Generalization Effects in Evaluative Conditioning: Evidence for Attitude Transfer Effects from Single Exemplars to Social Categories

Abstract

The present research investigated whether evaluatively conditioned attitudes toward members of a social category (CSs) generalize to other stimuli belonging to the same category as the CSs (generalization at the stimulus level) and to the category itself (generalization at the category level). In four experiments, USs were paired with schematic or naturalistic CSs belonging to certain fictitious groups. Afterward, attitudes toward the CSs, toward non-presented exemplars of the CS category, and toward the CS category were assessed. Results revealed evidence for generalization effects in EC on both the stimulus and the category level. Transfer effects were greater when participants’ awareness of the CS–US contingency (CA) was high. Moreover, we found differences in generalization between the stimulus and category level, indicating that different processes might contribute to the effects. Theoretical and practical implications such as using EC as a tool for changing attitudes toward social groups will be discussed.

Category-bounded emotional enhancement: spillover effects in the valuation of public goods

“…Studies have shown that evaluative conditioning works: after a conditioning phase, people's attitudes toward the CS change in correspondence to the valence of the US (see Hoffman, De Houwer, Perugini, Baeyens, & Crombez, 2010). Related to the present research, studies have also shown that the attitude change due the evaluative conditioning generalizes to other stimuli that are similar to the CS (similar to the current assimilation effects), but also to the whole category in which the CS belongs (Glaser & Kuchenbrandt, 2017). However, most of these studies have used relatively simple stimuli such as single words ("Denmark") or schematic representations of imaginary creatures.…”

“…For a parallel distinction in inductive inference refer to the category-based induction model of Osherson, Smith, Wilkie, López, and Shafir (1990; fits with the category transmission view) and the feature-based induction model of Sloman (1993; fits with the feature transmission view). It is also noteworthy to mention that a parallel issue has been addressed in studies on evaluative conditioning with mixed results: some findings appear to support a category-based transmission (e.g., Glaser & Kuchenbrandt, 2017), while others have highlighted the importance of shared features (e.g., Spruyt, Klauer, Gast, Schryver, & De Houwer, 2014).…”

Evaluative conditioning of artificial grammars: Evidence that subjectively-unconscious structures bias affective evaluations of novel stimuli.

“…However, in order for EC to have a significant impact in daily life, the effect must also generalize to previously unseen stimuli that share similarities with the presented CSs (Hütter & Tigges, 2019;Unkelbach, Stahl, & Förderer, 2012). Moreover, a core objective for EC research has been to determine whether the conditioning (e.g., Sweldens, Corneille, & Yzerbyt, 2014) and the generalization (e.g., Glaser & Kuchenbrandt, 2017) effects can appear without the awareness of the contingencies formed between the CSs and the USs, because evaluations that are based on unconscious knowledge might have different properties from consciously based ones. For example, the former could be more difficult to change, to integrate with infor-mation from other sources (cf.…”

“…In order to test whether generalization can occur without awareness, Glaser and Kuchenbrandt (2017) presented participants with two groups of alien creatures. On their heads, members of one group had an antenna, whereas members of the other group had a triangle-shaped object.…”

“…We argue that even if generalization of EC would be possible without awareness, most studies present participants with simple regularities that are easy to detect consciously: The CSs paired with a positive valence are easily distinguishable from those paired with negative valence (for example, they belong to clearly different and salient categories [e.g., men vs. women in Hütter et al, 2014] and/or are differentiated by obvious cues [e.g., Glaser & Kuchenbrandt, 2017]). Furthermore, these generalization studies typically use so-called "objective" measures of awareness, which assume that if participants accurately identify the valence the CSs have been paired with, they have conscious memory of the CS-US contingencies.…”