Priming Effects on Subsequent Episodic Memory: Testing Attentional Accounts

Prior work has shown that priming improves subsequent episodic memory, i.e., memory for the context in which an item is presented is improved if that item has been seen previously. We previously attributed this effect of "Priming on Subsequent Episodic Memory" (PSEM) to a sharpening of the perceptual/conceptual representation of an item, which improves its associability with an (arbitrary) background context, by virtue of increasing prediction error (Greve et al, 2017). However, an alternative explanation is that priming reduces the attentional resources needed to process an item, leaving more residual resources to encode its context. We report four experiments that tested this alternative, resource-based hypothesis, based on the assumption that reducing the available attentional resources by a concurrent load would reduce the size of the PSEM. In no experiment was there an interaction between attentional load and priming on mean memory performance, nor a consistent correlation across participants between priming and PSEM, failing to support the resource account. However, formal modelling revealed that a resource account is not, in fact, inconsistent with our data, by confirming that nonlinear (sigmoidal) resource-performance functions can reproduce any interaction with load, and, more strikingly, any pattern of correlation between priming and PSEM. This work reinforces not only the difficulty of refuting attentional resource accounts of memory encoding, but also questions the value of load manipulations more generally.

The repetition decrement effect in recognition memory: The influence of prime-target spacing

Rosner, Lopez-Benitez, D'Angelo, Thomson, and Milliken (2018, see also Collins, Rosner & Milliken, 2018) reported a novel recognition memory effect. In an incidental study phase, participants saw prime-target word pairs and were asked to name aloud just the target. Primes were unmasked, but participants were not required to attend to them. On repeated trials the prime and target were the same word, whereas on not-repeated trials the prime and target were different words. In the following test phase, recognition memory was better for not-repeated targets than for repeated targets. The present study explores whether this effect is influenced by the spacing between primes and targets. The results replicated prior studies in that immediate repetition resulted in a repetition decrement effect, but spaced repetition (by about 10 min) resulted in the opposite effect - better recognition for repeated than not-repeated targets. The results are discussed in relation to deficient processing theories of the spacing effect.

Congruency Encoding Effects on Recognition Memory: A Stage-Specific Account of Desirable Difficulty

Recent research suggests that selectively attending to relevant stimuli while having to ignore or resist conflicting stimuli can lead to improvements in learning. While mostly discussed within a broader "desirable difficulty" framework in the memory and education literatures, some recent work has focused on more mechanistic questions of how processing conflict (e.g., from incongruent primes) might elicit increased attention and control, producing enhanced incidental encoding of high-conflict stimuli. This encoding benefit for high-control-demand or high-difficulty situations has been broadly conceptualized as a task-general property, with no strong prediction of what particular task elements should produce this effect. From stage processing models of single- and dual-task performance, we propose that memory-enhancing difficulty manipulations should strongly depend on inducing additional cognitive control at particular processing stages. Over six experiments, we show that a memory benefit is produced when increased cognitive control (via incongruency priming) focuses additional processing on the core meaning of to-be-tested stimuli at the semantic categorization stage. In contrast, incongruency priming targeted at response selection within the same task produces similar effects on initial task performance, but gives no memory benefit for high-conflict trials. We suggest that a simple model of limited-capacity and stage-specific cognitive control allocation can account for and predict where and when conflict/difficulty encoding benefits will occur, and may serve as a model for desirable difficulty effects more broadly.

Selective attention effects on recognition: the roles of list context and perceptual difficulty

Two recent studies reported superior recognition memory for items that were incongruent targets than for items that were congruent targets in a prior incidental study phase (Krebs et al. in Cereb Cortex (New York, NY) 25(3):833-843, 2015; Rosner et al. in Psychol Res 79(3):411-424, 2015). The present study examined this effect further by addressing two issues. First, we examined whether this effect is sensitive to the list context in which congruent and incongruent items are presented. In Experiment 1, this issue was addressed by manipulating the relative proportions of congruent and incongruent trials in the study phase. In Experiments 2A and 2B, the same issue was examined by contrasting randomly intermixed and blocked manipulations of congruency. The results of these experiments, as well as a trial-to-trial sequence analysis, demonstrate that the recognition advantage for incongruent over congruent items is robust and remarkably insensitive to list context. Second, we examined recognition of incongruent and congruent items relative to a single word baseline condition. Incongruent (Experiment 3A) and congruent (Experiment 3B) items were both better recognized than single word items, though this effect was substantially stronger for incongruent items. These results suggest that perceptual processing difficulty, rather than interference caused by different target and distractor identities on its own, contributes to the enhanced recognition of incongruent items. Together, the results demonstrate that processes that are sensitive to perceptual processing difficulty of items but largely insensitive to list context produce heightened recognition sensitivity for incongruent targets.