Adult Attachment Security and Automatic Relationship Attitudes: A Social Cognitive Measure of Working Models of Attachment Introduction Working models or mental representations of attachment experiences are often proposed to be the mechanism explaining how parent-child attachment continues to influence development beyond infancy and influences functioning in other close relationships including peer relationships (see Belsky & Cassidy, 1994, for discussion) and adult relationships (Baldwin et al., 1993; Collins, 1996; Feeney & Noller, 1996). However, most investigations of working models in infants and adults have inferred the content and organization of attachment relationship representations from the coherence of responses to open-ended projective measures (Bretherton et al., 1990; Mian et al., 1985) and in some cases from self reports on various scales (Feeney & Noller, 1996). Though it is important to identify differences in the coherence, openness, and valence of working models, it is also important to identify individual differences in the processing of attachment relevant information (e.g. influences on encoding, accessibility, interpretation, storage, and retrieval processes). Greenwald, McGhee, and Schwarts (1998) have developed a cognitive measure of attitudes regarding social categories, the Implicit Association Test (IAT). Using a PDP/Exemplar model of mental representations, the results of the IAT have been described by Greenwald et al. (1998) as the result of associations formed between categories of information (e.g. mom, dad, book, board) and attribute dimensions (i.e. pleasant vs. unpleasant) that are automatically activated when the target category is encountered. The automatic nature of category activation makes this procedure useful for identifying attitudes toward a target category that an individual may not be willing to self-report due to self-presentation demands. In the last few years IAT tests have been developed by at least 40 separate research groups world wide to test implicit attitudes toward such categories as race/ethnicity, sex/gender, political affiliation, self, and others (e.g. romantic partner and mother). The present study tests the hypothesis that information congruent with an individual’s working models of attachment relationships will be automatically accessed and responded to faster than incongruent information. More specifically, having a secure adult attachment style and therefore having working models of attachment that contain generally positive information about relationship experiences should facilitate the processing of positive relationship information. Having an insecure adult attachment style and therefore working models containing generally negative information about relationship experiences should facilitate the processing of negative relationship information. Method To test this hypothesis, 40 undergraduate (36 female and 4 male) introductory psychology students completed two laboratory sessions two weeks apart. As a measure of attachment security, once during each session participants completed the 4 item Relationship Questionnaire (RQ; Bartholomew & Horowitz, 1991) which provides self-report categories of attachment style (secure, dismissing, preoccupied, and fearful). Secure participants were identified as individuals who reported a secure attachment style at time 1 and time 2. Insecure participants were identifies as individuals who reported an insecure attachment style at either time 1 or time 2. Jeffery E. Aspelmeier Radford University As a measure of automatic accessibility of working model congruent relationship information, participants completed a modified version of the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998). This procedure asks participants to complete a series of tasks (see Figure 1) where they categorize words, presented one at a time on a computer screen, from two word lists (a relationship figures vs. Inanimate objects list and a pleasant vs. unpleasant list) by pressing one of two response keys (Left or Right). In the two critical trials (tasks 3 and 5) words from these two lists are alternately presented one at a time and each response key represents two categories (e.g. Left = relationship fig. and pleasant, Right = inanimate object and unpleasant). Task 5 represents a reversal of the combination a participant had in task 4, such that one hand will represent relationship figure and pleasant (RP) once and relationship figure and unpleasant (RU) once. The order and hand presentation was randomized across subjects. Facilitation is seen in that response latencies to combinations for one task are shorter than those seen in the alternate task (See Figure 2 for an example). A difference scores was constructed by subtracting the average log transformed response latencies for the RP task from the average log transformed response latencies for the RU task (RU-RP), such that a greater score indicates that the RP combination was responded to faster than the RU combination. It was expected that secure participants would have larger difference scores compared to insecure participants. Task 1 2 3 Task Description Target Associated Concept Attribute Discrimination Discrimination Results & Discussion Department of Psychology Abstract The present study investigates the association between attachment security and assessing working models of close relationships using a cognitive measure of implicit attitudes. In an ongoing study, 40 participants, completed Bartholomew’s (1990) Relationship Questionnaire (RQ) and an adapted version of Greenwald, McGhee, & Schwarts’ (1998) Implicit Association Test (IAT). The IAT asks participants to categorize words from separate word lists (list 1 : relationship figures vs. inanimate objects; list 2 : pleasant vs. unpleasant words). The results show that secure participants had faster reactions times when relationship figures and pleasant words were responded to using the same hand in comparison to trials where relationship figures and unpleasant words were responded to with the same hand. Implications for using the IAT as a measure of attachment security are discussed. To ensure that differences in response times were not the result of the order that combinations were presented in, associations between combination order (RP first or RU first) and IAT difference scores were calculated. Results of an independent sample t-test were not significant, t(38) = .21, ns. Showing that, whether participants received the relationship/pleasant combination first or the relationship/unpleasant combination first the IAT response latencies were not significantly affected. A fixed factor ANCOVA with sex included as a blocking variable was calculated to test the hypothesis that secure participants would have higher difference scores than insecure participants. A significant difference was found between the IAT difference scores of secure and insecure participants, F (1, 37) = 4.15, p < .05. Table 1 displays the untransformed mean response latencies for the IAT difference scores for secure and insecure participants. Overall secure participants responded significantly faster to the RP combination and significantly slower on the RU combination (as indicated by the larger difference score) than did insecure participants. These results suggest that mental representations of close relationships in adulthood influence the encoding and subsequent processing of relevant stimuli where information congruent with ones representations has an advantage over incongruent information. Figure 2 Example of predicted results for the IAT. Separate for self-reported security and insecurity. “A” and “5" represent response keys on a standard keyboard. 4 5 Initial Combined Task Reverse Target Concept Discrim. Reverse Combined Task Task Instructions * RF vs. Inanimate * Object * Pleasant vs. Unpleasant * *RF * Pleasant Object * Unpleasant * RF * vs. *Inanimate Object RF* * Pleasant * Object Unpleasant * Sample Stimuli * Lover Battery * * Fiancé * Partner Shelter * * Family * Mother Dough * Brick * * Soulmate - * Lucky * Honor Poison * Grief * * Gift Disaster * * Happy Hatred * Assault * * Miracle - * Friend * Pleasure Square * Evil * Boards * * Laughter * Sweetheart Mess * * Kinfolk Cotton * - Lover * * Cabinet Fiancé * Partner * * Label Family * Mother * * Shower * Creamer Soulmate* - * Peace Parent * Filth * * Number * Rainbow Darling * Accident * * Folder Vomit * Beloved * - Figure 1. Schematic description of the implicit attitudes test (IAT). Numbered columns correspond to each of the five tasks participants are asked to complete. Task 1 and 2 introduce a pair of target concepts (RF = relationship figure) and a pair of attribute dimensions. Stimulus words presented one at a time are categorized with a left or right hand response as indicated by the asterisks in rows four and five. The target concept and category lists are intermixed in task 3 and task 5 (after training on the reversal of target concept responses in task 4). Relationship Unpleasant (RU) Secure = Slow Response Latencies Insecure = Fast Response Latencies Relationship Pleasant (RP) Secure = Fast Response Latencies Insecure = Slow Response Latencies Table 1 Mean Untransformed IAT Difference Scores for Secure and Insecure Participants _ _ Attachment Classification Secure Insecure F R2 _ IAT Difference Score _ 448.17 (196.51) 309.70 (276.38) 4.15* .10 _ _ Note: * = p < .05. Standard Deviations appear in parentheses below mean values. Conclusions That this procedure is resistant to self-presentation strategies has significant implications for assessing working models of attachment relationships and possibly attachment styles. For example, dismissing (avoidant) attachment is associated with minimizing the importance of relationships and giving unrealistically positive reports of relationship in general and yet giving examples of very negative specific relationship experiences when probed during the Adult Attachment Interview (e.g. Main et al., 1985). Such presentation strategies threaten the validity of self report measures of Adult Attachment styles which do not probe for defensiveness. The IAT would be useful for identifying such defensive strategies. More generally, The IAT has the potential to serve as an effective assessment of attachment security and with more refinement may be useful as a classification tool. References Baldwin, M. W., Fehr, B., Keedian, E., Seidel, M., and Thomson, D. W. (1993). An exploration of the relational schemata underlying attachment styles: Self-report and lexical decision approaches. Personality and Social Psychology Bulletin, 19(6), 746-754. Bartholomew, K. and Horowitz, L. M. (1991), Attachment styles among young adults: A test of a fourcategory model. Journal of Personality and Social Psychology, 61(2), 226-244. Belsky, J., and Cassidy, J., (1994). Attachment: Theory and evidence, In M. Rutter, D. Hay, and S. Baron-Cohen (Eds.), Developmental principles and clinical issues in psychology and psychiatry. Oxford: Blackwell. Bretherton, I., Ridgeway, D., and Cassidy, J. (1990). Assessing internal working models of the attachment relationship: An attachment story completion task for 3-yearolds. In M. T. Greenburg, D. Cichetti, and E. M. Cummings (Eds.), Attachment in the Preschool Years. Collins, N. L. (1996). Working models of attachment: Implications for explanation, emotion and behavior. Journal of Personality and Social Psychology, 71(4), 810-832. Feeny, J., and Noller, P. (1996) Adult Attachment. Thousand Oaks, CA: Sage. Greenwald, McGhee, and Schwarz (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74(6), 1464-1480. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I. Bretherton & E. Waters (Eds.), Growing points of attachment theory and research, Monographs for the Society for Research in Child Development, 50(1-2, Serial No. 209), 66-106.