The impact of open and closed mindsets on evaluative priming

Abstract

In the present research, we argue that open versus closed mindsets, accompanying ongoing versus completed mental jobs on the prime, determine the size of congruity effects in the evaluative priming paradigm. More specifically, we hypothesised that disfluent primes that resist an easily completed encoding process should induce an open mindset and thereby result in stronger congruity effects than fluent primes that induce closed mindsets. Across two experiments, we applied two different manipulations of prime fluency: gradual demasking (Experiment 1) and colour contrast (Experiment 2). As expected, in both experiments we found robust congruity effects, but only on trials with disfluent (vs. fluent) primes. Results of a follow-up experiment suggest that these effects are not due to attentional processes. We conclude that the mindsets resulting from individuals' activities during encoding are crucial in determining the outcome of evaluative priming effects.

Keywords:

• Affective priming
• Compound-cue
• Evaluative priming
• Fluency
• Mindset

ACKNOWLEDGEMENTS

The research reported in this article was supported by a Koselleck grant from the Deutsche Forschungsgemeinschaft to KF and a Lavoisier grant from the French Ministry of Foreign and European Affairs to TA. This research was carried out when the first author was a postdoc at the Cognitive Research Group in Social Psychology at the University of Heidelberg.

We gratefully acknowledge helpful comments on a draft of this article by Debbie Prentice.

Notes

¹We chose to apply our mindset account to evaluative priming because of the variety of normal, eliminated or reversed priming effects documented in this domain. However, it could easily apply to any priming situation, including semantic priming. It is not a coincidence if we refer to the operation of mindsets in other concept or response activation paradigms (e.g., Huber et al., 2001; Martin, 1986; Sparrow & Wegner, 2006).

²Participants could thus initially see a brief presentation of the prime before its gradual demasking. This allowed us to compare two conditions differing in terms of encoding attempts that induce distinct mindsets and not in absolute terms of short versus long prime presentation.

³An analysis on latencies simply trimmed to the range between 300 ms and 1500 ms yielded a similar pattern of results. The stimulus visibility by priming interaction was marginally significant, F(1, 19) = 4.08, p = .057, d = 0.90. Again, the priming effect was significant for the delayed visibility condition, F(1, 19) = 8.07, p = .010, d = 1.27, but not for the intact condition, F <1.

⁴Because processing fluency is hedonically marked, with fluently processed information experienced positively (Winkielman & Cacioppo, 2001), and less fluently processed, novel information giving rise to a negative experience (Phaf & Rotteveel, 2005), we also performed an ANOVA with prime and target valence as factors, in order to check whether our fluency manipulation interacted with stimulus valence. A 2 (Prime Valence: positive vs. negative) by 2 (Target Valence: positive vs. negative) by 2 (Stimulus Quality: fluent vs. disfluent) ANOVA was conducted on data from Experiments 1 and 2. Importantly, both analyses yielded the predicted three-way interaction of Prime Valence, Target Valence and Stimulus Quality corresponding to our hypotheses. Additionally, in Experiment 2, Prime Valence interacted with Stimulus Quality, F(1, 24) = 8.43, p = .007, d = 1.18. Participants were slower in trials with a negative disfluent prime (M = 773 ms) than a positive disfluent (M = 729 ms) or fluent prime (M = 741 ms for positive and M = 723 ms for negative primes, respectively).

⁵Again, an analysis based on the more conventional 300 ms–1500 ms data trimming procedure did not affect the pattern of significant and nonsignificant effects. The stimulus quality by priming interaction was significant, F(1, 24) = 6.31, p = .019, d = 1.03. Similarly, the congruity effect was significant in the low-contrast condition, F(1, 24) = 8.31, p < .01, d = 1.19, but not in the medium-contrast condition, F < 1.

⁶We would like to thank Dirk Hermans and an anonymous reviewer for raising this point.

⁷We suspect that this absence of effect on reaction times (combined with the fact that we found significant effect on accuracies) could be due to slightly different instructions compared to the other reported experiments, which inadvertently emphasised speed over accuracy. Consistent with this, we found that reaction times of the follow-up experiment were generally faster (M = 706 ms) compared to those in Experiment 1 (M = 799 ms) or Experiment 2 (M = 746 ms). Another reason for this absence of effect could be the fact that the follow-up experiment contained very few trials (i.e., 48, 12 trials per condition taking into account the fluency and the congruency factors) in comparison to most priming tasks.

⁸Conservation could be seen as the “default” operation on primes yielding normal priming effects on a typical priming task. This explains, for example, why we obtained priming effects in the pilot study.