Aging memory for pictures: Using high-density event-related potentials to understand the effect of aging on the picture superiority effect

1. Introduction

For more than 30 years the picture superiority effect has demonstrated that subjects are more likely to remember items if they are presented as pictures versus words (Shepard, 1976; see Mintzer & Snodgrass, 1999, for review). Recent research has focused on understanding the memorial processes that underlie this picture superiority effect (Ally & Budson, 2007; Langley et al., in press). Investigations into episodic memory from a dual-process perspective have provided fairly convincing evidence that familiarity and recollection provide independent bases of the recognition of studied items (Yonelinas, 2002; Woodruff, Hayama, & Rugg, 2006). Familiarity has been described as an acontextual sense that a test item has been seen before, whereas recollection has been described as the retrieval of contextual information about a studied item at test (Woodruff et al., 2006). Using high-density event-related potentials (ERPs), Ally and Budson (2007) suggested that pictures enhanced recollection compared to words, serving as the neural basis of the picture superiority effect in young adults. Older adults also demonstrate the picture superiority effect (Park, Puglisi, & Sovacool, 1983; Winograd, et al., 1982). However, many studies have suggested that recollection is impaired in the course of healthy aging (Craik and Jennings, 1992; Light, 1991). The primary goal of the current study was to investigate the effect of aging on the neural correlates of the picture superiority effect. We hoped to determine whether the increased memory for pictures in older adults was due to enhanced recollection of studied items, or whether it was attributable to enhancement of other memorial processes, such as familiarity or post-retrieval verification and monitoring.

Although it is widely acknowledged that aging has an adverse effect on episodic memory (Balota, Dolan, & Duchek, 2000; Craik and Jennings, 1992; Light, 1991; Smith, 1996), exactly how aging affects the processes of recognition memory has been the subject of some debate. Researchers have demonstrated that recollection remains intact in some groups of healthy older adults (Cabeza, et al., 2002; Davidson & Glisky, 2002; Duarte, et al., 2006). However, the majority of studies report that recollection is generally impaired with age, while familiarity is spared (Daselaar, et al., 2006b; Howard, et al., 2006; Jacoby, 1999; Jennings & Jacoby, 1993, 1997; Rybash & Hoyer, 1996; Spencer & Raz, 1995; Titov & Knight, 1997; Yonelinas, 2001; but see Light, Prull, La Voie, & Healy, 1999).

Recent studies have used different methodologies to better understand familiarity in the context of aging (Duarte et al., 2006; Toth and Parks, 2006). The majority of the tasks used in the laboratory to estimate familiarity rely on the ability to effectively report a confidence judgment or one’s subjective feeling of familiarity. It has been suggested, however, that perhaps older adults are impaired in their ability to access or report on the phenomenological experience of familiarity (Prull et al., 2006). Yet familiarity is generally thought to be an automatic/obligate process, which likely occurs physiologically with or without one being able to report on the phenomenological experience of familiarity (Ally & Budson, 2007; Ecker, et al., 2007; Toth, 1996; Yonelinas, 2002; Yonelinas & Jacoby, 1996). In other words, the neural correlate of familiarity is likely to occur in older adults even if they are impaired in their ability to report the subjective feeling of familiarity. The use of a physiologic measure of familiarity might eliminate the need to use subjective reports.

Due to their excellent temporal resolution, ERPs have recently been used to dissociate the processes of familiarity and recollection with a variety of different stimuli and experimental manipulations (Ally & Budson, 2007; Duarte et al., 2006; Li, Morcom, & Rugg, 2004; Woodruff et al., 2006). In addition to helping us understand when specific cognitive processes may occur with precise time sensitivity, ERPs can help to elucidate the duration and interaction of these cognitive processes. The literature reports that ERPs elicited by correctly classified old items (studied) are more positive in nature than ERPs elicited by correctly classified new items (unstudied) (Friedman & Johnson, 2000). This ERP phenomenon is referred to as the “old/new effect.” Researchers suggest that a neural correlate of familiarity is an enhanced old/new effect peaking around 400 ms at bilateral frontal electrode sites, that appears to precede controlled attempts by the individual to recollect information (Curran, 2000; Friedman & Johnson, 2000; Rugg, et al., 1998). Some researchers have referred to this component as the “N400” or “FN400;” we will refer to this component as the early frontal effect. Previous research has shown that more familiar unstudied test items elicited a larger early frontal effect than unfamiliar unstudied items (Curran, 2000; Goldmann et al., 2003). Further, Duzel, et al. (1997) asked subjects to make either a “remember” judgment if they remembered specific details of an item’s presentation at study, or a “know” judgment if they simply had a feeling of “knowing” that an item was shown at study but could not recollect details of the item’s presentation. They found that responses associated with “know” judgments elicited a greater early frontal effect than “remember” judgments. Furthermore, a recent ERP study using confidence ratings suggested that the magnitude of the early frontal effect varied directly with familiarity strength (Woodruff et al., 2006).

Numerous studies have investigated the neural correlate of recollection. The literature describes this correlate as an enhanced old/new effect occurring maximally at left parietal regions between 500 and 800 ms after test stimulus onset. Researchers have often referred to this component as the “LPC;” we will refer to this component as the parietal effect. This effect is greater for items correctly identified as previously studied and is relatively insensitive to alterations in familiarity (Woodruff et al., 2006). Further, Duzel and colleagues (1997) found that the parietal effect was enhanced by responses associated with “remember” judgments compared to “know” judgments. Some researchers have argued that the parietal effect indexes the amount of information retrieved (Fjell et al., 2005; Vilberg et al., 2004). Although it remains unclear exactly what role the parietal cortex is playing in recognition memory, the parietal activity may reflect the reactivation of the stored memory representation or the actual matching of representations stored in memory with perceptual representations of the test items (Addis & McAndrews, 2006; Ally & Budson, 2007; Schnyer, McNicholls, & Verfaellie, 2005; Wagner, et al., 2005).

When recollection is poor, difficult, or additional information is needed at retrieval, subjects must engage in post-retrieval processing. It has been suggested that post-retrieval monitoring and verification operates on the product of a retrieval attempt by holding information in working memory while it is evaluated for task relevance (Achim & Lepage, 2005). ERP studies suggest that a late right frontal old/new effect between 1000 and 1800 ms is associated with the ongoing evaluation and monitoring of the product of the retrieval attempt, particularly when the contents of memory are evaluated for specific features or details (Allan, Wilding, & Rugg, 1998; Rugg & Wilding, 2000; Wilding & Rugg, 1996). More recently, investigators have reported evidence of this late frontal effect despite unsuccessful recollection, and suggested that in addition to ongoing evaluation, this activity may reflect an executive search function of the frontal lobes directing subsequent retrieval attempts (Ally & Budson, 2007; Budson, et al., 2005; Goldmann, et al., 2003; Li et al., 2004).

Research focusing on ERPs and recognition memory in older adults has mainly focused on the parietal and late frontal effects. Most of these studies have reported an attenuation of the parietal effect in older adults (Fjell, et al., 2005; Joyce, et al., 1998; Morcom & Rugg, 2004; Nielson-Bohlman & Knight, 1995; Rugg, et al., 1997; Senkfor & Van Petten, 1998; Swick & Knight, 1997; but see Mark & Rugg, 1998; Trott, et al., 1997), consistent with behavioral reports of impaired recollection. There has been a long-standing suggestion that a decline in the attentional resources allocated at encoding (Park, et al., 1989; Salthouse, 1994) may be responsible for memories with less supportive or contextual information than those of younger adults (Craik, et al., 2000; Li et al., 2004; Light, 1996). One explanation for diminished recollection, based on ERP findings in older adults, is that there is an age-related decrease in the quantity or quality of information retrieved (Fjell et al., 2005).

The majority of ERP studies have suggested that the late frontal effect associated with recognition memory is relatively unaffected by age when source information is not needed at test (Li et al, 2004; Mark & Rugg, 1998; Morcom & Rugg, 2004; Wolk, et al., 2005). These findings are consistent with imaging studies investigating episodic memory in older adults (Cabeza et al., 2002; Madden, et al., 1999). In fact, Cabeza et al. (2002) reported that older adults may actually engage bilateral frontal regions as opposed to the right hemispheric dominance observed in younger adults, presumably to compensate for impairment of other brain processes.

To our knowledge, only two studies have commented on the early frontal effect and familiarity in healthy older adults. Morcom and Rugg (2004) found no difference in ERP amplitude during the 300 to 500 ms time interval between older and younger adults, suggesting that familiarity is intact in older adults. However, using a remember/know paradigm to understand memory for grayscale pictures, Duarte et al. (2006) reported that older adults did not exhibit the neural activity associated with familiarity during a 100 to 500 ms post-stimulus interval that was observed in younger adults.

To achieve the overall goal of understanding the effect of aging on the picture superiority effect, we compared the behavioral performance and the electrical activity of the brain in healthy older adults to that of the younger adults from Ally and Budson (2007). Behaviorally, we expected that like younger adults, older adults would also demonstrate the picture superiority effect. For the ERP results, we expected that the early frontal effect would be similar for younger and older adults for both pictures and words, reflecting older adults’ intact ability to use familiarity when making memorial judgments. We also expected that the late frontal effect would be similar in both groups or possibly enhanced for the older adult group, due to older adults’ potential need for additional post-retrieval processing compared to younger adults. However, we predicted that the parietal effect would be significantly diminished for pictures and words for the older adults compared with the younger adults, reflecting the older adults’ relatively impaired recollection. Finally, we hypothesized that the increased memory for pictures in older adults would be due to the ability of pictures to enhance the early and late frontal effects relative to words. This would reflect and increased ability for older adults to rely on familiarity and post-retrieval processing to compensate for impaired recollection of pictures compared to words.

2. Materials and Methods

3. Results

Previous research has shown that when study and test modality are not matched, such as in the picture-word and word-picture conditions, the neural correlates of recognition memory are profoundly affected (Ally & Budson, 2007; Schloerscheidt & Rugg, 2004). These two conditions were not relevant to the current investigation, and therefore were not presented. To better understand the effect of aging on the neural correlates of the picture superiority effect, only the results of the word-word and picture-picture conditions were analyzed.

4. Discussion

The present study set out to better understand the effect of aging on the neural basis of the picture superiority effect. To accomplish this, we examined the recognition memory performance and the electrical activity of the brain during recognition of pictures and words in healthy older adults compared to younger adults. Our behavioral results showed that although both groups demonstrated a clear picture superiority effect, the older adults benefited from pictures to a greater extent than the younger adults. Memory accuracy was similar, but numerically better, for the older adult group compared to the younger adult group for pictures. However, for words, memory accuracy trended towards worse performance for the older adults compared to the younger adults. The ERP results helped to explain these behavioral findings. We found no differences in the early frontal effect, the parietal effect, or late frontal effect between-subjects for the picture-picture condition and the scalp topographies showed a similar pattern of activity throughout the sampling epoch for both younger and older adults. However, the results revealed several significant differences between the two groups for the word-word condition.

Between-group differences for the word-word condition were evident during the early frontal effect 300 to 500 ms time interval. Here, there was a larger old/new effect for the younger adults than for the older adults at left superior frontal electrode sites. The scalp topographies for the word-word condition showed a robust old/new effect for the younger adult group at bilateral frontal regions from approximately 300 to 450 ms. In contrast, the scalp topographies for the older adults during this time interval showed a more right lateralized old/new effect that was not as robust as the young adults. This time interval has been previously associated with familiarity, and the results suggest that perhaps the process of familiarity for words is relatively impaired in the older adult group. To further aid our understanding of the picture superiority effect in older adults, we turn now to our within-subject comparisons between the word-word and picture-picture conditions. Here, we found that pictures were able to “restore” the impaired old/new differences at bilateral superior frontal electrode sites observed with words.

We also found distinct parietal effect differences between the two groups for the word-word condition. Here, the old/new effect was larger at left parietal electrode sites for the younger adults compared to the older adults. However, the old/new effect was larger for the older adults compared to the younger adults at right occipital electrode sites. The scalp topographies showed that showed that the parietal effect begins earlier and is much more robust for the younger adults for the word-word condition. In addition, the older adults initially showed weak parietal activity and then showed pronounced occipital activity. This increase in parieto-occipital activity during a recognition memory task is consistent with recent fMRI (Daselaar et al., 2006a, 2006b; Yonelinas, et al., 2005) and magnetoencephalographic (Aine et al., 2006) evidence in healthy older adults. We believe that there are three likely hypotheses – which may be different facets of the same phenomena – as to what might be contributing to this robust posterior activity during the word-word condition.

First, this parieto-occipital activity might reflect differences in the process of recollection for older and younger adults. Li et al. (2004) suggested that older and younger adults might differ in the types of information they attempt to retrieve. These authors speculated that whereas younger adults tend to rely on retrieval of abstract or conceptual information, older adults tend to rely on retrieval of visual and perceptual information. This suggestion is supported by research demonstrating that aging affects conceptual, but not perceptual memory processes (Stuart et al., 2006). If older adults are attempting to retrieve more perceptual-type information, this may account for the shift to more posterior or occipital activity during recollection.

A second hypothesis for the enhanced parieto-occipital activity is that older adults are attempting to reactivate a visual representation or mental image of the studied item. Previous imaging research has shown that primary visual areas are activated when subjects engage in mental imagery (Kosslyn et al., 1995), and these regions are reactivated at retrieval (Johnson & Rugg, in press). If older adults are aware of their degraded verbal memory, perhaps mental imagery might be used as a strategy to help boost their verbal memory.

Finally, the enhanced occipital activity might reflect an increased activation of familiarity-based areas of the brain. Daselaar and colleagues (2006b) used fMRI to dissociate familiarity and recollection within posterior brain regions and found that activations associated with familiarity in parieto-occipital regions were preserved in aging, whereas activations associated with recollection in parieto-temporal regions were attenuated. In fact, the authors reported that parieto-occipital regions associated with familiarity actually showed increased activation, perhaps compensating for reduced activation of regions associated with recollection. The close proximity of the familiarity-related parieto-occipital region to secondary visual cortex may implicate this region in visuo-perceptual functions (Daselaar et al., 2006a, 2006b). This is consistent with Yonelinas (2002) who suggested familiarity is more dependent on perceptual processes. Whether one or all of these related hypotheses are correct remains unclear. Future studies examining these possibilities could help to explain the robust parieto-occipital activity seen in older adults during the 500 to 800 ms time interval.

Both a diminished parietal effect and impaired recollection of words have been reported in older adults (Daselaar et al., 2006a, 2006b; Fjell et al., 2005; Joyce et al., 1998; Madden et al., 1999; Morcom & Rugg, 2004; Rugg et al., 1997; Swick & Knight, 1999). Fjell et al. (2005) found that not only was the parietal effect for words diminished in older adults aged 70 to 82, but also that there was a linear decline in the parietal effect evident in middle-aged subjects from 46 to 69 years of age. The authors suggested that the amount of information retrieved likely decreases with age. A long-standing hypothesis for impaired recollection in older adults is that there is an age-related decline in attentional resources at encoding (Park et al., 1989; Salthouse, 1994). Indeed, if attentional resources were not effectively allocated during encoding, the retrieval of the studied item would be difficult. However, the retrieval of distinctive stimuli such as pictures may not have the same attentional requirements as word stimuli and may make it easier for older adults to recollect them. While we do not believe older adults are incapable of using recollection to support their memorial decisions regarding words, it is likely that the overall quantity and/or quality of their recollective experience is greatly diminished in this modality compared to pictures.

The late frontal effect has been associated with post-retrieval monitoring and verification, and is typically associated with a robust old/new effect at right prefrontal electrode sites (Allan, Wilding, & Rugg, 1998; Rugg & Wilding, 2000; Wilding & Rugg, 1996). We found no between-group differences at right frontal regions during the 1000 to 1800 ms time interval for either the word-word or picture-picture conditions. These results are consistent with numerous studies suggesting that the late frontal effect appears to be less affected by age (Daselaar, et al. 2003; Li et al., 2004; Mark & Rugg, 1998; Morcom & Rugg, 2004; Wolk et al., 2005). Recent research has suggested that the late frontal effect may reflect other processes of the frontal lobes not related to post-retrieval monitoring, such as decision making and memory confidence (Dobbins & Han, 2006; Fleck et al., 2006). Ally and Budson (2007) speculated that the late frontal effect may also reflect activity associated with executive processing that helps direct conscious memory retrieval or multiple memorial search attempts. For the older adults in the current study who appear to have impaired recollection or words, perhaps the late frontal activity reflects prefrontal regions working to confirm a response based on other processes such as familiarity. Indeed, the frontal lobes have been implicated in response inhibition (Shimamura, 1995), which is critical to the suppression of responding based on familiarity alone. Numerous investigations have found that the frontal lobes are involved in distinguishing between identical versus highly similar and familiar items, and thus are important in avoiding false recognition (Budson et al., 2002, Delbecq-Derouesne et al., 1990; Henson et al., 1999a; Henson et al., 1999b; Melo et al., 1999; Parkin et al., 1996; Parkin et al., 1999; Rapcsak et al., 1998; Rapcsak et al., 2001; Schacter et al., 1996).

In conclusion, the results of the present investigation show that older adult memory accuracy for pictures was comparable to that of younger adults. However, memory for words was somewhat impaired. Helping to explain these findings, our ERP results suggested that pictures enable older adults to engage memorial neural systems in a very similar manner to young adults. The early frontal effect, parietal effect, and late frontal effect, were all indistinguishable between older and younger adults for pictures. In contrast, for words, the early frontal and parietal effects were significantly diminished for the older adults compared to the younger adults. These two old/new effects have been linked to familiarity and recollection, respectively, and our results suggest that these processes are impaired for word-based memory in the course of healthy aging. Our results also demonstrated that the late frontal effect is indistinguishable for older adults and younger adults for both pictures and words. The late frontal old/new effect has been linked to several retrieval and post-retrieval based processes that may reflect executive involvement in memory related searches and decisions. The findings of this study suggest that pictures can allow older adults to compensate for their impaired memorial processes, and may allow memorial components to function more effectively in older adults. These findings help us to understand healthy memory in older adults, and may serve as a model to understand how memory is affected in diseases of aging such as Alzheimer’s. Future studies will be able to investigate whether older adults and memory-disordered patients can create their own pictures – by using mental imagery – to compensate for their impaired memory for words.