Attachment to Peers and Perception of Attachment to Parents in Adults

The attachment behavior is created in the infancy and accompanies the person throughout life. Considering that the attachment style may change, this article investigates the impact of a lasting love relationship and its influence in adults’ perception of attachment to parents in the past, as well as adults’ attachment to peers in the present. Using the romantic relationship as a mediator, this study assesses the correlations between the current adult attachment style to peers, as evaluated by the PAAS instrument (Portuguese Adult Attachment Scale), and the adult perceptions of their past attachment to parents, measured by FMAQ (Father and Mother Attachment Questionnaire). This research used a convenience sample consisting of 180 participants, with age between 19 and 49 years old. The results demonstrate that there are correlations between the current attachment to peers and the individuals’ perception of attachment to father and mother. At the same time, however, the results show that there are differences in the correlation between the group with a lasting love relationship (G1/LLR) and the group without one (G2/WLLR). Therefore, it can be concluded from the results of this research that a lasting love relationship has the possibility to modify internal working models, as well as a person’s attachment style.


Introduction
According to the theory of attachment, attachment patterns start up and are set with the primary caregivers early in life.Attachment style formed in childhood tends to endure over time, has a survival function and tends to remain constant in adulthood (Bowlby, 1990;Fraley, 2002;Sroufe, 2005;Scharfe & Bartholomew, 1994;Waters, Merrick, Treboux, Crowell, & Albersheim, 2000).When Bowlby (1990) discusses the internalization of emotions and relationships, he argues that internal working models are a result of established relations and of the internalization of emotions and behaviors, which create a representation of ourselves and others.Bowlby (1993) also points out that the child expectations about the caregivers and how the caregivers will react are created in childhood and tend to remain unchanged through life.In his views, the behavior of adults can be explained in terms of their experiences in childhood.According to him, we create internal models about us as deserving, or not, of support, comfort and care; similarly, we create models of others, and also about the behavior, confidence and attitudes that can be expected from them.Although internal working models created in childhood tend to continue in adulthood, Bowlby also advances the argument that these models may change: Occasionally, however, some major change in environment or in organism occurs: we get married, have a baby, or receive promotion at work; or, less happily, someone close to us departs or dies, a limb is lost, or sight fails.At those times radical changes of model are called for (Bowlby, 1983: p. 82).
In this line of thought, Hazan and Shaver (1987) argued that internal models are dynamic representations that can be changed, developed or reorganized according to the different relational experiences.This argument was also confirmed by Moreira (2006), who demonstrated that the attachment style is related to the relationship that the person has.In this way, internal models are not rigid and have the ability to accommodate information and be determined by the behavior and characteristics of relational actors.The individuals end up adapting their representations of relationships according to the different relational partner, regardless of the expectations that they have (Cook, 2000).Other researchers also point out that the attachment style can be changed due to significant interactions that the person experiences (Bowlby, 1990;Hamilton, 2000;Lewis, Feiring, & Rosenthal, 2000;Moreira, 2006;Weinfield, Sroufe, & Egeland, 2000).
When extending the attachment theory in childhood to later adult attachment to the pair-bond, Hazan and Shaver (1987) theorized that the romantic relationship could be an attachment relationship.Also, Brennan and Shaver (1995), Collins (1996) and Simpson (1990) demonstrated that there is a connection between the attachment styles in childhood and the ways in which the adult thinks and behaves in their romantic relationships.
According to Collins and Read (1990), parental relationships that were remembered as warm and accepting were associated with current perceptions of others as being available to give support.Adults who reported secure relationships in childhood with their mother, father and family members, also reported not to feel anxious about commitments or intentions of their partners.They also noted that adults who had good relations with their mothers displayed an in-crease in self-esteem, benevolence, awareness of others, and saw others as help, confidence and assertiveness, showing the mother's importance during childhood for the relationships in adulthood, whether such relationships are romantic or not.
Similarly, Bowlby (1988), Crowell, Treboux andWaters, (2002), andMain (1999) suggested that people have the ability to change their attachment pattern provided that there is a secure and reliable connection, which may be a romantic relationship.However, more research is necessary to understand changes in attachment style and how these modifications work (Fraley, 2010).
While most studies on adult attachment elaborate on how the interactions with parents during childhood influence adult relationships (Auslander, Short, Succop, & Rosenthal, 2009;Ávila, Cabral, & Matos, 2011;Feeney, 2004;Feeney & Noller, 1990;Fraley, 2010), this research proposes that a lasting love romantic relationship has the ability to change the correlation between the adult attachment style to peers and the adult perceptions of their past attachment to parents.
In our study we examine the relation between the current attachment to peers and the perception of attachment to parents, in a retrospective version, using the romantic relationship as a mediator.
Beyond aiming to achieve a greater understanding about attachment dynamics, we seek to investigate whether the models of attachment are maintained or not, and whether the current attachment to peers correlates with the perception of attachment to parents in the past.For this purpose, we used two instruments.
For assessing the adult attachment to peers in the present, we used the Portuguese Adult Attachment Scale (PAAS) from Canavarro (1999).For measuring the perception that the person has about the attachment he had with their parents in the past, we used the Father and Mother Attachment Questionnaire (FMAQ) from Matos and Costa (2001).
With these two instruments, this study assesses the impact of a lasting love relationship on adult attachment by evaluating: 1) the person's perception of attachment to their peers in the present (PAAS) and 2) their perception of attachment to parents in the past (FMAQ).For the purposes of this research, a lasting love relationship is defined as a romantic relationship that lasts for 24 months or more.The use of this benchmark of 24 months was based on Cassidy (2001) and Hazan and Shaver (1994), who stated the need of this length of time for establishing an attachment relationship.Therefore, the subjects of this study are divided into two groups: those who are in a lasting love relationship for 24 months or more (G1/LLR), and those without a love relationship, or in a love relationship for less than 24 months (G2/WLLR).
This study aims to add to the literature on attachment theory because it explores the possible impact of a lasting romantic relationship on the relations between a person's present attachment style to peers and the perception of the past attachment to their parents.We consider that this study is innovative as our research has not encountered empirical studies in Portuguese language that make a correlation between the past perceptions of attachment to parents and the present attachment to peers with those two instruments.
The objectives of this study were: 1) to increase the understanding of the attachment phenomenon; 2) to investigate the relations between the early models of attachment and current models in adults; and 3) to explore the impact of a lasting love relationship as a potential factor of change in the internal models of attachment.
With these objectives in mind we posed the following hypothesis: H1.Since the attachment models tend to continue throughout life (Fraley, 2002;Scharfe & Bartholomew, 1994;Sroufe, 2005;Waters et al., 2000), the current attachment between adults should be similar to the perception that the person has of their attachment to their parents in the past.Therefore, we raise the following hypotheses: 1) The quality of the emotional bond (QEB) from FMAQ has a positive association with the comfort with proximity (PAAS) and trust in others (PAAS) and a negative association with anxiety (PAAS).
2) The inhibition of exploration and individuality (IEI) from FMAQ presents a positive association with anxiety (PAAS) and a negative association with the comfort with proximity (PAAS) and the trust in others (PAAS).
3) The separation anxiety (SA) from FMAQ presents a positive association with anxiety (from PAAS) and a negative association with the comfort with proximity and trust in others (both from PAAS).H2.Because the presence of a love relationship, according to some models, can modify the internal models of attachment (Brehm et al., 2002 andCrowell et al., 2002), it is expected that the correlations between the dimensions from PAAS and FMAQ are different between individuals who are in a stable and lasting love relationship for equal to or more than 24 months, when compared to individuals with no love relationship or who are in a love relationship for less than 24 months.

Method
Taking into account these hypotheses, we used the correlational method to verify the relation between the dimensions of the two instruments used (PAAS and FMAQ).

Participants
Participants were included in this study if they met the following criteria: 1) they must be psychology students of the University Lusófona of Humanities and Technologies in Lisbon, Portugal, since this research project was approved by the ethics committee of the Psychology Department of that university; 2) the participants age must be between 18 and 60 years old; 3) they must give their voluntary agreement to participate; 4) they must completely fill both instruments and the demographic data sheet.The exclusion criteria are the students that do not belong to a psychology class.
With the criteria in place, we conducted this research in 2012, with a convenience sample consisting of university students from various years of psychology course of the University Lusófona of Humanities and Technologies in Lisbon, Portugal.
After asking students for their agreement to this study, we passed the questionnaires.From 182 students that filled out all the questionnaires, we removed two of them because they did not have any recollection of their parents, who died when the participants were young.We reached around 3/4 of the total of psychology students of that university, making a good representative sample of them.
From the remaining 180 participants, their age were between 19 and 49 years, with M = 27.82 and SD = 8.205.From this sample of 180 participants, 140 were females (77.8%) and 40 were males (22.2%).Out of the 180 participants, 53 of them (29.4%) were in a love relationship for 24 months or more, and 127 (70.5%) were not in a romantic relationship or were in a love relationship for less than 24 months.

The Informed Consent Form
The informed consent form identifies the name of the investigator and his affiliation, the name of the project and what the study is about, the research purpose, the intervention procedures, duration, risks, benefits, and confidentiality.It requests the voluntary participation of the subjects, who must sign the consent form if they want participate.

Socio-Demographic Data Sheet
The socio-demographic data sheet is a questionnaire easy to fill out and collects information on the subject, such as gender, age, marital status, whether he was in a love relationship and for how long; as well as other information.

The Portuguese Adult Attachment Scale
The adult attachment scale has its origins in Collins and Read (1990) who, using the questionnaires of Hazan and Shaver (1987), turned such questionnaires into multiple items that could be independently assessed through a Likert scale, resulting in three dimensions.These dimensions were produced by exploratory factor analysis (PAF) with VARIMAX rotation (Canavarro, Dias, & Lima, 2006).These multiple items were transformed into 18 statements to be assessed through a Likert scale, and were adapted to Portugal by Canavarro (1999).This Portuguese Adult Attachment Scale (PAAS) assesses the relations with peers, although it is also generalizable to relations with other significant figures.This instrument allows us to characterize the perception of individuals about how they feel in their relationships, identifying three dimensions: 1) anxiety, 2) comfort with proximity and 3) trust in others.
This scale consists of 18 items, assessed on a Likert 5-point scale, ranging from 1 to 5 being the possible answers: 1) not at all characteristic of me; 2) somewhat characteristic of me; 3) characteristic of me; 4) very characteristic of me and 5) extremely characteristic of me.Each dimension of the scale is evaluated by 6 items.The anxiety dimension refers to the degree of anxiety experienced by the individual, related to interpersonal issues of fear of abandonment or not to be well-liked.The comfort with proximity dimension assesses the degree to which the individual feels comfortable with proximity and intimacy.Lastly, the factor trust in others concerns the degree of confidence that individuals have in others, as well as the availability of others when individuals feel like they need them (Canavarro et al., 2006).
According to Canavarro et al. (2006), the Cronbach's alpha for the three dimensions was good, meaning that the scale was reliable for use.We decided to test the internal consistency of the dimensions (Cronbach's alpha) in our sample and we obtained the following values: anxiety = .81,comfort with proximity = .67and trust in others = .75.
We should note that we removed the item 6 from the dimension comfort with proximity, as well as items 2 and 5 from the dimension trust in others, because they significantly lowered the values of internal consistency.

Father and Mother Attachment Questionnaire
By Matos and Costa (2001), the Father and Mother Attachment Questionnaire (FMAQ) was used in its retrospective version to evaluate the perception of representations that the subject has about the relationship he had with his parental figures in the past.This parental attachment questionnaire measures three dimensions: 1) the quality of the emotional bond (QEB); 2) the inhibition of exploration and individuality (IEI); and 3) the separation anxiety (SA).These three dimensions were identified through exploratory factor analysis (PAF) with VARIMAX rotation, both for the father and the mother data banks.
This questionnaire allows us to understand the perception of individuals on their relationship and behaviors with their parents in their relations with them in the past.It contains 30 questions, ten for each dimension, to be answered retrospectively on both the father and the mother, through a Likert scale of 6 points that varies according to the following possibilities: 1) I totally disagree, 2) I disagree, 3) I disagree moderately, 4) I agree moderately 5) I agree, 6) I totally agree.These values are organized into two columns, one for the father and one for the mother, each of which is to be answered independently to the father and to the mother (Matos & Gouveia, 2011).
The quality of the emotional bond dimension (QEB) is obtained through items 2, 5, 8, 11, 14, 17, 20, 23, 27 and 30.It evaluates the importance of parental figure as an attachment figure, perceived by the individual as unique and essential to his development, to whom he turns to in difficult situations and with whom he projects a lasting relationship (Matos & Gouveia, 2011).
As for the inhibition of exploration and individuality dimension (IEI), it is measured by items 1, 4, 7, 10, 13, 16, 19, 22, 25 and 28.This dimension concerns the perceptions of the individual with regard to restrictions on the expression of his individuality.These restrictive perceptions can be externalized as difficulties in expressing their views or differing views from the parental figure, or the lack of support for exploratory initiatives, or even the unwanted interference in matters which the individual consider as personal (Matos & Gouveia, 2011).
The dimension of the separation anxiety (SA) is obtained through the items 3, 6, 9, 12, 15, 18, 21, 24, 26 and 29.It provides access to the individual's perception regarding anxiety experiences and fear of separation from the parental attachment figure, which may be revealing of a relationship of dependency (Matos & Gouveia, 2011).
With regard to the internal consistency of the dimensions, in our study we obtained the following values of Cronbach's alpha for each dimension: QEB father = .95,QEB mother = .93,IEI father = .89,IEI mother = .88,SA father = .87and SA mother = .84.We should note that we removed item 5 from QEB father because it significantly lowered the internal consistency of the questionnaire.

Procedure
After obtaining the necessary approvals and fulfilling all legal and ethical procedures, the questionnaires were distributed in several class sections of the various years of the course of Psychology, where the collaboration of students in answering the questionnaires was requested and the consent form was submitted.Furthermore, an explanation was given about the purpose of the study and instructions on filling in the questionnaires.The collected data were subsequently analyzed into the statistical analysis program SPSS 20, where we calculated: 1) means and standard deviations; and 2) the correlation between the dimensions of the two applied scales (PAAS and the FMAQ): i) for the total sample, and ii) for each of the two groups -one group with individuals who are in a lasting love relationship for 24 months or more (G1/LLR), and another group of individuals without a love relationship or in a love relationship for less than 24 months (G2/WLLR).
Since we propose to explore the impact of a lasting love relationship as a factor of change in the internal models of attachment, we sought to answer our hypotheses by making correlations between the three dimensions of PAAS and the three dimensions of the FMAQ.
As some models suggest that the presence of a love relationship can modify the internal models of attachment (Brehm et al., 2002 andCrowell et al., 2002), it is expected that the correlations between the dimensions from PAAS and FMAQ are different between individuals who are in a stable and lasting love relationship for equal to or more than 24 months, when compared to individuals with no love relationship or who are in a love relationship for less than 24 months.

Results
The results are presented for three situations: 1) total sample (TS), 2) subjects in a lasting love relationship for 24 months or more (G1/LLR), 3) subjects without a love relationship or in a love relationship for less than 24 months (G2/WLLR).

Correlations between the Dimensions of the Two Scales
In Table 1, we have the correlations between the dimensions of the two scales for the three situations: 1) the total sample (TS, n = 180), 2) the group of individuals in a lasting love relationship (G1/LLR, n = 53), and 3) the group of individuals without a lasting love relationship (G2/WLLR, n = 127).

Correlations between PAAS Scale and FMAQ Scale for the Total Sample
Between anxiety (PAAS) and FMAQ dimensions, we have a positive correlation with IEI father/mother: for the father (r = .309,p = .000)and for the mother (r = .332,p = .000);similarly, we have a positive correlation with SA father/mother: for the father (r = .244,p = .001)and for the mother (r = .304,p = .000).No correlation was found between anxiety (PAAS) and QEB.
As for the comparison between comfort with proximity (PAAS) and FMAQ dimensions, we have a negative correlation with IEI father/mother: for the father (r = −.360,p = .000)and for the mother (r = −.321,p = .000).At the same time, we have a positive correlation between comfort with proximity (PAAS) and QEB father/mother: for the father (r = .192,p = .010)and for the mother (r = .222,p = .003).We found no correlation between comfort with proximity (PAAS) and SA.
Comparing trust in others (PAAS) and FMAQ dimensions, we have a negative correlation with IEI father/mother: for the father (r = −.371,p = .000)and for the mother (r = −.447,p = .000);on the other hand, we have a positive correlation with QEB father/mother: for the father (r = .189,p = .011)and for the mother (r = .192,p = .010).No correlation was found between trust in others (PAAS) and SA dimension in FMAQ.

Correlations between the PAAS Scale and the FMAQ Scale in the Group of Subjects in a Lasting Love Relationship for 24 Months or More (G1/LLR)
For individuals belonging to G1/LLR, the data show that there is a significant positive correlation between the anxiety dimension in (PAAS) and SA father/mother dimension (FMAQ): for the father (r = .363,p = .008)and for the mother (r = .377,p = .005).No correlations were found between anxiety (PAAS) and QEB, neither IEI.
Between comfort with proximity (PAAS) and FMAQ dimensions, we have a negative correlation with IEI father/mother: for the father (r = −.444,p = .001)and for the mother (r = −.343,p = .012).On the other hand, we have a positive correlation between comfort with proximity (PAAS) and QEB for the mother (r = .528,p = .000),but no correlation with QEB for the father.No correlations were found between comfort with proximity (PAAS) and SA dimension in FMAQ.
In relation to the comparison between trust in others (PAAS) and FMAQ, we have a negative correlation with IEI father/mother: for the father (r = −.381,p = .005)and for the mother (r = −.402,p = .003).Also, we have a positive correlation between trust in others (PAAS) with QEB mother (r = .338,p = .013),but none for the QEB father.We found no correlation between trust in others (PAAS) and SA dimension in FMAQ.

Correlations between the PAAS and FMAQ Scales in the Group of Subjects Not in a Love Relationship or in a Love Relationship for less than 24 Months (G2/WLLR)
In G2/WLLR, when comparing anxiety (PAAS) with FMAQ, we found a positive correlation with SA father/mother: for the father (r = .192,p = .031)and for the mother (r = .266,p = .002).Likewise, we have a positive correlation between anxiety (PAAS) with IEI father/mother, for the father (r = 323, p = .000)and for the mother (r = 348, p = .000).No correlation was found between anxiety (PAAS) and QEB dimension in FMAQ.
Between the dimensions of comfort with proximity (PAAS) and FMAQ, we have a negative correlation with IEI father/mother: for the father (r = −.318,p = .000)and for the mother (r = −.305,p = .000).We found no correlation between comfort with proximity (PAAS) and SA, neither QEB.
Between the trust in others (PAAS) and FMAQ dimensions, we have a negative correlation with IEI father/mother: for the father (r = −.351,p = .000)and for the mother (r = −.457,p = .000).At the same time, we have a positive correlation between trust in others (PAAS) with QEB father (r = .179,p = .043),but none with the mother.No correlation between trust in others (PAAS) and SA in FMAQ was found.
Comparing the Three Situations: Total Sample, the Group of Subjects in a Love Relationship for 24 Months or More, and the Group of Subjects Not in a Love Relationship or in a Love Relationship for less than 24 Months When we consider the three dimensions in the total sample, we found correlations among almost all the dimensions of the scales, which confirms our hypo-thesis that the current attachment is similar to the perception that an adult has of their attachment to their parents in the past.However, as we encountered exceptions in the correlations between anxiety and QEB, between comfort with proximity and SA, and between trust with others and SA, all of which do not have any correlations, we conclude that our hypothesis is partially supported.
We also observe that we have correlations in the same dimensions for the father and for the mother.This finding means that there is a correlation between the dimensions of scales as we have hypothesized and that there are not any differences between the father and the mother in these dimensions.
Concerning the comparison between anxiety (PAAS) with FMAQ dimensions, we point out that the correlation with IEI father/mother exists for the total sample (TS) and for the group without a lasting love relationship (G2/WLLR), but not for the group with a lasting love relationship (G1/LLR).This means that a lasting love relationship can change the correlations and occasionally the attachment that the person has.However, this finding only occurs between anxiety and IEI father/mother dimensions and between the trust of others and QEB father, where G2/WLLR has a correlation, but the G1/LLR does not.
In addition, we note that there are also differences between QEB mother and comfort with proximity and trust in others between the groups G1/LLR and G2/WLLR, because there are no correlations in the G2/WLLR, but we find them in the G1/LLR.
When we analyze our hypothesis against our results, we find that: 1) in the total sample, the QEB correlates positively with comfort with proximity and with trust in others, but not with anxiety.This finding confirms partially our hypothesis.2) The IEI has a positive correlation with anxiety and negative correlation with comfort with proximity and with trust in others, which corroborates our hypothesis.3) The SA has a positive correlation only with anxiety but not with any other dimension, which partially confirms our hypothesis.4) Lastly, one of our hypothesis presumed that the dimensions' correlations would be different between G2/WLLR and G1/LLR, but we only encounter such differences in the correlations between IEI father/mother and anxiety, between QEB father and trust in others, and between QEB mother and comfort with proximity, as well as between QEB mother and trust in others.Therefore, this finding only partially confirms our hypothesis.From the data collected, we can conclude that some dimensions do not have any correlation, while others do.This result partially corroborates our hypothesis.Also, we observe that we have significant differences between G1/LLR and G2/WLLR.However, such differences do not occur in all situations, what partially support our hypothesis.

Discussion
Although not all data are in accordance with our hypothesis, we remark that there is a correlation between the scales dimensions that are similar (SA in FMAQ and anxiety in PAAS).That is, there is a relation between the data of the two scales when we measure the construct "anxiety".
These results point to the existence of continuity between the attachment models created in the relationship with primary caregivers (assuming that in the general population these are the father and the mother) and the attachment models in adult.In other words, there appears to be evidence that the attachment style tends to remain constant throughout life, as proposed by Bowlby (1990) and confirmed in previous studies (Hamilton, 2000;Hazan & Shaver, 1987;Waters et al., 2000).In addition, although not all expected correlations were corroborated, they are almost always present in relation to similar categories (e.g.separation anxiety FMAQ and anxiety PAAS).Moreover, all dimensions of the scales show significant correlations, as we expected.In fact, the data seem to indicate the continuity of attachment models created in childhood, as these authors argued.
Furthermore, it should be noted that these are significant correlations and there is no overlap among dimensions.These data show that, in a large number of individuals in this sample, a value of a dimension of the PAAS (current attachment) can predict a value of the corresponding dimension of FMAQ (perception of attachment to parents in the past) or even the reverse.Therefore, we can conclude that the attachment model remained stable for the total sample in general.
Nevertheless, when comparing the correlations between the two groups, the one with a long lasting relationship and the other one without a relationship or with a relationship with less them 24 months, we found that there are significant differences in the correlations between these two groups (G1/LLR and G2/WLLR).More specifically, there are positive correlations in G2/WLLR between anxiety and IEI father/mother, as well as between trust in others and QEB father, which do not exist in G1/LLR.That is, the correlation between anxiety and IEI father/mother in G2/WLLR ceases to exist in G1/LLR, when individuals are in a love relationship for 24 months or more.Such differences between these two groups demonstrate that the attachment models can be changed by a lasting love relationship.Therefore, this finding supports our hypothesis that a lasting love relationship can modify the internal attachment models, what is in accord with Bowlby (1983Bowlby ( , 1988)), and also with some other studies that argued that the attachment models can change with the events of life (Lewis, Feiring, &Rosenthal, 2000 andWeinfield et al., 2000).
Similarly, there are positive correlations in G1/LLR between QEB mother and comfort with proximity, as well as between QEB mother and trust in others in PAAS, but there is not a similar correlation in G2/WLLR.Such positive correlations exist only in G1/LLR, and not in G2/WLLR, which means that the mother assume a relevant role when the individuals are in a long last relationship.We can also notice that QEB father has a positive correlation with trust in others only in G2/WLLR, and ceases to exist in G1/LLR, which indicates that the father has an important role in trust in this group but not in the other group.This brings us to the fact that the correlations between the dimensions of the two instruments are different from one group to another.We can thus conclude that there are differences between the two groups, confirming our hypothesis that a lasting love relationship can modify the internal attachment models.
Those individuals not in a love relationship for 24 months or more (G2/WLLR) have, above all, positive correlations in "negative" dimensions (anxiety/IEI), both for the mother and the father.In other words, the negative models appear to continue in G2/WLLR, which does not occur in the G1/LLR group.It seems that individuals in G2/WLLR maintain the same attachment style that they probably had in their childhood.
In the group of individuals in a love relationship for 24 months or more (G1/LLR), there are positive correlations among the "positive" dimensions in relation to the mother, but not to the father.We can therefore infer that the mother gains importance in the presence of a lasting love relationship.We can also infer that when there is a perception of a good relationship with the mother, these individuals tend to keep this model in adults when in G1/LLR.This is not true in the other group.
Thus, there seems to be a continuation of the models in G2/WLLR, in the "negative" sense (anxiety/IEI -insecure models tend to remain stable), and a continuation of the models in G1/LLR and G2/WLLR, in the "positive" sense (Comfort/IEI and Trust/IEI -secure models tend to remain stable).The data also demonstrate that the "negative" models may be modified in the presence of a lasting love relationship.From these findings, we can observe the stability of the attachment style as found by Hamilton (2000), and Waters et al. (2000), but also a change in it, as Crowell et al. (2002) and Moreira (2006) remarked.We can therefore state that the discontinuance of the attachment is as legitimate as its continuity, and that such change can occur in the presence of a lasting love relationship in adulthood.
Furthermore, the findings of G1/LLR and G2/WLLR corroborate the argument that people have the ability to change their attachment representation provided that there is a secure and trustworthy environment, as suggested by Bowlby (1988), Crowell et al. (2002), andMain (1999).
From the analysis of the G1/LLR and G2/WLLR results, the attachment representations are interdependent and are also permeable to different relational experiences that the person encounters.This means that the global attachment would be representative of the average of the various relationships that the individual has.In other words, the aspects of a specific relationship may influence the attachment style (Moreira, 2006).
Similarly, the results of G1/LLR and G2/WLLR seem to support the hypothesis that a lasting love relationship can behave as a compensatory attachment for the possible failure of the main attachment figures (parents), allowing the person to change their attachment style.

Conclusion
After considering all of the above, it seems clear that attachment style may change in accordance with a lasting love relationship.Our study shows the sig-nificant differences between the two analyzed groups and demonstrates the importance of a lasting love relationship, which may influence or change the type of attachment that the individual has.
Our study also draws attention to the differences found in the relations between the perception of attachment to parents in the past and the current attachment in the adult, taking into account the presence of a lasting love relationship.In other words, the data seem to indicate that the presence of an important relationship, such as a lasting love relationship, can modify the dynamic and global internal model and, consequently, influence how the individual sees and experiences the world around him.These data also suggest that the presence of a lasting love relationship can change the internal models that the person has in relation to their parents, and also possibly to other relationships.This research also remarks that the specific internal model of a relationship, which can be different from the initial models, can change the global attachment models.That is, the global internal model will result from the relations among various specific models and will depend in part on the relationship type the pair bond has.
Moreover, this study shows that a lasting love relationship can modify the internal models and thus can change adult behavior and relationships.However, we also realize that such love relationship is the result of the attachment in the past with his attachment figures.We therefore conclude that there is a bidirectional relation, where a lasting love relationship affects the perception of attachment to parents in the past and, at the same, is also a result from the attachment to parents in the past.
According to the results obtained in this study, we can also conclude that there are correlations between the current attachment to peers and the perception of attachment to parents, which indicate a continuity of attachment.This continuity shows the importance of attachment and relationships early in life as influencing factors in the infant attachment and behaviors, but also in the attachment style and relationships later in adulthood.The results also show us that a lasting love relationship ultimately modifies the perception the individual has about the attachment he had to his parents in the past.
Therefore, these data seem important and with practical implications, particularly in the study of marital relations and psychotherapy.We also emphasize the idea of bidirectional influence of the various models: the individual models appear to influence the models of a given relationship, and the other way around may also occur.
Hopefully, this study will enrich the understanding of the attachment behavior and of the relationships that an individual has established, and will help to comprehend the type of attachment he had in the past and how all is interconnected, which turns out to be a significant contribution to the field, especially in clinical terms for assessing relationships' problems.
In addition, we also consider this study a relevant one because it provides significant findings for clinical assessment of relationship's problems to promote relational well-being.Based on the data we have discussed, there seems to be no doubt that the type of attachment may change, and this is also one of the goals of therapy-to help change the internal models that we have about ourselves and others.Such change occurs not only in therapy, but it can occur whenever there are conditions and events that may cause a person to reformulate their cognitive structure and internal models, as in a lasting love relationship.
It should be noted that, in this study, we used a retrospective questionnaire to assess the individual's perception of attachment to parents.The specific internal models of these relationships-with the father and the mother-may also have changed over time in the presence of other relationships or relevant secondary figures.In other words, this study only allows us to correlate the current attachment model and the current perception of attachment to parents in the past.
Modifications of the specific internal models of the main attachment figures can be accomplished through longitudinal studies.
Lastly, we should note that this study has some limitations since the sample (a convenient one) was restricted to a particular group (psychology students) in a particular university (Lusófona University).For further research, the sample should include a more balanced gender representation, students from other courses and from other universities.Moreover, we should remark that we used the correlational method in order to analyze the results, but the results can also be analyzed with a canonical correlation analysis or other statistics methods.

Table 1 .
Correlations between the dimensions of QVPM and of PAAS.