Symposium

Symposium 1: The Information Exchange Between Working Memory and Long-Term Memory   
Friday, November 5, 10 AM - 12 PM CT 

Chairs: Lea Bartsch (University of Zurich), Vanessa Loaiza (University of Essex), and Eda Mizrak (University of Zurich)
A fundamental and enduring question since the beginning of memory research concerns the overlap and reciprocal impacts of briefly keeping information active in mind (working memory) and the durable retention of that information (long-term memory). Much research has addressed this question by considering how working memory processes constrain and shape subsequent long-term memory, and, concomitantly, how stored information in long-term memory may enhance or interfere with working memory. This work has significant theoretical implications for many models of memory regarding whether working memory and long-term memory are dissociable systems and to what extent. In this symposium, we consider the latest findings that speak to the bidirectional relationship between working memory and long-term memory in an effort to bring consensus to a field that has been wrought with debate for over 100 years.

Is Active Engagement in Working Memory Important for Long-Term Episodic Memory? 

Presenter: Vanessa M. Loaiza (University of Essex)
Many theories assume that actively maintaining information in working memory (WM) predicts its retention in episodic memory (EM), as revealed by the beneficial effects of more WM time. Here, we examined whether affording more time for intentional WM maintenance does indeed drive EM. Participants either incidentally or intentionally encoded and maintained words presented during trials of simple, complex, and slow span. Interleaving presentation of the words, a pause of equivalent duration entailed a blank screen or an arithmetic problem to read aloud and solve during slow and complex span trials, respectively. To ensure similar encoding of the words across groups, participants decided whether each word was a living or nonliving thing (i.e., an animacy judgment). A surprise delayed recall test at the end of the experiment assessed EM. Unexpectedly, there were no advantages of complex and slow span (long maintenance) over simple span (short maintenance) in delayed recall for either group, suggesting that the animacy judgment abolished these effects. Thus, if more WM time is used to keep information active via elaboration, then imposing elaboration across conditions equates them in terms of process engagement irrespective of WM time.

Developmental Working Memory Limits across the Human Lifespan: Implications for Long-Term Memory 

Presenter: Alicia Forsberg (University of Missouri)
Healthy young adults typically outperform both children and older adults on tests of working memory. We explored whether age-related working memory limitations also constrain the amount of information that children and older adults commit to long-term memory. Our participants (school-aged children, college students, and healthy older adults) performed a working memory task with unique everyday items, presented in groups of two, four, or six items. Presentation time was adjusted according to the number of items. Then, we tested participants' long-term memory for items from the working memory task. While young adults on average held more information in working memory, the ability to transfer information from working to long-term memory generally appeared consistent across the lifespan. We also explore age-related differences in confidence ratings in the working and long-term memory tasks. These results suggest that experimental, individual, and age-related working memory limitations act as a bottleneck for long-term memory performance and have theoretical and practical implications for cognitive development and cognitive aging.

When Does Episodic Long-Term Memory Contribute To Performance in a Working Memory Task? 

Presenter: Lea Bartsch (University of Zurich)
Previously, we found that episodic long-term memory (eLTM) enhances WM performance when both novel and previously learnt word pairs must be retained on a short-term basis. However, there is uncertainty regarding how and when WM draws on eLTM. Two possibilities are that (a) people draw on eLTM only if WM capacity is exceeded; or that (b) there is always a contribution of eLTM to WM performance. The former implies that, when the load on WM is low, performance is driven by representations stored in WM only, whereas at higher loads, people draw on both LTM and WM. We tested this prediction by familiarizing participants with some items before they completed a separate WM task. In accordance with possibility (a) performance deteriorated with the addition of stimuli from eLTM when WM load was low, but not when it was high; and it remained superior to performance with the matched set size comprising only new stimuli. In further experiments, we build on the assumption that under circumstances where eLTM contributes to WM task performance, these eLTM traces could benefit or hamper performance depending on the match between LTM traces and to-be stored information in WM, yielding proactive facilitation and interference, respectively.

Improving Working Memory Capacity Through Learned Knowledge Is Not Done Via Pointers To Long-Term Memories, But Rich, Hierarchical Memory Traces

Presenter: Timothy Brady (University of California, San Diego)
Previously acquired knowledge improves our ability to represent information in working memory, including via chunking. When we use chunks, it is often assumed we are reducing the load on working memory by storing only pointers to previously acquired long-term memories. By contrast, our recent work suggests a fundamentally different view: that chunks are not content-free pointers, but instead serve as cues that facilitate the encoding and retention into memory of additional perceptual details as part of structured, hierarchical memories. This provides a contrast to accounts in which working memory capacity is assumed to be exhaustively described by the number of chunks remembered. In particular, we find that participants actually have enhanced knowledge of the trial-specific perceptual details of items that can be chunked, rather than losing such details; and that stimuli that are recognized as a real object (i.e., a face) elicit more working memory usage (as though more features can be stored for them), rather than less working memory usage (as expected if pointers were used to store them). Overall, we suggest a rethinking of the nature of chunking and the role of previous knowledge in working memory.

How Does Working Memory Benefit from Long-Term Memory? 

Presenter: Eda Mizrak (University of Zurich)
Working memory (WM) can harness knowledge from long-term memory (LTM) when it is useful. For example, WM for novel stimuli is better when they are encoded in a list with familiar chunks compared to a list with no familiar chunks. LTM could benefit WM in two ways: 1) familiar chunks are encoded in a compressed form that takes less WM capacity (e.g., a pointer to LTM), freeing WM capacity for other items; 2) familiar chunks are easier to encode, leaving more encoding resources for other items. These accounts make different predictions and we tested them across multiple experiments by manipulating the presence of familiar chunks and free time during encoding. According to the encoding resource account (Popov & Reder, 2020), encoding resources replenish with free time. The first account predicts that the LTM benefit on WM should not change with extra free time; the second account predicts that the effects of extra free time and LTM benefit on WM should interact. We found that the LTM benefit did not change with extra free time. However, it is possible that additional mechanisms could mask the interaction. These results raise more questions than answers and we discuss additional steps required to distinguish between the two hypotheses.