Affective priming in the valent/neutral categorisation task is due to affective matching, not encoding facilitation: Reply to Spruyt

Abstract

Spruyt obtained an affective congruency effect in a valent/neutral categorisation task, which contrasts with the absence of such an effect in the same task that was reported by Werner and Rothermund. The crucial difference between the two studies is that Spruyt presented only valent primes, whereas Werner and Rothermund presented equal amounts of valent and neutral primes and targets in their experiments. Removing the neutral primes introduces a confound of affective matches with the required response. Affective congruency effects in Spruyt's study can be explained straightforwardly in terms of such an affective matching strategy. To demonstrate the influence of matching strategies in the valent/neutral task without neutral primes, we conducted an experiment in which we induced an affective mismatching strategy. In support of our reasoning, this study revealed an affective incongruency effect in the valent/neutral categorisation task. We conclude that affective congruency as well as incongruency effects in the valent/neutral categorisation task reflect post-lexical affective (mis-)matching strategies rather than encoding facilitation.

Keywords:

• Affective priming
• Affective matching
• Encoding
• Evaluation
• Categorisation

Notes

¹ Participants who were excluded responded much faster than the rest of the sample in both tasks (on average, response times (RTs) were more 200–250 ms faster for excluded participants), indicating that these participants were distracted and less motivated to perform well on the complex combined task, trading accuracy for speed. Including all participants into the analyses renders the incongruency effect in the latency data non-significant (though there is still an advantage for incongruent trials in the means), whereas the accuracy-based incongruency effect remains significant and even becomes somewhat stronger.

² RTs were treated as outliers if they were more than three interquartile ranges above the third quartile of the overall response time distribution (Tukey, 1977).

³ We observed a large number of errors in the valent/neutral task in our study, which is due to the complex nature of the combined task. On the one hand, error percentages are typically higher in a task switch situation compared to a single task. Second, and more importantly, by introducing the mismatch-identification task, we in fact created a conflict between the two competing response strategies: Within the valent/neutral task, there is still a strong incentive to search for affective matches, and to respond with the valent (‘YES’) key for affective matches. This is the very process that has driven the congruent effect in the study by Spruyt (in press). However, exactly the opposite strategy is required in the mismatch-identification task, in which participants have to search for affective mismatches and have to press the ‘YES’ key for these trials. Introducing a conflicting response strategy with our additional task increases the level of errors in the valent/neutral task because it interferes with the easy strategy to shortcut response identification.