Perceiving to Provide: How Partner Attachment Perceptions Inform Reassurance Provision in Romantic Relationships

Abstract

Romantic partners can play a key role in buffering one another’s attachment insecurities, but how they perceive one another’s attachment and whether perceptions predict behavior remains unclear. This research examined how accurately individuals perceive their partner’s global and relationship-specific attachment anxiety and avoidance, what biases shape these perceptions, and whether perceptions predict reassurance provision, a recommended buffering strategy for anxiously attached partners. Using the Truth and Bias Model, two studies (Study 1: N = 108 couples; Study 2: N = 147 couples) showed that individuals perceive their partner’s attachment with moderate accuracy, though they also overestimate insecurity and are biased to assume both similarity and complementarity in attachment. Moreover, people who perceived their partners to be higher in attachment anxiety provided greater reassurance during personal stressor discussions and in daily life (10-day ecological momentary assessment), suggesting that attachment perceptions may play a key role in initiating and calibrating buffering strategies.

Keywords

adult attachment interpersonal perception reassurance provision truth and bias model of judgment

Introduction

Although attachment insecurity is associated with poorer relationship functioning for individuals and their partners (e.g., Candel & Turliuc, 2019), a burgeoning literature on insecurity buffering identifies specific strategies that mitigate these costs (Arriaga & Kumashiro, 2019; Overall et al., 2022). Specifically, romantic partners can buffer one another’s attachment insecurities when they enact behaviors matched to the form of insecurity present (e.g., offering extra reassurance to an anxiously attached partner). Despite evidence for the effectiveness of insecurity buffering behaviors, how partners come to use these strategies remains unclear. The present research addresses two questions: First, to what extent do individuals perceive their partner’s attachment orientation accurately, and what factors account for systematic biases in these perceptions? Second, do perceptions of a partner’s attachment insecurity uniquely predict whether individuals offer reassurance that the relationship is safe (e.g., expressions of commitment, love, and affection), a key insecurity-buffering strategy targeted to attachment anxiety?

Accuracy and Bias in Attachment Perceptions

Individual differences in beliefs, expectations, and behavioral tendencies in close relationships are systematically described by adult attachment orientations (Brennan et al., 1998). One dimension of attachment insecurity, attachment anxiety, is characterized by heightened concerns about one’s own lovability, intense desire for closeness, fear of abandonment, and chronic preoccupation with a partner’s availability (Fraley & Shaver, 2000; Mikulincer & Shaver, 2016). The other dimension of attachment insecurity, attachment avoidance, is characterized by discomfort with intimacy, deactivation of relatedness needs, excessive self-reliance, and low trust in others (Fraley & Shaver, 2000; Mikulincer & Shaver, 2016). Individuals vary along both dimensions of attachment insecurity, with secure attachment represented by absent or low levels of both forms of insecurity. Attachment orientations are conceptualized hierarchically, with global attachment reflecting a dispositional tendency across close relationships and relationship-specific (RS) attachment reflecting expectations and behaviors in one specific relationship (Fraley et al., 2011; Overall et al., 2003). Global attachment correlates moderately with RS attachment (e.g., romantic, parental; Fraley et al., 2011).

To understand how individuals perceive their partner’s attachment orientation, we draw on the Truth and Bias model (LaBuda & Gere, 2023; West & Kenny, 2011). The Truth and Bias model is a comprehensive framework and the current best practice for simultaneously evaluating both the accuracy of and systematic biases in interpersonal perceptions (Fletcher, 2015; Stern & West, 2018; West & Kenny, 2011). We use this framework to define accuracy and bias and to review relevant literature.

Romantic partners’ judgments about one another’s traits and behaviors are often quite accurate, with accuracy (also called “correlational accuracy”) defined as the alignment between a partner’s perception of a trait and the “truth” (typically measured by the target’s self-report of that same trait; Fletcher, 2015; West & Kenny, 2011). For instance, Fletcher and Kerr (2010) found evidence of moderate accuracy for personality traits (e.g., a partner’s agreeableness). This alignment is expected, as these judgments are based on a shared history (Gagné & Lydon, 2004), and people are typically motivated to monitor their partner’s internal and external states to better understand them, adjust their behavior accordingly, and foster desired outcomes such as relationship satisfaction (Fletcher, 2015).

It is also plausible that people would have accurate perceptions of their partner’s attachment orientation, given that attachment orientations are manifested through observable emotional and behavioral patterns, such as “clinginess” in those high in anxiety or emotional distancing in those high in avoidance. There is some evidence that strangers can detect attachment insecurity (perhaps only attachment anxiety) in one another after only a brief interaction (Banai et al., 1998; Tu et al., 2022). Moreover, using a correlational approach, Molero et al. (2016) provided initial evidence for accuracy in ongoing relationships by showing that individuals’ perceptions of their romantic partner’s attachment tendencies were correlated with their partner’s self-reports. The current research will extend this initial evidence by using the Truth and Bias model to assess accuracy while teasing apart systematic sources of bias.

Systematic biases—defined as consistent deviations from the actual correspondence between judgments and “truth” (Fletcher, 2015; West & Kenny, 2011)—are important to understand because perceptions may serve as the basis for enacting insecurity buffering strategies. One such bias is directional bias, in which individuals perceive their partners as higher or lower on a given trait than the partners report themselves to be. Individuals typically view their partners through “rose-colored glasses,” over-perceiving positive attributes (e.g., intelligence) and under-perceiving negative traits (e.g., disorganization; Fletcher & Kerr, 2010; Murray et al., 1996). However, research suggests that when evaluating partners’ beliefs, attitudes, or behaviors directed toward oneself or the relationship, biased perceptions often shift in a more negative direction (e.g., under-perceiving a partner’s relationship satisfaction; Fletcher & Kerr, 2010). Negatively biased perceptions may serve an adaptive function by motivating greater vigilance and relational effort, thereby reducing the risk of complacency. This reasoning aligns with error management theory, which posits that in domains of high evolutionary or relational cost, it is often less risky to overestimate potential threats than to underestimate them (Fletcher & Kerr, 2010; Haselton & Buss, 2000). Attachment insecurity is a likely candidate for over-perception, since detecting even subtle signs provides partners with opportunities to mitigate its impacts.

Another common source of systematic bias in partner perception is projection bias (also called assumed similarity), the tendency to assume that one’s partner shares similar traits to oneself (Fletcher & Kerr, 2010). Projection may arise from basic egocentric inference processes (Epley et al., 2004), wherein individuals use their own characteristics as an anchor when making judgments about others. In romantic relationships, this bias may stem from the belief in a shared reality or sense of oneness between partners (Mashek et al., 2003; Rossignac-Milon & Higgins, 2018). Prior research has found significant associations between perceivers’ own attachment security and their perceptions of partners’ attachment security (Cobb et al., 2001; Ruvolo & Fabin, 1999), suggesting the possibility that individuals may project their own attachment onto their partner.

In the context of attachment, a unique form of bias may also emerge: complementarity bias, whereby anxiously attached individuals perceive their partners as more avoidant, and avoidantly attached individuals perceive their partners as more anxious. This bias may arise because the behavioral manifestations of attachment anxiety and avoidance often occur in opposing directions, as they entail hyperactivating and deactivating strategies, respectively (Mikulincer & Shaver, 2016). For example, attachment anxiety prompts excessive disclosure, whereas attachment avoidance features restrained disclosure (Snyder et al., 2024; Sun & Jakubiak, 2024). Therefore, an anxiously attached person may be more likely to perceive their partner as avoidantly attached (non-disclosive) relative to their preferences and expectations, whereas an avoidantly attached person may be more likely to perceive their partner as anxiously attached (over-disclosive). Consequently, anxiously attached individuals may perceive their partner as particularly emotionally distant, whereas avoidantly attached individuals may see their partner as excessively needy. Because people often judge others relative to their own dispositions (Epley et al., 2004), rather than considering the full range of possible behaviors, this logic likely applies to a range of attachment-relevant behaviors. One study found that one’s own attachment insecurity significantly predicted greater perceived partner insecurity in the other dimension (Strauss et al., 2012) in a broader mediation model, but complementarity has yet to be tested as part of the Truth and Bias model. Doing so allows us to isolate complementarity from other biases such as the tendency to perceive similarity.

Although some previous studies have explored the accuracy of partner attachment perceptions, earlier studies often treated accuracy and bias in perceptions as opposing constructs, assuming that greater accuracy necessarily means less bias. In contrast, contemporary perspectives recognize that individuals can be both accurate and biased in their perceptions. By applying the Truth and Bias framework (West & Kenny, 2011) to independently estimate accuracy and multiple forms of systematic bias, the present research can clarify how truth and bias co-occur in partner attachment perceptions. Critically, perceptions of a partner’s attachment (whether accurate or biased) may guide the initiation of insecurity buffering strategies.

Implications of Perceptions for “Safe Strategy” Buffering Behaviors

Both attachment anxiety and avoidance have been consistently associated with poorer relationship outcomes, including elevated conflict, diminished intimacy, and lower satisfaction (e.g., Candel & Turliuc, 2019; Li & Chan, 2012; Mikulincer & Shaver, 2016). Fortunately, emerging evidence suggests that the negative relational consequences of attachment insecurity can be minimized when partners enact specific “buffering” strategies that help to regulate an insecure individual’s attachment-related concerns and distress (e.g., Arriaga et al., 2018; Overall et al., 2022; Simpson & Overall, 2014).

For individuals high in attachment anxiety, partners are theorized to buffer their insecurity by enacting “safe strategies,” behaviors that offer reassurance that the relationship is safe (e.g., Arriaga et al., 2018). Partners can provide this reassurance in many ways, such as by expressing affection (through words, physical affection, or generous acts) or by directly stating their intention to maintain the relationship (Arriaga & Kumashiro, 2019; Overall et al., 2022). Although specific manifestations of reassurance may vary by context, these behaviors each focus on saliently communicating love, care, and continued commitment, which counters anxiously attached individuals’ hyperactivated monitoring for relational threats. Other positive relational behaviors—such as providing practical support, validating a partner’s experience, encouraging goal pursuit, and showing understanding—may incidentally or indirectly communicate care and commitment, but reassurance provision is distinct in that it communicates care and continued commitment directly and saliently. Reassurance provision is therefore matched to the unique insecurities of anxiously attached individuals (doubts about lovability, worries about abandonment), which then likely leads to perceptions that the reassuring partner is responsive (i.e., the partner understands and values them; Arriaga & Kumashiro, 2019) in addition to reducing perceived threat.

In contrast, partners of avoidantly attached individuals are posited to buffer insecurity by respecting autonomy and toning down the emotional charge and intimacy of interactions, which prevents the deactivation characteristic of attachment avoidance (e.g., Arriaga et al., 2018). As the current investigation focuses only on the enactment of “safe strategies” (due to the data available in the current research studies), we refer interested readers to the following reviews for evidence on the effectiveness of “soft strategies” (Arriaga et al., 2018; Overall et al., 2022).

Empirical evidence supports the effectiveness of targeted reassurance provision in mitigating the damaging effects associated with attachment anxiety. For example, individuals with higher attachment anxiety felt more secure in their relationships on days when their partners exaggerated affection for them (Lemay & Dudley, 2011). Affectionate touch has also been found to buffer the jealousy associated with attachment anxiety when individuals observe their partner viewing photos of attractive alternatives (Kim et al., 2018). What remains unclear, however, is whether such partner buffering behaviors are enacted strategically, based on individuals’ perceptions of their partner’s attachment orientation, or whether they arise incidentally from the support-provider’s own goals or preferences. Although partner buffering behaviors may sometimes occur without conscious reflection or intention (e.g., automatically offering a hug to a jealous partner), effective partner buffering in the long term is likely facilitated by partners being able to recognize and respond to one another’s specific attachment-related needs.

If partners of anxiously attached individuals do strategically use “safe” strategies to buffer attachment anxiety, then their perceptions of their partner’s attachment anxiety should be a key predictor of reassurance provision. Greater reassurance can help reduce the fear of abandonment experienced by anxiously attached individuals, both chronically and in situations that trigger attachment-related threat or distress (Arriaga et al., 2018; Overall et al., 2022; Simpson & Overall, 2014). Because the attachment alarm of anxiously attached individuals is chronically hyperactivated, they tend to overinterpret the significance of daily relationship events (Campbell et al., 2005). Consequently, partners who perceive them as high in attachment anxiety may provide greater reassurance during everyday interactions to soothe their heightened insecurity. Moreover, anxiously attached people also struggle more severely in stressful contexts, in which their need to depend on partners is amplified. Accordingly, partners of anxiously attached people may provide greater reassurance under such circumstances as well.

Current Research

The current research involved two studies—one with a sample of undergraduate couples and one with a sample of community couples—aimed at clarifying how partners may come to initiate attachment-matched buffering strategies. We had two goals in each study. First, we sought to systematically assess and describe the accuracy and forms of bias in perceptions of a partner’s attachment (Q1). Second, we aimed to investigate how perceptions of a partner’s attachment relate to the provision of reassurance, a “safe” partner buffering strategy targeted toward attachment anxiety (Q2).

We assessed the accuracy and bias in perceptions of RS attachment anxiety and avoidance (Studies 1 and 2) and global attachment anxiety and avoidance (Study 2). We predicted that romantic partners perceive their partner’s RS and global attachment insecurity with some degree of accuracy. At the same time, we expected perceptions of both RS and global attachment to show systematic bias, including overestimation of insecurity, projection of one’s own insecurity onto partner perceptions, and complementary perceptions (i.e., one’s own insecurity predicts viewing one’s partner as higher on the opposite dimension of insecurity). We also explored whether perceptions of RS attachment and global attachment differed in the extent of accuracy and bias. If partners over-perceive attachment insecurity as a relationship-protective strategy (e.g., Haselton & Buss, 2000), this tendency may be particularly apparent for RS perceptions because overlooking attachment insecurity in the immediate relationship likely carries greater costs than overlooking dispositional insecurity. Moreover, perceptions of RS attachment may be more susceptible to projection and complementarity biases than perceptions of global attachment, as the shared romantic relationship context could provide fertile ground for these biases to develop.¹

The current research also examined whether perceived partner attachment predicted reassurance provision across three contexts: couples’ personal goal discussions (Study 1), personal stressor discussions (Study 2), and daily life (Study 2). Couple discussions that center on personal goals or stressors offer prime opportunities to observe whether attachment-matched buffering occurs; as such, discussions often activate attachment insecurities that can be buffered in the moment (Overall et al., 2022). Moreover, daily life involves numerous opportunities for partners to demonstrate commitment and care, not only during moments of distress but also as ongoing reassurance to preemptively soften moments when insecurities will be activated. Assessing these processes in daily life allows for the capture of spontaneous relationship dynamics in naturalistic settings (Bolger et al., 2003), complementing the observations gathered in laboratory-based discussions. We predicted that partners would provide more reassurance during laboratory discussions and in daily life if they perceived their partner as high in RS and/or global attachment anxiety.

Study 1

In Study 1, we examined how accurately individuals perceive their partner’s RS attachment orientation, the factors that bias these perceptions, and whether such perceptions were linked to reassurance provision during couples’ personal goal discussions.

Method

Neither this study nor Study 2 was pre-registered, but our hypotheses were derived from theory, and all analyses are reported herein or in the Online Supplemental Materials.² The study designs and sample sizes were determined by other aims of the broader projects from which these data were drawn. We report all manipulations, measures (see Online Supplemental Materials for relevant measures and OSF https://osf.io/qps3a/overview for full measures), and exclusions.

Participants

The sample was collected between 2022 and 2024 from an undergraduate participant pool at a private university in the United States. Eligible participants needed to be in a committed romantic relationship for at least 6 months and have a partner willing to participate with them. Student participants were compensated with partial course credit, and their partners were either awarded the same or entered in a raffle to win one of four $50 gift cards. Couples could complete the study either in the laboratory (N = 17) or remotely via Zoom (N = 138). After excluding couples who did not consent to the use of their data (N = 47), the final analytic sample consisted of 108 couples. On average, couples had been in a romantic relationship for 1.5 years (SD = 1.3), and the vast majority were seriously dating (91.2%). Participants had a mean age of 19.2 years (SD = 1.8), approximately half identified as cisgender women (48.1%), and most identified as White (65.3%). See Supplemental Table 1 for additional demographic details. A sensitivity power analysis indicated that the effective sample size of 166 (accounting for dyadic nonindependence³) was sensitive to detect a correlation of r = .22 with 80% power (α = .05, two-tailed).

Procedure and Measures

All procedures and materials were approved by the university’s institutional review board. Descriptive statistics, reliability estimates, and zero-order correlations for primary variables are presented in Table 1.

Table 1.

Descriptive Statistics, Reliability Estimates, and Zero-Order Correlations (Study 1).

Study variable	α	M	SD	1	2	3	4	5	6
1. Own RS attachment anxiety	.85	2.35	1.38	−
2. Own RS attachment avoidance	.77	1.77	0.81	.47***	−
3. Partners’ RS attachment anxiety	−	−	−	.22***	.22***	−
4. Partners’ RS attachment avoidance	−	−	−	.22***	.14*	.47***	−
5. Perceived partner RS attachment anxiety	.87	2.65	1.54	.39***	.33***	.36***	.18**	−
6. Perceived partner RS attachment avoidance	.82	2.05	0.96	.50***	.43***	.25***	.38***	.32***	−
7. Reassurance provision	.81	3.95	1.07	−.10	−.21**	−.06	−.01	−.14*	−.20**

Note. The reliabilities, means, and standard deviations of partners’ reports were identical to those of self-reports and therefore not reported. RS = relationship-specific.

p < .05. **p < .01. ***p < .001.

Participants first completed a background survey via Qualtrics. This survey assessed demographics and self-report measures of participants’ own RS attachment orientations and their perceptions of their partner’s RS attachment orientation.

Attachment Orientation

We assessed self-reported and perceived partner RS attachment using the Experiences in Close Relationships—Relationship Structures scale (ECR-RS; Fraley et al., 2011) and an adapted version of the ECR-RS. In each scale, four items assessed RS attachment anxiety, and six items measured RS attachment avoidance.⁴ To measure perceived partner attachment, items were adapted to consider the partner’s experiences. For instance, the attachment anxiety item, “I am afraid that [partner] may abandon me,” was adapted to “[partner] is afraid that I may abandon them,” and the attachment avoidance item, “I prefer not to show [partner] how I feel deep down,” was adapted to “[partner] prefers not to show me how they feel deep down.” Participants were instructed to rate each item on a scale from 1 (Strongly disagree) to 7 (Strongly agree). The items for RS attachment anxiety and avoidance were averaged separately for each scale.

Reassurance Provision During Goal Discussion

At the end of the background survey, participants were asked to list up to five personal goals they intended to make significant progress on over the next 6 months. These goals were to be individual in nature—pertaining to the participant personally and not requiring the active involvement of their partner. As the topic for discussion, participants selected their most important goal to pursue over the next 6 months, from those rated as taking at least minimal time away from their relationship. On average, participants were moderately worried about accomplishing their goals (M = 4.0), and they planned to expend close to the highest possible effort (M = 6.2) toward achieving their goals over the next 6 months, both measured on a 7-point scale. Each partner took turns discussing their selected personal goal for 8 min while the interaction was video recorded (order randomly determined). Following each discussion, partners completed a set of post-discussion questionnaires reflecting their respective roles as either the goal achiever (i.e., discussed their own goal) or the goal supporter (i.e., discussed their partner’s goal).

We focused on supporters’ reassurance provision while discussing their partner’s personal goal, indexed by an average of three items measured in the post-discussion questionnaires. Specifically, participants rated the extent to which they told their partner that they care about them, reassured their partner that they want to maintain the relationship, and showed their partner that they value the relationship during the goal discussion on a 5-point scale (1 = Not at all to 5 = Very much).

Data Analytic Strategy

We used Truth and Bias models and linear mixed-effects models to test our hypotheses. To account for the dyadic nature of the data (i.e., individuals nested within couples), we modeled nonindependence using a compound symmetry covariance structure (Stern & West, 2018). Dyads were treated as indistinguishable in all models because of the presence of same-gender couples.

Truth and Bias models examined the degree to which individuals are accurate and biased in their judgments of their partner’s attachment orientation (Q1). Separate models were estimated to predict perceptions of attachment anxiety and attachment avoidance. Following West and Kenny (2011), perceivers’ ratings of their partner’s attachment dimension (e.g., attachment anxiety) were regressed on the partner’s self-report of that same attachment dimension (i.e., the “truth”) and the perceiver’s own self-rating on that dimension (i.e., a potential source of projection bias). We also included the perceiver’s self-reported level of the other attachment dimension to examine complementarity bias (e.g., did individuals high in avoidance tend to perceive their partners as high in anxiety?). These three predictors were each grand-mean centered on the partner’s self-reported attachment (i.e., the “truth”) across dyads, such that model intercepts represented directional bias—the degree to which individuals systematically overestimate or underestimate their partner’s attachment insecurity dimension. The slope of the partner’s self-reported attachment reflected correlational accuracy; the slope of the perceiver’s own attachment reflected assumed similarity (i.e., projection; Fletcher & Kerr, 2010); and the slope of the perceiver’s other attachment dimension reflected complementarity bias. We reported both unstandardized and standardized coefficients as indicators of effect sizes for accuracy, projection, and complementarity bias, and we provided Cohen’s d (mean difference divided by the standard deviation of the difference scores) as an indicator of effect size for directional bias (LaBuda & Gere, 2023).

To test whether perceived partner attachment predicted supporters’ reassurance provision during goal discussions (Q2), we used linear models accounting for the dyadic data structure. Specifically, supporters’ reassurance provision was regressed onto their perceptions of their partner’s attachment anxiety, controlling for perceptions of partner’s attachment avoidance and both the supporters’ and partners’ self-reported attachment anxiety and avoidance. All predictors were grand-mean centered. These variables were included so that we could assess whether perceptions of a partner’s attachment anxiety predict reassurance provision above and beyond other attachment factors that likely predict reassurance provision (see Tu et al., 2022 for a similar approach). We used perceptions themselves as the focal predictors rather than accuracy or bias components because from a buffering perspective, it should be the absolute level of perceived insecurity that drives buffering, not the accuracy or inaccuracy of attachment perceptions. We estimated these models using the gls function in R 4.3.1 (R Core Team, 2023; see Online Supplemental Materials for example syntax and list of R packages used).

Results

Q1: Accuracy and Bias in Perceived Partner Attachment

The full results for accuracy and bias in perceptions of partners’ RS attachment are presented in Table 2. Perceptions of partners’ RS attachment anxiety were positively associated with partners’ self-reported RS attachment anxiety, reflecting significant correlational accuracy. Perceptions of partners’ RS attachment anxiety were also significantly biased. People, on average, overestimated their partner’s RS attachment anxiety (as indicated by the significant and positive intercept reflecting positive directional bias). In addition, people higher in RS attachment anxiety or avoidance also tended to perceive their partner as more anxiously attached, suggesting both projection bias and complementarity bias.

Table 2.

Accuracy and Bias in Perceptions of Partners’ RS Attachment (Study 1).

Predictor	b [95% CI]	SE	p	β	d
Attachment anxiety
Intercept (directional bias)	0.44 [0.23, 0.65]	0.11	<.001	—	0.18
Partner’s attachment anxiety (correlational accuracy)	0.30 [0.16, 0.44]	0.07	<.001	.27
Perceivers’ own attachment anxiety (projection bias)	0.30 [0.14, 0.45]	0.08	<.001	.26
Perceiver’s own attachment avoidance (complementarity bias)	0.26 [0.01, 0.50]	0.13	.044	.13
Attachment avoidance
Intercept (directional bias)	0.15 [0.04, 0.27]	0.06	.011	—	0.29
Partner’s attachment avoidance (correlational accuracy)	0.32 [0.19, 0.45]	0.07	<.001	.27
Perceiver’s own attachment avoidance (projection bias)	0.28 [0.14, 0.42]	0.07	<.001	.24
Perceiver’s own attachment anxiety (complementarity bias)	0.23 [0.14, 0.31]	0.04	<.001	.33

Note. RS = relationship-specific.

As shown in Table 2, perceptions of partners’ RS attachment avoidance followed similar patterns. Revealing correlational accuracy, perceptions of partners’ RS attachment avoidance were positively associated with partners’ self-reported RS attachment avoidance. Indicating directional bias, people, on average, overestimated their partner’s RS attachment avoidance. Signifying projection and complementarity bias, perceptions of partners’ RS attachment avoidance were positively associated with people’s own RS attachment avoidance as well as anxiety.

Q2: Reassurance Provision

As shown in Table 3, perceptions of partners’ RS attachment anxiety were not significantly associated with reassurance provision. Of note, these findings were observed after controlling for perceptions of partners’ RS attachment avoidance and both partners’ self-reported RS attachment anxiety and avoidance, which were also not significantly associated with reassurance provision.⁵

Table 3.

Perceptions of Partners’ RS Attachment Predicting Reassurance Provision (Study 1).

Predictor	b [95% CI]	SE	p	β
Perceptions of partners’ attachment anxiety	−0.06 [−0.16, 0.04]	0.05	.207	−.09
Perceptions of partners’ attachment avoidance	−0.15 [−0.33, 0.03]	0.09	.094	−.14
Partners’ attachment anxiety	0.00 [−0.12, 0.12]	0.06	.996	.00
Partners’ attachment avoidance	0.08 [−0.13, 0.29]	0.11	.442	.06
Perceivers’ own attachment anxiety	0.05 [−0.08, 0.17]	0.06	.471	.06
Perceivers’ own attachment avoidance	−0.20 [−0.41, 0.00]	0.11	.056	−.15

Note. RS = relationship-specific.

Brief Discussion

Study 1 provided initial evidence that romantic partners do, to some extent, accurately perceive their partner’s RS attachment anxiety and avoidance, with effect sizes comparable to those found for negative relationship-relevant traits in a recent meta-analysis (LaBuda & Gere, 2023). Attachment perceptions also demonstrated evidence of directional bias (perceiving partners as more insecure than partners rated themselves), projection bias (assumed similarity), and complementarity bias (perceiving partners as higher on the opposite attachment dimension).

Regarding our second key question, we did not observe the expected association between perceived partner RS attachment anxiety and reassurance provision. This could indicate that the goal discussions may not have produced enough visible momentary distress to lead partners to enact safe strategies. This could be due to the fact that 88% of the discussions took place over Zoom, which might have hindered emotional expression and observation (Rollings et al., 2024). Additionally, although goal achievers were moderately worried about accomplishing their personal goals, goal supporters might not have perceived the context as particularly stressful for the achiever and, therefore might not have felt that reassurance was necessary. Furthermore, given that this sample consisted of young adults in relatively short relationships, these couples may not yet have developed the caregiving repertoires needed to enact safe strategies in response to perceived attachment anxiety.

Study 2

Rather than serving as a simple replication, Study 2 complemented and extended Study 1 in several ways. First, Study 2 assessed accuracy and bias in both RS and global attachment perceptions. Second, it also introduced two additional contexts to examine reassurance provision as a potential implication of these perceptions: personal stressor discussions and daily interaction. Finally, Study 2 used a more diverse community sample. Based on Study 1, we predicted that partner attachment perceptions would reflect both accuracy and systematic biases including overestimation of insecurity, projection, and complementarity. Although Study 1 did not find evidence that attachment perceptions predicted safe strategies during goal discussions, we reasoned that reassurance provision may be more evident in contexts that more directly activate attachment concerns—namely, when partners discuss personal stressors or navigate the varied insecurity-triggering situations of daily life. Thus, we predicted that greater perceived partner attachment anxiety would predict greater reassurance provision in these contexts. Although RS attachment typically predicts RS behaviors more strongly (Klohnen et al., 2005), we also tested whether perceiving a partner as anxious about close relationships in general (i.e., global attachment anxiety) would similarly predict buffering behaviors within the romantic relationship.

Method

Participants

Romantic couples were recruited from local communities in the northeast United States between 2024 and 2025. Couples were eligible if they had been in a romantic relationship for at least 1 year, saw each other in person at least 4 days per week, had access to a personal electronic device for completing surveys, and were able to attend in-person sessions on campus. Our final sample included 147 couples,⁶ most of whom were married (67.3%). Participants had been in a committed romantic relationship for an average of 12.3 years (SD = 9.8). Participants’ ages ranged from 20 to 73 (M = 38.8, SD = 11.6). Nearly half of the participants identified as women (49.7%), and the majority of participants identified as White (88.8%). See Supplemental Table 1 for detailed participant characteristics. A sensitivity power analysis (ICC_{affectionate touch} = 0.57, N_effective = 187) indicated that this effective sample size was sensitive to detect a correlation of r = .20 with 80% power (α = .05, two-tailed).

Procedure and Measures

Table 4.

Descriptive Statistics, Reliability Estimates, and Zero-Order Correlations (Study 2).

Study variable	α	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Own RS att. anxiety	.87	2.34	1.39	—
2. Own RS att. Avoidance	.85	2.10	0.98	.51***	—
3. Own global att. anxiety	.95	3.40	1.32	.62***	.30***	—
4. Own global att. avoidance	.93	3.66	1.11	.26***	.53***	.39***	—
5. Partners’ RS att. anxiety	—	—	—	.41***	.37***	.24***	.25***	—
6. Partners’ RS att. avoidance	—	—	—	.37***	.20***	.22***	.08	.51***	—
7. Partners’ global att. anxiety	—	—	—	.24***	.22***	.28***	.17**	.62***	.30***	—
8. Partners’ global att. avoidance	—	—	—	.25***	.08	.17**	−.03	.26***	.53***	.39***	—
9. Perceived partner RS att. anxiety	.87	2.85	1.53	.44***	.43***	.28***	.34***	.47***	.22***	.38***	.11	—
10. Perceived partner RS att. avoidance	.83	2.40	0.97	.44***	.37***	.31***	.18**	.29***	.48***	.09	.30***	.32***	—
11. Perceived partner global att. anxiety	.90	3.86	1.74	.25***	.27***	.33***	.30***	.32***	.15*	.53***	.11	.52***	.09	—
12. Perceived partner global att. avoidance	.86	3.88	1.36	.26***	.07	.28***	.03	.08	.35***	.09	.46***	.06	.45***	.04	—
13. Affection provision	.80	4.18	2.17	−.09***	−.12***	.01	−.03**	−.06***	−.13***	.08***	−.03***	.01	−.15***	.13***	−.02*	—
14. Reassurance provision	.86	5.23	1.43	−.17**	−.20***	.04	−.04	−.06	−.10	.11	−.02	.01	−.16**	.16**	−.02	.28***	—

Note. The reliabilities, means, and standard deviations of partners’ reports were identical to those of self-reports and therefore not reported. Att. = attachment; RS = relationship-specific.

p < .05. **p < .01. ***p < .001.

Couples first completed a background survey—either in the research lab or remotely via Zoom—that included demographic questionnaires, self-reports of their own and perceptions of their partner’s attachment, and other surveys beyond the scope of the current research.

Attachment Orientation

Participants’ own global attachment orientations were assessed using the Experiences in Close Relationships Questionnaire—Revised (ECR-R; Fraley et al., 2000). Global attachment anxiety (e.g., “I’m afraid that I will lose people’s love”) and avoidance (e.g., “I prefer not to be too close to people”) were each assessed with 18 items on a 7-point scale (1 = Strongly disagree to 7 = Strongly agree). Items were averaged separately for global attachment anxiety and avoidance.

Perceived global attachment, self-reported RS attachment, and perceived partner RS attachment orientations were assessed using the same measure described in Study 1, the ECR-RS (Fraley et al., 2011). Instructions for the global measure were modified to ask participants to think of their important relationships in general rather than about their relationship with their partner.

Reassurance Provision in Daily Life

Following the background survey, participants completed a 10-day ecological momentary assessment (EMA), during which each partner received four surveys per day (morning, afternoon, evening, and bedtime) at scheduled times they selected (as a couple) during the background session. Submissions were required within 1 hr of the scheduled time to be considered valid.⁷ In these surveys, partners independently rated the extent to which they showed affection (love, care, and fondness) for their partner by (a) touching them non-sexually (e.g., hugging, kissing, holding hands), (b) saying or writing things to them (e.g., saying “I love you,” giving a compliment, expressing gratitude), and (c) doing helpful, kind, or attentive things for them (e.g., doing chores, prioritizing time together, listening carefully to them). These ratings for affectionate touch, verbal affection, and practical affection, respectively, were made on a 9-point scale (1 = Not at all to 9 = Very much). Since the three affection behaviors were highly intercorrelated (rs > .70, all ps < .001), we calculated a composite affection provision score for each submission (up to 40 submissions per participant; N = 11,068).

Reassurance Provision During Stressor Discussion

After the EMA phase, couples visited the campus laboratory to participate in a series of activities and discussions, including two 7-min personal stressor discussions (order randomly determined). Participants were asked to identify an important problem or concern in their lives (a personal, non-relational issue) that they had not yet discussed, or had discussed the least, with their partner. On average, participants were moderately worried (M = 4.0) and stressed out (M = 4.0) thinking about their stressors, both measured on a 7-point scale. Each partner took turns discussing their personal stressor and then completed post-discussion questionnaires. We focused on reassurance provision while discussing the partner’s personal stressor, indexed by an average of two items measured in the post-discussion questionnaires. Specifically, participants rated the extent to which they showed that they love and/or care for their partner and comforted or reassured their partner during the discussion on a 7-point scale (1 = Not at all to 7 = Very much).

Data Analytic Strategy

We applied the same analytic strategy as in Study 1 to examine accuracy and bias in attachment perceptions and to test whether perceived partner attachment predicted reassurance provision during couple discussions. First, we examined perceptions of RS and global attachment in separate models. We then compared model parameters across attachment domains to test whether the extent of accuracy and bias differed between them. Specifically, we included an effect-coded attachment domain variable (RS = −1, global = 1) and its interactions with truth and bias components to test whether accuracy and bias estimates varied as a function of attachment domain. Second, for the EMA analyses, we estimated multilevel models with random intercepts for dyads, persons nested within dyads, and the dyad-by-time-point interaction to account for nonindependence of partners within couples, repeated observations within individuals, and partners’ responses at the same time point, respectively (del Rosario & West, 2025). This multilevel framework accommodates unequal numbers of observations across participants, allowing all available daily reports to contribute to parameter estimation without requiring imputation.

Results

Q1: Accuracy and Bias in Perceived Partner Attachment

As shown in Table 5, the patterns in perceptions of partners’ RS attachment were consistent with those found in Study 1. Regarding perceptions of partners’ RS attachment anxiety and attachment avoidance, people showed significant accuracy, a positive directional bias (i.e., overestimation), a projection bias, and a complementarity bias. Perceptions of partners’ global attachment anxiety followed a similar pattern: people demonstrated significant accuracy, a positive directional bias (i.e., overestimation), a projection bias, and a complementarity bias (see Table 5). For global attachment avoidance, people also showed significant accuracy, a positive directional bias (i.e., overestimation), and a complementarity bias. However, they did not exhibit a significant projection bias (i.e., assumed similarity).

Table 5.

Accuracy and Bias in Perceptions of Partners’ Attachment (Study 2).

Predictor	b [95% CI]	SE	p	β	d
RS attachment anxiety
Intercept (directional bias)	0.59 [0.45, 0.74]	0.07	<.001	—	0.34
Partners’ attachment anxiety (correlational accuracy)	0.33 [0.22, 0.45]	0.06	<.001	.30
Perceivers’ own attachment anxiety (projection bias)	0.23 [0.11, 0.36]	0.07	<.001	.21
Perceivers’ own attachment avoidance (complementarity bias)	0.32 [0.14, 0.50]	0.09	<.001	.20
RS attachment avoidance
Intercept (directional bias)	0.27 [0.19, 0.36]	0.04	<.001	—	0.31
Partners’ attachment avoidance (correlational accuracy)	0.37 [0.27, 0.47]	0.05	<.001	.37
Perceivers’ own attachment avoidance (projection bias)	0.20 [0.09, 0.31]	0.06	<.001	.20
Perceivers’ own attachment anxiety (complementarity bias)	0.13 [0.06, 0.21]	0.04	<.001	.19
Global attachment anxiety
Intercept (directional bias)	0.40 [0.25, 0.55]	0.08	<.001	—	0.30
Partners’ attachment anxiety (correlational accuracy)	0.62 [0.48, 0.75]	0.07	<.001	.47
Perceivers’ own attachment anxiety (projection bias)	0.18 [0.04, 0.32]	0.07	.011	.14
Perceivers’ own attachment avoidance (complementarity bias)	0.23 [0.08, 0.39]	0.08	.004	.15
Global attachment avoidance
Intercept (directional bias)	0.27 [0.15, 0.39]	0.06	<.001	—	0.17
Partners’ attachment avoidance (correlational accuracy)	0.53 [0.40, 0.65]	0.06	<.001	.43
Perceivers’ own attachment avoidance (projection bias)	−0.03 [-0.17, 0.10]	0.07	.615	−.03
Perceivers’ own attachment anxiety (complementarity bias)	0.20 [0.09, 0.31]	0.06	<.001	.19

Note. RS = relationship-specific.

We also examined whether accuracy and bias estimates varied as a function of attachment domain (RS vs. global). As shown in Supplemental Table 8, the interactions between attachment domain and partners’ attachment anxiety and avoidance were significant, suggesting that individuals demonstrated significantly greater correlational accuracy when perceiving their partner’s global attachment anxiety and avoidance compared to their RS attachment. Moreover, the interaction between the attachment domain and perceivers’ own attachment avoidance was significant, suggesting that participants exhibited significant projection bias when perceiving their partner’s RS attachment avoidance, but not when judging global avoidance. No other differences emerged across domains.

Q2: Reassurance Provision

Next, we examined whether perceptions of partners’ RS and global attachment anxiety (tested in separate models) predicted affection provision in daily life. As shown in Table 6, consistent with our hypotheses, people who perceived their partners to have greater RS attachment anxiety and/or greater global attachment anxiety showed significantly more affection provision in daily life than those who perceived their partners to be less anxiously attached in the relationship or globally. These findings were observed after controlling for perceived partner attachment avoidance and both partners’ self-reports of their own attachment anxiety and avoidance. Among these covariates, people who perceived their partners to have greater RS attachment avoidance showed significantly less affection provision in daily life (perceptions of partners’ global attachment avoidance did not predict affection provision), and perceivers’ own RS attachment avoidance and partners’ self-reported RS attachment avoidance significantly predicted less affection provision.

Table 6.

Perceptions of Partners’ Attachment Predicting Affection Provision (Study 2).

	Daily life				Stressor discussions
Predictor	b [95% CI]	SE	p	β	b [95% CI]	SE	p	β
RS attachment predictors
Perceptions of partners’ attachment anxiety	0.14 [0.03, 0.26]	0.06	.016	.10	0.16 [0.03, 0.29]	0.07	.019	.17
Perceptions of partners’ attachment avoidance	−0.20 [−0.38, −0.02]	0.09	.037	−.09	−0.13 [−0.33, 0.06]	0.10	.185	−.09
Partners’ attachment anxiety	0.03 [−0.12, 0.18]	0.07	.710	.02	0.02 [−0.13, 0.17]	0.08	.820	.02
Partners’ attachment avoidance	−0.23 [−0.45, −0.02]	0.11	.036	−.10	−0.04 [−0.25, 0.18]	0.11	.753	−.02
Perceivers’ own attachment anxiety	0.01 [−0.13, 0.16]	0.07	.869	.01	−0.11 [−0.25, 0.04]	0.07	.159	−.10
Perceivers’ own attachment avoidance	−0.29 [−0.50, −0.08]	0.10	.006	−.13	−0.28 [−0.49, −0.08]	0.11	.008	−.19
Global attachment predictors
Perceptions of partners’ attachment anxiety	0.17 [0.07, 0.27]	0.05	.001	.13	0.14 [0.03, 0.25]	0.06	.015	.17
Perceptions of partners’ attachment avoidance	−0.02 [−0.14, 0.11]	0.06	.802	−.01	−0.01 [−0.14, 0.13]	0.07	.911	−.01
Partners’ attachment anxiety	0.08 [−0.07, 0.24]	0.08	.320	.05	0.06 [−0.10, 0.21]	0.08	.488	.05
Partners’ attachment avoidance	−0.12 [−0.30, 0.07]	0.09	.213	−.06	−0.08 [−0.26, 0.11]	0.10	.427	−.06
Perceivers’ own attachment anxiety	0.00 [−0.14, 0.14]	0.07	.979	.00	0.02 [−0.12, 0.17]	0.07	.745	.02
Perceivers’ own attachment avoidance	−0.16 [−0.33, 0.03]	0.09	.068	−.08	−0.14 [−0.31, 0.03]	0.09	.102	−.11

Note. RS = relationship-specific.

Finally, we examined whether perceptions of partners’ RS and global attachment (tested in separate models) predicted supporters’ reassurance provision while discussing their partner’s personal stressors. As shown in Table 6, people who perceived their partners to have greater RS attachment anxiety and/or greater global attachment anxiety reported greater reassurance provision. Perceived partner avoidance was not significantly associated with reassurance provision, and perceivers’ own RS attachment avoidance was significantly negatively associated with their reassurance provision. These findings are consistent with our hypotheses, suggesting that people who perceived their partner as more anxiously attached, either to them specifically or in close relationships in general, provided more reassurance.⁸

Brief Discussion

Study 2 replicated evidence of accuracy, overestimation of insecurity, and both projection and complementarity bias for RS attachment (Q1). In addition, perceptions of partners’ global attachment also demonstrated significant accuracy, overestimation of insecurity, and complementarity bias. However, projection bias emerged only for perceptions of global attachment anxiety, not global attachment avoidance (unlike RS avoidance). Notably, people were significantly more accurate in judging their partner’s global attachment anxiety and avoidance compared to their RS attachment.

Consistent with our predictions for our second research question (Q2), individuals who perceived their partner as higher in attachment anxiety (either globally or specifically in their relationship) provided more reassurance in daily life and during partners’ personal stressor discussions. These findings suggest that perceptions of partners’ attachment anxiety—whether driven by accurate detection or biased judgment—may lead partners to initiate reassurance as a buffering strategy targeted to attachment anxiety.

General Discussion

While prior research has highlighted the benefits of attachment-matched buffering (e.g., Arriaga et al., 2018), it remains unclear whether partners accurately perceive one another’s attachment insecurity and tailor their behavior based on these perceptions. In the present research, we examined the extent to which partners accurately perceive one another’s RS and global attachment anxiety and avoidance, and we identified systematic biases in these perceptions. We also tested whether these perceptions predict partners’ provision of reassurance, a “safe” strategy that buffers attachment anxiety.

Accuracy and Bias in Attachment Perceptions

We found that individuals were able to perceive their partner’s attachment anxiety and avoidance with moderate accuracy (βs = .27–.47), with greater accuracy for global perceptions (βs = .43–.47) than RS perceptions (βs = .27–.37). To contextualize these effect sizes, we can compare them to LaBuda and Gere’s (2023) meta-analysis of accuracy and bias in romantic partner perceptions, where they observed an average accuracy effect size of β = .34 for personality traits and slightly lower effect sizes for positive and negative interaction traits (βs = .28, .24, respectively). If we consider global attachment to be akin to a personality trait and RS attachment to be a more contextually variable relational behavior, our effect sizes are similar to those observed in past work (LaBuda & Gere, 2023). This moderate degree of accuracy suggests that partners detect some aspects of attachment insecurity (e.g., amplified distress, clinginess) but that other aspects of attachment insecurity (or the extent of insecurity) may be unobservable or misinterpreted (accidentally or strategically).

Indeed, across both studies, people tended to perceive their partners as more insecurely attached than their partners reported themselves to be. This directional bias aligns with prior meta-analytic findings indicating that people hold negatively-biased relational perceptions that allow them to detect and mitigate potential risks (e.g., under-perceiving a partner’s relationship satisfaction; Fletcher & Kerr, 2010; LaBuda & Gere, 2023). Over-perceiving attachment insecurity may serve a regulatory function by sensitizing individuals to potential threats and prompting them to engage in pro-relationship behaviors (e.g., providing reassurance). The degree of directional bias did not significantly differ between RS and global attachment perceptions, suggesting that partners are similarly sensitive to detecting insecurity within their specific relationship and across close relationships more broadly, potentially enabling partners to buffer insecurities that arise in either domain.

The current investigation also found consistent evidence of projection bias (assumed similarity) in perceptions of partners’ RS attachment anxiety and avoidance. In contrast, projection only emerged for global attachment anxiety, not global attachment avoidance. Prior research indicates that projection is more likely when the focal characteristic is shared between perceiver and target (e.g., evaluations of a shared romantic relationship) and less likely when the characteristic is uniquely tied to the target (e.g., personality traits; Schul & Vinokur, 2000), likely because individuals are more inclined to use their own experiences and emotions as benchmarks when the perceived domain involves shared meaning or context. This past work helps to clarify why partners may assume greater similarity in RS attachment than global attachment. Our findings also reflect a pattern observed in the meta-analysis (LaBuda & Gere, 2023), which found that projection biases tend to be strongest when accuracy is lowest. This inverse association, evident in the contrast between RS and global attachment perceptions, may indicate that when people feel less confident in their ability to judge their partner accurately (e.g., RS attachment), they rely more heavily on their own internal states as a heuristic for perception.

Finally, this research offers the first empirical test of a bias uniquely relevant to the attachment context: complementarity bias. Across both studies, individuals’ own insecurity predicted viewing their partners as higher on the opposite dimension of insecurity than themselves (e.g., anxious individuals perceiving their partners as more avoidant). This pattern emerged for both RS and global attachment perceptions. Importantly, this bias appeared independently of actual partner characteristics (i.e., accuracy), a general tendency to overperceive insecurity, and assumed similarity between partners. Both assumed similarity and this inclination to view one’s partner as embodying the opposite form of insecurity may interfere with accurate understanding of the partner’s emotions, needs, and expectations. The complementarity bias may help explain the lower relationship functioning often observed in anxious–avoidant pairings relative to other dyadic combinations (e.g., anxious–anxious) and may contribute to the broader demand–withdraw pattern commonly observed in distressed relationships (Beck et al., 2013; Overall et al., 2013).

Implications of Perceptions for Safe Strategy Buffering Behaviors

We hypothesized that participants would provide greater reassurance when they perceived their partner as higher in attachment anxiety. This pattern would reflect a strategic form of attachment-matched buffering, as offering reassurance that the relationship is safe (i.e., a “safe strategy”) is theorized to effectively regulate anxiously attached individuals’ doubts about their partner’s commitment or regard, thereby protecting the relationship from more destructive consequences (Arriaga et al., 2018). This prediction was not supported in Study 1’s goal discussions but was supported in several unique analyses in Study 2. Specifically, participants who perceived their partner as higher in either global or RS attachment anxiety provided more reassurance in daily life and during laboratory-based personal stressor discussions. Contextual differences between studies may explain these disparate results. The personal goal discussions in Study 1 may not have strongly activated attachment anxiety, resulting in less observable need for reassurance. Additionally, as most of the Study 1 discussions occurred over Zoom, there may have been limited opportunity for emotional expression and responsiveness, further dampening opportunities for reassurance provision, as some forms of reassurance (e.g., affectionate touch) can only be provided in person. Differences in sample characteristics may also have contributed to different results across studies. Couples in Study 1 were, on average, in earlier stages of their relationships compared to those in Study 2, and thus may have relied on their own preferences rather than partner attachment perceptions when offering reassurance. Notably, supplementary analyses revealed that perceived partner attachment anxiety did not predict general supportive or pro-relational behaviors in either goal or stressor discussions (e.g., trying to understand the partner’s feelings, offering encouragement or support; see Supplemental Tables 9 and 10). This specificity is theoretically meaningful: it suggests that people who perceive their partners as anxiously attached do not simply increase positive behavior broadly, but selectively provide reassurance of their love, care, and continued commitment to directly address attachment anxiety’s core concerns.

The current research contributes to the insecurity buffering literature by helping to explicate the process of partner buffering. Whereas past work has shown that reassurance effectively buffers attachment anxiety (Overall et al., 2022), a critical but largely unexamined question is how some partners come to offer reassurance in the first place. By demonstrating that partners perceive each other’s attachment tendencies with some accuracy (and some bias) and then use these perceptions to guide their behavior, the present findings illuminate the perceptual pathway underlying partner buffering. Moreover, the daily diary findings suggest that buffering is not solely a reactive response to acute distress but may be an ongoing relational maintenance process, with partners providing elevated reassurance across the routine of daily life when they perceive their partner as anxiously attached. These findings suggest that effective partner buffering could be initiated by helping partners to interpret one another’s attachment orientations more accurately. Although the directional bias we observed likely enables partners to “err on the side of caution” and offer more reassurance than may be necessary to mitigate relational risks (Haselton & Buss, 2000), the projection and complementary biases may simply misdirect effective buffering behaviors. It is important to note that while reassurance provision buffers concurrent attachment anxiety, excessive reassurance may not only tax the provider but may also reinforce insecurity over time (e.g., Arriaga et al., 2018). Experiences that allow people to change the underlying beliefs about themselves and others are needed to enhance security (Arriaga & Kumashiro, 2019; Arriaga et al., 2018; Jakubiak, Fuentes, & Feeney, 2023; Jakubiak, Fuentes, Sun, & Feeney, 2023).

Strengths and Limitations

The current research employs the Truth and Bias Model (West & Kenny, 2011) to rigorously examine both the accuracy and the multiple sources of bias in attachment perceptions. By assessing both global and RS attachment, this work also provides a more nuanced understanding of how individuals perceive their partner’s attachment tendencies across different relational contexts. Importantly, the research goes beyond perception alone by testing whether individuals strategically modulate their behavior based on these perceptions, offering an initial step toward addressing a key gap in the partner buffering literature. Finally, by examining these processes across three distinct contexts—personal goal discussions, personal stressor discussions, and daily life—the findings support more generalizable conclusions.

Several limitations offer avenues for future research. First, we operationalized the “truth” of attachment insecurity using individuals’ self-reports, but people may not always be fully aware of their own insecurity. Past research shows that partner buffering strategies are effective when matched to a target’s self-reported attachment (e.g., Kim et al., 2018; Overall et al., 2014), suggesting that self-reports have value as a form of truth. Nonetheless, future research should incorporate alternative indicators, such as behavioral observations, informant reports, or implicit measures, to offer a more comprehensive account of accuracy and bias in partner perceptions. Second, we assessed partner reassurance via participants’ self-reports because our primary question concerns whether partners regulate their behavior based on perceived partner attachment. Future research could examine whether self-reported reassurance translates into externally observable behaviors coded by independent raters. Because participants anticipated receiving surveys throughout the day and following discussions, overall levels of reassurance provision may be higher in these studies than in other contexts. This limitation is unlikely to undermine our findings, as such effects should operate uniformly across participants rather than covary with attachment perceptions, but may skew descriptives. Additionally, our operationalizations of reassurance provision varied across contexts. Although these different manifestations all saliently demonstrate love, care, and commitment, future research would benefit from more comprehensive measurement across contexts. Finally, this study focused exclusively on safe strategies for buffering attachment anxiety because our datasets include measures of reassurance provision but do not include measures to assess soft strategies relevant to buffering avoidance (e.g., downplaying the emotional intensity of conflicts). Future work should examine a wider range of safe- and soft-strategy buffering behaviors across diverse relational contexts to assess whether perceptions of partner attachment guide buffering behaviors broadly.

In conclusion, across two studies, the present research demonstrated that individuals perceive their romantic partner’s global and RS attachment tendencies with moderate accuracy. At the same time, they tend to overestimate their partner’s attachment insecurity, project their own attachment orientation onto their partner, and view partners as more insecure on the opposite dimension than themselves. Importantly, perceptions of a partner’s attachment anxiety were associated with greater provision of reassurance both during personal stressor discussions and in daily life more broadly. Together, these findings underscore the nuanced ways in which people perceive their partner’s attachment tendencies and highlight the potential significance of these perceptions in buffering relationship insecurities.

Supplemental Material

sj-docx-1-psp-10.1177_01461672261448771 – Supplemental material for Perceiving to Provide: How Partner Attachment Perceptions Inform Reassurance Provision in Romantic Relationships

Supplemental material, sj-docx-1-psp-10.1177_01461672261448771 for Perceiving to Provide: How Partner Attachment Perceptions Inform Reassurance Provision in Romantic Relationships by Elina R. Sun, Xiangjing Kong, Jason A. Mitala, Jeewon Oh and Brett K. Jakubiak in Personality and Social Psychology Bulletin

Footnotes

ORCID iDs

Elina R. Sun

Xiangjing Kong

Jason A. Mitala

Jeewon Oh

Brett K. Jakubiak

Ethical Considerations

All procedures and materials were approved by the university’s institutional review board (Study 1: IRB No. 21-331 by Syracuse University; Study 2: IRB No. 23-229 by Syracuse University).

Consent to Participate

All participants provided written and verbal consent.

Consent for Publication

All participants provided written consent.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This material is based upon work supported by the National Science Foundation under Grant No. BCS-2140978.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data and syntax to reproduce the results are available at the Online Supplemental Material and

Supplemental Material

Supplemental material is available online with this article.

Notes

References

Arriaga

X. B.

Kumashiro

(2019). Walking a security tightrope: Relationship-induced changes in attachment security. Attachment in Adulthood, 25, 121–126. https://doi.org/10.1016/j.copsyc.2018.04.016

Arriaga

X. B.

Kumashiro

Simpson

J. A.

Overall

N. C.

(2018). Revising working models across time: Relationship situations that enhance attachment security. Personality and Social Psychology Review, 22(1), 71–96. https://doi.org/10.1177/1088868317705257

Banai

Weller

Mikulincer

(1998). Inter-judge agreement in evaluation of adult attachment style: The impact of acquaintanceship. British Journal of Social Psychology, 37(1), 95–109. https://doi.org/10.1111/j.2044-8309.1998.tb01159.x

Beck

L. A.

Pietromonaco

P. R.

DeBuse

C. J.

Powers

S. I.

Sayer

A. G.

(2013). Spouses’ attachment pairings predict neuroendocrine, behavioral, and psychological responses to marital conflict. Journal of Personality and Social Psychology, 105(3), 388–424. https://doi.org/10.1037/a0033056

Bolger

Davis

Rafaeli

(2003). Diary methods: Capturing life as it is lived. Annual Review of Psychology, 54, 579–616. https://doi.org/10.1146/annurev.psych.54.101601.145030

Brennan

K. A.

Clark

C. L.

Shaver

P. R.

(1998). Self-report measurement of adult attachment: An integrative overview. In Simpson

J. A.

Rholes

W. S.

(Eds.), Attachment theory and close relationships (pp. 46–76). The Guilford Press.

Campbell

Simpson

J. A.

Boldry

Kashy

D. A.

(2005). Perceptions of conflict and support in romantic relationships: The role of attachment anxiety. Journal of Personality and Social Psychology, 88(3), 510–531. https://doi.org/10.1037/0022-3514.88.3.510

Candel

O.-S.

Turliuc

M. N.

(2019). Insecure attachment and relationship satisfaction: A meta-analysis of actor and partner associations. Personality and Individual Differences, 147, 190–199. https://doi.org/10.1016/j.paid.2019.04.037

Cobb

R. J.

Davila

Bradbury

T. N.

(2001). Attachment security and marital satisfaction: The role of positive perceptions and social support. Personality and Social Psychology Bulletin, 27(9), 1131–1143. https://doi.org/10.1177/0146167201279006

10.

del Rosario

K. S.

West

T. V.

(2025). A practical guide to specifying random effects in longitudinal dyadic multilevel modeling. Advances in Methods and Practices in Psychological Science, 8(3), Article 25152459251351286. https://doi.org/10.1177/25152459251351286

11.

Epley

Keysar

Van Boven

Gilovich

(2004). Perspective taking as egocentric anchoring and adjustment. Journal of Personality and Social Psychology, 87(3), 327–339. https://doi.org/10.1037/0022-3514.87.3.327

12.

Fletcher

G. J. O.

(2015). Accuracy and bias of judgments in romantic relationships. Current Directions in Psychological Science, 24(4), 292–297. https://doi.org/10.1177/0963721415571664

13.

Fletcher

G. J. O.

Kerr

P. S. G.

(2010). Through the eyes of love: Reality and illusion in intimate relationships. Psychological Bulletin, 136(4), 627–658. https://doi.org/10.1037/a0019792

14.

Fraley

R. C.

Heffernan

M. E.

Vicary

A. M.

Brumbaugh

C. C.

(2011). The Experiences in Close Relationships—Relationship Structures Questionnaire: A method for assessing attachment orientations across relationships. Psychological Assessment, 23(3), 615–625. https://doi.org/10.1037/a0022898

15.

Fraley

R. C.

Shaver

P. R.

(2000). Adult romantic attachment: Theoretical developments, emerging controversies, and unanswered questions. Review of General Psychology, 4(2), 132–154. https://doi.org/10.1037/1089-2680.4.2.132

16.

Fraley

R. C.

Waller

N. G.

Brennan

K. A.

(2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78(2), 350–365. https://doi.org/10.1037/0022-3514.78.2.350

17.

Gagné

F. M.

Lydon

J. E.

(2004). Bias and accuracy in close relationships: An integrative review. Personality and Social Psychology Review, 8(4), 322–338. https://doi.org/10.1207/s15327957pspr0804_1

18.

Gordon

A. M.

Cross

Ascigil

Balzarini

Luerssen

Muise

(2022). Feeling appreciated buffers against the negative effects of unequal division of household labor on relationship satisfaction. Psychological Science, 33(8), 1313–1327. https://doi.org/10.1177/09567976221081872

19.

Haselton

M. G.

Buss

D. M.

(2000). Error management theory: A new perspective on biases in cross-sex mind reading. Journal of Personality and Social Psychology, 78(1), 81–91. https://doi.org/10.1037/0022-3514.78.1.81

20.

Jakubiak

B. K.

Fuentes

J. D.

Feeney

B. C.

(2023). Perceptions of oneself and one’s spouse following a stressor discussion predicting attachment insecurity over one year. Journal of Social and Personal Relationships, 40(10), 3195–3219. https://doi.org/10.1177/02654075231170145

21.

Jakubiak

B. K.

Fuentes

J. D.

Sun

E. R.

Feeney

B. C.

(2023). Social support is a balancing act: Mitigating attachment anxiety by supporting independence and dependence. Journal of Social and Personal Relationships, 40(7), 2127–2148. https://doi.org/10.1177/02654075221142000

22.

Kim

K. J.

Feeney

B. C.

Jakubiak

B. K.

(2018). Touch reduces romantic jealousy in the anxiously attached. Journal of Social and Personal Relationships, 35(7), 1019–1041. https://doi.org/10.1177/0265407517702012

23.

Klohnen

E. C.

Weller

J. A.

Luo

Choe

(2005). Organization and predictive power of general and relationship-specific attachment models: One for all, and all for one? Personality and Social Psychology Bulletin, 31(12), 1665–1682. https://doi.org/10.1177/0146167205278307

24.

LaBuda

J. E.

Gere

(2023). A meta-analytic review of accuracy and bias in romantic partner perceptions. Psychological Bulletin, 149(9–10), 580–610. https://doi.org/10.1037/bul0000405

25.

Lemay

E. P.

Jr. Dudley

K. L

. (2011). Caution: Fragile! Regulating the interpersonal security of chronically insecure partners. Journal of Personality and Social Psychology, 100(4), 681–702. https://doi.org/10.1037/a0021655

26.

Chan

D. K.-S.

(2012). How anxious and avoidant attachment affect romantic relationship quality differently: A meta-analytic review. European Journal of Social Psychology, 42(4), 406–419. https://doi.org/10.1002/ejsp.1842

27.

Mashek

D. J.

Aron

Boncimino

(2003). Confusions of self with close others. Personality and Social Psychology Bulletin, 29(3), 382–392. https://doi.org/10.1177/0146167202250220

28.

Mikulincer

Shaver

P. R.

(2016). Attachment in adulthood: Structure, dynamics, and change (2nd ed.). The Guilford Press.

29.

Molero

Shaver

P. R.

Fernández

Alonso-Arbiol

Recio

(2016). Long-term partners’ relationship satisfaction and their perceptions of each other’s attachment insecurities. Personal Relationships, 23(1), 159–171. https://doi.org/10.1111/pere.12117

30.

Murray

S. L.

Holmes

J. G.

Griffin

D. W.

(1996). The benefits of positive illusions: Idealization and the construction of satisfaction in close relationships. Journal of Personality and Social Psychology, 70(1), 79–98. https://doi.org/10.1037/0022-3514.70.1.79

31.

Overall

N. C.

Fletcher

G. J. O.

Friesen

M. D.

(2003). Mapping the intimate relationship mind: Comparisons between three models of attachment representations. Personality and Social Psychology Bulletin, 29(12), 1479–1493. https://doi.org/10.1177/0146167203251519

32.

Overall

N. C.

Girme

Y. U.

Lemay

E. P.

Jr. Hammond

M. D

. (2014). Attachment anxiety and reactions to relationship threat: The benefits and costs of inducing guilt in romantic partners. Journal of Personality and Social Psychology, 106(2), 235–256. https://doi.org/10.1037/a0034371

33.

Overall

N. C.

Pietromonaco

P. R.

Simpson

J. A.

(2022). Buffering and spillover of adult attachment insecurity in couple and family relationships. Nature Reviews Psychology, 1(2), 101–111. https://doi.org/10.1038/s44159-021-00011-1

34.

Overall

N. C.

Simpson

J. A.

Struthers

(2013). Buffering attachment-related avoidance: Softening emotional and behavioral defenses during conflict discussions. Journal of Personality and Social Psychology, 104(5), 854–871. https://doi.org/10.1037/a0031798

35.

Rollings

Kavanagh

Balabanova

Keane

Waller

B. M.

(2024). Facial behaviour and first impressions in computer mediated communication. Computers in Human Behavior, 161, Article 108391. https://doi.org/10.1016/j.chb.2024.108391

36.

Rossignac-Milon

Higgins

E. T.

(2018). Epistemic companions: Shared reality development in close relationships. Shared Reality, 23, 66–71. https://doi.org/10.1016/j.copsyc.2018.01.001

37.

Ruvolo

A. P.

Fabin

L. A.

(1999). Two of a kind: Perceptions of own and partner’s attachment characteristics. Personal Relationships, 6(1), 57–79. https://doi.org/10.1111/j.1475-6811.1999.tb00211.x

38.

Schul

Vinokur

A. D.

(2000). Projection in person perception among spouses as a function of the similarity in their shared experiences. Personality and Social Psychology Bulletin, 26(8), 987–1001. https://doi.org/10.1177/01461672002610008

39.

Simpson

J. A.

Overall

N. C.

(2014). Partner buffering of attachment insecurity. Current Directions in Psychological Science, 23(1), 54–59. https://doi.org/10.1177/0963721413510933

40.

Snyder

M. D.

Sun

E. R.

Nie

Jakubiak

B. K.

(2024). When and how do people share? Attachment anxiety predicts support-seeking strategies. Personal Relationships, 31(4), 1113–1132. https://doi.org/10.1111/pere.12573

41.

Stern

West

T. V.

(2018). Assessing accuracy in close relationships research: A truth and bias approach. Journal of Social and Personal Relationships, 35(1), 89–111. https://doi.org/10.1177/0265407517712901

42.

Strauss

Morry

M. M.

Kito

(2012). Attachment styles and relationship quality: Actual, perceived, and ideal partner matching. Personal Relationships, 19(1), 14–36. https://doi.org/10.1111/j.1475-6811.2010.01333.x

43.

Sun

E. R.

Jakubiak

B. K.

(2024). Attachment avoidance predicts limited and selective sharing of personal events in close relationships. Personal Relationships, 31(2), 397–419. https://doi.org/10.1111/pere.12537

44.

Maxwell

J. A.

Kim

J. J.

Peragine

Impett

E. A.

Muise

(2022). Is my attachment style showing? Perceptions of a date’s attachment anxiety and avoidance and dating interest during a speed-dating event. Journal of Research in Personality, 100, Article 104269. https://doi.org/10.1016/j.jrp.2022.104269

45.

West

T. V.

Kenny

D. A.

(2011). The truth and bias model of judgment. Psychological Review, 118(2), 357–378. https://doi.org/10.1037/a0022936

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.40 MB

0.00 MB