Effects of age on retrieval cue processing as revealed by ERPs

Abstract

The electrophysiological correlates of retrieval cue processing were investigated in healthy young (18–30 years) and older (63–75 years) subjects (n=16 per group). Retrieval orientation—the differential processing of cues according to the form of the sought-for information—and retrieval difficulty were manipulated in a factorial design. In separate study-test cycles, subjects studied either words or pictures, and performed a yes/no recognition memory task with words as the test items. ERPs elicited by correctly classified new words differed markedly according to study material in the young subjects, replicating previous findings. In the older subjects, this effect was smaller than in the young, and had a later onset and earlier offset. The scalp topography of the effect was however statistically indistinguishable in the two groups. These age-related ERP differences were unmodulated by task difficulty, and remained reliable when recognition performance was matched across the groups. By contrast, the magnitude and timing of ERP difficulty effects were unaffected by age. The findings suggest that older subjects are less able than young individuals to vary their processing of retrieval cues in response to different retrieval demands.

Introduction

Healthy ageing is associated with a decline in the ability to remember recently experienced events (Craik & Jennings, 1992; Light, 1991). This is found predominantly in tasks requiring intentional retrieval (Light & Singh, 1987), and is associated with a greater impairment of ‘recollection’ than ‘familiarity’ (Yonelinas, 2001) and, consequently, a greater difficulty in remembering details of the context in which items were encountered than in remembering the items themselves (Spencer & Raz, 1995). It remains uncertain to what extent this age-related impairment in episodic memory is due to changes in the capacity to encode new information (e.g. Craik & Rabinowitz, 1985; Glisky, Rubin, & Davidson, 2001; Perfect, Williams, & Anderton-Brown, 1995), as opposed to changes in processes operating at the time of retrieval (e.g. Burke & Light, 1981; Craik and McDowd, 1987, Craik and Rabinowitz, 1985).

With regard to retrieval, ageing might influence one or more of a number of functionally distinct component processes. Aspects of a prior event are reconstructed via an interaction between a ‘retrieval cue’ (either self-generated or provided by the environment) and the memory representation of the episode (Tulving, 1983). Rugg and Wilding (2000) identified three kinds of putative ’pre-retrieval’ process which precede this interaction. Firstly, ’retrieval mode’ establishes a cognitive state enabling events to be processed specifically as episodic retrieval cues (Wheeler, Stuss, & Tulving, 1997). Secondly, ‘retrieval effort’ mobilises attentional and processing resources in support of a retrieval attempt. Thirdly, ‘retrieval orientation’ biases the processing of retrieval cues so as to meet the demands of a specific retrieval task (for example Robb & Rugg, 2002; Rugg, Allan, & Birch, 2000). Pre-retrieval processes are held to operate regardless of whether a retrieval attempt is successful. Other, ‘post-retrieval’ processes are then brought to bear whenever a cue has initiated contact with stored information about a prior episode (‘synergistic ecphory’; Tulving, 1983). These post-retrieval processes support the representation and subsequent evaluation of retrieved episodic information.

It has been suggested previously that older subjects have difficulty with episodic retrieval because they cannot effectively specify the necessary mental operations (‘self-generation’ of retrieval cues) unless the test stimuli provide ‘environmental support’ for such operations (Craik, 1983, Craik, 2000). As has frequently been pointed out, this is consistent with the generally better performance of older adults on recognition compared with free recall tests, and on data-driven indirect memory tests than on both of the latter (La Voie & Light, 1994). However, other data suggest that older adults’ difficulty may instead be with deliberate recollection, regardless of the strength of external cues (Park & Shaw, 1992; see Light, 1996). This fits with evidence that although overall recognition performance is relatively preserved with increasing age, this reflects a predominance of familiarity-based responding (Yonelinas, 2001). Thus even when retrieval cues are identical to to-be-retrieved items, recollection is markedly impaired, indicating a possible difficulty in the processing of episodic retrieval cues themselves.

To study the different categories of retrieval processes, it is necessary to distinguish them from each other and from factors that influence encoding. Scalp-recorded event-related potentials (ERPs) are particularly valuable in this context, allowing the dissociation of the neural correlates of pre-retrieval processes from those of ‘retrieval success’, and providing detailed information about their temporal characteristics. ERP studies of the effects of ageing on episodic retrieval have so far focused on processing associated with retrieval success, and their results provide some, although not unequivocal, support for the notion that its neural correlates differ according to age (for review see Friedman, 2000). Age-related differences have been reported in two principal ERP correlates of retrieval success, both of which take the form of positive-going deflections in the waveforms elicited by correctly identified old items relative to correctly identified new items. The ‘left parietal old/new effect’, which onsets around 400–500 ms post stimulus, persists for about 400 ms, and is maximal over the left parietal scalp, is thought to be a neural correlate of the retrieval and representation of episodic information (e.g. Wilding & Rugg, 1996). A reduction in its magnitude in older compared with younger adults has been found in some studies using simple recognition judgements (Rugg, Mark, Gilchrist, & Roberts, 1997; Swick & Knight, 1997), but not where procedures are employed to ensure that ERP differences between young and older do not simply reflect a greater reliance in the latter on familiarity-based responding (e.g. Mark & Rugg, 1998; Trott, Friedman, Ritter, & Fabiani, 1997). The second, ‘right frontal’ old/new effect onsets later, exhibits a more sustained time-course (>1 s), and has a maximum over the right frontal scalp. This effect is apparent particularly in tests of source memory (see Rugg, Otten, & Henson, 2002). The right frontal effect also has been reported to be attenuated in older individuals, and it has been claimed that this reflects an age-related reduction in the engagement of frontally-mediated evaluative and monitoring processes (e.g. Trott et al., 1997; Wegesin, Friedman, Varughese, & Stern, 2002; but see Mark & Rugg, 1998 for contradictory findings).

The present study goes beyond previous work to investigate the possibility that ageing affects the neural correlates of retrieval orientation. If the differential processing of retrieval cues is attenuated or altered in older adults, this could contribute to age-related retrieval difficulties independently of any changes in the processes that follow successful retrieval. The present study takes as it starting point recent findings in young subjects which demonstrate that ERPs elicited by recognition memory test items vary according to the nature of the sought-for information (Herron & Rugg, 2003; Robb & Rugg, 2002). In Robb and Rugg’s (2002) experiment, participants studied lists of pictures or words, followed in each case by recognition memory tests in which all items were words. ERPs elicited by unstudied items (i.e. ‘new’ test words) were more positive-going when participants were trying to retrieve words than pictures (see Herron & Rugg, 2003, for replication and extension of this finding). The effect onset around 250 ms post-stimulus, lasted for around 1000 ms, and was diffusely distributed over the scalp. Since the effect was elicited by correctly classified new items, it could not have been due to differences in the content of retrieved information. Importantly, Robb and Rugg (2002) factorially crossed study material (and task) with a manipulation of task difficulty, and demonstrated that the effect of material did not interact with the effects of difficulty. Accordingly, they concluded that the material effect was a neural correlate of the adoption of different retrieval orientations, reflecting differences in the processing accorded ostensibly identical retrieval cues.

The design and procedure of the present study closely followed Robb and Rugg (2002). ERPs elicited by recognition test items (words) were compared after study of lists of words or pictures, in both ‘easy’ and ‘hard’ tests. By including a difficulty manipulation, it was possible to unconfound effects of age from those of difficulty on group differences in ERP effects (c.f. Morcom, Good, Frackowiak, & Rugg, 2003; see Rugg & Morcom, in press, for discussion of this issue). The principal question was whether, relative to those of younger subjects, the ERPs of older subjects would show evidence of a reduced capacity to process retrieval cues differentially according to the requirements of the two recognition tasks (retrieving pictures versus retrieving words). A secondary question was whether differences between ERP old/new effects in the two age groups, and the influence on such differences of study material, might shed light on the relationship between age-related effects on retrieval orientation and on retrieval success.