Two retrievals from a single cue: A bottleneck persists across episodic and semantic memory

Abstract

There is evidence in the literature that two retrievals from long-term memory cannot occur in parallel. To date, however, that work has explored only the case of two retrievals from newly acquired episodic memory. These studies demonstrated a retrieval bottleneck even after dual-retrieval practice. That retrieval bottleneck may be a global property of long-term memory retrieval, or it may apply only to the case of two retrievals from episodic memory. In the current experiments, we explored whether that apparent dual-retrieval bottleneck applies to the case of one retrieval from episodic memory and one retrieval from highly overlearned semantic memory. Across three experiments, subjects learned to retrieve a left or right keypress response form a set of 14 unique word cues (e.g., black-right keypress). In addition, they learned a verbal response which involved retrieving the antonym of the presented cue (e.g., black-"white"). In the dual-retrieval condition, subjects had to retrieve both the keypress response and the antonym word. The results suggest that the retrieval bottleneck is superordinate to specific long-term memory systems and holds across different memory components. In addition, the results support the assumption of a cue-level response chunking account of learned retrieval parallelism.

Adjustment of cognitive control to the frequency of dual-task interference

“…Importantly, as we already mentioned in the introduction section, participants can flexibly adopt a parallel or serial processing strategy depending on task requirements (Lehle, Steinhauser, & Hübner, 2009;Orscheschek, Strobach, Schubert, & Rickard, 2018;Strobach, Pashler, Schubert, & Rickard, 2014). In line with previous observations of , current research suggests that a parallel strategy is preferred over a serial strategy, even if it leads to higher costs of primary task performance.…”

“…Despite their differences, both models assume that the presence of a secondary task results in between-task interference and a decrease in performance, which is either due to suspended processing of the second task, or limited resources which need to be shared between both tasks. Interestingly, recent research suggests that participants can process dual-tasks in either a serial (one task at a time) or a parallel (sharing capacity) manner (see Lehle, Steinhauser, & Hübner, 2009;Orscheschek, Strobach, Schubert, & Rickard, 2018;Strobach, Pashler, Control adjustment in dual-tasking Schubert, & Rickard, 2014). However, the underlying conditions of the adopted processing strategy remain unclear.…”

The dual-task practice advantage: Empirical evidence and cognitive mechanisms

“…Because response chunking is an associative memory process that would occur during a dual-retrieval trial, this response chunking account suggests that learned retrieval parallelism occurs at the individual cue-response level (i.e., for old cues only) rather than at the task level (i.e., for old and new cues). Consistent with that possibility, Orscheschek et al (2019) and Strobach, Schubert, et al (2014) have shown that this parallelism did not transfer to cues for which only single-retrieval practice occurred during the practice phase, with dual-retrieval trials introduced only during a transfer phase. In this way, empirical evidence in dual tasks with two long-term memory retrieval tasks showed the existence of the DTPA phenomenon and provided evidence that is consistent with the integration hypothesis (i.e., the chunking mechanism; Nino & Rickard, 2003).…”

“…The findings of the dual-retrieval practice studies (Orscheschek et al, 2019;Strobach, Schubert, et al, 2014) are incompatible with assumptions of the allocation and scheduling hypothesis since there is no evidence for transfer of learned retrieval parallelism between old and new cues within the same blocks of dual-memory retrieval (i.e., old and new cues were presented under very similar conditions). Alternatively, the literature on dual tasks with two long-term memory tasks discusses one variation of the integration hypothesis.…”

“…The investigation of the DTPA phenomenon and its transfer effects in the context with two long-term memory tasks has a crucial difference from the previously reported dual-task combinations in that dual-task practice and single-task practice were not realized in different groups of participants. In this case, dual-task and single-task practice were realized in the same of group of participants, but in different of cues on which long-term memory practice was performed (Orscheschek, Strobach, Schubert, & Rickard, 2019;Strobach, Schubert, Pashler, & Rickard, 2014b). 1 In detail, participants had to associate each of a set of 14 color-word cues with a unique spatial key-press or a unique vocal-digit response in the beginning of practice on the long-term memory tasks.…”