Report
When linking is stronger than thinking: Associative transfer of valence disrupts the emergence of cognitive balance after attitude change

https://doi.org/10.1016/j.jesp.2009.07.005Get rights and content

Abstract

The present research investigated the role of cognitive balance vs. associative transfer of valence in attitude change. Participants first formed positive or negative attitudes toward several source individuals. Subsequently, participants were shown source–target pairs along with information about the source–target relationship (‘likes’/‘dislikes’). Afterwards, participants’ attitudes toward the sources were changed by means of information that was opposite to the initially induced attitude. In a control condition, initial source attitudes remained unqualified. Results in the control condition showed that initially formed attitudes and available relationship information produced target evaluations that were consistent with the notion of cognitive balance. However, when attitudes toward the sources changed, target evaluations directly matched attitudes toward individually associated sources, irrespective of the relation between source and target. These results suggest that associative transfer of valence can disrupt the emergence of cognitive balance after attitude change.

Introduction

Our social network consists of people we like and feel close to (e.g., our spouse, best friends, children, parents). Other people are not as close and we either like or dislike them, but they are nonetheless an important part of our network (e.g., acquaintances, neighbors, colleagues). In addition, there are individuals we hardly know, but we may still have a positive or negative attitude toward them. This could be the case when someone we like or dislike feels either positively or negatively about these individuals (e.g., Aronson and Cope, 1968, Gawronski et al., 2005). For instance, imagine that your favorite colleague has some friends that you do not really know well, but because your colleague is very fond of them, you also have a favorable attitude toward them. The opposite is likely true for people your colleague feels negatively about.

Now imagine that you have an argument with your colleague and the situation gets so bad that your attitude toward your colleague becomes highly negative. Will your new attitude toward your colleague also change your attitudes toward your colleague’s friends and enemies? Would you now start to dislike your colleague’s friends, but like your colleague’s foes? The question of what happens to attitudes in social networks after change occurs in one part of the network is interesting not only from a real-world perspective; it also has a strong theoretical significance, as there are two potential factors that may influence attitudes in social networks after attitude change: (a) the simple transfer of valence through associative links in memory, and (b) the desire to hold attitudes that are in line with the principles of cognitive balance. These two mechanisms can lead to converging outcomes under some conditions but to opposite outcomes in others. The main goal of the present research was to test the different predictions implied by the two mechanisms to provide deeper insights into the dynamics of attitudes in social networks after attitude change.

One major social psychological theory that directly addresses the structure of attitudes in social networks is Heider’s (1958) theory of cognitive balance. According to balance theory, people strive for a pattern of interpersonal relations that can be described as balanced. In their simplest form, these balanced patterns include triads of relations between three individuals in which: (a) people like individuals who are liked by their friends, (b) people dislike individuals who are disliked by friends, (c) people dislike individuals who are liked by those whom they personally dislike, or (d) people like individuals who are disliked by those whom they personally dislike. According to Heider, a triad of interpersonal relations is balanced if it includes either no or an even number of negative relations (i.e., people disliking each other) and imbalanced if it contains an odd number of negative relations.

Research guided by balance theory provided important insights into the structure of attitudes in social networks. Specifically, the desire to maintain balanced relations has been shown to influence attitudes toward unfamiliar individuals, even when there was no information about these individuals other than their relation to a positively or negatively evaluated familiar individual. In such cases, mere knowledge about this relation has been shown to create an attitude toward the unfamiliar individual, such that the resulting structure of attitudes formed a balanced triad (e.g., Aronson and Cope, 1968, Gawronski et al., 2005). In the present study, we were interested in whether these principles still hold when attitudes toward one person in the triad have changed. This is not as obvious as it may seem, as there is an alternative mechanism that may in fact disrupt the emergence of cognitive balance after attitude change.

The notion of associative transfer of valence is most prominently reflected in research on evaluative conditioning (EC) (for reviews, see De Houwer et al., 2001, Walther and Langer, 2008, Walther et al., 2005). EC effects refer to changes in liking that are due to the pairing of stimuli (De Houwer, 2007). In a prototypical EC study, a neutral conditioned stimulus (CS) is repeatedly paired with a positive or negative unconditioned stimulus (US). The typical result is a shift in the valence of the formerly neutral CS, such that it acquires the valence of the US. A common explanation of EC effects is that repeated pairings of a CS with a given US create a mental link between the CS and the US in memory. As such, subsequent activation of the CS in memory may associatively spread to the US, which in turn activates the evaluation of the US. The result is an evaluative response to the CS that directly corresponds to the one toward the US.

Empirical evidence for such associative transfers of valence comes from research on US-revaluation (e.g., Baeyens et al., 1992, Walther et al., 2009). US-revaluation means that subsequent changes in the valence of an originally positive or negative US lead to corresponding changes in the valence of pre-associated CSs (Rescorla, 1974). For example, Walther et al. (2009) employed an evaluative learning paradigm in which neutral faces (CS) were repeatedly paired with either positive or negative faces (US). Subsequently, the valence of the US faces was changed by presenting positive faces with negative information and negative faces with positive information. In a control condition, US faces were presented with neutral information. Results showed that revaluation of the US not only led to a reversal in the valence of the US faces; it also led to corresponding changes in the valence of the pre-associated CS faces. These findings suggest that the CS faces acquired their valence indirectly by virtue of their mental association to a given US. Moreover, these results provide further evidence that attitude changes can be due to associative transfers of valence resulting from established links between two stimuli in memory.

The notions of cognitive balance and associative transfer of valence seem particularly important in the context of attitude change in social networks, as the two mechanisms can lead to opposite outcomes under certain conditions. To illustrate these conditions, imagine that you like or dislike a person named Peter and that you have learned that Peter likes or dislikes another unfamiliar person named Mike. According to balance theory, you should like Mike if: (a) you like Peter and Peter likes Mike, or (b) you dislike Peter and Peter dislikes Mike. However, balance theory predicts that you should dislike Mike if (c) you like Peter and Peter dislikes Mike, or (d) you dislike Peter and Peter likes Mike (e.g., Aronson and Cope, 1968, Gawronski et al., 2005). Importantly, the predicted attitudes toward Mike should also emerge if you just changed your attitude toward Peter from positive to negative or from negative to positive. What matters is your current attitude toward Peter. Thus, if the abovementioned case was implemented in a 2 (attitudes toward Peter: positive vs. negative) × 2 (Peter’s attitude toward Mike: positive vs. negative) experimental design, attitudes toward Peter and knowledge about Peter’s attitude toward Mike should produce a cross-over interaction, and this interaction pattern should be directly reversed if your attitudes toward Peter changed (see Fig. 1, upper panel).

These predictions stand in contrast to the ones derived from the notion of associative transfer. If your attitude toward Peter changes, the mental link between Peter and Mike that is created during the learning of their relation may lead to an associative transfer of the newly acquired attitude toward Peter, such that Mike acquires whatever valence is associated with Peter. In this case, Mike should acquire a positive valence if your attitude toward Peter changes from negative to positive, and a negative valence if your attitude toward Peter changes from positive to negative. Importantly, such associative transfers of valence may occur regardless of whether Peter likes or dislikes Mike, as they result from the simple associative link between Peter and Mike in memory. Thus, associative transfer of valence implies a simple main effect of attitudes toward Peter, such that Mike is liked when attitudes toward Peter change from negative to positive and disliked when they change from positive to negative (see Fig. 1, lower panel).

To test the differential predictions implied by the two accounts, we combined an impression formation paradigm derived from research on cognitive balance (Gawronski et al., 2005) with the revaluation paradigm used by Walther et al. (2009). In this combined paradigm, participants first formed positive or negative attitudes toward a given set of “source” individuals. Subsequently, participants were shown pairs of “source” and “target” individuals with the additional information of whether the source likes or dislikes the target. In a third phase, participants’ original attitudes toward the sources were changed by means of information that was evaluatively opposite to the information presented in the first phase. In a control condition, initial attitudes were left unchanged. Finally, participants evaluated all sources and targets. Drawing on earlier evidence for the impact of cognitive balance on social attitudes (e.g., Aronson and Cope, 1968, Gawronski et al., 2005), it was expected that participants in the control condition (no revaluation of source valence) would form attitudes toward the target individuals that are in line with the principles of cognitive balance. Specifically, participants should show favorable attitudes toward targets that are liked by positive sources or disliked by negative sources. Further, they should show unfavorable attitudes toward targets that are liked by negative sources or disliked by positive sources. Of higher importance are the evaluations in the revaluation condition, which speak to the present question of how cognitive balance vs. associative transfer of valence influence social attitudes after attitude change. Whereas cognitive balance predicts a full reversal of the two-way interaction pattern predicted for the control condition (see Fig. 1, upper panel), associative transfer of valence predicts a simple main effect of source valence, such that target evaluations should directly correspond to the new attitudes toward the sources (see Fig. 1, lower panel).

Section snippets

Participants and design

Forty students (26 females, 14 males) drawn from a volunteer pool took part in a study on impression formation. Participants received partial credit towards a course requirement. The experiment employed a 2 (original valence of source: positive vs. negative) × 2 (revaluation of source: opposite valence vs. control) × 2 (source–target relation: likes vs. dislikes) within-subjects design.

Procedure and materials

Upon arrival, participants were greeted by an experimenter and seated in front of a computer screen. The

Attitudes toward sources

To confirm the effectiveness of our manipulation of source valence, evaluations of the source individuals were submitted to a 2 (original source valence: positive vs. negative) × 2 (revaluation: opposite valence vs. control) × 2 (source–target relation: likes vs. dislikes) ANOVA with repeated measurements on all factors. This analysis revealed a significant main effect of original source valence, F(1, 39) = 38.22, p < .001, η2 = .49, a significant main effect of revaluation, F(1, 39) = 98.00, p < .001, η2 = .72,

Discussion

The main goal of the present research was to investigate the impact of cognitive balance vs. associative transfer of valence on the structure of social attitudes after attitude change. Specifically, do interpersonal attitudes still form a balanced triad after the attitude toward one member of the triad has changed, or can associative transfer of valence disrupt the emergence of balanced attitudes after attitude change? Expanding on the notion of associative transfer in the EC literature (e.g.,

Acknowledgments

This research was supported by a grant from the German Science Foundation (DFG) to the second author. We would like to thank Rebecca Weil for helpful comments on an earlier version of this article, and Tuba Cay and Oliver Weigelt for their help in collecting the data. Correspondence concerning this article should be addressed to Eva Walther, Department of Psychology, University of Trier, 54286 Trier, Germany, Email: [email protected].

References (24)

  • B. Gawronski et al.

    Attitudes and cognitive consistency: The role of associative and propositional processes

  • A.G. Greenwald et al.

    A unified theory of implicit attitudes, stereotypes, self-esteem, and self-concept

    Psychological Review

    (2002)
  • Cited by (17)

    • Attitudes beyond associations: On the role of propositional representations in stimulus evaluation

      2020, Advances in Experimental Social Psychology
      Citation Excerpt :

      Within this line of research, there is one other finding that seems to challenge a propositional perspective. Langer, Walther, Gawronski, and Blank (2009) first presented unknown individuals with positive or negative behavioral descriptions. Pictures of these individuals were used as USs in a second phase during which participants saw pairs of pictures showing a novel individual (CS) and a positive or negative individual (US).

    • Catching nonconscious goals in the act of decision making

      2014, Organizational Behavior and Human Decision Processes
    • The associative-propositional evaluation model. Theory, evidence, and open questions

      2011, Advances in Experimental Social Psychology
      Citation Excerpt :

      Conversely, participants showed less favorable evaluations of targets that were disliked by a positive source compared with targets that were disliked by a negative source (see Fig. 2.6, left panel; see also Aronson & Cope, 1968; Gawronski, Walther, & Blank, 2005). The critical question in Langer et al.'s (2009) study was whether the crossover interaction obtained in the control condition would be reversed when participants’ initial source attitudes were changed into the opposite direction, as predicted by the notion of cognitive balance. Interestingly, this was not the case.

    • Unintentional Influences in Intentional Impression Formation

      2022, The Handbook of Impression Formation: A Social Psychological Approach
    View all citing articles on Scopus
    View full text