Tanya R. Jonker

Putting Retrieval-Induced Forgetting in Context: An Inhibition-Free, Context-Based Account

We present a new theoretical account of retrieval-induced forgetting (RIF) together with new experimental evidence that fits this account and challenges the dominant inhibition account. RIF occurs when the retrieval of some material from memory produces later forgetting of related material. The inhibition account asserts that RIF is the result of an inhibition mechanism that acts during retrieval to suppress the representations of interfering competitors. This inhibition is enduring, such that the suppressed material is difficult to access on a later test and is, therefore, recalled more poorly than baseline material. Although the inhibition account is widely accepted, a growing body of research challenges its fundamental assumptions. Our alternative account of RIF instead emphasizes the role of context in remembering. According to this context account, both of 2 tenets must be met for RIF to occur: (a) A context change must occur between study and subsequent retrieval practice, and (b) the retrieval practice context must be the active context during the final test when testing practiced categories. The results of 3 experiments, which directly test the divergent predictions of the 2 accounts, support the context account but cannot be explained by the inhibition account. In an extensive discussion, we survey the literature on RIF and apply our context account to the key findings, demonstrating the explanatory power of context.

Enhancing SART validity by statistically controlling speed-accuracy trade-offs

Numerous studies focused on elucidating the correlates, causes, and consequences of inattention/attention-lapses employ the Sustained Attention to Response Task (SART), a GO-NOGO task with infrequent withholds. Although the SART has become popular among inattention researchers, recent work has demonstrated its susceptibility to speed-accuracy trade-offs (SATOs), rendering its assessment of inattention problematic. Here, we propose and illustrate methods to statistically control for the occurrence of SATOs during SART performance. The statistical solutions presented here can be used to correct standard SART-error scores, including those of already-published data, thereby allowing researchers to re-examine existing data, and to more sensitively evaluate the validity of earlier conclusions.

Can Research Participants Comment Authoritatively on the Validity of Their Self-Reports of Mind Wandering and Task Engagement?

The study of mind wandering rests upon the assumption that people are able to consistently and accurately introspect and report on these sorts of mental experiences. Although there is some initial evidence that people can indeed accurately report on the subjective experience of mind wandering, to date, no work has directly examined peoples degree of confidence in their self-reports of mind wandering and the effects that confidence has on the accuracy of such reports. In the present study, participants completed a sustained-attention task during which they intermittently provided assessments of task engagement (i.e., whether they were focused on the task or mind wandering), as well as reports of confidence in the accuracy of their assessments. This study yielded 3 key findings: We found substantial between- and within-subject variability in both (a) reported mind wandering and (b) confidence in mind-wandering reports, and, most critically, (c) we found that the relation of reported mind wandering and task performance varied as a function of confidence. We discuss the implications of these findings in the context of the literature on mind wandering.

Performance reactivity in a continuous-performance task: implications for understanding post-error behavior.

Although there has been considerable interest in the effects of errors on subsequent performance, relatively few studies have considered the effects of non-error events that contain some performance-relevant information, such as correct performance on critical trials. In the present article, we propose and assess a hypothesis of performance reactivity. In support of this hypothesis, we provide evidence of performance decrements following both incorrect and correct responses but not following performance-irrelevant events. More specifically, in a continuous response task (Sustained Attention to Response Task), we (1) replicate previous findings that errors of commission on rare NOGO trials produce decrements in subsequent performance, and (2) observe that correct withholds to NOGO trials produce decrements in subsequent accuracy relative to task-irrelevant tones. These results corroborate a hypothesis that some error-related effects on subsequent performance are not unique, but are instead a particularly salient version of a more general performance-reactivity effect.

Retrieval-induced forgetting: testing the competition assumption of inhibition theory.

Practicing the retrieval of some information can lead to poorer retrieval of other related information, a phenomenon called retrieval-induced forgetting (RIF). This pattern has been explained as the result of inhibition of the related information during practice (Anderson, 2003). A core assumption of this inhibition account is that, to be suppressed, the related information must compete with the target information at the time of retrieval practice. Three experiments are reported that test this competition assumption. One experiment showed that RIF did not occur without specific retrieval practice of the target items when semantic generation of subordinates was performed. However, in 2 further experiments, RIF did occur when the semantic generation task was paired with category retrieval. Although there was no need for competition between target information and related information in these experiments, RIF was observed. These experiments undermine the competition assumption and hence the inhibition account.

Production improves memory equivalently following elaborative vs non-elaborative processing

Words that are read aloud are better remembered than those read silently. Recent research has suggested that, rather than reflecting a benefit for produced items, this production effect may reflect a cost to reading silently in a list containing both aloud and silent items (Bodner, Taikh, & Fawcett, 2013). This cost is argued to occur because silent items are lazily read, receiving less attention than aloud items which require an overt response. We examined the possible role of lazy reading in the production effect by testing whether the effect would be reduced under elaborative encoding, which precludes lazy reading of silent items. Contrary to a lazy reading account, we found that production benefited generated words as much as read words (Experiment 1) and deeply imagined words as much as shallowly imagined words (Experiment 2). We conclude that production stands out as equally distinct—and consequently as equally memorable—regardless of whether it accompanies deep or shallow processing, evidence that is inconsistent with a lazy reading account.

Testing the item-order account of design effects using the production effect.

A number of memory phenomena evident in recall in within-subject, mixed-lists designs are reduced or eliminated in between-subject, pure-list designs. The item-order account (McDaniel & Bugg, 2008) proposes that differential retention of order information might underlie this pattern. According to this account, order information may be encoded when a common form of processing is used alone in a list (e.g., reading), but not when an unusual form of processing is used (e.g., generation) or when a common form and an unusual form are mixed within a list. The production effect--better memory for words said aloud than for words read silently--shows this same design-contingent pattern. In 2 experiments, we investigated whether differential order retention might underlie the production effect. Consistent with the item-order account, we found that retention of order information was better in pure silent lists than in either pure aloud lists or mixed lists, as measured using an order reconstruction test. Moreover, in Experiment 2, order was better preserved in free recall of pure silent lists than of either pure aloud or mixed lists. Thus, production joins the set of tasks identified by McDaniel and Bugg (2008), and our findings suggest a role for order processing in explaining the production effect.

Less we forget: Retrieval cues and release from retrieval-induced forgetting

Retrieving some items from memory can impair the subsequent recall of other related but not retrieved items, a phenomenon called retrieval-induced forgetting (RIF). The dominant explanation of RIF—the inhibition account—asserts that forgetting occurs because related items are suppressed during retrieval practice to reduce retrieval competition. This item inhibition persists, making it more difficult to recall the related items on a later test. In our set of experiments, each category was designed such that each exemplar belonged to one of two subcategories (e.g., each BIRD exemplar was either a bird of prey or a pet bird), but this subcategory information was not made explicit during study or retrieval practice. Practicing retrieval of items from only one subcategory led to RIF for items from the other subcategory when cued only with the overall category label (BIRD) at test. However, adapting the technique of Gardiner, Craik, and Birtwistle (Journal of Learning and Verbal Behavior 11:778–783, 1972), providing subcategory cues during the final test eliminated RIF. The results challenge the inhibition account’s fundamental assumption of cue independence but are consistent with a cue-based interference account.

Retrieval-Induced Forgetting and Context

Retrieving information can result in the forgetting of related information, a phenomenon referred to as retrieval-induced forgetting (RIF). To date, the dominant explanation of RIF has been an inhibition account, which emphasizes long-term suppression of interfering memories. As one alternative, some have advocated for a strength-based interference account, which emphasizes the role of strengthening associations. More recently, we have proposed a context account, which emphasizes the role of context change and context reinstatement. In this article, we outline these three accounts of RIF and demonstrate that there is substantial evidence that uniquely supports our context account.

Not All Order Memory Is Equal: Test Demands Reveal Dissociations in Memory for Sequence Information.

Remembering the order of a sequence of events is a fundamental feature of episodic memory. Indeed, a number of formal models represent temporal context as part of the memory system, and memory for order has been researched extensively. Yet, the nature of the code(s) underlying sequence memory is still relatively unknown. Across 4 experiments that manipulated encoding task, we found evidence for 3 dissociable facets of order memory. Experiment 1 introduced a test requiring a judgment of which of 2 alternatives had immediately followed a word during encoding. This measure revealed better retention of interitem associations following relational encoding (silent reading) than relatively item-specific encoding (judging referent size), a pattern consistent with that observed in previous research using order reconstruction tests. In sharp contrast, Experiment 2 demonstrated the reverse pattern: Memory for the studied order of 2 sequentially presented items was actually better following item-specific encoding than following relational encoding. Experiment 3 reproduced this dissociation in a single experiment using both tests. Experiment 4 extended these findings by further dissociating the roles of relational encoding and item strength in the 2 tests. Taken together, these results indicate that memory for event sequence is influenced by (a) interitem associations, (b) the emphasized directionality of an association, and (c) an item’s strength independent of other items. Memory for order is more complicated than has been portrayed in theories of memory and its nuances should be carefully considered when designing tests and models of temporal and relational memory.