Relational Memory Reconsolidation: A New Lens on Change Mechanisms in EFT for Couples

ABSTRACT

This paper presents a new framework to account for change mechanisms in Emotion‐Focused Therapy for Couples (EFT). Building on Lane et al.'s (2015) integrated memory model, the Relational Memory Reconsolidation model described here posits that EFT reshapes partners' relational memory structures through emotionally intense moments of expressed vulnerability and responsiveness. These vulnerability‐responsiveness events allow partners to access and transform memories of past relational experiences (i.e., episodic memory) as well as associated beliefs about themselves and the relationship (i.e., semantic memory). This process of emotional activation and reconsolidation allows maladaptive interaction patterns, which are rigid and stressful, to evolve into more adaptive, emotionally responsive exchanges. These emotional moments activate and transform relational self‐states that are responsive to interpersonal cues and can shape partners' self‐concepts in a relational context. As therapy progresses and emotionally significant interactions are repeated, the coactivation of adaptive and maladaptive self‐states updates neocortical memory traces, leading to changes in higher‐order relational semantic structures (i.e., the relational self). These moments foster alignment in partners' perceptions of their shared reality, characterized by mutuality in feelings, practices, memories, goals, and identity. Clinically, the framework highlights the need to elicit one partner's core vulnerable emotions and, within the reconsolidation window, disconfirm related expectations (e.g., rejection) through the other partner's empathic and supportive response. This process may transform key components of partners' relational memory, including relational episodic memories, semantic structures, and procedural emotional responses.

1. Introduction

Emotion‐focused therapy (EFT) for couples is an evidence‐based model (Rathgeber et al. 2019; Spengler et al. 2022) that integrates systemic, humanistic, and experiential therapies (Greenberg and Goldman 2008; Johnson 2020). The original EFT model for couples was developed by Greenberg and Johnson (1988). Over time, their joint model evolved into two distinct therapeutic approaches: Johnson's Emotionally Focused Couple Therapy (EFCT; Johnson 2020), which is primarily grounded in attachment theory, and Greenberg and Goldman's Emotion‐Focused Couple Therapy (EFT‐C; Greenberg and Goldman 2008), which is mainly rooted in dyadic regulation and affect transformation (Rathgeber et al. 2019).

Throughout this paper, we use the term EFT as an umbrella term to refer to both Greenberg and Goldman's EFT‐C and Johnson's EFCT, while preserving their conceptual distinctions where appropriate. Notwithstanding differences in theory and terminology, both approaches emphasize that the key therapeutic process leading to positive treatment outcomes is in‐session events involving the expression of and responsiveness to partners' emotional vulnerabilities (Bradley and Furrow 2004; Greenman and Johnson 2013; McKinnon and Greenberg 2013; Tilley and Palmer 2013). Drawing on recent models of memory and theories emphasizing the interpersonal nature of the self, vulnerability‐responsiveness events are posited to be a common change mechanism across EFT models that exert their effects through the meta‐process of relational memory reconsolidation.

2. EFT Model of Distress and Change

The two main theoretical strands that guide the practice of EFT consist of distinct yet related conceptualizations of relational dysfunction. Both frameworks suggest that relational distress is maintained by behavioral patterns of rigid, self‐reinforcing interactional cycles driven by the partners' underlying emotions.

Johnson's EFCT framework (Johnson 2020) is primarily rooted in adult attachment theory (Bowlby 1969), which posits that relational distress emerges when partners' attachment needs (e.g., accessibility, responsiveness, engagement) are unmet, due to maladaptive attachment‐related cycles (Johnson 2020). For example, in a typical pursuer‐withdrawer cycle, one partner's core vulnerability may be loneliness and a fear of rejection; however, rather than revealing these primary emotions, the partner reacts by displaying secondary (“reactive”) emotions, such as anger through criticism and demand. The more partner A demands and criticizes, the more partner B feels emotionally overwhelmed and withdraws; however, the more partner B disengages, the more partner A feels abandoned and demands more. Couples who engage in this self‐perpetuating interactional pattern struggle to feel closeness, proximity, and attachment security in their relationship and, rather, experience anxiety, loneliness, and despair (Halchuk et al. 2010). In EFCT, relationship distress is fundamentally framed as separation distress stemming from perceived threats to attachment security. Such threats activate insecure attachment strategies, most commonly anxious pursuit or avoidant withdrawal, shaping the couple's interactional cycle. In other words, primary emotions are understood as arising from attachment distress (e.g., loneliness, separation anxiety), while secondary emotions reflect attachment coping strategies (e.g., angry protest, despair for connection).

Greenberg and Goldman's EFT‐C framework (Greenberg and Goldman 2008) primarily focuses on the processes of dyadic affect regulation and transformation. It is based on the assumption that relational distress occurs when partners do not experience the fulfillment of their emotion‐based needs within their relationships. In contrast to Johnson's EFCT, EFT‐C conceptualizes attachment as one motivational system among others, gives distinct emphasis to identity maintenance processes, and highlights self‐soothing as complementary to interpersonal soothing within couple interactions (Goldman and Greenberg 2013). Specifically, EFT‐C centers on three affect‐laden motivational/need systems: (a) the attachment system, which is related to the motivation to feel relational security, and to soothe or transform feelings of sadness, anxiety, and loneliness; (b) the identity system, which is related to the motivation to maintain a sense of self‐worth, and downregulate feelings of shame; and (c) the attraction system, which is related to the motivation to feel relational excitement, and upregulate positive feelings (Greenberg and Goldman 2008). Dyadic affect dysregulation results from partners' difficulties addressing emotion‐based motivations (e.g., feeling secure and validated) when interacting with each other, as a result of maladaptive relational cycles.

Notably, Greenberg and Goldman (2008) distinguish between pursue‐distance attachment‐based cycles (which map closely onto EFCT's pursuer‐withdrawer cycles) and dominance‐submission identity‐based cycles. The cycles in the domain of identity often involve shame and inadequacy as partners' core vulnerability, anger and fear as secondary reactive emotions, and domineering and submissive secondary behaviors as partners' ineffective efforts at self‐protection (Greenberg and Goldman 2008; also referred to as attempted solutions; Fisch et al. 1982). These attempts inadvertently maintain and exacerbate the vicious cycle, leading to dyadic dysregulation (Greenberg and Goldman 2008).

Greenberg and Johnson (1988) initially conceptualized EFT for couples as consisting of a 9‐step model of change, subdivided into three overarching phases. EFCT maintained this initial structure, while also incorporating findings from attachment research (Johnson 2020) and empirically validating the benefits of choreographing key moments of change, such as blamer softening events (Burgess Moser et al. 2015). EFT‐C expanded the original 3‐phase model into a five‐stage model that places greater emphasis on the interplay between interpersonal and intrapersonal emotional processes in adult relationships (Greenberg and Goldman 2008).

Despite such variations in emphasis, both theoretical perspectives consider that the treatment process unfolds across the two critical phases. The first phase of change, cycle de‐escalation (the first stage in EFCT, the second stage in EFT‐C), can be regarded as a first‐order systemic change, during which the partners are helped to observe and externalize their distress resulting from the interactional cycle (Johnson 2020). The partners learn to identify their habitual interactional cycles as the source of the problem rather than continuing to view each other as the enemy (Johnson 2020). The end state of this phase is often characterized by mitigated levels of distress and the partners' increased motivation to explore new ways of interacting with each other (Wiebe and Johnson 2016).

The second phase of change can be regarded as a second‐order systemic change, during which the partners transform their habitual interactional positions. Across both models, partners are helped to access more vulnerable emotions (e.g., shame, fear of abandonment), express these to each other, and respond with compassion and care (Kula et al. 2021). In EFCT, this phase is termed restructuring interactions (Stage 2) and focuses on restructuring the partners' positions within the negative interactional cycle, particularly through key change events such as withdrawer re‐engagement and blamer softening (Wiebe and Johnson 2016). In withdrawer re‐engagement, the more withdrawn partner becomes more emotionally accessible and responsive, whereas in blamer softening, the more critical or pursuing partner shifts from protest and blame toward the expression of more vulnerable attachment needs. In EFT‐C, the comparable phase, termed restructuring the negative interaction (Stage 4), similarly reflects enactments aimed at promoting expression and acceptance of vulnerable feelings, thereby transforming partners' emotion‐based interactions. These in‐session transactions, broadly conceptualized as change events or vulnerability‐responsiveness events, may produce temporary distress but have been shown to promote greater long‐term relational security (Burgess Moser et al. 2015).

The current paper puts forward an integrative conceptual framework grounded in the notion of relational memory reconsolidation to better understand how these key change processes transform the relational system. Lane et al. (2015) argued that therapeutic change can be understood as the transformation of emotionally‐laden memories that alters the self by changing clients' episodic memories, semantic knowledge (of the self and the world), and habitual procedural emotional responses. Building on this concept and taking into account the relational nature of the self (Andersen and Chen 2002; Andersen et al. 2017), the Relational Memory Reconsolidation model posits that vulnerability‐responsiveness events allow the partners to revisit implicit and explicit memory structures concerning the self in general and the relational self in particular. This process transforms the partners' feelings and experiences while in each other's presence.

3. Memory Reconsolidation

Two main competing neurobiological models have been put forward to explain how long‐term memories are consolidated. The first, known as the standard theory (Squire and Alvarez 1995), considered that consolidation starts in the hippocampus. Given the hippocampus' plasticity, namely, its faster synaptic changes as compared to the neocortex, it was thought to provide the first crucial yet temporary window for information encoding. In this model, the neural connections mediating memory retrieval are believed to gradually shift from the hippocampus to the neocortex until both episodic and semantic memory become independent of hippocampal activity (McClelland et al. 1995).

Mounting evidence, however, suggests that the hippocampal region and its neural connections to the neocortex play a vital role in long‐term episodic memory long after the initial consolidation stage (Nadel et al. 2000). To incorporate these findings, Nadel and Moscovitch (1997) put forward Multiple Trace Theory (MTT), which posits that the hippocampus becomes reactivated after each memory retrieval (Gilboa et al. 2004), thus creating new memory traces. Here, the hippocampus is seen as the host of spatial‐contextual memory representations (Ryan, Cox, et al. 2008) that continually links and indexes memory traces dispersed in the neocortex, which primarily hosts content memory representations (Nadel and Hardt 2011). Given the varying contexts of memory retrieval, each newly created trace is somewhat different and thus may strengthen or alter the original consolidated memory. It was suggested that memory representations are modified over time through a cumulative process involving overlapping memory traces (Nadel and Hardt 2011). This process alters the likelihood that these memories will be successfully retrieved in the future by enhancing or diminishing the chances of recall (Nadel et al. 2000). Thus, unlike the standard model, MTT views memory as inherently malleable, with the potential for fundamental change upon its reactivation (Elsey and Kindt 2017).

4. Memory Reconsolidation in Therapy

This model of memory reconsolidation has since been explicitly applied to inform treatment design, such as Reconsolidation of Traumatic Memories (RTM; Gray et al. 2019; see also Coherence Therapy; Ecker et al. 2012). However, beyond informing the development of new treatment modalities, Lane et al. (2015) proposed that the process of memory reconsolidation could underlie evidence‐based therapies. Lane et al. presented the integrated memory model to explain the mechanism of therapeutic change in terms of memory reconsolidation processes. It involved three associated components: (a) episodic‐autobiographical memory covering specific and personal events which, when remembered, are relived from a first‐person perspective, (b) semantic structures, which refer to general factual knowledge devoid of contextual or experienced content, and (c) emotional responses pertaining to nondeclarative procedural processes. Given that all three components are inextricably linked (Hoscheidt et al. 2013; Ryan, Hoscheidt, and Nadel 2008), a change in one memory component should lead to changes in the other two.

Lane et al. (2015) also posited that different types of therapy access one of the three associative components (i.e., episodic, semantic, and emotional responses) as an entry point to transform the integrated memory structure. For example, experiential therapies work through activating and changing emotional responses. Individual Emotion‐Focused Therapy is one such empirically‐based type of experiential therapy (Greenberg 2017). Therapeutic change in this model can be conceptualized as a process of intrapersonal affect transformation, where the objective is changing emotion with emotion (Greenberg 2021). Over the course of therapy, emotional transformation takes place by making the often‐avoided vulnerable emotion (e.g., feeling unwanted and rejected) accessible and fully felt. While the painful emotion is activated, the therapist helps the clients get in touch with their associated core needs (e.g., feeling accepted and validated), which often helps clients access and reintegrate more adaptive emotions (e.g., empowering anger for setting limits, sadness for the grieving of loss, self‐compassion for accepting oneself). Through this process of coactivation of painful and more adaptive emotions, individuals' automatic procedural emotional responses are expected to be transformed (Greenberg 2017). This newly synthesized emotional state is believed to drive long‐term changes in one's non‐declarative emotional patterns.

This experiential process provides access to integrated memory structures primarily by eliciting mental and bodily emotional responses (Lane et al. 2015). Importantly, this is often accompanied by the activation of painful episodic memories (e.g., a criticizing parent) that now lend themselves to reconsolidation. Through the activation of memories in which similar vulnerable emotions were experienced, the new emotional elements (e.g., self‐compassion) can be incorporated and re‐encoded as new memory traces. Similarly, this process often transforms one's semantic knowledge of the self (e.g., “I'm an adequate person”) and the world (e.g., “others will not reject me”).

5. Relational Memory Reconsolidation

Lane et al.'s (2015) model only indirectly acknowledges the interpersonal nature of the memory reconsolidation process. The relational aspect in their model is apparent mainly in their identification of the need for an emotionally attentive therapist to foster corrective emotional experiences. Across the variety of therapeutic modalities, the therapist is tasked with creating a safe and accepting environment in which the client can recall, process, and transform painful memories. However, what is lacking in this formulation is the acknowledgment of the extent to which memories (Hirst and Echterhoff 2012; Pasupathi et al. 1998, 2009), emotions (Parkinson 1996), and the self (Andersen and Chen 2002) are relationally co‐constructed. This is particularly crucial in the context of couple therapy, where change in the relational arena is at the epicenter.

Here, we suggest that, across various forms of couple therapy, the target of change can be more accurately framed as a transformation in relational integrated memory structures. In other words, the aim of couple therapy can be understood as producing a change in partners' (a) relational episodic memories, (b) semantic knowledge concerning their relational self (i.e., who I am as a partner) and relationship, and (c) dyadic emotional responses. As noted, in the context of EFT for couples, relational memory reconsolidation primarily relies on in‐session episodes of vulnerability‐responsiveness, in which one partner discloses vulnerable emotions and receives compassionate, empathic responses from the other partner. In other words, whereas in individual EFT, the primary change process of one's integrated memory structure can be regarded as intrapersonal affect‐transformation (changing emotion with another emotion; Greenberg 2021), in the context of EFT‐C, it involves interpersonal affect‐transformation (changing emotion with the partner's emotions). Relatedly, in the EFCT model, partners' disclosure of attachment‐related vulnerabilities paves the way for corrective attachment experiences of emotional availability and caring, which then modify their relational integrated memory structures (Johnson 2020).

This modification of integrated memory structures may also be understood as the updating of internal working models in accordance with Bowlby's (1969) developmental framework. Internal working models can be understood as relationally integrated memory structures, insofar as they organize episodic memories of attachment interactions, semantic beliefs about the self and others (e.g., “I am unworthy,” “others are unavailable”), and associated procedural emotional responses (e.g., pursuing closeness, withdrawal from connection). From a reconsolidation perspective, vulnerability‐responsiveness events may destabilize these attachment‐based expectations and update both self‐ and other‐representations. For example, a partner who expects rejection (“I am too much; others withdraw”) may, through repeated corrective experiences of responsiveness, revise this model toward “I can be vulnerable and remain connected.”

This reconceptualization of internal working models as relational integrated memory structures also clarifies how different therapeutic approaches may access and modify distinct components of this system. In line with Lane et al.'s (2015) formulation of individual EFT, we identify both EFT‐C and EFCT as two experiential models that primarily target the associative memory component of experiential‐emotional responses as they manifest explicitly and relationally. In a similar vein, other couple therapy modalities may explain change as related to other associative components of the relational integrated memory structure, such as cognitive modifications of semantic structures (e.g., CBCT; Fischer et al. 2016), or behavioral changes of implicit procedural responses (BCT; Liberman 1970). That is, the relational memory reconsolidation framework may certainly extend to other couple therapeutic contexts, and certain modalities are likely to operate beyond modifying a single associative component (e.g., IBCT aimed at altering both dysfunctional semantic structures and emotional responses; Christensen et al. 2020).

More broadly, since the relational memory reconsolidation framework is concerned with explaining the intervention rather than the therapy model per se, emotional processing has been proposed as a transtheoretical mechanism (Pascual‐Leone 2018). To the extent that any therapeutic intervention incorporates experiential components, it may engage similar reconsolidation processes by activating and modifying emotional responses, regardless of its theoretical origin. However, a detailed elucidation of the specific processes underlying other evidence‐based couple (or family) therapeutic modalities is beyond the scope of this paper, which restricts itself to models that explicitly target change through the couple's experiential work (i.e., EFCT and EFT‐C).

Our relational reconsolidation framework aligns with the constructivist and systemic traditions, which argue that couples and families collectively co‐construct memories (Dallos and Draper 2024) that influence shared recollection (Harris et al. 2014, 2017). The relational reconsolidation model grounds such theoretical claims in recent findings from social, personality, and clinical neuropsychological research that emphasize the primacy of shared group‐level processes unfolding between rather than within individuals (Gross and Medina‐DeVilliers 2020; Wilson and Coan 2021).

Specifically, we draw on three main empirical frameworks to describe this process: (a) the self‐states perspective, which emphasizes the contextual and temporal nature of personality (Lazarus and Rafaeli 2023); (b) the relational self, a specific self‐state pertaining to one's interpersonal self‐concept that is activated by the presence of a particular figure (e.g., one's partner) or cues in the environment associated with this figure (Andersen and Chen 2002); and (c) shared reality, the perceived commonality of partners' inner states, including the development of a shared identity, where partners integrate each other's perspectives and identities into their sense of self (Rossignac‐Milon and Higgins 2018).

5.1. The Self‐States Perspective

Various classical psychological theories (e.g., James 1890/1950; Kelly 1955; Putnam 1988), as well as modern ones (e.g., Lazarus et al. 2020), assume that the self is not a monolithic entity but rather a plurality of self‐parts or self‐states that dynamically interact with each other and the environment (for recent reviews, see both Dimaggio and Stiles 2007; Lazarus and Rafaeli 2023). In the social and personality literature, for example, this multiplicity perspective emerged from a long‐standing debate on the extent to which traits can reliably predict one's behavior (Kashima 2019).

In response to traditional trait theory's dubious assumption regarding within‐person trait stability across contexts, Mischel and Shoda (1995) suggested viewing personality as a system of cognitive‐affective mediating units (self‐states) that interact with environmental cues to produce behavior. For example, two individuals may exhibit the same “average” level of trait neuroticism. However, one may report distress primarily during social situations while remaining stoic during a challenging cognitive task, whereas the other may exhibit the opposite pattern. This unitary trait‐level approach thus fails to explain individual, context‐dependent differences and highlights the need for an approach that accounts for variability across situations. Fleeson (2001) put forward one such model that conceptualizes personality traits in terms of density distributions. In this model, personality is characterized by central tendencies (e.g., the extent to which one tends to be neurotic) and state‐level distributions (e.g., the extent to which one shows between‐situation variability in neuroticism).

In the clinical literature, this multiplicity of the self is recognized across various theoretical schools (Lazarus et al. 2020). This is referred to as objects (Klein 1952), self‐objects (Kohut 1977), voices (Stiles 1999), parts (Schwartz and Sweezy 2019, internal family systems therapy), activated working selves (Markus and Wurf 1987), modes (Mitchell 2000, relational psychodynamic therapy; Rafaeli et al. 2010, schema therapy), and self‐states (Bromberg 1996). These emotion schemes are the implicit sources of experience that may be triggered by specific internal or external stimuli (Pos and Greenberg 2007; Smith and Greenberg 2007).

One example is the self‐state of shame, which can be triggered by perceptual‐situational cues, such as criticism from one's partner, leading to the resurfacing of episodic memories of failing to live up to the expectations of others or oneself. Activating this state may further exacerbate and reinforce one's symbolic‐conceptual self‐representation (e.g., “you are a failure and always were”), accompanied by painful bodily sensations (e.g., pressure in the upper chest). In an attempt to obtain safety, an action tendency to hide and withdraw is likely to be triggered, which is reflected, for example, in self‐silencing (Elliott et al. 2004). In other words, self‐states (i.e., emotion schemes) go beyond one's subjective emotional experience, in that they are intertwined with episodic, cognitive, and behavioral elements. Consequently, as adaptive and maladaptive self‐states are coactivated and synthesized, new higher‐order self‐states may propel significant shifts in one's self‐organization (Greenberg 2017).

Importantly, one's “self‐states” are often embedded and triggered within an interpersonal context (Bromberg 1996; Mitchell 2000; Sullivan 1953/2013). A recent study found that a substantial portion of the within‐person variability in one's personality traits is relational and uniquely expressed in the presence of specific others (Lakey et al. 2022). We assume that in distressed relationships, undesired relational self‐states such as those of inadequacy and insecurity are repeatedly triggered in the presence of one's partner. Therefore, by activating specific self‐states in couple therapy and promoting supportiveness and empathy toward these states on the part of the partner, the therapist can help the partners transform their self‐states, especially those that are relational in nature.

5.2. The Relational Self

The “relational self” is a concept introduced by Andersen and Chen (2002), who argue that the self is inextricably linked to one's representations of significant others. The relational self refers to the connections in memory between self‐knowledge and knowledge of significant others (e.g., me with my wife), including attributes and role‐based aspects of that specific relationship (e.g., being a caring husband). The relational self exists at differing levels of specificity, ranging from the most general (me across relationships) to the most specific (me with my partner when we argue about household chores). The relational self is activated by internal or external cues of the significant other (Chen et al. 2006). Over time, repeated activation of relationship‐specific selves (e.g., last weekend, my mother criticized me again) crystallizes into more generalized relational selves (e.g., I am unworthy of others' love). These differing levels of the relational self, from specific state‐like to general trait‐like, align with the continuum of memory abstraction described by Renoult et al. (2012), where unique episodic events related to hippocampal activation increasingly transform into semantic memories stored in prefrontal and parietal cortical areas (Renoult et al. 2016). Therefore, the relational self reflects a continuum in which specific dyadic experiences are interlinked with broader semantic self‐schemas, thus shaping relational expectations and identity over time.

The relational self‐model is heavily grounded in the psychodynamic notion of transference, where past emotions, expectations, and the behavior of significant others are projected onto new relationships (Glassman and Andersen 1999). Transference is often accompanied by specific experiences of the self (e.g., “unworthy”; Hinkley and Andersen 1996) along with interpersonal expectations (e.g., “to be rejected”; Berk and Andersen 2000). EFT models for couples are grounded in experiential‐humanistic and systemic theories (Greenberg and Goldman 2008; Johnson 2020). The focus is on understanding present interactional cycles between partners rather than unconscious projections of past internalized figures. However, transference can be discussed in terms of the partners' past sensitivities (Greenberg and Goldman 2008) or attachment histories (Johnson 2020); that is, the unresolved relational pain each partner carries from early bonds. Couples' negative interactional cycles are maintained by the residual influence of these historical wounds that shape their behavior, perceptions, and interpretations of the relationship and the other (e.g., my partner is as withdrawn as my depressed mother was).

These relational expectations, shaped by prior experiences, can amplify the negative meanings associated with partners' behaviors when activated in current interactions. Specifically, certain behaviors by one's partner come to be identified as threatening (e.g., rejection) and thus trigger vulnerable emotions, such as a fear of abandonment or inadequacy. A partner experiencing such vulnerable emotions often responds with related secondary emotions, like anger. This response, in turn, invokes the other's vulnerable emotions, based on this person's past wounds and attachment injuries (e.g., excessive guilt and shame). In this manner, the maladaptive interactional cycle persists.

As noted, the primary change mechanism in EFT is considered to unfold during in‐session sequences of vulnerability‐responsiveness events. In these moments, the partners' higher‐order semantic structures of their specific (e.g., “I'm a failing partner”) and general (e.g., “I'm unworthy of respect”) relational selves are activated. Thus, having new emotionally corrective experiences with one's partner will likely transform the memory structure of one's relational self at various levels of specificity. Studies suggest that in EFT, partners' attachment style (i.e., their internal representation of the self‐with‐other) can become more secure (Burgess Moser et al. 2015). Notably, not only what we feel but also what we believe to be true, including the reality or nature of ourselves, can be shaped and shared with those closest to us.

5.3. Shared Reality

Shared reality is experienced when people's inner states are perceived as aligned with those of others (Echterhoff et al. 2009). It is the experience of sharing feelings, judgments, and evaluations regarding some aspects of the world (Andersen and Przybylinski 2018). The construction of a shared reality satisfies two primary psychological needs: the epistemic need to understand what is real and true, and the relational need to be connected and to belong. Stimuli, particularly social ones, are often ambiguous and thus open to interpretation. Consequently, people strive to achieve epistemic certainty by constructing a shared reality with others (Echterhoff and Higgins 2017; Rossignac‐Milon and Higgins 2018). Social constructions often determine what is seen as “real” in the world (Asch 1951), memory (Echterhoff et al. 2009), and the understanding of oneself (Hardin and Higgins 1996).

Infants form a shared reality with their primary caregivers via prelinguistic communication (e.g., infant pointing; Tomasello et al. 2007). This, in turn, helps establish what is real or important and satisfies the need for connectedness (Higgins 2016). For example, when an adult names an object that the infant has just pointed to, it communicates that this object is worthy of notice (Higgins 2016). Caregivers' micro‐facial behaviors (e.g., a smile or a frown) can help the infant determine the “reality” of stimuli in the environment in terms of safety or danger (Sorce et al. 1985). Shared reality continues to be pursued throughout life, mainly via adults' close relationships (Echterhoff et al. 2009).

Rossignac‐Milon and Higgins (2018) presented a developmental model describing how intimate partners construct a sense of shared reality over time. According to this model, shared reality is constructed in four accumulating phases: shared feelings, shared practices, shared coordination, and shared identity. Early in the relationship formation stage, partners first construct a sense of shared feelings, consisting of perceived commonalities in evaluative experiences, as revealed both explicitly (e.g., “It's amazing we both love Tarantino movies”) and implicitly through more subtle evaluative behaviors (e.g., shared laughing). In the second phase, termed shared practices, unique dyad‐specific activities emerge (e.g., “Friday night is movie night”). When repeated, these rituals form dyadic‐specific cultures and a unique relationship language (e.g., invented idioms, inside jokes). In the third stage, partners construct a shared coordination of their memories and future goals. This process, often shaped by anticipating one another's perspectives, influences how they recall experiences (French et al. 2008; Harris et al. 2014; Hirst and Echterhoff 2012) and make plans together to meet one another's needs and aspirations (Fitzsimons et al. 2015). Intimate partners' frequent sharing and joint reconstruction of the past shape their episodic memory (Hülür et al. 2022). Additionally, relationship‐related memories help foster intimacy and relationship satisfaction, suggesting that shared episodic memories support the development of a more cohesive relational narrative and enhance couple functioning (Aydin and Buyukcan‐Tetik 2021).

Shared memories, in which the basic psychological needs of autonomy, competence, and relatedness are met, are particularly important, as they predict future relationship health and durability (Philippe et al. 2013; Lejeune et al. 2021). Remarkably, one partner's recollection of a shared experience has been found to predict the other's level of commitment, even beyond the impact of their own memory of the same event (Guilbault and Philippe 2017). This reciprocal influence is indicative of a collaborative process, akin to a single transactive memory system in which couples depend on one another to encode, store, and access significant personal experiences (Hirst and Echterhoff 2012). These systems evolve, particularly in long‐term relationships, whose shared reminiscing influences their remembering (Barnier et al. 2018), at times at the expense of accuracy (Hirst and Echterhoff 2012). This body of research suggests that episodic memory serves not only as an individual cognitive process but also as a relational mechanism that promotes attachment security (Cao et al. 2020) and collaborative future planning in intimate relationships (Cao et al. 2018).

In the final stage of shared reality development, partners form a shared identity, where partners' minds begin to merge (Rossignac‐Milon et al. 2021). This shared identity is a core theme in relationship science (Finkel et al. 2017) and is discussed in terms of relational schemas (Baldwin 1992), cognitive interdependence (Agnew et al. 1998), self‐other integration (Slotter and Gardner 2009), relational self‐construal (Cross et al. 2000), inclusion of other in the self (IOS; Aron et al. 1992), and the relational self (Andersen and Chen 2002). Despite variations in emphasis, these constructs all refer to how partners' cognitions, experiences, and self‐concepts become increasingly interwoven to the extent that they sometimes blur the distinctions between partners (Rossignac‐Milon and Higgins 2018; Slotter and Gardner 2009). For example, romantic partners, especially those who are satisfied, often use “we‐ness” versus “I‐ness” phrasing spontaneously (Agnew et al. 1998; Alea et al. 2015; Gildersleeve et al. 2017). Studies utilizing implicit test paradigms have shown that romantic partners often fail to distinguish between traits attributed to themselves and those attributed to their partner, suggesting overlapping implicit representations of the self and the partner (Mashek et al. 2003; Slotter and Gardner 2009).

Early in therapy, couples often maintain a poor shared reality, holding diverging and conflicting perceptions of their relationship and each other's behaviors, intentions, and traits. They may also struggle to coordinate future goals, and their recollections of past relational events are frequently misaligned. A key example involves attachment injuries within the relationship (e.g., instances of abandonment or betrayal; Johnson et al. 2001), which often disrupt the coherence of the relational narrative, the couple's shared identity, and their sense of “we‐ness,” resulting in ongoing relational distress. Because attachment injuries involve acute breaches of trust that reorganize partners' expectations of safety and responsiveness, constructing a coherent shared account of the injury itself becomes central to resolution.

The forgiveness process unfolds through episodes of vulnerability and responsiveness, in which the injured partner expresses pain and unmet needs, while the offending partner acknowledges the impact of their actions and responds with emotional engagement and accountability (Makinen and Johnson 2006). Through these interactions, partners begin to co‐construct a new understanding of the injury and its meaning within the relationship, creating conditions that support forgiveness and reconciliation (Zuccarini et al. 2013).

For example, in one (hypothetical) session, Susan spoke about the loneliness she experienced after losing her mother while feeling that George was emotionally absent and preoccupied with his new career. As Susan shared this painful relational memory and associated feelings, George remained attuned and empathized with her sadness. He also expressed remorse, acknowledging how deeply he regretted not being there for her during that difficult time. Such vulnerability–responsiveness events often recur across sessions and may gradually reshape partners' shared reality (Rossignac‐Milon and Higgins 2018).

When partners engage emotionally around previously painful relational memories, new shared emotional experiences can emerge (e.g., empathic sadness concerning Susan's experience, “shared feelings”). These in‐session events can foster reciprocal disclosures (Dindia 2002) and support more adaptive communication patterns, or “shared practices.” Revisiting emotionally salient autobiographical memories within therapy (e.g., grieving alone) creates opportunities for dyadic affect regulation (EFT‐C; Greenberg and Goldman 2008), as partners emotionally re‐engage with the event and reshape its meaning within the relationship. Over time, such experiences may transform how partners organize and interpret relational memories associated with the attachment injury (e.g., changing Susan's recollection of George from one of indifference to one of remorse, thereby aligning their past relational memories and resulting in greater “shared coordination”).

Comparable to the ways in which attachment‐based working models shape and bias relational expectations (Simpson et al. 2010), altering the meaning of emotionally significant relationship events is expected to shift partners' relationship‐specific working models (EFCT; Johnson 2020). As a result, new semantic representations of the relationship may emerge with respect to the self (e.g., “I am worthy of love”), the other (e.g., “despite his actions, he genuinely cares”), and the relationship (e.g., “there is warmth in our relationship”). These changes are likely to foster a stronger shared identity, in which partners' self‐concepts increasingly overlap both explicitly (Gildersleeve et al. 2017) and implicitly (Mashek et al. 2003; Slotter and Gardner 2009).

Beyond transforming each partner's individual integrated memory structure, the concept of shared reality suggests that partners have a kind of relational integrated memory structure. As couples increasingly share feelings, habits, language, memories, goals, and identity (Rossignac‐Milon and Higgins 2018), the various components of their memory structure (episodic memories, semantic memories, and emotional responses) begin to overlap and be shaped by each other (Echterhoff and Higgins 2017; Guilbault and Philippe 2017). This idea is congruent with relationality thinking, which posits that “the best way to understand persons is not in isolation, but in the context of their relations with others, past and present, internal and external, actual and fantasized” (Mitchell 2000, 107). This definition does not exclude individuals but instead stresses the governing context of their relationship. As Buber (1970, 39) famously argued, in a fundamental sense, “Spirit is not in the I but between I and thou”. This perspective advocates viewing the self, particularly within couple therapy, as a consistently relational and context‐based phenomenon (Galovan and Schramm 2018). Through this co‐created shared reality space, each partner understands the other and the world, not only from the vantage point of a single mind but also from the joint merging of subjectivities (Rossignac‐Milon et al. 2021).

We suggest that for partners' minds to become more aligned, paradoxically, they must first recognize and appreciate the other's subjective differences. Partners' subjectivities, including gaps in their experiences, are fleshed out and revealed in vulnerability‐responsiveness events. In Buber's terminology, this engagement in an open dialogue often facilitates a movement from the instrumental “I‐it relationship” to the unmediated “I‐thou relationship” (Buber 1970). The prerequisite for moving from the I‐it position to the I‐thou position is an authentic, present, compassionate, and responsive heart (Galovan and Schramm 2018). Couples shift from reducing each other to their surface traits and functionalities to recognizing the subjectivity and infinite complexity of the other (Levinas 1961/1969; Sayre and Kunz 2005). Since the self is continually defined by its relations with the other, this relational change is concurrent with a change in oneself (Galovan and Schramm 2018). This, in turn, allows the partners to construct a sense of shared reality in which each dyad member's subjective experience is both recognized and expanded by integrating the other's perspective.

To summarize, we propose that a relational memory reconsolidation process underlies vulnerability‐responsiveness events in both models of EFT for couples. This process assumes that the self is not a static, monolithic entity but rather comprises various dynamic relational self‐states. The emotional activation of relational self‐states allows individuals to tap into and modify the meanings associated with specific episodic memories (e.g., emotional or attachment injuries) and semantic structures (i.e., who I am as a partner). The aim of therapy then becomes utilizing the reconsolidation window to align partners' diverging experiences by reconstructing their shared reality. In the next section, we discuss the potential clinical and research implications of this framework.

6. Clinical Implications

The Relational Memory Reconsolidation framework has several clinical implications. The first is the hopeful message that relational memories are not static but rather dynamic traces that are continuously reshaped by the relational context in which they are recalled (Nadel and Hardt 2011). Whether premeditated or spontaneous, emotional activation is critical in EFT models of couple therapy since it allows for the reconsolidation of negative relational memories. Process research based on EFT‐C (Greenberg et al. 2010) and EFCT (Zuccarini et al. 2013) indicates that the deepening of vulnerable emotional experience, particularly when accompanied by new affiliative responses from the partner, is critical for therapeutic change (for a review, see Greenman and Johnson 2013). This process provides an opportunity to transform and update deep‐seated memories (e.g., “going alone to a parent's funeral”) along with their associated negative semantic structures (e.g., “I am alone in this relationship”). Even if the couple is not struggling with a specific emotional or attachment injury (e.g., betrayal), distressed partners are often caught in negative emotional interaction cycles, where negative relational episodic memories are activated (e.g., “being put down during a heated argument”). Therefore, therapists are encouraged to intentionally create opportunities for emotional activation between partners that elicit relevant emotional memories, and to restructure interactions primarily through vulnerability‐responsiveness events. These key moments of heightened emotionality open a neurobiological window of opportunity to reconsolidate differences in subjectivities as they are expressed and listened to attentively.

Beyond general emotional activation, however, the reconsolidation framework specifies additional conditions for durable change. Change is hypothesized to occur when emotional engagement destabilizes the original memory traces sufficiently for new disconfirming experiences to integrate and update the underlying memory structure (Moscovitch et al. 2023). Accordingly, such updating is most likely when corrective experiences are introduced shortly after the original relational expectation has been reactivated, engendering prediction error: a mismatch between what is expected and what occurs (Exton‐McGuinness et al. 2015; Moscovitch et al. 2023). Therapists must facilitate full engagement with core relational expectations, whether attachment‐based (e.g., anticipated abandonment; Johnson 2020) or identity‐based (e.g., anticipated invalidation or shame; Greenberg and Goldman 2008), and then strategically introduce corrective interpersonal experiences that directly contradict these expectations. This requires careful tracking of activation markers, precise timing, and structured integration, ensuring both partners are prepared to (1) emotionally engage with their underlying primary emotions and (2) remain receptive to supporting their partner during critical moments of vulnerability.

To illustrate, consider a withdrawing partner who, through therapist‐guided exploration of a relevant episodic memory (e.g., reaching out during a moment of distress and being dismissed) and its associated emotions and bodily sensations, accesses an underlying fear of abandonment and the core relational expectation “if I show my needs, I will be rejected.” If the therapist identifies markers of destabilization (e.g., a shift from defensive withdrawal to affective arousal of vulnerability) and guides the other partner to respond with genuine reassurance rather than criticism, a prediction error is generated: the anticipated rejection does not occur. This mismatch between the expected and actual relational response is what the reconsolidation framework posits as the mechanism enabling lasting memory updating.

However, such moments are inherently fragile (Kula et al. 2026). If the activated vulnerability is met with defensiveness, withdrawal, or invalidation, the original relational expectation may instead be reinforced, strengthening the very pattern therapy seeks to transform. In these moments of within‐system rupture, the therapist's role becomes critical (Sherlow‐Levin et al. 2024). Rather than relying solely on the partner's responsiveness, the therapist must temporarily assume the function of a corrective attachment figure, validating the vulnerable experience while simultaneously tracking and exploring the processes that block the other partner's responsiveness. By doing so, the therapist helps stabilize the emotional engagement, maintain the reconsolidation window, and scaffold the emergence of a new, more responsive interaction pattern within the dyad.

Such moments, handled successfully, may begin to integrate partners' diverging semantic structures regarding themselves, the other, and the relationship. Recognizing, expanding, and differentiating subjectivities by accessing primary emotions are essential to successful personal and interpersonal integration. During these processes, therapists should encourage the listening partner to acknowledge and respect the other's perspective within the relational dynamic. As therapists process and reflect with the partners on successful vulnerability‐responsiveness events, they should explore shifts in the semantic knowledge regarding the self, the other, and the relationship. This may enable the reframing of previously rigid semantic structures in light of new perspectives and emotional experiences. In this way, the couple can solidify a sense of togetherness through attunement and support, thus eventually leading to a more harmonious and integrated shared reality (Wiesel et al. 2021).

Along with vulnerability‐responsiveness events, externalizing interactional cycles is another focal change process in both models of EFT for couples. By externalizing the problem to the “cycle,” instead of assigning individual blame (Johnson 2020), the partners can move from their narrow personal realities into a wider shared reality that they can both sense and agree on: the problem is not you or me; it is the dance we are sharing. Therapists should aim to identify and elaborate on couples' unique interactional cycles during sessions, thereby helping them move from a rigid, blame‐oriented perspective to a more flexible, shared narrative of their relational concerns. Indeed, restoring the shared reality between partners can be seen as another vital goal in therapy. Progress can be monitored by observing cues of shared reality, such as narrative alignment (Linhof and Allan 2019), spontaneous use of “we‐ness” phrasing (Alea et al. 2015), and greater coordination of feelings, practices, goals, and identity (Rossignac‐Milon and Higgins 2018).

7. Research Implications

The somewhat abstract nature of systemic and interpersonal phenomena, such as the relational integrated memory structure, poses a significant challenge for research. This highlights the need to utilize methods that can capture dyadic phenomena, considering new directions for measuring how couples reciprocally influence one another and participate in the same processes. These include subjective measures that quantify partners' overlapping self‐concepts (e.g., IOS; Aron et al. 1992) and shared reality experience (SR‐G; Rossignac‐Milon et al. 2021). New techniques are increasingly being introduced to gauge relational experience, including measuring couples' emotional synchrony via neurobiological (e.g., Goldstein et al. 2018), linguistic (e.g., Ireland and Pennebaker 2010), autonomic (e.g., Bar‐Kalifa et al. 2023), and behavioral (e.g., Ramseyer and Tschacher 2011; Rossignac‐Milon et al. 2021) markers.

Johnson et al. (2013) provide an illustrative example of a dyadic neuroimaging design examining interpersonal regulatory processes following EFCT. In their sample of 24 heterosexual couples, the female partners underwent fMRI scanning before and following therapy. Female partners showed significant reductions in activation in threat‐related limbic and prefrontal regions during handholding while anticipating threat. These findings are consistent with the Relational Memory Reconsolidation model, where observable changes in emotional responses occur during the activation of relational self‐states, such as the experience of safety when in contact with one's romantic partner. Recently, significant advances have been made in measuring synchrony during psychotherapeutic sessions (Zilcha‐Mano 2025) as well as in EFT sessions (Başer et al. 2025; Conroy et al. 2022). Given the evolutionary and neurological underpinnings of attachment‐based needs (Greenman and Johnson 2013; Greenman et al. 2017; Feldman et al. 2016), further implementation of these dyadic behavioral neuroscientific indices may facilitate empirical validation of the neurobiological basis of the proposed Relational Memory Reconsolidation model.

In the relational integrated memory framework, memory plays a crucial role in causing and maintaining relational distress. Although not all couples struggle with severe forms of emotional injuries, such challenges are not uncommon. For example, infidelity is a frequent issue in many relationships and a common presenting concern for couples seeking therapy (Gordon et al. 2023; Labrecque and Whisman 2017; Rokach and Chan 2023). Such emotional injuries are phenomenologically similar to traumatic events and can elicit similar posttraumatic symptoms (Anders et al. 2012; Glass and Wright 1997; Halchuk et al. 2010). In this trauma framework, emotional injuries can explain partners' avoidance of confrontation, intrusive relational imagery, hypervigilance, and changes in perceptions of themselves, others, and the world (De Stefano and Oala 2008; Johnson et al. 2001). Johnson et al. (2001) considered that attachment injury, consisting of the failure of a partner to provide caring and support during a time of distress, can overwhelm the partner in a way similar to a traumatic flashback. Therefore, just as traumatic memory content and structure tend to change with treatment (Booker et al. 2020), future studies should aim to uncover changes over time in how partners discuss memories of emotional injuries. Analyzing themes of forgiveness and reconciliation may prove to be a particularly productive direction, since these are often viewed as key therapeutic components in both resolving attachment injuries in EFCT (Johnson et al. 2001; Zuccarini et al. 2013) and emotional injuries in EFT‐C (Greenberg et al. 2010; Meneses and Greenberg 2011). Thematic and sentiment analyses based on advanced natural language processing methods (e.g., word2Vec‐based BERT models of semantic similarity) may show how couples' narratives evolve over time (Fenerci et al. 2025). This can shed light not only on how partners' terminology and speech patterns diverge or converge (e.g., Alea et al. 2015) but also on how partners' emotional responses, semantic structures, and episodic memories fluctuate and change.

Despite its parsimonious explanatory power across various psychotherapeutic approaches, Lane et al.'s (2015) memory reconsolidation model has yet to be systematically examined in individual therapy, let alone couple therapy. One key unexplored area concerns how the different associative components of memory interact with one another, as well as their order of change in different therapy modalities. For example, EFT models for couple therapy and cognitive behavioral couple therapy (CBCT; Baucom and Epstein 1990; Epstein and Zheng 2017) have demonstrated similar efficacy in reducing relational distress (Bodenmann et al. 2020). However, since EFT techniques focus on eliciting emotional responses and CBCT pays greater attention to transforming semantic structures (Bodenmann et al. 2020), differences in the sequence of activation patterns in each modality are likely to be present. For example, CBCT may follow a top‐down pattern, first targeting and modifying maladaptive beliefs in the semantic memory structure before working on partners' procedural emotional responses. EFT, in contrast, may adopt a more bottom‐up approach, initiating change by engaging emotional responses directly and activating the emotional memory system before accessing cognitive or semantic structures. These different sequences may create distinct pathways for change.

Notably, Integrative Behavioral Couple Therapy (IBCT; Christensen et al. 2020), another evidence‐based model for couple therapy, integrates both cognitively oriented and emotionally focused techniques. Although the primary aim of this article was to discuss how vulnerability‐responsiveness events, the key process in EFT models for couple therapy, drive change in partners' functioning, the relational memory reconsolidation framework may have broader relevance. Specifically, it may offer a theoretical bridge for exploring the commonalities and differences in mechanisms of change, and their sequencing across the two models of EFT (Greenberg and Goldman 2008; Johnson 2020), as well as other empirically supported couple therapy models.

Finally, although the present framework is not theoretically expected to vary by couple configuration, most empirical work in this area has relied on heterosexual, cisgender samples. Consistent with calls in the literature for broader inclusion of sexual and gender minority (SGM) couples—who remain significantly underrepresented in couple therapy research (Randall et al. 2022; Spengler et al. 2020; Whitton et al. 2023)—future research should examine relational memory reconsolidation processes across diverse couple types.

8. Summary

The Relational Memory Reconsolidation framework suggests that models of EFT for couples (Greenberg and Goldman 2008; Johnson 2020) facilitate meaningful change by helping partners reconsolidate their relational integrated memory structures through corrective interpersonal interactions. During vulnerability‐responsiveness events, and specifically those concerning negative relational episodic memories, partners are guided to reveal their deeper feelings while receiving attuned support. Engaging with partners' emotional responses activates relevant relational self‐states regarding their beliefs and perceptions of the relationship. Through repeated sequences of vulnerability and responsiveness, partners begin to update negative relational memory traces, such as feelings of rejection or fear, with more positive ones, such as experiences of acceptance and emotional safety. This transformation gradually alters their integrated memory structure and can foster lasting change.

Throughout this process, they cease to “totalize” one another; that is, they cease to view one another as objects (Galovan and Schramm 2018). Instead, they recognize the other's subjectivity and complexity (Levinas 1961/1969). Ironically, this recognition of each other's different subjective experiences allows the partners to reconstruct a stronger, more aligned shared reality and identity. Expressing and attending to these emotional experiences reconsolidates a sense of mutual understanding and connection, thus leading to enduring changes in how the relationship is experienced, perceived, and remembered. Future research is needed to substantiate these processes and develop a more precise empirical model of couples' relational dynamics, experiences, and memories.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.