Overcoming the Negative Consequences of Interference From Recognition Memory Testing

Retrieving effectively from source memory: Evidence for differentiation and local matching processes

The ability to distinguish between different explanations of human memory abilities continues to be the subject of many ongoing theoretical debates. These debates attempt to account for a growing corpus of empirical phenomena in item-memory judgments, which include the list strength effect, the strength-based mirror effect, and output interference. One of the main theoretical contenders is the Retrieving Effectively from Memory (REM) model. We show that REM, in its current form, has difficulties in accounting for source-memory judgments - a situation that calls for its revision. We propose an extended REM model that assumes a local-matching process for source judgments alongside source differentiation. We report a first evaluation of this model's predictions using three experiments in which we manipulated the relative source-memory strength of different lists of items. Analogous to item-memory judgments, we observed a null list strength effect and a strength-based mirror effect in the case of source memory. In a second evaluation, which relied on a novel experiment alongside two previously published datasets, we evaluated the model's predictions regarding the manifestation of output interference in item and lack of it in source memory judgments. Our results showed output interference severely affecting the accuracy of item-memory judgments but having a null or negligible impact when it comes to source-memory judgments. Altogether, these results support REM's core notion of differentiation (for both item and source information) as well as the concept of local matching proposed by the present extension.

Modeling the continuous recognition paradigm to determine how retrieval can impact subsequent retrievals

There are several ways in which retrieval during a memory test can harm memory: (a) retrieval can cause an increase in interference due to the storage of additional information (i.e., item-noise); (b) retrieval can decrease accessibility to studied items due to context drift; and (c) retrieval can result in a trade in accuracy for speed as testing progresses. While these mechanisms produce similar outcomes in a study-test paradigm, they are dissociated in the 'continuous' recognition paradigm, where items are presented continuously and a participant's task is to detect a repeat of an item. In this paradigm, context drift results in worse performance with increasing study-test lag (the lag effect), whereas increasing item-noise is evident in a decrease in performance for later test trials in the sequence (the test position effect [TPE]). In the present investigation, we measured the influences of item-noise, context drift, and decision-related factors in a novel continuous recognition dataset using variants of the Osth et al. (2018) global matching model. We fit both choice and response times at the single trial level using state-of-the-art hierarchical Bayesian methods while incorporating crucial amendments to the modeling framework, including multiple context scales and sequential effects. We found that item-noise was responsible for producing the TPE, context drift decreased the magnitude of the TPE (by diminishing the impact of item-noise), and speed-accuracy changes had some minor effects that varied across participants.

Test position effects on hit and false alarm rates in recognition memory for paintings and words

When old/new recognition memory is tested with equal numbers of studied and nonstudied items and no rewards or instructions that favour one response over the other, there is no obvious reason for response bias. In line with this, Canadian undergraduates have shown, on average, a neutral response bias when we tested them on recognition of common English words. By contrast, most subjects we have tested on recognition of richly detailed images have shown a conservative bias: they more often erred by missing a studied image than by judging a nonstudied image as studied. Here, in an effort to better understand these materials-based bias effects (MBBEs), we examined changes in hit and false alarm (FA) rates (and in sensitivity and bias) from the first to fourth quartile of a recognition memory test in eight experiments in which undergraduates studied words and/or images of paintings. Response bias for images tended to increase across quartiles, whereas bias for words showed no consistent pattern across quartiles. This pattern could be described as an increase in the MBBE over the course of the test, but the underlying patterns for hits and FAs are not easily reconciled with this interpretation. Hit rates decreased over the course of the test for both materials types, with that decline tending to be steeper for images than words. For words, FA rates tended to increase across quartiles, whereas for paintings FA rates did not increase across quartiles. We discuss implications of these findings for theoretical accounts of the MBBE.

Why seemingly more difficult test conditions produce more accurate recognition of semantic prototype words: A recognition memory paradox?

Subjects studied Deese-Roediger-McDermott semantic-associate lists and took a recognition test. The makeup and number of test probes were manipulated. In Experiments 1 and 2A, one of three or all three distractors were semantically related to the list theme. In Experiment 2B, 6 or 30 related probes were used at test. Results showed that semantically related distractors and a longer list of test words both had a beneficial effect on the accurate discrimination of the prototype lures from the studied semantic associates and on the discrimination of studied from unstudied prototype words. These findings are inconsistent with predictions of memory interference and activation theories. We propose that the counterintuitive findings can be explained by the notion of old/new recognition as categorization learning and that relatedness and a larger number of test probes provide more accurate information about the prototype lure as a distractor, thereby improving its classification as a distractor.

Modeling the dynamics of recognition memory testing with an integrated model of retrieval and decision making

A robust finding in recognition memory is that performance declines monotonically across test trials. Despite the prevalence of this decline, there is a lack of consensus on the mechanism responsible. Three hypotheses have been put forward: (1) interference is caused by learning of test items (2) the test items cause a shift in the context representation used to cue memory and (3) participants change their speed-accuracy thresholds through the course of testing. We implemented all three possibilities in a combined model of recognition memory and decision making, which inherits the memory retrieval elements of the Osth and Dennis (2015) model and uses the diffusion decision model (DDM: Ratcliff, 1978) to generate choice and response times. We applied the model to four datasets that represent three challenges, the findings that: (1) the number of test items plays a larger role in determining performance than the number of studied items, (2) performance decreases less for strong items than weak items in pure lists but not in mixed lists, and (3) lexical decision trials interspersed between recognition test trials do not increase the rate at which performance declines. Analysis of the model's parameter estimates suggests that item interference plays a weak role in explaining the effects of recognition testing, while context drift plays a very large role. These results are consistent with prior work showing a weak role for item noise in recognition memory and that retrieval is a strong cause of context change in episodic memory.

Release from output interference in recognition memory: A test of the attention hypothesis

Retrieval results in both costs and benefits to episodic memory. Output interference (OI) refers to the finding that episodic memory accuracy decreases with increasing test trials. Release from OI is the restoration of original accuracy at some point during the test. For example, a release from OI in recognition memory testing occurs when the semantic similarity between stimuli decreases midway through testing, suggesting that item representations stored on early trials cause interference on tests occurring on later trials to the extent that the earlier items share features with the latter items. In two recognition memory experiments, we demonstrate release from OI for words and faces. We also test whether release from OI is the result of interference or is due to a boost in attention caused by reorienting to a novel stimulus type. A test for the foils presented during the initial test list supports the interference account of OI. Implications for models of memory are discussed.

Towards a theory of individual differences in statistical learning

In recent years, statistical learning (SL) research has seen a growing interest in tracking individual performance in SL tasks, mainly as a predictor of linguistic abilities. We review studies from this line of research and outline three presuppositions underlying the experimental approach they employ: (i) that SL is a unified theoretical construct; (ii) that current SL tasks are interchangeable, and equally valid for assessing SL ability; and (iii) that performance in the standard forced-choice test in the task is a good proxy of SL ability. We argue that these three critical presuppositions are subject to a number of theoretical and empirical issues. First, SL shows patterns of modality- and informational-specificity, suggesting that SL cannot be treated as a unified construct. Second, different SL tasks may tap into separate sub-components of SL that are not necessarily interchangeable. Third, the commonly used forced-choice tests in most SL tasks are subject to inherent limitations and confounds. As a first step, we offer a methodological approach that explicitly spells out a potential set of different SL dimensions, allowing for better transparency in choosing a specific SL task as a predictor of a given linguistic outcome. We then offer possible methodological solutions for better tracking and measuring SL ability. Taken together, these discussions provide a novel theoretical and methodological approach for assessing individual differences in SL, with clear testable predictions.This article is part of the themed issue 'New frontiers for statistical learning in the cognitive sciences'.

Dynamic memory searches: Selective output interference for the memory of facts

The benefits of testing on later memory performance are well documented; however, the manner in which testing harms memory performance is less well understood. This research is concerned with the finding that accuracy decreases over the course of testing, a phenomena termed "output interference" (OI). OI has primarily been investigated with episodic memory, but there is limited research investigating OI in measures of semantic memory (i.e., knowledge). In the current study, participants were twice tested for their knowledge of factual questions; they received corrective feedback during the first test. No OI was observed during the first test, when participants presumably searched semantic memory to answer the general-knowledge questions. During the second test, OI was observed. Conditional analyses of Test 2 performance revealed that OI was largely isolated to questions answered incorrectly during Test 1. These were questions for which participants needed to rely on recent experience (i.e., the feedback in episodic memory) to respond correctly. One possible explanation is that episodic memory is more susceptible to the sort of interference generated during testing (e.g., gradual changes in context, encoding/updating of items) relative to semantic memory. Alternative explanations are considered.

The survival effect in memory: does it hold into old age and non-ancestral scenarios?

The survival effect in memory refers to the memory enhancement for materials encoded in reference to a survival scenario compared to those encoded in reference to a control scenario or with other encoding strategies [1]. The current study examined whether this effect is well maintained in old age by testing young (ages 18–29) and older adults (ages 65–87) on the survival effect in memory for words encoded in ancestral and/or non-ancestral modern survival scenarios relative to a non-survival control scenario. A pilot study was conducted to select the best matched comparison scenarios based on potential confounding variables, such as valence and arousal. Experiment 1 assessed the survival effect with a well-matched negative control scenario in both young and older adults. The results showed an age-equivalent survival effect across an ancestral and a non-ancestral modern survival scenario. Experiment 2 replicated the survival effect in both age groups with a positive control scenario. Taken together, the data suggest a robust survival effect that is well preserved in old age across ancestral and non-ancestral survival scenarios.

Strength cues and blocking at test promote reliable within-list criterion shifts in recognition memory

In seven experiments, we explored the potential for strength-based, within-list criterion shifts in recognition memory. People studied a mix of target words, some presented four times (strong) and others studied once (weak). In Experiments 1, 2, 4A, and 4B, the test was organized into alternating blocks of 10, 20, or 40 trials. Each block contained lures intermixed with strong targets only or weak targets only. In strength-cued conditions, test probes appeared in a unique font color for strong and weak blocks. In the uncued conditions of Experiments 1 and 2, similar strength blocks were tested, but strength was not cued with font color. False alarms to lures were lower in blocks containing strong target words, as compared with lures in blocks containing weak targets, but only when strength was cued with font color. Providing test feedback in Experiment 2 did not alter these results. In Experiments 3A-3C, test items were presented in a random order (i.e., not blocked by strength). Of these three experiments, only one demonstrated a significant shift even though strength cues were provided. Overall, the criterion shift was larger and more reliable as block size increased, and the shift occurred only when strength was cued with font color. These results clarify the factors that affect participants' willingness to change their response criterion within a test list.

Null category-length and target–lure relatedness effects in episodic recognition: A constraint on item-noise interference models

There are two main classes of model of interference effects in recognition memory: item-noise and context-noise. Item-noise models predict that a loss of memory discriminability will occur with an increase in the number of studied items from the same taxonomic category (category length, CL) and that forced-choice recognition performance will be higher when the target and lure are related rather than unrelated. Context-noise models, however, predict null effects for both of these manipulations. Although results from some recent experiments suggest that CL and target–lure relatedness have a trivial or no effect on memory discriminability when the related items from the same taxonomic category are “not back to back in the study list but are separated (spaced) by interleaving items from other semantic categories,” these experiments have methodological limitations that were eliminated in the present experiment in which exemplars representing category lengths of 2, 8, or 14 were presented spaced apart within the same study list. Recognition was tested using a yes/no recognition test or a two-alternative forced-choice recognition test in which the target and lure were either related or unrelated. In yes/no recognition, d′ decreased as CL increased, replicating prior research. However, when the slope of the z-ROC function is less than 1.0, as is typically so and was so in the present results, d′ differences can arise due to criterion shifts and are not necessarily due to memory discriminability differences. When the more appropriate measure of memory discriminability, d a, was computed, CL had no effect in yes/no recognition, nor did it have an effect in forced-choice recognition, which also was not affected by target–lure relatedness. Thus, the present results are congruent with context-noise models and pose a challenge for item-noise models.