Age-related Differences in Brain Activity during True and False Memory Retrieval Nancy A. Dennis1, Hongkeun Kim2, and Roberto Cabeza1 Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/20/8/1390/1759633/jocn.2008.20096.pdf by guest on 18 May 2021 Abstract & Compared to young adults, older adults show not only reduction is consistent with age-related deficits in recollec- a reduction in true memories but also an increase in false tion, whereas the retrosplenial increase suggests compensa- memories. We investigated the neural bases of these age ef- tory recruitment of alternative recollection-related regions. fects using functional magnetic resonance imaging and a false During incorrect recognition of critical lures (false memory), memory task that resembles the Deese–Roediger–McDermott older adults displayed stronger activity than young adults in (DRM) paradigm. Young and older participants were scanned the left lateral temporal cortex, a region involved in semantic during a word recognition task that included studied words processing and semantic gist. Taken together, the results sug- and new words that were strongly associated with studied gest that older adults’ deficits in true memories reflect a de- words (critical lures). During correct recognition of studied cline in recollection processes mediated by the hippocampus, words (true memory), older adults showed weaker activity whereas their increased tendency to have false memories re- than young adults in the hippocampus but stronger activity flects their reliance on semantic gist mediated by the lateral than young adults in the retrosplenial cortex. The hippocampal temporal cortex. & INTRODUCTION lure, and this tendency is stronger for older than for young adults (Watson, McDermott, & Balota, 2004; Balota One of the most frequent cognitive complaints in older et al., 1999; Tun et al., 1998). Age-related increases in false adults is poor memory for everyday events. Supporting memories have been attributed to both a deficit in mem- this casual observation, research has shown a decline in ory for item-specific details (LaVoie & Faulkner, 2000; Tun episodic memory performance across the lifespan (for et al., 1998; Koutstaal & Schacter, 1997; Norman & reviews, see Prull, Gabrieli, & Bunge, 2000; Zacks, Hasher, Schacter, 1997; Spencer & Raz, 1995) and to an increased & Li, 2000). Furthermore, in addition to exhibiting deficits reliance on semantic gist (Balota et al., 1999; Tun et al., in true memories, older adults are also more prone to 1998). false memories than young adults (Tun, Wingfield, Rosen, Behavioral studies have demonstrated that older adults & Blanchard, 1998; Koutstaal & Schacter, 1997; Norman are impaired in memory for item-specific details, and & Schacter, 1997). Although the neural bases of age- neuroimaging evidence has associated this deficit with a related deficits in true memory has been previously in- dysfunction of the hippocampus (Cabeza, 2006). Com- vestigated using functional neuroimaging (for a review, pared to young adults, older adults show difficulties in see Dennis & Cabeza, 2008), to our knowledge, no neu- retrieving the context (Bayen, Phelps, & Spaniol, 2000; roimaging study has examined the age-related changes Spencer & Raz, 1995; Park & Puglisi, 1985) and percep- in brain activity that are associated with false memories. tual features of studied items (Bastin & Van der Linden, This was the goal of the present functional magnetic 2003). Retrieval of such item-specific details is defined as resonance imaging (fMRI) study. ‘‘recollection,’’ whereas memory in the absence of item- We investigated true and false memory retrieval in the specific contextual information is referred to as ‘‘fa- scanner using a false memory task that resembles the miliarity.’’ Neuropsychological and neuroimaging studies Deese–Roediger–McDermott (DRM) paradigm (Roediger indicate that recollection and familiarity are two separate, & McDermott, 1995; Deese, 1959). In the typical DRM dissociable mechanisms and depend on different brain paradigm, participants study lists of words in which all the regions (Yonelinas, 2002). In the case of true memories, words in the list are semantically related to a word that aging has been shown to impair recollection more so is not presented (the critical lure). At test, participants than familiarity (Bastin & Van der Linden, 2003; Davidson show a tendency to falsely recall or recognize the critical & Glisky, 2002; Parkin & Walter, 1992). Lesion and func- tional neuroimaging studies have associated recollec- 1Duke University, Durham, NC, 2Daegu University, Daegu, South tion with the hippocampus (Eichenbaum, Yonelinas, & Korea Ranganath, 2007; Yonelinas, 2002; Aggleton & Brown, D 2008 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 20:8, pp. 1390–1402 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/jocn.2008.20096 by guest on 24 September 2021 1999), which is a region that shows significant atrophy in coded item-specific traces should result in the successful older adults (Raz, Rodrigue, Kennedy, & Acker, 2007; Raz recognition of studied items, whereas recovery of gist et al., 2005; Raz, Rodrigue, Head, Kennedy, & Acker, traces may result in the tendency to respond ‘‘old’’ not 2004). Linking behavioral and neuroscientific evidence, only to studied items but also to the nonstudied critical functional neuroimaging studies have associated older lures, which strongly matches the semantic theme of the adults’ recollection deficits with reductions in hippocam- studied list. The fuzzy trace theory can account for age- pal activity (Daselaar, Fleck, Dobbins, Madden, & Cabeza, related increases in false memories by postulating that 2006; Cabeza et al., 2004). older adults have a deficit in memory for item-specific Despite the aforementioned deficits in recollection traces but not for gist traces (Tun et al., 1998). and hippocampal functioning during episodic memory Given that semantic processes are relatively well pre- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/20/8/1390/1759633/jocn.2008.20096.pdf by guest on 18 May 2021 tasks, older adults compared to young adults often show served in older adults (Light, 1992; Light & Burke, 1988; increased recruitment of other brain regions such as fron- Salthouse, 1982), they may rely more on these processes tal lobes (e.g., Dennis, Daselaar, & Cabeza, 2006; Gutchess to compensate for deficits in episodic memory. Although et al., 2005; Cabeza et al., 2004) or cortical medial- relying on semantic gist may, at times, enhance true temporal lobe (MTL) regions (e.g., rhinal cortex: Daselaar, memories, it may also lead to false memories when lures Fleck, Dobbins, et al., 2006). The idea that older adults are semantically associated to encoded items. Functional compensate for declining processes with the recruitment neuroimaging studies have associated semantic pro- of resources not typically recruited by young adults is cessing with the left temporal cortex (for reviews, see a common finding in the aging literature (for a review, Thompson-Schill, Kan, & Oliver, 2006; Wise & Price, 2006). see Dennis & Cabeza, 2008). However, direct evidence Supporting the role of this area in processing semantic linking performance and neural compensation is scarce. gist, patients with semantic dementia (and damage to Although the current study will investigate age deficits in this region) are impaired at extracting and/or utilizing hippocampal-mediated recollection, it will also investigate semantic gist (Simons, Verfaellie, et al., 2005). Finally, the notion of compensation within the retrieval network, there is also some evidence that older adults show en- specifically related to recollection and retrieval of item- hanced activation in this region during lexical decision specific details. tasks (Whiting et al., 2003; Madden et al., 2002). In As noted above, however, older adults’ memory impair- sum, increased false memories in older adults may reflect ments are assumed to reflect not only a deficit in memory greater reliance on semantic gist processes mediated by for item-specific details associated with true memories the left temporal cortex. but also greater reliance on semantic gist associated with The current study used fMRI and a categorized word- false memories (Balota et al., 1999; Kensinger & Schacter, list task, which, similar to the DRM paradigm, has been 1999; Koutstaal & Schacter, 1997). According to fuzzy studied extensively in the context of false memories and trace theory, two different kinds of memory traces are fuzzy trace theory (Brainerd & Reyna, 2007; Budson created during encoding: item-specific traces and gist et al., 2006; Brainerd, Wright, Reyna, & Mojardin, 2001; traces (Schacter, Verfaellie, & Pradere, 1996; Brainerd & Brainerd, Reyna, & Mojardin, 1999) in order to investi- Reyna, 1990). Item-specific traces retain the distinctive gate the effects of aging on retrieval activity associated features of the individual items, whereas gist traces retain with both true and false memories. As illustrated by only the general meaning of the event, lacking any Figure 1, on each encoding trial, participants studied a perceptual details or information pertaining to the encod- ‘‘mini word-list’’ comprising four instances (e.g., horse, ing event.1 In the case of DRM lists, item-specific traces chicken, sheep, goat) of a semantic category (e.g., farm are stored for the details associated with each item in the animals). During the memory test,