La mémoire autobiographique épisodique permet de remonter dans le temps pour revivre mentalement (conscience autonoétique) un événement personnellement vécu. La revue de la littérature sur le fonctionnement de la mémoire autobiographique épisodique au cours de la dépression met en évidence trois résultats principaux : la surgénéralisation des souvenirs, la congruence à l'humeur et la fréquence des souvenirs intrusifs. La surgénéralisation, c'est-à-dire le rappel préférentiel de souvenirs portant sur des événements répétés ou étendus, est relativement spécifique de la dépression. Cet effet a été répliqué quasi constamment dans la littérature depuis une quinzaine d'années. Il constitue un facteur de mauvais pronostic de l'épisode dépressif et possède probablement les caractéristiques d'un trait, notamment en cas de trouble dépressif récurrent. Les données les plus récentes suggèrent l'existence d'un trait acquis, équivalent d'une séquelle cognitive de la dépression. Il existe des liens entre évitement des souvenirs intrusifs et surgénéralisation. Ces liens demandent à être explorés au sein d'un modèle intégrant self, fonctions exécutives et mémoire autobiographique épisodique dans le domaine de la dépression.

Autobiographical memory and personal identity (self) are linked by a reciprocal relationship. Autobiographical memory is critical for both grounding and changing the self. Individuals' current self-views, beliefs, and goals influence their recollections of the past. According to Tulving, episodic memory is characterized by autonoetic consciousness, which is associated with a sense of the self in the past (emotions and goals) and mental reliving of an experience. Its close relationship with self and emotion strongly involves episodic autobiographical memory in the psychopathology of depression. However, due to methodological and conceptual issues, little attention has been paid to episodic autobiographical memory in depression. Since the seminal work of Williams et al. 15 years ago, there is now growing interest around this issue.

We reviewed the evidence for three major features of autobiographical memory functioning in depression : an increase in general memory retrieval (overgenerality), a mood-congruent memory effect and the high occurrence of intrusive memories of stressful events. Although it was first observed among suicidal patients, overgenerality is actually associated with both depression and post-traumatic stress disorder. Overgenerality is not associated with anxious disorders other than post-traumatic stress disorder, obsessive-compulsive disorder, or borderline personality disorder. Most of controlled studies carried out on autobiographical memory in depression rely on the Williams' Autobiographical Memory Test (AMT). When presented with positive and negative cue words and asked to retrieve specific personal events, depressed patients (unlike matched controls) are less specific in their memories. They tend to recall repeated events (categorical overgeneral memories) rather than single episodes (specific memories). Overgenerality in depression is : 1) more evident with positive than with negative events (mood-congruent memory effect) ; 2) related to avoidance of intrusive memories ; 3) quite stable over time, ie, remaining after remission ; and 4) related to short-term prognosis in depression. Although it is not clear whether overgenerality is a cause or an effect of depression, there is some evidence to suggest that overgenerality is a trait marker indicating vulnerability to persistent depression. Mood-congruent effect, a well-known effect in depression, has been addressed in both autobio-graphical and non-autobiographical memory. Depressed patients spontaneously recall more negative than positive memories. With the AMT, depressed patients take longer to respond to positive than to negative cues, whereas controls do the opposite. Depression is also associated with a high occurrence of spontaneous intrusive memories of stressful life events. Studies found intrusions and related avoidance, as measured by the Impact of Event Scale, to be positively correlated with overgenerality, whereas there was no direct link between performance on the Autobiographical Memory Test and stressful life events per se. Both Williams' mnemonic interlock model and Conwayʼs self-memory system are useful models to address the complexity of findings regarding autobiographical memory and depression.

According to Williams, repeated avoidance of stressful memories leads depressed patients to have an autobiographical memory functioning characterized by iterative retrievals of categorical overgeneral memories, producing an enduring overgeneral retrieval style. According to Conway, the recollection of autobiographical memories requires a retrieval process that provides access to sensory/perceptual event-specific knowledge (ie perceptions and feelings) via a personal semantic knowledge base (ie lifetime periods and generic events). This retrieval process (generative retrieval mode) relies on both executive functioning and current self-view, namely the working-self. Spontaneous memories, usually vivid, result from a direct retrieval mode in which event-specific knowledge is directly triggered. In line with this model, episodic autobiographical memory impairment in state depression may arise from the working self rather than from autobiographical knowledge. The mood-congruent effect may be explained by the current (depressed) self. The high occurrence of intrusive memories may be explained by lack of executive control during direct retrieval. Overgenerality may rely on the interaction of both executive dysfunction and current (depressed) self, within the working-self, during generative retrieval. Our review suggests that further evidence is needed to address the relationship between executive functioning, self and autobiographical memory in depression.