I like you, I like you not: Understanding the formation of context-dependent automatic attitudes

Abstract

Previous research has shown that automatic evaluations can be highly context dependent. Expanding on past research demonstrating context effects for existing attitudes toward familiar objects, the present research examined basic principles that guide the formation of context-dependent versus context-independent automatic attitudes. Results from four experiments showed that: (a) newly formed attitudes generalised to novel contexts when prior experiences with the attitude object were evaluatively homogeneous; (b) when prior experiences were evaluatively heterogeneous, automatic evaluations became context sensitive, such that they reflected the contingency between the valence of prior experiences and the context in which these experiences occurred; and (c) when prior experiences were evaluatively heterogeneous across different contexts, novel contexts elicited automatic evaluations that reflected the valence of first experiences with the attitude object. Implications for research on automatic evaluation and attitude change are discussed.

Keywords:

• Attitude formation
• Attitude change
• Automatic evaluation
• Context effects
• Implicit measures

Acknowledgements

This research was supported by PHS Grant MH 63762, Canada Research Chairs Program Grant 202555, Social Sciences and Humanities Research Council of Canada Grant 410-2005-1339, and Early Researcher Award ER07-03-067 from the Ontario Ministry for Research and Innovation.

We would like to thank the 18th and 19th Annual Duck Conference on Social Cognition for facilitating this research.

Notes

¹ In the present article, we use the term automatic in the sense of unintentional (i.e., spontaneous evaluative responses to an object that do not require an intention to evaluate that object).

² Note that such primacy effects of earlier acquired information in novel contexts may also occur when evaluatively inconsistent information is acquired in the same learning context and the novel context is sufficiently dissimilar to the context during learning (e.g., Bouton & Ricker, 1994). Even though this possibility is not explicitly addressed in the present studies, it seems an interesting question for future research.

³ Note that responses to positive target words are typically faster than responses to negative words, thereby promoting scores higher than zero for the present scoring. Thus, the resulting priming scores should not be interpreted in an absolute manner, such that scores higher than zero would indicate a positive response and scores lower than zero would indicate a negative response. Instead, priming scores should only be interpreted in a relative manner, such that higher scores indicate more positive responses.

⁴ Note, however, that the acquisition of evaluatively inconsistent information in the same learning context may not necessarily change the original evaluative response in novel contexts. Instead, it seems possible that novel contexts may still reflect the valence of the earlier acquired response, even when that response is changed for the learning context (see Bouton & Ricker, 1994).