Multiple-trace theory and semantic dementia: Response to K.S. Graham (1999).

Abstract

We are indebted to Graham for raising the issue of semantic dementia and its implications for multiple-trace theory [Graham, K.S. (1999) Semantic dementia: a challenge to the multiple-trace theory of memory consolidation? Trends Cognit. Sci. 3, 85–87]1. Investigating people with semantic dementia to study hippocampal function and episodic memory is an innovative departure from the typical strategy of studying people with amnesia who have lesions to the medial temporal lobe and diencephalon. By capitalizing on the observation that the medial temporal lobes are spared in the early to middle stages of semantic dementia, one can evaluate their contribution to episodic (and semantic) memory when the lateral inferotemporal cortex, which supports semantic memory, is damaged. In their investigations, Graham and Hodges2 confirmed and extended Snowden et al’s3 initial observations that episodic and semantic memories based on recent experiences are relatively spared in semantic dementia, whereas remote memories are deteriorated or lost. Graham and Hodges2,4 concluded that these results provided evidence for traditional consolidation theory, which states that the medial temporal lobes or hippocampal complex play only a time-limited role in memory. By this view, once memories are consolidated, they are retrieved directly from the neocortex and other structures in which the memory or engram is represented. Because semantic dementia results from neocortical degeneration, only these remote, consolidated memories would be lost. More recent memories would be spared because they had not yet been consolidated and could still be recovered via the medial temporal lobe structures that are intact. We believe this interpretation is neither the only, nor the best, available for data from these intriguing cases.