Episodic and semantic memory influences on picture naming in Alzheimer’s disease

Abstract

This study investigated the relationship between semantic and episodic memory as they support lexical access by healthy younger and older adults and individuals with Alzheimer’s disease (AD). In particular, we were interested in examining the pattern of semantic and episodic memory declines in AD (i.e., word-finding difficulty and impaired recent memory) vis-à-vis more preserved remote memories. We administered a picture naming task in which the episodic period of the pictures and whether the pictured items were unique to one period or commonly used across periods were varied. Groups of younger adults (N = 40), healthy older adults (N = 20) and older adults with AD (N = 18) were asked to name drawings of objects in four conditions: dated unique, contemporary unique, dated common, and contemporary common. The results indicated that all participants named items that were common to both episodic periods more successfully than items unique to one period. An interaction was observed such that the healthy older and AD groups were more successful in retrieving names of objects presented in the dated compared to contemporary unique conditions, whereas the younger adults showed the reverse pattern. These results indicate that naming ability is affected both by the cumulative frequency of using an item over a lifetime and by when an item was first acquired. The findings support a theoretical stance which proposes an enduring reciprocal link between semantic and episodic memory. This theoretical relationship has practical implications for the development of intervention strategies when interacting with persons who have AD.

Introduction

Anomia, or word-finding difficulty, is an early appearing symptom of Alzheimer’s disease (AD). Explanations for the source of this word-level impairment have appealed to the diminished integrity of semantic representations in memory (e.g., Garrard et al., 2005, Hodges et al., 1992), to difficulty accessing those representations due to the weakening of neural network connectivity and/or cognitive processing limitations (e.g., Bonilla and Johnson, 1995, Kempler et al., 1995), or to both “storage” and “access” deficits (e.g., Chenery, Murdoch, & Ingram, 1996). Other lexical factors that have been found to reliably influence naming ability in AD include word frequency, age of acquisition, and familiarity (e.g., Silveri et al., 2002, Taylor, 1998).

One promising area of investigation that has received less attention until recently is the relationship between difficulties persons with AD have in retrieving information from semantic memory and declines in their episodic memory (Nebes, 1992). Whereas semantic memory holds general (time-independent) knowledge of the world (e.g., who Abraham Lincoln was) and lexical–semantic information (e.g., the attributes of “tree”), episodic memory holds information linked to a specific event and period of time in one’s past (e.g., the day one graduated from high school) (Tulving, 2002). Within the domain of episodic memory in AD, recent episodic memories are more negatively affected than remote memories (Lipinska et al., 1992, Sadek et al., 2004, Sagar et al., 1988, Sagar et al., 1991, Small and Perry, 2005). This relative sparing of remote memory has been referred to as a temporally graded memory impairment in AD (Westmacott, Freedman, Black, Stokes, & Moscovitch, 2004). The neuropathological profile leading to this pattern of memory impairments indicates more extensive abnormalities in the hippocampal formation (supporting recent episodic memory) than in the association cortices (supporting semantic memory and remote episodic memory). The question addressed in the present study is to what extent these memory systems are interdependent, and whether successful retrieval of an object’s name from semantic memory might interact with episodic information related to that object.

Previous research provides evidence for a strong association between certain episodic and semantic memories (Kazui et al., 2003, Ratcliff and McKoon, 1986, Snowden et al., 1999, Tulving, 1986). For example, Westmacott and Moscovitch (2003) reported that younger adults’ retrieval of information from semantic memory (famous names) was influenced by the autobiographical significance of the names (episodic memory). In another study, Westmacott, Black, Freedman, and Moscovitch (2003) presented the same materials to persons with AD. Their initial analysis failed to find a similar association between retrieval of famous names and autobiographical significance of those names. However, because the number of famous names that AD participants could recognize overall was smaller than that of the controls, the restricted range of stimuli may have led to the observed floor performance and null effects. To evaluate this possibility, the authors re-analyzed only the famous names that AD participants’ recognized, and they found that the AD group showed similar benefits from autobiographical significance when recalling famous names. In a later study, Westmacott et al. (2004) reported evidence of temporally graded semantic memory loss in AD, with famous names and words from the remote past being more preserved than recent names and words. These findings coincide with research on autobiographical memories which shows a “reminiscence bump” of increased recall of events by elderly participants from their early adulthood (i.e., 20–30 years of age) (Berntsen and Rubin, 2002, Fromholt et al., 2003, Sagar et al., 1991).

In the present study, we investigated whether the retrieval of object names from semantic memory by persons with AD would be affected by episodic memory, and in particular, the “episodic” depiction of the object to be named. Namely, will objects depicted from an earlier period in the participants’ lives be more readily named by persons with AD than objects depicted in their contemporary form? An affirmative response to this question was predicted based on the temporally graded memory deficits observed in AD, as well as on evidence that early acquired words are more robust in the face of brain impairments than later acquired words (Hirsh and Funnell, 1995, Taylor, 1998). This prediction was also motivated by Poon and Fozard’s (1978) study comparing younger and older adults’ naming of objects drawn from different parts of the 20th century. These authors reported that older adults named objects more readily when they were unique to an earlier era, whereas younger adults showed facilitated naming of objects unique to contemporary times.

Most research has examined retrograde memory deficits in AD in terms of autobiographical memories. The research findings of Westmacott and others, however, have demonstrated a strong interdependency between episodic and semantic memory. This relationship is not unexpected when one considers that much of semantic knowledge is acquired and reinforced through autobiographical experience. As Snowden, Griffiths, and Neary (1994) state, “autobiographical experience links the word with the object…meaning is invested by the experience” (p. 284). Although it may be the case that memories of facts often become generic (i.e., lose their episodic character) as they become used over and over in different contexts (Tulving, 2002), certain dimensions (e.g., visual features) may be retained according to one’s episodic experiences at a particular time in life (McClelland et al., 1995, Westmacott and Moscovitch, 2003). For example, one may first learn the word “wise” when hearing a story about grandparents, but through subsequent exposure to and use of this word in other contexts it takes on additional meanings. In this sense, implicitly stored episodic traces (e.g., wise ∼ grandparents) may be associated with much of the vocabulary we know. It is conceivable, then, that certain episodic traces may facilitate access to a lexical form, especially if those traces encode attributes of the word across time. In this study, we examined whether the visual appearance of a lexical item in different episodic periods of the 20th century would influence older adults’ retrieval of the word label from semantic memory. We proposed that recognizing and naming a familiar object may be facilitated by using a picture of an object that represents the object in a form when it was first learned and/or most frequently used (Poon & Fozard, 1978). While we did not measure the participants’ autobiographical experiences with each item, the set of items was selected from readily available public sources and was thus considered to consist of vocabulary terms common used by individuals who lived at given periods through the 20th century.

Similar to Poon and Fozard’s (1978) findings, when older persons with AD see an object in a form that they would have repeatedly seen earlier in their lives, they may be more successful in naming it compared to the same object appearing in its contemporary form (e.g., picture of a stove from the 1920’s vs. the 1990’s). Moreover, it is possible that the severity of the retrograde episodic memory decline in AD may, to an even greater extent than in normal older adults, affect the retrieval of information from semantic memory as it varies along episodic dimensions (i.e., the era from which the picture of an object is taken). The following are four hypotheses which we tested in an experiment involving a picture naming task administered to older adults with AD, older healthy adults, and younger healthy adults. The third and fourth hypotheses seek to replicate the findings of Poon and Fozard (1978) and are motivated by the different amounts of experience that older and younger adults have with dated vs. contemporary objects.

• In line with the temporally graded episodic memory deficits in AD (i.e., better memory for remote than recent information), naming of pictures of early 20th century objects is predicted to be better than that of later 20th century objects.

• The effect in (1) should be largest for lexical items that are common to both early and late periods because repeated use of these words over the lifespan is predicted to produce stronger representations and connections in semantic memory (McClelland et al., 1995, Westmacott et al., 2004).

• Older healthy adults will be better than younger adults at naming pictures from an earlier period when the items are unique to that period.

• Younger adults will be better than healthy older adults at naming pictures from a contemporary period when the items are unique to that period.