Neuroimaging studies have often implicated encoding processes in age-related episodic memory decline (see Maillet and Rajah, 2014, for a meta-analysis; Wang and Cabeza, 2016, for review), whereas many of the neural correlates of retrieval processing appear to be relatively stable throughout much of the adult lifespan (e.g., Dulas and Duarte, 2011; de Chastelaine et al., 2016; de Chastelaine et al., 2017). For example, fMRI studies have reported retrieval effects in the ‘core recollection network’ (Kim, 2010, Rugg and Vilberg, 2013, King et al., 2015) to be of equal or near equal magnitude in young and older participants (e.g., Duarte et al., 2008, Dulas and Duarte, 2012, Dulas and Duarte, 2016, Hou et al., 2021, Wang and Giovanello, 2016, Wang et al., 2016). Furthermore, the neural correlates of retrieval processes thought to place high demands on executive control, such as post-retrieval monitoring, also appear to be minimally affected in healthy aging (e.g., Fandakova et al., 2013; Dulas and Duarte, 2014; Wang et al., 2016; Horne et al., 2021). Older adults’ adoption of specific ‘retrieval orientations’ – task sets that bias retrieval cue processing in service of a retrieval goal (Rugg, 2004) – appears, however, to be less spontaneous than is the case in young adults and may occur only when the retrieval task explicitly requires recollection of episodic content (Morcom and Rugg, 2004, Jacoby et al., 2005, Duverne et al., 2009).

Findings from a recent fMRI study investigating episodic memory retrieval in young and older adults suggest that another aspect of retrieval processing, ‘retrieval gating’, might be particularly vulnerable to the effects of increasing age (Srokova et al., 2021). Retrieval gating refers to the ability to control the contents of a retrieved memory so as to align them with the retrieval goal (Elward and Rugg, 2015, Elward et al., 2021). Retrieval gating has been proposed to depend on a combination of attentional suppression and amplification, whereby gating weakens cortical representations of goal-irrelevant information, especially when the information might interfere or compete with goal-relevant information, while strengthening the representation of relevant information (Elward and Rugg, 2015). In these prior studies, and in the present study also, the amount of retrieved content was indexed by the strength of retrieval-related cortical reinstatement, that is, the strength of reactivation of content-selective patterns of neural activity that were elicited during the encoding of a study event (for review, see Rugg and Srokova, 2024). On the widely held assumption that reinstatement supplies the content of a retrieved memory (Rugg and Srokova, 2024), retrieval gating can be operationalized in terms of the goal-dependent modulation of reinstatement.

In the study of Srokova et al. (2021), retrieval gating was examined in samples of young and older adults using an experimental procedure closely similar to that employed by Elward and colleagues (Elward and Rugg, 2015, Elward et al., 2021). Words were presented at study superimposed over an image of an urban scene, a rural scene or a scrambled background. The word-image pairs appeared at one of three locations on the presentation monitor. The subsequent scanned memory test comprised two sub-tasks which were administered in short interleaved ‘mini-blocks’. In the ‘location’ task, the requirement was to judge the location in which recognized test words had been studied. In the ‘background’ task, the source memory judgment was with respect to the nature of the image that had been paired with the test word. In both age groups, retrieval-related scene reinstatement (operationalized as greater neural activity for words studied with scenes than with a scrambled background) was identified in the background task in two scene-selective cortical regions, the parahippocampal place area (PPA) and medial place area (MPA; also referred to as the retrosplenial complex). Replicating prior findings (Elward and Rugg, 2015, Elward et al., 2021), reinstatement effects in the young participants were markedly attenuated in the location task - the hallmark of retrieval gating. Crucially, this finding was not evident in the older adults in whom reinstatement effects did not significantly differ between the two memory tasks. Additionally, in the young age group only, PPA scene reinstatement in the background task correlated positively across participants with source memory performance in that task (as was also reported by Elward et al., 2021), as well as with item memory more generally.

One interpretation of the absent retrieval gating in older adults reported by Srokova et al. (2021) is that the ability to control the content of recollected information declines with age. This interpretation is consistent with other evidence pointing to an age-related decline in cognitive control (e.g., Yuan and Raz, 2014; Zanto and Gazzaley, 2019), including the kinds of inhibitory control processes that might play a role in ‘suppressing’ otherwise memorable mnemonic content (Gazzaley et al., 2005, Gazzaley et al., 2008, Chadick et al., 2014, Campbell et al., 2020, Weeks et al., 2020). Relatedly, it has been proposed that older adults encode irrelevant features of an event into memory to a greater extent than younger adults, leading to ‘cluttered memory representations’ that interfere with the retrieval of target information (Amer et al., 2022). Such representations can include recently activated information that is no longer relevant but which remains bound to target information because of reduced cognitive control (a phenomenon related to age-related ‘hyper-binding’; e.g., Campbell et al., 2010; Campbell and Hasher, 2018). Indeed, Amer et al. (2022) proposed that the absent retrieval gating in older adults that was reported by Srokova et al. (2021) could be accounted for by such a mechanism.

It is noteworthy, however, that Srokova et al. (2021) reported that source memory accuracy was markedly lower in the older than the young age group. The relatively weak contextual memory in the older participants raises the possibility that their failure to demonstrate retrieval gating in the location task (i.e., when the background images were goal-irrelevant) did not reflect an inability to engage a gating strategy but, rather, a lack of motivation to do so. By this argument, the relatively weak memories for the background images were insufficiently intrusive to motivate engagement of retrieval gating when background information was goal irrelevant. This possibility is reminiscent of the finding that the adoption of specific retrieval orientations in older adults occurs only when there is a motivation to do so (Duverne et al., 2009). We examined this possibility in the present study by employing the same experimental design as Srokova et al. (2021), but with a study manipulation that strengthened memory for the background contexts so that performance in the older adults matched that of the younger adults in the prior study. If the failure in that study to identify retrieval gating effects in the older age group does indeed reflect weak memory for the background images, then we would expect to find evidence for the engagement of a gating strategy in the older age group employed in the present study. (Note that, as in Elward and colleagues 2015, 2021 and Srokova et al., 2021, MRI data were not acquired during the encoding phase, obviating an analysis of encoding-retrieval overlap and the direct assessment of reinstatement effects (cf. Rugg and Srokova, 2024). Thus, what is referred to here and in those prior reports as ‘retrieval-related reinstatement’ is arguably more accurately described as content-dependent retrieval. However, given the extensive evidence that the regions of interest employed in the present study and that of Srokova et al. 2021 exhibit strong scene selectivity, and consistently demonstrate scene-related reinstatement (e.g., Folville et al., 2020; Bainbridge et al., 2021; Hill et al., 2021), we continue our prior practice of referring to scene-selective retrieval effects in these regions as reinstatement).

As in Srokova et al. (2021), young and older adults studied a series of concrete nouns superimposed on a rural scene, an urban scene, or a scrambled background. Each word-image pair was studied twice, in contrast with the single study presentations employed previously. The subsequent scanned test phase was identical to that employed in the prior study: two interleaved memory tasks were employed, one of which probed memory for the test word’s studied background while the other tested location memory. We expected to identify scene reinstatement effects in the PPA and MPA in both age groups. We also expected that younger adults would demonstrate a retrieval gating effect in the form of attenuated scene reinstatement in the location task relative to the background task. As noted previously, if the failure of older adults to demonstrate retrieval gating in the study of Srokova et al. (2021) reflected a lack of incentive to employ gating because of weak memory for the backgrounds, the strengthening of these memories should motivate the adoption of a gating strategy and, hence, lead to attenuated scene reinstatement in the location task.