Memory for emotional words: The role of semantic relatedness, encoding task and affective valence

Abstract

Emotional stimuli have been repeatedly demonstrated to be better remembered than neutral ones. The aim of the present study was to test whether this advantage in memory is mainly produced by the affective content of the stimuli or it can be rather accounted for by factors such as semantic relatedness or type of encoding task. The valence of the stimuli (positive, negative and neutral words that could be either semantically related or unrelated) as well as the type of encoding task (focused on either familiarity or emotionality) was manipulated. The results revealed an advantage in memory for emotional words (either positive or negative) regardless of semantic relatedness. Importantly, this advantage was modulated by the encoding task, as it was reliable only in the task which focused on emotionality. These findings suggest that congruity with the dimension attended at encoding might contribute to the superiority in memory for emotional words, thus offering us a more complex picture of the underlying mechanisms behind the advantage for emotional information in memory.

Keywords:

• Memory
• Semantic relatedness
• Positive words
• Negative words
• Encoding tasks

This paper is dedicated to the memory of our beloved friend and colleague Rosa Sánchez-Casas. We are grateful for everything she taught us.

This research was funded by the Spanish Ministry of Economy and Competitiveness [PSI2012-37623], by the Autonomous Government of Catalonia [2014SGR-1444], by the Regional Government of Galicia [INCITE09204014PR] and by the ELC research grant [CN2011/011].

This paper is dedicated to the memory of our beloved friend and colleague Rosa Sánchez-Casas. We are grateful for everything she taught us.

This research was funded by the Spanish Ministry of Economy and Competitiveness [PSI2012-37623], by the Autonomous Government of Catalonia [2014SGR-1444], by the Regional Government of Galicia [INCITE09204014PR] and by the ELC research grant [CN2011/011].

Notes

¹ In this baseline experiment, 31 naive participants performed the emotionality task with a list containing 28 unrelated neutral words. We only tested the unrelated condition because it was not possible to find a new set of animal names matched in semantic relatedness. The baseline list comprised the set of 14 neutral unrelated words used in our experiment plus a new set of 14 neutral unrelated words (they were matched in all the relevant variables with the positive, negative and neutral experimental sets: concreteness, length, frequency and semantic relatedness). The following additional set of 14 neutral unrelated words were used in the baseline experiment:

Albañil (builder), archivador (file), barril (barrel), radiador (radiator), cerilla (match), chaleco (vest), cortina (curtain), dardo (dart), estatua (statue), póster (poster), lavabo (sink), martillo (hammer), pañuelo (handkerchief), and paraguas (umbrella).

We analysed together the results of both experiments (only including the emotionality task and the unrelated condition). We performed a mixed ANOVA with the variables set (set 1 consisting of the original neutral words and set 2 consisting of either the new neutral words or the original positive or negative words, depending on the list) and list (list 1 with positive and neutral words, list 2 with negative and neutral words and list 3 with two sets of neutral words). The first one was a within-subjects variable and the second one was a between-subjects variable. We obtained a significant effect of set, F(1,69) = 16.96, MSE = 121.35, p < .001, η² = .19, as well as list, F(2,69) = 10.21, MSE = 146.51, p < .001, η² = .23. Importantly, there was a significant interaction between both variables, F(2,69) = 3.53, MSE = 121.35, p < .05, η² = .09. Pairwise comparisons revealed that when the lists included a set of emotional (either positive or negative) and a set of neutral words, the former were significantly better remembered than the later (both ps < .005, see Table 2 for means). Conversely, there was not any difference in recall when the two sets were neutral (M = 16.4, SD = 10.9 and M = 17.5, SD = 12.1 for the old and the new set of neutral words, respectively). When we compared memory for positive, negative and neutral words across lists we found that positive words (M = 31.1, SD = 14.6), but not negative words (M =22.8, SD = 11.5), were significantly better remembered than their matched neutral baseline words (M = 17.5, SD = 12.1, p < .001). Regarding neutral words, they were remembered worse when they appeared in a list mixed with a set of negative words (M = 11.4, SD = 10.5) than with a set of positive words (M = 21.1, SD = 9.1), p < .01. However, the difference was not significant when the comparison was with a list including only neutral words (M = 16.4, SD = 10.9).