Introduction

Baldwin (1992) used the transtheoretical term “relational schema” to capture ideas from various theories that all describe mental structures that inform individuals on how they think about themselves, others, and the rules of interaction between the two. Relational schemas influence perception and behavior of individuals which recursively influences the nature of the relational schema. However, formation, function, stability, and breadth of life situations and relationships the relational schema impacts differ depending on the theory being utilized.

Chen et al. (2006) built on Baldwin’s (1992) work and used the term “relational self” in their effort to synthesize theories that postulate relational schemas are formed from, and dependent on, a person’s relationships. The “relational self” concept holds that ideas about one’s own value, capability, and identity are directly tied to the interaction with another person. The “relational self” is adaptable and fluid because individuals are in multiple relationships throughout life. Readers are encouraged to review both Baldwin (1992) and Chen et al. (2006) for fuller coverage of the theories which informed their syntheses. Attachment theory, which can be categorized under “relational schema” and “relational self” syntheses, will provide the theoretical support for the present article.

Bowlby set out in his attachment and loss series (Bowlby 1969, 1982, 1973, 1980) to create a theory of human functioning that was grounded in the central relationships of life. Over the years, attachment researchers have developed into many different branches with unique strengths and perspectives on Bowlby’s original conceptualizations and predictions. It is not surprising that as time has gone on the different branches have developed different methods on how attachment should be measured, studied, and what aspects of attachment the empirical evidence supports. Each of the different branches of attachment research claim connection to Bowlby’s original work and ideas.

Mikulincer and Shaver (2008) suggested that Bowlby’s concept of internal working models bridges the gap between many of the different extensions and branches of Bowlby’s original work. Indeed, Bowlby (1969, 1982) in describing internal working models, stated that attachment behavior is “controlled by a behavioural system conceived as an organisation existing within the child… This organisation, conceived as a permanent, indeed a central, feature of the child’s personality, is never idle” (sic, p. 373).

Bowlby (1982) went on to reinforce that this control system has two main functions in identifying internal or external “danger or stress” and the “whereabouts and accessibility of the attachment figure” (p. 373). Based on the appraisal of these two areas, the individual then makes decisions on how to behave “until such time as the system’s sensors indicate that the child’s situation has changed appropriately, experienced by him as feeling comforted and secure (p. 373).

Internal working models are not limited to child development. In 1973, Bowlby dedicated several pages to the concept of working models in reference to attachment. In the 1973 work, Bowlby discusses how individuals can change their internal working models over time, even into adulthood. Bowlby stated that for adults, the “presence” and “availability” of attachment figures is based on the perceived “accessibility” and “responsiveness” of attachment figures rather than the physical presence of the attachment figure (Bowlby 1973). It is in the conditions of these working models, based on countless interactions with attachment figures in childhood and beyond, that an individual gains confidence within themselves and others. Bowlby postulated that there was an internal working model for the self as well as others. Bowlby 1969, 1982) recognized that the internal working model of the self and others were differentiated conceptually and recursive experientially.

The working modes of self and others deeply influence attachment behavior and how individuals handle perceived stressors of life.

[Each] individual builds working models of the world and of himself in it, with the aid of which he perceives events, forecasts the future, and constructs his plans. In the working model of the world that anyone builds, a key feature is his notion of who his attachment figures are, where they may be found, and how they may be expected to respond. Similarly, the working model of the self that anyone builds a key feature is his notion of how acceptable or unacceptable he himself is in the eyes of his attachment figures. On the structure of these complementary models are based that persons’ forecasts of how accessible and responsive his attachment figures are likely to be should he turn to them for support. And, in terms of the theory now advanced, it is on the structure of those models that depends, also, whether he feels confident that his attachment figures are in general readily available or whether he is more or less afraid that they will not be available-occasionally, frequently, or most of the time. (Bowlby 1973, p. 203)

The implications for understanding the scope and impact of internal working models are especially profound in settings such as the therapy room. Individuals, couples, and families attend therapy and frequently express intense affect in the presence of the clinician. If the clinician perceives the intense affect of the family as stressful, or a threat, the attachment system within the therapist would be activated and subsequently effect the therapist’s perception, evaluation, and interventions. This study seeks to better understand the perception of conflict external to an observer in the context of attachment anxiety and avoidance.

Two main branches of understanding of attachment, and internal working models, have been developed over the years. George et al. (as cited in Hesse, 2008) developed the Adult Attachment Interview (AAI) to explore patterns in how adults recall experiences in their childhood. How the individual describes their childhood experiences, and responses to their childhood experiences, is then categorized into themes connected to attachment styles. The AAI requires extensive training in administering and interpreting responses to the interview and has shown to have strong reliability and validity (see Hesse 2008). The AAI was originally codified to correlate with Ainsworth’s Secure, Anxious/ambivalent, and Avoidant categorization of attachment styles.

Hazan and Shaver (1987) developed a pencil paper assessment for adults to identify their own relationship with one of three descriptions consistent with Ainsworth’s attachment style typology. Interestingly, Hazan and Shaver also argued for Bowlby’s original position that internal working models are in effect over the life of the individual as justification for studying adult romantic attachment. Their questionnaire was based on available models of attachment style at the time, i.e., Ainsworth’s typology (Hazan and Shaver, 1987).

Bartholomew (1990), and Bartholomew and Horowitz (1991) reaffirmed Bowlby’s internal working models in the study of adult romantic attachment. They created a punnet square with one side of the square representing the working models of self (dependence) and the other side representing the working model of others (avoidance). While Bartholomew’s (1990) conceptualization was categorical in nature, it set the stage for later research that placed attachment anxiety (Bartholomew’s dependence concept) and attachment avoidance on continuous dimensions (See Fraley et al. 2000). Mikulincer and Shaver (2008) also emphasized the connection between working models of self as the underlying mechanism behind the dimension of attachment anxiety while the internal working model of others was conceptualized to underlie attachment avoidance. Through using attachment dimensions in research, as opposed to a categorical approach, a deeper understanding of attachment in adulthood can be achieved (Fraley et al. 2000).

As Bowlby originally postulated, internal working models affect the way an individual perceives the world and their relative efficacy in that world. A large and growing literature has given credence to this hypothesis. Research in the social psychological branch of attachment has explored how attachment dimensions affect social perception (see Mikulincer and Shaver 2008 for an extensive review), and processing of attachment related stimuli (see Edelstien and Gillath 2007 for an example).

Edelstien and Gillath 2007 illustrated that when under a stressful experimental procedure utilizing attachment related words (Emotional Stroop Test), participants higher in attachment avoidance performed more poorly in the test when compared to neutral words presented in the same test. Mikulincer et al. (2009) gave key words to participants and found individuals with low attachment anxiety and low attachment avoidance created narratives that reflected a secure attachment “script.”

Of particular relevance to the current study, a growing body of literature has explored the effects of attachment dimensions, and styles, on recognition of facial expressions (e.g., Maier et al. 2005; Niedenthal et al. 2002; Dewitte and De Houwer 2008; Suslow et al. 2010; Cooper et al. 2009). A brief review of the more recent literature on attachment and processing of social information follows.

Maier et al. (2005) showed participants 38 different photographs of facial expressions. The expressions ranged from negative to positive with neutral facial expression occupying the middle of the spectrum. Photos illustrating a variety of social interactions (e.g., partner violence or a couple kissing) and neutral scenes (e.g., mountains and animals) were also presented to subjects.

Subjects were asked to identify the content of the photo, and if the photo contained a person, they were to identify the emotion or action being expressed. Photos were displayed for 15 ms at a time with increasingly longer durations of exposure. Experimenters would document at what point in time the participants correctly identified the image, action, and emotion expressed (see Maier et al. 2005 for further methodological details).

Maier et al. (2005) predicted that preoccupied individuals, characterized by high attachment anxiety and low avoidance, would be faster at identifying negative affect and instead found support for dismissing styles to be faster at identifying negative affect in others. They did note however, that the correlation between preoccupied styles and speed of identification was in the same direction for the dismissing style.

Bowlby originally hypothesized that the attachment system would be engaged when individuals were in distress subsequently affecting perception. Niedenthal et al. (2002) found that in the nondistressed condition, fearfully attached individuals stated that happy and angry (both approach oriented emotions) disappeared sooner in the slide-show as compared to secure, preoccupied and dismissive styles. Participants identified as preoccupied perceived the emotions to stay for longer periods of time followed by dismissive style. When in distress, participants identified as preoccupied and dismissive perceived sadness and anger to disappear, or “offset,” more quickly than fearful attachment. Niedenthal et al. (2002) suggested that the attachment processes operate at automatic as well as controlled levels. They postulated that each style had similar rapid recognition of attachment triggers (automatic process) in order to determine which response to make behaviorally (controlled process in order to approach or avoid). Contrary to Niedenthal et al. (2002) and Maier et al. (2005); Cooper et al. (2009) did not find connections between perception and attachment styles within the different experiments of their own study.

Silva et al. (2012) gave participants a series of unrelated images including: nature scenes, animals, and facial expressions. Subjects were asked to identify at what point a previously identified target image appeared. The experimental condition involved placing a photo of different facial expressions immediately preceding the target image. Accuracy in identifying the target image was then used as the dependent variable. Their research showed that when “negative distracters” immediately preceded the target image, participants identified as anxiously attached were less accurate whereas identified avoidant individuals showed no such interference in the task (Silva et al. 2012). The researchers concluded that avoidant styles were able to suppress the negative affect to perform the task while anxious styles would become more focused on the negative affect “distracter”.

Dewitte and De Houwer (2008) found similar results showing that a combination of high attachment anxiety and high avoidance led to decreased attention on happy faces and high attachment avoidance showed a reduced attention for angry faces. Of interest, Dewitte and De Houwer (2008)were also able to model the inter-relatedness of attachment anxiety and avoidance dimensions as measured by the ECR without reducing data into categories.

The current study seeks to contribute to the literature by exploring attachment dimensions and social information processing by exposing participants to photographs and videos of couples interacting with each other. It was hypothesized that individuals endorsing increased attachment anxiety would be more likely to perceive greater negative affect and less positive affect in couples. It was also hypothesized that individuals that endorse higher attachment avoidance would minimize negative affect while perceiving greater positive affect in couples.

Method

Sample

Forty participants responded to recruitment strategies that sought to collect a sample from the general public, psychotherapy graduate students, and licensed mental health professionals. Of the 40 original participants, 39 participants (11 male, 28 female) had some data for analysis purposes. Participants were an average of 28.77 years old with 16.82 years of education. Participants reported being single (35.9 %), 5.1 % as divorced, and 59 % reporting that they were either married (33.4 %) or in a committed relationship (25.6 %). It is important to note the gender composition of the general population group and therapist group. There were nine male and 10 female participants in the general population group while there were two males and 18 females in the therapy and graduate student group. Ethnicity of the entire sample was 19.4 % (n = 7) African American, 2.8 % (n = 1) Arab-American, 71.8 % (n = 28) Caucasian, and 7.7 % (n = 3) did not identify their ethnicity. Ethnicity was distributed across the therapist and general population groups with two of the seven African Americans and 15 of the 28 Caucasians in the therapist group.

Procedure

After receiving approval from the appropriate institutional review boards, participants were recruited via advertisements in undergraduate and graduate classrooms, fliers distributed to different groups and clubs on a southeastern university campus, and fliers sent to community mental health agencies. Three rounds of recruitment materials were also mailed to all licensed mental health professionals listed in the local yellow pages.

Participants arrived at the lab and upon consenting to participate were oriented to the equipment and research procedures. Basic demographic information was recorded prior to the start of the laboratory procedure. All other pencil/paper questionnaires were completed after the laboratory procedure to avoid any priming effect the questionnaires may have had. Neurophysiological measurements were taken during the course of the study including galvanic skin response, heart rate, and 19 channels of EEG data. Placement of electrodes was accomplished via a specialized cap following the international 10–20 placement system. EEG data was recorded via Nexus 32 (Mind Media, The Netherlands) channel units (see Werner-Wilson et al. 2011) for technical details of electrode placement and measurement. Individuals participating in the study experienced 4 min of eyes closed and 4 min of eyes open relaxation prior to being exposed to the stimuli to establish baseline measurements. Although neurophysiological data was not reported in this study, procedures related to neurophysiological measurement may potentially act as a stressor which could activate the attachment system.

Once physiological measures were recording accurately and baseline data was recorded, participants completed a prediction and evaluation task by observing still images and videos of three different couples engaged in a problem solving task. If the participant recognized any person in couple, that couple’s video was not used. Videos were randomly presented to each participant to counter any practice effects or variance introduced by presenting the same ordering of couples to every participant.

The videos, photographs, and associated data from the couples observed were collected previously as a part of another study and the couples agreed to have their data be available for future research. As a part of the other study each couple filled out the revised dyadic adjustment scale (RDAS; Busby et al. 1995). This data was used for the purposes of the current study to determine if the couple had good relational adjustment (both individuals scored at least 50 on the RDAS, identified in the present study as Couple A), ambivalent adjustment (both individuals scored on the RDAS between 46 and 50, identified in the present study as Couple B), or poorly adjusted (both scored at least below 48 and 46 on the RDAS, identified in the present study as Couple C).

Prediction Task

One still photograph was extracted every 10 s from a video of a couple engaging in a problem solving task. Participants were given instruction that they would see a photograph of a couple engaging in a problem solving task for 5 s and were asked to make a prediction, based on the photograph they were observing, of the direction of the conflict to come. Without seeing the next image, participants predicted if the next image in the sequence would increase in conflict intensity, decrease in conflict intensity, or maintain the current level of conflict they just saw. Images were shown in sequence of the interaction that occurred. Participants recorded their responses via a wireless 10-key device. Colored paint that rose when dry (e.g., “puff paint”) was placed on the 4, 5, and 6 keys of the key pad to facilitate tactile awareness of the keys in an attempt to minimize head and eye movement that would introduce artifact in the EEG. Participants had 5 s to record their prediction and then had a five second rest period before seeing the next photo.

Evaluation Task

Following the prediction task, the participant watched 5 min of video, including audio, of the same couple engaged in a problem solving task. The still images were culled from the videos observed. Immediately following the 5 min video, participants were asked to rate the overall intensity of the couples interactions for the 5 min of video they had just observed. Participants rated 18 different affective or interaction based dimensions on a scale from 0 to 9 with 9 being the strongest expression of the target dimension (e.g., anger, humor, warm, contemptuous, etc.) and 0 indicating the absence of the dimension. Individuals used the 10-key pad to record their responses. The 18 dimensions and scaling tasks were based off of the dimensions used by Waldinger et al. (2004). No time limits were given to subjects in rating the couple on each dimension. Lab assistants in an adjoining room ensured that the subjects’ responses were recorded accurately.

Participants were reimbursed $40 dollars for the approximately 90 min they spent in the laboratory setting. In an effort to keep participants blind to the purposes of the study, they were given a questionnaire packet to take home, or fill out after they completed the laboratory portion of the study. Individuals received an additional $20 when the questionnaire packet was returned. Questionnaires contained detailed questions about experiences from their family of origin, attachment, and differentiation.

Measures

The experiences in close relationships—revised (ECR-R; Fraley et al. 2000) resulted from using item response theory on items from four of the most researched adult romantic attachment questionnaires. Fraley et al. 2000 then established the final version of the ECR-R which contains 36 items that reliably assess attachment dimensions of anxiety and avoidance in reference to adult relationships. Much of the literature reviewed pertaining to the current study used the original ECR or one of the other three scales the ECR-R was based from. This allows the results of the ECR-R to be put more firmly into context with previous work. Additionally, Mikulincer and Shaver (2008) assert that attachment anxiety as measured by instruments such as the ECR-R is consistent with Bowlby’s internal working model of the self with attachment avoidance is representative of the internal working model of others.

The ECR-R was scored so that high scores indicated high attachment anxiety and high attachment avoidance. Consistent with research using the ECR or ECR-R, alpha reliability for the anxiety subscale measured in the current study was .877 while the avoidance subscale was .830 (Dewitte and De Houwer 2008; Niedenthal et al. 2002; Fraley et al. 2000).

Couple Affect and Interaction Evaluation Device

Waldinger et al. (2004) introduced an observational coding system that did not require specific training to use. Naïve coders in Waldinger et al. (2004) rated couples using 18 different affect (e.g., humor, defensiveness) or interaction (e.g., “turned into each other”) categories on a 0–9 scale. Results from their original study showed strong correlations from highly trained coders utilizing SPAFF (Gottman et al., 1995).

In a departure from Waldinger, et al. (2004), participants rated the 5 min video segment of the couple on 18 different affective categories rather than rating each partner on each category over 30 s clips. However, the categories and range of responses used were similar to Waldinger et al. (2004). Each category was rated on a scale from 0 to 9 with high scores indicating high expressions of the category and low scores representing low intensity of the category. Depending on the category, high ratings may be positive in the case of positive affect while high ratings of negative affect would indicate higher amounts of negative affect or interactions. Twelve of the categories were determined to be examples of negative affect or interactions (e.g., defensiveness, criticism, contemptuous, etc.). Six of the categories were identified as positive affect or interactions (e.g., humorous, warm, caring).

Results

To minimize possible co-linearity and consistent with other studies (e.g., Dewitte and De Houwer 2008; Cooper et al. 2009), scores on the ECR-R scales were centered on their mean prior to inclusion in the analysis.

Prediction Task

The dependent measure for the prediction task was the overall trend of predictions participants generated for each couple. Trends were identified by the total score at the end of each prediction task. There were 28 predictions made for each couple observed. If the participant made a prediction that the couple conflict would escalate, a score of −1 was recorded. If the participant predicted that the couple conflict would de-escalate, they received 1 point for each prediction. A zero was coded for each prediction of “no change.” Scores were then totaled. If the accompanying total score was a negative number, it would indicate that the participant made more predictions of deterioration relative to predictions of de-escalation while positive scores would indicate the reverse. Scores closer to zero imply an equal balance of predictions.

To test the hypothesis that attachment dimensions would affect overall trends of participant predictions, a separate path model for each couple observed was fitted to the data. No relationships between attachment dimensions and prediction trends were found.

Evaluation Task

The dependent measures for the evaluation condition consisted of the participant’s total score of the 12 negative affect and interaction (NA) categories and a total score of the 6 positive affect and interaction (PA) categories recorded for each couple. High scores on the NA scale indicate that the participant perceived high intensity of negative affect and interactions after watching the couple’s video. High scores on the PA scale suggest that the participant perceived high intensity of positive affect and interactions in the video.

To test the hypothesis that attachment dimensions would affect perception of specific kinds of affect/interactions, a fully saturated path model utilizing maximum likelihood with attachment dimensions as predictors and PA and NA as response variables were performed. A path model was fitted for each couple observed by participants.

All paths for Couple A (nondistressed as determined by RDAS scores) and Couple B (marital distress was ambivalent as determined by RDAS scores) were non-significant. While none of the paths in the models for Couple A and Couple B were significant, it is interesting to note the directions of the paths in the Couple A and Couple B models. When observing the nondistressed couple, higher attachment anxiety was minimally related to higher perceived PA (b = .08) and lower NA (b = −.23). Higher attachment avoidance was associated with lower PA (b = −.15) and NA scores (b = −25). However, all the paths in the ambivalently stressed couple were identical to path directions found in Couple C.

The path model for Couple C (distressed couple as defined by RDAS scores) showed a significant path between attachment anxiety and NA (b = .329) indicating that as attachment anxiety increases, participants were more likely to perceive greater intensity of negative affect/interactions in the couple after holding attachment avoidance constant (see Fig. 1). Another path from attachment anxiety to PA was also significant (b = −.222), indicating that as attachment anxiety increased, perception of positive affect/interactions decreased after holding attachment avoidance constant (see Fig. 1). The non-significant paths would suggest a minimal association between increases in attachment avoidance and higher ratings of PA (b = .07) and lower ratings of NA (b = −.16).

Fig. 1
figure 1

Attachment dimensions and perception of a distressed couple. Bolded lines indicate significant paths. Standardized coefficients are italicized and in parentheses. * p < .05. ** p < .01

Full size image

Due to the nature of the fully saturated models, fit statistics were not available. The path between attachment avoidance and PA was removed as it was not significant in the model and previous research has not shown a consistent connection between attachment avoidance and the perception of positive affect. Removing this path also gained a degree of freedom and subsequent model showed a good fit (Chi square = .523, p = .469, CFI = 1.00, RSMEA = .000). The significant paths from the original model between attachment anxiety, PA, and NA remained significant in this model.

Discussion

Prior work has shown relative consistency in finding that attachment dimensions are associated with changes in the nature and timing of perceived affect in others. This study was the first known to use couple interactions as the stimuli for participants. This study also has provocative implications for clinical and family life education settings as these professionals partially rely on their perception of couples in making decisions.

Discussion of the results of the current study attempt to follow recommendations by Fraley et al. (2000) to use the ECR-R in dimension form rather than reducing dimensions into categories and subsequently losing meaningful variance in the process. The challenge of placing the current results in the literature lies in comparing dimension based findings with literature that use a categorical operationalization of attachment such as secure, preoccupied, etc.

Attachment anxiety as measured in this study is synonymous with Bowlby’s internal working model of self (Mikulincer and Shaver 2008). The model of self refers to the individual’s belief in their own value, as based in their perception of being valued by a primary other such as a parent or romantic partner. When individuals have a strong working model of self, they tend to be more confident in approaching the world. The inverse relationship would also be accurate. Specifically, as an individual is less confident in their value, confidence in approaching novel or stressful situations decrease. It would make theoretical sense that individuals higher in attachment anxiety are more watchful for “threats” in their environment, (Bowlby 1973; Simpson and Belsky 2008).

The findings from the current study suggest that as attachment anxiety increased, perceived intensity of negative relationship dynamics and/or affect also increased when observing a distressed couple. The nonsignificant path directions when observing the ambivalently distressed couple indicated a similar relationship. It could also be stated that negative affect would be perceived as less intense as attachment anxiety decreased. Silva et al. (2012) found that individuals with “anxious attachment” were distracted by facial expressions containing negative affect. “Distraction” in the Silva et al. (2012) study was operationalized as focusing longer on facial expression containing negative affect instead of identifying the target image which followed. Langer (2009) may argue that instead of being “distracted,” participants could be said to be “otherwise attracted” toward the negative affect and maintained their focus there.

In the distressed condition of the Niedenthal et al. (2002) study, participants identified as fearfully attached (high anxiety, high avoidance) were the last of the attachment style groups to see negative expressions disappear. Stated in another way, those subjects perceived negative affect longer than individuals categorized into other attachment styles. Consistent with the current study, Dykas and Cassidy (2011) summarized findings indicating that individuals higher in attachment anxiety are more likely to identify negative emotional content.

The relationship between attachment anxiety and the minimization of positive dynamics/affect also has some previous support. In the current study, participants that scored higher in attachment anxiety perceived positive emotions as less intense in a distressed couple, with non-significant trends in the same direction when observing the ambivalently distressed couple. Consistent with these results, Dewitte and De Houwer (2008) found that high attachment anxiety, when combined with high attachment avoidance, was associated with decreased attention of positive affect.

In the nondistressed condition of Niedenthal et al. (2002), individuals classified as fearfully attached perceived positive affect to disappear more quickly than other attachment styles. Couple A, the nondistressed couple, provides a possible corollary to Niedenthal et al.’s (2002) nondistressed condition. It should be reinforced that all the paths in the model for Couple A were not significant, and as such, implications of the findings of this model are very tenuous. These findings are somewhat encouraging in that as attachment avoidance increased, the intensity of perceived negative affect in the nondistressed couple decreased which is similar to previous work.

While the paths between attachment avoidance and affective categories for the ambivalent and distressed couples were not significant, it is instructive to note the direction of associations present. The results show that as attachment avoidance increased, the perceived intensity of negative affect drops while perceived intensity of positive affect is increased. The pattern of attachment avoidance and decreased attention, memory, or identification of negative affect is consistent with the theoretical and empirical literature. Suslow et al. (2010) found that attachment avoidance negatively correlated with correctly identifying sad faces. Silva et al. (2012) found that participants higher in attachment avoidance were able to successfully identify the target image in spite of negative affect distractors.

Implications

Bowlby believed that internal working models form part of the foundation of a person’s personality and is reminiscent of a trait description. While internal working models are adaptable, they can consistently impact how an individual may perceive and respond to the world around them. As such, the impact of internal working models on social information processing has profound implications for mental health professions such as licensed marriage and family therapists, clinical supervision, and family life educators as accurate perception is crucial to assessment, treatment, supervision, and education.

Therapy and Clinical Supervision

Historically, proponents of differing therapeutic models would refer to minimization or exacerbation of observed phenomenon as “projection,” “transference,” or “counter-transference.” However, Bowlby 1969, 1982 rejected these types of conceptualizations considering individuals’ responses as normative and based on how attachment experiences formed internal working models.

The distinction between normative responses versus defense mechanisms can have a powerful effect on the clinician/client relationship as well as the supervisor/supervisee relationship. Defense mechanisms could be perceived to be a deficit implying the client or supervise will need to be “fixed.” Normative responses to internal working models can be “understood” and are, by definition, adaptable. Clients and supervisees in this frame are given a nurturing environment and given permission to explore their patterns and increase awareness.

Framing a client or supervisee’s perception as a normative response to attachment experience ideally would lead to safer environment from which clients and supervisees can better understand their responses. While the therapist/client and supervisor/supervisee relationship would not meet Bowlby’s criteria of a primary attachment relationship, therapists and supervisors would do well to value the individuals they work with and be appropriately responsive to their needs.

Family Life Education

Accurate perception is a crucial component in the provision of educational and other non-clinical intervention programs. It is important for family life educators to understand their own attachment based tendencies as their assessments can determine the appropriateness of a couple for a particular program. Misidentification of couples may adversely affect group/class dynamics and the ability of the couple(s) to utilize the class experiences to its fullest. Additionally, couples that are clinically distressed are often not appropriate for many psychoeducational interventions (Markman and Rhoades 2012). Beyond screening for appropriateness, assessment can assist the educator in tailoring the program to meet the specific needs of the couple(s).

Limitations

This study found evidence for attachment anxiety (internal working model of self) influencing the evaluation of distressed couples and there are limitations that should be addressed in future research along these lines.

The gender composition of the study was heavily skewed toward females, over two-thirds of which had some therapy training. Unfortunately, the available graduate programs, agencies, and private practices that were solicited for participation had significantly more females present. While none of the previous research would indicate a strong gender effect within attachment dimensions, it would be stronger methodologically to be more balanced to rule out any gender effect that may be present.

The sample size in this study was small given the nature of the statistical analysis used. However, the sample herein was consistent with previous work cited (e.g., Maier et al., n = 57, Silva et al., n = 54, Dewitte & Houwer, n = 42) and represents a comparatively large sample for research that involved neurophysiological data collection. Larger sample sizes in the future will have the ability to fully test the relationships between attachment dimensions and evaluations of distressed, nondistressed, and ambiguous couples.

Another potential limitation of the study was the sole use of the RDAS to determine couple distress. Couples may rate themselves as distressed, or nondistressed, and behave differently when observed. Future research should include expert coding of the video to confirm the reported distress level of the couple. Clearer distinctions between nondistressed, ambivalent, and distressed couples prior to participants observing them would also help tease out effects of attachment dimensions and perception of couple affect and interactions.

Beyond simply having participants evaluate couple interactions, future research could “trigger” the attachment mechanisms via direct or subliminal stressors prior to showing a video to be evaluated (see Mulkincer’s work and others for examples). A possible trigger for an experiment with therapy implications could be to ask participants to imagine that they are going to be the couple’s therapist and that their evaluation they are about to perform will be the basis of therapy.

Conclusion

Bowlby’s observations and articulation of attachment theory continues to be highly influential in understanding the nature and power of relationships. This study sought to explore the impact of attachment anxiety and avoidance on observations of relationships beyond those the participant was involved in. The overall pattern from the current study would suggest that individuals high in attachment anxiety perceive less positivity and more negativity in distressed couples. Future work needs to continue to explore the influence of internal working models individuals use to navigate everyday life and in helping relationships that emphasize accuracy in perception.