Implicit processing in the cued recall of young and old adults.

Two cued recall experiments were reported in which younger and older subjects studied target words varying in number of preexperimental associates. In Experiment 1, targets were studied in either the absence or presence of meaning-related context cues, with recall always prompted by the cues. In the absence of context, words with smaller sets of associates were easier to recall than those with larger sets, but this effect was reduced for older subjects. The presence of a study context cue facilitated recall and eliminated the effect of associative set size for both ages. In Experiment 2, targets were studied and tested in the presence of unrelated words. In this situation, words with smaller sets of associates were less likely to be recalled than words with larger sets; again the effect was reduced for older subjects. The results are interpreted as an age decrement in processing implicitly activated information.     

Prior knowledge: the effects of natural category size on memory for implicitly encoded concepts

Studying a familiar word activates its associates, and these associates affect the likelihood of recalling the studied word in a cued recall task. These experiments examined variables that normally affect memory for the studied word to determine if they have similar effects on memory for the word's associates. Memory for associates was tested by cued recall (Experiments 1-3) or by recognition (Experiments 4-5), with the number and strength of the associates varied in all experiments. In different experiments, test instructions (direct-indirect), distractor tasks, lag, and amount of practice were manipulated. Provided that subjects were not distracted prior to test, the probability of recalling associates of the studied word decreased with the number of associates activated and with their strength under all conditions. The strength of the associates but not their number affected recognition. In general, variables that affected recall and recognition of studied words had parallel effects on their associates.     

Aging and Inhibition of Unconsciously Processed Information: No Apparent Deficit

Young and older adults were compared on a cued recall task designed to show effects of activation and inhibition of unconsciously processed information. The number of associates related to the target words by prior experience was varied, and the targets were studied either in the absence or the presence of meaningfully related context cues. The number of associates of the target had an effect on recall when context cues were absent during study, suggesting that the associates had been activated and encoded during study. In contrast, when context cues were present during study, no effect of the number of target associates was observed, suggesting that the associates had been inhibited during study. Comparable activation and inhibition effects were obtained for younger and older subjects. The results suggested little or no age-related deficit in inhibition when the to-be-inhibited information was unconsciously processed and not brought to conscious awareness.     

The fSAM model of false recall

The authors report a new theory of false memory building upon existing associative memory models and implemented in fSAM, the first fully specified quantitative model of false recall. Participants frequently intrude unstudied critical words while recalling lists comprising their strongest semantic associates but infrequently produce other extralist and prior-list intrusions. The authors developed the theory by simulating recall of such lists, using factorial combinations of semantic mechanisms operating at encoding, retrieval, or both stages. During encoding, unstudied words' associations to list context were strengthened in proportion to their strength of semantic association either to each studied word or to all co-rehearsed words. During retrieval, words received preference in proportion to their strength of semantic association to the most recently recalled single word or multiple words. The authors simulated all intrusion types and veridical recall for lists varying in semantic association strength among studied and critical words from the same and different lists. Multiplicative semantic encoding and retrieval mechanisms performed well in combination. Using such combined mechanisms, the authors also simulated several core findings from the Deese-Roediger-McDermott paradigm literature, including developmental patterns, specific list effects, association strength effects, and true-false correlations. These results challenge existing false-memory theories.     

Prior knowledge and cued recall: category size and dominance

Previous work indicates that our ability to recall a recently experienced word is reduced by the number of related concepts that it activates in permanent memory. The purpose of this research was to evaluate the possibility that dominance of primary associates, and not category size, is responsible for this observation. The results of three experiments involving manipulations of target and cue set size, as well as meaning and rhyme, indicate that category size effects are independent of dominance. In fact, the advantage of smaller categories and fewer activated items is substantially reduced for words having very dominant primary associates. The findings are discussed in relation to the Sensory-Semantic model.     

Can test list context manipulations improve recognition accuracy in the DRM paradigm?

Only test-based manipulations can be used to help people distinguish accurate from false memories once events have been encoded. In two experiments we examined how the type of studied words (weak vs strong associates, or less vs more memorable associates) and nonstudied lure words (related vs unrelated lures) on the test list affect recognition accuracy in the Deese-Roediger-McDermott paradigm. False recognition of critical lures decreased substantially in the related-lure context, but so did correct recognition of studied words. False recognition was little affected by the studied-word manipulations. In general, participants claimed to recognise critical lures as often as weak associates or less memorable studied words but less often than either strong associates or more memorable studied words. The test-list context affected how participants classified their recognition experiences but it did not systematically change their overall memory accuracy.     

Can test list context manipulations improve recognition accuracy in the DRM paradigm?

Only test-based manipulations can be used to help people distinguish accurate from false memories once events have been encoded. In two experiments we examined how the type of studied words (weak vs strong associates, or less vs more memorable associates) and nonstudied lure words (related vs unrelated lures) on the test list affect recognition accuracy in the Deese-Roediger-McDermott paradigm. False recognition of critical lures decreased substantially in the related-lure context, but so did correct recognition of studied words. False recognition was little affected by the studied-word manipulations. In general, participants claimed to recognise critical lures as often as weak associates or less memorable studied words but less often than either strong associates or more memorable studied words. The test-list context affected how participants classified their recognition experiences but it did not systematically change their overall memory accuracy.     

Prior knowledge and memory: the influence of natural category size as a function of intention and distraction

Words having larger numbers of related associates are more difficult to recall under conditions of intentional learning. This finding indicates that memory performance is inversely related to natural category size. The purpose of the experiments presented in this article was to evaluate the potential influence of orienting task and attention shifts associated with delayed testing on the natural category size effect. The results of Experiments 1 and 2 indicate that it is obtained regardless of task orientation. However, the results of Experiments 2 and 3 indicate that the effect is easily disrupted. Even after only 5 min of solving multiplication problems, words having larger sets of associates are no more difficult to recall than those having smaller sets. These findings suggest that related associates are automatically activated and encoded whenever a familiar concept is experienced in the absence of specific semantic cues. However, words having larger sets will be more difficult to recall only when these related associates are in a state of current activation.     

Evidence that nonconscious processes are sufficient to produce false memories

Are nonconscious processes sufficient to cause false memories of a nonstudied event? To investigate this issue, we controlled and measured conscious processing in the DRM task, in which studying associates (e.g., bed, rest, awake...) causes false memories of nonstudied associates (e.g., sleep). During the study phase, subjects studied visually masked associates at extremely rapid rates, followed by immediate recall. After this initial phase, nonstudied test words were rapidly presented for perceptual identification, followed by recognition memory judgments. On the perceptual identification task, we found significant priming of nonstudied associates, relative to control words. We also found significant false recognition of these nonstudied associates, even when subjects did not recall this word at study or identify it at test, indicating that nonconscious processes can cause false recognition. These recognition effects were found immediately after studying each list of associates, but not on a delayed test that occurred after the presentation of several intervening lists. Nonconscious processes are sufficient to cause this memory illusion on immediate tests, but may be insufficient for more vivid and lasting false memories.     

How does delayed testing reduce effects of implicit memory: Context infusion or cuing with context?

Studying a word implicitly activates related associates that affect its recall in the extralist cuing task. Recall is more likely when one of these associates becomes the test cue and when other associates of the studied word activate this associate. Context disruptions occurring between study and test reduce these effects, suggesting that implicitly activated memories are linked to context. The context infusion hypothesis assumes that context information spreads throughout a word's associative network during study, decreasing with distance from the studied word. Interactive cuing assumes that context is linked only to the studied word and that recall is based on retrieving information from the test cue and the context. The infusion hypothesis predicts that effects of disruptions will depend on link distance whereas interactive cuing predicts that distance will have no effect. The experiments evaluate these explanations by manipulating target-to-cue strength, associate-to-cue strength, and context disruption. Experiment 1 varies disruption by testing under the same or different conditions (room, modality, experimenter). Experiment 2 tests recall immediately or after 5 m, 10 m, or 20 m of multiplication. The results are inconsistent with context infusion and support the interactive cuing explanation.     

ASSOCIATIVE PROCESSES IN FALSE RECALL AND FALSE RECOGNITION

Abstract— Studying a list of words associated to a critical nonpresented word results in high rates of false recall and false recognition for that nonpresented item (Roediger & McDermott, 1995) Two experiments examined the effect of manipulating the number of associates presented on false recall and later false recognition of a nonpresented item. In Experiment 1, associate lists of varying lengths were studied, in Experiment 2, list length was held constant and the number of associates within the list was manipulated. In both experiments, the rate of critical intrusions in recall increased steadily with increasing number of associates studied Most notably, the filler words used in Experiment 2 to equate the list lengths did not affect the rate of critical intrusions, although they did depress recall of studied words. False recall and false recognition appear to be tied to the total, not the mean, associative strength of items in the list.     

Incidental formation of episodic associations: The importance of sentential context

The influence of relevant semantic context on the incidental formation of episodic associations between words was probed in two experiments. In Experiment 1, we examined the influence of associations formed incidentally between unrelated words presented either in isolation or embedded in a sentential context on subsequent explicit paired-associate leaning tested by cued recall. The results of Experiment 1 showed that the cued-recall rate of words studied in sentential context was higher than that of words co-occurring in isolated pairs. A subsequent single-items recognition test showed equal item memory for words studied in sentences than for words studied in isolated pairs, suggesting that the sentential context effect in cued recall indeed reflected stronger associations between paired words rather than better memory for single words. In Experiment 2, we ruled out memory for the entire sentence as an alternative explanation for the results of Experiment 1. We suggest two possible mechanisms to account for this advantage: First, pairs embedded in a sentence undergo semantic elaboration that might lead to the incidental formation of an association between them. Second, words embedded in a sentence enjoy the conjoint activation of compatible semantic features, a fact that may also facilitate the formation of an episodic association between them. The implications of these results for computational models using word representations based on co-occurrence data are discussed.     

Flexibility of information processing and the memory of field-independent and field-dependent learners

This study examined memory differences between 48 field-independent and 48 field-dependent learners, and it hypothesized that rigid information processing by field-dependent learners interferes with cue efficiency during memory recall. The field-independent and field-dependent learners were presented with paired associates in which the target recall words were homographs, and accompanying associates emphasized one of the homograph's two meanings. Homograph recall occurred under three conditions: (a) same cue recall, (b) different cue recall (emphasized other meaning of homograph), and (c) free recall. This study found memory differences favoring the field-independent learner and findings consistent with the rigid information processing hypothesis of field-dependent learners.     

Influences of intentional and unintentional forgetting on false memories

In 2 experiments, we examined the interplay of 2 types of memory errors: forgetting and false memory--errors of omission and commission, respectively. We examined the effects of 2 manipulations known to inhibit retrieval of studied words--directed forgetting and part-list cuing--on the false recall of an unstudied "critical" word following study of its 15 strongest associates. Participants cued to forget the 1st of 2 studied lists before studying the 2nd recalled fewer List 1 words but intruded the missing critical word more often than did participants cued to remember both lists. By contrast, providing some studied words as cues during recall reduced both recall of the remaining studied words and intrusions of the critical word. The results suggest that forgetting can increase or decrease false memories, depending on whether such forgetting reflects impaired access to an entire episode or retrieval competition among elements of an episode.     

The effect of divided attention on false memory depends on how memory is tested

In three experiments, we investigated the effects of divided attention on false memory, using the Deese/Roediger-McDermott (DRM) paradigm. In Experiments 1 and 2, participants studied six DRM lists with full attention and six in one of two divided-attention conditions (random number generation or digit monitoring). Both divided-attention conditions increased false recall of related words (Experiment 1) but reduced false recognition (Experiment 2). These results were confirmed in Experiment 3, in which the type of secondary task was manipulated within groups. We argue that the increase in false recall with divided attention reflects a change in participants' response criterion, whereas the decrease in false recognition occurs because the secondary tasks prevent participants from generating associates of the words presented at study.     

Sentence encoding and implicitly activated memories

Some words have fewer direct associates than others, and, when words varying in set size are studied in a list-learning task, those with smaller sets are more likely to be recalled. This set-size effect is found in cued recall when the words are studied in the absence of related words, but not when studied in the presence of related words. Related words provide context and theoretically inhibit irrelevant associates. The present research determined that set-size effects are found when words are encoded in sentence contexts. In contrast to list-learning experiments, the results of three experiments found such effects even when lexically related words were present in the sentences. Other findings indicated that target-set-size effects were determined by the proximity of related words in the sentence and the nature of the test cue. The results are discussed in relation to a model for explaining set-size effects and to selective findings from the sentence-comprehension literature.     

False recollection in children with reading comprehension difficulties

Children with reading comprehension difficulties display impaired performance on semantic processing tasks. These impairments are assumed to reflect weaker knowledge about abstract semantic associations between words in poor comprehenders [Nation, K., and Snowling, M. (1999). Developmental differences in sensitivity to semantic relations among good and poor comprehenders: evidence from semantic priming. Cognition, 19, B1-B13.]. We examined the performance of poor comprehenders on the Deese/Roediger/McDermott (DRM) paradigm. Children studied spoken words that were semantic associates (e.g., bed, rest, and awake) or phonological associates (e.g., pole, bowl, and hole) followed by free recall and a recognition test containing nonstudied critical words (e.g., sleep and roll). Results showed reduced recall and recognition of critical words in the semantic condition but not in the phonological condition for poor comprehenders. We argue that poor comprehenders are less sensitive to abstract semantic associations between words because of reduced gist memory.     

Strengthening the activation of unconsciously activated memories

Reading a word activates a set of associated words-for example, reading CAB activates car, driver, and other associates. This research was conducted to determine whether the unconscious activation of these associates adds activation to other associates in the same set. Words were studied, and recall was tested with cues comprised of associates directly activated by the target. On the basis of preexisting links, the associates serving as test cues also received strong (e.g., car) or weak (e.g., driver) additional inputs from other associates in the set. Cues receiving stronger inputs from other associates were more effective, indicating that unconsciously activated associates strengthen one another. This effect declined when conditions disrupted the retrieval of context features linked to the episode. Attending to another task during or after study or receiving implicit test instructions decreased the effects of input from other associates. Importantly, disruptions affected input from other associates more than direct input from the target, suggesting that the influence of context wanes as the links in the network become more distant.     

Spreading activation or spooky action at a distance?

How do preexisting connections among a word's associates facilitate its cued recall and recognition? A spreading-activation model assumes activation spreads to, among, and from a studied word's associates, and that its return is what strengthens its representation. An activation-at-a-distance model assumes strengthening is produced by the synchronous activation of the word's associates. The spread model predicts that connections among the studied word's associates will have a greater effect on memory when more of its associates return activation. The distance model predicts that total connections are important, not their direction. The results of cued recall experiments supported the distance model in showing that that connections among the associates facilitated recall regardless of the number of returning connections.     

Cue set size effects: Sampling activated associates or cross-target interference?

Previous findings indicate that test cues linked to more associates (more knowledge) produce lower levels of recall than cues with fewer associates. One hypothesis attributes this effect to cross-target interference arising during retrieval on the assumption that cues with more associates are more likely to be indirectly connected to studied words other than the target. Another attributes the effect to sampling associates of the cue on the assumption that the probability of sampling the target declines as more associates are activated. Findings from four experiments showed that recall varied with cue set size, and, more importantly, that cue set size affected recall independently of the interference produced by cross-target connections. These results were interpreted as supporting a model that attributes cue set size effects to sampling processes associated with the intersection of the test cue and its associates with the target and its associates.