An Emotional Mediation Theory of Differential Age Effects in Episodic and Semantic Memories

ABSTRACT

Although there is a large decrement in central episodic memory processes as adults age, there is no appreciable decrement in central semantic memory processes (Allen et al., Journal of Gerontology: Psychological Sciences, 57B, P173–P186, 2002; Allen et al., Experimental Aging Research, 28, 111–142, 2002; Mitchell, Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 31–49, 1989). The authors develop a theory of episodic memory's connections to cognitive, emotional, and motivational systems to explain these differential age effects. The theory is discussed within the context of the cognitive neuroscience research regarding limbic system connectivity in conjunction with Damasio's notion of somatic markers (Descartes’ error: Emotion, reason, and the human brain, New York: Grosset/Putnam, 1994). The central hypothesis is that elements of limbic system circuitry, including portions of the medial temporal lobes and frontal cortex, are associated with both working and long-term episodic memory performance, and by extension, with the capacity to engage in emotion-guided, self-regulatory processes that depend heavily on episodic memory. In contrast, the semantic memory system may have less shared interface with episodic and affective networks (i.e., the limbic-related system), and therefore remain independent of neurocognitive changes impacting emotional states and episodic-type memory processes. Accordingly, this framework may account for the pattern of age-related declines in episodic relative to semantic memory, particularly if older adults experience less emotional activation, and therefore fewer somatic markers, than younger adults. An initial empirical examination of this emotional mediation theory is presented, using preexisting data that include indicators of age, chronic tendency to focus on negative emotional stimuli (neuroticism), and working memory performance.

This research was supported by National Institutes of Health/National Institute on Aging Grant AG09282 to Philip A. Allen. The authors wish to acknowledge useful comments from Marty Murphy, Ray Sanders, and Karen Kopera-Frye on earlier drafts of this paper.

Notes

Note. Decimals have been eliminated from correlations to save space. HVS = Hybrid Visual Search; MemSrch = Memory Search; WAIS Vocab = Vocabulary subscale of Wechsler Adult Intelligence Scale; LexDec = Lexical Decision Task; MultVer = Multiplication Verification Task

Note. RMSEA = root mean square error of approximation; NNFI = non-normed fit index; CFI = comparative fit index. The χ² goodness-of-fit index for all three models is not statistically significant, indicating acceptable fit of the models to the data *p < .05.