Abstract
The frequency and characteristics of involuntary autobiographical memories were compared in 25 stable dysphoric and 28 non-dysphoric participants, using a new laboratory-based task (Schlagman & Kvavilashvili, 2008). Participants detected infrequent target stimuli (vertical lines) in a simple vigilance task and recorded any involuntary autobiographical memories that came to mind, mostly in response to irrelevant words presented on the screen. Dysphoric participants reported involuntary memories as frequently and as quickly as non-dysphoric participants and their memories were not repetitive intrusive memories of negative or traumatic events. Additional content analysis showed that dysphoric participants did not recall more memories of objectively negative events (e.g., accidents, illnesses, deaths) than non-dysphoric participants. However, significant group differences emerged in terms of a mood congruency effect whereby dysphoric participants rated their memories as more negative than non-dysphoric participants. Moreover, the proportion of negatively rated involuntary memories was related to lower mood ratings at the end of the session in the dysphoric but not in the non-dysphoric group. Finally, groups did not differ on several memory characteristics such as vividness, specificity (high in both groups) and rates of rehearsal (low in both groups). Theoretical and practical implications of these findings for research on depression and autobiographical memory are discussed.
Acknowledgements
Simone Schlagman is now at Inter-Research Science Centre, Oldendorf/Luhe, Germany.
The research presented in this paper was supported by a doctoral studentship awarded to the second author from the Economic and Social Research Council (UK) and a British Academy/Leverhulme Trust (UK) senior research fellowship awarded to the first author. Portions of research described in this paper were presented at the Autobiographical Memory Meeting, Oxford University, UK (January 2005) and the British Psychological Society Cognitive Section Conference, Leeds University, UK (September 2005). We are grateful to Kim Meredith for helping us out with data collection and analyses.
Notes
1The presence of cues indicates that involuntary memories are “not spontaneous in the sense of being cue-independent, but only in the sense of being unintended” (Berntsen, 1996, p. 441).
2Although it is customary to use a cut-off point of 9 and above on BDI to class participants as dysphoric or depressed (see Cox, Enns, Borger, & Parker, 1999), we followed the recommendation of Vredenburg, Flett, and Krames (1993) and used a cut-off point of 16 and above, which corresponds to moderate depression (Beck, 1967) and reduces the chances of participants being classed as non-depressed when tested a second time (see Zimmerman, 1986).
3There is some variability in guidelines as to what counts as good reliability in case of Cohen's Kappa. However, as a rule of thumb, reliability between 0.60 and 0.80 is considered as good. For example, Landis and Koch (1977) consider reliability between 0.60 and 0.80 as substantial, and between 0.81 to 1.00 as perfect agreement.
4We also calculated the mean number of involuntary memories that were triggered by negative, positive, and neutral cues and entered them into a 2 (group)×3 (cue-valence) mixed ANOVA with repeated measures on the last factor. This resulted in a significant main effect of cue-valence F(2, 84) = 11.46, p<.005, ηp 2=.21, with the number of memories triggered by negative cues (M=3.25) being reliably higher than the number of memories triggered by either neutral (M=1.18) or positive cues (M=1.80) (both ps<.001), which did not differ from each other (p=.24). There was no main effect of group (F <1) or group by cue-valence interaction (F=1.22), indicating the absence of mood congruency effect for the valence of cues.
5This was a fairly large category encompassing a variety of everyday activities such as shopping, gardening, watching TV, exercising, playing chess, etc.
6Similar results were obtained when we calculated correlations between the proportion of negative memories and participants’ mood-change scores. These were obtained by subtracting mood rating at the start of the session from the mood rating at the end of session. A score of 0 would indicate no change, a negatively signed mood score would indicate the worsening of the mood, and a positively signed score an improvement of mood over the session. For the dysphoric group there was a significant negative correlation between mood-change and the proportion of negative memories, r(22) = −.46, p=.03. Thus the higher the proportion of negative memories reported, the lower the mood-change score (i.e., the more negative the mood became). For the non-dysphoric group, the correlation was not significant, r(26) = −.11, p=.61.
7Relevant previous research does not usually require participants to self-rate the specificity of their memories: the memories are coded by independent raters. Therefore to further support the analysis of participants’ own ratings, two independent coders also rated each memory as to whether it concerned a general or specific event and inter-rater reliability was high ( Kappa=.90, SE=.04). The mean proportion of memories rated as specific were .82 (SD=0.20) for the non-dysphoric group and .77 (SD=0.21) for the dysphoric group and did not differ (F <1).