Refresh my memory: Episodic memory reinstatements intrude on working memory maintenance

Abigail Novick Hoskin, Aaron M. Bornstein, Kenneth A. Norman, Jonathan D. Cohen

TL;DR

This paper reveals that our long-term episodic memories (EM) can "intrude" into our short-term working memory (WM) even when WM isn't actively being disrupted. This intrusion, driven by the reactivation of contextual associations from EM, can lead to errors and slower responses in WM tasks, suggesting a more intertwined relationship between these memory systems than previously thought.

What Problem Does This Paper Address?

A central question in cognitive neuroscience is how different memory systems, particularly working memory (WM) and episodic memory (EM), interact. Traditionally, WM was seen as a temporary buffer for information actively being processed, while EM stored more permanent, event-specific memories. Research often focused on how EM interfered with WM only when WM was overloaded or distracted. However, studies also showed that brain regions supporting EM are active even during rest, suggesting that EM content might be spontaneously reactivated. This raised a crucial question: could these spontaneous EM reinstatements influence WM even when WM is not disrupted? The authors aimed to investigate this by exploring how reactivated EM content, specifically contextual associations, might "leak" into WM and affect performance.

Methods

The researchers employed a multi-pronged approach across three experiments:

  • Experiment 1 (Substitution Errors): Participants performed a short-term recall task with varying levels of distraction during the delay period. They also studied words associated with specific "context pictures." The key was to see if errors (substitutions) in recall were more likely to be words from the same context as the target words, even in the absence of distraction.
  • Experiment 2 (Reaction Time - RT): This experiment used a delayed nonmatch-to-sample (DNMS) task with no distraction during the delay. Participants remembered target words, and then a probe word was presented. The researchers measured RTs to determine if probe words sharing the encoding context with the targets (but not being targets themselves, termed "lure probes") led to slower responses. This was a more sensitive measure than substitution errors.
  • Experiment 3 (fMRI & MVPA): This experiment replicated Experiment 2 but with functional magnetic resonance imaging (fMRI). They used multivariate pattern analysis (MVPA) to identify neural patterns indicative of context reinstatement during the delay period. This allowed them to directly link neural activity reflecting EM reinstatement to behavioral measures (RTs).

Datasets

The study used word lists and scene pictures as stimuli. The words were drawn from the Medical Research Council Psycholinguistic Database. No new benchmarks were introduced; rather, the study utilized established experimental paradigms (delayed recall, DNMS) to probe memory interactions.

Key Insights and Ideas

  • EM isn't just a backup for disrupted WM: The most significant insight is that EM can influence WM performance even when WM is functioning normally, without explicit distraction. This challenges the view that EM interference is solely a consequence of WM overload.
  • Context is the key intruder: The mechanism of intrusion appears to be the reactivation of contextual associations linked to the target items in EM. These associations, when brought into WM, can create confusion.
  • Subtle effects matter: While Experiment 1 showed substitution errors, Experiment 2 demonstrated that even when accuracy is high, RTs can reveal the impact of EM intrusions. This highlights the importance of using multiple measures to understand cognitive processes.
  • Neural evidence for context reinstatement: Experiment 3 provided direct neural evidence that specific patterns of brain activity during the delay period reflect the reinstatement of contextual information, and this reinstatement is linked to behavioral slowing.

Key Results

  • Experiment 1: Participants made significantly more substitution errors from the same encoding context as the target words, even in the no-distraction condition, though the effect was strongest with distraction.
  • Experiment 2: Participants were significantly slower to respond to "lure probes" (words from the same context as targets but not targets themselves) compared to probes from unrelated contexts. This slowing occurred despite near-ceiling accuracy.
  • Experiment 3: fMRI data showed that neural activity patterns during the delay period could predict context reinstatement. This reinstatement of the target's context was specifically associated with slower reaction times on lure trials, and this effect was stronger when the lure word had a higher "overlap" (was more frequently encountered with target words during learning).

Limitations and Future Work

  • Generalizability of Context: The study focused on visual scene contexts. Future work could explore other types of contextual information (e.g., temporal, emotional).
  • Strategic vs. Incidental Reinstatement: While the study suggests spontaneous reinstatement, it acknowledges that some reinstatements might be strategic. Further research could disentangle these processes.
  • Adaptive Nature of Interaction: The authors raise questions about the adaptive value of this EM-WM interaction. Is it a bug or a feature? Future work could explore how this interaction might support learning, generalization, or decision-making in more complex, naturalistic environments.
  • Individual Differences: While the study controlled for some individual differences (e.g., RTs), further investigation into how individual memory characteristics influence these interactions could be valuable.

Summary updated

Abstract

A fundamental question in memory research is how different forms of memory interact. Previous research has shown that people rely on working memory (WM) in short-term recognition tasks; a common view is that episodic memory (EM) only influences performance on these tasks when WM maintenance is disrupted. However, retrieval of memories from EM has been widely observed during brief periods of quiescence, raising the possibility that EM retrievals during maintenance-critically, before a response can be prepared-might affect short-term recognition memory performance even in the absence of distraction. We hypothesized that this influence would be mediated by the lingering presence of reactivated EM content in WM. We obtained support for this hypothesis in three experiments, showing that delay-period EM reactivation introduces incidentally associated information (context) into WM, and that these retrieved associations negatively impact subsequent recognition, leading to substitution errors (Experiment 1) and slowing of accurate responses (Experiment 2). FMRI pattern analysis showed that slowing is mediated by the content of EM reinstatement (Experiment 3). These results expose a previously hidden influence of EM on WM, raising new questions about the adaptive nature of their interaction.
BibTeX
@inproceedings{hoskin2018refresh,
  author = {Abigail Novick Hoskin and Aaron M. Bornstein and Kenneth A. Norman and Jonathan D. Cohen},
  title = {{Refresh my memory: Episodic memory reinstatements intrude on working memory maintenance}},
  booktitle = {Cognitive},
  year = {2018},
  doi = {10.3758/s13415-018-00674-z},
}
Figure