Updating Memory and Remembering Changes
Navigating a changing world requires updating memory. Earlier memories often interfere with memory for new, similar events, but our work shows that this interference can be eliminated, and even reversed, when people notice how current events differ from the past and later remember those changes. Our theoretical account proposes that when a new experience reminds people of a related past event, that memory guides predictions about what will happen next. When those predictions fail, people can integrate the old and new details, along with the fact that something changed, into a memory that preserves their order.
We test this account with naturalistic events by showing viewers movies of an actor performing everyday activities that sometimes change from one day to the next. Using fMRI, eye tracking, and overt predictions, we have found that reinstating memories of earlier activities, looking toward where the actor acted before, and detecting and later remembering changes all predict better memory for changed activities.
We also study updating with word pairs whose responses change across presentations, which allows precise control over what changes and when. Recent studies show that reinstating the context of an earlier event helps people retrieve it during a later event, which improves memory for both. People must also attend to both the earlier and later events to benefit from noticing changes. Finally, change awareness determines whether earlier memories help or hurt memory for later events, and the reverse, as time passes.
We have also applied this work to correcting misinformation, in research previously supported by the National Science Foundation. Corrections that briefly remind people of a false claim before presenting the truth improve memory and belief accuracy more than corrections that present only the truth. Eye tracking shows that these reminders direct attention to the corrective details, and improvements in belief accuracy last for weeks to a month when people confidently remember that a claim was corrected.
A central question across these lines of work is how memory updating changes with age. Older adults are often thought to be more susceptible to interference from outdated information, and in our everyday-events studies their retrieval of earlier activities was less strongly linked to successful updating. However, techniques that bring earlier information to mind can help both older and younger adults. Testing memory for earlier information before new learning improved memory for the more recent information and for what had changed. Similarly, reminding people of false headlines during corrections improved memory for the true information and reduced intrusions of false information in both age groups. Older adults were less able to recognize which headlines had been corrected, which suggests that their memories may still lack some of the details that signal whether information is true.
Mnemonic Discrimination
Mnemonic discrimination is the ability to distinguish current experiences from memories of similar but not identical events. It may be supported by hippocampal pattern separation, which encodes similar experiences as distinct representations. We measure it with recognition tasks in which people study objects and later must distinguish them from similar lures (e.g., a different glass of red wine). Using this approach, we have identified lure discrimination impairments in healthy older adults and in children and adolescents at risk for schizophrenia, and have shown that intrinsic functional connectivity in the default mode network predicts lure discrimination ability.
In collaboration with the PAAD-2 (Protecting Against Alzheimer's Disease) trial at UNCG, we are examining whether a year-long physical activity program improves hippocampal memory in sedentary middle-aged adults with a family history of dementia. Preliminary results suggest that exercise improved lure discrimination, mainly for older participants who carry the APOE4 allele, the strongest common genetic risk factor for late-onset Alzheimer's disease. This benefit did not extend to traditional recognition or perceptual discrimination, and it grew with greater exercise adherence.
Our work also asks what successful lure discrimination reflects. People who reported being focused during study later rejected more lures, especially when they recollected the studied object, but they also more often mistook lures for studied objects. When people explained their decisions in their own words, they often described rejecting lures by recalling the studied object and noticing what had changed, and this language predicted accuracy better than confidence ratings did. More broadly, we have proposed a unified theory in which memory for similar events reflects three distinct processes, false recollection, false familiarity, and recollection rejection, each of which can be measured from recognition confidence. This framework suggests, for example, that older adults' difficulty with similar lures reflects both more false recollection and less recollection rejection.
With support from the National Science Foundation, we are collaborating with researchers at UNC Chapel Hill to collect fMRI data using a new ultra-high-resolution 3T protocol. Standard 3T protocols cannot distinguish the dentate gyrus and CA3 hippocampal subfields, but this protocol separates them while also imaging the whole brain. Using a standard mnemonic similarity task, this work aims to establish that the protocol can isolate each subfield's contribution to lure discrimination and its interactions with cortical networks. In complementary work here at UNCG, we are planning to use pupillometric measures of sustained attention and memory reinstatement to further examine these processes.
Individual and Age Differences in Memory and Control
Context theories of episodic memory propose that remembering a specific past event depends on reinstating the context in which it occurred. Free and cued recall rely heavily on this self-initiated reinstatement, making them ideal for studying how people constrain memory search. The ability to use context varies across people: those with higher working memory capacity and better sustained attention tend to reinstate context more effectively and remember more. It also changes with age. Older adults' memory deficits are most pronounced when little environmental support is available and when similar memories compete during retrieval. We study these differences by analyzing how recall unfolds over time, including where people begin recalling, how they move from one memory to the next, and when they report memories from the wrong event.
In collaboration with a computational modeling lab at another university, we have used these recall dynamics to develop a model of how people overcome the pull of recent memories to retrieve earlier ones. The model proposes that before retrieving, people reinstate the context of the target event, such as the instructions that accompanied an earlier list, and use it to guide memory search. This mechanism captured recall patterns that an alternative account, in which people mentally step back through recent memories to reach earlier ones, could not.
We are now testing whether this reinstatement reflects broader cognitive control. In a large individual differences study, participants completed a dual-list free recall task along with multiple measures of working memory capacity and sustained attention, including thought probes about their attentional states. This allows us to ask whether people with stronger control abilities reinstate the target context more effectively and better resist interference from competing lists. The same participants also completed a cued recall task in which word pairs changed across lists, like the tasks in our memory updating research. We are examining whether working memory capacity and sustained attention also predict how well people update memory when events change, which would link the control processes that constrain memory search to those that support memory updating.