Identifying the Basis for the Word Frequency Effect in Recognition Memory

Abstract

Recent theories of recognition memory have identified two bases on which recognition-memory judgments may be made: recollection, which involves retrieval of contextual information from an earlier episode of stimulus presentation; and familiarity, which is distinguished by a general sense of familiarity in the absence of recollection. Four experiments were conducted to test whether the word frequency effect (WFE) in recognition memory (superior performance with low- in comparison with high-frequency targets) results from recollection-based processes, familiarity-based processes, or both. In two of the experiments, superior memory for aspects of the study context was found for low-frequency in comparison with high-frequency words, suggesting frequency-related differences in recollection. The other two experiments used Jacoby's (1991) inclusion/exclusion paradigm to provide estimates of the contribution of recollection and familiarity to recognition. In both experiments the data suggested that the WFE is primarily a recollection-based phenomenon. These findings suggest that the recognition memory WFE for old items results primarily from the effects of word frequency on recollection. The implications of these findings for theories of recognition memory are discussed.     

Retention of word pairs as a function of level of processing,instruction to remember,and delay

ABSTRACT

Modern research on the efficacy of levels of processing (LoP) tasks on memory has focused on less than 1-day retention delays, while assuming that the observed benefits of deep tasks will continue across remote delays. However, direct tests of the continued benefits of deep processes for accuracy and organisation in remote memory are rare. The current set of experiments, using auditorily-presented lists of scrambled word pairs, tested whether deep LoP tasks produced better free recall and organisation over one week (Experiment 1) and four weeks (Experiments 2 and 3). All experiments revealed significant LoP effects on free recall and organisation at immediate and delayed test, with no effect of intention to remember. However, Experiments 2 and 3 revealed poor recall and organisation at the delayed tests among all of the LoP groups, suggesting that deep processing may not produce highly accessible memories over very long delays.     

Episodic intertrial learning of younger and older adults: Effects of word frequency

ABSTRACT

Older adults have a demonstrable episodic memory deficit. The present study aimed to investigate whether the age deficit in episodic memory was influenced by stimulus characteristics known to produce differences in memory performance in younger adults, specifically word frequency. An intertrial paradigm was used whereby participants studied high- or low-frequency lists over several study-test trials, and the loss and gain of individual items was measured across trials; putative measures of consolidation and encoding. The results show that high-frequency words are recalled significantly better than low-frequency words. Older adults acquired high-frequency words at a greater rate across trials than they did for low-frequency words, an effect not evident in the younger adults. Older adults were found to have deficits in both encoding and consolidation as measured by losses and gains of items across trials. The results support the inter-item association theory of the word frequency effect on recall, with the age differences suggesting that memory deficits are sensitive to stimuli characteristics – one interpretation being that the ease of processing of the stimuli at encoding facilitates later recall.     

Implicit and explicit memory in young and older adults

In three experiments, young and older adults were compared on both implicit and explicit memory tasks. The size of repetition priming effects in word completion and in perceptual identification tasks did not differ reliably across ages. However, age-related decrements in performance were obtained in free recall, cued recall, and recognition. These results, similar to those observed in amnesics, suggest that older adults are impaired on tasks which require conscious recollection but that memory which depends on automatic activation processes in relatively unaffected by age.     

Memory for unidentified items: Evidence for the use of letter information in familiarity processes

Participants can give accurate recognition judgments to word fragments that they are unable to complete. In three experiments, the generality of this finding was examined across tasks. Accurate memory judgments in the absence of identification were obtained in item recognition and judgments of presentation frequency but not in associative recognition or list discrimination. The former two tasks are thought to involve the use of familiarity; the latter two are thought to rely on recollection. The present results are consistent with the claim that recognition without identification reflects familiarity processes.     

Does test‐induced priming play a role in the creation of false memories?

We investigated the role of test‐induced priming in creating false memories in the Deese/Roediger‐McDermott (DRM) paradigm, in which subjects study lists of related words (bed, rest, awake) and then falsely recall or recognise a related word (sleep) on a later test. However, in experiments using three different procedures, we found that the number of related words tested prior to the critical word had surprisingly little impact on false recall and recognition. We manipulated the location of the critical item in tests of yes/no recognition, word‐stem cued recall, and part‐set cued recall. We consistently obtained high probabilities of false recall and recognition, but the probability was unaffected by the number of related items presented prior to the test of the critical item. Surprisingly, test‐induced priming of the critical item does not seem to play a large role in this memory illusion.     

No Benefit of Eye-Closure Rehearsal in a Unimodal Recognition Memory Test for Word Items

Recent studies have suggested that closing the eyes helps memory retrieval in recall tests for audiovisual clips that contain multimodal information. In two experiments, we examined whether eye-closure improves recognition memory performance for word lists presented unimodally (i.e., visually or aurally). In the encoding phase, participants saw or heard a list of unrelated, meaningful word items. After a fixed retention interval of 1 week (Experiment 1, n = 110) and 5 min (Experiment 2, n = 44), the participants were asked to mentally rehearse the items with their eyes open or closed, and then they performed a recognition test. The results revealed no effect of eye-closure rehearsal on recognition performance. We discuss the possible reasons why no eye-closure benefit was found in recognition memory tests for unrelated word items.     

Recall of word lists is enhanced with increased spreading activation

ABSTRACT

Some evidence exists supporting a relationship between spreading activation in semantic/lexical memory networks and episodic memory. However, the results have been mixed and there have been no investigations examining whether a relationship exists between variability in spreading activation and episodic memory. Hence, we sought to investigate these potential relationships. Thirteen individuals were administered the Hopkins Verbal Learning Test–Revised (HVLT-R) and the Controlled Oral Word Association Test (COWAT). The average word frequency of all the “F” words generated on the COWAT was used as a measure of spreading activation. Variability in spreading activation was assessed by calculating the variability of the word frequencies from the COWAT across time. The results confirmed our hypotheses, with significant negative correlations found between free recall on the HVLT-R and both the average word frequency and measures of variability in spreading activation.     

Word length and age influences on forward and backward immediate serial recall

The present research is aimed at understanding the processes involved in short-term memory and how they interact with age. Specifically, word length effects were examined under forward serial recall, backward serial recall, and item recognition tasks, with performance being interpreted within an item-order theoretical framework. The interaction of age, word length, and direction of recall was examined in two experiments, the first of which confirmed that the word length was present with forward recall and absent with backward recall. In addition, age effects were stronger in backward recall than in forward recall. In the second experiment, an item-order trade-off methodology was utilized with backward recall. When order memory was required, there was no word length effect and strong age effects. When memory was tested via an item recognition test, there was a reverse word length effect and no age effect. While word length effects can be interpreted within the item-order framework, age effects cannot.     

What's in a name? access to information from people's names

Abstract

The processing of people's names is contrasted with face recognition and word recognition. The effects of the familiarity of initial and surnames and frequency of surnames (the number of people with the same surname) were investigated in several tasks. It was found that the effects of name familiarity and surname frequency were analogous to the effects of word frequency in tasks which did not require access to memory for individuals (a nationality decision and naming latency). In tasks which do require access to memory for individuals (familiarity decision and a semantic classification), the effect of surname frequency was analogous to the effect of distinctive-ness in face recognition. The results are discussed in terms of a functional model of name processing in which name recognition units mediate between the output of word recognition units and access to identity-specific semantics.     

Serial reconstruction of order and serial recall in verbal short-term memory

We carried out a series of experiments on verbal short-term memory for lists of words. In the first experiment, participants were tested via immediate serial recall, and word frequency and list set size were manipulated. With closed lists, the same set of items was repeatedly sampled, and with open lists, no item was presented more than once. In serial recall, effects of word frequency and set size were found. When a serial reconstruction-of-order task was used, in a second experiment, robust effects of word frequency emerged, but set size failed to show an effect. The effects of word frequency in order reconstruction were further examined in two final experiments. The data from these experiments revealed that the effects of word frequency are robust and apparently are not exclusively indicative of output processes. In light of these findings, we propose a multiple-mechanisms account in which word frequency can influence both retrieval and preretrieval processes.     

Depth of processing and recall of threat material in fearful and nonfearful individuals

Abstract

Although many studies have examined the nature of memory distortions in anxious individuals, few have considered biases in specific memory processes, such as encoding or retrieval. To investigate whether the presentation of threat material facilitates encoding biases, spider fearful (n=63), blood fearful (n=73), and nonfearful (n=75) participants encoded spider related, blood related, and neutral words as a function of three levels of processing (i.e., structural, semantic, and self referent). Participants subsequently completed either a free recall or a recognition task. All participants demonstrated a partial depth of processing effect, such that they recalled more words encoded in the self referent condition than in the other two conditions, but groups did not differ in their recall of stimuli as a function of word type. Relative to participants in the other groups, spider fearful participants had fewer spider related intrusions in the recall condition, and they made fewer errors in responding to structural and semantic encoding questions when spider related words were presented. These results contribute to an increasingly large body of literature suggesting that anxious individuals are not characterized by a memory bias toward threat, and they raise the possibility that individuals with spider fears process threat-relevant information differently than individuals with blood fears.     

Maintenance rehearsal: the key to the role attention plays in storage and forgetting

Research with the maintenance-rehearsal paradigm, in which word pairs are rehearsed as distractor material during a series of digit recall trials, has previously indicated that low frequency and new word pairs capture attention to a greater degree than high frequency and old word pairs. This impacts delayed recognition of the pairs and interferes with immediate digit recall. The effect on immediate digit recall may provide the missing converging evidence for the role of attention in memory. In the current study, 3 experiments were conducted to further investigate the role of attention capture and novelty in storage and forgetting. In addition to the previously established effects, the novelty of switching rehearsal between 2 pairs was found to impair both digit recall and memory for the first pair. The attentional effects we obtained were dependent upon participant expectation, and forgetting appears to be due to interference with consolidation rather than decay or traditional associative interference. Finally, the attentional effects we observed in associative recognition were primarily reflected in a lowering of the false alarm rate with increases in the strength of the parent pairs. Although dual-process models can accommodate this finding on the assumption that recollection is invoked at test alongside familiarity, we showed that the level of recall in this paradigm is so small that recollection can be ruled out. Accordingly, our results are challenging for the existing models of associative recognition to accommodate.     

Word-List-Learning Performance in Younger and Older Adults: Intra-Individual Performance Variability and False Memory

ABSTRACT

We investigated the word-list-learning performance of younger and older adults over 4 consecutive days at different times of day to study age-related differences in consistency of performance over time and the influence of circadian variation on performance. Eighteen younger (M age, 23.4 years) and 18 older (M age, 73.3 years) men and women participated. The start time of testing alternated between morning and early evening across the 4 days of testing. On each test day, participants learned a different list of 15 unrelated words over four learning trials. As expected, younger adults performed better than older adults on immediate recall, delayed recall, and recognition. Contrary to our expectations, time of day did not significantly influence recall or recognition performance in either the older or younger adults. Older adults did show a greater incidence of false memory (i.e., previously learned list intrusions in free recall and false alarms in recognition) than younger adults. Older adults also exhibited greater intra-individual performance variability on the measures of false memory across test days. This variability was not related to circadian variation. False memory and variability of performance have both been linked to frontal systems dysfunction. The findings presented here are consistent with the notion that changes in cognition with aging in part reflect age-related decline in frontal lobe function.     

Objects and properties: A study of the breakdown of semantic memory

Abstract

This paper reports a study of the breakdown of semantic memory in the case of a subject with semantic dementia. The first experiment shows that the subject failed to comprehend words of low familiarity and word frequency, even though the spoken word forms were recognised as familiar. Experiments 2 and 3 showed (a) that the recall of word meanings in definition tasks did not vary with the generality of the word meaning (e.g. category, basic level, or subordinate property) but varied instead with the concept familiarity and frequency of the name; (b) that the ability to verify properties of basic-level objects was not affected by the ability to comprehend the property name, but depended instead on the degree of knowledge demonstrated for the object name in definition tasks; (c) that properties were frequently verified correctly when the object had been defined only to the superordinate level. It is argued that the results do not support the widely held view that, in general, specific information is lost first when semantic memory breaks down. The selective failure to recall specific information for some word meanings is discussed with reference to two theoretical accounts.     

Presentation format and its effect on working memory

In three experiments, we examined the separate cognitive demands of processing and storage in working memory and looked at how effective the coordination was when items for storage varied in format/modality. A sentence verification task involving arithmetic facts was combined with a span task involving two to six items presented in picture, printed word, or spoken word format. The first two experiments were the same, except for the added requirement of articulation of the math sentence in Experiment 2. Experiment 3 varied the length of the span item and compared recall with recognition performance. The results showed that both spoken words and picture produced superior recall and recognition, as compared with printed words, and are consistent with Baddeley and Logie's (1999) and Mayer's (2001) models of working memory. Also, the differences in processing performance across spans varied with the difficulty of the task but showed the strongest support for the resource allocation model (Foos 1995).     

Effects of orthographic and phonological word length on memory for lists shown at RSVP and STM rates

This article reports 3 experiments in which effects of orthographic and phonological word length on memory were examined for short lists shown at rapid serial visual presentation (RSVP) and short-term memory (STM) rates. Only visual-orthographic length reduced RSVP serial recall, whereas both orthographic and phonological length lowered recall for STM lists in Experiment 1. Word-length effects may arise from output processes or from the temporal duration of output in recall. In 2 further experiments, output demands were reduced through the use of a recognition test. Recognition accuracy was impaired only by orthographic length for RSVP lists and by phonological length for STM lists in both experiments. The results demonstrate 2 item length effects not simply attributable to increased output time in recall, and implications for theories of STM are considered.     

Serial position-dependent false memory effects*

ABSTRACT

Evidence for false recognition within seconds of encoding suggests that semantic-associative influences are not restricted to long-term memory, consistent with unitary memory accounts but contrary to dual store models. The present study sought further relevant evidence using a modified free recall converging associates task where participants studied 12-item lists composed of 3 semantically distinct quartets (sublists) related to a separate, non-presented theme word (i.e., words 1–4/theme1, 5–8/theme2, and 9–12/theme3). This list construction permits assessment of false recall errors from each sublist, and, particularly, the primacy and recency sublists that have been linked to long- and short-term memory stores. Experiment 1 tested immediate free recall for items. Associative false memories were evident from all sublists, however, significantly less so from the recent sublist, which also showed the highest levels of veridical memory. By inserting a brief (3?s) distractor prior to recall, Experiment 2 selectively reduced veridical memory and increased false memory for the recent sublist while leaving the primacy sublist unaffected. These recall results converge with prior evidence indicating the immediacy of false recognition, and can be understood within a unitary framework where the differential availability of verbatim features and gist-based cues affect memory for primacy and recency sublists.     

The importance of cue familiarity and cue distinctiveness in prospective memory

Abstract

Both retrospective cued-memory tasks and event-based prospective memory tasks require that cue and target information be associated, and that aspects of that association be reinstated for successful remembering. These functional similarities between retrospective memory and prospective memory were the bases for the hypothesis that the familiarity and the distinctiveness of the target event (cue) would influence prospective memory performance. Experiment 1, focusing on target familiarity, found a nominal advantage in prospective memory with unfamiliar target events. Experiment 2 showed a significant benefit for unfamiliar target events, as well as for target events that were distinctive relative to the local context. Additionally, prospective memory performance did not reliably correlate with explicit retrospective memory tasks (recall and recognition), but did correlate with an indirect retrospective memory task (word fragment completion). This pattern suggests and helps specify the general view that prospective memory processes may be similar to those involved in both direct and indirect tests of retrospective memory.     

Implicit memory bias,explicit memory bias,and anxiety

Abstract

Normals high, medium, and low in trait anxiety performed two encoding tasks (one predominantly data-driven and the other conceptually driven) on threat-related and neutral words, followed by tests of word completion, cued recall, and free recall. Memory performance indicated the existence of negative memory biases in the high trait-anxious group, but it was generally not possible to decide whether the biases were associated with trait anxiety rather than with depression. The biases were obtained mainly when there was a match between the processes at encoding and those at the time of test, whether the matching processes were predominantly data-driven or conceptually driven. Implications of these findings for implicit and explicit memory biases associated with high trait anxiety are discussed.