Fluency affects source memory for familiar names in younger and older adults: Evidence from event-related brain potentials

Highlights

• We examined ERP correlates of source memory for familiar names.

• We observed an N400 effect when modality did not change from study to test.

• This effect occurred for correct source decisions only.

• Similar ERP results were obtained in older participants.

• Our findings suggest a prominent role of fluency for source memory.

Abstract

A current debate in memory research is whether and how the access to source information depends not only on recollection, but on fluency-based processes as well. In three experiments, we used event-related brain potentials (ERPs) to examine influences of fluency on source memory for famous names. At test, names were presented visually throughout, whereas visual or auditory presentation was used at learning. In Experiment 1, source decisions following old/new judgments were more accurate for repeated relative to non-repeated visually and auditorily learned names. ERPs were more positive between 300 and 600 ms for visually learned as compared to both auditorily learned and new names, resembling an N400 priming effect. In Experiment 2, we omitted the old/new decision to more directly test fast-acting fluency effects on source memory. We observed more accurate source judgments for repeated versus non-repeated visually learned names, but no such effect for repeated versus non-repeated auditorily learned names. Again, an N400 effect (300–600 ms) differentiated between visually and auditorily learned names. Importantly, this effect occurred for correct source decisions only. We interpret it as indexing fluency arising from within-modality priming of visually learned names at test. This idea was further supported in Experiment 3, which revealed an analogous pattern of results in older adults, consistent with the assumption of spared fluency processes in older age. In sum, our findings suggest that fluency affects person-related source memory via within-modality repetition priming in both younger and older adults.

Introduction

Memory for whether or not we have heard a certain name before is crucial for daily-life social interactions. Often, however, it is important not only to recognize that name, but also to remember in which context the name had occurred. Confusing such information can be embarrassing, and thus prejudicial to social interaction—an issue, which may be particularly disadvantageous for older adults, as social interactions form an important part of their everyday life (Leirer et al., 1990). The present study examined neural correlates of remembering context information for familiar names in both young and older adults.

Our episodic memory system is not only constituted by storage and retrieval of a particular item, but may also yield related information, such as the spatial and temporal context or perceptual information. Such detail is often termed source information as it informs about the conditions under which an item originated (Johnson et al., 1993, Mitchell and Johnson, 2009, Mollison and Curran, 2012). Typically, source memory has been examined by experimentally varying different aspects during encoding (e.g. such as an item was presented visually or auditorily during learning, see Wilding et al., 1995). Importantly for the present study, previous research suggests that source memory is substantially reduced in older adults (Anderson and Craik, 2000, Cansino et al., 2013, Spencer and Raz, 1995).

Source memory is commonly tested in addition to item memory, and in order to determine how a stimulus was remembered. Many current models of recognition memory assume two processes of retrieval: familiarity and recollection (Yonelinas, 2002). Familiarity is defined as a fast-acting context-free ‘feeling of knowing’, whereas recollection is assumed to enable the re-experience of an episode, and thus necessarily involves context or source information which is accessed more slowly (e.g. Mandler, 1980, Yonelinas, 2002). Until recently, there has been scant disagreement that source memory predominantly relies on recollection. Recent studies, however, suggest a prominent role of familiarity for source decisions as well (Diana et al., 2008).

Event-related potentials (ERPs) may contribute to specify the role of familiarity during the retrieval of source information. It is well-established that correctly recognized items (i.e., hits) elicit more positive ERP amplitudes than correct rejections (Rugg, 1995), a phenomenon known as the old/new effect. More specifically, a relatively early (approximately 300–500 ms) frontal old/new effect, sometimes referred to as FN400, has been related to familiarity (e.g. Curran, 2000). Alternatively, others related this effect to conceptual priming (Paller et al., 2007), and thus to implicit memory. A subsequent left parietal old/new effect (approximately 500–800 ms) is commonly related to recollection (Rugg and Curran, 2007). In line with the idea of recollection contributing to source memory, this left parietal positivity is larger for hits with correct as opposed to incorrect or missing source information (Senkfor and Van Petten, 1998, Wilding and Rugg, 1996). Finally, a late frontal old/new effect (LFE), starting at approximately 500–800 ms and lasting for several hundred milliseconds, is assumed to index retrieval monitoring (Cruse and Wilding, 2009, Wilding and Rugg, 1996).

A number of previous studies examined ERP old/new effects in older adults. Whereas some researchers observed similar parietal effects in young and older adults (e.g. Mark and Rugg, 1998, Osorio et al., 2009, Trott et al., 1997), others found this effect to be reduced or absent in older groups (Nessler et al., 2007, Swick and Knight, 1997, Wang et al., 2012). While these latter findings are generally in line with the suggestion of reduced recollection in older adults (e.g. Anderson and Craik, 2000), the discrepancy in ERP results may be partly related to task effects (Wolk et al., 2009). For instance, old/new effects of older and young adults become more similar with stronger environmental support (e.g., with longer word stems in a cued recall task; Angel et al., 2010). In addition, several studies found the earlier frontal old/new effect to be absent in older adults (in a deep but not in a shallow encoding condition, see Osorio et al., 2009, Wang et al., 2012, Wolk et al., 2009). As familiarity is typically assumed to be intact in this age group, this finding has been interpreted as reflecting qualitative differences in the neural signals reflecting familiarity in young versus older adults (Wang et al., 2012). Finally, with respect to late frontal effects, results are again mixed, with some studies showing similar LFEs in young and older adults (e.g. Mark and Rugg, 1998), while others reported age-related changes (Cansino et al., 2012, Swick et al., 2006).

A number of previous studies examined these ERP effects to clarify the specific processes related to source retrieval. Supporting the notion of beneficial familiarity effects on source memory in young adults, Diana and colleagues (Diana et al., 2011) observed a mid-frontal ERP effect for correct versus incorrect source judgments. This effect was restricted to a condition emphasizing unitization of an item with its context (such as representing an object with the background color as one single unit). It should be noted, however, that the frontal effect in this study occurred substantially later (i.e. 700–1000 ms) than the typical frontal old/new effect described above.

Several ERP studies suggested contributions of familiarity on source judgments independent of unitization (e.g. Addante et al., 2012). For instance, Peters and Daum (2009) reported a more positive FN400 for correct relative to incorrect source retrieval and for incorrect trials relative to correct rejections. Mollison and Curran (2012) observed that correct source memory for spatial location (but not frame color) was accompanied by an increase in behavioral measures of familiarity. Similarly, an FN400 modulation, which differentiated between correct versus incorrect source responses was only apparent when source memory for spatial information was tested.

While familiarity reflects processes of explicit memory, implicit processes, such as perceptual priming, have also been reported to affect source memory (Kurilla, 2011). Interestingly, both familiarity and priming are assumed to be related to fluency. Fluency is typically defined as a feeling of ease associated with a cognitive operation, and has been shown to influence a variety of judgments based on, for instance, conceptual or perceptual similarity (for overview, see Oppenheimer, 2008). It has been proposed that enhanced fluency manifests in priming and can concomitantly be experienced as (or translated into) familiarity during recognition (Hayes and Verfaellie, 2012). Others have suggested that fluency signals are implicit in nature and therefore separate from familiarity, but may (accidentally) contaminate explicit forms of memory, i.e. familiarity-based recognition (for a similar line of argument, see Lucas et al., 2012, Voss et al., 2012).

Independent of the exact interrelation of familiarity and priming, fluency-related processes may not only contribute to the accessibility of source information in younger adults, but also affect source memory in older age. Bastin et al. (2013) reported beneficial effects of familiarity-based processes on source memory in older adults when applying the unitization procedure previously introduced by Diana et al., 2008, Diana et al., 2011. Moreover, in an fMRI study, (Dulas and Duarte, 2012) observed activity in the perirhinal cortex, a structure associated with familiarity-based processes, to increase with enhanced source memory performance in older adults. This finding was interpreted as reflecting greater reliance on familiarity with higher age. Corresponding evidence from ERP studies is lacking.

Taken together, growing evidence suggests a prominent role of fluency for source memory, but the precise circumstances under which fluency may or may not enhance source decisions are not clear at present. We reasoned that using familiar names as stimulus material would be particularly promising in this context. In contrast to (abstract or concrete) nouns or objects, personal names are relatively pure referents, with little or no inherent meaning (Wittgenstein, 1922), pointing to unique mental representations for individual persons (Burton et al., 1990, Valentine et al., 1991). Fluency-based processes, such as modality-specific long-term repetition priming, easily activate these representations (e.g. Bruce and Valentine, 1985). Whereas priming has been investigated in a number of previous studies, source memory for person-related information has been examined relatively scarcely (Dywan et al., 2002, Yovel and Paller, 2004).

Previous studies on name priming yield valuable information for the present purposes. Participants are faster to indicate that a name is familiar if it has been presented previously (Bruce and Valentine, 1985, Pickering and Schweinberger, 2003). Importantly, effects of long-term repetition priming of names (and faces) do not cross input modality (see e.g. Burton et al., 1990), which means that, e.g., the prior presentation of a written name results in a faster decision if that same written name is presented again, whereas a spoken name should not prime a familiarity decision on a written name. This is generally in line with reduced or even absent cross-modality priming with auditory versus visual word presentation (e.g. Roediger and Blaxton, 1987), but in contrast to cross-modality priming in both younger and older adults for ecologically valid stimuli presented in the visual and auditory, as well as in the visual and tactual modality (Ballesteros et al., 2009, Reales and Ballesteros, 1999). In an ERP study, repeated relative to non-repeated names elicited more positive amplitudes between 500–600 ms (Schweinberger et al., 2002), resembling an N400 effect (Kutas and Federmeier, 2011), which was interpreted as reflecting the facilitated access to person-specific representations. Accordingly, an influence of repetition priming on source memory judgments for person-related information may manifest in N400-like ERP effects.

In the present study, we examined source memory for famous names. The use of these ecologically valid stimuli appeared particularly important for examining older adults who often complain about difficulties remembering names (see e.g. Cohen and Burke, 1993, Cohen and Faulkner, 1986, Old and Naveh-Benjamin, 2012). Moreover, on the one hand age-related source memory deficits are typically observed in laboratory experiments, while on the other hand a considerable discrepancy between lab and daily-life situations has been noted (see e.g. Verhaeghen et al., 2012). In the present experiments, names were either presented visually or auditorily during learning. To test for potential effects of repetition priming on source judgments, all names were presented visually at test, and accordingly half of the names were repeated within the same modality between learning and test, whereas the other half was presented in a different modality. Thus, in context of the present study, fluency was conceptualized as the potentially facilitated processing for visually relative to auditorily learned names resulting from an overlap between study and test modality. Moreover, we presented half of the names repeatedly during learning, a manipulation that has been described to particularly enhance the recollection of the items (Dewhurst and Anderson, 1999). In all experiments reported below, we assessed neural correlates of the accompanying processes using ERPs. We reasoned that if fluency affected source memory in terms of within-modality repetition priming, more positive amplitudes for visually learned names as compared to both auditorily learned and new names should occur in an early time window (reflecting an N400/FN400). Moreover, if repetition during learning enhanced recollection for both visually and auditorily learned names (see Dewhurst and Anderson, 1999), larger old/new effects in the 600–900 ms time range for items repeated during learning, independent of study-test modality overlap, should be observed.