Evaluative conditioning effects are modulated by the nature of contextual pairings

ABSTRACT

Across two studies participants completed a learning phase comprised of two types of trials: context pairing trials in which two (valenced or non-valenced) words were identical or opposite to one another and evaluative conditioning (EC) trials in which a CS was paired with a US. Based on the idea that EC occurs because CS-US pairings function as a symbolic cue about the relation between the CS and the US, we hypothesised that the nature of context pairings (identical or opposite) might moderate EC effects. Results indicate that identity-based context pairs led to typical assimilative explicit and implicit effects whereas opposition-based pairs led to attenuated effects. Implications and different accounts of our findings are discussed.

KEYWORDS:

• Evaluative conditioning
• symbolic
• relational qualifier
• implicit evaluation

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Jan De Houwer http://orcid.org/0000-0003-0488-5224

Notes

1 Note that our definition of a “symbol” and “symbolic meaning construction” is situated at a very abstract level of analysis. We view symbols as stimuli which are in some ways functionally substitutable for other stimuli in the environment (i.e. something that can stand for something else) and symbolic meaning construction as the transformation of a proximal event into a symbol by organisms with the aforementioned ability. By operating at a high level of abstraction, we hope to obtain new insights and reach consensus that might not be achievable when operating at the level of specific theories about learning and liking. That said, there are theories at both the functional (e.g. Hayes et al., 2001) and mental levels (e.g. Deacon, 1997) which offer insight into the origins and nature of symbols and symbolic meaning construction and which seem compatible with the general position we forward here.

2 The priming manipulation of Corneille et al. (2009) could been seen as another way of altering the symbolic meaning of pairings via contextual manipulations (i.e. they directed attention towards the similarities or differences between one set of paired events [pictures] and then introduced a second set of such events [CS-US pairings]). Although the priming manipulation could have exerted an impact because of this reason, it is important to note that this manipulation was designed explicitly with the aim of changing the perceptual processing of stimuli. Hence, the effects of the priming manipulation could also have merely influenced perceptual processes.

3 At the mental level of analysis our symbolic view is distinct from, but certainly compatible with, propositional models of EC. Given that our symbolic view considers pairings as cues which specify how stimuli are related, and that propositional representations are necessary in order to encode [higher-order] relational information in a mental system, it seems likely that propositions are a necessary component for symbolic meaning construction to occur. Critically, however, our symbolic view extends beyond propositional models by highlighting that pairings themselves can function as symbols (not just the CSs and US which are paired) and that pairings can also influence the relational content of the propositional belief.

4 We ran two initial pilot studies (n = 41 and n = 60) that attempted to influence EC effects by manipulating the nature of context pairs (words vs. pictures), and context pair training (blocked vs. interspersed). Results indicated that implicit and explicit evaluations were consistently biased by context pairs, such that assimilative EC effects emerged when the context pairs signaled that pairings were a cue for sameness whereas attenuated effects emerged when pairings were a cue for opposition. Experimental scripts, raw data, and analyses for these two pilot studies can also be found on the OSF website (https://osf.io/bn2q3/).

5 Note that including the data for all participants in the analyses did not result in a shift in significance for any of the reported effects (in Experiments 1 or 2). That said, we decided to continue excluding these participants to be consistent with our initial data-analytic plan.

Additional information

Funding

This research was conducted with the support of Fonds Wetenschappelijk Onderzoek [grant number BOF16/MET_V/002] to JDH.