Abstract

Researchers have been interested in what retrieval process is responsible for overgeneral autobiographical memories (OGM) in depression. Previous cross-sectional studies demonstrated that, for negatively valenced cues, directly retrieved OGM, rather than generatively retrieved OGM, are associated with depression. However, longitudinal evidence of this relationship is still lacking and needs to be tested. We conducted a re-analysis of the online computerised memory specificity training (c-MeST) data to examine whether directly retrieved OGM for negative cues prospectively predicts high levels of depression 1 month later. Participants who met the criteria of current major depressive disorders (N = 116; n = 58 in the c-MeST group and n = 58 in the control group) recalled autobiographical memories for positive and negative cues and judged each retrieval process (i.e. direct or generative). The results supported our prediction, and directly retrieved OGM for negative cues predicted high levels of depressive symptoms 1 month later even after controlling for the group effect, baseline depressive symptoms, executive functioning and rumination. Exploratory analysis showed that direct retrieval of specific memories prospectively predicted low levels of depression. These results support the theory that elevated accessibility of negatively valenced general memory representations is a vulnerability factor for depressive symptoms.