Increased Direct Retrieval of Overgeneral Categoric Memory in Individuals with Dysphoria and a History of Major Depression

A one-factor model of OGM was proposed which claimed that OGM occurs independently of cue valence (Griffith et al. 2009, 2012a; Takano et al. 2017). Our prediction that increased categoric memory could be obtained only in negative cues appears to contradict this one-factor model. We believe that the OGM originating from the low executive control surely occurs regardless of cue valence. Functional avoidance and memory suppression generally occur in negative memory, but if they become too generalized, OGM may occur regardless of cue valence (Williams et al. 2007; see also Hulbert et al. 2016, who implied the generalization of memory suppression). However, we believe that OGM induced by depressogenic processing and dysfunctional schema is biased to negative tendency. The AMT traditional instruction (Williams and Broadbent 1986) should approach a one-factor model because this instruction strongly reflects executive control in that it forces the retrieval of specific memories (see Discussion in Matsumoto and Mochizuki 2019). However, when minimal instruction, which is an instruction that more easily reflects schema-based direct retrieval rather than generative retrieval under the monitoring of executive control, is applied to individuals with depression, a 2-factor model may be established. The 2-factor model also indicates adaptation when minimal instruction is used for undergraduate students (Griffith et al. 2009; Takano et al. 2017) or traditional instruction is used for participants who experienced childhood abuse (Griffith et al. 2012a). In future studies, it may be an interesting idea to examine the factor structure of minimal instruction in consideration of depression and traumatic experiences.

We also performed hypothesis-driven analyses. One-way ANOVA was performed to determine the proportion of categoric memories recalled via direct retrieval in response to negative cues and showed a significant group difference (F (2, 81) = 4.32, p = .017, partial η² = .10). A multiple comparisons analysis showed that the past MDE group produced more categoric memories than the healthy control group (t = 2.74, p = .007, d = 0.78) and the matched control group (t = 2.55, p = .013, d = 0.72). These results supported our hypothesis.

Although direct retrieval of categoric memory was observed regardless of cue valence, this result is difficult to interpret using the one-factor model of OGM (Griffith et al. 2009). This is because the impairment of executive control or functional avoidance, which may be the basis of the one-factor model, causes the attenuation of generation retrieval but does not contribute much to direct retrieval (Eade et al. 2006; Watson et al. 2013; Williams et al. 1999).

As in Study 2, individuals with current and past depression were more likely than the general population because they were encouraged to join the studies.