Mood-congruent true and false memory: Effects of depression

Abstract

The Deese/Roediger-McDermott paradigm was used to investigate the effect of depression on true and false recognition. In this experiment true and false recognition was examined across positive, neutral, negative, and depression-relevant lists for individuals with and without a diagnosis of major depressive disorder. Results showed that participants with major depressive disorder falsely recognised significantly more depression-relevant words than non-depressed controls. These findings also parallel recent research using recall instead of recognition and show that there are clear mood congruence effects for depression on false memory performance.

Keywords:

• False memory
• Depression
• Mood congruence

Acknowledgements

Order of authorship was determined alphabetically. Both authors contributed equally to the research and the writing of this article.

Notes

¹Although norms exist for neutral, positive, and negative DRM lists, depression-relevant lists have not been previously rated. For the purposes of the current experiment we created these depression-relevant lists using standard word association norms (Moss & Older, 1996) to ensure their compatibility with other DRM lists. In addition, a separate group of 68 non-depressed participants rated these new words in terms of their relevance to depression, as the average person would rate them. The rating scale had seven options ranging from very irrelevant to very relevant (to depression). The final depression-relevant word lists (see Appendix) were chosen when 70% or more of the participants rated the critical targets and at least 6 of the 10 words in the lists as very relevant to depression. Care was taken to ensure there was minimal overlap between the concepts in the separate lists.

²Not all of the words on the various lists were available in all of the different sets of norms for valence, word frequency, or backward associative strength. Indeed, although the majority of the words could be rated across list type, a minority could not. The analyses we report are for those words where norms were available.

³Signal-detection analyses were also conducted. The recognition results were the same regardless of whether raw scores or signal detection measures (d′ and C) were used in the analyses. As there were no differences in the outcome of these analyses, and because we wanted to make contact with other studies in the literature, we report the untransformed analyses.