Attention please: No affective priming effects in a valent/neutral-categorisation task

Abstract

Affective congruency effects in the evaluation task can be explained by either spreading of activation or response competition. Eliminating effects of response compatibility by using other tasks (semantic categorisation, naming task) typically also eliminates affective congruency effects. However, there is no need for processing the affective information of the stimuli in these tasks either, which could be necessary for an affectively mediated spreading of activation (Spruyt et al., 2007, 2009, 2012). We introduced a new task to further test this hypothesis. The valent/neutral-categorisation task does not confound affective congruency with response compatibility, but still requires a processing of the stimuli's valence. No affective congruency effect was obtained with this task in two experiments, disfavouring a conditional spreading activation account. On the other hand, a significant priming effect was found for associated word pairs in Experiment 1, providing evidence for the sensitivity of the task to detect spreading activation processes.

Keywords:

• Affective priming
• Attention
• Spreading activation
• Encoding, Response compatibility
• Response conflict
• Evaluation
• Categorisation

Notes

¹More recent variants of this account replaced the semantic-network analogy of the original spreading activation account with assumptions regarding distributed representations of stimuli (Masson, 1995). Since a detailed distinction between variants of accounts postulating facilitated encoding of affectively congruent targets is not the main focus of this paper, we will summarise these accounts with the term “spreading activation”, which is commonly used in the literature.

²Everaert et al. (2011) varied the proportion of valent stimuli rather than using a secondary task to manipulate attention towards valence. They found affective priming effects in the high valence proportion, but not in the low valence proportion condition. At least in this case our hypothetical considerations do not seem to be applicable.

³RTs were treated as outliers if they were below 400 ms or more than one and a half interquartile ranges above the third quartile of the individual response time distribution (Tukey, 1977). The relatively high lower cut-off was chosen on the basis of a close inspection of the overall RT distribution, which suggested this threshold. However, replacing the 400 ms cut-off with a 250 ms threshold yielded virtually identical results. In particular, no significant affective congruency effect was obtained for the valent trials, t(59) < 1, whereas a significant associative priming effect was obtained for the neutral materials, t(59) = 2.80, p<.01.

⁴As in Experiment 1 RTs were treated as outliers if they were below 400 ms or more than one and a half interquartile ranges above the third quartile of the individual response time distribution (Tukey, 1977). Replacing the 400 ms cut-off with a 250 ms threshold yielded the same results. No significant affective congruency effect was obtained, t(64) = 1.60, p=.114.