests reciprocal paths; number and density of observations | Temporal precedence or reciprocal prediction, not causation unless stronger assumptions are justified |
| **Cross-sectional observational study** | Sampling; construct specificity; shared self-report; partner or observer corroboration; confounding; clinical versus community base rates | Association and construct mapping, not direction or maintenance |
| **Psychometric study** | Item provenance and content coverage; sample adequacy; exploratory versus confirmatory structure; independent replication; reliability; measurement invariance; convergent, discriminant, incremental, predictive, and criterion validity; circularity of self-identified “known groups” | What the instrument reliably scores in the tested sample, not proof of a natural syndrome |
| **Qualitative study** | Sampling and self-selection; analytic transparency and reflexivity; evidential grounding; negative cases; transferability across role, gender, culture, and clinical context | Meaning, experience, and candidate process; not prevalence, typicality, or causal direction |
| **Theory or dynamical-methods paper** | Operational definitions; falsifiability; match between model and sampling density; stationarity assumptions; sensitivity to analytic choices; evidence of dwell, return, perturbation, or transition rather than visual resemblance | A disciplined model or method; “attractor,” “metastability,” and related terms only when the data and analysis earn them |

Across designs, additional weight was given to direct behavioral or partner-reported measures, repeated observation, credible temporal ordering, relevant comparators, transparent limitations, and replication across methods. Weight was reduced for convenience samples, single-informant cross-sectional data, small analysis cells, post hoc theorizing, poorly discriminated constructs, unmodeled dyadic non-independence, and clinical or gender-specific samples generalized as universal. No study was treated as decisive solely because it was highly cited or published in a prestigious venue.

### Deduplication, adjudication, and contradiction search

DOI was the primary deduplication key; PMID and normalized author–year–title records were secondary checks. Multiple papers from one program were retained only when they answered different questions. A review and one of its included primary studies could both remain when the review supplied coverage or a pooled estimate and the primary paper supplied a decisive design detail, temporal test, mechanism, or contradiction. When two records performed the same argumentative job, priority went to the more direct, methodologically credible, or better documented source. Bibliographic discrepancies were resolved against the primary publication or authoritative publisher record where possible; unresolved uncertainty was recorded rather than silently harmonized.

Selection and evidence grading were editorial and qualitative, not blinded duplicate screening; no inter-rater statistic is available. A source remained core only if removing it would materially weaken a mechanism judgment, contradiction, measurement conclusion, boundary condition, or historical claim. Disagreement across studies was not converted into consensus by vote: the claim was narrowed, moderators were specified, or the evidence grade was lowered. The exact deduplicated core total and source-level records are reported with the final bibliography; diagnostic/classification pages, mutual-aid history, popular books, and purely bibliographic authority records are identified separately from that count.

Contradiction search was built into every stream. Positive mechanism terms were paired with terms such as `null`, `non-significant`, `reverse direction`, `bidirectional`, `prospective`, `moderator`, `heterogeneity`, `critique`, `replication`, and `construct validity`; anchor papers were also followed to later tests and critiques. Particular effort was made to find evidence that symptoms elicit accommodation rather than only the reverse; reassurance does not uniformly produce rejection; practical and responsive support can improve goal pursuit; sacrifice can be beneficial under some motives and conditions; autonomic synchrony is heterogeneous rather than intrinsically healthy; family involvement can improve addiction outcomes; and active-comparator trials can attenuate early CRAFT findings. These findings were retained because they change the model, not treated as exceptions to be explained away.

### Source-access transparency and limitations

The companion evidence register marks access status: **FT/OA-FT** for a full report directly consulted, **A+** for an abstract-level record cross-checked against a full review, author manuscript, or publisher record, and **ABS** when the inference is confined to the verified abstract. Abstract-only records are not used for unreported sample details, factor structures, analytic decisions, or causal interpretation. Authority pages are used only for the fact they are positioned to establish, such as current diagnostic classification. Every core record reports the study type, population and sample size where relevant, finding used here, major limitation, and its nonredundant argumentative role.

The principal method limitations are the non-exhaustive and iterative search; English-language emphasis; absence of prospective screening counts, duplicate independent review, and a common formal risk-of-bias instrument; uneven full-text access; and substantial construct and population heterogeneity. Several pivotal literatures—especially codependency, differentiation, and adult dyadic rigidity—remain dominated by self-report and cross-sectional designs. The synthesis can therefore identify the most defensible mechanisms and the limits of current claims; it cannot estimate prevalence, demonstrate that all eligible studies were captured, or validate *co-constriction* as a unitary construct.

---

## 1. Executive scientific synthesis

**Bottom line.** The literature does not support *codependency* as a single, validated disorder or a clearly bounded latent trait. It is more scientifically useful as a historically influential umbrella that should be decomposed into better-defined processes: family accommodation, excessive reassurance seeking, attachment-related hyperactivation, interpersonal dependency, contingent self-worth, unmitigated communion or self-sacrifice, low differentiation, controlling or overprotective support, and reciprocal dyadic influence (Hands & Dear, 1994; Gotham & Sher, 1996; Molina et al., 2026).

The strongest evidence for a “care becomes control” mechanism comes from family accommodation in obsessive-compulsive and anxiety disorders, reassurance seeking, overprotective support, and behaviorally oriented addiction-family interventions. Across these literatures, help can produce immediate relief and can sometimes reduce exposure to corrective learning, consequences, or mastery. But three cautions are essential. First, much of the maintenance claim is correlational. Second, distress often elicits accommodation more strongly than accommodation subsequently predicts distress. Third, responsive and practical support often improves goal progress; practical help is not intrinsically autonomy-reducing (Salkovskis & Kobori, 2015; Zniva et al., 2017; Bertelsen et al., 2023; Hermida-Barros et al., 2024; Vowels & Carnelley, 2022).

The “borrowed self” is best understood not as a discovered entity but as a convergence among partially overlapping constructs. Relationship-contingent self-esteem shows that self-evaluation can fluctuate with relationship events. Unmitigated communion captures self-neglect and overinvolvement. Differentiation research concerns maintaining an “I-position” and regulating emotional reactivity while remaining connected. Attachment anxiety and excessive reassurance seeking help explain vigilance to abandonment cues and attempts to recruit certainty from another person. These strands make the proposed phenomenon recognizable, but no study has established a single pathway in which “being needed” progressively becomes a person's organizing identity (Knee et al., 2008; Starr & Davila, 2008; Calatrava et al., 2022; Helgeson et al., 2015).

Dyadic research justifies speaking about reciprocal influence, temporal coupling, recurrent interaction states, and flexibility versus rigidity. It does **not** establish a neural “codependency circuit,” literal control of one nervous system by another, or a generic codependent attractor. In relationship research, an *attractor* is defensible only when repeated observations and an explicit dynamical model identify a recurrent region of state space or a return tendency. Elsewhere it should be marked as a metaphor. Synchrony is likewise not a health metric: coupling can be adaptive, neutral, or maladaptive depending on context, timescale, direction, and the behavior being synchronized (Butler, 2011; Hollenstein, 2007; Mayo et al., 2021; Brinberg et al., 2025).

Addiction evidence most directly contradicts the idea that family involvement is inherently enabling. Significant-other-involved treatments yield modest advantages over individually delivered active treatment, and Community Reinforcement and Family Training (CRAFT) can increase treatment entry in some trials. Yet effects vary widely by modality and population, and a recent active-comparator trial found no advantage for CRAFT. Family members should not be framed as causing addiction. The scientifically safer distinction is between **causing a disorder** and **altering a specific behavioral contingency**—for example, removing a consequence, increasing access to reinforcement, or making treatment engagement more likely (Miller et al., 1999; Roozen et al., 2010; Ariss & Fairbairn, 2020; Siljeholm et al., 2024).

The healthy countermodel is **flexible interdependence**: responsiveness plus preserved authorship. Support is capacity-building when it is calibrated to need, requested or negotiated where possible, proportional to actual risk and present capacity, compatible with the recipient's goals, and followed by restored choice and action. The contrast is not dependence versus independence. Secure dependence can support exploration; the relevant contrast is support that expands a person's feasible actions versus support that repeatedly performs the action, decision, exposure, or emotional labor in their place (Feeney, 2004, 2007; Koestner et al., 2012; Vowels & Carnelley, 2022).

**Verdict on co-constriction.** As a Flow Hijacked synthesis, *co-constriction* is promising if presented as a falsifiable process model rather than a diagnosis or established construct. Immediate relief, recipient-side repetition, recurrence, and some bidirectional influence have reasonable support in specific clinical literatures; reinforcement of the provider's intervention is plausible but directly undermeasured. Progressive loss of behavioral options, role rigidity, and narrowing of perceived futures are compelling interpretations but are not yet directly demonstrated as a unified longitudinal sequence. The model therefore requires narrower wording and explicit boundary conditions.

---

## 2. The status of codependency

### 2.1 History: four traditions that should not be collapsed

The contemporary word carries at least four histories.

**Recovery and popular usage.** Al-Anon began in 1951 as mutual aid for relatives and friends of people with alcohol problems (Al-Anon Family Group Headquarters, n.d.). Early professional and recovery discourse used *co-alcoholic* for a spouse or relative whose life had become organized around the drinker. During the late 1970s and early 1980s, *co-dependent* broadened the reference beyond alcohol and beyond spouses (Haaken, 1990; Hands & Dear, 1994). Popular books—especially in the mid-1980s—then widened it further to people-pleasing, rescuing, poor boundaries, compulsive relationships, self-neglect, and “loss of self” (Beattie, 1986). Beattie popularized the term but did not coin it. Popular use is experiential and elastic; its reach is part of its appeal and part of its scientific problem.

**Clinical tradition.** Addiction professionals tried to describe recurring difficulties among family members: external focus, self-worth tied to managing another, attempts to control, emotional inhibition, and continued investment despite harm. Cermak's 1986 diagnostic proposal used criteria modeled partly on personality-disorder logic (Cermak, 1986). The proposal was influential, but there is no evidence that it became an official diagnosis or that a unitary syndrome was validated. It is safer to say the proposal was not adopted than to repeat the common but poorly documented story that the American Psychiatric Association formally “rejected” it.

**Family-systems formulations.** Bowenian and structural traditions supplied ideas of fusion, low differentiation, diffuse boundaries, triangulation, and system-stabilizing behavior. These ideas helped move attention from a defective spouse to a recurrent relational pattern. Yet *codependency* is not simply another name for any one of them: differentiation and enmeshment have their own theories and measures, and family homeostasis is a systems proposition rather than a diagnosis.

**Empirically operationalized constructs.** Researchers created scales with names such as the Spann–Fischer Codependency Scale, Codependency Assessment Tool, Friel inventory, Codependency Inventory, Codependency Assessment Questionnaire, Codependency Clinical Rating Scale, and Composite Codependency Scale. These instruments made empirical work possible, but they did not converge on a single content domain. Some emphasize external focus and self-worth; others add family-of-origin experiences, medical problems, concealment, control, self-sacrifice, or emotional suppression.

The historical conclusion is therefore not that one stable phenomenon was gradually discovered. Rather, a relational and cultural concern was repeatedly redescribed by recovery communities, clinicians, systems theorists, and measurement researchers.

### 2.2 Definitions: recurrent themes without a common boundary

Hands and Dear's critical review found no consensus definition and little evidence that the proposed syndrome was unique to relatives of people with alcohol problems. A later thematic analysis of published definitions identified four recurring elements—external focus, self-sacrifice, interpersonal control, and suppression of emotion—but recurrence across definitions is not proof that the elements form one latent construct. The most recent integrative review likewise concludes that clinical salience has outrun conceptual and empirical unity (Molina et al., 2026).

The 2026 integrative review by Molina, Taiwo, and Grey is the most current synthesis. It followed PRISMA 2020, was preregistered, searched seven databases through September 2024, and retained 30 studies. The paper itself reports two inconsistent study-type tallies—14 quantitative, nine qualitative, and seven theoretical in the abstract, versus 17 quantitative, five qualitative, and eight conceptual in the body—illustrating the difficulty of classifying this mixed literature. Six perspectives emerged: sociocultural, relational, addiction/pathology, developmental, psychoanalytic, and cognitive-personality. Across studies, codependency scores or self-identification were associated with distress, disrupted identity, and relationship problems.

That review does **not** validate codependency as a single disorder. Most quantitative work was correlational and self-report; only 11 included studies measured mental-health outcomes; attachment findings were limited and inconsistent; samples and methods varied; cultural coverage was narrow; and a single reviewer conducted risk-of-bias appraisal. The review deliberately restricted the main search to 2013–2024, so it cannot replace appraisal of foundational measurement studies. Its strongest conclusion for the present project is conceptual fragmentation plus clinical salience—not construct unity or causal direction.

### 2.3 Measurement: reliability is not construct validity

Several scales show acceptable internal consistency, known-groups differences, or correlations with theoretically adjacent variables. The revised Composite Codependency Scale is among the clearest contemporary examples. In an exploratory study of 301 community participants plus 49 members of a codependency self-help group, Marks et al. supported three factors—self-sacrifice, interpersonal control, and emotional suppression—and reported group and concurrent validity. A Norwegian study of affected family members later found that codependency-related characteristics were associated with poorer quality of life (Vederhus et al., 2019).

These findings are useful but limited in four ways.

1. **Content heterogeneity.** Factor structures and item content vary. A high score can mean self-neglect, control, suppression, distress, family history, dependency, or some mixture.
2. **Criterion circularity.** “Known groups” are often people who already self-identify with codependency or attend groups organized around the concept. Group discrimination may show that a scale captures the group's self-understanding, not a naturally bounded syndrome.
3. **Convergent without discriminant validity.** Correlation with depression, anxiety, low self-esteem, attachment insecurity, dependency, self-sacrifice, or family dysfunction confirms relevance but may show redundancy. Few studies establish incremental prediction beyond these constructs.
4. **Sampling and method bias.** Convenience samples, women partners or mothers, cross-sectional designs, and single-informant questionnaires dominate. Distress caused by living with addiction can be mistaken for a stable personal pathology.

Cullen and Carr's systemic study is an especially important falsification attempt: participants high in measured codependency were not more likely to report parental substance problems. Lindley et al. documented psychometric problems and construct overlap across measures. Bacon et al.'s interpretative phenomenological study gave depth to the experience of self-identified codependency; its purposive, self-labeling sample cannot decide whether codependency is a distinct construct.

The correct psychometric verdict is therefore **partial reliability, uncertain boundaries, and inadequate discriminant and causal validity**.

### 2.4 Overlap with better-defined constructs

For this review, most recurring content attributed to codependency can be redescribed using constructs with clearer definitions:

- **attachment anxiety / hyperactivation:** vigilance to abandonment and intensified proximity or reassurance strategies;
- **interpersonal dependency:** difficulty functioning or deciding without another's guidance or approval;
- **relationship-contingent self-esteem:** self-worth that fluctuates with relationship events;
- **unmitigated communion / self-sacrifice:** overinvolvement with others at the expense of self-care;
- **family accommodation and excessive reassurance seeking:** specific behaviors that reduce distress or prevent exposure;
- **low differentiation / fusion:** difficulty maintaining an “I-position” and regulating reactivity while connected;
- **controlling or overprotective support:** help delivered in ways that constrain choice, competence, or self-efficacy;
- **caregiver burden and stress-coping:** strain and adaptations produced by living with serious problems;
- **interpersonal control:** attempts to regulate another's behavior, affect, or access to consequences.

This is an analytic decomposition, not the result of one comprehensive item-level redundancy study. None is a perfect replacement because *codependency* combines behavior, motive, identity, relationship structure, and distress. That is precisely why it performs poorly as a scientific unit.

### 2.5 Diagnostic status

Codependency is not a diagnosis in the current *Diagnostic and Statistical Manual of Mental Disorders* (DSM-5-TR, with cumulative updates through September 2025) or the World Health Organization's ICD-11 (American Psychiatric Association, 2022, 2025; World Health Organization, 2026). Absence from these systems does not prove that the described suffering is unreal. It means there is no officially accepted criterion set, diagnostic threshold, or evidence base sufficient to treat *codependency* as a recognized disorder. Clinicians may use the word in formulation or psychoeducation; that use should not be mistaken for diagnostic standing.

### 2.6 Gender, culture, and the risk of pathologizing care

Feminist critiques identified a structural confound: behaviors called codependent—emotional labor, loyalty, self-sacrifice, responsibility for family stability—have been disproportionately expected of women (Haaken, 1990; Anderson, 1994). A construct developed largely from women partners of men with alcohol problems can convert unequal care burdens into a defect located inside the caregiver. The reverse error is also possible: because a behavior is socially expected or initially adaptive, it is not necessarily harmless.

Cultural norms also change what obligation, privacy, family decision-making, and autonomy mean. Differentiation research remains culturally concentrated, and adolescent research has separated cohesion from enmeshment while showing that enmeshment's association with wellbeing can vary by country (Manzi et al., 2006; Calatrava et al., 2022). The question cannot be whether a family is “too close” by an individualist template; it must be whether closeness permits volition, truthful feedback, role change, and safety within that context.

### 2.7 Scientific disposition of the term

The evidence favors **retain, subdivide, and demote**:

- **Retain** *codependency* as a historical, cultural, and lived-experience term that helps people name a recognizable cluster of problems.
- **Subdivide** it in scientific explanation into operational mechanisms and outcomes.
- **Demote** it from diagnosis, causal explanation, personality type, or presumption about motive.

The term should appear once near the entrance to the Flow Hijacked publication, then give way to precise language. The central phenomenon is not “a codependent person.” It is the possibility that particular forms of connection, under particular contingencies, become mutually organizing and hard to vary.

## 3. When Care Becomes Control

### Scientific synthesis

Care does not become control merely because it is intense, repetitive, practical, or relieving. The strongest literature instead supports a **functional and temporal distinction**. Help is more likely to increase capacity when it is responsive to the recipient's stated aims, proportionate to actual need, preserves meaningful choice, supports tolerable contact with difficulty, and leaves the recipient with more skill, confidence, or room to act. The same broad class of behavior becomes concerning when it is unrequested or coercive, organized mainly around ending the helper's own alarm, supplies certainty that must soon be supplied again, participates in ritual or avoidance, conceals behavior or repeatedly prevents informative consequences, or chronically performs functions the other person could safely practice. The scientific question is therefore not “Was help given?” but **what did this particular response make more and less likely, for each person, at the next relevant time point?**

Four conclusions survive close scrutiny.

1. **Immediate relief is real but not diagnostically decisive.** In OCD and related anxiety samples, reassurance is commonly followed by an immediate reduction in discomfort, and accommodation is often explicitly undertaken to reduce distress or prevent escalation (Salkovskis & Kobori, 2015; Halldorsson et al., 2016). Relief can be humane and necessary. It becomes evidence for a possible maintenance process only when the relieving response also blocks corrective learning, autonomous action, or exposure to consequences and becomes more likely to recur.
2. **Accommodation is associated with greater symptoms and impairment, but much of the evidence is correlational and reciprocal.** The most comprehensive meta-analysis found a moderate accommodation–OCD severity association (*r* = .42), yet baseline accommodation did not predict symptom improvement across treatment (*g* = −.03). Accommodation also declined during individual treatment that did not necessarily target the family (Hermida-Barros et al., 2024). Session-level and adult longitudinal work indicates influence in both directions; symptoms can elicit accommodation at least as clearly as accommodation can precede symptoms (Bertelsen et al., 2023; Kelley et al., 2024).
3. **There is limited causal evidence for “capacity substitution,” not a general law that practical help disables people.** In a small randomized laboratory study, close friends who supplied arithmetic solutions rather than procedural steps produced greater subjective and cardiovascular stress and lower perceived control when recipients later worked alone (Zniva et al., 2017). This demonstrates that the form of help can alter a recipient's response to the next solo attempt under specific conditions; it does not establish impaired performance, progressive dependence, or the same effect in long relationships. In the broader couple literature, partner support is positively associated with goal outcomes, including practical support; responsive, action-facilitating, and autonomy-supportive forms are generally beneficial (Overall et al., 2010; Koestner et al., 2012; Vowels & Carnelley, 2022).
4. **Dependable support can increase rather than reduce autonomy.** Feeney's “dependency paradox” studies found that accepting a partner's legitimate need and responding contingently predicted greater exploration, goal pursuit, and self-efficacy, including prospectively (Feeney, 2007). The relevant contrast is not dependence versus independence. It is **flexible, responsive interdependence versus rigid regulation that repeatedly substitutes for choice, learning, or role flexibility**.

This literature supports a narrow formulation of the proposed care-to-control transition:

> Care becomes controlling in function when a response that is meant to help persistently governs the other person's behavior, removes meaningful choice or feedback, or substitutes for capacities that could safely be exercised—and when the response is difficult to revise despite changing needs or accumulating costs.

“Control” here describes a function and outcome, not necessarily hostile intent. A frightened partner can constrain another person's action while trying sincerely to protect them; a reassuring family member can participate in a compulsion without endorsing it; and a person in acute danger may appropriately require directive or substitutive care. Motive, behavior, and effect must be assessed separately.

### 3.1 An evidence-informed functional heuristic: support, scaffolding, accommodation, or substitution?

Observable topography is not enough. Driving someone to an appointment may expand access, enable avoidance, or be a neutral division of labor. Paying a bill may preserve housing during a crisis, remove a consequence that would otherwise prompt change, or meet a disability-related access need. Reassurance may communicate attachment and reality-based information, or it may become a repeated certainty ritual. Monitoring may be consensual harm reduction, an agreed treatment aid, anxious checking, or coercive surveillance.

The following dimensions are evidence-informed questions, not a validated assessment instrument. They are better discriminators than labels such as *helping*, *rescuing*, or *enabling*.

| Dimension | More capacity-building | More capacity-substituting or controlling |
|---|---|---|
| **Consent and ownership** | The recipient can request, refuse, revise, and remain author of the goal. | Help is imposed, refusal is punished, or the helper becomes the de facto owner of the problem. |
| **Fit to actual capacity** | Assistance fills a genuine gap or supplies a temporary scaffold. | The helper takes over a safely manageable action or assumes incapacity without testing it. |
| **Form of aid** | Emotion is acknowledged; steps, information, transport, practice, or resources are offered. | The answer, decision, certainty, ritual, concealment, or complete performance is repeatedly supplied. |
| **Relationship to learning** | Help permits graded exposure, error, feedback, mastery, and attribution of success to the recipient. | Help reliably terminates uncertainty, ritualizes certainty, or prevents the recipient from contacting manageable outcomes. |
| **Time course** | The dose can increase in crisis and decrease as capacity returns; roles are revisable. | A temporary response becomes the only response; hand-back repeatedly fails despite adequate capacity and safety. |
| **Contingency** | Support reinforces approach, treatment participation, sober/valued activity, and problem solving. | Support is delivered contingently on avoidance, compulsions, use-related behavior, threats, or escalating distress and thereby may strengthen that sequence. |
| **Truth and consequences** | The helper protects safety without falsifying records, lying, or absorbing every ordinary consequence. | The helper repeatedly conceals behavior or prevents proportionate, non-dangerous feedback from reaching the actor. |
| **Effect on both people** | Both retain other roles, supports, and the ability to pause or renegotiate. | One becomes indispensable/vigilant and the other managed/reliant; either person's distress makes the arrangement difficult to vary. |

No single cell proves pathology. A person with severe illness or disability may appropriately receive long-term task substitution without loss of volition. Conversely, formally “offering choices” can remain controlling when choices are illusory or refusal is unsafe. Assessment requires the person's actual capacities, preferences, risks, resources, culture, and power position.

### 3.2 Accommodation: the clearest bounded model

Family accommodation in OCD is the most mature empirical analogue for the proposed mechanism. It has a defined behavioral referent: relatives change routines, facilitate avoidance, provide repeated reassurance, or participate in rituals to prevent or reduce obsessional distress. Calvocoressi et al. (1999) created the Family Accommodation Scale in 36 adult patient–caregiver dyads; the scale showed good reliability and clinically coherent associations. That achievement matters because it moves the analysis from a global identity such as “codependent relative” to countable responses during a specified period.

Across 41 studies, Wu et al. (2016) estimated a moderate association between accommodation and OCD symptom severity (*r* = .42). The updated preregistered review by Hermida-Barros et al. (2024) retained essentially the same estimate across 108 studies and 8,928 participants. Accommodation is therefore not an anecdotal feature of severe OCD. However, neither correlation identifies direction. Greater symptoms create more situations in which relatives are asked to reassure, alter routines, or join rituals; caregiver burden and conflict may affect both reports; and many studies rely on a single informant.

The newer temporal evidence is especially important. During ten sessions of youth CBT, prior accommodation predicted slightly higher anxiety at the next session, but prior anxiety predicted more subsequent accommodation with approximately twice the standardized coefficient (Bertelsen et al., 2023). In adults receiving intensive or residential OCD treatment, symptom reduction from admission to discharge predicted lower subsequent accommodation, whereas increased accommodation from discharge to one month predicted increased symptoms from one to six months (Kelley et al., 2024). These findings support a reciprocal system rather than a one-directional family cause. They remain observational within active treatment, vulnerable to time-varying treatment exposure, attrition, shared method, and the limits of cross-lagged models.

The updated meta-analysis supplies an important contradiction: baseline accommodation did **not** predict pre–post change in OCD severity across treatment, even though accommodation decreased during both individual and family-focused CBT (Hermida-Barros et al., 2024). This pattern is compatible with at least three accounts: reducing accommodation helps some patients; symptom improvement makes accommodation less necessary; or both respond to treatment. It rules out presenting accommodation as a settled primary driver.

The rare experimental evidence is promising but small. Thompson-Hollands et al. (2015) randomized 18 adult OCD–relative dyads, all receiving exposure and response prevention, to a two-session family accommodation intervention or treatment as usual. The adjunct produced large between-condition differences in accommodation and symptoms by week eight, and accommodation change temporally preceded later symptom change. The study was explicitly preliminary, extremely underpowered, demographically homogeneous, and unable to isolate which element of the family intervention mattered. Couple-integrated CBT studies likewise suggest improvement in symptoms, accommodation, and relationship outcomes, but the meta-analysis included only 15 reports, used lenient inclusion criteria, and combined many weak designs (Stewart et al., 2020).

The defensible conclusion is not “families maintain OCD.” It is:

> In OCD, some family responses enter the disorder's behavioral contingencies. Participation in rituals, facilitation of avoidance, and repeated certainty-giving can sometimes reduce distress immediately while interfering with exposure and response prevention. Symptom severity also elicits these responses, and structured family involvement can be therapeutic.

This conclusion cannot be transferred wholesale to addiction, depression, ordinary attachment needs, disability, or relationship conflict. The relevant contingency must be identified in each domain.

### 3.3 Reassurance: relief, recurrence, and the limits of the evidence

Reassurance is not one behavior. It may provide new information, communicate availability, correct a realistic misunderstanding, or function as repeated checking. Kobori and Salkovskis (2013) found distinguishable reassurance patterns in OCD, panic disorder, and healthy controls: clinical groups sought more reassurance, while OCD was marked by greater intensity and carefulness and more self-reassurance. This cross-sectional measurement work supports specificity but cannot show maintenance.

The best direct evidence for the proposed short timescale is retrospective. In 153 people with OCD, 50 with panic disorder, and 52 controls, participants recalled natural reassurance episodes. The clinical groups reported immediate relief followed by return of discomfort and renewed urges; controls did not show the same resurgence (Salkovskis & Kobori, 2015). This fits a negative-reinforcement account, but the study did not observe reassurance as it occurred, manipulate it, or show that relief increased later response probability. Memory, demand, and disorder severity could shape the reported sequence.

The provider side is even less directly tested. In qualitative interviews with ten long-term caregivers of adults with OCD, Halldorsson et al. (2016) described shared immediate relief, fear of provoking distress or anger, lack of alternatives, frustration, and relationship costs. These accounts make it plausible that reassurance is negatively reinforcing for **both** people: the recipient's doubt subsides and the provider's exposure to distress or conflict ends. A ten-person purposive sample cannot estimate prevalence, sequence, or causality. It should be used to formulate the mechanism, not certify it.

Evidence from excessive reassurance seeking in depression further weakens any universal claim. Starr and Davila's (2008) meta-analysis found a moderate association with depressive symptoms but only a small association with interpersonal rejection (about *r* = .14), stronger when rejection was self-reported; prospective findings were inconsistent. In two couple studies, Shaver et al. (2005) found that attachment anxiety explained substantial overlap between reassurance seeking and depression, and diary associations varied by partner, gender, and attachment. The popular cascade—reassurance inevitably causes rejection, which inevitably causes more reassurance—is not established.

Nor is the humane alternative emotional withdrawal. In a small counterbalanced scenario experiment with 36 people with OCD, participants imagined receiving either certainty-oriented reassurance or emotional support. Support reduced the anticipated urge to seek reassurance and was rated as more calming, connecting, and helpful for managing emotion (Causier & Salkovskis, 2025). Because the outcomes were imagined, immediate, and self-reported, this is preliminary evidence rather than a clinical efficacy trial. It nevertheless demonstrates an important design principle: **a partner can refuse to certify certainty while still joining the person emotionally**.

For the lecture, negative reinforcement should therefore be expressed conditionally:

> When reassurance or intervention reliably ends uncertainty or distress for one or both people, the relief may increase the likelihood that the same interpersonal response is used again. In OCD, several components of this sequence are supported; the complete two-person reinforcement loop has not yet been demonstrated prospectively across repeated natural episodes.

### 3.4 Overprotection and capacity substitution

“Capacity substitution” is useful only if kept behavioral. It means that one person supplies the answer, action, decision, or regulation that the other person could safely attempt with less intrusive support. It does **not** refer to necessary assistance, consensual division of labor, supported decision making, emergency intervention, or disability accommodation.

Zniva et al. (2017) provide the cleanest experimental analogue. Forty-four undergraduates completed arithmetic tasks while a close friend provided no support, autonomy support (procedural steps), or overprotective support (solutions). In a later solo stress challenge, those who had received solutions showed greater subjective stress, heart rate, and diastolic blood-pressure responses and felt less control; procedural support did not produce these adverse responses. Because the manipulation held the task and relationship roughly constant while changing the form of help, it supports a narrow short-term causal claim: receiving solutions rather than procedural guidance can increase stress and reduce perceived control during the next unaided task. The study does not by itself show impaired task performance. The tiny student sample, artificial task, uncertain matching to recipient preferences, and absence of follow-up prohibit claims about chronic dependency or close relationships generally.

The broader support literature is a necessary corrective. Vowels and Carnelley's (2022) preregistered multilevel meta-analysis synthesized 195 effects from 36 samples and 10,130 participants. Partner support was positively associated with goal outcomes overall (*r* = .25); responsive and practical support were positive, while negative support—control, coercion, interference, or criticism—was negative. Practical support was less strongly tied to self-efficacy than responsive support, a pattern consistent with but not proof of substitution. Only about one fifth of included studies were longitudinal, measurement quality was often weak, and causal direction remained unclear: people who make progress may elicit more support.

Prospective and observational studies sharpen the distinction. Across three studies, Koestner et al. (2012) found that autonomy support—perspective taking, acknowledging choice, and supporting self-endorsed goals—predicted goal progress and relationship wellbeing more consistently than directive positive guidance. In 47 couples observed discussing desired self-improvement and followed quarterly for one year, nurturing and action-facilitating support predicted growth and relationship quality, whereas criticism and invalidation did not (Overall et al., 2010). Feeney and Collins (2003) showed in 194 couples that caregiving motives and styles are heterogeneous: responsive care was distinguishable from compulsive or controlling care and predicted trust and healthier relationship functioning. These studies are mostly self-report or observational and cannot fully separate partner effects from recipient traits, goal feasibility, or pre-existing relationship quality.

Feeney's (2007) dependency-paradox studies provide direct counterevidence to the idea that receiving help necessarily narrows agency. Across observational, prospective, and assistance paradigms, responsive acceptance of need predicted greater independent exploration, self-efficacy, and goal completion. The work did not test unlimited or needless dependence and mostly sampled different-gender US couples. Its bounded implication is nevertheless important: **reliable access to another person can make autonomous action safer.**

Support delivery also matters. A 32-day diary study of 68 couples facing a bar examination found that provider-reported support that recipients did not register as an explicit transaction was associated with better adjustment, whereas consciously received support sometimes carried affective costs (Bolger et al., 2000). Three experiments with 257 women anticipating a speech found that subtle or “invisible” support reduced distress without the same efficacy threat produced by conspicuous aid (Bolger & Amarel, 2007). These results should not be converted into advice to manipulate or hide help. Visibility was confounded with evaluative meaning in some paradigms, the samples and stressors were narrow, and later work shows motivation and responsiveness can change the effect. The safer inference is that **help can imply “you cannot cope” unless its delivery preserves competence and dignity**.

### 3.5 When monitoring or protection becomes control

The core source set does not support a universal threshold at which monitoring becomes control. A defensible analysis asks what monitoring is for, how agreement is obtained, what happens after refusal, and whether surveillance expands or contracts as risk changes.

Monitoring may be justified when there is immediate overdose risk, suicidality, severe incapacity, cognitive impairment, danger to children or dependent adults, or an agreed treatment plan. It may also be a consensual harm-reduction tool: medication support, shared financial limits, breath testing, or check-ins can increase safety and agency when roles, duration, data access, and exit conditions are jointly specified. Monitoring becomes more controlling when it is secret, disproportionate, punitive, difficult to refuse, used beyond the period of danger, or replaces the monitored person's participation in decisions. The present source set can support this normative-functional framework, but it does not supply a validated “monitoring-to-control” dose curve.

The same caution applies to overprotection. The laboratory evidence suggests that solving a manageable problem for someone can harm the next solo attempt; it does not justify withdrawing support from a person whose current capacity is genuinely impaired. Overprotection is defined relative to need and preference. Without an independent assessment of capacity, risk, and consent, the label can simply encode an outsider's preference for independence.

### 3.6 Moderators and exceptions

The transition from care to constriction is conditional on at least eight moderators.

- **Actual danger and incapacity.** In overdose risk, psychosis, medical crisis, severe withdrawal, violence, exploitation, or impaired decision-making, directive action can be necessary. Immediate safety and later agency-building are different time scales.
- **Requestedness, responsiveness, and fit.** Help that matches the recipient's stated need tends to function differently from unsolicited correction. Responsive support can include practical aid; autonomy support does not mean verbal encouragement while withholding needed resources.
- **Task learnability.** Substitution is most likely to displace capacity when the task is safe, within reach, and learnable. It is less meaningful where a stable impairment makes unaided performance impossible or disproportionately costly.
- **Chronicity and hand-back.** Temporary takeover in crisis is not the same as a fixed role. Concern rises when roles cannot be renegotiated after danger or incapacity changes.
- **The exact contingency.** Providing warmth after distress is not equivalent to providing certainty after a compulsion, money following substance use, or concealment after a harmful act. Only the latter response–outcome relations can be evaluated as maintaining a specified behavior.
- **Helper fear, burden, and alternatives.** Accommodation may be the safest response available in a household with threats, escalation, scarce treatment access, financial dependence, or caregiving overload. Calling it “control” can erase constraint.
- **Power and relationship safety.** Mutual influence is not equal responsibility. A response made under coercion, fear, or material captivity cannot be interpreted as voluntary participation in a symmetrical loop.
- **Culture and disability.** Interdependence, obligation, and shared decision making have different meanings across cultures and access contexts. Autonomy is volition and authorship, not performing every task alone.

These moderators rule out simple behavioral commandments. “Stop rescuing,” “let consequences happen,” and “do not reassure” are not scientific conclusions. Each can be unsafe or countertherapeutic when detached from risk, function, treatment plan, and available alternatives.

### 3.7 Contradictions that must remain visible

1. **Symptoms evoke care.** The association between accommodation and severity cannot be read only as care causing symptoms. In the strongest session-level test, anxiety more strongly predicted subsequent accommodation than accommodation predicted anxiety (Bertelsen et al., 2023).
2. **Accommodation can fall because symptoms improve.** Baseline accommodation did not predict treatment change in the updated meta-analysis, while accommodation declined during individual CBT (Hermida-Barros et al., 2024).
3. **Structured family involvement can improve outcomes.** Couple- and family-integrated CBT can reduce symptoms and accommodation; “family involvement” is not the risk factor. The risk lies in particular contingencies (Stewart et al., 2020).
4. **Practical support usually correlates with better goal outcomes.** The meta-analytic average is positive, not harmful. Control, coercion, criticism, and interference show the negative association (Vowels & Carnelley, 2022).
5. **Dependable care can enable exploration.** Accepting legitimate dependence can increase later independence, the opposite of a simple use-it-and-lose-it account (Feeney, 2007).
6. **Reassurance does not uniformly generate rejection.** The rejection association in the ERS literature is small and method-dependent, and daily effects vary across partners (Starr & Davila, 2008; Shaver et al., 2005).
7. **Immediate regulation is not evidence of long-term harm.** Calming another person may prevent overwhelm and enable later action. Harm requires evidence that the particular response blocks learning, choice, feedback, or role revision—not merely that distress decreased.
8. **The provider's motive cannot be inferred from effect.** A response may constrain agency without a conscious need to dominate; conversely, apparently autonomy-supportive language can be coercive when refusal has costs.

### 3.8 Evidence ruling for the care-to-control transition

| Proposed component | Ruling | Boundary of the claim |
|---|---|---|
| A partner response can reduce distress immediately. | **Strongly supported in bounded contexts** | Directly supported for recalled reassurance episodes and consistent with accommodation reports; not every kind of help or diagnosis. |
| Recipient relief can make reassurance, avoidance, or help-seeking recur through negative reinforcement. | **Reasonably supported in bounded anxiety/OCD paradigms; naturalistic causal evidence remains incomplete** | Immediate relief and renewed urges are documented, but repeated natural episodes rarely test whether relief changes the recipient's next-response probability. |
| Recipient or provider relief reinforces the helper's accommodation, reassurance, or rescue. | **Plausible but directly undermeasured** | Small qualitative studies describe provider relief and conflict reduction; prospective studies rarely test whether either form of relief increases the provider's next response. |
| Accommodation can maintain avoidance, ritual, or anxiety. | **Reasonably supported for OCD/anxiety** | Association, temporal evidence, and a small adjunctive RCT converge; reverse influence and shared treatment change are substantial. |
| Practical help can replace rather than build capacity. | **Plausible with one clean short-term experiment** | Supplying solutions increased stress and reduced perceived control in a subsequent solo task; impaired performance and chronic relational dependence were not demonstrated. |
| Responsive and autonomy-supportive care can build capacity. | **Reasonably to strongly supported** | Meta-analytic, prospective, observational, and experimental evidence converges, though measures and samples are often narrow. |
| Reassurance progressively disables autonomous regulation. | **Not established as a general claim** | Supported only under specific ritualized/avoidant functions; ordinary reassurance may inform, soothe, reconnect, or enable approach. |
| Monitoring/overprotection inevitably narrows agency. | **Unsupported without context** | Risk, consent, actual capacity, treatment agreement, and duration determine function. |
| The helper and recipient mutually maintain a fixed rescue loop. | **Plausible but not directly demonstrated in full** | Reciprocal accommodation is supported; the full sequence from distress through dual relief, recurrence, reduced coping, and role rigidity has not been measured. |
| Reduced consequences cause recovery or behavior change. | **Unsupported as a general rule** | “Natural consequences” is not a standardized intervention and may create danger. Contingency must be specified; safety cannot be treated as reinforcement noise. |

### 3.9 Publication formulation

> Helping is not the opposite of agency. Good support often lends calm, information, labor, or confidence until a person can move again. The risk begins when relief becomes the only criterion of success: the doubt must be answered, the consequence erased, the task taken over, the distress stopped now. In some clinical settings—most clearly OCD—those responses can enter a reciprocal learning loop. The distressed person may learn where certainty or escape can be obtained; the helper may learn how to end alarm, conflict, or helplessness, although this provider-side learning remains undermeasured. But the loop is not universal, and symptoms also call forth care. The question is whether support leaves each person with more possible action afterward, or whether both become increasingly dependent on repeating the same roles.

The section should not imply that love “becomes control” whenever one person calms, rescues, monitors, or performs tasks for another. Nor should it claim that the helper causes the disorder, secretly needs the other person to remain ill, or should withdraw support so that “natural consequences” can work. The scientifically defensible target is a **specified, repeated, context-dependent contingency** that reduces immediate distress while interfering with choice, learning, truthful feedback, or safe role revision.

## 4. The Borrowed Self

“The borrowed self” is a **Flow Hijacked metaphor**, not an established construct, diagnosis, or validated personality type. Its most defensible referent is not a person who “has no self,” but a configuration in which self-worth, voice, felt safety, and available action become unusually contingent on a close other’s state. The relevant literatures do not converge on one syndrome. They describe partially overlapping processes: low differentiation, relationship-contingent self-esteem, self-silencing, unmitigated communion, submissive dependency, attachment hyperactivation, excessive reassurance seeking, and compulsive or controlling caregiving. None alone is equivalent to codependency, and their correlations do not establish a single developmental or causal pathway.

The evidence supports a narrower proposition: some people have more difficulty preserving self-definition, volition, and role flexibility when a close other is distressed, dissatisfied, withdrawing, or in need. It supports much less strongly the claim that one person’s identity is progressively “taken over” by the relationship, and it does not yet demonstrate that two people mutually produce a unified “borrowed self” over time. The scientific task is therefore to specify what is contingent, whose behavior changes, and over what timescale.

### 4.1 Differentiation: a capacity within connection, not freedom from need

Bowen-derived differentiation combines an intrapersonal capacity—regulating emotion without losing reflective choice—with an interpersonal capacity: remaining connected while holding an “I-position.” The Differentiation of Self Inventory and its revision operationalize emotional reactivity, emotional cutoff, fusion with others, and I-position. The measurement history is cautionary. The original Fusion With Others subscale had weak reliability and partly confused maladaptive fusion with mature concern; revision improved internal consistency but did not eliminate conceptual overlap with attachment insecurity (Skowron & Friedlander, 1998; Skowron & Schmitt, 2003). A later validation supported two higher-order domains, affect regulation and interdependent relating, rather than one simple essence (Jankowski & Hooper, 2012).

**Directly demonstrated:** self-reported differentiation is repeatedly associated with personal and relationship adjustment. A two-country, two-wave study of 479 heterosexual couples found that baseline couple-level differentiation predicted some later relationship outcomes in Spain and the United States (Rodríguez-González et al., 2023). **But the causal claim is weak:** Calatrava et al.’s (2022) scoping review found that 250 of 295 studies were cross-sectional, all differentiation measurement was self-report, and intervention and stability evidence was scarce. The longitudinal couple study used different measures and follow-up intervals between countries, had only two waves, and operationalized “couple differentiation” by summing partner scores. It establishes temporal ordering for some associations, not a mechanism by which a self becomes borrowed.

Nor is enmeshment a synonym for closeness. In 264 married couples, intrusiveness was empirically separable from closeness-caregiving, and many spouse-behavior dimensions were not reciprocal (Werner et al., 2001). In British and Italian adolescents, cohesion predicted better well-being in both samples, while enmeshment predicted poorer well-being only in Britain (Manzi et al., 2006). The Enmeshed subscale of FACES-IV has also shown weak reliability and construct problems in a pediatric-oncology sample (Marsac & Alderfer, 2011). These findings directly challenge any inference from emotional intensity, family contact, loyalty, or sacrifice to pathology.

The best definition for this publication is therefore: **differentiation is the capacity to remain emotionally connected while preserving regulation, self-definition, and volitional choice.** It should not be translated into emotional distance, rigid boundaries, reduced need, or Western individualism. Self-determination theory defines autonomy as volition rather than separateness (Ryan & Deci, 2017). Across student samples in the United States, Russia, Turkey, and South Korea, self-endorsed motivation for both individualist and collectivist practices predicted well-being (Chirkov et al., 2003).

### 4.2 A relational self is not necessarily a contingent self

The cleanest evidence for the metaphor concerns **relationship-contingent self-esteem** (RCSE). Across four studies, including daily and twice-daily assessment, Knee et al. (2008) found that people higher in RCSE showed larger momentary self-esteem changes as relationship events became more positive or negative, beyond attachment anxiety, general contingent self-worth, inclusion of the other in self, and baseline self-esteem. This is a comparatively strong within-person demonstration that relational signals can become amplified inputs into self-evaluation.

It is not evidence that relational identity is inherently fragile. Cross et al. (2000) found that a relational-interdependent self-construal—defining oneself partly through close relationships—was associated with closeness, commitment, considering others in decisions, and, in an interaction study, partner-perceived openness and responsiveness. The distinction is not between a self “inside” versus “outside” relationships. It is between:

- a **relational self**, in which close others meaningfully inform identity while worth and action retain multiple supports; and
- **contingent self-organization**, in which relationship cues disproportionately govern worth, voice, safety, or available action.

Even RCSE does not map cleanly onto dysfunction. In Knee et al.’s dyadic study, couples in which both partners were high in RCSE reported greater commitment, without greater satisfaction or closeness. The samples were predominantly young, female, US, and heterosexual, and the studies were nonexperimental. **Strong inference:** relational contingency can increase volatility and vulnerability. **Not directly demonstrated:** that it necessarily reduces agency, produces enmeshment, or narrows a person’s future.

The popular idea of an identity organized around “being needed” is more weakly supported. Themes of responsibility and indispensability appear in measures of unmitigated communion, self-silencing, and caregiving motives. They have not been consolidated into a well-validated unitary construct with demonstrated incremental validity. Flow Hijacked can use “need to be needed” as a phenomenological description or hypothesis about motive, but should not present it as an established mechanism.

### 4.3 When care excludes the caregiver’s self

Unmitigated communion (UC) was designed to distinguish positive communion from concern for others that excludes concern for self. Multi-method work links UC—not communion itself—to negative self-view, reliance on others for self-evaluation, self-neglect, overinvolvement, interpersonal difficulty, and distress. Aubé’s (2008) research included peer report, a community follow-up, and a diary study. In a ten-day diary, people higher in UC reported more support provision and more interpersonal difficulty; support and over-nurturance were positively related to well-being mainly among those lower in UC (Helgeson et al., 2015). This is consistent with the possibility that the same observable behavior—helping—functions differently depending on whether self-concern and choice remain available.

The limitation is not incidental. UC scales combine motive, worry, attention, low self-regard, and behavior; many studies use small Western student samples; and shared negative-affect content can inflate associations with distress. **Directly demonstrated:** a measurable pattern of self-excluding other-focus covaries with poorer adjustment. **Theoretically plausible:** this pattern makes a caregiving role unusually hard to relinquish. **Not established:** that UC is codependency, that it develops because another person depends on the caregiver, or that care itself produces loss of self.

Self-silencing research adds voice and power. The Silencing the Self Scale was developed from women’s accounts of suppressing expression to preserve relationships and initially showed associations with depression (Jack & Dill, 1992). A recent synthesis of 126 studies found recurrent associations with depressive symptoms, conflict, inequity, and low mutuality, but causal direction remained uncertain because most evidence was cross-sectional and Western, White, heterosexual, and binary-gendered (Jack et al., 2026). The scale does not function identically across genders: in one validation, some items appeared to measure intimacy avoidance or threatened independence among men rather than silence used to preserve connection (Remen et al., 2002).

Silence also cannot be interpreted without power and safety. Withholding speech can be strategic protection where disclosure risks retaliation, exclusion, financial harm, or violence. Locating the problem solely in the silenced person’s “boundary difficulty” would mistake adaptation to constraint for an intrapsychic deficit. The appropriate question is whether nonexpression is freely chosen and revisable, or compelled by fear, dependence, unequal power, or actual danger.

Sacrifice supplies an essential contradiction to a simple loss-of-self narrative. A meta-analysis of 82 datasets and 32,053 participants found that willingness and satisfaction with sacrifice were generally positively associated with personal and relationship well-being; enacted sacrifice had a small negative association with personal well-being, and perceived costs were negative (Righetti et al., 2020). In 80 couples, approach-motivated sacrifice—moving toward valued relational outcomes—predicted better emotional and relational outcomes, whereas avoidance-motivated sacrifice—preventing conflict, rejection, or guilt—predicted worse outcomes for giver and recipient (Impett et al., 2014). These findings are prospective and dyadic but not randomized tests of motive.

Caregiving studies point in the same direction. Security has been associated with responsive rather than compulsive caregiving, with effects moderated by gender and marriage duration (Feeney, 1996). A multimethod study of 194 couples found that caregiving behavior reflected caregiver-side knowledge, motives, trust, and interdependence as well as the recipient's need (Feeney & Collins, 2001). In a later study of 194 young couples followed for two to three months, distinct caregiving motives predicted different styles of care; responsive caregiving predicted recipient trust and relationship functioning, whereas compulsive and controlling caregiving were distinguishable (Feeney & Collins, 2003). The meaningful contrast is therefore not care versus selfhood, but **responsive, consent-sensitive care versus care that becomes inflexible, intrusive, self-regulatory for the giver, or substitutive for the receiver**.

### 4.4 Dependency, attachment hyperactivation, and reassurance

Dependency is heterogeneous. Pincus and Wilson (2001) distinguished submissive dependency, exploitable dependency, love dependency, and low dependency. Submissive dependency was more closely linked to fearful attachment and loneliness; love dependency could coexist with secure and affiliative patterns. The Relationship Profile Test likewise separates destructive overdependence, dysfunctional detachment, and healthy dependency (Bornstein et al., 2002). These are primarily cross-sectional validation studies, so they establish discriminability more securely than developmental cause. They nevertheless invalidate a one-dimensional scale running from dependent/bad to independent/good.

Attachment anxiety offers one account of why a close other’s availability can become unusually salient. Hyperactivating strategies include intensified attention to rejection cues, proximity seeking, and reassurance seeking. But attachment categories should not be used as fixed identities, and reassurance is not inherently pathological. Starr and Davila’s (2008) meta-analysis found a moderate concurrent association between excessive reassurance seeking and depressive symptoms across 38 studies (aggregate *N* = 6,973), but a much smaller association with interpersonal rejection across 16 studies (mean *r* approximately .14). Associations were stronger when rejection was self-reported; prospective results were inconsistent. In two couple studies, attachment anxiety accounted for substantial ERS–depression overlap, while daily reassurance did not uniformly erode relationship quality; effects varied by attachment anxiety and gender (Shaver et al., 2005).

This is a direct challenge to an intuitively appealing loop in which reassurance inevitably creates rejection and greater dependency. Depression, conflict, or actual partner unreliability may elicit reassurance; reassurance may sometimes repair uncertainty; and effects may be reciprocal or conditional. **Reasonably supported:** attachment-related vigilance and repeated reassurance can organize attention and behavior around another person’s signals. **Plausible but indirect:** repeated external regulation can reduce opportunities for self-generated coping in some contexts. **Unsupported as a general law:** reassurance progressively disables autonomous regulation.

### 4.5 Evidence ruling: what “the borrowed self” may safely mean

The strongest formulation is a **loss of flexibility under relational threat**. A person’s self-evaluation may fluctuate sharply with relationship events; voice may be suppressed to preserve connection; care may become difficult to withhold even when intrusive or substitutive; and dependency may become submissive, exclusive, or fear-governed rather than responsive. Those processes can plausibly stabilize a role in which one person is chronically needed and the other chronically managed. Existing studies, however, rarely measure both partners repeatedly enough to show that role stabilization, and they do not establish a single syndrome.

Three guardrails are therefore non-negotiable.

1. **Connection is not the pathology.** Closeness, relational identity, willing sacrifice, reassurance, and dependence can all be adaptive. The target is rigidity, coercion, or loss of alternatives—not relational intensity.
2. **Culture, gender, and power change meaning.** Expectations of caregiving and emotional labor are gendered; much of the evidence comes from Western, heterosexual, self-report samples. Autonomy should be assessed as volition within a cultural context, not resemblance to individualist independence. Silence, deference, or material support may also reflect constrained options rather than psychological fusion.
3. **Mutuality must be demonstrated, not presumed.** One person’s measure of fusion, contingent worth, or distress cannot establish a two-person feedback loop. “Borrowed self” should describe a proposed pattern of contingent organization, not assign a character defect or claim that one partner literally owns the other’s identity.

The healthy endpoint is **flexible interdependence**: both people can need, give, refuse, disagree, receive, repair, and change roles without making the relationship—or one person’s psychological viability—depend on a single fixed arrangement. The self is not purified of relationship. It becomes more capable of remaining in relationship without surrendering authorship.

## 5. Two Nervous Systems, One Attractor

The dyadic literature supports a restrained systems claim: close partners can become temporally interdependent, and some interaction patterns become recurrent, narrow, and difficult to vary. It does **not** support the literal image of two nervous systems merging into one, nor does it establish the proposed distress–rescue–relief–reliance sequence as a single demonstrated mechanism. “One attractor” can therefore remain in the title as a disciplined metaphor, provided the publication defines what has actually been measured and marks the untested links.

### 5.1 What dyadic regulation means—and does not mean

Dyadic regulation is scientifically earned when temporally ordered data show that one person's state or action helps predict change in the other's beyond each person's own prior trajectory. Depending on the design, researchers can estimate lagged influence, contingent transitions, damping or amplification around an equilibrium, occupancy of joint states, and return to a previously visited state. Such work demonstrates more than similarity between partners: it tests whether their time series are statistically coupled (Butler, 2011; Butler & Randall, 2013; Helm et al., 2014; Reed et al., 2015; Randall et al., 2021).

Temporal dependence is not automatically a behavioral contingency. A lagged association shows that one time series helps predict another; a functional reinforcement claim additionally requires a specified response–consequence relation and evidence that the consequence changes the later probability of the response. This review therefore uses **temporally contingent responding** for sequential dependence and reserves **behavioral contingency** for the narrower functional claim.

The finding is real but bounded. In a laboratory study of 42 different-gender couples discussing a shared stressor, both partners' recalled stress tended to damp, but the estimated cross-partner effect ran mainly from women's stress to men's. Positive dyadic-coping reports did not moderate the dynamics as predicted; negative coping did (Randall et al., 2021). In 101 couples, cardiovascular dynamics were more stabilizing during support than control discussions, especially when men received support, yet there was substantial between-couple heterogeneity (Zee & Bolger, 2023). These studies make short-term interpersonal stabilization **reasonably supported**. They do not show that stabilization increased later capacity, reduced it, or was caused by a behavior properly described as rescue.

The correct empirical unit is always a particular signal, behavior, task, time scale, model, and sample. A cross-lag in respiratory sinus arrhythmia, a transition between coded affect states, and a coupled oscillator coefficient do not measure one generic substance called “co-regulation.” Model assumptions also matter. Oscillator models presuppose approximately oscillatory dynamics; fixed-parameter models can conceal relationships that change within an episode or across days. Simulations and an idiographic couple series show that a stationary vector-autoregressive model can obscure time-varying partner effects, while the more flexible model requires much longer series and sacrifices precision (Bringmann et al., 2018). Model fit is evidence for a specified representation, not proof that a relationship “is” an oscillator.

Mutuality likewise must be demonstrated rather than assumed. Some models find bidirectional influence; others find a stronger path in one direction, changing direction over time, or no reliable cross-partner effect. “Dyadically maintained” can mean that each person's behavior enters the other's predictive sequence or, where consequences are measured, the other's behavioral contingencies. It does not mean equal causal contribution, equal power, equal freedom to act, or equal responsibility for harm. This distinction is indispensable wherever there is coercion, dependency created by disability or material conditions, or a marked power asymmetry.

### 5.2 Synchrony is coordination, not a health score

Behavioral and physiological covariation is often treated as if it directly indexed intimacy, empathy, or relational health. The evidence does not permit that interpretation. Couple studies have reported linkage in blood pressure, heart-rate indices, respiratory sinus arrhythmia, electrodermal activity, respiration, and cortisol. But the direction, lag, and meaning vary across signals and contexts (Levenson & Gottman, 1983; Liu et al., 2013; Reed et al., 2013; Helm et al., 2014; Timmons et al., 2015; Palumbo et al., 2017).

The strongest focused quantitative synthesis found only a small, statistically nonsignificant overall association between autonomic synchrony and relational outcomes, with high heterogeneity; sympathetic and parasympathetic indices did not behave alike (Mayo et al., 2021). Several findings directly contradict “more synchrony is better.” Physiological linkage was greater during marital conflict and among less satisfied couples in a foundational laboratory study (Levenson & Gottman, 1983). In another study, perceived partner influence and demand–withdraw behavior shifted blood-pressure linkage toward in-phase coordination, without the same pattern across every physiological channel (Reed et al., 2013). Naturalistic cortisol data from 19 couples showed same-day similarity, especially under strain, but no significant next-day cross-lagged partner effect (Liu et al., 2013). In two larger experiments, inducing co-rumination did not increase sympathetic covariation, and some dyads showed negative rather than positive covariation (DiGiovanni et al., 2024).

Synchrony can reflect responsive attunement, shared escalation, turn-taking, attempted influence, or simple exposure to the same room, topic, posture, breathing rhythm, meal, waking time, or stressor. Autocorrelation and analytic choices can also generate apparent linkage. Accordingly, synchrony is **directly demonstrated as a context-dependent pattern of covariation**, but its psychological function is underdetermined without concurrent behavior, temporal direction, comparison against appropriate null models, and downstream outcomes. It should never be used as shorthand for secure connection, pathology, a shared neural state, or one person “regulating another's brain.”

### 5.3 Rigidity, flexibility, and observed behavioral repertoire

State-space grids provide the clearest empirical bridge to the idea of a relationship that has become narrow. Two synchronized behavior streams are plotted as joint states. Investigators can then quantify the number of states visited, concentration of occupancy, transition frequency, dispersion or entropy, dwell time, and return. These are genuine measures of the dyad's **observed behavioral repertoire during a defined task** (Hollenstein, 2007). They are not measures of agency, identity, imagination, or “possible futures” unless those outcomes are assessed separately.

Most prospective evidence comes from parent–child research, not adult caregiving relationships. In an at-risk community cohort initially comprising 270 families, interaction rigidity at age five was modestly associated with higher and increasing externalizing problems across the kindergarten year (Hollenstein et al., 2004). In small clinical treatment studies, improvement sometimes coincided with increased interactional flexibility or with a temporary peak in recurrence entropy (Granic et al., 2007; Lichtwarck-Aschoff et al., 2012). A perturbation—asking a parent and child to end an interaction—also revealed different reorganizations in two clinical subgroups: one remained permissive, whereas another shifted toward mutual hostility (Granic & Lamey, 2002). These findings show that a dyad's response to disturbance can expose organization not visible in average behavior.

They do not justify “maximum variability is health.” In 998 community families, short conflict bouts and the combination of peaceful-resolution content with *low* transition entropy predicted lower later antisocial behavior (Dishion et al., 2012). In another longitudinal study, flexibility interacted beneficially with shared positive affect, but mother–child and father–child flexibility showed different main effects (Lunkenheimer et al., 2011). A dyad that efficiently settles into calm problem-solving may display a concentrated repertoire because the task has been solved; a dyad chaotically moving among hostile states may look statistically flexible while functioning poorly.

The defensible principle is therefore **adaptive range**, not variability for its own sake: enough organization to coordinate, enough flexibility to change strategy when context or goals change. A relational pattern becomes concerning when its narrowness is content-specific, persistent despite changing demands, costly to one or both people, and associated with impaired recovery, role change, or independent action. Evidence for this principle is **reasonable in developmental family research but indirect for adult codependent-like relationships**.

### 5.4 What “attractor” can legitimately mean

In classical dynamical systems, an attractor implies convergence from a surrounding set of starting conditions and stability under perturbation. Much relational research uses the word more modestly. In a state-space grid, an “attractor” may be an investigator-defined or data-derived region with disproportionate occupancy, long dwell, or repeated return. That is an operational description of recurrence, not evidence of a hidden force, a neural basin, inevitability, or a self-sustaining relationship essence.

Brinberg et al. (2025) offer an instructive boundary case. Eighty-one friend dyads spent 59% of support conversations in an a priori “problem-description attractor,” yet time in that region, timing of exit, and entropy generally did not predict emotional improvement or perceived support quality. An operational attractor can therefore be common without being harmful or outcome-relevant. Likewise, treatment studies in small, uncontrolled parent–child samples report “phase-transition-like” reorganization, but they cannot demonstrate a physical phase transition or establish it as the mechanism of therapeutic change (Granic et al., 2007; Lichtwarck-Aschoff et al., 2012).

For Flow Hijacked, the preferred phrase is **attractor-like interaction pattern**: repeated return to a limited set of observable responses. Unqualified technical use of *attractor* should be reserved for a methods passage that states how the region was defined, what recurrence or convergence was measured, over what time scale, and whether the pattern survived perturbation or out-of-sample testing. *Metastability* and *phase transition* do not yet earn substantive roles in the lecture. They may be useful research hypotheses, but the present evidence does not make them necessary to explain the central relational mechanism.

The same restraint applies to **family homeostasis**. The idea is historically influential, but adult codependency studies rarely operationalize a set point, perturbation, return trajectory, or stability parameter. Resistance to change is not evidence that a family “needs” a symptom, and recurrence alone does not establish a homeostatic mechanism. In the publication, *homeostasis* should remain historical or metaphorical; where data permit, use the narrower language of observed recurrence, dwell, return, or recovery after perturbation.

### 5.5 Testing the proposed rescue loop

The proposed sequence is: distress → monitoring → intervention or rescue → immediate relief → reduced exposure to consequences or autonomous coping → increased future reliance → caregiver vigilance or resentment → instability → renewed distress → renewed intervention. Existing dyadic evidence supports only portions of this sequence.

| Proposed link | Evidential status | What can be said |
|---|---|---|
| Distress or perturbation → temporally dependent partner response | **Reasonably supported** | Conflict, mutual-stress, support, and health-disagreement tasks produce temporally coupled affective, behavioral, or physiological change. Most studies do not specifically measure vigilance or monitoring, and temporal coupling alone does not establish reinforcement. |
| Intervention or support → immediate relief | **Reasonably supported for some support; indirect for rescue** | Subjective stress or physiology sometimes dampens during supportive interaction. Damping can also reflect time, habituation, task completion, or self-regulation. |
| Recipient relief → repetition of the helper's behavior | **Currently speculative** | The core studies do not show that recipient relief reinforces the same caregiver response in a later episode. This remains a testable negative-reinforcement hypothesis. |
| Rescue → reduced consequences or autonomous coping | **Not established in this stream** | Later unaided coping, self-efficacy, and exposure to consequences are generally unmeasured. Physiological calming is not evidence of diminished capacity. |
| Reduced coping → greater future reliance | **Currently unsupported** | No prospective episode-to-episode study here tests this mediation path. |
| Helper relief, role identity, or burden → vigilance or resentment | **Plausible but indirect** | Other literatures address these variables; coupling studies do not establish the proposed temporal order. |
| Instability → recurrence of the pattern | **Reasonably supported for recurrent interaction states; speculative for this exact loop** | State-space and recurrence methods demonstrate dwell and return, but they have not tracked repeated rescue episodes with the proposed mediators. |
| Mutual maintenance | **Reasonably supported as a possible architecture** | Bidirectional effects occur, but asymmetric and null effects are common. Mutual maintenance never entails equal responsibility. |
| Narrowing of behavioral repertoire | **Reasonably supported as an observational analogue** | Fewer states, concentrated occupancy, or fewer transitions can be measured. Subjective agency, role flexibility, and perceived alternatives require separate measures. |

No identified study measures, across repeated natural episodes, all of the necessary elements: precipitating distress; monitoring; the exact form of help, accommodation, or rescue; immediate relief in each person; change in the probability of repeating the response; autonomous coping or exposure to consequences; later reliance; and contraction of each person's repertoire. The complete loop is therefore a **Flow Hijacked mechanism map**, not a discovered law.

### 5.6 Scientific ruling for “Two Nervous Systems, One Attractor”

The strongest defensible synthesis is that some dyads repeatedly coordinate distress and response in ways that become predictable and behaviorally restrictive. Direct evidence establishes cross-partner temporal dependence and the measurable recurrence or concentration of joint behavior. It is a **bounded integrative inference across adjacent literatures**, not a result demonstrated within one study stream, that a pattern can stabilize a dyad in the short term while making alternative responses less practiced or less available within that interaction. Whether the same process reduces autonomous coping, freezes identities, or narrows perceived futures over months and years remains plausible but incompletely tested.

The title must therefore be explained, not literalized:

> “One attractor” is a metaphor for a recurrent interaction pattern, not a claim that two brains become one system. In research, an attractor-like pattern means that a dyad repeatedly returns to a limited set of observable states; whether that pattern helps or harms depends on its content, context, flexibility, power structure, and consequences.

This yields a narrower, testable definition of the systems claim: **co-constriction is a proposed process in which mutually contingent responses become concentrated into a limited interaction repertoire that supplies short-term predictability or relief and may, under some conditions, reduce opportunities for autonomous coping and flexible role change.** The evidence directly supports temporal coupling, some forms of temporally contingent responding, recurrence, and restricted observed repertoires. Functional reinforcement contingencies require additional evidence that consequences change later response probability. The literature only indirectly supports the progression from relief to reinforcement to reliance, and it does not yet demonstrate narrowing of subjective possibility. That evidential boundary should remain visible in the lecture rather than being repaired with neuroscientific language.

## 6. Addiction and the family/partner system

### 6.1 Replace blame with contingency analysis

Neither the codependency literature nor family-treatment research supports saying that relatives *cause* an addiction. Addiction has multiple determinants; a partner's response is neither necessary nor sufficient for the disorder. What relatives can sometimes alter is narrower and more observable: the immediate cost or reward attached to a particular act, access to money or substances, concealment from employers or clinicians, exposure to consequences, reinforcement for sober activity, or the probability of entering treatment.

This distinction matters because *enabling* has accumulated several incompatible meanings: any continued relationship with the person, material help, rescuing from a consequence, tolerance of substance use, denial, or an intrapsychic need to be needed. Rotunda and Doman's critical review found a thin and inconsistent empirical base beneath the label. The term can also hide context: paying rent, offering transport, or allowing someone home can preserve safety and treatment access in one situation while financing use or removing informative feedback in another. The scientific unit should therefore be a specified behavior, its timing, and its measurable contingency—not the moral identity “enabler.”

**Safe formulation:** a family response may maintain or interrupt a *particular interpersonal contingency*; this does not establish that the family caused the underlying disorder. **Evidential status:** structured behavioral treatments show that some social consequences and treatment-entry contingencies are modifiable, but the naturalistic causal effects of rescue, concealment, monitoring, or shielding are much less directly studied.

This distinction limits what can be inferred from intervention trials. CRAFT and significant-other-involved treatment show that deliberately changing relational responses can improve some outcomes. They do **not** demonstrate that every naturally occurring act popularly called “enabling” maintains substance use, or identify rescue as the mediator of addiction persistence. Naturalistic behaviors must be specified and studied rather than back-inferred from treatment efficacy.

### 6.2 Families are affected people, not merely treatment instruments

The Stress–Strain–Coping–Support model treats living with a relative's addiction as a chronic stressor rather than evidence of family pathology (Orford et al., 2010a). A synthesis of two decades of qualitative studies described uncertainty, conflict over money and possessions, threats to home life, worry, and often inadequate social support; women partners and mothers were overrepresented, but the corpus included multiple family roles and countries (Orford et al., 2010b). In 168 help-seeking family members—87.5% women—perceived problem severity, burden, tolerant-inactive coping, and hopelessness fit the proposed pathways, but the cross-sectional design could not establish temporal order (Horváth & Urbán, 2019).

Interventions directed to affected relatives can improve their wellbeing independently of whether the person using substances changes. Rushton et al.'s preregistered review included 19 studies; only ten contributed to meta-analyses, and small, methodologically weak studies limited certainty. Individually administered interventions reduced depression and distress in three RCTs, while group and pre–post evidence also suggested improvement (Rushton et al., 2023). A broader review across substance use, gambling, and gaming similarly found benefits for some affected-other, relationship, and treatment-entry outcomes, but emphasized few studies and methodological limitations (Merkouris et al., 2022). This evidence supports offering relatives help in their own right; it does not validate “codependency” as their diagnosis.

### 6.3 Structured significant-other involvement can improve treatment outcomes

The clearest contradiction to a blanket “family involvement sustains addiction” claim is the treatment literature. Ariss and Fairbairn pooled 16 randomized trials (2,115 participants; 77 effect sizes) comparing significant-other-involved treatment with active individual treatment. The pooled advantage for substance use and related problems was small but reliable (*d* = 0.242), persisted 12–18 months, and corresponded to roughly three fewer weeks of use per year. This is evidence that structured relational involvement can improve outcomes, not proof that every form of involvement helps.

CRAFT is particularly relevant because it replaces confrontation or detachment with skills for reinforcing non-use, allowing selected consequences, communicating, protecting the concerned other, and inviting treatment. In Miller, Meyers, and Tonigan's trial of 130 concerned significant others, CRAFT produced more treatment entry (64%) than Al-Anon facilitation (13%) or a Johnson confrontation preparation (30%); all three groups showed similar improvement in family-member functioning and relationship quality. Roozen et al.'s review of four high-quality RCTs (264 concerned others) estimated substantially greater treatment engagement for CRAFT than Al-Anon/Nar-Anon facilitation or Johnson intervention, while concerned others improved across conditions. Kirby et al. later found full CRAFT and a shorter treatment-entry component produced similar treatment entry (62–63%) and both exceeded Al-Anon/Nar-Anon facilitation (37%) among 115 concerned others; substance use and family-member functioning improved over time without between-condition differences.

The effect is not invariant. Archer et al.'s review of 14 studies (691 concerned others, predominantly women spouses or parents) found treatment-entry rates ranging from 12.5% to 86% and lower results in gambling adaptations; the comprehensive formats appeared strongest, but uncontrolled designs and delivery differences complicated interpretation. In a Swedish trial of 113 parents of treatment-refusing young adults, CRAFT did not outperform manualized counselling at 24 weeks (33% versus 31% treatment entry; OR 0.84, 95% CI 0.35–1.99); recruitment stopped after a post-launch stopping rule and the sample was mostly mothers of sons (Siljeholm et al., 2024). The responsible conclusion is that CRAFT is an evidence-based option with context-dependent effects, not a universally superior method.

### 6.4 Al-Anon and “tough love” require narrower claims

Al-Anon is culturally important and many participants report benefit, but its evidence base cannot carry causal claims about detachment, codependency, or treatment entry. In early CRAFT trials, Al-Anon facilitation conditions often improved concerned-other wellbeing while producing lower entry of the person with addiction. In a later multisite RCT of 279 concerned others, an intensive Al-Anon referral increased relationship resources but did not significantly increase Al-Anon attendance relative to usual care (28% versus 21%; Timko et al., 2022). Observational studies of attendance are vulnerable to self-selection and survivorship; they cannot identify which doctrine caused improvement.

“Natural consequences” is not a single intervention. It can mean stopping deception, declining to finance use, not completing another adult's obligations, or—much more dangerously—withdrawal of housing, contact, naloxone, transport, or emergency aid. In 22 qualitative interviews conducted in British Columbia's toxic-drug crisis, “tough love” was understood inconsistently and decisions were shaped by perceived harms, values, agency beliefs, and available resources; participants described uncertainty and regret, and found negotiated boundaries more usable than a binary of enabling versus tough love (Hawkins et al., 2026). This small, context-specific qualitative study does not decide effectiveness, but it exposes the ethical and ecological information erased by slogans.

No contingency principle warrants withholding immediate protection in overdose risk, violence, exploitation, severe incapacity, or danger to children or dependent adults. Safety actions and longer-term autonomy building operate on different timescales.

### 6.5 Mutual influence is real but not the proposed rescue loop

Longitudinal alcohol research supports reciprocal partner influence at a modest level. Muyingo et al.'s meta-analysis of 17 longitudinal studies (10,553 couples) estimated partner effects on later drinking in both directions, somewhat larger from women to men (β = .19) than men to women (β = .12), with effects moderated by design and relationship characteristics. These studies show social influence; they do not identify reassurance, shielding, or rescue as the mediator, and the samples were largely different-sex couples. They therefore support the *mutuality* premise of co-constriction more than its proposed sequence.

### 6.6 Flow Hijacked evidence-informed decision heuristic

The useful question is not “Am I enabling?” but the more specific sequence below. This is an evidence-informed Flow Hijacked heuristic, not a validated clinical decision rule or a substitute for safety assessment.

1. **What precise behavior am I responding to?**
2. **What immediate outcome does my response produce for each person?**
3. **Which consequence, learning opportunity, or source of safety does it change?**
4. **Is the response reinforcing treatment, non-use, honesty, and self-directed action—or making harmful behavior easier?**
5. **Is there acute risk that makes protection non-negotiable?**
6. **Can the same care be delivered in a way that returns authorship and action to the other person?**

The heuristic is compatible with behavioral family interventions and avoids diagnosing love, loyalty, or distress; it has not itself been validated as an assessment or treatment protocol.

## 7. Healthy interdependence

### 7.1 The alternative to rigid dependency is not detachment

Attachment and support research supplies a crucial correction: responsive care and autonomous action are compatible. Feeney's “dependency paradox” studies combined partner reports, observed couple behavior, an experimental assistance manipulation, and follow-up evidence; acceptance of legitimate dependency needs was associated with greater independent functioning rather than helplessness (Feeney, 2007). Her secure-base studies likewise link responsive partner support with exploration and goal striving (Feeney, 2004; Feeney & Thrush, 2010). These demographically bounded studies do not show that support always creates autonomy, but they provide direct counterevidence to the premise that healthy functioning requires minimizing dependence.

A preregistered multilevel meta-analysis of partner support and goal outcomes pooled 195 effects from 36 samples and 10,130 people in romantic relationships. The overall association was *r* = .25. Responsive support (*r* = .27) and practical support (*r* = .22) were positively associated with progress, commitment, or self-efficacy, whereas negative support—control, coercion, interference—was negatively associated (*r* = −.14). Responsive support related similarly to all goal outcomes; practical help related less strongly to self-efficacy than to progress or commitment (Vowels & Carnelley, 2022).

That meta-analysis is both the strongest positive evidence and an important brake on certainty. Only about one fifth of effects were longitudinal; cross-sectional effects were larger; few measures had prior validation; and heterogeneity was substantial. It shows that support usually accompanies better goal outcomes. It cannot specify the optimal dose, prove directionality, or tell when repeated practical help substitutes for competence.

Support can also succeed on one outcome and fail on another. In a 31-day diary study around the New York bar examination, received emotional support was associated, on average, with both greater partner closeness and greater negative mood; distressed days may of course have elicited more support (Gleason et al., 2008). In 67 cohabiting couples, the benefit of visible or invisible help depended less on its mere occurrence than on whether the recipient experienced it as understanding, validating, and caring (Maisel & Gable, 2009). These studies make an important distinction for this review: affiliative success (“I feel cared for”), proximal regulation (“I feel calmer”), performance (“the task was completed”), and capacity (“I can act more flexibly later”) are different outcomes and may diverge.

### 7.2 Studied support features and Flow Hijacked design principles

The most directly studied features across secure-base, goal-support, and self-determination work are:

- **Responsiveness:** accurately understanding the person's state and communicating care (Maisel & Gable, 2009; Vowels & Carnelley, 2022).
- **Autonomy support:** taking the person's perspective, acknowledging choice and ambivalence, and avoiding pressure as the primary instrument (Koestner et al., 2012).
- **Goal alignment:** supporting goals the recipient endorses, rather than recruiting the recipient into the supporter's preferred future (Koestner et al., 2012; Overall et al., 2010).
- **Action-facilitating or practical support:** supplying useful steps, information, labor, or resources without assuming that more aid is always better (Overall et al., 2010; Vowels & Carnelley, 2022).
- **Availability:** being reliably reachable in ways that can support exploration and goal pursuit (Feeney, 2004, 2007; Feeney & Thrush, 2010).
- **Low control or interference:** avoiding coercion, criticism, pressure, and obstruction as the principal means of influence (Vowels & Carnelley, 2022).

This review derives a second set of **Flow Hijacked design principles** from those findings. These are evidence-informed integrations, not independently validated support constructs:

- **Calibration:** match practical help to actual capacity, danger, skill, and request.
- **Competence return:** where possible, leave the decision, exposure, practice, or next action with the recipient.
- **Availability without takeover:** remain reachable while allowing tolerable uncertainty and effort.
- **Temporal flexibility:** permit intensive substitution during genuine incapacity and graded handback as capacity returns.
- **Reciprocity and repair:** allow roles to change, refusals to be negotiated, and connection to resume after a limit or mistake without fixing either person in one role.

In three prospective studies, autonomy support—operationalized as empathic perspective-taking—predicted three-month goal progress and was associated with relationship quality and wellbeing; directive support showed at most marginal links to progress and none to relationship quality or wellbeing (Koestner et al., 2012). This does not make advice harmful. It suggests that positive guidance without perspective-taking is a weaker relational resource than help that preserves ownership.

### 7.3 The clearest experimental contrast is substitution versus scaffolding

In a small laboratory experiment, 44 students solved arithmetic tasks with a close friend providing either autonomy support (calculation steps), overprotective support (solutions), or no support. During a subsequent solo stress task, the overprotection group showed greater heart rate, diastolic blood pressure, subjective stress, and lower perceived control; autonomy support did not produce those effects (Zniva et al., 2017). The sample and task are narrow, so the study cannot stand in for long relationships. It nevertheless isolates a bounded contrast central to the lecture: receiving solutions rather than procedural guidance was followed by greater stress and lower perceived control when the recipient next worked alone. The study did not establish poorer task performance or progressive dependence.

The distinction also appears in clinical accommodation. The goal is not cold refusal. In a preliminary experiment with 36 people with OCD, imagining emotional support rather than reassurance reduced the anticipated urge to seek reassurance and was rated as more acceptable, calming, connecting, and consistent with “fighting OCD together” (Causier & Salkovskis, 2025). Because the scenarios were imagined and outcomes anticipated, clinical efficacy remains unproven. What the study contributes is a humane alternative: validate distress and connection without certifying the feared proposition or completing the compulsion.

### 7.4 A functional definition of expanding connection

For this publication, **connection that expands behavioral options** is an interpretive shorthand, not a validated measure: after support, does the person have more usable options, more confidence or skill, better access to values and goals, and a greater ability to act without that exact intervention next time? This is not a moment-by-moment requirement. Care may temporarily substitute for capacity during illness, crisis, disability, grief, or overload. The relevant pattern is whether the relationship can shift modes as risk and capacity change.

By contrast, agency-replacing regulation is suggested when the same interpersonal move becomes compulsory across contexts; the receiver has fewer opportunities for mastery; the caregiver cannot tolerate nonintervention; help is increasingly organized around preventing either person's distress; and attempts to change the pattern produce rapid reversion, escalation, or identity threat. Several of these indicators are plausible integrations rather than validated diagnostic criteria.

### 7.5 Flow Hijacked reflection heuristic for the healthy countermodel

The following questions translate the evidence into a Flow Hijacked reflection heuristic. They do not operationalize a validated construct and should not be used as diagnostic criteria or as a substitute for individualized safety assessment.

- Does care respond to the person's expressed goal, current capacity, and actual danger?
- Does it reduce shame while preserving truthful feedback?
- Does it offer a step, tool, presence, or choice rather than automatically supplying the whole solution?
- Can the receiver decline, revise, or reciprocate the support?
- Can the supporter tolerate the receiver's ordinary distress and imperfect learning?
- After acute need passes, is responsibility deliberately handed back?
- Can both people occupy more than one role—strong, dependent, uncertain, caring, cared for?
- Does the relationship remain safe when a boundary is set?

The last question is non-negotiable: “flexibility” is not a reason to remain exposed to abuse or coercion. **Boundary** and **repair** are used here descriptively, not as mechanisms independently validated by this bounded review. A boundary means an explicit limit on one's own participation rather than emotional withdrawal; repair means restoring truthful coordination after a rupture or limit. The broad literatures on boundaries and conflict repair were outside scope, so these terms should not carry causal claims in the publication.

## 8. Evidence audit of the Co-Constriction hypothesis

### 8.1 Proposed construct

> **Co-constriction (Flow Hijacked working hypothesis):** a relational process in which two people repeatedly regulate one another in ways that reduce immediate distress or uncertainty but, over time, narrow behavioral flexibility, agency, role flexibility, and the range of psychologically accessible responses available to both people.

The definition has four virtues: it is process-based rather than person-blaming; it makes mutuality possible without requiring equal power or equal contribution; it focuses on temporal trade-offs; and it specifies outcomes that could in principle be measured. Its chief weakness is that it currently joins findings from different populations and levels of analysis. No included study tested the entire sequence, and *psychologically accessible futures* has no agreed operational measure in this literature.

### 8.2 Component-by-component ruling

| Proposed component | Grade | What the evidence permits | What it does not permit |
|---|---|---|---|
| 1. Short-term reduction of distress or uncertainty | **Strongly supported in bounded contexts** | Reassurance commonly produces immediate relief in OCD/anxiety; accommodation is explicitly undertaken to reduce distress; responsive support can calm and aid coping (Salkovskis & Kobori, 2015; Causier & Salkovskis, 2025). | Generalizing the same time course to all partners, all “codependent” behavior, or addiction without measuring it. |
| 2a. Recipient-side repetition after relief | **Reasonably supported in bounded reassurance/avoidance paradigms** | Immediate relief and renewed urges fit negative reinforcement; treatment models also target repeated avoidance and accommodation (Salkovskis & Kobori, 2015). | Claiming that naturalistic relief was shown prospectively to cause the next reassurance or help-seeking episode across diagnoses. |
| 2b. Provider-side repetition after recipient or provider relief | **Plausible but directly undermeasured** | Qualitative caregiver accounts describe relief, conflict avoidance, and lack of alternatives; behavioral family treatments show that some social contingencies are modifiable (Halldorsson et al., 2016; Miller et al., 1999). | Claiming that recipient relief or reduced provider guilt/anxiety has been shown to reinforce the same provider response across repeated natural episodes. |
| 3. Increased recurrence of a relational pattern | **Plausible to reasonably supported; causal link uneven** | Reassurance urges return in anxious samples; accommodation and symptoms show temporal persistence; observed dyads occupy recurrent states (Salkovskis & Kobori, 2015; Hollenstein, 2007; Brinberg et al., 2025). | Inferring that immediate relief caused a stable relationship-wide loop from persistence, a cross-sectional correlation, or retrospective report. |
| 4. Decreasing behavioral flexibility | **Plausible but mostly indirect** | Accommodation can reduce exposure and mastery opportunities; state-space work can quantify restricted repertoires; overprotective support increased stress and reduced perceived control during a later solo task in one experiment (Hollenstein, 2007; Zniva et al., 2017). | Saying codependent relationships have been shown longitudinally to lose “degrees of freedom.” That trajectory has not been tested as such. |
| 5. Greater role rigidity | **Plausible but weakly demonstrated** | Qualitative and family-systems accounts describe fixed rescuer, responsible, ill, or dependent positions; contingent self-worth can make role change threatening (Knee et al., 2008; Bacon et al., 2020). | Treating “overfunctioner/underfunctioner” as a validated dyadic phenotype or assuming roles are symmetric. |
| 6. Reduced autonomous coping | **Plausible with narrow direct support; indirect in most accommodation research** | One small experiment found greater later stress and lower perceived control after solutions rather than procedural guidance; reduced exposure and mastery are credible but usually inferred routes (Zniva et al., 2017). Secure-base and autonomy-support studies show the opposite is possible (Feeney, 2004). | Assuming any relief or practical help reduces competence; treating physiological stress as impaired performance; or extrapolating child-anxiety findings directly to adult addiction relationships. |
| 7. Narrowing of perceived alternatives | **Currently speculative as a distinct outcome** | Contingent self-worth, attachment hyperactivation, hopelessness, and rigid state repertoires are compatible with a narrower field of action (Knee et al., 2008; Horváth & Urbán, 2019). | Claiming the range of imagined futures has been measured or shown to contract because of interpersonal regulation. |
| 8. Mutual rather than purely individual maintenance | **Reasonably supported for reciprocal influence; weak for the exact loop** | Cross-lagged accommodation/anxiety findings, longitudinal partner alcohol effects, and dyadic regulation research show bidirectional influence (Bertelsen et al., 2023; Muyingo et al., 2020; Butler & Randall, 2013). | Assuming equality of responsibility, confusing reciprocal influence with consent, or claiming the proposed distress→rescue→dependence sequence has been validated intact. |

### 8.3 Overall classification

- **Strongly supported:** immediate regulation can reduce distress; close others affect behavior and outcomes; responsive, autonomy-supportive care can expand capacity.
- **Reasonably supported:** recipient-side reassurance or avoidance can become self-repeating in bounded clinical contexts; some processes are bidirectional; practical substitution can worsen later perceived control and stress under narrow experimental conditions.
- **Plausible but indirect:** provider-side reinforcement by recipient or provider relief; reduced autonomous coping outside the narrow experiment; and the conjunction of caregiver vigilance, recipient reliance, role lock-in, and reduced interactional flexibility.
- **Currently speculative:** progressive narrowing of each person's perceived future; a single transdiagnostic co-constriction trajectory; a characteristic attractor shared by “codependent” relationships.
- **Contradicted or requiring reformulation:** care, dependence, co-regulation, accommodation, or synchrony are not inherently harmful; family involvement does not generally worsen addiction; the distressed person is not simply made dependent by the caregiver; caregiver behavior is often an effect as well as a cause of distress; practical support often improves goal progress.

### 8.4 Recommended revision

Use co-constriction in the publication only with the following qualification:

> **Co-constriction is a proposed relational process—not a diagnosis—in which a narrow set of reciprocal responses becomes increasingly likely because it solves an immediate problem for one or both people, while repeatedly displacing opportunities for autonomous coping, honest feedback, role variation, or alternative action. Whether constriction occurs depends on risk, capacity, consent, timing, culture, and whether support is later handed back.**

This version removes the unmeasured claim that futures necessarily narrow, makes contingency and recurrence explicit, and preserves the stronger idea—that recurrent relief can trade short-term stability for long-term flexibility—as a testable hypothesis.

## 9. Evidence-to-lecture map

| Movement | Scientific payload | Defensible dramatic turn | Guardrail | Reopening move |
|---|---|---|---|---|
| **When Care Becomes Control** | Reassurance and accommodation can relieve immediate distress; repeated substitution can reduce exposure, feedback, mastery, or self-efficacy; responsive/practical support often helps. | “The same act can be care at one timescale and constriction at another.” | Do not infer harm from intensity, sacrifice, or relief alone. Ask what capacity, danger, consent, and learning opportunity were present. | Change the function of help: comfort without false certainty, scaffolding rather than supplying, safety without concealment. |
| **The Borrowed Self** | Relationship-contingent self-esteem, unmitigated communion, attachment hyperactivation, low differentiation, and caregiver identity describe distinct routes by which another's state can organize one's own regulation. | “I no longer only respond to your distress; my worth and stability begin to require a role in it.” | This is an integrative interpretation, not a unitary syndrome. Avoid gendered caricatures and mind-reading about motive. | Recover multiple sources of worth and role; name one's position; make care chosen, bounded, and revisable. |
| **Two Nervous Systems, One Attractor** | Dyadic influence, temporal coupling, repeated interaction states, and bidirectional symptom–accommodation paths are demonstrable; restricted repertoires can be modeled. | “A response can become likely not because either person chose the whole pattern, but because each step makes the next step easier.” | “Nervous systems” is embodied shorthand, not literal fusion. “Attractor” is technical only when repeated states and return tendencies are modeled. | Interrupt one transition; lengthen the pause; add a third response; measure recovery and role switching rather than pursuing perfect calm. |
| **Reopening the Relationship** | Secure-base responsiveness, autonomy support, CRAFT, graded reduction of accommodation, and competence-preserving practical help show that connection and agency can rise together. | “The exit is not less love; it is care that returns authorship.” | Do not prescribe exposure to abuse or danger. Capacity-building can include intensive care during incapacity and deliberate handback later. | Negotiate goals; separate support from certainty; preserve truthful consequences; build tolerated distress, competence, reciprocity, and repair. |

### Recommended narrative sequence

1. Open with a recognizable act of loving relief, not a pathological person.
2. Show two clocks: **what works in the next five minutes** and **what becomes more likely next month**.
3. Reveal that the caregiver is also being regulated—relief, certainty, worth, and role—not merely controlling the other.
4. Zoom out from two motives to one recurring transition structure.
5. Stop before the metaphor hardens: state exactly which links are evidenced and which are Flow Hijacked's synthesis.
6. Reopen possibilities through one altered response at a time, preserving attachment and safety.

### A scientifically faithful core sentence for the lecture

> Love narrows the future not when people need one another, but when a small set of responses becomes compulsory—because it relieves something now—while repeatedly displacing truthful feedback, role flexibility, autonomous practice, and other ways of staying connected.

## 10. Claims Flow Hijacked can safely make

The following are publication-ready if their qualifiers remain intact.

| # | Safe claim | Strongest support |
|---:|---|---|
| 1 | **“Codependency” is a historically influential but scientifically fragmented umbrella, not a validated unitary disorder.** | Hands & Dear (1994); Molina et al. (2026) |
| 2 | **The term moved from describing relatives of people with alcohol problems to a much broader account of self, relationship, and control—faster than its evidence base developed.** | Hands & Dear (1994); Dear et al. (2004); Molina et al. (2026) |
| 3 | **Codependency scales measure partly different mixtures of self-sacrifice, interpersonal control, emotional suppression, external focus, and distress.** | Lindley et al. (1999); Marks et al. (2012); Vederhus et al. (2019) |
| 4 | **It is usually more informative to name the behavior and mechanism—such as accommodation or reassurance—than to label a person codependent.** | Hands & Dear (1994); Marks et al. (2012); Molina et al. (2026) |
| 5 | **Reassurance can reduce anxiety quickly while the urge for further reassurance later returns, particularly in OCD and panic samples.** | Salkovskis & Kobori (2015) |
| 6 | **Family accommodation is moderately associated with OCD severity, but association alone does not show that accommodation caused the symptoms.** | Wu et al. (2016); Hermida-Barros et al. (2024) |
| 7 | **In the largest recent OCD synthesis, baseline accommodation did not predict symptom improvement, even though accommodation and severity were correlated and both often fell during treatment.** | Hermida-Barros et al. (2024) |
| 8 | **Distress can evoke accommodation as well as accommodation preceding distress; at least one repeated-measures treatment study found both directions, with the symptom-to-accommodation path larger.** | Bertelsen et al. (2023) |
| 9 | **Support should not be equated with reassurance: in a preliminary imagined-scenario OCD study, emotional support was rated as more connecting and calming and produced a lower anticipated urge for reassurance than certainty-giving.** | Causier & Salkovskis (2025) |
| 10 | **Supplying solutions can differ from scaffolding competence; in a small experiment, receiving solutions was followed by greater stress and lower perceived control during a later solo task.** | Zniva et al. (2017) |
| 11 | **Partner support is, on average, associated with better goal progress, commitment, and self-efficacy; control, coercion, and interference are associated with worse outcomes.** | Vowels & Carnelley (2022) |
| 12 | **Practical help often improves progress, yet it appears less consistently related to self-efficacy than responsive support.** | Vowels & Carnelley (2022) |
| 13 | **Secure dependence and autonomy are compatible: responsive acceptance of legitimate need can support later independent action.** | Feeney (2004, 2007); Feeney & Thrush (2010) |
| 14 | **In prospective goal studies, autonomy support defined through empathic perspective-taking predicted goal progress and relational wellbeing more consistently than directive positive guidance.** | Koestner et al. (2012) |
| 15 | **Self-worth can become contingent on relationship events, producing stronger fluctuations and preoccupation without being reducible to relationship satisfaction or commitment.** | Knee et al. (2008) |
| 16 | **Self-neglecting overinvolvement is not the same as warmth or communion; “unmitigated communion” was developed precisely to separate them.** | Aubé (2008); Helgeson et al. (2015) |
| 17 | **Differentiation concerns maintaining a coherent position while emotionally connected; it should not be translated as detachment or radical self-sufficiency.** | Skowron & Friedlander (1998); Skowron & Schmitt (2003); Calatrava et al. (2022) |
| 18 | **Higher differentiation is associated with better relationship functioning and lower attachment anxiety, but much of the literature remains self-report and non-causal.** | Calatrava et al. (2022); Rodríguez-González et al. (2023) |
| 19 | **Attachment anxiety helps explain hypervigilance to rejection and excessive reassurance seeking, but reassurance seeking does not uniformly damage every relationship.** | Shaver et al. (2005); Starr & Davila (2008) |
| 20 | **A caregiver's intervention may regulate the caregiver's uncertainty or exposure to conflict as well as the recipient's distress; small qualitative OCD evidence makes this plausible, but provider-side reinforcement should be measured rather than presumed.** | Halldorsson et al. (2016) |
| 21 | **Family responses can alter local consequences and treatment-engagement contingencies around substance use; these findings do not establish that relatives caused the underlying addiction.** | Rotunda & Doman (2001); Miller et al. (1999); Ariss & Fairbairn (2020) |
| 22 | **Involving a significant other in substance-use treatment produces a small average advantage over active individual treatment.** | Ariss & Fairbairn (2020) |
| 23 | **CRAFT can increase treatment entry, but effects vary by delivery, population, comparator, and study; it is not universally superior.** | Roozen et al. (2010); Archer et al. (2020); Siljeholm et al. (2024) |
| 24 | **Interventions directed to affected family members can improve their wellbeing independently of whether the person with the addictive behavior enters treatment; this supports treating relatives as people with needs of their own, not only as levers for someone else's change.** | Orford et al. (2010a, 2010b); Merkouris et al. (2022); Rushton et al. (2023) |
| 25 | **Partners influence one another over time, but reciprocal influence is not the same as equal responsibility.** | Muyingo et al. (2020); Butler (2011); Butler & Randall (2013) |
| 26 | **Repeated-observation studies can quantify concentrated repertoires and recurrent joint states; calling such a pattern an “attractor” requires an explicit operational definition and dense temporal data, and adult evidence specific to codependent-like couples is lacking.** | Hollenstein (2007); Granic & Lamey (2002); Butler (2011); Brinberg et al. (2025) |
| 27 | **Synchrony is not intrinsically healthy: its meaning depends on what is coupled, in which direction, at what lag, and in what context.** | Butler (2011); Palumbo et al. (2017); Timmons et al. (2015) |
| 28 | **Co-constriction is best presented as a new process hypothesis whose early links have support, while progressive role and future narrowing remain to be tested directly.** | This synthesis; see Section 8 |

## 11. Claims Flow Hijacked should not make

Do not publish any of the following without replacing it with the safer formulation indicated.

| Tempting formulation | Why it fails | Safer replacement |
|---|---|---|
| “Codependency is a mental disorder.” | It is not an accepted DSM-5-TR or ICD-11 diagnosis and lacks a stable operational core. | “Codependency is a contested umbrella used in clinical and self-help traditions.” |
| “Codependents have a recognizable personality type.” | Scale content overlaps broad distress, dependency, self-sacrifice, attachment, and control; specificity is poor. | Describe the measured trait or behavior. |
| “Family members enable or cause addiction.” | Conflates etiology with a local contingency and assigns blame unsupported by evidence. | “Some responses can alter the consequences or reinforcement surrounding a behavior.” |
| “If you rescue someone, they will become dependent.” | Outcome depends on danger, capacity, timing, task, consent, and handback; support often helps. | “Repeated substitution can sometimes reduce opportunities for mastery or feedback.” |
| “Accommodation maintains anxiety.” | Often plausible, but much evidence is correlational and direction is bidirectional. | “Accommodation is associated with severity and may maintain avoidance in some contexts.” |
| “Never reassure an anxious person.” | Ordinary reassurance can be adaptive; abrupt refusal can be distressing; support may replace certainty-giving. | Distinguish emotional support from repeated certainty rituals. |
| “Natural consequences are always necessary for recovery.” | “Natural” is undefined; consequences can include preventable death, homelessness, or harm to dependants. | Evaluate each action by safety, contingency, proportionality, and alternatives. |
| “Boundaries mean withdrawing love or contact.” | Boundaries can specify what one will do while preserving care and connection. | “A boundary protects truthful participation and role clarity.” |
| “Differentiation means needing no one.” | Reverses the construct; differentiation is self-position *within connection*. | “Differentiation permits closeness without losing authorship.” |
| “Strong attachment is enmeshment.” | Attachment security, intimacy, and cohesion are not diffuse boundaries or fusion. | Identify the specific autonomy or boundary impairment. |
| “Anxious attachment is a fixed identity.” | Attachment dimensions are probabilistic, contextual, and changeable. | “Attachment-related anxiety can heighten specific responses under threat.” |
| “The caregiver secretly needs the other person to stay ill.” | Motive cannot be inferred from behavior; the statement is stigmatizing and often unfalsifiable. | “The caregiving role may also regulate the caregiver's uncertainty or self-worth; assess rather than assume.” |
| “Two nervous systems become one.” | Biologically false and suggestive of literal fusion/control. | “Two embodied people can influence one another's affect, physiology, and behavior over time.” |
| “Synchrony shows healthy co-regulation.” | Synchrony may index shared threat, escalation, task demands, or adaptation. | Interpret coupling by context, lag, direction, and outcome. |
| “The couple is trapped in an attractor.” | Unless recurrent states and return dynamics were measured, this is metaphor, not result. | “The interaction repeatedly returns to a narrow pattern”; label attractor language as analogy unless modeled. |
| “Co-constriction reduces neural degrees of freedom.” | No neural evidence tests the construct. | Refer to observable behavioral repertoire or perceived options. |
| “Short-term relief causes long-term dysfunction.” | Temporal trade-off is mechanism- and context-specific; causality is rarely demonstrated across the whole chain. | “Short-term relief can reinforce a response that, when repeated, may impede learning or flexibility.” |
| “Both partners are equally responsible for the loop.” | Reciprocal influence does not erase power, coercion, disability, danger, or unequal choice. | “Both behavior streams may matter without being equally voluntary, powerful, or blameworthy.” |
| “Healthy love tolerates any distress.” | Can rationalize neglect, abuse, or preventable danger. | “Healthy support tolerates *ordinary and safe* distress in the service of capacity; it acts on danger.” |
| “Co-constriction is the mechanism behind codependency.” | The unified sequence has not been tested. | “Co-constriction is Flow Hijacked's integrative hypothesis for a subset of codependent-like patterns.” |

## 12. Terminology audit

| Term | Origin and best available definition | Empirical status and overlap | Recommendation for Flow Hijacked |
|---|---|---|---|
| **Codependency** | Addiction-treatment, Al-Anon, and self-help traditions; definitions variously emphasize external focus, self-sacrifice, control, emotional suppression, and identity through another. | No stable unitary definition; multiple scales with changing factor structures; overlaps attachment anxiety, dependency, self-sacrifice, interpersonal control, low self-esteem, and distress; not a DSM-5-TR or ICD-11 diagnosis. | **Use as a historical/popular doorway only.** State that the review decomposes it into mechanisms; never diagnose a person or relationship with it. |
| **Dependency** | Broad clinical, personality, attachment, and social-psychological traditions; reliance on another for emotional, instrumental, or decisional needs. | Not inherently pathological; measures often conflate need, helplessness, reassurance, and attachment insecurity. | **Use only with a qualifier** (instrumental dependency, interpersonal dependency, dependency need). Avoid as the opposite of health. |
| **Interdependence** | Social psychology/interdependence theory; outcomes and behavior are mutually affected across repeated interaction. | Strong theoretical and empirical tradition; mutual influence says nothing by itself about equality, health, or autonomy. | **Use technically.** Prefer *flexible interdependence* for the healthy countermodel and define what flexibility means. |
| **Interdependency** | Common-language variant often used as a synonym for interdependence; it does not name a distinct, agreed scientific construct in the reviewed relationship literature. | Conceptually overlaps almost completely with *interdependence* and can be confused with pathological dependency. | **Avoid as the technical term.** Use *interdependence* for the established tradition; retain *interdependency* only in quotations or explicitly nontechnical prose. |
| **Enmeshment** | Structural/family-systems tradition; diffuse boundaries and overinvolvement that constrain individuation or subsystem functioning. | Clinically influential but inconsistently operationalized; overlaps fusion, psychological control, cohesion, and cultural norms of family closeness. | **Use descriptively and sparingly.** Never equate closeness or collectivist family practice with pathology. Specify the boundary or autonomy problem. |
| **Differentiation of self** | Bowen family-systems theory; capacity to maintain a coherent position and regulate emotional reactivity while staying emotionally connected. | Commonly measured by DSI/DSI-R domains such as emotional reactivity, I-position, fusion, and cutoff; associations with attachment and relationship quality are substantial, but causal/longitudinal evidence is sparse and cultural assumptions remain. | **Use technically with definition and caveat.** Explicitly say differentiation is not detachment or individualism. |
| **Accommodation** | OCD/anxiety clinical research; changes in another person's behavior or routines intended to prevent or reduce symptoms, fear, or distress. | Well operationalized by disorder-specific measures; associated with severity and responsive to treatment, but much evidence is cross-sectional and baseline accommodation does not consistently predict symptom change. | **Use technically in the populations studied.** Elsewhere call the analogy “accommodation-like” and specify the action. |
| **Enabling** | Addiction clinical and recovery traditions; usually means behavior that makes continued harmful use easier or shields consequences. | Definitions and measures are inconsistent; easily becomes moral or tautological (“it enabled because use continued”); overlaps accommodation, tolerance, protection, and ordinary caregiving. | **Avoid as an analytic term.** Replace with the exact behavior and contingency. It may appear in quotations or historical discussion. |
| **Overfunctioning** | Family-systems and clinical vernacular; taking on responsibilities, decisions, or regulation beyond one's role or another's current need. | Face-valid but weakly standardized; often paired rhetorically with “underfunctioning”; overlaps overprotection, compulsive caregiving, and role overload. | **Descriptive only.** Name what is being taken over, for whom, and with what effect. |
| **Compulsive caregiving** | Attachment and clinical theory; persistent care provision organized partly around the caregiver's anxiety, inhibited care-seeking, or need to preserve proximity/control. | Conceptually important but far less securely measured than adult attachment dimensions; early evidence often clinical, retrospective, or gendered. | **Use cautiously and only when compulsion is evidenced.** Do not infer motive from the amount of care. |
| **Self-sacrifice** | Personality/schema and social-personality research; prioritizing another's needs at cost to one's own. A related, better-specified construct is *unmitigated communion*: concern for others that excludes adequate concern for self, distinct from healthy communion. | Measurable and associated with distress in several populations; motives, voluntariness, culture, burden, and reciprocity matter. | **Use technically when a cited measure is involved; otherwise descriptive.** Separate chosen sacrifice from identity-bound or coerced sacrifice. |
| **Interpersonal emotion regulation** | Emotion-regulation science; attempts to influence one's own or another person's emotional state through social interaction. | Active field with heterogeneous taxonomies and methods; can be adaptive or maladaptive, deliberate or automatic. | **Use technically as the broad process.** Always specify target, strategy, timescale, and outcome. |
| **Co-regulation** | Developmental, relationship, and psychophysiological research; coordinated, often bidirectional adjustment across time that contributes to regulation. | Definitions vary from temporal covariation to damping/return toward equilibrium; synchrony alone does not demonstrate regulation or benefit. | **Use only with an operational definition.** Never use as a synonym for harmony, health, or nervous-system fusion. |
| **Co-dysregulation** | Emerging clinical/developmental shorthand for reciprocal escalation or failure to recover. | No single accepted definition; often inferred from escalation or maladaptive synchrony. | **Descriptive, not diagnostic.** Prefer the observed pattern: escalation, mutual amplification, failure to recover. |
| **Relational rigidity** | Dynamic-systems and clinical process traditions; a restricted interaction repertoire, low context-sensitive switching, or persistent return to a small set of states. | Can be quantified using repeated observations, entropy, transition probabilities, or state-space grids; its meaning depends on context and timescale. | **Use technically when measured; descriptively otherwise.** Flexibility, not maximal variability, is the healthier comparator. |
| **Dyadic regulation** | Relationship science; regulation in which each person's behavior, affect, or physiology is both input to and outcome of the other's behavior over time. | Supported as reciprocal influence in many designs, but actor–partner association is not automatically a regulatory mechanism. | **Use technically with direction and lag specified.** Avoid implying equal power or responsibility. |
| **Co-constriction** | Flow Hijacked synthesis proposed in this review. | Not an established construct or measure. Components receive unequal support; the unified longitudinal pathway is untested. | **Use as an explicitly new, falsifiable process hypothesis or organizing metaphor.** Do not use as a diagnosis, trait, or settled mechanism. |

### Terminological hierarchy recommended for the publication

1. Begin with **codependency** as the familiar but contested cultural term.
2. Move quickly to the measurable **behavior**: reassurance, accommodation, monitoring, consequence shielding, practical substitution, controlling support.
3. Identify the proposed **learning or regulatory process**: negative reinforcement, reduced exposure/mastery, contingent self-worth, attachment hyperactivation, reciprocal influence.
4. Describe the **dyadic property** only if supported: recurrence, restricted repertoire, escalation, low context sensitivity.
5. Introduce **co-constriction** as Flow Hijacked's testable synthesis, with its uncertainty visible.

## 13. Research gaps

Only gaps that bear directly on the proposed model are included.

1. **The complete temporal chain has not been tested.** Intensive dyadic studies should measure distress, monitoring, intervention, relief for both parties, exposure/consequences, autonomous action, and later recurrence within the same design.
2. **Provider reinforcement is undermeasured.** Accommodation research usually measures the recipient's symptoms and the provider's behavior, not whether intervening reduces the provider's anxiety, guilt, uncertainty, or identity threat.
3. **Agency needs behavioral indicators.** Most work relies on self-report. Studies need choice generation, independent task initiation, persistence, help-seeking flexibility, competence transfer, and context-sensitive role switching.
4. **“Narrowing futures” needs an operational definition.** Candidate outcomes include fewer generated options, lower perceived controllability, reduced possible-self diversity, constricted goal portfolios, or a smaller observed response repertoire; none yet defines co-constriction.
5. **Role rigidity lacks validated dyadic measurement.** Overfunctioner/underfunctioner language needs repeated observation, both partners' reports, and tests of whether roles fail to switch when context changes.
6. **Adult partner accommodation is understudied outside OCD.** Child-anxiety findings should not be assumed to generalize to adult addiction, depression, chronic illness, disability, or everyday dependence.
7. **Support versus substitution needs dose and timing research.** The same practical act may scaffold a novice, protect during incapacity, or displace competence after capacity returns. Studies should manipulate timing, visibility, request, and handback.
8. **Dyadic models need asymmetry.** Reciprocal influence models should incorporate differences in power, risk, resources, impairment, coercion, and responsibility instead of treating bidirectionality as equality.
9. **Culture is not a nuisance moderator.** Differentiation, boundaries, obligation, and self-sacrifice carry different meanings across kinship systems and material conditions. Cross-cultural measurement invariance and indigenous/local constructs are needed.
10. **Gender sampling remains skewed.** Codependency and affected-family studies have often recruited women partners or mothers, risking conversion of gendered care burdens into a female pathology. Studies need men, queer and gender-diverse participants, and varied relationship structures.
11. **Addiction research needs behavior-level contingency mapping.** “Enabling” should be replaced with preregistered definitions of specific responses and their near- and long-term outcomes, including harm-reduction and overdose contexts.
12. **CRAFT mechanisms remain uncertain.** Component studies suggest treatment-entry training may be sufficient for the best-established outcome, while a recent active-comparator trial was null. Mediation, fidelity, context, and who benefits require better trials.
13. **Co-regulation outcomes need multiple timescales.** Immediate calming, end-of-conversation recovery, next-day functioning, and month-scale autonomy may point in different directions; a single synchrony coefficient cannot adjudicate them.
14. **Dynamic-systems claims need direct tests.** Dense time series, perturbations, state-space or recurrence analyses, and out-of-sample prediction are required before calling a codependent-like pattern an attractor or a phase transition.
15. **Healthy “handback” has barely been studied.** Interventions should test how support can be tapered or roles renegotiated without abandonment signals, abrupt withdrawal, or loss of necessary care.

## 14. Final curated bibliography and core-source evidence register

The final deduplicated core set contains exactly **110 sources**. Historical and diagnostic-classification authorities are not counted as core references and are listed separately below. Detailed design, population, sample-size, finding, limitation, and argument-relevance records for the core set are provided in the companion `Core_Source_Evidence_Register.md`.

### Core references (110)

1. Ançel, G., & Kabakçi, E. (2009). Psychometric properties of the Turkish form of Codependency Assessment Tool. *Archives of Psychiatric Nursing, 23*(6), 441–453. [https://doi.org/10.1016/j.apnu.2008.10.004](https://doi.org/10.1016/j.apnu.2008.10.004). PMID 19926026. [Abstract only]

2. Anderson, S. C. (1994). A critical analysis of the concept of codependency. *Social Work, 39*(6), 677–685. [https://doi.org/10.1093/sw/39.6.677](https://doi.org/10.1093/sw/39.6.677). PMID 7992137. [Abstract only]

3. Archer, M., Harwood, H., Stevelink, S., Rafferty, L., & Greenberg, N. (2020). Community reinforcement and family training and rates of treatment entry: A systematic review. *Addiction, 115*(6), 1024–1037. https://doi.org/10.1111/add.14901

4. Ariss, T., & Fairbairn, C. E. (2020). The effect of significant-other involvement in treatment for substance use disorders: A meta-analysis. *Journal of Consulting and Clinical Psychology, 88*(6), 526–540. https://doi.org/10.1037/ccp0000495

5. Aubé, J. (2008). Balancing concern for other with concern for self: Links between unmitigated communion, communion, and psychological well-being. *Journal of Personality, 76*(1), 101–134. [doi:10.1111/j.1467-6494.2007.00481.x](https://doi.org/10.1111/j.1467-6494.2007.00481.x); PMID 18186712.

6. Bacon, I., McKay, E., Reynolds, F., & McIntyre, A. (2020). The lived experience of codependency: An interpretative phenomenological analysis. *International Journal of Mental Health and Addiction, 18*(3), 754–771. [https://doi.org/10.1007/s11469-018-9983-8](https://doi.org/10.1007/s11469-018-9983-8). [Full text]

7. Bertelsen, T. B., Himle, J. A., & Håland, Å. T. (2023). Bidirectional relationship between family accommodation and youth anxiety during cognitive-behavioral treatment. *Child Psychiatry & Human Development, 54*(3), 905–912. https://doi.org/10.1007/s10578-021-01304-5 (PMID: 34997892; PMCID: PMC10140011).

8. Boeding, S. E., Paprocki, C. M., Baucom, D. H., Abramowitz, J. S., Wheaton, M. G., Fabricant, L. E., & Fischer, M. S. (2013). Let me check that for you: Symptom accommodation in romantic partners of adults with obsessive-compulsive disorder. *Behaviour Research and Therapy, 51*(6), 316–322. https://doi.org/10.1016/j.brat.2013.03.002 (PMID: 23567474).

9. Bolger, N., & Amarel, D. (2007). Effects of social support visibility on adjustment to stress: Experimental evidence. *Journal of Personality and Social Psychology, 92*(3), 458–475. https://doi.org/10.1037/0022-3514.92.3.458 (PMID: 17352603).

10. Bolger, N., Zuckerman, A., & Kessler, R. C. (2000). Invisible support and adjustment to stress. *Journal of Personality and Social Psychology, 79*(6), 953–961. https://doi.org/10.1037/0022-3514.79.6.953 (PMID: 11138764).

11. Bornstein, R. F., Geiselman, K. J., Eisenhart, E. A., & Languirand, M. A. (2002). Construct validity of the Relationship Profile Test: Links with attachment, identity, relatedness, and affect. *Assessment, 9*(4), 373–381. [doi:10.1177/1073191102238195](https://doi.org/10.1177/1073191102238195); PMID 12462757.

12. Brinberg, M., Solomon, D. H., Bodie, G. D., Jones, S. M., & Ram, N. (2025). Using state space grids to quantify and examine dynamics of dyadic conversation. *Communication Methods and Measures, 19*(1), 1–23. https://doi.org/10.1080/19312458.2024.2413973

13. Bringmann, L. F., Ferrer, E., Hamaker, E. L., Borsboom, D., & Tuerlinckx, F. (2018). Modeling nonstationary emotion dynamics in dyads using a time-varying vector-autoregressive model. *Multivariate Behavioral Research, 53*(3), 293–314. https://doi.org/10.1080/00273171.2018.1439722

14. Butler, E. A. (2011). Temporal interpersonal emotion systems: The “TIES” that form relationships. *Personality and Social Psychology Review, 15*(4), 367–393. https://doi.org/10.1177/1088868311411164

15. Butler, E. A., & Randall, A. K. (2013). Emotional coregulation in close relationships. *Emotion Review, 5*(2), 202–210. https://doi.org/10.1177/1754073912451630

16. Butler, E. A., Hollenstein, T., Shoham, V., & Rohrbaugh, M. J. (2014). A dynamic state-space analysis of interpersonal emotion regulation in couples who smoke. *Journal of Social and Personal Relationships, 31*(7), 907–927. https://doi.org/10.1177/0265407513508732

17. Calatrava, M., Martins, M. V., Schweer-Collins, M., Duch-Ceballos, C., & Rodríguez-González, M. (2022). Differentiation of self: A scoping review of Bowen Family Systems Theory’s core construct. *Clinical Psychology Review, 91*, 102101. [doi:10.1016/j.cpr.2021.102101](https://doi.org/10.1016/j.cpr.2021.102101); PMID 34823190.

18. Calvocoressi, L., Mazure, C. M., Kasl, S. V., Skolnick, J., Fisk, D., Vegso, S. J., Van Noppen, B. L., & Price, L. H. (1999). Family accommodation of obsessive-compulsive symptoms: Instrument development and assessment of family behavior. *The Journal of Nervous and Mental Disease, 187*(10), 636–642. https://doi.org/10.1097/00005053-199910000-00008 (PMID: 10535658).

19. Causier, C., & Salkovskis, P. M. (2025). Fighting OCD together: An experimental study of the effectiveness and acceptability of seeking and receiving emotional support for OCD. *Journal of Behavior Therapy and Experimental Psychiatry, 86*, 101987. https://doi.org/10.1016/j.jbtep.2024.101987 (PMID: 39232282).

20. Cermak, T. L. (1986). Diagnostic criteria for codependency. *Journal of Psychoactive Drugs, 18*(1), 15–20. [https://doi.org/10.1080/02791072.1986.10524475](https://doi.org/10.1080/02791072.1986.10524475). PMID 3701499. [Abstract/verified secondary full-text reconstruction of criteria]

21. Chirkov, V. I., Ryan, R. M., Kim, Y., & Kaplan, U. (2003). Differentiating autonomy from individualism and independence: A self-determination theory perspective on internalization of cultural orientations and well-being. *Journal of Personality and Social Psychology, 84*(1), 97–110. [doi:10.1037/0022-3514.84.1.97](https://doi.org/10.1037/0022-3514.84.1.97); PMID 12518973.

22. Cross, S. E., Bacon, P. L., & Morris, M. L. (2000). The relational-interdependent self-construal and relationships. *Journal of Personality and Social Psychology, 78*(4), 791–808. [doi:10.1037/0022-3514.78.4.791](https://doi.org/10.1037/0022-3514.78.4.791); PMID 10794381.

23. Cullen, J., & Carr, A. (1999). Codependency: An empirical study from a systemic perspective. *Contemporary Family Therapy, 21*(4), 505–526. [https://doi.org/10.1023/A:1021627205565](https://doi.org/10.1023/A:1021627205565). [Full text]

24. Dear, G. E. (2002). The Holyoake Codependency Index: Further evidence of factorial validity. *Drug and Alcohol Review, 21*(1), 47–52. [https://doi.org/10.1080/09595230220119354](https://doi.org/10.1080/09595230220119354). PMID 12189004. [Abstract only]

25. Dear, G. E., & Roberts, C. M. (2000). The Holyoake Codependency Index: Investigation of the factor structure and psychometric properties. *Psychological Reports, 87*(3), 991–1002. [https://doi.org/10.2466/pr0.2000.87.3.991](https://doi.org/10.2466/pr0.2000.87.3.991). PMID 11191419. [Abstract only]

26. Dear, G. E., & Roberts, C. M. (2005). Validation of the Holyoake Codependency Index. *The Journal of Psychology, 139*(4), 293–314. [https://doi.org/10.3200/JRLP.139.4.293-314](https://doi.org/10.3200/JRLP.139.4.293-314). PMID 16097271. [Abstract only]

27. Dear, G. E., Roberts, C. M., & Lange, L. (2004). Defining codependency: A thematic analysis of published definitions. In S. P. Shohov (Ed.), *Advances in Psychology Research* (Vol. 34, pp. 189–205). Nova Science Publishers. Institutional record: [https://ro.ecu.edu.au/ecuworks/7153/](https://ro.ecu.edu.au/ecuworks/7153/). [Institutional abstract]

28. DiGiovanni, A. M., Peters, B. J., Tudder, A., Gresham, A. M., & Bolger, N. (2024). Physiological synchrony in supportive discussions: An examination of co-rumination, relationship type, and heterogeneity. *Psychophysiology, 61*(7), e14554. https://doi.org/10.1111/psyp.14554

29. Dishion, T. J., Forgatch, M., Van Ryzin, M. J., & Winter, C. (2012). The nonlinear dynamics of family problem solving in adolescence: The predictive validity of a peaceful resolution attractor. *Nonlinear Dynamics, Psychology, and Life Sciences, 16*(3), 331–352. PMID: https://pubmed.ncbi.nlm.nih.gov/22695152/ ; full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC3547648/

30. Feeney, B. C. (2004). A secure base: Responsive support of goal strivings and exploration in adult intimate relationships. *Journal of Personality and Social Psychology, 87*(5), 631–648. [https://doi.org/10.1037/0022-3514.87.5.631](https://doi.org/10.1037/0022-3514.87.5.631). PMID: [15535776](https://pubmed.ncbi.nlm.nih.gov/15535776/)

31. Feeney, B. C. (2007). The dependency paradox in close relationships: Accepting dependence promotes independence. *Journal of Personality and Social Psychology, 92*(2), 268–285. https://doi.org/10.1037/0022-3514.92.2.268 (PMID: 17279849).

32. Feeney, B. C., & Collins, N. L. (2001). Predictors of caregiving in adult intimate relationships: An attachment theoretical perspective. *Journal of Personality and Social Psychology, 80*(6), 972–994. [https://doi.org/10.1037/0022-3514.80.6.972](https://doi.org/10.1037/0022-3514.80.6.972). PMID: [11414378](https://pubmed.ncbi.nlm.nih.gov/11414378/)

33. Feeney, B. C., & Collins, N. L. (2003). Motivations for caregiving in adult intimate relationships: Influences on caregiving behavior and relationship functioning. *Personality and Social Psychology Bulletin, 29*(8), 950–968. [doi:10.1177/0146167203252807](https://doi.org/10.1177/0146167203252807); PMID 15189615.

34. Feeney, B. C., & Thrush, R. L. (2010). Relationship influences on exploration in adulthood: The characteristics and function of a secure base. *Journal of Personality and Social Psychology, 98*(1), 57–76. [https://doi.org/10.1037/a0016961](https://doi.org/10.1037/a0016961). PMID: [20053031](https://pubmed.ncbi.nlm.nih.gov/20053031/); PMCID: [PMC2805473](https://pmc.ncbi.nlm.nih.gov/articles/PMC2805473/)

35. Feeney, J. A. (1996). Attachment, caregiving, and marital satisfaction. *Personal Relationships, 3*(4), 401–416. [doi:10.1111/j.1475-6811.1996.tb00124.x](https://doi.org/10.1111/j.1475-6811.1996.tb00124.x).

36. Fischer, J. L., Spann, L., & Crawford, D. (1991). Measuring codependency. *Alcoholism Treatment Quarterly, 8*(1), 87–100. [https://doi.org/10.1300/J020V08N01_06](https://doi.org/10.1300/J020V08N01_06). [Full text visually inspected]

37. Fuller, J. A., & Warner, R. M. (2000). Family stressors as predictors of codependency. *Genetic, Social, and General Psychology Monographs, 126*(1), 5–22. [https://pubmed.ncbi.nlm.nih.gov/10713899/](https://pubmed.ncbi.nlm.nih.gov/10713899/). PMID 10713899; no DOI located. [Abstract only]

38. Gleason, M. E. J., Iida, M., Shrout, P. E., & Bolger, N. (2008). Receiving support as a mixed blessing: Evidence for dual effects of support on psychological outcomes. *Journal of Personality and Social Psychology, 94*(5), 824–838. [https://doi.org/10.1037/0022-3514.94.5.824](https://doi.org/10.1037/0022-3514.94.5.824). PMCID: [PMC3898885](https://pmc.ncbi.nlm.nih.gov/articles/PMC3898885/)

39. Gotham, H. J., & Sher, K. J. (1996). Do codependent traits involve more than basic dimensions of personality and psychopathology? *Journal of Studies on Alcohol, 57*(1), 34–39. [https://doi.org/10.15288/jsa.1996.57.34](https://doi.org/10.15288/jsa.1996.57.34). PMID 8747499. [Abstract only]

40. Granic, I., & Lamey, A. V. (2002). Combining dynamic systems and multivariate analyses to compare the mother-child interactions of externalizing subtypes. *Journal of Abnormal Child Psychology, 30*(3), 265–283. https://doi.org/10.1023/A:1015106913866

41. Granic, I., O'Hara, A., Pepler, D., & Lewis, M. D. (2007). A dynamic systems analysis of parent-child changes associated with successful “real-world” interventions for aggressive children. *Journal of Abnormal Child Psychology, 35*, 845–857. https://doi.org/10.1007/s10802-007-9133-4

42. Grumi, S., Pettenati, G., Manfredini, V., & Provenzi, L. (2022). Flexibility and organization in parent-child interaction through the lens of the dynamic system approach: A systematic review of State Space Grid studies. *Infant Behavior and Development, 67*, 101722. https://doi.org/10.1016/j.infbeh.2022.101722

43. Haaken, J. (1990). A critical analysis of the co-dependence construct. *Psychiatry, 53*(4), 396–406. [https://doi.org/10.1080/00332747.1990.11024522](https://doi.org/10.1080/00332747.1990.11024522). PMID 2263680. [Full text]

44. Halldorsson, B., Salkovskis, P. M., Kobori, O., & Pagdin, R. (2016). “I do not know what else to do”: Caregivers' perspective on reassurance seeking in obsessive-compulsive disorder. *Journal of Obsessive-Compulsive and Related Disorders, 8*, 21–30. https://doi.org/10.1016/j.jocrd.2015.11.003.

45. Hands, M., & Dear, G. E. (1994). Co-dependency: A critical review. *Drug and Alcohol Review, 13*(4), 437–445. [https://doi.org/10.1080/09595239400185571](https://doi.org/10.1080/09595239400185571). PMID 16818359. [Full text]

46. Hawkins, J., Salmon, A., Fernando, S., Battle, C., Esau, S., Snyder, D., & Sikora, M. (2026). “I don’t know what we should have done differently”: A qualitative study on the dilemmas of “tough love” and toxic drugs in British Columbia, Canada. *Drugs: Education, Prevention and Policy, 33*(2), 156–165. Advance online publication 2 May 2025. https://doi.org/10.1080/09687637.2025.2493140

47. Helgeson, V. S., Swanson, J., Ra, O., Randall, H., & Zhao, Y. (2015). Links between unmitigated communion, interpersonal behaviors and well-being: A daily diary approach. *Journal of Research in Personality, 57*, 53–60. [doi:10.1016/j.jrp.2014.12.007](https://doi.org/10.1016/j.jrp.2014.12.007).

48. Helm, J. L., Sbarra, D. A., & Ferrer, E. (2014). Coregulation of respiratory sinus arrhythmia in adult romantic partners. *Emotion, 14*(3), 522–531. https://doi.org/10.1037/a0035960

49. Hermida-Barros, L., Primé-Tous, M., García-Delgar, B., Forcadell, E., Lera-Miguel, S., Fernández de la Cruz, L., Vieta, E., Radua, J., Lázaro, L., & Fullana, M. A. (2024). Family accommodation in obsessive-compulsive disorder: An updated systematic review and meta-analysis. *Neuroscience & Biobehavioral Reviews, 161*, 105678. https://doi.org/10.1016/j.neubiorev.2024.105678 (PMID: 38621516).

50. Hollenstein, T. (2007). State space grids: Analyzing dynamics across development. *International Journal of Behavioral Development, 31*(4), 384–396. https://doi.org/10.1177/0165025407077765

51. Hollenstein, T., Granic, I., Stoolmiller, M., & Snyder, J. (2004). Rigidity in parent-child interactions and the development of externalizing and internalizing behavior in early childhood. *Journal of Abnormal Child Psychology, 32*(6), 595–607. https://doi.org/10.1023/B:JACP.0000047209.37650.41

52. Horváth, Z., & Urbán, R. (2019). Testing the stress–strain–coping–support (SSCS) model among family members of an alcohol misusing relative: The mediating effect of burden and tolerant-inactive coping. *Addictive Behaviors, 89*, 200–205. https://doi.org/10.1016/j.addbeh.2018.10.010

53. Hughes-Hammer, C., Martsolf, D. S., & Zeller, R. A. (1998). Development and testing of the Codependency Assessment Tool. *Archives of Psychiatric Nursing, 12*(5), 264–272. [https://doi.org/10.1016/S0883-9417(98)80036-8](https://doi.org/10.1016/S0883-9417(98)80036-8). PMID 9793213. [Abstract only]

54. Impett, E. A., Gere, J., Kogan, A., Gordon, A. M., & Keltner, D. (2014). How sacrifice impacts the giver and the recipient: Insights from approach–avoidance motivational theory. *Journal of Personality, 82*(5), 390–401. [doi:10.1111/jopy.12070](https://doi.org/10.1111/jopy.12070); PMID 24007506.

55. Irwin, H. J. (1995). Codependence, narcissism, and childhood trauma. *Journal of Clinical Psychology, 51*(5), 658–665. [https://doi.org/10.1002/1097-4679(199509)51:5%3C658::AID-JCLP2270510511%3E3.0.CO;2-N](https://doi.org/10.1002/1097-4679(199509)51:5%3C658::AID-JCLP2270510511%3E3.0.CO;2-N). PMID 8801242. [Abstract only]

56. Jack, D. C., & Dill, D. (1992). The Silencing the Self Scale: Schemas of intimacy associated with depression in women. *Psychology of Women Quarterly, 16*(1), 97–106. [doi:10.1111/j.1471-6402.1992.tb00242.x](https://doi.org/10.1111/j.1471-6402.1992.tb00242.x).

57. Jack, D. C., Brody, L. R., & Sikov, A. (2026). Advancing the next generation of research on self-silencing and depression: A narrative review and synthesis of three decades of research. *Sex Roles, 92*, Article 7. [doi:10.1007/s11199-025-01637-8](https://doi.org/10.1007/s11199-025-01637-8).

58. Jankowski, P. J., & Hooper, L. M. (2012). Differentiation of self: A validation study of the Bowen theory construct. *Couple and Family Psychology: Research and Practice, 1*(3), 226–243. [doi:10.1037/a0027469](https://doi.org/10.1037/a0027469).

59. Kelley, K. N., Caporino, N. E., & Falkenstein, M. J. (2024). Family accommodation in intensive/residential treatment for adults with OCD: A cross-lagged panel analysis. *Behavior Therapy, 55*(2), 391–400. https://doi.org/10.1016/j.beth.2023.07.012 (PMID: 38418048).

60. Kirby, K. C., Benishek, L. A., Kerwin, M. E., Dugosh, K. L., Carpenedo, C. M., Bresani, E., Haugh, J. A., Washio, Y., & Meyers, R. J. (2017). Analyzing components of Community Reinforcement and Family Training (CRAFT): Is treatment entry training sufficient? *Psychology of Addictive Behaviors, 31*(7), 818–827. https://doi.org/10.1037/adb0000306

61. Kluwer, E. S., Karremans, J. C., Riedijk, L., & Knee, C. R. (2020). Autonomy in relatedness: How need fulfillment interacts in close relationships. *Personality and Social Psychology Bulletin, 46*(4), 603–616. [doi:10.1177/0146167219867964](https://doi.org/10.1177/0146167219867964); PMID 31390934; PMCID PMC7057354.

62. Knee, C. R., Canevello, A., Bush, A. L., & Cook, A. (2008). Relationship-contingent self-esteem and the ups and downs of romantic relationships. *Journal of Personality and Social Psychology, 95*(3), 608–627. [doi:10.1037/0022-3514.95.3.608](https://doi.org/10.1037/0022-3514.95.3.608); PMID 18729698.

63. Kobori, O., & Salkovskis, P. M. (2013). Patterns of reassurance seeking and reassurance-related behaviours in obsessive-compulsive disorder and anxiety disorders. *Behavioural and Cognitive Psychotherapy, 41*(1), 1–23. https://doi.org/10.1017/S1352465812000665 (PMID: 22948342).

64. Koestner, R., Powers, T. A., Carbonneau, N., Milyavskaya, M., & Chua, S. N. (2012). Distinguishing autonomous and directive forms of goal support: Their effects on goal progress, relationship quality, and subjective well-being. *Personality and Social Psychology Bulletin, 38*(12), 1609–1620. https://doi.org/10.1177/0146167212457075 (PMID: 22930370).

65. Levenson, R. W., & Gottman, J. M. (1983). Marital interaction: Physiological linkage and affective exchange. *Journal of Personality and Social Psychology, 45*(3), 587–597. https://doi.org/10.1037/0022-3514.45.3.587

66. Lichtwarck-Aschoff, A., Hasselman, F., Cox, R., Pepler, D., & Granic, I. (2012). A characteristic destabilization profile in parent-child interactions associated with treatment efficacy for aggressive children. *Nonlinear Dynamics, Psychology, and Life Sciences, 16*(3), 353–379. PMID: https://pubmed.ncbi.nlm.nih.gov/22695153/

67. Lindley, N. R., Giordano, P. J., & Hammer, E. D. (1999). Codependency: Predictors and psychometric issues. *Journal of Clinical Psychology, 55*(1), 59–64. [https://doi.org/10.1002/(SICI)1097-4679(199901)55:1%3C59::AID-JCLP5%3E3.0.CO;2-M](https://doi.org/10.1002/(SICI)1097-4679(199901)55:1%3C59::AID-JCLP5%3E3.0.CO;2-M). PMID 10100831. [Abstract only]

68. Liu, S., Rovine, M. J., Cousino Klein, L., & Almeida, D. M. (2013). Synchrony of diurnal cortisol pattern in couples. *Journal of Family Psychology, 27*(4), 579–588. https://doi.org/10.1037/a0033735

69. Lunkenheimer, E. S., Olson, S. L., Hollenstein, T., Sameroff, A. J., & Winter, C. (2011). Dyadic flexibility and positive affect in parent-child coregulation and the development of child behavior problems. *Development and Psychopathology, 23*(2), 577–591. https://doi.org/10.1017/S095457941100006X

70. Maisel, N. C., & Gable, S. L. (2009). The paradox of received social support: The importance of responsiveness. *Psychological Science, 20*(8), 928–932. [https://doi.org/10.1111/j.1467-9280.2009.02388.x](https://doi.org/10.1111/j.1467-9280.2009.02388.x). PMID: [19549083](https://pubmed.ncbi.nlm.nih.gov/19549083/)

71. Manzi, C., Vignoles, V. L., Regalia, C., & Scabini, E. (2006). Cohesion and enmeshment revisited: Differentiation, identity, and well-being in two European cultures. *Journal of Marriage and Family, 68*(3), 673–689. [doi:10.1111/j.1741-3737.2006.00282.x](https://doi.org/10.1111/j.1741-3737.2006.00282.x).

72. Marks, A. D. G., Blore, R. L., Hine, D. W., & Dear, G. E. (2012). Development and validation of a revised measure of codependency. *Australian Journal of Psychology, 64*(3), 119–127. [https://doi.org/10.1111/j.1742-9536.2011.00034.x](https://doi.org/10.1111/j.1742-9536.2011.00034.x). [Extended abstract/publisher preview]

73. Marsac, M. L., & Alderfer, M. A. (2011). Psychometric properties of the FACES-IV in a pediatric oncology population. *Journal of Pediatric Psychology, 36*(5), 528–538. [doi:10.1093/jpepsy/jsq003](https://doi.org/10.1093/jpepsy/jsq003); PMID 20147435; PMCID PMC3131703.

74. Martsolf, D. S., Hughes-Hammer, C., Estok, P., & Zeller, R. A. (1999). Codependency in male and female helping professionals. *Archives of Psychiatric Nursing, 13*(2), 97–103. [https://doi.org/10.1016/S0883-9417(99)80026-0](https://doi.org/10.1016/S0883-9417(99)80026-0). PMID 10222638. [Abstract only]

75. Mayo, O., Lavidor, M., & Gordon, I. (2021). Interpersonal autonomic nervous system synchrony and its association to relationship and performance—A systematic review and meta-analysis. *Physiology & Behavior, 235*, 113391. https://doi.org/10.1016/j.physbeh.2021.113391

76. Merkouris, S. S., Rodda, S. N., & Dowling, N. A. (2022). Affected other interventions: A systematic review and meta-analysis across addictions. *Addiction, 117*(9), 2393–2414. https://doi.org/10.1111/add.15825

77. Miller, W. R., Meyers, R. J., & Tonigan, J. S. (1999). Engaging the unmotivated in treatment for alcohol problems: A comparison of three strategies for intervention through family members. *Journal of Consulting and Clinical Psychology, 67*(5), 688–697. https://doi.org/10.1037/0022-006X.67.5.688

78. Molina, E., Taiwo, A. O., & Grey, B. (2026). Co-dependency revisited: An integrative review of conceptualisations and mental health outcomes. *Clinical Psychology & Psychotherapy, 33*(2), e70265. [https://doi.org/10.1002/cpp.70265](https://doi.org/10.1002/cpp.70265). PMID 41958380; PROSPERO CRD42024575573. [Full text]

79. Muyingo, L., Smith, M. M., Sherry, S. B., McEachern, E., Leonard, K. E., & Stewart, S. H. (2020). Relationships on the rocks: A meta-analysis of romantic partner effects on alcohol use. *Psychology of Addictive Behaviors, 34*(6), 629–640. https://doi.org/10.1037/adb0000578

80. Orford, J., Copello, A., Velleman, R., & Templeton, L. (2010a). Family members affected by a close relative’s addiction: The stress–strain–coping–support model. *Drugs: Education, Prevention and Policy, 17*(sup1), 36–43. https://doi.org/10.3109/09687637.2010.514801

81. Orford, J., Velleman, R., Copello, A., Templeton, L., & Ibanga, A. (2010b). The experiences of affected family members: A summary of two decades of qualitative research. *Drugs: Education, Prevention and Policy, 17*(sup1), 44–62. https://doi.org/10.3109/09687637.2010.514192

82. Overall, N. C., Fletcher, G. J. O., & Simpson, J. A. (2010). Helping each other grow: Romantic partner support, self-improvement, and relationship quality. *Personality and Social Psychology Bulletin, 36*(11), 1496–1513. https://doi.org/10.1177/0146167210383045 (PMID: 20858888).

83. Palumbo, R. V., Marraccini, M. E., Weyandt, L. L., Wilder-Smith, O., McGee, H. A., Liu, S., & Goodwin, M. S. (2017). Interpersonal autonomic physiology: A systematic review of the literature. *Personality and Social Psychology Review, 21*(2), 99–141. https://doi.org/10.1177/1088868316628405

84. Pincus, A. L., & Wilson, K. R. (2001). Interpersonal variability in dependent personality. *Journal of Personality, 69*(2), 223–251. [doi:10.1111/1467-6494.00143](https://doi.org/10.1111/1467-6494.00143); PMID 11339797.

85. Randall, A. K., Tao, C., Leon, G., & Duran, N. D. (2021). Couples' co-regulation dynamics as a function of perceived partner dyadic coping. *Anxiety, Stress, & Coping, 34*(6), 597–611. https://doi.org/10.1080/10615806.2021.1912740

86. Reed, R. G., Barnard, K., & Butler, E. A. (2015). Distinguishing emotional coregulation from codysregulation: An investigation of emotional dynamics and body weight in romantic couples. *Emotion, 15*(1), 45–60. https://doi.org/10.1037/a0038561

87. Reed, R. G., Randall, A. K., Post, J. H., & Butler, E. A. (2013). Partner influence and in-phase versus anti-phase physiological linkage in romantic couples. *International Journal of Psychophysiology, 88*(3), 309–316. https://doi.org/10.1016/j.ijpsycho.2012.08.009

88. Remen, A. L., Chambless, D. L., & Rodebaugh, T. L. (2002). Gender differences in the construct validity of the Silencing the Self Scale. *Psychology of Women Quarterly, 26*(2), 151–159. [doi:10.1111/1471-6402.00053](https://doi.org/10.1111/1471-6402.00053).

89. Righetti, F., Sakaluk, J. K., Faure, R., & Impett, E. A. (2020). The link between sacrifice and relational and personal well-being: A meta-analysis. *Psychological Bulletin, 146*(10), 900–921. [doi:10.1037/bul0000297](https://doi.org/10.1037/bul0000297); PMID 32673005.

90. Rodríguez-González, M., Bell, C. A., Pereyra, S. B., Martínez-Díaz, M. P., Schweer-Collins, M., & Bean, R. A. (2023). Differentiation of self and relationship attachment, quality, and stability: A path analysis of dyadic and longitudinal data from Spanish and U.S. couples. *PLOS ONE, 18*(3), e0282482. [doi:10.1371/journal.pone.0282482](https://doi.org/10.1371/journal.pone.0282482); PMID 36862686; PMCID PMC9980780.

91. Roozen, H. G., de Waart, R., & van der Kroft, P. (2010). Community reinforcement and family training: An effective option to engage treatment-resistant substance-abusing individuals in treatment. *Addiction, 105*(10), 1729–1738. https://doi.org/10.1111/j.1360-0443.2010.03016.x

92. Rotunda, R. J., & Doman, K. (2001). Partner enabling of substance use disorders: Critical review and future directions. *The American Journal of Family Therapy, 29*(4), 257–270. https://doi.org/10.1080/01926180126496

93. Rushton, C., Kelly, P. J., Raftery, D., Beck, A., & Larance, B. (2023). The effectiveness of psychosocial interventions for family members impacted by another’s substance use: A systematic review and meta-analysis. *Drug and Alcohol Review, 42*(4), 960–977. https://doi.org/10.1111/dar.13607

94. Ryan, R. M., & Deci, E. L. (2017). *Self-determination theory: Basic psychological needs in motivation, development, and wellness*. Guilford Press. ISBN 978-1-4625-2876-9. Stable publisher record: [https://www.guilford.com/books/Self-Determination-Theory/Ryan-Deci/9781462538966](https://www.guilford.com/books/Self-Determination-Theory/Ryan-Deci/9781462538966)

95. Salkovskis, P. M., & Kobori, O. (2015). Reassuringly calm? Self-reported patterns of responses to reassurance seeking in obsessive compulsive disorder. *Journal of Behavior Therapy and Experimental Psychiatry, 49*(Part B), 203–208. https://doi.org/10.1016/j.jbtep.2015.09.002 (PMID: 26433701).

96. Shaver, P. R., Schachner, D. A., & Mikulincer, M. (2005). Attachment style, excessive reassurance seeking, relationship processes, and depression. *Personality and Social Psychology Bulletin, 31*(3), 343–359. [doi:10.1177/0146167204271709](https://doi.org/10.1177/0146167204271709); PMID 15657450.

97. Siljeholm, O., Edvardsson, K., Bergström, M., & Hammarberg, A. (2024). Community Reinforcement and Family Training versus counselling for parents of treatment-refusing young adults with hazardous substance use: A randomized controlled trial. *Addiction, 119*(5), 915–927. https://doi.org/10.1111/add.16429

98. Skowron, E. A., & Friedlander, M. L. (1998). The Differentiation of Self Inventory: Development and initial validation. *Journal of Counseling Psychology, 45*(3), 235–246. [doi:10.1037/0022-0167.45.3.235](https://doi.org/10.1037/0022-0167.45.3.235).

99. Skowron, E. A., & Schmitt, T. A. (2003). Assessing interpersonal fusion: Reliability and validity of a new DSI Fusion With Others subscale. *Journal of Marital and Family Therapy, 29*(2), 209–222. [doi:10.1111/j.1752-0606.2003.tb01201.x](https://doi.org/10.1111/j.1752-0606.2003.tb01201.x); PMID 12728779.

100. Starr, L. R., & Davila, J. (2008). Excessive reassurance seeking, depression, and interpersonal rejection: A meta-analytic review. *Journal of Abnormal Psychology, 117*(4), 762–775. [doi:10.1037/a0013866](https://doi.org/10.1037/a0013866); PMID 19025224.

101. Stewart, K. E., Sumantry, D., & Malivoire, B. L. (2020). Family and couple integrated cognitive-behavioural therapy for adults with OCD: A meta-analysis. *Journal of Affective Disorders, 277*, 159–168. https://doi.org/10.1016/j.jad.2020.07.140 (PMID: 32828003).

102. Thompson-Hollands, J., Abramovitch, A., Tompson, M. C., & Barlow, D. H. (2015). A randomized clinical trial of a brief family intervention to reduce accommodation in obsessive-compulsive disorder: A preliminary study. *Behavior Therapy, 46*(2), 218–229. https://doi.org/10.1016/j.beth.2014.11.001 (PMID: 25645170; PMCID: PMC4748371).

103. Timko, C., Grant, K. M., Han, X., Young, L. B., & Cucciare, M. A. (2022). Al-Anon Intensive Referral to facilitate concerned others’ participation in Al-Anon Family Groups: A randomized controlled trial. *Addiction, 117*(3), 590–599. https://doi.org/10.1111/add.15670

104. Timmons, A. C., Margolin, G., & Saxbe, D. E. (2015). Physiological linkage in couples and its implications for individual and interpersonal functioning: A literature review. *Journal of Family Psychology, 29*(5), 720–731. https://doi.org/10.1037/fam0000115

105. Vederhus, J.-K., Kristensen, Ø., & Timko, C. (2019). How do psychological characteristics of family members affected by substance use influence quality of life? *Quality of Life Research, 28*(8), 2161–2170. [https://doi.org/10.1007/s11136-019-02169-x](https://doi.org/10.1007/s11136-019-02169-x). PMID 30895489; PMCID PMC6620238. [Full text]

106. Vowels, L. M., & Carnelley, K. B. (2022). Partner support and goal outcomes: A multilevel meta-analysis and a methodological critique. *European Journal of Social Psychology, 52*(4), 679–694. https://doi.org/10.1002/ejsp.2846.

107. Werner, P. D., Green, R.-J., Greenberg, J., Browne, T. L., & McKenna, T. E. (2001). Beyond enmeshment: Evidence for the independence of intrusiveness and closeness-caregiving in married couples. *Journal of Marital and Family Therapy, 27*(4), 459–471. [doi:10.1111/j.1752-0606.2001.tb00340.x](https://doi.org/10.1111/j.1752-0606.2001.tb00340.x); PMID 11594014.

108. Wu, M. S., McGuire, J. F., Martino, C., Phares, V., Selles, R. R., & Storch, E. A. (2016). A meta-analysis of family accommodation and OCD symptom severity. *Clinical Psychology Review, 45*, 34–44. https://doi.org/10.1016/j.cpr.2016.03.003 (PMID: 27019367).

109. Zee, K. S., & Bolger, N. (2023). Physiological coregulation during social support discussions. *Emotion, 23*(3), 825–843. https://doi.org/10.1037/emo0001107

110. Zniva, R., Pauli, P., & Schulz, S. M. (2017). Overprotective social support leads to increased cardiovascular and subjective stress reactivity. *Biological Psychology, 123*, 226–234. https://doi.org/10.1016/j.biopsycho.2016.12.009 (PMID: 28017786).

### Non-core authority records

These sources establish cultural provenance or current diagnostic status; they are not part of the 110-source scientific core.

- **Al-Anon Family Group Headquarters.** *Al-Anon timeline* and *FAQ about Al-Anon’s history*. [Official timeline](https://al-anon.org/for-members/wso/archives/al-anon-timeline/) · [Official history FAQ](https://al-anon.org/for-members/wso/archives/faq-al-anons-history/). These authority webpages establish the organization’s 1951–1952 history, not the origin of the word *codependency*.
- **Beattie, M. (1986).** *Codependent No More: How to Stop Controlling Others and Start Caring for Yourself*. Hazelden. ISBN 0-89486-402-5. [Official publisher author/history page](https://www.hazelden.org/store/author/4?Melody-Beattie=). Retained as a popularizing self-help artifact, not as scientific validation; the defensible description is “first published by Hazelden in 1986; widely circulated 1987 edition.”
- **American Psychiatric Association. (2022).** *Diagnostic and Statistical Manual of Mental Disorders* (5th ed., text rev.; DSM-5-TR). APA Publishing. [https://doi.org/10.1176/appi.books.9780890425787](https://doi.org/10.1176/appi.books.9780890425787) · [Official DSM page](https://www.psychiatry.org/psychiatrists/practice/dsm/about-dsm). The manual and cumulative official updates through September 2025 contain no diagnosis named *codependency*; absence is non-recognition, not evidence of a formal committee rejection.
- **World Health Organization. (2024).** *Clinical descriptions and diagnostic requirements for ICD-11 mental, behavioural and neurodevelopmental disorders*. WHO. ISBN 978-92-4-007726-3. [Official publication](https://www.who.int/publications/i/item/9789240077263) · [ICD-11 MMS browser, January 2026 release](https://icd.who.int/browse/2026-01/mms/en). No *codependency* diagnosis appears; nearby relationship-distress codes are non-disorder health-status/contact categories.